Aren
/
NetBSD
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity
NetBSD/sys/dev/ic/rtw.c

3629 lines
88 KiB
C
Raw Normal View History

In rtw_detach, flag the rtw as "invalid" (RTW_F_INVALID). Do not try to power-down the RF section of an invalid rtw. The radiotap signal quality field is 16, not 8, bits wide. Convert the byte order.
2004-12-29 22:41:04 +03:00
/* $NetBSD: rtw.c,v 1.36 2004/12/29 19:41:04 dyoung Exp $ */
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
/*-
* Copyright (c) 2004, 2005 David Young. All rights reserved.
*
* Programmed for NetBSD by David Young.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
* 3. The name of David Young may not be used to endorse or promote
* products derived from this software without specific prior
* written permission.
*
* THIS SOFTWARE IS PROVIDED BY David Young ``AS IS'' AND ANY
* EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO,
* THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A
* PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL David
* Young BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
* EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED
* TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
* DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND
* ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
* OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
* OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY
* OF SUCH DAMAGE.
*/
/*
* Device driver for the Realtek RTL8180 802.11 MAC/BBP.
*/
#include <sys/cdefs.h>
In rtw_detach, flag the rtw as "invalid" (RTW_F_INVALID). Do not try to power-down the RF section of an invalid rtw. The radiotap signal quality field is 16, not 8, bits wide. Convert the byte order.
2004-12-29 22:41:04 +03:00
__KERNEL_RCSID(0, "$NetBSD: rtw.c,v 1.36 2004/12/29 19:41:04 dyoung Exp $");
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
#include "bpfilter.h"
#include <sys/param.h>
At last, I have rtw w/ Philips RF receiving packets. I added some sysctls to aid debugging: * hw.rtw.debug -- enable debugging * hw.rtw.flush_rfio -- Linux voodoo: possibly makes the MAC "flush" bits down the serial bus to the RF * hw.rtw.host_rfio: force the host to bang bits to the RF, instead of the MAC banging bits * hw.rtw.rfio_delay: after telling the MAC to bang bits to the RF front-end, delay rfio_delay microseconds. * hw.rtw.rfprog_fallback: there is this notion of the "RF programming method." I believe the choice influences the polarity/timing of the serial bus used to program the RF front-end. I know the correct choice for Intersil/RFMD/Philips front-ends, only. For all other front-ends, I "fallback" to rfprog_fallback. Make rtw_txdac_enable take an rtw_softc argument. I will probably revert this change. Add some Linux voodoo to rtw_continuous_tx_enable. I will probably revert this change. Important: add rtw_set_rfprog, which sets the correct RF programming method. This change and the following change are probably responsible for making the Philips RF work. Important: RTW_CONFIG1 is an 8-bit register, treat it that way! Important: RTW_BRSR is 16-bit, RTW_CRCOUNT, RTW_PHYDELAY, and RTW_MSR are 8-bit: treat them that way! Vastly simplify rtw_resume_ticks. Note to self: set the LED state to match the power state. Hedge against the possibility that RTW_MSR is protected as RTW_CONFIG[0123] are, meanwhile reworking that section of rtw_init a little. Add sc_anaparm, which isn't used, yet....
2004-12-13 03:48:02 +03:00
#include <sys/sysctl.h>
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
#include <sys/systm.h>
#include <sys/callout.h>
#include <sys/mbuf.h>
#include <sys/malloc.h>
#include <sys/kernel.h>
#include <sys/time.h>
#include <sys/types.h>
#include <machine/endian.h>
#include <machine/bus.h>
#include <machine/intr.h> /* splnet */
#include <uvm/uvm_extern.h>
#include <net/if.h>
#include <net/if_media.h>
#include <net/if_ether.h>
#include <net80211/ieee80211_var.h>
#include <net80211/ieee80211_compat.h>
#include <net80211/ieee80211_radiotap.h>
#if NBPFILTER > 0
#include <net/bpf.h>
#endif
#include <dev/ic/rtwreg.h>
#include <dev/ic/rtwvar.h>
#include <dev/ic/rtwphyio.h>
#include <dev/ic/rtwphy.h>
#include <dev/ic/smc93cx6var.h>
#define KASSERT2(__cond, __msg) \
do { \
if (!(__cond)) \
panic __msg ; \
} while (0)
At last, I have rtw w/ Philips RF receiving packets. I added some sysctls to aid debugging: * hw.rtw.debug -- enable debugging * hw.rtw.flush_rfio -- Linux voodoo: possibly makes the MAC "flush" bits down the serial bus to the RF * hw.rtw.host_rfio: force the host to bang bits to the RF, instead of the MAC banging bits * hw.rtw.rfio_delay: after telling the MAC to bang bits to the RF front-end, delay rfio_delay microseconds. * hw.rtw.rfprog_fallback: there is this notion of the "RF programming method." I believe the choice influences the polarity/timing of the serial bus used to program the RF front-end. I know the correct choice for Intersil/RFMD/Philips front-ends, only. For all other front-ends, I "fallback" to rfprog_fallback. Make rtw_txdac_enable take an rtw_softc argument. I will probably revert this change. Add some Linux voodoo to rtw_continuous_tx_enable. I will probably revert this change. Important: add rtw_set_rfprog, which sets the correct RF programming method. This change and the following change are probably responsible for making the Philips RF work. Important: RTW_CONFIG1 is an 8-bit register, treat it that way! Important: RTW_BRSR is 16-bit, RTW_CRCOUNT, RTW_PHYDELAY, and RTW_MSR are 8-bit: treat them that way! Vastly simplify rtw_resume_ticks. Note to self: set the LED state to match the power state. Hedge against the possibility that RTW_MSR is protected as RTW_CONFIG[0123] are, meanwhile reworking that section of rtw_init a little. Add sc_anaparm, which isn't used, yet....
2004-12-13 03:48:02 +03:00
int rtw_rfprog_fallback = 0;
int rtw_host_rfio = 0;
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
#ifdef RTW_DEBUG
Change rtw_debug from a debug level to a debug mask. Add a lot of debug flags. From Linux: handle an RTL8180 bug. Sometimes the NIC skips from the middle of the ring to the 0th rx descriptor. Now the driver resynchronizes. Handle a receive descriptor underrun or Rx FIFO overflow condition in the way that the Linux driver does. This kind of seems like overkill, but whatever. Protect rtw_ioctl with splnet(). Do not load a tx descriptor with a buffer shorter than 4 bytes. Handle a transmit timeout less disruptively.
2004-12-25 09:58:37 +03:00
int rtw_debug = 0;
If defined(RTW_DEBUG), provide a sysctl (hw.rtw.rxbufs_limit) for limiting the number of rx buffers an rtw may allocate. Use this sysctl to test the code that copes with buffer exhaustion. Allocate at most RTW_RXQLEN rx buffers, stopping at the sysctl limit. Record in sc_nrxdesc how many were allocated, and put the end-of-ring indication on sc_rxdesc[sc_nrxdesc - 1]. In rtw_init, if no rx buffers could be allocated, log a complaint, clear IFF_RUNNING, and exit with an error. Many changes to accomodate a short rx ring, mainly of the "add a rx-ring length argument" variety. XXX I really should consolidate all of the rx ring variables in one struct and pass that to the rx-ring subroutines. Bug fix: after calling MCLGET, use the (m->m_flags & M_EXT) idiom to check for success, instead of m != NULL. Bug fix: at the top of rtw_start, if IFF_RUNNING is not set, or IFF_OACTIVE is, get out.
2004-12-29 01:21:15 +03:00
int rtw_rxbufs_limit = RTW_RXQLEN;
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
#endif /* RTW_DEBUG */
Change rtw_debug from a debug level to a debug mask. Add a lot of debug flags. From Linux: handle an RTL8180 bug. Sometimes the NIC skips from the middle of the ring to the 0th rx descriptor. Now the driver resynchronizes. Handle a receive descriptor underrun or Rx FIFO overflow condition in the way that the Linux driver does. This kind of seems like overkill, but whatever. Protect rtw_ioctl with splnet(). Do not load a tx descriptor with a buffer shorter than 4 bytes. Handle a transmit timeout less disruptively.
2004-12-25 09:58:37 +03:00
#define NEXT_ATTACH_STATE(sc, state) do { \
DPRINTF(sc, RTW_DEBUG_ATTACH, \
("%s: attach state %s\n", __func__, #state)); \
sc->sc_attach_state = state; \
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
} while (0)
Remove useless debug sysctls. For AP scanning, change the channel dwell time from 1 second to 1/5 second.
2004-12-27 09:12:28 +03:00
int rtw_dwelltime = 200; /* milliseconds */
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
Add the transmit section of rtw(4): build up and tear down transmit descriptor rings, move packets from the transmit queues to the transmit rings, handle transmission-completed interrupts. My Linksys WPC11 ver. 4 with Maxim RF, generates interrupts like it is successfully transmitting packets. Clearly there are bugs: my G4 Powerbook locks up hard. I will debug tomorrow.
2004-12-19 11:19:25 +03:00
static void rtw_start(struct ifnet *);
At last, I have rtw w/ Philips RF receiving packets. I added some sysctls to aid debugging: * hw.rtw.debug -- enable debugging * hw.rtw.flush_rfio -- Linux voodoo: possibly makes the MAC "flush" bits down the serial bus to the RF * hw.rtw.host_rfio: force the host to bang bits to the RF, instead of the MAC banging bits * hw.rtw.rfio_delay: after telling the MAC to bang bits to the RF front-end, delay rfio_delay microseconds. * hw.rtw.rfprog_fallback: there is this notion of the "RF programming method." I believe the choice influences the polarity/timing of the serial bus used to program the RF front-end. I know the correct choice for Intersil/RFMD/Philips front-ends, only. For all other front-ends, I "fallback" to rfprog_fallback. Make rtw_txdac_enable take an rtw_softc argument. I will probably revert this change. Add some Linux voodoo to rtw_continuous_tx_enable. I will probably revert this change. Important: add rtw_set_rfprog, which sets the correct RF programming method. This change and the following change are probably responsible for making the Philips RF work. Important: RTW_CONFIG1 is an 8-bit register, treat it that way! Important: RTW_BRSR is 16-bit, RTW_CRCOUNT, RTW_PHYDELAY, and RTW_MSR are 8-bit: treat them that way! Vastly simplify rtw_resume_ticks. Note to self: set the LED state to match the power state. Hedge against the possibility that RTW_MSR is protected as RTW_CONFIG[0123] are, meanwhile reworking that section of rtw_init a little. Add sc_anaparm, which isn't used, yet....
2004-12-13 03:48:02 +03:00
static int rtw_sysctl_verify_rfio(SYSCTLFN_PROTO);
static int rtw_sysctl_verify_rfprog(SYSCTLFN_PROTO);
#ifdef RTW_DEBUG
Change rtw_debug from a debug level to a debug mask. Add a lot of debug flags. From Linux: handle an RTL8180 bug. Sometimes the NIC skips from the middle of the ring to the 0th rx descriptor. Now the driver resynchronizes. Handle a receive descriptor underrun or Rx FIFO overflow condition in the way that the Linux driver does. This kind of seems like overkill, but whatever. Protect rtw_ioctl with splnet(). Do not load a tx descriptor with a buffer shorter than 4 bytes. Handle a transmit timeout less disruptively.
2004-12-25 09:58:37 +03:00
static void rtw_print_txdesc(struct rtw_softc *, const char *,
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
struct rtw_txsoft *, struct rtw_txdesc_blk *, int);
At last, I have rtw w/ Philips RF receiving packets. I added some sysctls to aid debugging: * hw.rtw.debug -- enable debugging * hw.rtw.flush_rfio -- Linux voodoo: possibly makes the MAC "flush" bits down the serial bus to the RF * hw.rtw.host_rfio: force the host to bang bits to the RF, instead of the MAC banging bits * hw.rtw.rfio_delay: after telling the MAC to bang bits to the RF front-end, delay rfio_delay microseconds. * hw.rtw.rfprog_fallback: there is this notion of the "RF programming method." I believe the choice influences the polarity/timing of the serial bus used to program the RF front-end. I know the correct choice for Intersil/RFMD/Philips front-ends, only. For all other front-ends, I "fallback" to rfprog_fallback. Make rtw_txdac_enable take an rtw_softc argument. I will probably revert this change. Add some Linux voodoo to rtw_continuous_tx_enable. I will probably revert this change. Important: add rtw_set_rfprog, which sets the correct RF programming method. This change and the following change are probably responsible for making the Philips RF work. Important: RTW_CONFIG1 is an 8-bit register, treat it that way! Important: RTW_BRSR is 16-bit, RTW_CRCOUNT, RTW_PHYDELAY, and RTW_MSR are 8-bit: treat them that way! Vastly simplify rtw_resume_ticks. Note to self: set the LED state to match the power state. Hedge against the possibility that RTW_MSR is protected as RTW_CONFIG[0123] are, meanwhile reworking that section of rtw_init a little. Add sc_anaparm, which isn't used, yet....
2004-12-13 03:48:02 +03:00
static int rtw_sysctl_verify_debug(SYSCTLFN_PROTO);
If defined(RTW_DEBUG), provide a sysctl (hw.rtw.rxbufs_limit) for limiting the number of rx buffers an rtw may allocate. Use this sysctl to test the code that copes with buffer exhaustion. Allocate at most RTW_RXQLEN rx buffers, stopping at the sysctl limit. Record in sc_nrxdesc how many were allocated, and put the end-of-ring indication on sc_rxdesc[sc_nrxdesc - 1]. In rtw_init, if no rx buffers could be allocated, log a complaint, clear IFF_RUNNING, and exit with an error. Many changes to accomodate a short rx ring, mainly of the "add a rx-ring length argument" variety. XXX I really should consolidate all of the rx ring variables in one struct and pass that to the rx-ring subroutines. Bug fix: after calling MCLGET, use the (m->m_flags & M_EXT) idiom to check for success, instead of m != NULL. Bug fix: at the top of rtw_start, if IFF_RUNNING is not set, or IFF_OACTIVE is, get out.
2004-12-29 01:21:15 +03:00
static int rtw_sysctl_verify_rxbufs_limit(SYSCTLFN_PROTO);
At last, I have rtw w/ Philips RF receiving packets. I added some sysctls to aid debugging: * hw.rtw.debug -- enable debugging * hw.rtw.flush_rfio -- Linux voodoo: possibly makes the MAC "flush" bits down the serial bus to the RF * hw.rtw.host_rfio: force the host to bang bits to the RF, instead of the MAC banging bits * hw.rtw.rfio_delay: after telling the MAC to bang bits to the RF front-end, delay rfio_delay microseconds. * hw.rtw.rfprog_fallback: there is this notion of the "RF programming method." I believe the choice influences the polarity/timing of the serial bus used to program the RF front-end. I know the correct choice for Intersil/RFMD/Philips front-ends, only. For all other front-ends, I "fallback" to rfprog_fallback. Make rtw_txdac_enable take an rtw_softc argument. I will probably revert this change. Add some Linux voodoo to rtw_continuous_tx_enable. I will probably revert this change. Important: add rtw_set_rfprog, which sets the correct RF programming method. This change and the following change are probably responsible for making the Philips RF work. Important: RTW_CONFIG1 is an 8-bit register, treat it that way! Important: RTW_BRSR is 16-bit, RTW_CRCOUNT, RTW_PHYDELAY, and RTW_MSR are 8-bit: treat them that way! Vastly simplify rtw_resume_ticks. Note to self: set the LED state to match the power state. Hedge against the possibility that RTW_MSR is protected as RTW_CONFIG[0123] are, meanwhile reworking that section of rtw_init a little. Add sc_anaparm, which isn't used, yet....
2004-12-13 03:48:02 +03:00
#endif /* RTW_DEBUG */
/*
* Setup sysctl(3) MIB, hw.rtw.*
*
* TBD condition CTLFLAG_PERMANENT on being an LKM or not
*/
SYSCTL_SETUP(sysctl_rtw, "sysctl rtw(4) subtree setup")
{
int rc;
struct sysctlnode *cnode, *rnode;
if ((rc = sysctl_createv(clog, 0, NULL, &rnode,
CTLFLAG_PERMANENT, CTLTYPE_NODE, "hw", NULL,
NULL, 0, NULL, 0, CTL_HW, CTL_EOL)) != 0)
goto err;
if ((rc = sysctl_createv(clog, 0, &rnode, &rnode,
CTLFLAG_PERMANENT, CTLTYPE_NODE, "rtw",
"Realtek RTL818x 802.11 controls",
NULL, 0, NULL, 0, CTL_CREATE, CTL_EOL)) != 0)
goto err;
#ifdef RTW_DEBUG
/* control debugging printfs */
if ((rc = sysctl_createv(clog, 0, &rnode, &cnode,
CTLFLAG_PERMANENT|CTLFLAG_READWRITE, CTLTYPE_INT,
"debug", SYSCTL_DESCR("Enable RTL818x debugging output"),
rtw_sysctl_verify_debug, 0, &rtw_debug, 0,
CTL_CREATE, CTL_EOL)) != 0)
goto err;
If defined(RTW_DEBUG), provide a sysctl (hw.rtw.rxbufs_limit) for limiting the number of rx buffers an rtw may allocate. Use this sysctl to test the code that copes with buffer exhaustion. Allocate at most RTW_RXQLEN rx buffers, stopping at the sysctl limit. Record in sc_nrxdesc how many were allocated, and put the end-of-ring indication on sc_rxdesc[sc_nrxdesc - 1]. In rtw_init, if no rx buffers could be allocated, log a complaint, clear IFF_RUNNING, and exit with an error. Many changes to accomodate a short rx ring, mainly of the "add a rx-ring length argument" variety. XXX I really should consolidate all of the rx ring variables in one struct and pass that to the rx-ring subroutines. Bug fix: after calling MCLGET, use the (m->m_flags & M_EXT) idiom to check for success, instead of m != NULL. Bug fix: at the top of rtw_start, if IFF_RUNNING is not set, or IFF_OACTIVE is, get out.
2004-12-29 01:21:15 +03:00
/* Limit rx buffers, for simulating resource exhaustion. */
if ((rc = sysctl_createv(clog, 0, &rnode, &cnode,
CTLFLAG_PERMANENT|CTLFLAG_READWRITE, CTLTYPE_INT,
"rxbufs_limit",
SYSCTL_DESCR("Set rx buffers limit"),
rtw_sysctl_verify_rxbufs_limit, 0, &rtw_rxbufs_limit, 0,
CTL_CREATE, CTL_EOL)) != 0)
goto err;
At last, I have rtw w/ Philips RF receiving packets. I added some sysctls to aid debugging: * hw.rtw.debug -- enable debugging * hw.rtw.flush_rfio -- Linux voodoo: possibly makes the MAC "flush" bits down the serial bus to the RF * hw.rtw.host_rfio: force the host to bang bits to the RF, instead of the MAC banging bits * hw.rtw.rfio_delay: after telling the MAC to bang bits to the RF front-end, delay rfio_delay microseconds. * hw.rtw.rfprog_fallback: there is this notion of the "RF programming method." I believe the choice influences the polarity/timing of the serial bus used to program the RF front-end. I know the correct choice for Intersil/RFMD/Philips front-ends, only. For all other front-ends, I "fallback" to rfprog_fallback. Make rtw_txdac_enable take an rtw_softc argument. I will probably revert this change. Add some Linux voodoo to rtw_continuous_tx_enable. I will probably revert this change. Important: add rtw_set_rfprog, which sets the correct RF programming method. This change and the following change are probably responsible for making the Philips RF work. Important: RTW_CONFIG1 is an 8-bit register, treat it that way! Important: RTW_BRSR is 16-bit, RTW_CRCOUNT, RTW_PHYDELAY, and RTW_MSR are 8-bit: treat them that way! Vastly simplify rtw_resume_ticks. Note to self: set the LED state to match the power state. Hedge against the possibility that RTW_MSR is protected as RTW_CONFIG[0123] are, meanwhile reworking that section of rtw_init a little. Add sc_anaparm, which isn't used, yet....
2004-12-13 03:48:02 +03:00
#endif /* RTW_DEBUG */
/* set fallback RF programming method */
if ((rc = sysctl_createv(clog, 0, &rnode, &cnode,
CTLFLAG_PERMANENT|CTLFLAG_READWRITE, CTLTYPE_INT,
"rfprog_fallback",
SYSCTL_DESCR("Set fallback RF programming method"),
rtw_sysctl_verify_rfprog, 0, &rtw_rfprog_fallback, 0,
CTL_CREATE, CTL_EOL)) != 0)
goto err;
/* force host to control RF I/O bus */
if ((rc = sysctl_createv(clog, 0, &rnode, &cnode,
CTLFLAG_PERMANENT|CTLFLAG_READWRITE, CTLTYPE_INT,
"host_rfio", SYSCTL_DESCR("Enable host control of RF I/O"),
rtw_sysctl_verify_rfio, 0, &rtw_host_rfio, 0,
CTL_CREATE, CTL_EOL)) != 0)
goto err;
return;
err:
printf("%s: sysctl_createv failed (rc = %d)\n", __func__, rc);
}
static int
rtw_sysctl_verify(SYSCTLFN_ARGS, int lower, int upper)
{
int error, t;
struct sysctlnode node;
node = *rnode;
t = *(int*)rnode->sysctl_data;
node.sysctl_data = &t;
error = sysctl_lookup(SYSCTLFN_CALL(&node));
if (error || newp == NULL)
return (error);
if (t < lower || t > upper)
return (EINVAL);
*(int*)rnode->sysctl_data = t;
return (0);
}
static int
rtw_sysctl_verify_rfprog(SYSCTLFN_ARGS)
{
return rtw_sysctl_verify(SYSCTLFN_CALL(rnode), 0,
MASK_AND_RSHIFT(RTW_CONFIG4_RFTYPE_MASK, RTW_CONFIG4_RFTYPE_MASK));
}
static int
rtw_sysctl_verify_rfio(SYSCTLFN_ARGS)
{
return rtw_sysctl_verify(SYSCTLFN_CALL(rnode), 0, 1);
}
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
#ifdef RTW_DEBUG
At last, I have rtw w/ Philips RF receiving packets. I added some sysctls to aid debugging: * hw.rtw.debug -- enable debugging * hw.rtw.flush_rfio -- Linux voodoo: possibly makes the MAC "flush" bits down the serial bus to the RF * hw.rtw.host_rfio: force the host to bang bits to the RF, instead of the MAC banging bits * hw.rtw.rfio_delay: after telling the MAC to bang bits to the RF front-end, delay rfio_delay microseconds. * hw.rtw.rfprog_fallback: there is this notion of the "RF programming method." I believe the choice influences the polarity/timing of the serial bus used to program the RF front-end. I know the correct choice for Intersil/RFMD/Philips front-ends, only. For all other front-ends, I "fallback" to rfprog_fallback. Make rtw_txdac_enable take an rtw_softc argument. I will probably revert this change. Add some Linux voodoo to rtw_continuous_tx_enable. I will probably revert this change. Important: add rtw_set_rfprog, which sets the correct RF programming method. This change and the following change are probably responsible for making the Philips RF work. Important: RTW_CONFIG1 is an 8-bit register, treat it that way! Important: RTW_BRSR is 16-bit, RTW_CRCOUNT, RTW_PHYDELAY, and RTW_MSR are 8-bit: treat them that way! Vastly simplify rtw_resume_ticks. Note to self: set the LED state to match the power state. Hedge against the possibility that RTW_MSR is protected as RTW_CONFIG[0123] are, meanwhile reworking that section of rtw_init a little. Add sc_anaparm, which isn't used, yet....
2004-12-13 03:48:02 +03:00
static int
rtw_sysctl_verify_debug(SYSCTLFN_ARGS)
{
Change rtw_debug from a debug level to a debug mask. Add a lot of debug flags. From Linux: handle an RTL8180 bug. Sometimes the NIC skips from the middle of the ring to the 0th rx descriptor. Now the driver resynchronizes. Handle a receive descriptor underrun or Rx FIFO overflow condition in the way that the Linux driver does. This kind of seems like overkill, but whatever. Protect rtw_ioctl with splnet(). Do not load a tx descriptor with a buffer shorter than 4 bytes. Handle a transmit timeout less disruptively.
2004-12-25 09:58:37 +03:00
return rtw_sysctl_verify(SYSCTLFN_CALL(rnode), 0, RTW_DEBUG_MAX);
At last, I have rtw w/ Philips RF receiving packets. I added some sysctls to aid debugging: * hw.rtw.debug -- enable debugging * hw.rtw.flush_rfio -- Linux voodoo: possibly makes the MAC "flush" bits down the serial bus to the RF * hw.rtw.host_rfio: force the host to bang bits to the RF, instead of the MAC banging bits * hw.rtw.rfio_delay: after telling the MAC to bang bits to the RF front-end, delay rfio_delay microseconds. * hw.rtw.rfprog_fallback: there is this notion of the "RF programming method." I believe the choice influences the polarity/timing of the serial bus used to program the RF front-end. I know the correct choice for Intersil/RFMD/Philips front-ends, only. For all other front-ends, I "fallback" to rfprog_fallback. Make rtw_txdac_enable take an rtw_softc argument. I will probably revert this change. Add some Linux voodoo to rtw_continuous_tx_enable. I will probably revert this change. Important: add rtw_set_rfprog, which sets the correct RF programming method. This change and the following change are probably responsible for making the Philips RF work. Important: RTW_CONFIG1 is an 8-bit register, treat it that way! Important: RTW_BRSR is 16-bit, RTW_CRCOUNT, RTW_PHYDELAY, and RTW_MSR are 8-bit: treat them that way! Vastly simplify rtw_resume_ticks. Note to self: set the LED state to match the power state. Hedge against the possibility that RTW_MSR is protected as RTW_CONFIG[0123] are, meanwhile reworking that section of rtw_init a little. Add sc_anaparm, which isn't used, yet....
2004-12-13 03:48:02 +03:00
}
If defined(RTW_DEBUG), provide a sysctl (hw.rtw.rxbufs_limit) for limiting the number of rx buffers an rtw may allocate. Use this sysctl to test the code that copes with buffer exhaustion. Allocate at most RTW_RXQLEN rx buffers, stopping at the sysctl limit. Record in sc_nrxdesc how many were allocated, and put the end-of-ring indication on sc_rxdesc[sc_nrxdesc - 1]. In rtw_init, if no rx buffers could be allocated, log a complaint, clear IFF_RUNNING, and exit with an error. Many changes to accomodate a short rx ring, mainly of the "add a rx-ring length argument" variety. XXX I really should consolidate all of the rx ring variables in one struct and pass that to the rx-ring subroutines. Bug fix: after calling MCLGET, use the (m->m_flags & M_EXT) idiom to check for success, instead of m != NULL. Bug fix: at the top of rtw_start, if IFF_RUNNING is not set, or IFF_OACTIVE is, get out.
2004-12-29 01:21:15 +03:00
static int
rtw_sysctl_verify_rxbufs_limit(SYSCTLFN_ARGS)
{
return rtw_sysctl_verify(SYSCTLFN_CALL(rnode), 0, RTW_RXQLEN);
}
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
static void
rtw_print_regs(struct rtw_regs *regs, const char *dvname, const char *where)
{
Change rtw_debug from a debug level to a debug mask. Add a lot of debug flags. From Linux: handle an RTL8180 bug. Sometimes the NIC skips from the middle of the ring to the 0th rx descriptor. Now the driver resynchronizes. Handle a receive descriptor underrun or Rx FIFO overflow condition in the way that the Linux driver does. This kind of seems like overkill, but whatever. Protect rtw_ioctl with splnet(). Do not load a tx descriptor with a buffer shorter than 4 bytes. Handle a transmit timeout less disruptively.
2004-12-25 09:58:37 +03:00
#define PRINTREG32(sc, reg) \
RTW_DPRINTF(RTW_DEBUG_REGDUMP, \
("%s: reg[ " #reg " / %03x ] = %08x\n", \
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
dvname, reg, RTW_READ(regs, reg)))
Change rtw_debug from a debug level to a debug mask. Add a lot of debug flags. From Linux: handle an RTL8180 bug. Sometimes the NIC skips from the middle of the ring to the 0th rx descriptor. Now the driver resynchronizes. Handle a receive descriptor underrun or Rx FIFO overflow condition in the way that the Linux driver does. This kind of seems like overkill, but whatever. Protect rtw_ioctl with splnet(). Do not load a tx descriptor with a buffer shorter than 4 bytes. Handle a transmit timeout less disruptively.
2004-12-25 09:58:37 +03:00
#define PRINTREG16(sc, reg) \
RTW_DPRINTF(RTW_DEBUG_REGDUMP, \
("%s: reg[ " #reg " / %03x ] = %04x\n", \
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
dvname, reg, RTW_READ16(regs, reg)))
Change rtw_debug from a debug level to a debug mask. Add a lot of debug flags. From Linux: handle an RTL8180 bug. Sometimes the NIC skips from the middle of the ring to the 0th rx descriptor. Now the driver resynchronizes. Handle a receive descriptor underrun or Rx FIFO overflow condition in the way that the Linux driver does. This kind of seems like overkill, but whatever. Protect rtw_ioctl with splnet(). Do not load a tx descriptor with a buffer shorter than 4 bytes. Handle a transmit timeout less disruptively.
2004-12-25 09:58:37 +03:00
#define PRINTREG8(sc, reg) \
RTW_DPRINTF(RTW_DEBUG_REGDUMP, \
("%s: reg[ " #reg " / %03x ] = %02x\n", \
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
dvname, reg, RTW_READ8(regs, reg)))
Change rtw_debug from a debug level to a debug mask. Add a lot of debug flags. From Linux: handle an RTL8180 bug. Sometimes the NIC skips from the middle of the ring to the 0th rx descriptor. Now the driver resynchronizes. Handle a receive descriptor underrun or Rx FIFO overflow condition in the way that the Linux driver does. This kind of seems like overkill, but whatever. Protect rtw_ioctl with splnet(). Do not load a tx descriptor with a buffer shorter than 4 bytes. Handle a transmit timeout less disruptively.
2004-12-25 09:58:37 +03:00
RTW_DPRINTF(RTW_DEBUG_REGDUMP, ("%s: %s\n", dvname, where));
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
PRINTREG32(regs, RTW_IDR0);
PRINTREG32(regs, RTW_IDR1);
PRINTREG32(regs, RTW_MAR0);
PRINTREG32(regs, RTW_MAR1);
PRINTREG32(regs, RTW_TSFTRL);
PRINTREG32(regs, RTW_TSFTRH);
PRINTREG32(regs, RTW_TLPDA);
PRINTREG32(regs, RTW_TNPDA);
PRINTREG32(regs, RTW_THPDA);
PRINTREG32(regs, RTW_TCR);
PRINTREG32(regs, RTW_RCR);
PRINTREG32(regs, RTW_TINT);
PRINTREG32(regs, RTW_TBDA);
PRINTREG32(regs, RTW_ANAPARM);
PRINTREG32(regs, RTW_BB);
PRINTREG32(regs, RTW_PHYCFG);
PRINTREG32(regs, RTW_WAKEUP0L);
PRINTREG32(regs, RTW_WAKEUP0H);
PRINTREG32(regs, RTW_WAKEUP1L);
PRINTREG32(regs, RTW_WAKEUP1H);
PRINTREG32(regs, RTW_WAKEUP2LL);
PRINTREG32(regs, RTW_WAKEUP2LH);
PRINTREG32(regs, RTW_WAKEUP2HL);
PRINTREG32(regs, RTW_WAKEUP2HH);
PRINTREG32(regs, RTW_WAKEUP3LL);
PRINTREG32(regs, RTW_WAKEUP3LH);
PRINTREG32(regs, RTW_WAKEUP3HL);
PRINTREG32(regs, RTW_WAKEUP3HH);
PRINTREG32(regs, RTW_WAKEUP4LL);
PRINTREG32(regs, RTW_WAKEUP4LH);
PRINTREG32(regs, RTW_WAKEUP4HL);
PRINTREG32(regs, RTW_WAKEUP4HH);
PRINTREG32(regs, RTW_DK0);
PRINTREG32(regs, RTW_DK1);
PRINTREG32(regs, RTW_DK2);
PRINTREG32(regs, RTW_DK3);
PRINTREG32(regs, RTW_RETRYCTR);
PRINTREG32(regs, RTW_RDSAR);
PRINTREG32(regs, RTW_FER);
PRINTREG32(regs, RTW_FEMR);
PRINTREG32(regs, RTW_FPSR);
PRINTREG32(regs, RTW_FFER);
/* 16-bit registers */
PRINTREG16(regs, RTW_BRSR);
PRINTREG16(regs, RTW_IMR);
PRINTREG16(regs, RTW_ISR);
PRINTREG16(regs, RTW_BCNITV);
PRINTREG16(regs, RTW_ATIMWND);
PRINTREG16(regs, RTW_BINTRITV);
PRINTREG16(regs, RTW_ATIMTRITV);
PRINTREG16(regs, RTW_CRC16ERR);
PRINTREG16(regs, RTW_CRC0);
PRINTREG16(regs, RTW_CRC1);
PRINTREG16(regs, RTW_CRC2);
PRINTREG16(regs, RTW_CRC3);
PRINTREG16(regs, RTW_CRC4);
PRINTREG16(regs, RTW_CWR);
/* 8-bit registers */
PRINTREG8(regs, RTW_CR);
PRINTREG8(regs, RTW_9346CR);
PRINTREG8(regs, RTW_CONFIG0);
PRINTREG8(regs, RTW_CONFIG1);
PRINTREG8(regs, RTW_CONFIG2);
PRINTREG8(regs, RTW_MSR);
PRINTREG8(regs, RTW_CONFIG3);
PRINTREG8(regs, RTW_CONFIG4);
PRINTREG8(regs, RTW_TESTR);
PRINTREG8(regs, RTW_PSR);
PRINTREG8(regs, RTW_SCR);
PRINTREG8(regs, RTW_PHYDELAY);
PRINTREG8(regs, RTW_CRCOUNT);
PRINTREG8(regs, RTW_PHYADDR);
PRINTREG8(regs, RTW_PHYDATAW);
PRINTREG8(regs, RTW_PHYDATAR);
PRINTREG8(regs, RTW_CONFIG5);
PRINTREG8(regs, RTW_TPPOLL);
PRINTREG16(regs, RTW_BSSID16);
PRINTREG32(regs, RTW_BSSID32);
#undef PRINTREG32
#undef PRINTREG16
#undef PRINTREG8
}
#endif /* RTW_DEBUG */
void
Miscellaneous changes. Details below. Important changes flagged with []. Using the driver with my Linksys WPC11 ver. 4, it seems to be receiving packets for a change. The WPC11 ver. 4 has a Maxim RF section. My no-name rtw with Philips RF section still does not receive any packets. Keep access-level (analog params > config[0123] registers > none) in sc_access. Add rtw_set_access for changing the access level. Make rtw_continuous_tx_enable and other subroutines use rtw_set_access instead of rtw_config0123_enable and rtw_anaparm_enable. Factor part of the chip-reset code into rtw_chip_reset1. Change the 'struct foo (*bar)[N]'-style arguments to 'struct foo *bar'-style arguments. Consolidate software/hardware Tx/Rx ring setup in rtw_hwring_setup, rtw_swring_setup. Add a new constant, SA2400_OPMODE_DEFAULTS, for the bits that we *always* set in the SA2400 OPMODE register. Factor some code out into rtw_sa2400_calibrate. (Inspired by the Linux driver.) [] When the receiver goes into underrun/overflow state, call a new subroutine, rtw_kick() that stops the Rx/Tx processes, resets the chip, reinitializes the Tx/Rx rings, and restarts Rx/Tx processes. (Inspired by the Linux driver.) [] In rtw_intr_rx, check for too-short packets before calling ieee80211_find_rxnode. I believe this will prevent a repeat of the MCHK exception I saw once on macppc. [] Use seconds-elapased as well as microseconds-elapsed to set the next "due date" for the timeout interrupt. This keeps the driver from programming the timeout to expire too early. [] In rtw_intr, read RTW_ISR at most 10 times, then get out. If the interface is not enabled (RTW_F_ENABLED), then get out. [] In rtw_stop, get out if the interface is not enabled (RTW_F_ENABLED). Block IPL_NET interrupts. Don't read/write any registers if the interface is invalid (RTW_F_INVALID). [] Call rtw_stop in rtw_detach.
2004-12-12 09:37:59 +03:00
rtw_continuous_tx_enable(struct rtw_softc *sc, int enable)
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
{
Miscellaneous changes. Details below. Important changes flagged with []. Using the driver with my Linksys WPC11 ver. 4, it seems to be receiving packets for a change. The WPC11 ver. 4 has a Maxim RF section. My no-name rtw with Philips RF section still does not receive any packets. Keep access-level (analog params > config[0123] registers > none) in sc_access. Add rtw_set_access for changing the access level. Make rtw_continuous_tx_enable and other subroutines use rtw_set_access instead of rtw_config0123_enable and rtw_anaparm_enable. Factor part of the chip-reset code into rtw_chip_reset1. Change the 'struct foo (*bar)[N]'-style arguments to 'struct foo *bar'-style arguments. Consolidate software/hardware Tx/Rx ring setup in rtw_hwring_setup, rtw_swring_setup. Add a new constant, SA2400_OPMODE_DEFAULTS, for the bits that we *always* set in the SA2400 OPMODE register. Factor some code out into rtw_sa2400_calibrate. (Inspired by the Linux driver.) [] When the receiver goes into underrun/overflow state, call a new subroutine, rtw_kick() that stops the Rx/Tx processes, resets the chip, reinitializes the Tx/Rx rings, and restarts Rx/Tx processes. (Inspired by the Linux driver.) [] In rtw_intr_rx, check for too-short packets before calling ieee80211_find_rxnode. I believe this will prevent a repeat of the MCHK exception I saw once on macppc. [] Use seconds-elapased as well as microseconds-elapsed to set the next "due date" for the timeout interrupt. This keeps the driver from programming the timeout to expire too early. [] In rtw_intr, read RTW_ISR at most 10 times, then get out. If the interface is not enabled (RTW_F_ENABLED), then get out. [] In rtw_stop, get out if the interface is not enabled (RTW_F_ENABLED). Block IPL_NET interrupts. Don't read/write any registers if the interface is invalid (RTW_F_INVALID). [] Call rtw_stop in rtw_detach.
2004-12-12 09:37:59 +03:00
struct rtw_regs *regs = &sc->sc_regs;
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
u_int32_t tcr;
tcr = RTW_READ(regs, RTW_TCR);
tcr &= ~RTW_TCR_LBK_MASK;
if (enable)
tcr |= RTW_TCR_LBK_CONT;
else
tcr |= RTW_TCR_LBK_NORMAL;
RTW_WRITE(regs, RTW_TCR, tcr);
RTW_SYNC(regs, RTW_TCR, RTW_TCR);
Miscellaneous changes. Details below. Important changes flagged with []. Using the driver with my Linksys WPC11 ver. 4, it seems to be receiving packets for a change. The WPC11 ver. 4 has a Maxim RF section. My no-name rtw with Philips RF section still does not receive any packets. Keep access-level (analog params > config[0123] registers > none) in sc_access. Add rtw_set_access for changing the access level. Make rtw_continuous_tx_enable and other subroutines use rtw_set_access instead of rtw_config0123_enable and rtw_anaparm_enable. Factor part of the chip-reset code into rtw_chip_reset1. Change the 'struct foo (*bar)[N]'-style arguments to 'struct foo *bar'-style arguments. Consolidate software/hardware Tx/Rx ring setup in rtw_hwring_setup, rtw_swring_setup. Add a new constant, SA2400_OPMODE_DEFAULTS, for the bits that we *always* set in the SA2400 OPMODE register. Factor some code out into rtw_sa2400_calibrate. (Inspired by the Linux driver.) [] When the receiver goes into underrun/overflow state, call a new subroutine, rtw_kick() that stops the Rx/Tx processes, resets the chip, reinitializes the Tx/Rx rings, and restarts Rx/Tx processes. (Inspired by the Linux driver.) [] In rtw_intr_rx, check for too-short packets before calling ieee80211_find_rxnode. I believe this will prevent a repeat of the MCHK exception I saw once on macppc. [] Use seconds-elapased as well as microseconds-elapsed to set the next "due date" for the timeout interrupt. This keeps the driver from programming the timeout to expire too early. [] In rtw_intr, read RTW_ISR at most 10 times, then get out. If the interface is not enabled (RTW_F_ENABLED), then get out. [] In rtw_stop, get out if the interface is not enabled (RTW_F_ENABLED). Block IPL_NET interrupts. Don't read/write any registers if the interface is invalid (RTW_F_INVALID). [] Call rtw_stop in rtw_detach.
2004-12-12 09:37:59 +03:00
rtw_set_access(sc, RTW_ACCESS_ANAPARM);
At last, I have rtw w/ Philips RF receiving packets. I added some sysctls to aid debugging: * hw.rtw.debug -- enable debugging * hw.rtw.flush_rfio -- Linux voodoo: possibly makes the MAC "flush" bits down the serial bus to the RF * hw.rtw.host_rfio: force the host to bang bits to the RF, instead of the MAC banging bits * hw.rtw.rfio_delay: after telling the MAC to bang bits to the RF front-end, delay rfio_delay microseconds. * hw.rtw.rfprog_fallback: there is this notion of the "RF programming method." I believe the choice influences the polarity/timing of the serial bus used to program the RF front-end. I know the correct choice for Intersil/RFMD/Philips front-ends, only. For all other front-ends, I "fallback" to rfprog_fallback. Make rtw_txdac_enable take an rtw_softc argument. I will probably revert this change. Add some Linux voodoo to rtw_continuous_tx_enable. I will probably revert this change. Important: add rtw_set_rfprog, which sets the correct RF programming method. This change and the following change are probably responsible for making the Philips RF work. Important: RTW_CONFIG1 is an 8-bit register, treat it that way! Important: RTW_BRSR is 16-bit, RTW_CRCOUNT, RTW_PHYDELAY, and RTW_MSR are 8-bit: treat them that way! Vastly simplify rtw_resume_ticks. Note to self: set the LED state to match the power state. Hedge against the possibility that RTW_MSR is protected as RTW_CONFIG[0123] are, meanwhile reworking that section of rtw_init a little. Add sc_anaparm, which isn't used, yet....
2004-12-13 03:48:02 +03:00
rtw_txdac_enable(sc, !enable);
rtw_set_access(sc, RTW_ACCESS_ANAPARM); /* XXX Voodoo from Linux. */
Miscellaneous changes. Details below. Important changes flagged with []. Using the driver with my Linksys WPC11 ver. 4, it seems to be receiving packets for a change. The WPC11 ver. 4 has a Maxim RF section. My no-name rtw with Philips RF section still does not receive any packets. Keep access-level (analog params > config[0123] registers > none) in sc_access. Add rtw_set_access for changing the access level. Make rtw_continuous_tx_enable and other subroutines use rtw_set_access instead of rtw_config0123_enable and rtw_anaparm_enable. Factor part of the chip-reset code into rtw_chip_reset1. Change the 'struct foo (*bar)[N]'-style arguments to 'struct foo *bar'-style arguments. Consolidate software/hardware Tx/Rx ring setup in rtw_hwring_setup, rtw_swring_setup. Add a new constant, SA2400_OPMODE_DEFAULTS, for the bits that we *always* set in the SA2400 OPMODE register. Factor some code out into rtw_sa2400_calibrate. (Inspired by the Linux driver.) [] When the receiver goes into underrun/overflow state, call a new subroutine, rtw_kick() that stops the Rx/Tx processes, resets the chip, reinitializes the Tx/Rx rings, and restarts Rx/Tx processes. (Inspired by the Linux driver.) [] In rtw_intr_rx, check for too-short packets before calling ieee80211_find_rxnode. I believe this will prevent a repeat of the MCHK exception I saw once on macppc. [] Use seconds-elapased as well as microseconds-elapsed to set the next "due date" for the timeout interrupt. This keeps the driver from programming the timeout to expire too early. [] In rtw_intr, read RTW_ISR at most 10 times, then get out. If the interface is not enabled (RTW_F_ENABLED), then get out. [] In rtw_stop, get out if the interface is not enabled (RTW_F_ENABLED). Block IPL_NET interrupts. Don't read/write any registers if the interface is invalid (RTW_F_INVALID). [] Call rtw_stop in rtw_detach.
2004-12-12 09:37:59 +03:00
rtw_set_access(sc, RTW_ACCESS_NONE);
}
Make rtw(4) build w/o RTW_DEBUG. Thanks to Charles Hannum for pointing out my oversight.
2004-12-27 01:59:41 +03:00
#ifdef RTW_DEBUG
Miscellaneous changes. Details below. Important changes flagged with []. Using the driver with my Linksys WPC11 ver. 4, it seems to be receiving packets for a change. The WPC11 ver. 4 has a Maxim RF section. My no-name rtw with Philips RF section still does not receive any packets. Keep access-level (analog params > config[0123] registers > none) in sc_access. Add rtw_set_access for changing the access level. Make rtw_continuous_tx_enable and other subroutines use rtw_set_access instead of rtw_config0123_enable and rtw_anaparm_enable. Factor part of the chip-reset code into rtw_chip_reset1. Change the 'struct foo (*bar)[N]'-style arguments to 'struct foo *bar'-style arguments. Consolidate software/hardware Tx/Rx ring setup in rtw_hwring_setup, rtw_swring_setup. Add a new constant, SA2400_OPMODE_DEFAULTS, for the bits that we *always* set in the SA2400 OPMODE register. Factor some code out into rtw_sa2400_calibrate. (Inspired by the Linux driver.) [] When the receiver goes into underrun/overflow state, call a new subroutine, rtw_kick() that stops the Rx/Tx processes, resets the chip, reinitializes the Tx/Rx rings, and restarts Rx/Tx processes. (Inspired by the Linux driver.) [] In rtw_intr_rx, check for too-short packets before calling ieee80211_find_rxnode. I believe this will prevent a repeat of the MCHK exception I saw once on macppc. [] Use seconds-elapased as well as microseconds-elapsed to set the next "due date" for the timeout interrupt. This keeps the driver from programming the timeout to expire too early. [] In rtw_intr, read RTW_ISR at most 10 times, then get out. If the interface is not enabled (RTW_F_ENABLED), then get out. [] In rtw_stop, get out if the interface is not enabled (RTW_F_ENABLED). Block IPL_NET interrupts. Don't read/write any registers if the interface is invalid (RTW_F_INVALID). [] Call rtw_stop in rtw_detach.
2004-12-12 09:37:59 +03:00
static const char *
rtw_access_string(enum rtw_access access)
{
switch (access) {
case RTW_ACCESS_NONE:
return "none";
case RTW_ACCESS_CONFIG:
return "config";
case RTW_ACCESS_ANAPARM:
return "anaparm";
default:
return "unknown";
}
}
Make rtw(4) build w/o RTW_DEBUG. Thanks to Charles Hannum for pointing out my oversight.
2004-12-27 01:59:41 +03:00
#endif /* RTW_DEBUG */
Miscellaneous changes. Details below. Important changes flagged with []. Using the driver with my Linksys WPC11 ver. 4, it seems to be receiving packets for a change. The WPC11 ver. 4 has a Maxim RF section. My no-name rtw with Philips RF section still does not receive any packets. Keep access-level (analog params > config[0123] registers > none) in sc_access. Add rtw_set_access for changing the access level. Make rtw_continuous_tx_enable and other subroutines use rtw_set_access instead of rtw_config0123_enable and rtw_anaparm_enable. Factor part of the chip-reset code into rtw_chip_reset1. Change the 'struct foo (*bar)[N]'-style arguments to 'struct foo *bar'-style arguments. Consolidate software/hardware Tx/Rx ring setup in rtw_hwring_setup, rtw_swring_setup. Add a new constant, SA2400_OPMODE_DEFAULTS, for the bits that we *always* set in the SA2400 OPMODE register. Factor some code out into rtw_sa2400_calibrate. (Inspired by the Linux driver.) [] When the receiver goes into underrun/overflow state, call a new subroutine, rtw_kick() that stops the Rx/Tx processes, resets the chip, reinitializes the Tx/Rx rings, and restarts Rx/Tx processes. (Inspired by the Linux driver.) [] In rtw_intr_rx, check for too-short packets before calling ieee80211_find_rxnode. I believe this will prevent a repeat of the MCHK exception I saw once on macppc. [] Use seconds-elapased as well as microseconds-elapsed to set the next "due date" for the timeout interrupt. This keeps the driver from programming the timeout to expire too early. [] In rtw_intr, read RTW_ISR at most 10 times, then get out. If the interface is not enabled (RTW_F_ENABLED), then get out. [] In rtw_stop, get out if the interface is not enabled (RTW_F_ENABLED). Block IPL_NET interrupts. Don't read/write any registers if the interface is invalid (RTW_F_INVALID). [] Call rtw_stop in rtw_detach.
2004-12-12 09:37:59 +03:00
static void
rtw_set_access1(struct rtw_regs *regs,
enum rtw_access oaccess, enum rtw_access naccess)
{
KASSERT(naccess >= RTW_ACCESS_NONE && naccess <= RTW_ACCESS_ANAPARM);
KASSERT(oaccess >= RTW_ACCESS_NONE && oaccess <= RTW_ACCESS_ANAPARM);
if (naccess == oaccess)
return;
switch (naccess) {
case RTW_ACCESS_NONE:
switch (oaccess) {
case RTW_ACCESS_ANAPARM:
rtw_anaparm_enable(regs, 0);
/*FALLTHROUGH*/
case RTW_ACCESS_CONFIG:
rtw_config0123_enable(regs, 0);
/*FALLTHROUGH*/
case RTW_ACCESS_NONE:
break;
}
break;
case RTW_ACCESS_CONFIG:
switch (oaccess) {
case RTW_ACCESS_NONE:
rtw_config0123_enable(regs, 1);
/*FALLTHROUGH*/
case RTW_ACCESS_CONFIG:
break;
case RTW_ACCESS_ANAPARM:
rtw_anaparm_enable(regs, 0);
break;
}
break;
case RTW_ACCESS_ANAPARM:
switch (oaccess) {
case RTW_ACCESS_NONE:
rtw_config0123_enable(regs, 1);
/*FALLTHROUGH*/
case RTW_ACCESS_CONFIG:
rtw_anaparm_enable(regs, 1);
/*FALLTHROUGH*/
case RTW_ACCESS_ANAPARM:
break;
}
break;
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
}
}
Miscellaneous changes. Details below. Important changes flagged with []. Using the driver with my Linksys WPC11 ver. 4, it seems to be receiving packets for a change. The WPC11 ver. 4 has a Maxim RF section. My no-name rtw with Philips RF section still does not receive any packets. Keep access-level (analog params > config[0123] registers > none) in sc_access. Add rtw_set_access for changing the access level. Make rtw_continuous_tx_enable and other subroutines use rtw_set_access instead of rtw_config0123_enable and rtw_anaparm_enable. Factor part of the chip-reset code into rtw_chip_reset1. Change the 'struct foo (*bar)[N]'-style arguments to 'struct foo *bar'-style arguments. Consolidate software/hardware Tx/Rx ring setup in rtw_hwring_setup, rtw_swring_setup. Add a new constant, SA2400_OPMODE_DEFAULTS, for the bits that we *always* set in the SA2400 OPMODE register. Factor some code out into rtw_sa2400_calibrate. (Inspired by the Linux driver.) [] When the receiver goes into underrun/overflow state, call a new subroutine, rtw_kick() that stops the Rx/Tx processes, resets the chip, reinitializes the Tx/Rx rings, and restarts Rx/Tx processes. (Inspired by the Linux driver.) [] In rtw_intr_rx, check for too-short packets before calling ieee80211_find_rxnode. I believe this will prevent a repeat of the MCHK exception I saw once on macppc. [] Use seconds-elapased as well as microseconds-elapsed to set the next "due date" for the timeout interrupt. This keeps the driver from programming the timeout to expire too early. [] In rtw_intr, read RTW_ISR at most 10 times, then get out. If the interface is not enabled (RTW_F_ENABLED), then get out. [] In rtw_stop, get out if the interface is not enabled (RTW_F_ENABLED). Block IPL_NET interrupts. Don't read/write any registers if the interface is invalid (RTW_F_INVALID). [] Call rtw_stop in rtw_detach.
2004-12-12 09:37:59 +03:00
void
rtw_set_access(struct rtw_softc *sc, enum rtw_access access)
{
rtw_set_access1(&sc->sc_regs, sc->sc_access, access);
Change rtw_debug from a debug level to a debug mask. Add a lot of debug flags. From Linux: handle an RTL8180 bug. Sometimes the NIC skips from the middle of the ring to the 0th rx descriptor. Now the driver resynchronizes. Handle a receive descriptor underrun or Rx FIFO overflow condition in the way that the Linux driver does. This kind of seems like overkill, but whatever. Protect rtw_ioctl with splnet(). Do not load a tx descriptor with a buffer shorter than 4 bytes. Handle a transmit timeout less disruptively.
2004-12-25 09:58:37 +03:00
RTW_DPRINTF(RTW_DEBUG_ACCESS,
("%s: access %s -> %s\n", sc->sc_dev.dv_xname,
Miscellaneous changes. Details below. Important changes flagged with []. Using the driver with my Linksys WPC11 ver. 4, it seems to be receiving packets for a change. The WPC11 ver. 4 has a Maxim RF section. My no-name rtw with Philips RF section still does not receive any packets. Keep access-level (analog params > config[0123] registers > none) in sc_access. Add rtw_set_access for changing the access level. Make rtw_continuous_tx_enable and other subroutines use rtw_set_access instead of rtw_config0123_enable and rtw_anaparm_enable. Factor part of the chip-reset code into rtw_chip_reset1. Change the 'struct foo (*bar)[N]'-style arguments to 'struct foo *bar'-style arguments. Consolidate software/hardware Tx/Rx ring setup in rtw_hwring_setup, rtw_swring_setup. Add a new constant, SA2400_OPMODE_DEFAULTS, for the bits that we *always* set in the SA2400 OPMODE register. Factor some code out into rtw_sa2400_calibrate. (Inspired by the Linux driver.) [] When the receiver goes into underrun/overflow state, call a new subroutine, rtw_kick() that stops the Rx/Tx processes, resets the chip, reinitializes the Tx/Rx rings, and restarts Rx/Tx processes. (Inspired by the Linux driver.) [] In rtw_intr_rx, check for too-short packets before calling ieee80211_find_rxnode. I believe this will prevent a repeat of the MCHK exception I saw once on macppc. [] Use seconds-elapased as well as microseconds-elapsed to set the next "due date" for the timeout interrupt. This keeps the driver from programming the timeout to expire too early. [] In rtw_intr, read RTW_ISR at most 10 times, then get out. If the interface is not enabled (RTW_F_ENABLED), then get out. [] In rtw_stop, get out if the interface is not enabled (RTW_F_ENABLED). Block IPL_NET interrupts. Don't read/write any registers if the interface is invalid (RTW_F_INVALID). [] Call rtw_stop in rtw_detach.
2004-12-12 09:37:59 +03:00
rtw_access_string(sc->sc_access),
rtw_access_string(access)));
sc->sc_access = access;
}
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
/*
* Enable registers, switch register banks.
*/
void
rtw_config0123_enable(struct rtw_regs *regs, int enable)
{
u_int8_t ecr;
ecr = RTW_READ8(regs, RTW_9346CR);
ecr &= ~(RTW_9346CR_EEM_MASK | RTW_9346CR_EECS | RTW_9346CR_EESK);
if (enable)
ecr |= RTW_9346CR_EEM_CONFIG;
Add some bus barriers. Remark where they are needed, e.g., read_eeprom. Change some printfs to RTW_DPRINTF2s to quiet rtw(4) when debugging is disabled.
2004-12-20 04:13:45 +03:00
else {
RTW_WBW(regs, RTW_9346CR, MAX(RTW_CONFIG0, RTW_CONFIG3));
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
ecr |= RTW_9346CR_EEM_NORMAL;
Add some bus barriers. Remark where they are needed, e.g., read_eeprom. Change some printfs to RTW_DPRINTF2s to quiet rtw(4) when debugging is disabled.
2004-12-20 04:13:45 +03:00
}
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
RTW_WRITE8(regs, RTW_9346CR, ecr);
RTW_SYNC(regs, RTW_9346CR, RTW_9346CR);
}
/* requires rtw_config0123_enable(, 1) */
void
rtw_anaparm_enable(struct rtw_regs *regs, int enable)
{
u_int8_t cfg3;
cfg3 = RTW_READ8(regs, RTW_CONFIG3);
Miscellaneous changes. Details below. Important changes flagged with []. Using the driver with my Linksys WPC11 ver. 4, it seems to be receiving packets for a change. The WPC11 ver. 4 has a Maxim RF section. My no-name rtw with Philips RF section still does not receive any packets. Keep access-level (analog params > config[0123] registers > none) in sc_access. Add rtw_set_access for changing the access level. Make rtw_continuous_tx_enable and other subroutines use rtw_set_access instead of rtw_config0123_enable and rtw_anaparm_enable. Factor part of the chip-reset code into rtw_chip_reset1. Change the 'struct foo (*bar)[N]'-style arguments to 'struct foo *bar'-style arguments. Consolidate software/hardware Tx/Rx ring setup in rtw_hwring_setup, rtw_swring_setup. Add a new constant, SA2400_OPMODE_DEFAULTS, for the bits that we *always* set in the SA2400 OPMODE register. Factor some code out into rtw_sa2400_calibrate. (Inspired by the Linux driver.) [] When the receiver goes into underrun/overflow state, call a new subroutine, rtw_kick() that stops the Rx/Tx processes, resets the chip, reinitializes the Tx/Rx rings, and restarts Rx/Tx processes. (Inspired by the Linux driver.) [] In rtw_intr_rx, check for too-short packets before calling ieee80211_find_rxnode. I believe this will prevent a repeat of the MCHK exception I saw once on macppc. [] Use seconds-elapased as well as microseconds-elapsed to set the next "due date" for the timeout interrupt. This keeps the driver from programming the timeout to expire too early. [] In rtw_intr, read RTW_ISR at most 10 times, then get out. If the interface is not enabled (RTW_F_ENABLED), then get out. [] In rtw_stop, get out if the interface is not enabled (RTW_F_ENABLED). Block IPL_NET interrupts. Don't read/write any registers if the interface is invalid (RTW_F_INVALID). [] Call rtw_stop in rtw_detach.
2004-12-12 09:37:59 +03:00
cfg3 |= RTW_CONFIG3_CLKRUNEN;
if (enable)
cfg3 |= RTW_CONFIG3_PARMEN;
else
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
cfg3 &= ~RTW_CONFIG3_PARMEN;
RTW_WRITE8(regs, RTW_CONFIG3, cfg3);
RTW_SYNC(regs, RTW_CONFIG3, RTW_CONFIG3);
}
/* requires rtw_anaparm_enable(, 1) */
void
At last, I have rtw w/ Philips RF receiving packets. I added some sysctls to aid debugging: * hw.rtw.debug -- enable debugging * hw.rtw.flush_rfio -- Linux voodoo: possibly makes the MAC "flush" bits down the serial bus to the RF * hw.rtw.host_rfio: force the host to bang bits to the RF, instead of the MAC banging bits * hw.rtw.rfio_delay: after telling the MAC to bang bits to the RF front-end, delay rfio_delay microseconds. * hw.rtw.rfprog_fallback: there is this notion of the "RF programming method." I believe the choice influences the polarity/timing of the serial bus used to program the RF front-end. I know the correct choice for Intersil/RFMD/Philips front-ends, only. For all other front-ends, I "fallback" to rfprog_fallback. Make rtw_txdac_enable take an rtw_softc argument. I will probably revert this change. Add some Linux voodoo to rtw_continuous_tx_enable. I will probably revert this change. Important: add rtw_set_rfprog, which sets the correct RF programming method. This change and the following change are probably responsible for making the Philips RF work. Important: RTW_CONFIG1 is an 8-bit register, treat it that way! Important: RTW_BRSR is 16-bit, RTW_CRCOUNT, RTW_PHYDELAY, and RTW_MSR are 8-bit: treat them that way! Vastly simplify rtw_resume_ticks. Note to self: set the LED state to match the power state. Hedge against the possibility that RTW_MSR is protected as RTW_CONFIG[0123] are, meanwhile reworking that section of rtw_init a little. Add sc_anaparm, which isn't used, yet....
2004-12-13 03:48:02 +03:00
rtw_txdac_enable(struct rtw_softc *sc, int enable)
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
{
u_int32_t anaparm;
At last, I have rtw w/ Philips RF receiving packets. I added some sysctls to aid debugging: * hw.rtw.debug -- enable debugging * hw.rtw.flush_rfio -- Linux voodoo: possibly makes the MAC "flush" bits down the serial bus to the RF * hw.rtw.host_rfio: force the host to bang bits to the RF, instead of the MAC banging bits * hw.rtw.rfio_delay: after telling the MAC to bang bits to the RF front-end, delay rfio_delay microseconds. * hw.rtw.rfprog_fallback: there is this notion of the "RF programming method." I believe the choice influences the polarity/timing of the serial bus used to program the RF front-end. I know the correct choice for Intersil/RFMD/Philips front-ends, only. For all other front-ends, I "fallback" to rfprog_fallback. Make rtw_txdac_enable take an rtw_softc argument. I will probably revert this change. Add some Linux voodoo to rtw_continuous_tx_enable. I will probably revert this change. Important: add rtw_set_rfprog, which sets the correct RF programming method. This change and the following change are probably responsible for making the Philips RF work. Important: RTW_CONFIG1 is an 8-bit register, treat it that way! Important: RTW_BRSR is 16-bit, RTW_CRCOUNT, RTW_PHYDELAY, and RTW_MSR are 8-bit: treat them that way! Vastly simplify rtw_resume_ticks. Note to self: set the LED state to match the power state. Hedge against the possibility that RTW_MSR is protected as RTW_CONFIG[0123] are, meanwhile reworking that section of rtw_init a little. Add sc_anaparm, which isn't used, yet....
2004-12-13 03:48:02 +03:00
struct rtw_regs *regs = &sc->sc_regs;
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
anaparm = RTW_READ(regs, RTW_ANAPARM);
if (enable)
anaparm &= ~RTW_ANAPARM_TXDACOFF;
else
anaparm |= RTW_ANAPARM_TXDACOFF;
RTW_WRITE(regs, RTW_ANAPARM, anaparm);
RTW_SYNC(regs, RTW_ANAPARM, RTW_ANAPARM);
}
static __inline int
Get rid of the silly char (*dvname)[IFNAMSIZ] pattern. Replace it with const char *dvname. I'm not sure what I was thinking.
2004-12-20 03:28:02 +03:00
rtw_chip_reset1(struct rtw_regs *regs, const char *dvname)
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
{
u_int8_t cr;
Miscellaneous changes. Details below. Important changes flagged with []. Using the driver with my Linksys WPC11 ver. 4, it seems to be receiving packets for a change. The WPC11 ver. 4 has a Maxim RF section. My no-name rtw with Philips RF section still does not receive any packets. Keep access-level (analog params > config[0123] registers > none) in sc_access. Add rtw_set_access for changing the access level. Make rtw_continuous_tx_enable and other subroutines use rtw_set_access instead of rtw_config0123_enable and rtw_anaparm_enable. Factor part of the chip-reset code into rtw_chip_reset1. Change the 'struct foo (*bar)[N]'-style arguments to 'struct foo *bar'-style arguments. Consolidate software/hardware Tx/Rx ring setup in rtw_hwring_setup, rtw_swring_setup. Add a new constant, SA2400_OPMODE_DEFAULTS, for the bits that we *always* set in the SA2400 OPMODE register. Factor some code out into rtw_sa2400_calibrate. (Inspired by the Linux driver.) [] When the receiver goes into underrun/overflow state, call a new subroutine, rtw_kick() that stops the Rx/Tx processes, resets the chip, reinitializes the Tx/Rx rings, and restarts Rx/Tx processes. (Inspired by the Linux driver.) [] In rtw_intr_rx, check for too-short packets before calling ieee80211_find_rxnode. I believe this will prevent a repeat of the MCHK exception I saw once on macppc. [] Use seconds-elapased as well as microseconds-elapsed to set the next "due date" for the timeout interrupt. This keeps the driver from programming the timeout to expire too early. [] In rtw_intr, read RTW_ISR at most 10 times, then get out. If the interface is not enabled (RTW_F_ENABLED), then get out. [] In rtw_stop, get out if the interface is not enabled (RTW_F_ENABLED). Block IPL_NET interrupts. Don't read/write any registers if the interface is invalid (RTW_F_INVALID). [] Call rtw_stop in rtw_detach.
2004-12-12 09:37:59 +03:00
int i;
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
RTW_WRITE8(regs, RTW_CR, RTW_CR_RST);
RTW_WBR(regs, RTW_CR, RTW_CR);
Change rtw_debug from a debug level to a debug mask. Add a lot of debug flags. From Linux: handle an RTL8180 bug. Sometimes the NIC skips from the middle of the ring to the 0th rx descriptor. Now the driver resynchronizes. Handle a receive descriptor underrun or Rx FIFO overflow condition in the way that the Linux driver does. This kind of seems like overkill, but whatever. Protect rtw_ioctl with splnet(). Do not load a tx descriptor with a buffer shorter than 4 bytes. Handle a transmit timeout less disruptively.
2004-12-25 09:58:37 +03:00
for (i = 0; i < 1000; i++) {
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
if ((cr = RTW_READ8(regs, RTW_CR) & RTW_CR_RST) == 0) {
Change rtw_debug from a debug level to a debug mask. Add a lot of debug flags. From Linux: handle an RTL8180 bug. Sometimes the NIC skips from the middle of the ring to the 0th rx descriptor. Now the driver resynchronizes. Handle a receive descriptor underrun or Rx FIFO overflow condition in the way that the Linux driver does. This kind of seems like overkill, but whatever. Protect rtw_ioctl with splnet(). Do not load a tx descriptor with a buffer shorter than 4 bytes. Handle a transmit timeout less disruptively.
2004-12-25 09:58:37 +03:00
RTW_DPRINTF(RTW_DEBUG_RESET,
("%s: reset in %dus\n", dvname, i));
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
return 0;
}
RTW_RBR(regs, RTW_CR, RTW_CR);
Change rtw_debug from a debug level to a debug mask. Add a lot of debug flags. From Linux: handle an RTL8180 bug. Sometimes the NIC skips from the middle of the ring to the 0th rx descriptor. Now the driver resynchronizes. Handle a receive descriptor underrun or Rx FIFO overflow condition in the way that the Linux driver does. This kind of seems like overkill, but whatever. Protect rtw_ioctl with splnet(). Do not load a tx descriptor with a buffer shorter than 4 bytes. Handle a transmit timeout less disruptively.
2004-12-25 09:58:37 +03:00
DELAY(10); /* 10us */
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
}
Get rid of the silly char (*dvname)[IFNAMSIZ] pattern. Replace it with const char *dvname. I'm not sure what I was thinking.
2004-12-20 03:28:02 +03:00
printf("%s: reset failed\n", dvname);
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
return ETIMEDOUT;
}
Miscellaneous changes. Details below. Important changes flagged with []. Using the driver with my Linksys WPC11 ver. 4, it seems to be receiving packets for a change. The WPC11 ver. 4 has a Maxim RF section. My no-name rtw with Philips RF section still does not receive any packets. Keep access-level (analog params > config[0123] registers > none) in sc_access. Add rtw_set_access for changing the access level. Make rtw_continuous_tx_enable and other subroutines use rtw_set_access instead of rtw_config0123_enable and rtw_anaparm_enable. Factor part of the chip-reset code into rtw_chip_reset1. Change the 'struct foo (*bar)[N]'-style arguments to 'struct foo *bar'-style arguments. Consolidate software/hardware Tx/Rx ring setup in rtw_hwring_setup, rtw_swring_setup. Add a new constant, SA2400_OPMODE_DEFAULTS, for the bits that we *always* set in the SA2400 OPMODE register. Factor some code out into rtw_sa2400_calibrate. (Inspired by the Linux driver.) [] When the receiver goes into underrun/overflow state, call a new subroutine, rtw_kick() that stops the Rx/Tx processes, resets the chip, reinitializes the Tx/Rx rings, and restarts Rx/Tx processes. (Inspired by the Linux driver.) [] In rtw_intr_rx, check for too-short packets before calling ieee80211_find_rxnode. I believe this will prevent a repeat of the MCHK exception I saw once on macppc. [] Use seconds-elapased as well as microseconds-elapsed to set the next "due date" for the timeout interrupt. This keeps the driver from programming the timeout to expire too early. [] In rtw_intr, read RTW_ISR at most 10 times, then get out. If the interface is not enabled (RTW_F_ENABLED), then get out. [] In rtw_stop, get out if the interface is not enabled (RTW_F_ENABLED). Block IPL_NET interrupts. Don't read/write any registers if the interface is invalid (RTW_F_INVALID). [] Call rtw_stop in rtw_detach.
2004-12-12 09:37:59 +03:00
static __inline int
Get rid of the silly char (*dvname)[IFNAMSIZ] pattern. Replace it with const char *dvname. I'm not sure what I was thinking.
2004-12-20 03:28:02 +03:00
rtw_chip_reset(struct rtw_regs *regs, const char *dvname)
Miscellaneous changes. Details below. Important changes flagged with []. Using the driver with my Linksys WPC11 ver. 4, it seems to be receiving packets for a change. The WPC11 ver. 4 has a Maxim RF section. My no-name rtw with Philips RF section still does not receive any packets. Keep access-level (analog params > config[0123] registers > none) in sc_access. Add rtw_set_access for changing the access level. Make rtw_continuous_tx_enable and other subroutines use rtw_set_access instead of rtw_config0123_enable and rtw_anaparm_enable. Factor part of the chip-reset code into rtw_chip_reset1. Change the 'struct foo (*bar)[N]'-style arguments to 'struct foo *bar'-style arguments. Consolidate software/hardware Tx/Rx ring setup in rtw_hwring_setup, rtw_swring_setup. Add a new constant, SA2400_OPMODE_DEFAULTS, for the bits that we *always* set in the SA2400 OPMODE register. Factor some code out into rtw_sa2400_calibrate. (Inspired by the Linux driver.) [] When the receiver goes into underrun/overflow state, call a new subroutine, rtw_kick() that stops the Rx/Tx processes, resets the chip, reinitializes the Tx/Rx rings, and restarts Rx/Tx processes. (Inspired by the Linux driver.) [] In rtw_intr_rx, check for too-short packets before calling ieee80211_find_rxnode. I believe this will prevent a repeat of the MCHK exception I saw once on macppc. [] Use seconds-elapased as well as microseconds-elapsed to set the next "due date" for the timeout interrupt. This keeps the driver from programming the timeout to expire too early. [] In rtw_intr, read RTW_ISR at most 10 times, then get out. If the interface is not enabled (RTW_F_ENABLED), then get out. [] In rtw_stop, get out if the interface is not enabled (RTW_F_ENABLED). Block IPL_NET interrupts. Don't read/write any registers if the interface is invalid (RTW_F_INVALID). [] Call rtw_stop in rtw_detach.
2004-12-12 09:37:59 +03:00
{
uint32_t tcr;
/* from Linux driver */
tcr = RTW_TCR_CWMIN | RTW_TCR_MXDMA_2048 |
LSHIFT(7, RTW_TCR_SRL_MASK) | LSHIFT(7, RTW_TCR_LRL_MASK);
RTW_WRITE(regs, RTW_TCR, tcr);
RTW_WBW(regs, RTW_CR, RTW_TCR);
return rtw_chip_reset1(regs, dvname);
}
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
static __inline int
Get rid of the silly char (*dvname)[IFNAMSIZ] pattern. Replace it with const char *dvname. I'm not sure what I was thinking.
2004-12-20 03:28:02 +03:00
rtw_recall_eeprom(struct rtw_regs *regs, const char *dvname)
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
{
int i;
u_int8_t ecr;
ecr = RTW_READ8(regs, RTW_9346CR);
ecr = (ecr & ~RTW_9346CR_EEM_MASK) | RTW_9346CR_EEM_AUTOLOAD;
RTW_WRITE8(regs, RTW_9346CR, ecr);
RTW_WBR(regs, RTW_9346CR, RTW_9346CR);
/* wait 2.5ms for completion */
for (i = 0; i < 25; i++) {
ecr = RTW_READ8(regs, RTW_9346CR);
if ((ecr & RTW_9346CR_EEM_MASK) == RTW_9346CR_EEM_NORMAL) {
Change rtw_debug from a debug level to a debug mask. Add a lot of debug flags. From Linux: handle an RTL8180 bug. Sometimes the NIC skips from the middle of the ring to the 0th rx descriptor. Now the driver resynchronizes. Handle a receive descriptor underrun or Rx FIFO overflow condition in the way that the Linux driver does. This kind of seems like overkill, but whatever. Protect rtw_ioctl with splnet(). Do not load a tx descriptor with a buffer shorter than 4 bytes. Handle a transmit timeout less disruptively.
2004-12-25 09:58:37 +03:00
RTW_DPRINTF(RTW_DEBUG_RESET,
("%s: recall EEPROM in %dus\n", dvname, i * 100));
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
return 0;
}
RTW_RBR(regs, RTW_9346CR, RTW_9346CR);
DELAY(100);
}
Get rid of the silly char (*dvname)[IFNAMSIZ] pattern. Replace it with const char *dvname. I'm not sure what I was thinking.
2004-12-20 03:28:02 +03:00
printf("%s: recall EEPROM failed\n", dvname);
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
return ETIMEDOUT;
}
static __inline int
rtw_reset(struct rtw_softc *sc)
{
int rc;
At last, I have rtw w/ Philips RF receiving packets. I added some sysctls to aid debugging: * hw.rtw.debug -- enable debugging * hw.rtw.flush_rfio -- Linux voodoo: possibly makes the MAC "flush" bits down the serial bus to the RF * hw.rtw.host_rfio: force the host to bang bits to the RF, instead of the MAC banging bits * hw.rtw.rfio_delay: after telling the MAC to bang bits to the RF front-end, delay rfio_delay microseconds. * hw.rtw.rfprog_fallback: there is this notion of the "RF programming method." I believe the choice influences the polarity/timing of the serial bus used to program the RF front-end. I know the correct choice for Intersil/RFMD/Philips front-ends, only. For all other front-ends, I "fallback" to rfprog_fallback. Make rtw_txdac_enable take an rtw_softc argument. I will probably revert this change. Add some Linux voodoo to rtw_continuous_tx_enable. I will probably revert this change. Important: add rtw_set_rfprog, which sets the correct RF programming method. This change and the following change are probably responsible for making the Philips RF work. Important: RTW_CONFIG1 is an 8-bit register, treat it that way! Important: RTW_BRSR is 16-bit, RTW_CRCOUNT, RTW_PHYDELAY, and RTW_MSR are 8-bit: treat them that way! Vastly simplify rtw_resume_ticks. Note to self: set the LED state to match the power state. Hedge against the possibility that RTW_MSR is protected as RTW_CONFIG[0123] are, meanwhile reworking that section of rtw_init a little. Add sc_anaparm, which isn't used, yet....
2004-12-13 03:48:02 +03:00
uint8_t config1;
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
Get rid of the silly char (*dvname)[IFNAMSIZ] pattern. Replace it with const char *dvname. I'm not sure what I was thinking.
2004-12-20 03:28:02 +03:00
if ((rc = rtw_chip_reset(&sc->sc_regs, sc->sc_dev.dv_xname)) != 0)
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
return rc;
Get rid of the silly char (*dvname)[IFNAMSIZ] pattern. Replace it with const char *dvname. I'm not sure what I was thinking.
2004-12-20 03:28:02 +03:00
if ((rc = rtw_recall_eeprom(&sc->sc_regs, sc->sc_dev.dv_xname)) != 0)
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
;
At last, I have rtw w/ Philips RF receiving packets. I added some sysctls to aid debugging: * hw.rtw.debug -- enable debugging * hw.rtw.flush_rfio -- Linux voodoo: possibly makes the MAC "flush" bits down the serial bus to the RF * hw.rtw.host_rfio: force the host to bang bits to the RF, instead of the MAC banging bits * hw.rtw.rfio_delay: after telling the MAC to bang bits to the RF front-end, delay rfio_delay microseconds. * hw.rtw.rfprog_fallback: there is this notion of the "RF programming method." I believe the choice influences the polarity/timing of the serial bus used to program the RF front-end. I know the correct choice for Intersil/RFMD/Philips front-ends, only. For all other front-ends, I "fallback" to rfprog_fallback. Make rtw_txdac_enable take an rtw_softc argument. I will probably revert this change. Add some Linux voodoo to rtw_continuous_tx_enable. I will probably revert this change. Important: add rtw_set_rfprog, which sets the correct RF programming method. This change and the following change are probably responsible for making the Philips RF work. Important: RTW_CONFIG1 is an 8-bit register, treat it that way! Important: RTW_BRSR is 16-bit, RTW_CRCOUNT, RTW_PHYDELAY, and RTW_MSR are 8-bit: treat them that way! Vastly simplify rtw_resume_ticks. Note to self: set the LED state to match the power state. Hedge against the possibility that RTW_MSR is protected as RTW_CONFIG[0123] are, meanwhile reworking that section of rtw_init a little. Add sc_anaparm, which isn't used, yet....
2004-12-13 03:48:02 +03:00
config1 = RTW_READ8(&sc->sc_regs, RTW_CONFIG1);
RTW_WRITE8(&sc->sc_regs, RTW_CONFIG1, config1 & ~RTW_CONFIG1_PMEN);
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
/* TBD turn off maximum power saving? */
return 0;
}
static __inline int
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
rtw_txdesc_dmamaps_create(bus_dma_tag_t dmat, struct rtw_txsoft *descs,
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
u_int ndescs)
{
int i, rc = 0;
for (i = 0; i < ndescs; i++) {
rc = bus_dmamap_create(dmat, MCLBYTES, RTW_MAXPKTSEGS, MCLBYTES,
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
0, 0, &descs[i].ts_dmamap);
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
if (rc != 0)
break;
}
return rc;
}
static __inline int
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
rtw_rxdesc_dmamaps_create(bus_dma_tag_t dmat, struct rtw_rxsoft *descs,
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
u_int ndescs)
{
int i, rc = 0;
for (i = 0; i < ndescs; i++) {
rc = bus_dmamap_create(dmat, MCLBYTES, 1, MCLBYTES, 0, 0,
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
&descs[i].rs_dmamap);
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
if (rc != 0)
break;
}
return rc;
}
static __inline void
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
rtw_rxdesc_dmamaps_destroy(bus_dma_tag_t dmat, struct rtw_rxsoft *descs,
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
u_int ndescs)
{
int i;
for (i = 0; i < ndescs; i++) {
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
if (descs[i].rs_dmamap != NULL)
bus_dmamap_destroy(dmat, descs[i].rs_dmamap);
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
}
}
static __inline void
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
rtw_txdesc_dmamaps_destroy(bus_dma_tag_t dmat, struct rtw_txsoft *descs,
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
u_int ndescs)
{
int i;
for (i = 0; i < ndescs; i++) {
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
if (descs[i].ts_dmamap != NULL)
bus_dmamap_destroy(dmat, descs[i].ts_dmamap);
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
}
}
static __inline void
rtw_srom_free(struct rtw_srom *sr)
{
sr->sr_size = 0;
if (sr->sr_content == NULL)
return;
free(sr->sr_content, M_DEVBUF);
sr->sr_content = NULL;
}
static void
rtw_srom_defaults(struct rtw_srom *sr, u_int32_t *flags, u_int8_t *cs_threshold,
Get rid of the silly char (*dvname)[IFNAMSIZ] pattern. Replace it with const char *dvname. I'm not sure what I was thinking.
2004-12-20 03:28:02 +03:00
enum rtw_rfchipid *rfchipid, u_int32_t *rcr)
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
{
*flags |= (RTW_F_DIGPHY|RTW_F_ANTDIV);
*cs_threshold = RTW_SR_ENERGYDETTHR_DEFAULT;
*rcr |= RTW_RCR_ENCS1;
*rfchipid = RTW_RFCHIPID_PHILIPS;
}
static int
rtw_srom_parse(struct rtw_srom *sr, u_int32_t *flags, u_int8_t *cs_threshold,
enum rtw_rfchipid *rfchipid, u_int32_t *rcr, enum rtw_locale *locale,
Get rid of the silly char (*dvname)[IFNAMSIZ] pattern. Replace it with const char *dvname. I'm not sure what I was thinking.
2004-12-20 03:28:02 +03:00
const char *dvname)
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
{
int i;
const char *rfname, *paname;
char scratch[sizeof("unknown 0xXX")];
u_int16_t version;
u_int8_t mac[IEEE80211_ADDR_LEN];
*flags &= ~(RTW_F_DIGPHY|RTW_F_DFLANTB|RTW_F_ANTDIV);
*rcr &= ~(RTW_RCR_ENCS1 | RTW_RCR_ENCS2);
version = RTW_SR_GET16(sr, RTW_SR_VERSION);
Get rid of the silly char (*dvname)[IFNAMSIZ] pattern. Replace it with const char *dvname. I'm not sure what I was thinking.
2004-12-20 03:28:02 +03:00
printf("%s: SROM version %d.%d", dvname, version >> 8, version & 0xff);
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
if (version <= 0x0101) {
printf(" is not understood, limping along with defaults\n");
Get rid of the silly char (*dvname)[IFNAMSIZ] pattern. Replace it with const char *dvname. I'm not sure what I was thinking.
2004-12-20 03:28:02 +03:00
rtw_srom_defaults(sr, flags, cs_threshold, rfchipid, rcr);
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
return 0;
}
printf("\n");
for (i = 0; i < IEEE80211_ADDR_LEN; i++)
mac[i] = RTW_SR_GET(sr, RTW_SR_MAC + i);
Change rtw_debug from a debug level to a debug mask. Add a lot of debug flags. From Linux: handle an RTL8180 bug. Sometimes the NIC skips from the middle of the ring to the 0th rx descriptor. Now the driver resynchronizes. Handle a receive descriptor underrun or Rx FIFO overflow condition in the way that the Linux driver does. This kind of seems like overkill, but whatever. Protect rtw_ioctl with splnet(). Do not load a tx descriptor with a buffer shorter than 4 bytes. Handle a transmit timeout less disruptively.
2004-12-25 09:58:37 +03:00
RTW_DPRINTF(RTW_DEBUG_ATTACH,
("%s: EEPROM MAC %s\n", dvname, ether_sprintf(mac)));
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
*cs_threshold = RTW_SR_GET(sr, RTW_SR_ENERGYDETTHR);
if ((RTW_SR_GET(sr, RTW_SR_CONFIG2) & RTW_CONFIG2_ANT) != 0)
*flags |= RTW_F_ANTDIV;
Realtek sample sources are not consistent about the "Digital PHY" bit in the SROM. It seems as if it is set to 1 when the PHY is *analog*, not *digital*. Fix my sources. In rtw_intr_rx, use units of 500kb/s instead of 100kb/s for rate, to be consistent with net80211's expectations. Polish up some debugging ugly messages. Dump raw 802.11 packets if IFF_DEBUG|IFF_LINK2 and RTW_DEBUG is defined. Polish power-state (on/sleep/off) handling. Especially improve support for RFMD (totally untested) and Maxim. For Philips, take the Digital PHY property into account. Call the net80211 watchdog function from rtw_watchdog, so that we scan again if auth/assoc fails. Be a little more cautious about writing register[RTW_TPPOLL], since other drivers are.... Don't frob the high/low-priority queues right now, since I don't use them. Add rtw_join_bss which programs the card with the BSSID and other properties of a BSS. Use it on state transitions. Factor out rtw_set_nettype. Make rtw_recv_beacon call ieee80211_recv_mgmt instead of dropping beacons on the floor! TBD IBSS merges. Change some rtw_debug=2 printfs to rtw_debug=3 (RTW_DPRINTF3) printfs so the console doesn't get spammed so badly at rtw_debug=2. Change some debugging printfs to RTW_DPRINTFs. E.g., print the "RF programming method" only if debugging is enabled.
2004-12-21 02:05:41 +03:00
/* Note well: the sense of the RTW_SR_RFPARM_DIGPHY bit seems
* to be reversed.
*/
if ((RTW_SR_GET(sr, RTW_SR_RFPARM) & RTW_SR_RFPARM_DIGPHY) == 0)
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
*flags |= RTW_F_DIGPHY;
if ((RTW_SR_GET(sr, RTW_SR_RFPARM) & RTW_SR_RFPARM_DFLANTB) != 0)
*flags |= RTW_F_DFLANTB;
*rcr |= LSHIFT(MASK_AND_RSHIFT(RTW_SR_GET(sr, RTW_SR_RFPARM),
RTW_SR_RFPARM_CS_MASK), RTW_RCR_ENCS1);
*rfchipid = RTW_SR_GET(sr, RTW_SR_RFCHIPID);
switch (*rfchipid) {
case RTW_RFCHIPID_GCT: /* this combo seen in the wild */
rfname = "GCT GRF5101";
paname = "Winspring WS9901";
break;
case RTW_RFCHIPID_MAXIM:
rfname = "Maxim MAX2820"; /* guess */
paname = "Maxim MAX2422"; /* guess */
break;
case RTW_RFCHIPID_INTERSIL:
rfname = "Intersil HFA3873"; /* guess */
paname = "Intersil <unknown>";
break;
case RTW_RFCHIPID_PHILIPS: /* this combo seen in the wild */
rfname = "Philips SA2400A";
paname = "Philips SA2411";
break;
case RTW_RFCHIPID_RFMD:
/* this is the same front-end as an atw(4)! */
rfname = "RFMD RF2948B, " /* mentioned in Realtek docs */
"LNA: RFMD RF2494, " /* mentioned in Realtek docs */
"SYN: Silicon Labs Si4126"; /* inferred from
* reference driver
*/
paname = "RFMD RF2189"; /* mentioned in Realtek docs */
break;
case RTW_RFCHIPID_RESERVED:
rfname = paname = "reserved";
break;
default:
snprintf(scratch, sizeof(scratch), "unknown 0x%02x", *rfchipid);
rfname = paname = scratch;
}
Get rid of the silly char (*dvname)[IFNAMSIZ] pattern. Replace it with const char *dvname. I'm not sure what I was thinking.
2004-12-20 03:28:02 +03:00
printf("%s: RF: %s, PA: %s\n", dvname, rfname, paname);
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
switch (RTW_SR_GET(sr, RTW_SR_CONFIG0) & RTW_CONFIG0_GL_MASK) {
case RTW_CONFIG0_GL_USA:
*locale = RTW_LOCALE_USA;
break;
case RTW_CONFIG0_GL_EUROPE:
*locale = RTW_LOCALE_EUROPE;
break;
case RTW_CONFIG0_GL_JAPAN:
*locale = RTW_LOCALE_JAPAN;
break;
default:
*locale = RTW_LOCALE_UNKNOWN;
break;
}
return 0;
}
/* Returns -1 on failure. */
static int
rtw_srom_read(struct rtw_regs *regs, u_int32_t flags, struct rtw_srom *sr,
Get rid of the silly char (*dvname)[IFNAMSIZ] pattern. Replace it with const char *dvname. I'm not sure what I was thinking.
2004-12-20 03:28:02 +03:00
const char *dvname)
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
{
int rc;
struct seeprom_descriptor sd;
u_int8_t ecr;
(void)memset(&sd, 0, sizeof(sd));
ecr = RTW_READ8(regs, RTW_9346CR);
if ((flags & RTW_F_9356SROM) != 0) {
Change rtw_debug from a debug level to a debug mask. Add a lot of debug flags. From Linux: handle an RTL8180 bug. Sometimes the NIC skips from the middle of the ring to the 0th rx descriptor. Now the driver resynchronizes. Handle a receive descriptor underrun or Rx FIFO overflow condition in the way that the Linux driver does. This kind of seems like overkill, but whatever. Protect rtw_ioctl with splnet(). Do not load a tx descriptor with a buffer shorter than 4 bytes. Handle a transmit timeout less disruptively.
2004-12-25 09:58:37 +03:00
RTW_DPRINTF(RTW_DEBUG_ATTACH, ("%s: 93c56 SROM\n", dvname));
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
sr->sr_size = 256;
sd.sd_chip = C56_66;
} else {
Change rtw_debug from a debug level to a debug mask. Add a lot of debug flags. From Linux: handle an RTL8180 bug. Sometimes the NIC skips from the middle of the ring to the 0th rx descriptor. Now the driver resynchronizes. Handle a receive descriptor underrun or Rx FIFO overflow condition in the way that the Linux driver does. This kind of seems like overkill, but whatever. Protect rtw_ioctl with splnet(). Do not load a tx descriptor with a buffer shorter than 4 bytes. Handle a transmit timeout less disruptively.
2004-12-25 09:58:37 +03:00
RTW_DPRINTF(RTW_DEBUG_ATTACH, ("%s: 93c46 SROM\n", dvname));
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
sr->sr_size = 128;
sd.sd_chip = C46;
}
ecr &= ~(RTW_9346CR_EEDI | RTW_9346CR_EEDO | RTW_9346CR_EESK |
RTW_9346CR_EEM_MASK);
ecr |= RTW_9346CR_EEM_PROGRAM;
RTW_WRITE8(regs, RTW_9346CR, ecr);
sr->sr_content = malloc(sr->sr_size, M_DEVBUF, M_NOWAIT);
if (sr->sr_content == NULL) {
Get rid of the silly char (*dvname)[IFNAMSIZ] pattern. Replace it with const char *dvname. I'm not sure what I was thinking.
2004-12-20 03:28:02 +03:00
printf("%s: unable to allocate SROM buffer\n", dvname);
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
return ENOMEM;
}
(void)memset(sr->sr_content, 0, sr->sr_size);
/* RTL8180 has a single 8-bit register for controlling the
* 93cx6 SROM. There is no "ready" bit. The RTL8180
* input/output sense is the reverse of read_seeprom's.
*/
sd.sd_tag = regs->r_bt;
sd.sd_bsh = regs->r_bh;
sd.sd_regsize = 1;
sd.sd_control_offset = RTW_9346CR;
sd.sd_status_offset = RTW_9346CR;
sd.sd_dataout_offset = RTW_9346CR;
sd.sd_CK = RTW_9346CR_EESK;
sd.sd_CS = RTW_9346CR_EECS;
sd.sd_DI = RTW_9346CR_EEDO;
sd.sd_DO = RTW_9346CR_EEDI;
/* make read_seeprom enter EEPROM read/write mode */
sd.sd_MS = ecr;
sd.sd_RDY = 0;
Add some bus barriers. Remark where they are needed, e.g., read_eeprom. Change some printfs to RTW_DPRINTF2s to quiet rtw(4) when debugging is disabled.
2004-12-20 04:13:45 +03:00
/* TBD bus barriers */
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
if (!read_seeprom(&sd, sr->sr_content, 0, sr->sr_size/2)) {
Get rid of the silly char (*dvname)[IFNAMSIZ] pattern. Replace it with const char *dvname. I'm not sure what I was thinking.
2004-12-20 03:28:02 +03:00
printf("%s: could not read SROM\n", dvname);
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
free(sr->sr_content, M_DEVBUF);
sr->sr_content = NULL;
return -1; /* XXX */
}
/* end EEPROM read/write mode */
RTW_WRITE8(regs, RTW_9346CR,
(ecr & ~RTW_9346CR_EEM_MASK) | RTW_9346CR_EEM_NORMAL);
RTW_WBRW(regs, RTW_9346CR, RTW_9346CR);
if ((rc = rtw_recall_eeprom(regs, dvname)) != 0)
return rc;
#ifdef RTW_DEBUG
{
int i;
Change rtw_debug from a debug level to a debug mask. Add a lot of debug flags. From Linux: handle an RTL8180 bug. Sometimes the NIC skips from the middle of the ring to the 0th rx descriptor. Now the driver resynchronizes. Handle a receive descriptor underrun or Rx FIFO overflow condition in the way that the Linux driver does. This kind of seems like overkill, but whatever. Protect rtw_ioctl with splnet(). Do not load a tx descriptor with a buffer shorter than 4 bytes. Handle a transmit timeout less disruptively.
2004-12-25 09:58:37 +03:00
RTW_DPRINTF(RTW_DEBUG_ATTACH,
("\n%s: serial ROM:\n\t", dvname));
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
for (i = 0; i < sr->sr_size/2; i++) {
if (((i % 8) == 0) && (i != 0))
Change rtw_debug from a debug level to a debug mask. Add a lot of debug flags. From Linux: handle an RTL8180 bug. Sometimes the NIC skips from the middle of the ring to the 0th rx descriptor. Now the driver resynchronizes. Handle a receive descriptor underrun or Rx FIFO overflow condition in the way that the Linux driver does. This kind of seems like overkill, but whatever. Protect rtw_ioctl with splnet(). Do not load a tx descriptor with a buffer shorter than 4 bytes. Handle a transmit timeout less disruptively.
2004-12-25 09:58:37 +03:00
RTW_DPRINTF(RTW_DEBUG_ATTACH, ("\n\t"));
RTW_DPRINTF(RTW_DEBUG_ATTACH,
(" %04x", sr->sr_content[i]));
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
}
Change rtw_debug from a debug level to a debug mask. Add a lot of debug flags. From Linux: handle an RTL8180 bug. Sometimes the NIC skips from the middle of the ring to the 0th rx descriptor. Now the driver resynchronizes. Handle a receive descriptor underrun or Rx FIFO overflow condition in the way that the Linux driver does. This kind of seems like overkill, but whatever. Protect rtw_ioctl with splnet(). Do not load a tx descriptor with a buffer shorter than 4 bytes. Handle a transmit timeout less disruptively.
2004-12-25 09:58:37 +03:00
RTW_DPRINTF(RTW_DEBUG_ATTACH, ("\n"));
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
}
#endif /* RTW_DEBUG */
return 0;
}
At last, I have rtw w/ Philips RF receiving packets. I added some sysctls to aid debugging: * hw.rtw.debug -- enable debugging * hw.rtw.flush_rfio -- Linux voodoo: possibly makes the MAC "flush" bits down the serial bus to the RF * hw.rtw.host_rfio: force the host to bang bits to the RF, instead of the MAC banging bits * hw.rtw.rfio_delay: after telling the MAC to bang bits to the RF front-end, delay rfio_delay microseconds. * hw.rtw.rfprog_fallback: there is this notion of the "RF programming method." I believe the choice influences the polarity/timing of the serial bus used to program the RF front-end. I know the correct choice for Intersil/RFMD/Philips front-ends, only. For all other front-ends, I "fallback" to rfprog_fallback. Make rtw_txdac_enable take an rtw_softc argument. I will probably revert this change. Add some Linux voodoo to rtw_continuous_tx_enable. I will probably revert this change. Important: add rtw_set_rfprog, which sets the correct RF programming method. This change and the following change are probably responsible for making the Philips RF work. Important: RTW_CONFIG1 is an 8-bit register, treat it that way! Important: RTW_BRSR is 16-bit, RTW_CRCOUNT, RTW_PHYDELAY, and RTW_MSR are 8-bit: treat them that way! Vastly simplify rtw_resume_ticks. Note to self: set the LED state to match the power state. Hedge against the possibility that RTW_MSR is protected as RTW_CONFIG[0123] are, meanwhile reworking that section of rtw_init a little. Add sc_anaparm, which isn't used, yet....
2004-12-13 03:48:02 +03:00
static void
rtw_set_rfprog(struct rtw_regs *regs, enum rtw_rfchipid rfchipid,
const char *dvname)
{
u_int8_t cfg4;
const char *method;
cfg4 = RTW_READ8(regs, RTW_CONFIG4) & ~RTW_CONFIG4_RFTYPE_MASK;
switch (rfchipid) {
default:
cfg4 |= LSHIFT(rtw_rfprog_fallback, RTW_CONFIG4_RFTYPE_MASK);
method = "fallback";
break;
case RTW_RFCHIPID_INTERSIL:
cfg4 |= RTW_CONFIG4_RFTYPE_INTERSIL;
method = "Intersil";
break;
case RTW_RFCHIPID_PHILIPS:
cfg4 |= RTW_CONFIG4_RFTYPE_PHILIPS;
method = "Philips";
break;
case RTW_RFCHIPID_RFMD:
cfg4 |= RTW_CONFIG4_RFTYPE_RFMD;
method = "RFMD";
break;
}
RTW_WRITE8(regs, RTW_CONFIG4, cfg4);
Add some bus barriers. Remark where they are needed, e.g., read_eeprom. Change some printfs to RTW_DPRINTF2s to quiet rtw(4) when debugging is disabled.
2004-12-20 04:13:45 +03:00
RTW_WBR(regs, RTW_CONFIG4, RTW_CONFIG4);
Change rtw_debug from a debug level to a debug mask. Add a lot of debug flags. From Linux: handle an RTL8180 bug. Sometimes the NIC skips from the middle of the ring to the 0th rx descriptor. Now the driver resynchronizes. Handle a receive descriptor underrun or Rx FIFO overflow condition in the way that the Linux driver does. This kind of seems like overkill, but whatever. Protect rtw_ioctl with splnet(). Do not load a tx descriptor with a buffer shorter than 4 bytes. Handle a transmit timeout less disruptively.
2004-12-25 09:58:37 +03:00
RTW_DPRINTF(RTW_DEBUG_INIT,
("%s: %s RF programming method, %#02x\n", dvname, method,
Realtek sample sources are not consistent about the "Digital PHY" bit in the SROM. It seems as if it is set to 1 when the PHY is *analog*, not *digital*. Fix my sources. In rtw_intr_rx, use units of 500kb/s instead of 100kb/s for rate, to be consistent with net80211's expectations. Polish up some debugging ugly messages. Dump raw 802.11 packets if IFF_DEBUG|IFF_LINK2 and RTW_DEBUG is defined. Polish power-state (on/sleep/off) handling. Especially improve support for RFMD (totally untested) and Maxim. For Philips, take the Digital PHY property into account. Call the net80211 watchdog function from rtw_watchdog, so that we scan again if auth/assoc fails. Be a little more cautious about writing register[RTW_TPPOLL], since other drivers are.... Don't frob the high/low-priority queues right now, since I don't use them. Add rtw_join_bss which programs the card with the BSSID and other properties of a BSS. Use it on state transitions. Factor out rtw_set_nettype. Make rtw_recv_beacon call ieee80211_recv_mgmt instead of dropping beacons on the floor! TBD IBSS merges. Change some rtw_debug=2 printfs to rtw_debug=3 (RTW_DPRINTF3) printfs so the console doesn't get spammed so badly at rtw_debug=2. Change some debugging printfs to RTW_DPRINTFs. E.g., print the "RF programming method" only if debugging is enabled.
2004-12-21 02:05:41 +03:00
RTW_READ8(regs, RTW_CONFIG4)));
At last, I have rtw w/ Philips RF receiving packets. I added some sysctls to aid debugging: * hw.rtw.debug -- enable debugging * hw.rtw.flush_rfio -- Linux voodoo: possibly makes the MAC "flush" bits down the serial bus to the RF * hw.rtw.host_rfio: force the host to bang bits to the RF, instead of the MAC banging bits * hw.rtw.rfio_delay: after telling the MAC to bang bits to the RF front-end, delay rfio_delay microseconds. * hw.rtw.rfprog_fallback: there is this notion of the "RF programming method." I believe the choice influences the polarity/timing of the serial bus used to program the RF front-end. I know the correct choice for Intersil/RFMD/Philips front-ends, only. For all other front-ends, I "fallback" to rfprog_fallback. Make rtw_txdac_enable take an rtw_softc argument. I will probably revert this change. Add some Linux voodoo to rtw_continuous_tx_enable. I will probably revert this change. Important: add rtw_set_rfprog, which sets the correct RF programming method. This change and the following change are probably responsible for making the Philips RF work. Important: RTW_CONFIG1 is an 8-bit register, treat it that way! Important: RTW_BRSR is 16-bit, RTW_CRCOUNT, RTW_PHYDELAY, and RTW_MSR are 8-bit: treat them that way! Vastly simplify rtw_resume_ticks. Note to self: set the LED state to match the power state. Hedge against the possibility that RTW_MSR is protected as RTW_CONFIG[0123] are, meanwhile reworking that section of rtw_init a little. Add sc_anaparm, which isn't used, yet....
2004-12-13 03:48:02 +03:00
}
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
static __inline void
rtw_init_channels(enum rtw_locale locale,
struct ieee80211_channel (*chans)[IEEE80211_CHAN_MAX+1],
Get rid of the silly char (*dvname)[IFNAMSIZ] pattern. Replace it with const char *dvname. I'm not sure what I was thinking.
2004-12-20 03:28:02 +03:00
const char *dvname)
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
{
int i;
const char *name = NULL;
#define ADD_CHANNEL(_chans, _chan) do { \
(*_chans)[_chan].ic_flags = IEEE80211_CHAN_B; \
(*_chans)[_chan].ic_freq = \
ieee80211_ieee2mhz(_chan, (*_chans)[_chan].ic_flags);\
} while (0)
switch (locale) {
case RTW_LOCALE_USA: /* 1-11 */
name = "USA";
for (i = 1; i <= 11; i++)
ADD_CHANNEL(chans, i);
break;
case RTW_LOCALE_JAPAN: /* 1-14 */
name = "Japan";
ADD_CHANNEL(chans, 14);
for (i = 1; i <= 14; i++)
ADD_CHANNEL(chans, i);
break;
case RTW_LOCALE_EUROPE: /* 1-13 */
name = "Europe";
for (i = 1; i <= 13; i++)
ADD_CHANNEL(chans, i);
break;
default: /* 10-11 allowed by most countries */
name = "<unknown>";
for (i = 10; i <= 11; i++)
ADD_CHANNEL(chans, i);
break;
}
Get rid of the silly char (*dvname)[IFNAMSIZ] pattern. Replace it with const char *dvname. I'm not sure what I was thinking.
2004-12-20 03:28:02 +03:00
printf("%s: Geographic Location %s\n", dvname, name);
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
#undef ADD_CHANNEL
}
static __inline void
rtw_identify_country(struct rtw_regs *regs, enum rtw_locale *locale,
Get rid of the silly char (*dvname)[IFNAMSIZ] pattern. Replace it with const char *dvname. I'm not sure what I was thinking.
2004-12-20 03:28:02 +03:00
const char *dvname)
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
{
u_int8_t cfg0 = RTW_READ8(regs, RTW_CONFIG0);
switch (cfg0 & RTW_CONFIG0_GL_MASK) {
case RTW_CONFIG0_GL_USA:
*locale = RTW_LOCALE_USA;
break;
case RTW_CONFIG0_GL_JAPAN:
*locale = RTW_LOCALE_JAPAN;
break;
case RTW_CONFIG0_GL_EUROPE:
*locale = RTW_LOCALE_EUROPE;
break;
default:
*locale = RTW_LOCALE_UNKNOWN;
break;
}
}
static __inline int
rtw_identify_sta(struct rtw_regs *regs, u_int8_t (*addr)[IEEE80211_ADDR_LEN],
Get rid of the silly char (*dvname)[IFNAMSIZ] pattern. Replace it with const char *dvname. I'm not sure what I was thinking.
2004-12-20 03:28:02 +03:00
const char *dvname)
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
{
static const u_int8_t empty_macaddr[IEEE80211_ADDR_LEN] = {
0x00, 0x00, 0x00, 0x00, 0x00, 0x00
};
u_int32_t idr0 = RTW_READ(regs, RTW_IDR0),
idr1 = RTW_READ(regs, RTW_IDR1);
(*addr)[0] = MASK_AND_RSHIFT(idr0, BITS(0, 7));
(*addr)[1] = MASK_AND_RSHIFT(idr0, BITS(8, 15));
(*addr)[2] = MASK_AND_RSHIFT(idr0, BITS(16, 23));
(*addr)[3] = MASK_AND_RSHIFT(idr0, BITS(24 ,31));
(*addr)[4] = MASK_AND_RSHIFT(idr1, BITS(0, 7));
(*addr)[5] = MASK_AND_RSHIFT(idr1, BITS(8, 15));
if (IEEE80211_ADDR_EQ(addr, empty_macaddr)) {
printf("%s: could not get mac address, attach failed\n",
Get rid of the silly char (*dvname)[IFNAMSIZ] pattern. Replace it with const char *dvname. I'm not sure what I was thinking.
2004-12-20 03:28:02 +03:00
dvname);
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
return ENXIO;
}
Get rid of the silly char (*dvname)[IFNAMSIZ] pattern. Replace it with const char *dvname. I'm not sure what I was thinking.
2004-12-20 03:28:02 +03:00
printf("%s: 802.11 address %s\n", dvname, ether_sprintf(*addr));
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
return 0;
}
static u_int8_t
rtw_chan2txpower(struct rtw_srom *sr, struct ieee80211com *ic,
struct ieee80211_channel *chan)
{
u_int idx = RTW_SR_TXPOWER1 + ieee80211_chan2ieee(ic, chan) - 1;
KASSERT2(idx >= RTW_SR_TXPOWER1 && idx <= RTW_SR_TXPOWER14,
("%s: channel %d out of range", __func__,
idx - RTW_SR_TXPOWER1 + 1));
return RTW_SR_GET(sr, idx);
}
static void
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
rtw_txdesc_blk_init_all(struct rtw_txdesc_blk *tdb)
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
{
int pri;
u_int ndesc[RTW_NTXPRI] =
{RTW_NTXDESCLO, RTW_NTXDESCMD, RTW_NTXDESCHI, RTW_NTXDESCBCN};
for (pri = 0; pri < RTW_NTXPRI; pri++) {
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
tdb[pri].tdb_nfree = ndesc[pri];
tdb[pri].tdb_next = 0;
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
}
}
static int
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
rtw_txsoft_blk_init(struct rtw_txsoft_blk *tsb)
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
{
int i;
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
struct rtw_txsoft *ts;
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
SIMPLEQ_INIT(&tsb->tsb_dirtyq);
SIMPLEQ_INIT(&tsb->tsb_freeq);
for (i = 0; i < tsb->tsb_ndesc; i++) {
ts = &tsb->tsb_desc[i];
ts->ts_mbuf = NULL;
SIMPLEQ_INSERT_TAIL(&tsb->tsb_freeq, ts, ts_q);
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
}
return 0;
}
static void
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
rtw_txsoft_blk_init_all(struct rtw_txsoft_blk *tsb)
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
{
int pri;
Miscellaneous changes. Details below. Important changes flagged with []. Using the driver with my Linksys WPC11 ver. 4, it seems to be receiving packets for a change. The WPC11 ver. 4 has a Maxim RF section. My no-name rtw with Philips RF section still does not receive any packets. Keep access-level (analog params > config[0123] registers > none) in sc_access. Add rtw_set_access for changing the access level. Make rtw_continuous_tx_enable and other subroutines use rtw_set_access instead of rtw_config0123_enable and rtw_anaparm_enable. Factor part of the chip-reset code into rtw_chip_reset1. Change the 'struct foo (*bar)[N]'-style arguments to 'struct foo *bar'-style arguments. Consolidate software/hardware Tx/Rx ring setup in rtw_hwring_setup, rtw_swring_setup. Add a new constant, SA2400_OPMODE_DEFAULTS, for the bits that we *always* set in the SA2400 OPMODE register. Factor some code out into rtw_sa2400_calibrate. (Inspired by the Linux driver.) [] When the receiver goes into underrun/overflow state, call a new subroutine, rtw_kick() that stops the Rx/Tx processes, resets the chip, reinitializes the Tx/Rx rings, and restarts Rx/Tx processes. (Inspired by the Linux driver.) [] In rtw_intr_rx, check for too-short packets before calling ieee80211_find_rxnode. I believe this will prevent a repeat of the MCHK exception I saw once on macppc. [] Use seconds-elapased as well as microseconds-elapsed to set the next "due date" for the timeout interrupt. This keeps the driver from programming the timeout to expire too early. [] In rtw_intr, read RTW_ISR at most 10 times, then get out. If the interface is not enabled (RTW_F_ENABLED), then get out. [] In rtw_stop, get out if the interface is not enabled (RTW_F_ENABLED). Block IPL_NET interrupts. Don't read/write any registers if the interface is invalid (RTW_F_INVALID). [] Call rtw_stop in rtw_detach.
2004-12-12 09:37:59 +03:00
for (pri = 0; pri < RTW_NTXPRI; pri++)
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
rtw_txsoft_blk_init(&tsb[pri]);
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
}
static __inline void
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
rtw_rxdescs_sync(struct rtw_rxdesc_blk *rdb, int desc0, int nsync, int ops)
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
{
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
KASSERT(nsync <= rdb->rdb_ndesc);
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
/* sync to end of ring */
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
if (desc0 + nsync > rdb->rdb_ndesc) {
bus_dmamap_sync(rdb->rdb_dmat, rdb->rdb_dmamap,
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
offsetof(struct rtw_descs, hd_rx[desc0]),
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
sizeof(struct rtw_rxdesc) * (rdb->rdb_ndesc - desc0), ops);
nsync -= (rdb->rdb_ndesc - desc0);
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
desc0 = 0;
}
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
KASSERT(desc0 < rdb->rdb_ndesc);
KASSERT(nsync <= rdb->rdb_ndesc);
KASSERT(desc0 + nsync <= rdb->rdb_ndesc);
Change rtw_debug from a debug level to a debug mask. Add a lot of debug flags. From Linux: handle an RTL8180 bug. Sometimes the NIC skips from the middle of the ring to the 0th rx descriptor. Now the driver resynchronizes. Handle a receive descriptor underrun or Rx FIFO overflow condition in the way that the Linux driver does. This kind of seems like overkill, but whatever. Protect rtw_ioctl with splnet(). Do not load a tx descriptor with a buffer shorter than 4 bytes. Handle a transmit timeout less disruptively.
2004-12-25 09:58:37 +03:00
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
/* sync what remains */
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
bus_dmamap_sync(rdb->rdb_dmat, rdb->rdb_dmamap,
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
offsetof(struct rtw_descs, hd_rx[desc0]),
sizeof(struct rtw_rxdesc) * nsync, ops);
}
static void
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
rtw_txdescs_sync(struct rtw_txdesc_blk *tdb, u_int desc0, u_int nsync, int ops)
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
{
/* sync to end of ring */
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
if (desc0 + nsync > tdb->tdb_ndesc) {
bus_dmamap_sync(tdb->tdb_dmat, tdb->tdb_dmamap,
tdb->tdb_ofs + sizeof(struct rtw_txdesc) * desc0,
sizeof(struct rtw_txdesc) * (tdb->tdb_ndesc - desc0),
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
ops);
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
nsync -= (tdb->tdb_ndesc - desc0);
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
desc0 = 0;
}
/* sync what remains */
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
bus_dmamap_sync(tdb->tdb_dmat, tdb->tdb_dmamap,
tdb->tdb_ofs + sizeof(struct rtw_txdesc) * desc0,
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
sizeof(struct rtw_txdesc) * nsync, ops);
}
static void
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
rtw_txdescs_sync_all(struct rtw_txdesc_blk *tdb)
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
{
int pri;
for (pri = 0; pri < RTW_NTXPRI; pri++) {
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
rtw_txdescs_sync(&tdb[pri], 0, tdb[pri].tdb_ndesc,
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
BUS_DMASYNC_PREREAD|BUS_DMASYNC_PREWRITE);
}
}
static void
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
rtw_rxbufs_release(bus_dma_tag_t dmat, struct rtw_rxsoft *desc)
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
{
int i;
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
struct rtw_rxsoft *rs;
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
Change rtw_debug from a debug level to a debug mask. Add a lot of debug flags. From Linux: handle an RTL8180 bug. Sometimes the NIC skips from the middle of the ring to the 0th rx descriptor. Now the driver resynchronizes. Handle a receive descriptor underrun or Rx FIFO overflow condition in the way that the Linux driver does. This kind of seems like overkill, but whatever. Protect rtw_ioctl with splnet(). Do not load a tx descriptor with a buffer shorter than 4 bytes. Handle a transmit timeout less disruptively.
2004-12-25 09:58:37 +03:00
for (i = 0; i < RTW_RXQLEN; i++) {
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
rs = &desc[i];
if (rs->rs_mbuf == NULL)
If defined(RTW_DEBUG), provide a sysctl (hw.rtw.rxbufs_limit) for limiting the number of rx buffers an rtw may allocate. Use this sysctl to test the code that copes with buffer exhaustion. Allocate at most RTW_RXQLEN rx buffers, stopping at the sysctl limit. Record in sc_nrxdesc how many were allocated, and put the end-of-ring indication on sc_rxdesc[sc_nrxdesc - 1]. In rtw_init, if no rx buffers could be allocated, log a complaint, clear IFF_RUNNING, and exit with an error. Many changes to accomodate a short rx ring, mainly of the "add a rx-ring length argument" variety. XXX I really should consolidate all of the rx ring variables in one struct and pass that to the rx-ring subroutines. Bug fix: after calling MCLGET, use the (m->m_flags & M_EXT) idiom to check for success, instead of m != NULL. Bug fix: at the top of rtw_start, if IFF_RUNNING is not set, or IFF_OACTIVE is, get out.
2004-12-29 01:21:15 +03:00
continue;
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
bus_dmamap_sync(dmat, rs->rs_dmamap, 0,
rs->rs_dmamap->dm_mapsize, BUS_DMASYNC_POSTREAD);
bus_dmamap_unload(dmat, rs->rs_dmamap);
m_freem(rs->rs_mbuf);
rs->rs_mbuf = NULL;
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
}
}
static __inline int
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
rtw_rxsoft_alloc(bus_dma_tag_t dmat, struct rtw_rxsoft *rs)
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
{
int rc;
struct mbuf *m;
MGETHDR(m, M_DONTWAIT, MT_DATA);
if (m == NULL)
Bug fix: in rtw_rxbuf_alloc, unload the previous mbuf chain before loading the new chain.
2004-12-23 09:03:09 +03:00
return ENOBUFS;
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
MCLGET(m, M_DONTWAIT);
If defined(RTW_DEBUG), provide a sysctl (hw.rtw.rxbufs_limit) for limiting the number of rx buffers an rtw may allocate. Use this sysctl to test the code that copes with buffer exhaustion. Allocate at most RTW_RXQLEN rx buffers, stopping at the sysctl limit. Record in sc_nrxdesc how many were allocated, and put the end-of-ring indication on sc_rxdesc[sc_nrxdesc - 1]. In rtw_init, if no rx buffers could be allocated, log a complaint, clear IFF_RUNNING, and exit with an error. Many changes to accomodate a short rx ring, mainly of the "add a rx-ring length argument" variety. XXX I really should consolidate all of the rx ring variables in one struct and pass that to the rx-ring subroutines. Bug fix: after calling MCLGET, use the (m->m_flags & M_EXT) idiom to check for success, instead of m != NULL. Bug fix: at the top of rtw_start, if IFF_RUNNING is not set, or IFF_OACTIVE is, get out.
2004-12-29 01:21:15 +03:00
if ((m->m_flags & M_EXT) == 0) {
m_freem(m);
Bug fix: in rtw_rxbuf_alloc, unload the previous mbuf chain before loading the new chain.
2004-12-23 09:03:09 +03:00
return ENOBUFS;
If defined(RTW_DEBUG), provide a sysctl (hw.rtw.rxbufs_limit) for limiting the number of rx buffers an rtw may allocate. Use this sysctl to test the code that copes with buffer exhaustion. Allocate at most RTW_RXQLEN rx buffers, stopping at the sysctl limit. Record in sc_nrxdesc how many were allocated, and put the end-of-ring indication on sc_rxdesc[sc_nrxdesc - 1]. In rtw_init, if no rx buffers could be allocated, log a complaint, clear IFF_RUNNING, and exit with an error. Many changes to accomodate a short rx ring, mainly of the "add a rx-ring length argument" variety. XXX I really should consolidate all of the rx ring variables in one struct and pass that to the rx-ring subroutines. Bug fix: after calling MCLGET, use the (m->m_flags & M_EXT) idiom to check for success, instead of m != NULL. Bug fix: at the top of rtw_start, if IFF_RUNNING is not set, or IFF_OACTIVE is, get out.
2004-12-29 01:21:15 +03:00
}
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
m->m_pkthdr.len = m->m_len = m->m_ext.ext_size;
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
if (rs->rs_mbuf != NULL)
bus_dmamap_unload(dmat, rs->rs_dmamap);
Bug fix: in rtw_rxbuf_alloc, unload the previous mbuf chain before loading the new chain.
2004-12-23 09:03:09 +03:00
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
rs->rs_mbuf = NULL;
Bug fix: in rtw_rxbuf_alloc, unload the previous mbuf chain before loading the new chain.
2004-12-23 09:03:09 +03:00
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
rc = bus_dmamap_load_mbuf(dmat, rs->rs_dmamap, m, BUS_DMA_NOWAIT);
Bug fix: in rtw_rxbuf_alloc, unload the previous mbuf chain before loading the new chain.
2004-12-23 09:03:09 +03:00
if (rc != 0) {
m_freem(m);
return -1;
}
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
rs->rs_mbuf = m;
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
return 0;
}
static int
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
rtw_rxsoft_init_all(bus_dma_tag_t dmat, struct rtw_rxsoft *desc,
If defined(RTW_DEBUG), provide a sysctl (hw.rtw.rxbufs_limit) for limiting the number of rx buffers an rtw may allocate. Use this sysctl to test the code that copes with buffer exhaustion. Allocate at most RTW_RXQLEN rx buffers, stopping at the sysctl limit. Record in sc_nrxdesc how many were allocated, and put the end-of-ring indication on sc_rxdesc[sc_nrxdesc - 1]. In rtw_init, if no rx buffers could be allocated, log a complaint, clear IFF_RUNNING, and exit with an error. Many changes to accomodate a short rx ring, mainly of the "add a rx-ring length argument" variety. XXX I really should consolidate all of the rx ring variables in one struct and pass that to the rx-ring subroutines. Bug fix: after calling MCLGET, use the (m->m_flags & M_EXT) idiom to check for success, instead of m != NULL. Bug fix: at the top of rtw_start, if IFF_RUNNING is not set, or IFF_OACTIVE is, get out.
2004-12-29 01:21:15 +03:00
int *ndesc, const char *dvname)
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
{
If defined(RTW_DEBUG), provide a sysctl (hw.rtw.rxbufs_limit) for limiting the number of rx buffers an rtw may allocate. Use this sysctl to test the code that copes with buffer exhaustion. Allocate at most RTW_RXQLEN rx buffers, stopping at the sysctl limit. Record in sc_nrxdesc how many were allocated, and put the end-of-ring indication on sc_rxdesc[sc_nrxdesc - 1]. In rtw_init, if no rx buffers could be allocated, log a complaint, clear IFF_RUNNING, and exit with an error. Many changes to accomodate a short rx ring, mainly of the "add a rx-ring length argument" variety. XXX I really should consolidate all of the rx ring variables in one struct and pass that to the rx-ring subroutines. Bug fix: after calling MCLGET, use the (m->m_flags & M_EXT) idiom to check for success, instead of m != NULL. Bug fix: at the top of rtw_start, if IFF_RUNNING is not set, or IFF_OACTIVE is, get out.
2004-12-29 01:21:15 +03:00
int i, rc = 0;
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
struct rtw_rxsoft *rs;
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
Change rtw_debug from a debug level to a debug mask. Add a lot of debug flags. From Linux: handle an RTL8180 bug. Sometimes the NIC skips from the middle of the ring to the 0th rx descriptor. Now the driver resynchronizes. Handle a receive descriptor underrun or Rx FIFO overflow condition in the way that the Linux driver does. This kind of seems like overkill, but whatever. Protect rtw_ioctl with splnet(). Do not load a tx descriptor with a buffer shorter than 4 bytes. Handle a transmit timeout less disruptively.
2004-12-25 09:58:37 +03:00
for (i = 0; i < RTW_RXQLEN; i++) {
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
rs = &desc[i];
If defined(RTW_DEBUG), provide a sysctl (hw.rtw.rxbufs_limit) for limiting the number of rx buffers an rtw may allocate. Use this sysctl to test the code that copes with buffer exhaustion. Allocate at most RTW_RXQLEN rx buffers, stopping at the sysctl limit. Record in sc_nrxdesc how many were allocated, and put the end-of-ring indication on sc_rxdesc[sc_nrxdesc - 1]. In rtw_init, if no rx buffers could be allocated, log a complaint, clear IFF_RUNNING, and exit with an error. Many changes to accomodate a short rx ring, mainly of the "add a rx-ring length argument" variety. XXX I really should consolidate all of the rx ring variables in one struct and pass that to the rx-ring subroutines. Bug fix: after calling MCLGET, use the (m->m_flags & M_EXT) idiom to check for success, instead of m != NULL. Bug fix: at the top of rtw_start, if IFF_RUNNING is not set, or IFF_OACTIVE is, get out.
2004-12-29 01:21:15 +03:00
/* we're in rtw_init, so there should be no mbufs allocated */
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
KASSERT(rs->rs_mbuf == NULL);
If defined(RTW_DEBUG), provide a sysctl (hw.rtw.rxbufs_limit) for limiting the number of rx buffers an rtw may allocate. Use this sysctl to test the code that copes with buffer exhaustion. Allocate at most RTW_RXQLEN rx buffers, stopping at the sysctl limit. Record in sc_nrxdesc how many were allocated, and put the end-of-ring indication on sc_rxdesc[sc_nrxdesc - 1]. In rtw_init, if no rx buffers could be allocated, log a complaint, clear IFF_RUNNING, and exit with an error. Many changes to accomodate a short rx ring, mainly of the "add a rx-ring length argument" variety. XXX I really should consolidate all of the rx ring variables in one struct and pass that to the rx-ring subroutines. Bug fix: after calling MCLGET, use the (m->m_flags & M_EXT) idiom to check for success, instead of m != NULL. Bug fix: at the top of rtw_start, if IFF_RUNNING is not set, or IFF_OACTIVE is, get out.
2004-12-29 01:21:15 +03:00
#ifdef RTW_DEBUG
if (i == rtw_rxbufs_limit) {
printf("%s: TEST hit %d-buffer limit\n", dvname, i);
rc = ENOBUFS;
break;
}
#endif /* RTW_DEBUG */
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
if ((rc = rtw_rxsoft_alloc(dmat, rs)) != 0) {
printf("%s: rtw_rxsoft_alloc failed, %d buffers, "
If defined(RTW_DEBUG), provide a sysctl (hw.rtw.rxbufs_limit) for limiting the number of rx buffers an rtw may allocate. Use this sysctl to test the code that copes with buffer exhaustion. Allocate at most RTW_RXQLEN rx buffers, stopping at the sysctl limit. Record in sc_nrxdesc how many were allocated, and put the end-of-ring indication on sc_rxdesc[sc_nrxdesc - 1]. In rtw_init, if no rx buffers could be allocated, log a complaint, clear IFF_RUNNING, and exit with an error. Many changes to accomodate a short rx ring, mainly of the "add a rx-ring length argument" variety. XXX I really should consolidate all of the rx ring variables in one struct and pass that to the rx-ring subroutines. Bug fix: after calling MCLGET, use the (m->m_flags & M_EXT) idiom to check for success, instead of m != NULL. Bug fix: at the top of rtw_start, if IFF_RUNNING is not set, or IFF_OACTIVE is, get out.
2004-12-29 01:21:15 +03:00
"rc %d\n", dvname, i, rc);
break;
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
}
}
If defined(RTW_DEBUG), provide a sysctl (hw.rtw.rxbufs_limit) for limiting the number of rx buffers an rtw may allocate. Use this sysctl to test the code that copes with buffer exhaustion. Allocate at most RTW_RXQLEN rx buffers, stopping at the sysctl limit. Record in sc_nrxdesc how many were allocated, and put the end-of-ring indication on sc_rxdesc[sc_nrxdesc - 1]. In rtw_init, if no rx buffers could be allocated, log a complaint, clear IFF_RUNNING, and exit with an error. Many changes to accomodate a short rx ring, mainly of the "add a rx-ring length argument" variety. XXX I really should consolidate all of the rx ring variables in one struct and pass that to the rx-ring subroutines. Bug fix: after calling MCLGET, use the (m->m_flags & M_EXT) idiom to check for success, instead of m != NULL. Bug fix: at the top of rtw_start, if IFF_RUNNING is not set, or IFF_OACTIVE is, get out.
2004-12-29 01:21:15 +03:00
*ndesc = i;
return rc;
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
}
static __inline void
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
rtw_rxdesc_init(struct rtw_rxdesc_blk *rdb, struct rtw_rxsoft *rs,
Consolidate variables related to the rx ring in sc_rxdesc_blk, which is a struct rtw_rxdesc_blk. Put a copy of the DMA tag and the DMA map into the rx- and tx-ring blocks so that I don't have to pass them to subroutines all of the time.
2004-12-29 03:21:37 +03:00
int idx, int kick)
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
{
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
int is_last = (idx == rdb->rdb_ndesc - 1);
Change rtw_debug from a debug level to a debug mask. Add a lot of debug flags. From Linux: handle an RTL8180 bug. Sometimes the NIC skips from the middle of the ring to the 0th rx descriptor. Now the driver resynchronizes. Handle a receive descriptor underrun or Rx FIFO overflow condition in the way that the Linux driver does. This kind of seems like overkill, but whatever. Protect rtw_ioctl with splnet(). Do not load a tx descriptor with a buffer shorter than 4 bytes. Handle a transmit timeout less disruptively.
2004-12-25 09:58:37 +03:00
uint32_t ctl, octl, obuf;
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
struct rtw_rxdesc *rd = &rdb->rdb_desc[idx];
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
obuf = rd->rd_buf;
rd->rd_buf = htole32(rs->rs_dmamap->dm_segs[0].ds_addr);
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
ctl = LSHIFT(rs->rs_mbuf->m_len, RTW_RXCTL_LENGTH_MASK) |
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
RTW_RXCTL_OWN | RTW_RXCTL_FS | RTW_RXCTL_LS;
if (is_last)
ctl |= RTW_RXCTL_EOR;
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
octl = rd->rd_ctl;
rd->rd_ctl = htole32(ctl);
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
Make rtw(4) build w/o RTW_DEBUG. Thanks to Charles Hannum for pointing out my oversight.
2004-12-27 01:59:41 +03:00
RTW_DPRINTF(
kick ? (RTW_DEBUG_RECV_DESC | RTW_DEBUG_IO_KICK)
: RTW_DEBUG_RECV_DESC,
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
("%s: rd %p buf %08x -> %08x ctl %08x -> %08x\n", __func__, rd,
le32toh(obuf), le32toh(rd->rd_buf), le32toh(octl),
le32toh(rd->rd_ctl)));
Change rtw_debug from a debug level to a debug mask. Add a lot of debug flags. From Linux: handle an RTL8180 bug. Sometimes the NIC skips from the middle of the ring to the 0th rx descriptor. Now the driver resynchronizes. Handle a receive descriptor underrun or Rx FIFO overflow condition in the way that the Linux driver does. This kind of seems like overkill, but whatever. Protect rtw_ioctl with splnet(). Do not load a tx descriptor with a buffer shorter than 4 bytes. Handle a transmit timeout less disruptively.
2004-12-25 09:58:37 +03:00
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
/* sync the mbuf */
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
bus_dmamap_sync(rdb->rdb_dmat, rs->rs_dmamap, 0,
rs->rs_dmamap->dm_mapsize, BUS_DMASYNC_PREREAD);
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
/* sync the descriptor */
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
bus_dmamap_sync(rdb->rdb_dmat, rdb->rdb_dmamap,
Consolidate variables related to the rx ring in sc_rxdesc_blk, which is a struct rtw_rxdesc_blk. Put a copy of the DMA tag and the DMA map into the rx- and tx-ring blocks so that I don't have to pass them to subroutines all of the time.
2004-12-29 03:21:37 +03:00
RTW_DESC_OFFSET(hd_rx, idx), sizeof(struct rtw_rxdesc),
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
BUS_DMASYNC_PREREAD|BUS_DMASYNC_PREWRITE);
}
static void
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
rtw_rxdesc_init_all(struct rtw_rxdesc_blk *rdb, struct rtw_rxsoft *ctl, int kick)
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
{
int i;
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
struct rtw_rxdesc *rd;
struct rtw_rxsoft *rs;
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
for (i = 0; i < rdb->rdb_ndesc; i++) {
rd = &rdb->rdb_desc[i];
rs = &ctl[i];
rtw_rxdesc_init(rdb, rs, i, kick);
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
}
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
rdb->rdb_next = 0;
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
}
static void
rtw_io_enable(struct rtw_regs *regs, u_int8_t flags, int enable)
{
u_int8_t cr;
Change rtw_debug from a debug level to a debug mask. Add a lot of debug flags. From Linux: handle an RTL8180 bug. Sometimes the NIC skips from the middle of the ring to the 0th rx descriptor. Now the driver resynchronizes. Handle a receive descriptor underrun or Rx FIFO overflow condition in the way that the Linux driver does. This kind of seems like overkill, but whatever. Protect rtw_ioctl with splnet(). Do not load a tx descriptor with a buffer shorter than 4 bytes. Handle a transmit timeout less disruptively.
2004-12-25 09:58:37 +03:00
RTW_DPRINTF(RTW_DEBUG_IOSTATE, ("%s: %s 0x%02x\n", __func__,
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
enable ? "enable" : "disable", flags));
cr = RTW_READ8(regs, RTW_CR);
/* XXX reference source does not enable MULRW */
#if 0
/* enable PCI Read/Write Multiple */
cr |= RTW_CR_MULRW;
#endif
RTW_RBW(regs, RTW_CR, RTW_CR); /* XXX paranoia? */
if (enable)
cr |= flags;
else
cr &= ~flags;
RTW_WRITE8(regs, RTW_CR, cr);
RTW_SYNC(regs, RTW_CR, RTW_CR);
}
static void
rtw_intr_rx(struct rtw_softc *sc, u_int16_t isr)
{
Use a table to convert the hardware's bit rates (0, 1, 2, 3) to net80211 bit rates (2, 4, 11, 22). Count input errors (if_ierrors++) for "DMA error/FIFO overflow" and unknown bit rate.
2004-12-27 23:04:45 +03:00
static const int ratetbl[4] = {2, 4, 11, 22}; /* convert rates:
* hardware -> net80211
*/
Change rtw_debug from a debug level to a debug mask. Add a lot of debug flags. From Linux: handle an RTL8180 bug. Sometimes the NIC skips from the middle of the ring to the 0th rx descriptor. Now the driver resynchronizes. Handle a receive descriptor underrun or Rx FIFO overflow condition in the way that the Linux driver does. This kind of seems like overkill, but whatever. Protect rtw_ioctl with splnet(). Do not load a tx descriptor with a buffer shorter than 4 bytes. Handle a transmit timeout less disruptively.
2004-12-25 09:58:37 +03:00
u_int next, nproc = 0;
Add bpf(4) packet taps to rtw(4). Now you can tap DLT_EN10MB (Ethernet II), DLT_IEEE802_11 (802.11), and DLT_IEEE802_11_RADIO (radiotap + 802.11).
2004-12-29 01:30:07 +03:00
int hwrate, len, rate, rssi, sq;
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
u_int32_t hrssi, hstat, htsfth, htsftl;
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
struct rtw_rxdesc *rd;
struct rtw_rxsoft *rs;
struct rtw_rxdesc_blk *rdb;
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
struct mbuf *m;
struct ieee80211_node *ni;
struct ieee80211_frame *wh;
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
rdb = &sc->sc_rxdesc_blk;
Change rtw_debug from a debug level to a debug mask. Add a lot of debug flags. From Linux: handle an RTL8180 bug. Sometimes the NIC skips from the middle of the ring to the 0th rx descriptor. Now the driver resynchronizes. Handle a receive descriptor underrun or Rx FIFO overflow condition in the way that the Linux driver does. This kind of seems like overkill, but whatever. Protect rtw_ioctl with splnet(). Do not load a tx descriptor with a buffer shorter than 4 bytes. Handle a transmit timeout less disruptively.
2004-12-25 09:58:37 +03:00
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
KASSERT(rdb->rdb_next < rdb->rdb_ndesc);
Consolidate variables related to the rx ring in sc_rxdesc_blk, which is a struct rtw_rxdesc_blk. Put a copy of the DMA tag and the DMA map into the rx- and tx-ring blocks so that I don't have to pass them to subroutines all of the time.
2004-12-29 03:21:37 +03:00
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
for (next = rdb->rdb_next; ; next = (next + 1) % rdb->rdb_ndesc) {
rtw_rxdescs_sync(rdb, next, 1,
If defined(RTW_DEBUG), provide a sysctl (hw.rtw.rxbufs_limit) for limiting the number of rx buffers an rtw may allocate. Use this sysctl to test the code that copes with buffer exhaustion. Allocate at most RTW_RXQLEN rx buffers, stopping at the sysctl limit. Record in sc_nrxdesc how many were allocated, and put the end-of-ring indication on sc_rxdesc[sc_nrxdesc - 1]. In rtw_init, if no rx buffers could be allocated, log a complaint, clear IFF_RUNNING, and exit with an error. Many changes to accomodate a short rx ring, mainly of the "add a rx-ring length argument" variety. XXX I really should consolidate all of the rx ring variables in one struct and pass that to the rx-ring subroutines. Bug fix: after calling MCLGET, use the (m->m_flags & M_EXT) idiom to check for success, instead of m != NULL. Bug fix: at the top of rtw_start, if IFF_RUNNING is not set, or IFF_OACTIVE is, get out.
2004-12-29 01:21:15 +03:00
BUS_DMASYNC_POSTREAD|BUS_DMASYNC_POSTWRITE);
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
rd = &rdb->rdb_desc[next];
rs = &sc->sc_rxsoft[next];
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
hstat = le32toh(rd->rd_stat);
hrssi = le32toh(rd->rd_rssi);
htsfth = le32toh(rd->rd_tsfth);
htsftl = le32toh(rd->rd_tsftl);
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
Change rtw_debug from a debug level to a debug mask. Add a lot of debug flags. From Linux: handle an RTL8180 bug. Sometimes the NIC skips from the middle of the ring to the 0th rx descriptor. Now the driver resynchronizes. Handle a receive descriptor underrun or Rx FIFO overflow condition in the way that the Linux driver does. This kind of seems like overkill, but whatever. Protect rtw_ioctl with splnet(). Do not load a tx descriptor with a buffer shorter than 4 bytes. Handle a transmit timeout less disruptively.
2004-12-25 09:58:37 +03:00
RTW_DPRINTF(RTW_DEBUG_RECV_DESC,
("%s: rxdesc[%d] hstat %08x hrssi %08x htsft %08x%08x\n",
__func__, next, hstat, hrssi, htsfth, htsftl));
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
Change rtw_debug from a debug level to a debug mask. Add a lot of debug flags. From Linux: handle an RTL8180 bug. Sometimes the NIC skips from the middle of the ring to the 0th rx descriptor. Now the driver resynchronizes. Handle a receive descriptor underrun or Rx FIFO overflow condition in the way that the Linux driver does. This kind of seems like overkill, but whatever. Protect rtw_ioctl with splnet(). Do not load a tx descriptor with a buffer shorter than 4 bytes. Handle a transmit timeout less disruptively.
2004-12-25 09:58:37 +03:00
KASSERT((hstat & (RTW_RXSTAT_FS|RTW_RXSTAT_LS)) ==
(RTW_RXSTAT_FS|RTW_RXSTAT_LS));
++nproc;
/* still belongs to NIC */
if ((hstat & RTW_RXSTAT_OWN) != 0) {
if (nproc > 1)
break;
/* sometimes the NIC skips to the 0th descriptor */
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
rtw_rxdescs_sync(rdb, 0, 1,
If defined(RTW_DEBUG), provide a sysctl (hw.rtw.rxbufs_limit) for limiting the number of rx buffers an rtw may allocate. Use this sysctl to test the code that copes with buffer exhaustion. Allocate at most RTW_RXQLEN rx buffers, stopping at the sysctl limit. Record in sc_nrxdesc how many were allocated, and put the end-of-ring indication on sc_rxdesc[sc_nrxdesc - 1]. In rtw_init, if no rx buffers could be allocated, log a complaint, clear IFF_RUNNING, and exit with an error. Many changes to accomodate a short rx ring, mainly of the "add a rx-ring length argument" variety. XXX I really should consolidate all of the rx ring variables in one struct and pass that to the rx-ring subroutines. Bug fix: after calling MCLGET, use the (m->m_flags & M_EXT) idiom to check for success, instead of m != NULL. Bug fix: at the top of rtw_start, if IFF_RUNNING is not set, or IFF_OACTIVE is, get out.
2004-12-29 01:21:15 +03:00
BUS_DMASYNC_POSTREAD|BUS_DMASYNC_POSTWRITE);
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
rd = &rdb->rdb_desc[0];
if ((rd->rd_stat & htole32(RTW_RXSTAT_OWN)) != 0)
Change rtw_debug from a debug level to a debug mask. Add a lot of debug flags. From Linux: handle an RTL8180 bug. Sometimes the NIC skips from the middle of the ring to the 0th rx descriptor. Now the driver resynchronizes. Handle a receive descriptor underrun or Rx FIFO overflow condition in the way that the Linux driver does. This kind of seems like overkill, but whatever. Protect rtw_ioctl with splnet(). Do not load a tx descriptor with a buffer shorter than 4 bytes. Handle a transmit timeout less disruptively.
2004-12-25 09:58:37 +03:00
break;
The rx descriptor ring bug happens quite often, so printf'ing occurrences is annoying. Only printf about it when rtw_debug contains RTW_DEBUG_BUG.
2004-12-25 10:45:53 +03:00
RTW_DPRINTF(RTW_DEBUG_BUGS,
("%s: NIC skipped to rxdesc[0]\n",
sc->sc_dev.dv_xname));
Change rtw_debug from a debug level to a debug mask. Add a lot of debug flags. From Linux: handle an RTL8180 bug. Sometimes the NIC skips from the middle of the ring to the 0th rx descriptor. Now the driver resynchronizes. Handle a receive descriptor underrun or Rx FIFO overflow condition in the way that the Linux driver does. This kind of seems like overkill, but whatever. Protect rtw_ioctl with splnet(). Do not load a tx descriptor with a buffer shorter than 4 bytes. Handle a transmit timeout less disruptively.
2004-12-25 09:58:37 +03:00
next = 0;
continue;
}
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
if ((hstat & RTW_RXSTAT_IOERROR) != 0) {
printf("%s: DMA error/FIFO overflow %08x, "
"rx descriptor %d\n", sc->sc_dev.dv_xname,
hstat & RTW_RXSTAT_IOERROR, next);
Use a table to convert the hardware's bit rates (0, 1, 2, 3) to net80211 bit rates (2, 4, 11, 22). Count input errors (if_ierrors++) for "DMA error/FIFO overflow" and unknown bit rate.
2004-12-27 23:04:45 +03:00
sc->sc_if.if_ierrors++;
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
goto next;
}
IMPORTANT: set m->m_pkthdr.rcvif on rx'd packets. Illegal memory references abound without this fix. In rtw_intr_rx, skip to the next rx packet earlier if the rx packet length is too short. Also in rtw_intr_rx, if a packet is rx'd with unknown rate, printf a warning, drop the packet, recycle the rx buffer, skip to next rx packet.
2004-12-25 10:24:17 +03:00
len = MASK_AND_RSHIFT(hstat, RTW_RXSTAT_LENGTH_MASK);
if (len < IEEE80211_MIN_LEN) {
sc->sc_ic.ic_stats.is_rx_tooshort++;
goto next;
}
Use a table to convert the hardware's bit rates (0, 1, 2, 3) to net80211 bit rates (2, 4, 11, 22). Count input errors (if_ierrors++) for "DMA error/FIFO overflow" and unknown bit rate.
2004-12-27 23:04:45 +03:00
hwrate = MASK_AND_RSHIFT(hstat, RTW_RXSTAT_RATE_MASK);
if (hwrate >= sizeof(ratetbl) / sizeof(ratetbl[0])) {
IMPORTANT: set m->m_pkthdr.rcvif on rx'd packets. Illegal memory references abound without this fix. In rtw_intr_rx, skip to the next rx packet earlier if the rx packet length is too short. Also in rtw_intr_rx, if a packet is rx'd with unknown rate, printf a warning, drop the packet, recycle the rx buffer, skip to next rx packet.
2004-12-25 10:24:17 +03:00
printf("%s: unknown rate #%d\n", sc->sc_dev.dv_xname,
MASK_AND_RSHIFT(hstat, RTW_RXSTAT_RATE_MASK));
Use a table to convert the hardware's bit rates (0, 1, 2, 3) to net80211 bit rates (2, 4, 11, 22). Count input errors (if_ierrors++) for "DMA error/FIFO overflow" and unknown bit rate.
2004-12-27 23:04:45 +03:00
sc->sc_if.if_ierrors++;
IMPORTANT: set m->m_pkthdr.rcvif on rx'd packets. Illegal memory references abound without this fix. In rtw_intr_rx, skip to the next rx packet earlier if the rx packet length is too short. Also in rtw_intr_rx, if a packet is rx'd with unknown rate, printf a warning, drop the packet, recycle the rx buffer, skip to next rx packet.
2004-12-25 10:24:17 +03:00
goto next;
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
}
Use a table to convert the hardware's bit rates (0, 1, 2, 3) to net80211 bit rates (2, 4, 11, 22). Count input errors (if_ierrors++) for "DMA error/FIFO overflow" and unknown bit rate.
2004-12-27 23:04:45 +03:00
rate = ratetbl[hwrate];
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
#ifdef RTW_DEBUG
#define PRINTSTAT(flag) do { \
if ((hstat & flag) != 0) { \
printf("%s" #flag, delim); \
delim = ","; \
} \
} while (0)
Change rtw_debug from a debug level to a debug mask. Add a lot of debug flags. From Linux: handle an RTL8180 bug. Sometimes the NIC skips from the middle of the ring to the 0th rx descriptor. Now the driver resynchronizes. Handle a receive descriptor underrun or Rx FIFO overflow condition in the way that the Linux driver does. This kind of seems like overkill, but whatever. Protect rtw_ioctl with splnet(). Do not load a tx descriptor with a buffer shorter than 4 bytes. Handle a transmit timeout less disruptively.
2004-12-25 09:58:37 +03:00
if ((rtw_debug & RTW_DEBUG_RECV_DESC) != 0) {
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
const char *delim = "<";
printf("%s: ", sc->sc_dev.dv_xname);
if ((hstat & RTW_RXSTAT_DEBUG) != 0) {
Realtek sample sources are not consistent about the "Digital PHY" bit in the SROM. It seems as if it is set to 1 when the PHY is *analog*, not *digital*. Fix my sources. In rtw_intr_rx, use units of 500kb/s instead of 100kb/s for rate, to be consistent with net80211's expectations. Polish up some debugging ugly messages. Dump raw 802.11 packets if IFF_DEBUG|IFF_LINK2 and RTW_DEBUG is defined. Polish power-state (on/sleep/off) handling. Especially improve support for RFMD (totally untested) and Maxim. For Philips, take the Digital PHY property into account. Call the net80211 watchdog function from rtw_watchdog, so that we scan again if auth/assoc fails. Be a little more cautious about writing register[RTW_TPPOLL], since other drivers are.... Don't frob the high/low-priority queues right now, since I don't use them. Add rtw_join_bss which programs the card with the BSSID and other properties of a BSS. Use it on state transitions. Factor out rtw_set_nettype. Make rtw_recv_beacon call ieee80211_recv_mgmt instead of dropping beacons on the floor! TBD IBSS merges. Change some rtw_debug=2 printfs to rtw_debug=3 (RTW_DPRINTF3) printfs so the console doesn't get spammed so badly at rtw_debug=2. Change some debugging printfs to RTW_DPRINTFs. E.g., print the "RF programming method" only if debugging is enabled.
2004-12-21 02:05:41 +03:00
printf("status %08x", hstat);
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
PRINTSTAT(RTW_RXSTAT_SPLCP);
PRINTSTAT(RTW_RXSTAT_MAR);
PRINTSTAT(RTW_RXSTAT_PAR);
PRINTSTAT(RTW_RXSTAT_BAR);
PRINTSTAT(RTW_RXSTAT_PWRMGT);
PRINTSTAT(RTW_RXSTAT_CRC32);
PRINTSTAT(RTW_RXSTAT_ICV);
printf(">, ");
}
printf("rate %d.%d Mb/s, time %08x%08x\n",
Realtek sample sources are not consistent about the "Digital PHY" bit in the SROM. It seems as if it is set to 1 when the PHY is *analog*, not *digital*. Fix my sources. In rtw_intr_rx, use units of 500kb/s instead of 100kb/s for rate, to be consistent with net80211's expectations. Polish up some debugging ugly messages. Dump raw 802.11 packets if IFF_DEBUG|IFF_LINK2 and RTW_DEBUG is defined. Polish power-state (on/sleep/off) handling. Especially improve support for RFMD (totally untested) and Maxim. For Philips, take the Digital PHY property into account. Call the net80211 watchdog function from rtw_watchdog, so that we scan again if auth/assoc fails. Be a little more cautious about writing register[RTW_TPPOLL], since other drivers are.... Don't frob the high/low-priority queues right now, since I don't use them. Add rtw_join_bss which programs the card with the BSSID and other properties of a BSS. Use it on state transitions. Factor out rtw_set_nettype. Make rtw_recv_beacon call ieee80211_recv_mgmt instead of dropping beacons on the floor! TBD IBSS merges. Change some rtw_debug=2 printfs to rtw_debug=3 (RTW_DPRINTF3) printfs so the console doesn't get spammed so badly at rtw_debug=2. Change some debugging printfs to RTW_DPRINTFs. E.g., print the "RF programming method" only if debugging is enabled.
2004-12-21 02:05:41 +03:00
(rate * 5) / 10, (rate * 5) % 10, htsfth, htsftl);
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
}
#endif /* RTW_DEBUG */
if ((hstat & RTW_RXSTAT_RES) != 0 &&
sc->sc_ic.ic_opmode != IEEE80211_M_MONITOR)
goto next;
/* if bad flags, skip descriptor */
if ((hstat & RTW_RXSTAT_ONESEG) != RTW_RXSTAT_ONESEG) {
printf("%s: too many rx segments\n",
sc->sc_dev.dv_xname);
goto next;
}
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
bus_dmamap_sync(sc->sc_dmat, rs->rs_dmamap, 0,
rs->rs_dmamap->dm_mapsize, BUS_DMASYNC_POSTREAD);
Bug fix: in rtw_rxbuf_alloc, unload the previous mbuf chain before loading the new chain.
2004-12-23 09:03:09 +03:00
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
m = rs->rs_mbuf;
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
/* if temporarily out of memory, re-use mbuf */
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
switch (rtw_rxsoft_alloc(sc->sc_dmat, rs)) {
Bug fix: in rtw_rxbuf_alloc, unload the previous mbuf chain before loading the new chain.
2004-12-23 09:03:09 +03:00
case 0:
break;
case ENOBUFS:
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
printf("%s: rtw_rxsoft_alloc(, %d) failed, "
If defined(RTW_DEBUG), provide a sysctl (hw.rtw.rxbufs_limit) for limiting the number of rx buffers an rtw may allocate. Use this sysctl to test the code that copes with buffer exhaustion. Allocate at most RTW_RXQLEN rx buffers, stopping at the sysctl limit. Record in sc_nrxdesc how many were allocated, and put the end-of-ring indication on sc_rxdesc[sc_nrxdesc - 1]. In rtw_init, if no rx buffers could be allocated, log a complaint, clear IFF_RUNNING, and exit with an error. Many changes to accomodate a short rx ring, mainly of the "add a rx-ring length argument" variety. XXX I really should consolidate all of the rx ring variables in one struct and pass that to the rx-ring subroutines. Bug fix: after calling MCLGET, use the (m->m_flags & M_EXT) idiom to check for success, instead of m != NULL. Bug fix: at the top of rtw_start, if IFF_RUNNING is not set, or IFF_OACTIVE is, get out.
2004-12-29 01:21:15 +03:00
"dropping packet\n", sc->sc_dev.dv_xname, next);
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
goto next;
Bug fix: in rtw_rxbuf_alloc, unload the previous mbuf chain before loading the new chain.
2004-12-23 09:03:09 +03:00
default:
/* XXX shorten rx ring, instead? */
panic("%s: could not load DMA map\n",
sc->sc_dev.dv_xname);
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
}
if (sc->sc_rfchipid == RTW_RFCHIPID_PHILIPS)
rssi = MASK_AND_RSHIFT(hrssi, RTW_RXRSSI_RSSI);
else {
rssi = MASK_AND_RSHIFT(hrssi, RTW_RXRSSI_IMR_RSSI);
/* TBD find out each front-end's LNA gain in the
* front-end's units
*/
if ((hrssi & RTW_RXRSSI_IMR_LNA) == 0)
rssi |= 0x80;
}
Add bpf(4) packet taps to rtw(4). Now you can tap DLT_EN10MB (Ethernet II), DLT_IEEE802_11 (802.11), and DLT_IEEE802_11_RADIO (radiotap + 802.11).
2004-12-29 01:30:07 +03:00
sq = MASK_AND_RSHIFT(hrssi, RTW_RXRSSI_SQ);
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
/* Note well: now we cannot recycle the rs_mbuf unless
Add bpf(4) packet taps to rtw(4). Now you can tap DLT_EN10MB (Ethernet II), DLT_IEEE802_11 (802.11), and DLT_IEEE802_11_RADIO (radiotap + 802.11).
2004-12-29 01:30:07 +03:00
* we restore its original length.
*/
IMPORTANT: set m->m_pkthdr.rcvif on rx'd packets. Illegal memory references abound without this fix. In rtw_intr_rx, skip to the next rx packet earlier if the rx packet length is too short. Also in rtw_intr_rx, if a packet is rx'd with unknown rate, printf a warning, drop the packet, recycle the rx buffer, skip to next rx packet.
2004-12-25 10:24:17 +03:00
m->m_pkthdr.rcvif = &sc->sc_if;
m->m_pkthdr.len = m->m_len = len;
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
m->m_flags |= M_HASFCS;
wh = mtod(m, struct ieee80211_frame *);
/* TBD use _MAR, _BAR, _PAR flags as hints to _find_rxnode? */
ni = ieee80211_find_rxnode(&sc->sc_ic, wh);
sc->sc_tsfth = htsfth;
Realtek sample sources are not consistent about the "Digital PHY" bit in the SROM. It seems as if it is set to 1 when the PHY is *analog*, not *digital*. Fix my sources. In rtw_intr_rx, use units of 500kb/s instead of 100kb/s for rate, to be consistent with net80211's expectations. Polish up some debugging ugly messages. Dump raw 802.11 packets if IFF_DEBUG|IFF_LINK2 and RTW_DEBUG is defined. Polish power-state (on/sleep/off) handling. Especially improve support for RFMD (totally untested) and Maxim. For Philips, take the Digital PHY property into account. Call the net80211 watchdog function from rtw_watchdog, so that we scan again if auth/assoc fails. Be a little more cautious about writing register[RTW_TPPOLL], since other drivers are.... Don't frob the high/low-priority queues right now, since I don't use them. Add rtw_join_bss which programs the card with the BSSID and other properties of a BSS. Use it on state transitions. Factor out rtw_set_nettype. Make rtw_recv_beacon call ieee80211_recv_mgmt instead of dropping beacons on the floor! TBD IBSS merges. Change some rtw_debug=2 printfs to rtw_debug=3 (RTW_DPRINTF3) printfs so the console doesn't get spammed so badly at rtw_debug=2. Change some debugging printfs to RTW_DPRINTFs. E.g., print the "RF programming method" only if debugging is enabled.
2004-12-21 02:05:41 +03:00
#ifdef RTW_DEBUG
if ((sc->sc_if.if_flags & (IFF_DEBUG|IFF_LINK2)) ==
(IFF_DEBUG|IFF_LINK2)) {
ieee80211_dump_pkt(mtod(m, uint8_t *), m->m_pkthdr.len,
rate, rssi);
}
#endif /* RTW_DEBUG */
Add bpf(4) packet taps to rtw(4). Now you can tap DLT_EN10MB (Ethernet II), DLT_IEEE802_11 (802.11), and DLT_IEEE802_11_RADIO (radiotap + 802.11).
2004-12-29 01:30:07 +03:00
#if NBPFILTER > 0
if (sc->sc_radiobpf != NULL) {
struct ieee80211com *ic = &sc->sc_ic;
struct rtw_rx_radiotap_header *rr = &sc->sc_rxtap;
rr->rr_tsft =
htole64(((uint64_t)htsfth << 32) | htsftl);
if ((hstat & RTW_RXSTAT_SPLCP) != 0)
rr->rr_flags = IEEE80211_RADIOTAP_F_SHORTPRE;
rr->rr_flags = 0;
rr->rr_rate = rate;
rr->rr_chan_freq =
htole16(ic->ic_bss->ni_chan->ic_freq);
rr->rr_chan_flags =
htole16(ic->ic_bss->ni_chan->ic_flags);
rr->rr_antsignal = rssi;
In rtw_detach, flag the rtw as "invalid" (RTW_F_INVALID). Do not try to power-down the RF section of an invalid rtw. The radiotap signal quality field is 16, not 8, bits wide. Convert the byte order.
2004-12-29 22:41:04 +03:00
rr->rr_barker_lock = htole16(sq);
Add bpf(4) packet taps to rtw(4). Now you can tap DLT_EN10MB (Ethernet II), DLT_IEEE802_11 (802.11), and DLT_IEEE802_11_RADIO (radiotap + 802.11).
2004-12-29 01:30:07 +03:00
bpf_mtap2(sc->sc_radiobpf, (caddr_t)rr,
sizeof(sc->sc_rxtapu), m);
}
#endif /* NPBFILTER > 0 */
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
ieee80211_input(&sc->sc_if, m, ni, rssi, htsftl);
ieee80211_release_node(&sc->sc_ic, ni);
next:
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
rtw_rxdesc_init(rdb, rs, next, 0);
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
}
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
rdb->rdb_next = next;
Miscellaneous changes. Details below. Important changes flagged with []. Using the driver with my Linksys WPC11 ver. 4, it seems to be receiving packets for a change. The WPC11 ver. 4 has a Maxim RF section. My no-name rtw with Philips RF section still does not receive any packets. Keep access-level (analog params > config[0123] registers > none) in sc_access. Add rtw_set_access for changing the access level. Make rtw_continuous_tx_enable and other subroutines use rtw_set_access instead of rtw_config0123_enable and rtw_anaparm_enable. Factor part of the chip-reset code into rtw_chip_reset1. Change the 'struct foo (*bar)[N]'-style arguments to 'struct foo *bar'-style arguments. Consolidate software/hardware Tx/Rx ring setup in rtw_hwring_setup, rtw_swring_setup. Add a new constant, SA2400_OPMODE_DEFAULTS, for the bits that we *always* set in the SA2400 OPMODE register. Factor some code out into rtw_sa2400_calibrate. (Inspired by the Linux driver.) [] When the receiver goes into underrun/overflow state, call a new subroutine, rtw_kick() that stops the Rx/Tx processes, resets the chip, reinitializes the Tx/Rx rings, and restarts Rx/Tx processes. (Inspired by the Linux driver.) [] In rtw_intr_rx, check for too-short packets before calling ieee80211_find_rxnode. I believe this will prevent a repeat of the MCHK exception I saw once on macppc. [] Use seconds-elapased as well as microseconds-elapsed to set the next "due date" for the timeout interrupt. This keeps the driver from programming the timeout to expire too early. [] In rtw_intr, read RTW_ISR at most 10 times, then get out. If the interface is not enabled (RTW_F_ENABLED), then get out. [] In rtw_stop, get out if the interface is not enabled (RTW_F_ENABLED). Block IPL_NET interrupts. Don't read/write any registers if the interface is invalid (RTW_F_INVALID). [] Call rtw_stop in rtw_detach.
2004-12-12 09:37:59 +03:00
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
KASSERT(rdb->rdb_next < rdb->rdb_ndesc);
If defined(RTW_DEBUG), provide a sysctl (hw.rtw.rxbufs_limit) for limiting the number of rx buffers an rtw may allocate. Use this sysctl to test the code that copes with buffer exhaustion. Allocate at most RTW_RXQLEN rx buffers, stopping at the sysctl limit. Record in sc_nrxdesc how many were allocated, and put the end-of-ring indication on sc_rxdesc[sc_nrxdesc - 1]. In rtw_init, if no rx buffers could be allocated, log a complaint, clear IFF_RUNNING, and exit with an error. Many changes to accomodate a short rx ring, mainly of the "add a rx-ring length argument" variety. XXX I really should consolidate all of the rx ring variables in one struct and pass that to the rx-ring subroutines. Bug fix: after calling MCLGET, use the (m->m_flags & M_EXT) idiom to check for success, instead of m != NULL. Bug fix: at the top of rtw_start, if IFF_RUNNING is not set, or IFF_OACTIVE is, get out.
2004-12-29 01:21:15 +03:00
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
return;
}
Add the transmit section of rtw(4): build up and tear down transmit descriptor rings, move packets from the transmit queues to the transmit rings, handle transmission-completed interrupts. My Linksys WPC11 ver. 4 with Maxim RF, generates interrupts like it is successfully transmitting packets. Clearly there are bugs: my G4 Powerbook locks up hard. I will debug tomorrow.
2004-12-19 11:19:25 +03:00
static void
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
rtw_txsoft_release(bus_dma_tag_t dmat, struct ieee80211com *ic,
struct rtw_txsoft *ts)
Add the transmit section of rtw(4): build up and tear down transmit descriptor rings, move packets from the transmit queues to the transmit rings, handle transmission-completed interrupts. My Linksys WPC11 ver. 4 with Maxim RF, generates interrupts like it is successfully transmitting packets. Clearly there are bugs: my G4 Powerbook locks up hard. I will debug tomorrow.
2004-12-19 11:19:25 +03:00
{
struct mbuf *m;
struct ieee80211_node *ni;
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
m = ts->ts_mbuf;
ni = ts->ts_ni;
Change rtw_debug from a debug level to a debug mask. Add a lot of debug flags. From Linux: handle an RTL8180 bug. Sometimes the NIC skips from the middle of the ring to the 0th rx descriptor. Now the driver resynchronizes. Handle a receive descriptor underrun or Rx FIFO overflow condition in the way that the Linux driver does. This kind of seems like overkill, but whatever. Protect rtw_ioctl with splnet(). Do not load a tx descriptor with a buffer shorter than 4 bytes. Handle a transmit timeout less disruptively.
2004-12-25 09:58:37 +03:00
KASSERT(m != NULL);
KASSERT(ni != NULL);
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
ts->ts_mbuf = NULL;
ts->ts_ni = NULL;
Add the transmit section of rtw(4): build up and tear down transmit descriptor rings, move packets from the transmit queues to the transmit rings, handle transmission-completed interrupts. My Linksys WPC11 ver. 4 with Maxim RF, generates interrupts like it is successfully transmitting packets. Clearly there are bugs: my G4 Powerbook locks up hard. I will debug tomorrow.
2004-12-19 11:19:25 +03:00
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
bus_dmamap_sync(dmat, ts->ts_dmamap, 0, ts->ts_dmamap->dm_mapsize,
Add the transmit section of rtw(4): build up and tear down transmit descriptor rings, move packets from the transmit queues to the transmit rings, handle transmission-completed interrupts. My Linksys WPC11 ver. 4 with Maxim RF, generates interrupts like it is successfully transmitting packets. Clearly there are bugs: my G4 Powerbook locks up hard. I will debug tomorrow.
2004-12-19 11:19:25 +03:00
BUS_DMASYNC_POSTWRITE);
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
bus_dmamap_unload(dmat, ts->ts_dmamap);
Add the transmit section of rtw(4): build up and tear down transmit descriptor rings, move packets from the transmit queues to the transmit rings, handle transmission-completed interrupts. My Linksys WPC11 ver. 4 with Maxim RF, generates interrupts like it is successfully transmitting packets. Clearly there are bugs: my G4 Powerbook locks up hard. I will debug tomorrow.
2004-12-19 11:19:25 +03:00
m_freem(m);
ieee80211_release_node(ic, ni);
}
static void
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
rtw_txsofts_release(bus_dma_tag_t dmat, struct ieee80211com *ic,
struct rtw_txsoft_blk *tsb)
Add the transmit section of rtw(4): build up and tear down transmit descriptor rings, move packets from the transmit queues to the transmit rings, handle transmission-completed interrupts. My Linksys WPC11 ver. 4 with Maxim RF, generates interrupts like it is successfully transmitting packets. Clearly there are bugs: my G4 Powerbook locks up hard. I will debug tomorrow.
2004-12-19 11:19:25 +03:00
{
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
struct rtw_txsoft *ts;
Add the transmit section of rtw(4): build up and tear down transmit descriptor rings, move packets from the transmit queues to the transmit rings, handle transmission-completed interrupts. My Linksys WPC11 ver. 4 with Maxim RF, generates interrupts like it is successfully transmitting packets. Clearly there are bugs: my G4 Powerbook locks up hard. I will debug tomorrow.
2004-12-19 11:19:25 +03:00
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
while ((ts = SIMPLEQ_FIRST(&tsb->tsb_dirtyq)) != NULL) {
rtw_txsoft_release(dmat, ic, ts);
SIMPLEQ_REMOVE_HEAD(&tsb->tsb_dirtyq, ts_q);
SIMPLEQ_INSERT_TAIL(&tsb->tsb_freeq, ts, ts_q);
Add the transmit section of rtw(4): build up and tear down transmit descriptor rings, move packets from the transmit queues to the transmit rings, handle transmission-completed interrupts. My Linksys WPC11 ver. 4 with Maxim RF, generates interrupts like it is successfully transmitting packets. Clearly there are bugs: my G4 Powerbook locks up hard. I will debug tomorrow.
2004-12-19 11:19:25 +03:00
}
}
static __inline void
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
rtw_collect_txpkt(struct rtw_softc *sc, struct rtw_txdesc_blk *tdb,
struct rtw_txsoft *ts, int ndesc)
Add the transmit section of rtw(4): build up and tear down transmit descriptor rings, move packets from the transmit queues to the transmit rings, handle transmission-completed interrupts. My Linksys WPC11 ver. 4 with Maxim RF, generates interrupts like it is successfully transmitting packets. Clearly there are bugs: my G4 Powerbook locks up hard. I will debug tomorrow.
2004-12-19 11:19:25 +03:00
{
Extract the transmission status from a tx descriptor, and convert to host byte-order, once.
2004-12-23 08:44:39 +03:00
uint32_t hstat;
Add the transmit section of rtw(4): build up and tear down transmit descriptor rings, move packets from the transmit queues to the transmit rings, handle transmission-completed interrupts. My Linksys WPC11 ver. 4 with Maxim RF, generates interrupts like it is successfully transmitting packets. Clearly there are bugs: my G4 Powerbook locks up hard. I will debug tomorrow.
2004-12-19 11:19:25 +03:00
int data_retry, rts_retry;
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
struct rtw_txdesc *tdn;
Add the transmit section of rtw(4): build up and tear down transmit descriptor rings, move packets from the transmit queues to the transmit rings, handle transmission-completed interrupts. My Linksys WPC11 ver. 4 with Maxim RF, generates interrupts like it is successfully transmitting packets. Clearly there are bugs: my G4 Powerbook locks up hard. I will debug tomorrow.
2004-12-19 11:19:25 +03:00
const char *condstring;
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
rtw_txsoft_release(sc->sc_dmat, &sc->sc_ic, ts);
Add the transmit section of rtw(4): build up and tear down transmit descriptor rings, move packets from the transmit queues to the transmit rings, handle transmission-completed interrupts. My Linksys WPC11 ver. 4 with Maxim RF, generates interrupts like it is successfully transmitting packets. Clearly there are bugs: my G4 Powerbook locks up hard. I will debug tomorrow.
2004-12-19 11:19:25 +03:00
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
tdb->tdb_nfree += ndesc;
Add the transmit section of rtw(4): build up and tear down transmit descriptor rings, move packets from the transmit queues to the transmit rings, handle transmission-completed interrupts. My Linksys WPC11 ver. 4 with Maxim RF, generates interrupts like it is successfully transmitting packets. Clearly there are bugs: my G4 Powerbook locks up hard. I will debug tomorrow.
2004-12-19 11:19:25 +03:00
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
tdn = &tdb->tdb_desc[ts->ts_last];
Add the transmit section of rtw(4): build up and tear down transmit descriptor rings, move packets from the transmit queues to the transmit rings, handle transmission-completed interrupts. My Linksys WPC11 ver. 4 with Maxim RF, generates interrupts like it is successfully transmitting packets. Clearly there are bugs: my G4 Powerbook locks up hard. I will debug tomorrow.
2004-12-19 11:19:25 +03:00
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
hstat = le32toh(tdn->td_stat);
Extract the transmission status from a tx descriptor, and convert to host byte-order, once.
2004-12-23 08:44:39 +03:00
rts_retry = MASK_AND_RSHIFT(hstat, RTW_TXSTAT_RTSRETRY_MASK);
data_retry = MASK_AND_RSHIFT(hstat, RTW_TXSTAT_DRC_MASK);
Add the transmit section of rtw(4): build up and tear down transmit descriptor rings, move packets from the transmit queues to the transmit rings, handle transmission-completed interrupts. My Linksys WPC11 ver. 4 with Maxim RF, generates interrupts like it is successfully transmitting packets. Clearly there are bugs: my G4 Powerbook locks up hard. I will debug tomorrow.
2004-12-19 11:19:25 +03:00
sc->sc_if.if_collisions += rts_retry + data_retry;
Extract the transmission status from a tx descriptor, and convert to host byte-order, once.
2004-12-23 08:44:39 +03:00
if ((hstat & RTW_TXSTAT_TOK) != 0)
Add the transmit section of rtw(4): build up and tear down transmit descriptor rings, move packets from the transmit queues to the transmit rings, handle transmission-completed interrupts. My Linksys WPC11 ver. 4 with Maxim RF, generates interrupts like it is successfully transmitting packets. Clearly there are bugs: my G4 Powerbook locks up hard. I will debug tomorrow.
2004-12-19 11:19:25 +03:00
condstring = "ok";
else {
sc->sc_if.if_oerrors++;
condstring = "error";
}
Change rtw_debug from a debug level to a debug mask. Add a lot of debug flags. From Linux: handle an RTL8180 bug. Sometimes the NIC skips from the middle of the ring to the 0th rx descriptor. Now the driver resynchronizes. Handle a receive descriptor underrun or Rx FIFO overflow condition in the way that the Linux driver does. This kind of seems like overkill, but whatever. Protect rtw_ioctl with splnet(). Do not load a tx descriptor with a buffer shorter than 4 bytes. Handle a transmit timeout less disruptively.
2004-12-25 09:58:37 +03:00
DPRINTF(sc, RTW_DEBUG_XMIT_DESC,
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
("%s: ts %p txdesc[%d, %d] %s tries rts %u data %u\n",
sc->sc_dev.dv_xname, ts, ts->ts_first, ts->ts_last,
Add the transmit section of rtw(4): build up and tear down transmit descriptor rings, move packets from the transmit queues to the transmit rings, handle transmission-completed interrupts. My Linksys WPC11 ver. 4 with Maxim RF, generates interrupts like it is successfully transmitting packets. Clearly there are bugs: my G4 Powerbook locks up hard. I will debug tomorrow.
2004-12-19 11:19:25 +03:00
condstring, rts_retry, data_retry));
}
/* Collect transmitted packets. */
static __inline void
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
rtw_collect_txring(struct rtw_softc *sc, struct rtw_txsoft_blk *tsb,
struct rtw_txdesc_blk *tdb)
Add the transmit section of rtw(4): build up and tear down transmit descriptor rings, move packets from the transmit queues to the transmit rings, handle transmission-completed interrupts. My Linksys WPC11 ver. 4 with Maxim RF, generates interrupts like it is successfully transmitting packets. Clearly there are bugs: my G4 Powerbook locks up hard. I will debug tomorrow.
2004-12-19 11:19:25 +03:00
{
int ndesc;
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
struct rtw_txsoft *ts;
Add the transmit section of rtw(4): build up and tear down transmit descriptor rings, move packets from the transmit queues to the transmit rings, handle transmission-completed interrupts. My Linksys WPC11 ver. 4 with Maxim RF, generates interrupts like it is successfully transmitting packets. Clearly there are bugs: my G4 Powerbook locks up hard. I will debug tomorrow.
2004-12-19 11:19:25 +03:00
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
while ((ts = SIMPLEQ_FIRST(&tsb->tsb_dirtyq)) != NULL) {
ndesc = 1 + ts->ts_last - ts->ts_first;
if (ts->ts_last < ts->ts_first)
ndesc += tdb->tdb_ndesc;
Add the transmit section of rtw(4): build up and tear down transmit descriptor rings, move packets from the transmit queues to the transmit rings, handle transmission-completed interrupts. My Linksys WPC11 ver. 4 with Maxim RF, generates interrupts like it is successfully transmitting packets. Clearly there are bugs: my G4 Powerbook locks up hard. I will debug tomorrow.
2004-12-19 11:19:25 +03:00
Assert that I'm DMA-syncing a non-negative number of transmit descriptors. As I free dirty transmit buffers, move them to the free list, not back to the dirty list! Add a 5-second transmit timeout.
2004-12-20 03:16:21 +03:00
KASSERT(ndesc > 0);
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
rtw_txdescs_sync(tdb, ts->ts_first, ndesc,
Add the transmit section of rtw(4): build up and tear down transmit descriptor rings, move packets from the transmit queues to the transmit rings, handle transmission-completed interrupts. My Linksys WPC11 ver. 4 with Maxim RF, generates interrupts like it is successfully transmitting packets. Clearly there are bugs: my G4 Powerbook locks up hard. I will debug tomorrow.
2004-12-19 11:19:25 +03:00
BUS_DMASYNC_POSTREAD|BUS_DMASYNC_POSTWRITE);
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
if ((tdb->tdb_desc[ts->ts_last].td_stat &
Add the transmit section of rtw(4): build up and tear down transmit descriptor rings, move packets from the transmit queues to the transmit rings, handle transmission-completed interrupts. My Linksys WPC11 ver. 4 with Maxim RF, generates interrupts like it is successfully transmitting packets. Clearly there are bugs: my G4 Powerbook locks up hard. I will debug tomorrow.
2004-12-19 11:19:25 +03:00
htole32(RTW_TXSTAT_OWN)) != 0)
break;
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
rtw_collect_txpkt(sc, tdb, ts, ndesc);
SIMPLEQ_REMOVE_HEAD(&tsb->tsb_dirtyq, ts_q);
SIMPLEQ_INSERT_TAIL(&tsb->tsb_freeq, ts, ts_q);
Add the transmit section of rtw(4): build up and tear down transmit descriptor rings, move packets from the transmit queues to the transmit rings, handle transmission-completed interrupts. My Linksys WPC11 ver. 4 with Maxim RF, generates interrupts like it is successfully transmitting packets. Clearly there are bugs: my G4 Powerbook locks up hard. I will debug tomorrow.
2004-12-19 11:19:25 +03:00
sc->sc_if.if_flags &= ~IFF_OACTIVE;
}
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
if (ts == NULL)
tsb->tsb_tx_timer = 0;
Add the transmit section of rtw(4): build up and tear down transmit descriptor rings, move packets from the transmit queues to the transmit rings, handle transmission-completed interrupts. My Linksys WPC11 ver. 4 with Maxim RF, generates interrupts like it is successfully transmitting packets. Clearly there are bugs: my G4 Powerbook locks up hard. I will debug tomorrow.
2004-12-19 11:19:25 +03:00
}
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
static void
rtw_intr_tx(struct rtw_softc *sc, u_int16_t isr)
{
Add the transmit section of rtw(4): build up and tear down transmit descriptor rings, move packets from the transmit queues to the transmit rings, handle transmission-completed interrupts. My Linksys WPC11 ver. 4 with Maxim RF, generates interrupts like it is successfully transmitting packets. Clearly there are bugs: my G4 Powerbook locks up hard. I will debug tomorrow.
2004-12-19 11:19:25 +03:00
int pri;
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
struct rtw_txsoft_blk *tsb;
struct rtw_txdesc_blk *tdb;
Add the transmit section of rtw(4): build up and tear down transmit descriptor rings, move packets from the transmit queues to the transmit rings, handle transmission-completed interrupts. My Linksys WPC11 ver. 4 with Maxim RF, generates interrupts like it is successfully transmitting packets. Clearly there are bugs: my G4 Powerbook locks up hard. I will debug tomorrow.
2004-12-19 11:19:25 +03:00
for (pri = 0; pri < RTW_NTXPRI; pri++) {
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
tsb = &sc->sc_txsoft_blk[pri];
tdb = &sc->sc_txdesc_blk[pri];
Add the transmit section of rtw(4): build up and tear down transmit descriptor rings, move packets from the transmit queues to the transmit rings, handle transmission-completed interrupts. My Linksys WPC11 ver. 4 with Maxim RF, generates interrupts like it is successfully transmitting packets. Clearly there are bugs: my G4 Powerbook locks up hard. I will debug tomorrow.
2004-12-19 11:19:25 +03:00
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
rtw_collect_txring(sc, tsb, tdb);
Add the transmit section of rtw(4): build up and tear down transmit descriptor rings, move packets from the transmit queues to the transmit rings, handle transmission-completed interrupts. My Linksys WPC11 ver. 4 with Maxim RF, generates interrupts like it is successfully transmitting packets. Clearly there are bugs: my G4 Powerbook locks up hard. I will debug tomorrow.
2004-12-19 11:19:25 +03:00
Change rtw_debug from a debug level to a debug mask. Add a lot of debug flags. From Linux: handle an RTL8180 bug. Sometimes the NIC skips from the middle of the ring to the 0th rx descriptor. Now the driver resynchronizes. Handle a receive descriptor underrun or Rx FIFO overflow condition in the way that the Linux driver does. This kind of seems like overkill, but whatever. Protect rtw_ioctl with splnet(). Do not load a tx descriptor with a buffer shorter than 4 bytes. Handle a transmit timeout less disruptively.
2004-12-25 09:58:37 +03:00
if ((isr & RTW_INTR_TX) != 0)
rtw_start(&sc->sc_if);
Add the transmit section of rtw(4): build up and tear down transmit descriptor rings, move packets from the transmit queues to the transmit rings, handle transmission-completed interrupts. My Linksys WPC11 ver. 4 with Maxim RF, generates interrupts like it is successfully transmitting packets. Clearly there are bugs: my G4 Powerbook locks up hard. I will debug tomorrow.
2004-12-19 11:19:25 +03:00
}
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
/* TBD */
return;
}
static void
rtw_intr_beacon(struct rtw_softc *sc, u_int16_t isr)
{
/* TBD */
return;
}
static void
rtw_intr_atim(struct rtw_softc *sc)
{
/* TBD */
return;
}
Change rtw_debug from a debug level to a debug mask. Add a lot of debug flags. From Linux: handle an RTL8180 bug. Sometimes the NIC skips from the middle of the ring to the 0th rx descriptor. Now the driver resynchronizes. Handle a receive descriptor underrun or Rx FIFO overflow condition in the way that the Linux driver does. This kind of seems like overkill, but whatever. Protect rtw_ioctl with splnet(). Do not load a tx descriptor with a buffer shorter than 4 bytes. Handle a transmit timeout less disruptively.
2004-12-25 09:58:37 +03:00
#ifdef RTW_DEBUG
static void
rtw_dump_rings(struct rtw_softc *sc)
{
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
struct rtw_txdesc_blk *tdb;
struct rtw_rxdesc *rd;
struct rtw_rxdesc_blk *rdb;
Change rtw_debug from a debug level to a debug mask. Add a lot of debug flags. From Linux: handle an RTL8180 bug. Sometimes the NIC skips from the middle of the ring to the 0th rx descriptor. Now the driver resynchronizes. Handle a receive descriptor underrun or Rx FIFO overflow condition in the way that the Linux driver does. This kind of seems like overkill, but whatever. Protect rtw_ioctl with splnet(). Do not load a tx descriptor with a buffer shorter than 4 bytes. Handle a transmit timeout less disruptively.
2004-12-25 09:58:37 +03:00
int desc, pri;
if ((rtw_debug & RTW_DEBUG_IO_KICK) == 0)
return;
for (pri = 0; pri < RTW_NTXPRI; pri++) {
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
tdb = &sc->sc_txdesc_blk[pri];
Change rtw_debug from a debug level to a debug mask. Add a lot of debug flags. From Linux: handle an RTL8180 bug. Sometimes the NIC skips from the middle of the ring to the 0th rx descriptor. Now the driver resynchronizes. Handle a receive descriptor underrun or Rx FIFO overflow condition in the way that the Linux driver does. This kind of seems like overkill, but whatever. Protect rtw_ioctl with splnet(). Do not load a tx descriptor with a buffer shorter than 4 bytes. Handle a transmit timeout less disruptively.
2004-12-25 09:58:37 +03:00
printf("%s: txpri %d ndesc %d nfree %d\n", __func__, pri,
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
tdb->tdb_ndesc, tdb->tdb_nfree);
for (desc = 0; desc < tdb->tdb_ndesc; desc++)
rtw_print_txdesc(sc, ".", NULL, tdb, desc);
Change rtw_debug from a debug level to a debug mask. Add a lot of debug flags. From Linux: handle an RTL8180 bug. Sometimes the NIC skips from the middle of the ring to the 0th rx descriptor. Now the driver resynchronizes. Handle a receive descriptor underrun or Rx FIFO overflow condition in the way that the Linux driver does. This kind of seems like overkill, but whatever. Protect rtw_ioctl with splnet(). Do not load a tx descriptor with a buffer shorter than 4 bytes. Handle a transmit timeout less disruptively.
2004-12-25 09:58:37 +03:00
}
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
rdb = &sc->sc_rxdesc_blk;
Consolidate variables related to the rx ring in sc_rxdesc_blk, which is a struct rtw_rxdesc_blk. Put a copy of the DMA tag and the DMA map into the rx- and tx-ring blocks so that I don't have to pass them to subroutines all of the time.
2004-12-29 03:21:37 +03:00
Change rtw_debug from a debug level to a debug mask. Add a lot of debug flags. From Linux: handle an RTL8180 bug. Sometimes the NIC skips from the middle of the ring to the 0th rx descriptor. Now the driver resynchronizes. Handle a receive descriptor underrun or Rx FIFO overflow condition in the way that the Linux driver does. This kind of seems like overkill, but whatever. Protect rtw_ioctl with splnet(). Do not load a tx descriptor with a buffer shorter than 4 bytes. Handle a transmit timeout less disruptively.
2004-12-25 09:58:37 +03:00
for (desc = 0; desc < RTW_RXQLEN; desc++) {
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
rd = &rdb->rdb_desc[desc];
If defined(RTW_DEBUG), provide a sysctl (hw.rtw.rxbufs_limit) for limiting the number of rx buffers an rtw may allocate. Use this sysctl to test the code that copes with buffer exhaustion. Allocate at most RTW_RXQLEN rx buffers, stopping at the sysctl limit. Record in sc_nrxdesc how many were allocated, and put the end-of-ring indication on sc_rxdesc[sc_nrxdesc - 1]. In rtw_init, if no rx buffers could be allocated, log a complaint, clear IFF_RUNNING, and exit with an error. Many changes to accomodate a short rx ring, mainly of the "add a rx-ring length argument" variety. XXX I really should consolidate all of the rx ring variables in one struct and pass that to the rx-ring subroutines. Bug fix: after calling MCLGET, use the (m->m_flags & M_EXT) idiom to check for success, instead of m != NULL. Bug fix: at the top of rtw_start, if IFF_RUNNING is not set, or IFF_OACTIVE is, get out.
2004-12-29 01:21:15 +03:00
printf("%s: %sctl %08x rsvd0/rssi %08x buf/tsftl %08x "
Change rtw_debug from a debug level to a debug mask. Add a lot of debug flags. From Linux: handle an RTL8180 bug. Sometimes the NIC skips from the middle of the ring to the 0th rx descriptor. Now the driver resynchronizes. Handle a receive descriptor underrun or Rx FIFO overflow condition in the way that the Linux driver does. This kind of seems like overkill, but whatever. Protect rtw_ioctl with splnet(). Do not load a tx descriptor with a buffer shorter than 4 bytes. Handle a transmit timeout less disruptively.
2004-12-25 09:58:37 +03:00
"rsvd1/tsfth %08x\n", __func__,
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
(desc >= rdb->rdb_ndesc) ? "UNUSED " : "",
le32toh(rd->rd_ctl), le32toh(rd->rd_rssi),
le32toh(rd->rd_buf), le32toh(rd->rd_tsfth));
Change rtw_debug from a debug level to a debug mask. Add a lot of debug flags. From Linux: handle an RTL8180 bug. Sometimes the NIC skips from the middle of the ring to the 0th rx descriptor. Now the driver resynchronizes. Handle a receive descriptor underrun or Rx FIFO overflow condition in the way that the Linux driver does. This kind of seems like overkill, but whatever. Protect rtw_ioctl with splnet(). Do not load a tx descriptor with a buffer shorter than 4 bytes. Handle a transmit timeout less disruptively.
2004-12-25 09:58:37 +03:00
}
}
#endif /* RTW_DEBUG */
Miscellaneous changes. Details below. Important changes flagged with []. Using the driver with my Linksys WPC11 ver. 4, it seems to be receiving packets for a change. The WPC11 ver. 4 has a Maxim RF section. My no-name rtw with Philips RF section still does not receive any packets. Keep access-level (analog params > config[0123] registers > none) in sc_access. Add rtw_set_access for changing the access level. Make rtw_continuous_tx_enable and other subroutines use rtw_set_access instead of rtw_config0123_enable and rtw_anaparm_enable. Factor part of the chip-reset code into rtw_chip_reset1. Change the 'struct foo (*bar)[N]'-style arguments to 'struct foo *bar'-style arguments. Consolidate software/hardware Tx/Rx ring setup in rtw_hwring_setup, rtw_swring_setup. Add a new constant, SA2400_OPMODE_DEFAULTS, for the bits that we *always* set in the SA2400 OPMODE register. Factor some code out into rtw_sa2400_calibrate. (Inspired by the Linux driver.) [] When the receiver goes into underrun/overflow state, call a new subroutine, rtw_kick() that stops the Rx/Tx processes, resets the chip, reinitializes the Tx/Rx rings, and restarts Rx/Tx processes. (Inspired by the Linux driver.) [] In rtw_intr_rx, check for too-short packets before calling ieee80211_find_rxnode. I believe this will prevent a repeat of the MCHK exception I saw once on macppc. [] Use seconds-elapased as well as microseconds-elapsed to set the next "due date" for the timeout interrupt. This keeps the driver from programming the timeout to expire too early. [] In rtw_intr, read RTW_ISR at most 10 times, then get out. If the interface is not enabled (RTW_F_ENABLED), then get out. [] In rtw_stop, get out if the interface is not enabled (RTW_F_ENABLED). Block IPL_NET interrupts. Don't read/write any registers if the interface is invalid (RTW_F_INVALID). [] Call rtw_stop in rtw_detach.
2004-12-12 09:37:59 +03:00
static void
rtw_hwring_setup(struct rtw_softc *sc)
{
struct rtw_regs *regs = &sc->sc_regs;
RTW_WRITE(regs, RTW_RDSAR, RTW_RING_BASE(sc, hd_rx));
RTW_WRITE(regs, RTW_TLPDA, RTW_RING_BASE(sc, hd_txlo));
RTW_WRITE(regs, RTW_TNPDA, RTW_RING_BASE(sc, hd_txmd));
RTW_WRITE(regs, RTW_THPDA, RTW_RING_BASE(sc, hd_txhi));
RTW_WRITE(regs, RTW_TBDA, RTW_RING_BASE(sc, hd_bcn));
Add some bus barriers. Remark where they are needed, e.g., read_eeprom. Change some printfs to RTW_DPRINTF2s to quiet rtw(4) when debugging is disabled.
2004-12-20 04:13:45 +03:00
RTW_SYNC(regs, RTW_TLPDA, RTW_RDSAR);
Change rtw_debug from a debug level to a debug mask. Add a lot of debug flags. From Linux: handle an RTL8180 bug. Sometimes the NIC skips from the middle of the ring to the 0th rx descriptor. Now the driver resynchronizes. Handle a receive descriptor underrun or Rx FIFO overflow condition in the way that the Linux driver does. This kind of seems like overkill, but whatever. Protect rtw_ioctl with splnet(). Do not load a tx descriptor with a buffer shorter than 4 bytes. Handle a transmit timeout less disruptively.
2004-12-25 09:58:37 +03:00
RTW_DPRINTF(RTW_DEBUG_XMIT_DESC,
("%s: reg[TLPDA] <- %" PRIxPTR "\n", __func__,
(uintptr_t)RTW_RING_BASE(sc, hd_txlo)));
RTW_DPRINTF(RTW_DEBUG_XMIT_DESC,
("%s: reg[TNPDA] <- %" PRIxPTR "\n", __func__,
(uintptr_t)RTW_RING_BASE(sc, hd_txmd)));
RTW_DPRINTF(RTW_DEBUG_XMIT_DESC,
("%s: reg[THPDA] <- %" PRIxPTR "\n", __func__,
(uintptr_t)RTW_RING_BASE(sc, hd_txhi)));
RTW_DPRINTF(RTW_DEBUG_XMIT_DESC,
("%s: reg[TBDA] <- %" PRIxPTR "\n", __func__,
(uintptr_t)RTW_RING_BASE(sc, hd_bcn)));
RTW_DPRINTF(RTW_DEBUG_RECV_DESC,
("%s: reg[RDSAR] <- %" PRIxPTR "\n", __func__,
(uintptr_t)RTW_RING_BASE(sc, hd_rx)));
Miscellaneous changes. Details below. Important changes flagged with []. Using the driver with my Linksys WPC11 ver. 4, it seems to be receiving packets for a change. The WPC11 ver. 4 has a Maxim RF section. My no-name rtw with Philips RF section still does not receive any packets. Keep access-level (analog params > config[0123] registers > none) in sc_access. Add rtw_set_access for changing the access level. Make rtw_continuous_tx_enable and other subroutines use rtw_set_access instead of rtw_config0123_enable and rtw_anaparm_enable. Factor part of the chip-reset code into rtw_chip_reset1. Change the 'struct foo (*bar)[N]'-style arguments to 'struct foo *bar'-style arguments. Consolidate software/hardware Tx/Rx ring setup in rtw_hwring_setup, rtw_swring_setup. Add a new constant, SA2400_OPMODE_DEFAULTS, for the bits that we *always* set in the SA2400 OPMODE register. Factor some code out into rtw_sa2400_calibrate. (Inspired by the Linux driver.) [] When the receiver goes into underrun/overflow state, call a new subroutine, rtw_kick() that stops the Rx/Tx processes, resets the chip, reinitializes the Tx/Rx rings, and restarts Rx/Tx processes. (Inspired by the Linux driver.) [] In rtw_intr_rx, check for too-short packets before calling ieee80211_find_rxnode. I believe this will prevent a repeat of the MCHK exception I saw once on macppc. [] Use seconds-elapased as well as microseconds-elapsed to set the next "due date" for the timeout interrupt. This keeps the driver from programming the timeout to expire too early. [] In rtw_intr, read RTW_ISR at most 10 times, then get out. If the interface is not enabled (RTW_F_ENABLED), then get out. [] In rtw_stop, get out if the interface is not enabled (RTW_F_ENABLED). Block IPL_NET interrupts. Don't read/write any registers if the interface is invalid (RTW_F_INVALID). [] Call rtw_stop in rtw_detach.
2004-12-12 09:37:59 +03:00
}
If defined(RTW_DEBUG), provide a sysctl (hw.rtw.rxbufs_limit) for limiting the number of rx buffers an rtw may allocate. Use this sysctl to test the code that copes with buffer exhaustion. Allocate at most RTW_RXQLEN rx buffers, stopping at the sysctl limit. Record in sc_nrxdesc how many were allocated, and put the end-of-ring indication on sc_rxdesc[sc_nrxdesc - 1]. In rtw_init, if no rx buffers could be allocated, log a complaint, clear IFF_RUNNING, and exit with an error. Many changes to accomodate a short rx ring, mainly of the "add a rx-ring length argument" variety. XXX I really should consolidate all of the rx ring variables in one struct and pass that to the rx-ring subroutines. Bug fix: after calling MCLGET, use the (m->m_flags & M_EXT) idiom to check for success, instead of m != NULL. Bug fix: at the top of rtw_start, if IFF_RUNNING is not set, or IFF_OACTIVE is, get out.
2004-12-29 01:21:15 +03:00
static int
Miscellaneous changes. Details below. Important changes flagged with []. Using the driver with my Linksys WPC11 ver. 4, it seems to be receiving packets for a change. The WPC11 ver. 4 has a Maxim RF section. My no-name rtw with Philips RF section still does not receive any packets. Keep access-level (analog params > config[0123] registers > none) in sc_access. Add rtw_set_access for changing the access level. Make rtw_continuous_tx_enable and other subroutines use rtw_set_access instead of rtw_config0123_enable and rtw_anaparm_enable. Factor part of the chip-reset code into rtw_chip_reset1. Change the 'struct foo (*bar)[N]'-style arguments to 'struct foo *bar'-style arguments. Consolidate software/hardware Tx/Rx ring setup in rtw_hwring_setup, rtw_swring_setup. Add a new constant, SA2400_OPMODE_DEFAULTS, for the bits that we *always* set in the SA2400 OPMODE register. Factor some code out into rtw_sa2400_calibrate. (Inspired by the Linux driver.) [] When the receiver goes into underrun/overflow state, call a new subroutine, rtw_kick() that stops the Rx/Tx processes, resets the chip, reinitializes the Tx/Rx rings, and restarts Rx/Tx processes. (Inspired by the Linux driver.) [] In rtw_intr_rx, check for too-short packets before calling ieee80211_find_rxnode. I believe this will prevent a repeat of the MCHK exception I saw once on macppc. [] Use seconds-elapased as well as microseconds-elapsed to set the next "due date" for the timeout interrupt. This keeps the driver from programming the timeout to expire too early. [] In rtw_intr, read RTW_ISR at most 10 times, then get out. If the interface is not enabled (RTW_F_ENABLED), then get out. [] In rtw_stop, get out if the interface is not enabled (RTW_F_ENABLED). Block IPL_NET interrupts. Don't read/write any registers if the interface is invalid (RTW_F_INVALID). [] Call rtw_stop in rtw_detach.
2004-12-12 09:37:59 +03:00
rtw_swring_setup(struct rtw_softc *sc)
{
If defined(RTW_DEBUG), provide a sysctl (hw.rtw.rxbufs_limit) for limiting the number of rx buffers an rtw may allocate. Use this sysctl to test the code that copes with buffer exhaustion. Allocate at most RTW_RXQLEN rx buffers, stopping at the sysctl limit. Record in sc_nrxdesc how many were allocated, and put the end-of-ring indication on sc_rxdesc[sc_nrxdesc - 1]. In rtw_init, if no rx buffers could be allocated, log a complaint, clear IFF_RUNNING, and exit with an error. Many changes to accomodate a short rx ring, mainly of the "add a rx-ring length argument" variety. XXX I really should consolidate all of the rx ring variables in one struct and pass that to the rx-ring subroutines. Bug fix: after calling MCLGET, use the (m->m_flags & M_EXT) idiom to check for success, instead of m != NULL. Bug fix: at the top of rtw_start, if IFF_RUNNING is not set, or IFF_OACTIVE is, get out.
2004-12-29 01:21:15 +03:00
int rc;
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
struct rtw_rxdesc_blk *rdb;
Consolidate variables related to the rx ring in sc_rxdesc_blk, which is a struct rtw_rxdesc_blk. Put a copy of the DMA tag and the DMA map into the rx- and tx-ring blocks so that I don't have to pass them to subroutines all of the time.
2004-12-29 03:21:37 +03:00
Miscellaneous changes. Details below. Important changes flagged with []. Using the driver with my Linksys WPC11 ver. 4, it seems to be receiving packets for a change. The WPC11 ver. 4 has a Maxim RF section. My no-name rtw with Philips RF section still does not receive any packets. Keep access-level (analog params > config[0123] registers > none) in sc_access. Add rtw_set_access for changing the access level. Make rtw_continuous_tx_enable and other subroutines use rtw_set_access instead of rtw_config0123_enable and rtw_anaparm_enable. Factor part of the chip-reset code into rtw_chip_reset1. Change the 'struct foo (*bar)[N]'-style arguments to 'struct foo *bar'-style arguments. Consolidate software/hardware Tx/Rx ring setup in rtw_hwring_setup, rtw_swring_setup. Add a new constant, SA2400_OPMODE_DEFAULTS, for the bits that we *always* set in the SA2400 OPMODE register. Factor some code out into rtw_sa2400_calibrate. (Inspired by the Linux driver.) [] When the receiver goes into underrun/overflow state, call a new subroutine, rtw_kick() that stops the Rx/Tx processes, resets the chip, reinitializes the Tx/Rx rings, and restarts Rx/Tx processes. (Inspired by the Linux driver.) [] In rtw_intr_rx, check for too-short packets before calling ieee80211_find_rxnode. I believe this will prevent a repeat of the MCHK exception I saw once on macppc. [] Use seconds-elapased as well as microseconds-elapsed to set the next "due date" for the timeout interrupt. This keeps the driver from programming the timeout to expire too early. [] In rtw_intr, read RTW_ISR at most 10 times, then get out. If the interface is not enabled (RTW_F_ENABLED), then get out. [] In rtw_stop, get out if the interface is not enabled (RTW_F_ENABLED). Block IPL_NET interrupts. Don't read/write any registers if the interface is invalid (RTW_F_INVALID). [] Call rtw_stop in rtw_detach.
2004-12-12 09:37:59 +03:00
rtw_txdesc_blk_init_all(&sc->sc_txdesc_blk[0]);
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
rtw_txsoft_blk_init_all(&sc->sc_txsoft_blk[0]);
Miscellaneous changes. Details below. Important changes flagged with []. Using the driver with my Linksys WPC11 ver. 4, it seems to be receiving packets for a change. The WPC11 ver. 4 has a Maxim RF section. My no-name rtw with Philips RF section still does not receive any packets. Keep access-level (analog params > config[0123] registers > none) in sc_access. Add rtw_set_access for changing the access level. Make rtw_continuous_tx_enable and other subroutines use rtw_set_access instead of rtw_config0123_enable and rtw_anaparm_enable. Factor part of the chip-reset code into rtw_chip_reset1. Change the 'struct foo (*bar)[N]'-style arguments to 'struct foo *bar'-style arguments. Consolidate software/hardware Tx/Rx ring setup in rtw_hwring_setup, rtw_swring_setup. Add a new constant, SA2400_OPMODE_DEFAULTS, for the bits that we *always* set in the SA2400 OPMODE register. Factor some code out into rtw_sa2400_calibrate. (Inspired by the Linux driver.) [] When the receiver goes into underrun/overflow state, call a new subroutine, rtw_kick() that stops the Rx/Tx processes, resets the chip, reinitializes the Tx/Rx rings, and restarts Rx/Tx processes. (Inspired by the Linux driver.) [] In rtw_intr_rx, check for too-short packets before calling ieee80211_find_rxnode. I believe this will prevent a repeat of the MCHK exception I saw once on macppc. [] Use seconds-elapased as well as microseconds-elapsed to set the next "due date" for the timeout interrupt. This keeps the driver from programming the timeout to expire too early. [] In rtw_intr, read RTW_ISR at most 10 times, then get out. If the interface is not enabled (RTW_F_ENABLED), then get out. [] In rtw_stop, get out if the interface is not enabled (RTW_F_ENABLED). Block IPL_NET interrupts. Don't read/write any registers if the interface is invalid (RTW_F_INVALID). [] Call rtw_stop in rtw_detach.
2004-12-12 09:37:59 +03:00
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
rdb = &sc->sc_rxdesc_blk;
if ((rc = rtw_rxsoft_init_all(sc->sc_dmat, sc->sc_rxsoft, &rdb->rdb_ndesc,
sc->sc_dev.dv_xname)) != 0 && rdb->rdb_ndesc == 0) {
If defined(RTW_DEBUG), provide a sysctl (hw.rtw.rxbufs_limit) for limiting the number of rx buffers an rtw may allocate. Use this sysctl to test the code that copes with buffer exhaustion. Allocate at most RTW_RXQLEN rx buffers, stopping at the sysctl limit. Record in sc_nrxdesc how many were allocated, and put the end-of-ring indication on sc_rxdesc[sc_nrxdesc - 1]. In rtw_init, if no rx buffers could be allocated, log a complaint, clear IFF_RUNNING, and exit with an error. Many changes to accomodate a short rx ring, mainly of the "add a rx-ring length argument" variety. XXX I really should consolidate all of the rx ring variables in one struct and pass that to the rx-ring subroutines. Bug fix: after calling MCLGET, use the (m->m_flags & M_EXT) idiom to check for success, instead of m != NULL. Bug fix: at the top of rtw_start, if IFF_RUNNING is not set, or IFF_OACTIVE is, get out.
2004-12-29 01:21:15 +03:00
printf("%s: could not allocate rx buffers\n",
sc->sc_dev.dv_xname);
return rc;
}
Consolidate variables related to the rx ring in sc_rxdesc_blk, which is a struct rtw_rxdesc_blk. Put a copy of the DMA tag and the DMA map into the rx- and tx-ring blocks so that I don't have to pass them to subroutines all of the time.
2004-12-29 03:21:37 +03:00
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
rdb = &sc->sc_rxdesc_blk;
rtw_rxdescs_sync(rdb, 0, rdb->rdb_ndesc,
If defined(RTW_DEBUG), provide a sysctl (hw.rtw.rxbufs_limit) for limiting the number of rx buffers an rtw may allocate. Use this sysctl to test the code that copes with buffer exhaustion. Allocate at most RTW_RXQLEN rx buffers, stopping at the sysctl limit. Record in sc_nrxdesc how many were allocated, and put the end-of-ring indication on sc_rxdesc[sc_nrxdesc - 1]. In rtw_init, if no rx buffers could be allocated, log a complaint, clear IFF_RUNNING, and exit with an error. Many changes to accomodate a short rx ring, mainly of the "add a rx-ring length argument" variety. XXX I really should consolidate all of the rx ring variables in one struct and pass that to the rx-ring subroutines. Bug fix: after calling MCLGET, use the (m->m_flags & M_EXT) idiom to check for success, instead of m != NULL. Bug fix: at the top of rtw_start, if IFF_RUNNING is not set, or IFF_OACTIVE is, get out.
2004-12-29 01:21:15 +03:00
BUS_DMASYNC_POSTREAD|BUS_DMASYNC_POSTWRITE);
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
rtw_rxdesc_init_all(rdb, sc->sc_rxsoft, 1);
Miscellaneous changes. Details below. Important changes flagged with []. Using the driver with my Linksys WPC11 ver. 4, it seems to be receiving packets for a change. The WPC11 ver. 4 has a Maxim RF section. My no-name rtw with Philips RF section still does not receive any packets. Keep access-level (analog params > config[0123] registers > none) in sc_access. Add rtw_set_access for changing the access level. Make rtw_continuous_tx_enable and other subroutines use rtw_set_access instead of rtw_config0123_enable and rtw_anaparm_enable. Factor part of the chip-reset code into rtw_chip_reset1. Change the 'struct foo (*bar)[N]'-style arguments to 'struct foo *bar'-style arguments. Consolidate software/hardware Tx/Rx ring setup in rtw_hwring_setup, rtw_swring_setup. Add a new constant, SA2400_OPMODE_DEFAULTS, for the bits that we *always* set in the SA2400 OPMODE register. Factor some code out into rtw_sa2400_calibrate. (Inspired by the Linux driver.) [] When the receiver goes into underrun/overflow state, call a new subroutine, rtw_kick() that stops the Rx/Tx processes, resets the chip, reinitializes the Tx/Rx rings, and restarts Rx/Tx processes. (Inspired by the Linux driver.) [] In rtw_intr_rx, check for too-short packets before calling ieee80211_find_rxnode. I believe this will prevent a repeat of the MCHK exception I saw once on macppc. [] Use seconds-elapased as well as microseconds-elapsed to set the next "due date" for the timeout interrupt. This keeps the driver from programming the timeout to expire too early. [] In rtw_intr, read RTW_ISR at most 10 times, then get out. If the interface is not enabled (RTW_F_ENABLED), then get out. [] In rtw_stop, get out if the interface is not enabled (RTW_F_ENABLED). Block IPL_NET interrupts. Don't read/write any registers if the interface is invalid (RTW_F_INVALID). [] Call rtw_stop in rtw_detach.
2004-12-12 09:37:59 +03:00
Consolidate variables related to the rx ring in sc_rxdesc_blk, which is a struct rtw_rxdesc_blk. Put a copy of the DMA tag and the DMA map into the rx- and tx-ring blocks so that I don't have to pass them to subroutines all of the time.
2004-12-29 03:21:37 +03:00
rtw_txdescs_sync_all(&sc->sc_txdesc_blk[0]);
If defined(RTW_DEBUG), provide a sysctl (hw.rtw.rxbufs_limit) for limiting the number of rx buffers an rtw may allocate. Use this sysctl to test the code that copes with buffer exhaustion. Allocate at most RTW_RXQLEN rx buffers, stopping at the sysctl limit. Record in sc_nrxdesc how many were allocated, and put the end-of-ring indication on sc_rxdesc[sc_nrxdesc - 1]. In rtw_init, if no rx buffers could be allocated, log a complaint, clear IFF_RUNNING, and exit with an error. Many changes to accomodate a short rx ring, mainly of the "add a rx-ring length argument" variety. XXX I really should consolidate all of the rx ring variables in one struct and pass that to the rx-ring subroutines. Bug fix: after calling MCLGET, use the (m->m_flags & M_EXT) idiom to check for success, instead of m != NULL. Bug fix: at the top of rtw_start, if IFF_RUNNING is not set, or IFF_OACTIVE is, get out.
2004-12-29 01:21:15 +03:00
return 0;
Miscellaneous changes. Details below. Important changes flagged with []. Using the driver with my Linksys WPC11 ver. 4, it seems to be receiving packets for a change. The WPC11 ver. 4 has a Maxim RF section. My no-name rtw with Philips RF section still does not receive any packets. Keep access-level (analog params > config[0123] registers > none) in sc_access. Add rtw_set_access for changing the access level. Make rtw_continuous_tx_enable and other subroutines use rtw_set_access instead of rtw_config0123_enable and rtw_anaparm_enable. Factor part of the chip-reset code into rtw_chip_reset1. Change the 'struct foo (*bar)[N]'-style arguments to 'struct foo *bar'-style arguments. Consolidate software/hardware Tx/Rx ring setup in rtw_hwring_setup, rtw_swring_setup. Add a new constant, SA2400_OPMODE_DEFAULTS, for the bits that we *always* set in the SA2400 OPMODE register. Factor some code out into rtw_sa2400_calibrate. (Inspired by the Linux driver.) [] When the receiver goes into underrun/overflow state, call a new subroutine, rtw_kick() that stops the Rx/Tx processes, resets the chip, reinitializes the Tx/Rx rings, and restarts Rx/Tx processes. (Inspired by the Linux driver.) [] In rtw_intr_rx, check for too-short packets before calling ieee80211_find_rxnode. I believe this will prevent a repeat of the MCHK exception I saw once on macppc. [] Use seconds-elapased as well as microseconds-elapsed to set the next "due date" for the timeout interrupt. This keeps the driver from programming the timeout to expire too early. [] In rtw_intr, read RTW_ISR at most 10 times, then get out. If the interface is not enabled (RTW_F_ENABLED), then get out. [] In rtw_stop, get out if the interface is not enabled (RTW_F_ENABLED). Block IPL_NET interrupts. Don't read/write any registers if the interface is invalid (RTW_F_INVALID). [] Call rtw_stop in rtw_detach.
2004-12-12 09:37:59 +03:00
}
static void
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
rtw_txdesc_blk_reset(struct rtw_txdesc_blk *tdb)
Change rtw_debug from a debug level to a debug mask. Add a lot of debug flags. From Linux: handle an RTL8180 bug. Sometimes the NIC skips from the middle of the ring to the 0th rx descriptor. Now the driver resynchronizes. Handle a receive descriptor underrun or Rx FIFO overflow condition in the way that the Linux driver does. This kind of seems like overkill, but whatever. Protect rtw_ioctl with splnet(). Do not load a tx descriptor with a buffer shorter than 4 bytes. Handle a transmit timeout less disruptively.
2004-12-25 09:58:37 +03:00
{
int i;
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
(void)memset(tdb->tdb_desc, 0,
sizeof(tdb->tdb_desc[0]) * tdb->tdb_ndesc);
for (i = 0; i < tdb->tdb_ndesc; i++)
tdb->tdb_desc[i].td_next = htole32(RTW_NEXT_DESC(tdb, i));
tdb->tdb_nfree = tdb->tdb_ndesc;
tdb->tdb_next = 0;
Change rtw_debug from a debug level to a debug mask. Add a lot of debug flags. From Linux: handle an RTL8180 bug. Sometimes the NIC skips from the middle of the ring to the 0th rx descriptor. Now the driver resynchronizes. Handle a receive descriptor underrun or Rx FIFO overflow condition in the way that the Linux driver does. This kind of seems like overkill, but whatever. Protect rtw_ioctl with splnet(). Do not load a tx descriptor with a buffer shorter than 4 bytes. Handle a transmit timeout less disruptively.
2004-12-25 09:58:37 +03:00
}
static void
rtw_txdescs_reset(struct rtw_softc *sc)
Miscellaneous changes. Details below. Important changes flagged with []. Using the driver with my Linksys WPC11 ver. 4, it seems to be receiving packets for a change. The WPC11 ver. 4 has a Maxim RF section. My no-name rtw with Philips RF section still does not receive any packets. Keep access-level (analog params > config[0123] registers > none) in sc_access. Add rtw_set_access for changing the access level. Make rtw_continuous_tx_enable and other subroutines use rtw_set_access instead of rtw_config0123_enable and rtw_anaparm_enable. Factor part of the chip-reset code into rtw_chip_reset1. Change the 'struct foo (*bar)[N]'-style arguments to 'struct foo *bar'-style arguments. Consolidate software/hardware Tx/Rx ring setup in rtw_hwring_setup, rtw_swring_setup. Add a new constant, SA2400_OPMODE_DEFAULTS, for the bits that we *always* set in the SA2400 OPMODE register. Factor some code out into rtw_sa2400_calibrate. (Inspired by the Linux driver.) [] When the receiver goes into underrun/overflow state, call a new subroutine, rtw_kick() that stops the Rx/Tx processes, resets the chip, reinitializes the Tx/Rx rings, and restarts Rx/Tx processes. (Inspired by the Linux driver.) [] In rtw_intr_rx, check for too-short packets before calling ieee80211_find_rxnode. I believe this will prevent a repeat of the MCHK exception I saw once on macppc. [] Use seconds-elapased as well as microseconds-elapsed to set the next "due date" for the timeout interrupt. This keeps the driver from programming the timeout to expire too early. [] In rtw_intr, read RTW_ISR at most 10 times, then get out. If the interface is not enabled (RTW_F_ENABLED), then get out. [] In rtw_stop, get out if the interface is not enabled (RTW_F_ENABLED). Block IPL_NET interrupts. Don't read/write any registers if the interface is invalid (RTW_F_INVALID). [] Call rtw_stop in rtw_detach.
2004-12-12 09:37:59 +03:00
{
Add the transmit section of rtw(4): build up and tear down transmit descriptor rings, move packets from the transmit queues to the transmit rings, handle transmission-completed interrupts. My Linksys WPC11 ver. 4 with Maxim RF, generates interrupts like it is successfully transmitting packets. Clearly there are bugs: my G4 Powerbook locks up hard. I will debug tomorrow.
2004-12-19 11:19:25 +03:00
int pri;
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
struct rtw_txdesc_blk *tdb;
Change rtw_debug from a debug level to a debug mask. Add a lot of debug flags. From Linux: handle an RTL8180 bug. Sometimes the NIC skips from the middle of the ring to the 0th rx descriptor. Now the driver resynchronizes. Handle a receive descriptor underrun or Rx FIFO overflow condition in the way that the Linux driver does. This kind of seems like overkill, but whatever. Protect rtw_ioctl with splnet(). Do not load a tx descriptor with a buffer shorter than 4 bytes. Handle a transmit timeout less disruptively.
2004-12-25 09:58:37 +03:00
for (pri = 0; pri < RTW_NTXPRI; pri++) {
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
tdb = &sc->sc_txdesc_blk[pri];
rtw_txsofts_release(sc->sc_dmat, &sc->sc_ic,
&sc->sc_txsoft_blk[pri]);
rtw_txdesc_blk_reset(tdb);
rtw_txdescs_sync(tdb, 0, tdb->tdb_ndesc,
Simplify the reset we do on a transmit timeout. Really, this could be as simple as rtw_stop(, 0), rtw_init().
2004-12-27 22:49:16 +03:00
BUS_DMASYNC_PREWRITE|BUS_DMASYNC_PREREAD);
Change rtw_debug from a debug level to a debug mask. Add a lot of debug flags. From Linux: handle an RTL8180 bug. Sometimes the NIC skips from the middle of the ring to the 0th rx descriptor. Now the driver resynchronizes. Handle a receive descriptor underrun or Rx FIFO overflow condition in the way that the Linux driver does. This kind of seems like overkill, but whatever. Protect rtw_ioctl with splnet(). Do not load a tx descriptor with a buffer shorter than 4 bytes. Handle a transmit timeout less disruptively.
2004-12-25 09:58:37 +03:00
}
}
static void
rtw_rxdescs_reset(struct rtw_softc *sc)
{
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
rtw_rxdesc_init_all(&sc->sc_rxdesc_blk, &sc->sc_rxsoft[0], 1);
Change rtw_debug from a debug level to a debug mask. Add a lot of debug flags. From Linux: handle an RTL8180 bug. Sometimes the NIC skips from the middle of the ring to the 0th rx descriptor. Now the driver resynchronizes. Handle a receive descriptor underrun or Rx FIFO overflow condition in the way that the Linux driver does. This kind of seems like overkill, but whatever. Protect rtw_ioctl with splnet(). Do not load a tx descriptor with a buffer shorter than 4 bytes. Handle a transmit timeout less disruptively.
2004-12-25 09:58:37 +03:00
}
static void
rtw_intr_ioerror(struct rtw_softc *sc, uint16_t isr)
{
Miscellaneous changes. Details below. Important changes flagged with []. Using the driver with my Linksys WPC11 ver. 4, it seems to be receiving packets for a change. The WPC11 ver. 4 has a Maxim RF section. My no-name rtw with Philips RF section still does not receive any packets. Keep access-level (analog params > config[0123] registers > none) in sc_access. Add rtw_set_access for changing the access level. Make rtw_continuous_tx_enable and other subroutines use rtw_set_access instead of rtw_config0123_enable and rtw_anaparm_enable. Factor part of the chip-reset code into rtw_chip_reset1. Change the 'struct foo (*bar)[N]'-style arguments to 'struct foo *bar'-style arguments. Consolidate software/hardware Tx/Rx ring setup in rtw_hwring_setup, rtw_swring_setup. Add a new constant, SA2400_OPMODE_DEFAULTS, for the bits that we *always* set in the SA2400 OPMODE register. Factor some code out into rtw_sa2400_calibrate. (Inspired by the Linux driver.) [] When the receiver goes into underrun/overflow state, call a new subroutine, rtw_kick() that stops the Rx/Tx processes, resets the chip, reinitializes the Tx/Rx rings, and restarts Rx/Tx processes. (Inspired by the Linux driver.) [] In rtw_intr_rx, check for too-short packets before calling ieee80211_find_rxnode. I believe this will prevent a repeat of the MCHK exception I saw once on macppc. [] Use seconds-elapased as well as microseconds-elapsed to set the next "due date" for the timeout interrupt. This keeps the driver from programming the timeout to expire too early. [] In rtw_intr, read RTW_ISR at most 10 times, then get out. If the interface is not enabled (RTW_F_ENABLED), then get out. [] In rtw_stop, get out if the interface is not enabled (RTW_F_ENABLED). Block IPL_NET interrupts. Don't read/write any registers if the interface is invalid (RTW_F_INVALID). [] Call rtw_stop in rtw_detach.
2004-12-12 09:37:59 +03:00
struct rtw_regs *regs = &sc->sc_regs;
Add the transmit section of rtw(4): build up and tear down transmit descriptor rings, move packets from the transmit queues to the transmit rings, handle transmission-completed interrupts. My Linksys WPC11 ver. 4 with Maxim RF, generates interrupts like it is successfully transmitting packets. Clearly there are bugs: my G4 Powerbook locks up hard. I will debug tomorrow.
2004-12-19 11:19:25 +03:00
Change rtw_debug from a debug level to a debug mask. Add a lot of debug flags. From Linux: handle an RTL8180 bug. Sometimes the NIC skips from the middle of the ring to the 0th rx descriptor. Now the driver resynchronizes. Handle a receive descriptor underrun or Rx FIFO overflow condition in the way that the Linux driver does. This kind of seems like overkill, but whatever. Protect rtw_ioctl with splnet(). Do not load a tx descriptor with a buffer shorter than 4 bytes. Handle a transmit timeout less disruptively.
2004-12-25 09:58:37 +03:00
if ((isr & RTW_INTR_TXFOVW) != 0)
printf("%s: tx fifo overflow\n", sc->sc_dev.dv_xname);
if ((isr & (RTW_INTR_RDU|RTW_INTR_RXFOVW)) == 0)
return;
Disable interrupts and stop transmit DMA before disabling the transmitter and receiver.
2004-12-23 08:54:54 +03:00
The rx descriptor ring bug happens quite often, so printf'ing occurrences is annoying. Only printf about it when rtw_debug contains RTW_DEBUG_BUG.
2004-12-25 10:45:53 +03:00
RTW_DPRINTF(RTW_DEBUG_BUGS, ("%s: restarting xmit/recv\n",
sc->sc_dev.dv_xname));
Disable interrupts and stop transmit DMA before disabling the transmitter and receiver.
2004-12-23 08:54:54 +03:00
Make rtw(4) build w/o RTW_DEBUG. Thanks to Charles Hannum for pointing out my oversight.
2004-12-27 01:59:41 +03:00
#ifdef RTW_DEBUG
Change rtw_debug from a debug level to a debug mask. Add a lot of debug flags. From Linux: handle an RTL8180 bug. Sometimes the NIC skips from the middle of the ring to the 0th rx descriptor. Now the driver resynchronizes. Handle a receive descriptor underrun or Rx FIFO overflow condition in the way that the Linux driver does. This kind of seems like overkill, but whatever. Protect rtw_ioctl with splnet(). Do not load a tx descriptor with a buffer shorter than 4 bytes. Handle a transmit timeout less disruptively.
2004-12-25 09:58:37 +03:00
rtw_dump_rings(sc);
Make rtw(4) build w/o RTW_DEBUG. Thanks to Charles Hannum for pointing out my oversight.
2004-12-27 01:59:41 +03:00
#endif /* RTW_DEBUG */
Disable interrupts and stop transmit DMA before disabling the transmitter and receiver.
2004-12-23 08:54:54 +03:00
rtw_io_enable(regs, RTW_CR_RE | RTW_CR_TE, 0);
Change rtw_debug from a debug level to a debug mask. Add a lot of debug flags. From Linux: handle an RTL8180 bug. Sometimes the NIC skips from the middle of the ring to the 0th rx descriptor. Now the driver resynchronizes. Handle a receive descriptor underrun or Rx FIFO overflow condition in the way that the Linux driver does. This kind of seems like overkill, but whatever. Protect rtw_ioctl with splnet(). Do not load a tx descriptor with a buffer shorter than 4 bytes. Handle a transmit timeout less disruptively.
2004-12-25 09:58:37 +03:00
/* Collect rx'd packets. Refresh rx buffers. */
rtw_intr_rx(sc, 0);
/* Collect tx'd packets. */
rtw_intr_tx(sc, 0);
RTW_WRITE16(regs, RTW_IMR, 0);
RTW_SYNC(regs, RTW_IMR, RTW_IMR);
rtw_chip_reset1(regs, sc->sc_dev.dv_xname);
rtw_rxdescs_reset(sc);
rtw_txdescs_reset(sc);
Miscellaneous changes. Details below. Important changes flagged with []. Using the driver with my Linksys WPC11 ver. 4, it seems to be receiving packets for a change. The WPC11 ver. 4 has a Maxim RF section. My no-name rtw with Philips RF section still does not receive any packets. Keep access-level (analog params > config[0123] registers > none) in sc_access. Add rtw_set_access for changing the access level. Make rtw_continuous_tx_enable and other subroutines use rtw_set_access instead of rtw_config0123_enable and rtw_anaparm_enable. Factor part of the chip-reset code into rtw_chip_reset1. Change the 'struct foo (*bar)[N]'-style arguments to 'struct foo *bar'-style arguments. Consolidate software/hardware Tx/Rx ring setup in rtw_hwring_setup, rtw_swring_setup. Add a new constant, SA2400_OPMODE_DEFAULTS, for the bits that we *always* set in the SA2400 OPMODE register. Factor some code out into rtw_sa2400_calibrate. (Inspired by the Linux driver.) [] When the receiver goes into underrun/overflow state, call a new subroutine, rtw_kick() that stops the Rx/Tx processes, resets the chip, reinitializes the Tx/Rx rings, and restarts Rx/Tx processes. (Inspired by the Linux driver.) [] In rtw_intr_rx, check for too-short packets before calling ieee80211_find_rxnode. I believe this will prevent a repeat of the MCHK exception I saw once on macppc. [] Use seconds-elapased as well as microseconds-elapsed to set the next "due date" for the timeout interrupt. This keeps the driver from programming the timeout to expire too early. [] In rtw_intr, read RTW_ISR at most 10 times, then get out. If the interface is not enabled (RTW_F_ENABLED), then get out. [] In rtw_stop, get out if the interface is not enabled (RTW_F_ENABLED). Block IPL_NET interrupts. Don't read/write any registers if the interface is invalid (RTW_F_INVALID). [] Call rtw_stop in rtw_detach.
2004-12-12 09:37:59 +03:00
rtw_hwring_setup(sc);
Change rtw_debug from a debug level to a debug mask. Add a lot of debug flags. From Linux: handle an RTL8180 bug. Sometimes the NIC skips from the middle of the ring to the 0th rx descriptor. Now the driver resynchronizes. Handle a receive descriptor underrun or Rx FIFO overflow condition in the way that the Linux driver does. This kind of seems like overkill, but whatever. Protect rtw_ioctl with splnet(). Do not load a tx descriptor with a buffer shorter than 4 bytes. Handle a transmit timeout less disruptively.
2004-12-25 09:58:37 +03:00
Make rtw(4) build w/o RTW_DEBUG. Thanks to Charles Hannum for pointing out my oversight.
2004-12-27 01:59:41 +03:00
#ifdef RTW_DEBUG
Change rtw_debug from a debug level to a debug mask. Add a lot of debug flags. From Linux: handle an RTL8180 bug. Sometimes the NIC skips from the middle of the ring to the 0th rx descriptor. Now the driver resynchronizes. Handle a receive descriptor underrun or Rx FIFO overflow condition in the way that the Linux driver does. This kind of seems like overkill, but whatever. Protect rtw_ioctl with splnet(). Do not load a tx descriptor with a buffer shorter than 4 bytes. Handle a transmit timeout less disruptively.
2004-12-25 09:58:37 +03:00
rtw_dump_rings(sc);
Make rtw(4) build w/o RTW_DEBUG. Thanks to Charles Hannum for pointing out my oversight.
2004-12-27 01:59:41 +03:00
#endif /* RTW_DEBUG */
Change rtw_debug from a debug level to a debug mask. Add a lot of debug flags. From Linux: handle an RTL8180 bug. Sometimes the NIC skips from the middle of the ring to the 0th rx descriptor. Now the driver resynchronizes. Handle a receive descriptor underrun or Rx FIFO overflow condition in the way that the Linux driver does. This kind of seems like overkill, but whatever. Protect rtw_ioctl with splnet(). Do not load a tx descriptor with a buffer shorter than 4 bytes. Handle a transmit timeout less disruptively.
2004-12-25 09:58:37 +03:00
Miscellaneous changes. Details below. Important changes flagged with []. Using the driver with my Linksys WPC11 ver. 4, it seems to be receiving packets for a change. The WPC11 ver. 4 has a Maxim RF section. My no-name rtw with Philips RF section still does not receive any packets. Keep access-level (analog params > config[0123] registers > none) in sc_access. Add rtw_set_access for changing the access level. Make rtw_continuous_tx_enable and other subroutines use rtw_set_access instead of rtw_config0123_enable and rtw_anaparm_enable. Factor part of the chip-reset code into rtw_chip_reset1. Change the 'struct foo (*bar)[N]'-style arguments to 'struct foo *bar'-style arguments. Consolidate software/hardware Tx/Rx ring setup in rtw_hwring_setup, rtw_swring_setup. Add a new constant, SA2400_OPMODE_DEFAULTS, for the bits that we *always* set in the SA2400 OPMODE register. Factor some code out into rtw_sa2400_calibrate. (Inspired by the Linux driver.) [] When the receiver goes into underrun/overflow state, call a new subroutine, rtw_kick() that stops the Rx/Tx processes, resets the chip, reinitializes the Tx/Rx rings, and restarts Rx/Tx processes. (Inspired by the Linux driver.) [] In rtw_intr_rx, check for too-short packets before calling ieee80211_find_rxnode. I believe this will prevent a repeat of the MCHK exception I saw once on macppc. [] Use seconds-elapased as well as microseconds-elapsed to set the next "due date" for the timeout interrupt. This keeps the driver from programming the timeout to expire too early. [] In rtw_intr, read RTW_ISR at most 10 times, then get out. If the interface is not enabled (RTW_F_ENABLED), then get out. [] In rtw_stop, get out if the interface is not enabled (RTW_F_ENABLED). Block IPL_NET interrupts. Don't read/write any registers if the interface is invalid (RTW_F_INVALID). [] Call rtw_stop in rtw_detach.
2004-12-12 09:37:59 +03:00
RTW_WRITE16(regs, RTW_IMR, sc->sc_inten);
Add some bus barriers. Remark where they are needed, e.g., read_eeprom. Change some printfs to RTW_DPRINTF2s to quiet rtw(4) when debugging is disabled.
2004-12-20 04:13:45 +03:00
RTW_SYNC(regs, RTW_IMR, RTW_IMR);
Miscellaneous changes. Details below. Important changes flagged with []. Using the driver with my Linksys WPC11 ver. 4, it seems to be receiving packets for a change. The WPC11 ver. 4 has a Maxim RF section. My no-name rtw with Philips RF section still does not receive any packets. Keep access-level (analog params > config[0123] registers > none) in sc_access. Add rtw_set_access for changing the access level. Make rtw_continuous_tx_enable and other subroutines use rtw_set_access instead of rtw_config0123_enable and rtw_anaparm_enable. Factor part of the chip-reset code into rtw_chip_reset1. Change the 'struct foo (*bar)[N]'-style arguments to 'struct foo *bar'-style arguments. Consolidate software/hardware Tx/Rx ring setup in rtw_hwring_setup, rtw_swring_setup. Add a new constant, SA2400_OPMODE_DEFAULTS, for the bits that we *always* set in the SA2400 OPMODE register. Factor some code out into rtw_sa2400_calibrate. (Inspired by the Linux driver.) [] When the receiver goes into underrun/overflow state, call a new subroutine, rtw_kick() that stops the Rx/Tx processes, resets the chip, reinitializes the Tx/Rx rings, and restarts Rx/Tx processes. (Inspired by the Linux driver.) [] In rtw_intr_rx, check for too-short packets before calling ieee80211_find_rxnode. I believe this will prevent a repeat of the MCHK exception I saw once on macppc. [] Use seconds-elapased as well as microseconds-elapsed to set the next "due date" for the timeout interrupt. This keeps the driver from programming the timeout to expire too early. [] In rtw_intr, read RTW_ISR at most 10 times, then get out. If the interface is not enabled (RTW_F_ENABLED), then get out. [] In rtw_stop, get out if the interface is not enabled (RTW_F_ENABLED). Block IPL_NET interrupts. Don't read/write any registers if the interface is invalid (RTW_F_INVALID). [] Call rtw_stop in rtw_detach.
2004-12-12 09:37:59 +03:00
rtw_io_enable(regs, RTW_CR_RE | RTW_CR_TE, 1);
}
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
static __inline void
rtw_suspend_ticks(struct rtw_softc *sc)
{
Change rtw_debug from a debug level to a debug mask. Add a lot of debug flags. From Linux: handle an RTL8180 bug. Sometimes the NIC skips from the middle of the ring to the 0th rx descriptor. Now the driver resynchronizes. Handle a receive descriptor underrun or Rx FIFO overflow condition in the way that the Linux driver does. This kind of seems like overkill, but whatever. Protect rtw_ioctl with splnet(). Do not load a tx descriptor with a buffer shorter than 4 bytes. Handle a transmit timeout less disruptively.
2004-12-25 09:58:37 +03:00
RTW_DPRINTF(RTW_DEBUG_TIMEOUT,
("%s: suspending ticks\n", sc->sc_dev.dv_xname));
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
sc->sc_do_tick = 0;
}
static __inline void
rtw_resume_ticks(struct rtw_softc *sc)
{
u_int32_t tsftrl0, tsftrl1, next_tick;
tsftrl0 = RTW_READ(&sc->sc_regs, RTW_TSFTRL);
tsftrl1 = RTW_READ(&sc->sc_regs, RTW_TSFTRL);
At last, I have rtw w/ Philips RF receiving packets. I added some sysctls to aid debugging: * hw.rtw.debug -- enable debugging * hw.rtw.flush_rfio -- Linux voodoo: possibly makes the MAC "flush" bits down the serial bus to the RF * hw.rtw.host_rfio: force the host to bang bits to the RF, instead of the MAC banging bits * hw.rtw.rfio_delay: after telling the MAC to bang bits to the RF front-end, delay rfio_delay microseconds. * hw.rtw.rfprog_fallback: there is this notion of the "RF programming method." I believe the choice influences the polarity/timing of the serial bus used to program the RF front-end. I know the correct choice for Intersil/RFMD/Philips front-ends, only. For all other front-ends, I "fallback" to rfprog_fallback. Make rtw_txdac_enable take an rtw_softc argument. I will probably revert this change. Add some Linux voodoo to rtw_continuous_tx_enable. I will probably revert this change. Important: add rtw_set_rfprog, which sets the correct RF programming method. This change and the following change are probably responsible for making the Philips RF work. Important: RTW_CONFIG1 is an 8-bit register, treat it that way! Important: RTW_BRSR is 16-bit, RTW_CRCOUNT, RTW_PHYDELAY, and RTW_MSR are 8-bit: treat them that way! Vastly simplify rtw_resume_ticks. Note to self: set the LED state to match the power state. Hedge against the possibility that RTW_MSR is protected as RTW_CONFIG[0123] are, meanwhile reworking that section of rtw_init a little. Add sc_anaparm, which isn't used, yet....
2004-12-13 03:48:02 +03:00
next_tick = tsftrl1 + 1000000;
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
RTW_WRITE(&sc->sc_regs, RTW_TINT, next_tick);
sc->sc_do_tick = 1;
Change rtw_debug from a debug level to a debug mask. Add a lot of debug flags. From Linux: handle an RTL8180 bug. Sometimes the NIC skips from the middle of the ring to the 0th rx descriptor. Now the driver resynchronizes. Handle a receive descriptor underrun or Rx FIFO overflow condition in the way that the Linux driver does. This kind of seems like overkill, but whatever. Protect rtw_ioctl with splnet(). Do not load a tx descriptor with a buffer shorter than 4 bytes. Handle a transmit timeout less disruptively.
2004-12-25 09:58:37 +03:00
RTW_DPRINTF(RTW_DEBUG_TIMEOUT,
("%s: resume ticks delta %#08x now %#08x next %#08x\n",
Add some bus barriers. Remark where they are needed, e.g., read_eeprom. Change some printfs to RTW_DPRINTF2s to quiet rtw(4) when debugging is disabled.
2004-12-20 04:13:45 +03:00
sc->sc_dev.dv_xname, tsftrl1 - tsftrl0, tsftrl1, next_tick));
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
}
static void
rtw_intr_timeout(struct rtw_softc *sc)
{
Change rtw_debug from a debug level to a debug mask. Add a lot of debug flags. From Linux: handle an RTL8180 bug. Sometimes the NIC skips from the middle of the ring to the 0th rx descriptor. Now the driver resynchronizes. Handle a receive descriptor underrun or Rx FIFO overflow condition in the way that the Linux driver does. This kind of seems like overkill, but whatever. Protect rtw_ioctl with splnet(). Do not load a tx descriptor with a buffer shorter than 4 bytes. Handle a transmit timeout less disruptively.
2004-12-25 09:58:37 +03:00
RTW_DPRINTF(RTW_DEBUG_TIMEOUT, ("%s: timeout\n", sc->sc_dev.dv_xname));
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
if (sc->sc_do_tick)
rtw_resume_ticks(sc);
return;
}
int
rtw_intr(void *arg)
{
Miscellaneous changes. Details below. Important changes flagged with []. Using the driver with my Linksys WPC11 ver. 4, it seems to be receiving packets for a change. The WPC11 ver. 4 has a Maxim RF section. My no-name rtw with Philips RF section still does not receive any packets. Keep access-level (analog params > config[0123] registers > none) in sc_access. Add rtw_set_access for changing the access level. Make rtw_continuous_tx_enable and other subroutines use rtw_set_access instead of rtw_config0123_enable and rtw_anaparm_enable. Factor part of the chip-reset code into rtw_chip_reset1. Change the 'struct foo (*bar)[N]'-style arguments to 'struct foo *bar'-style arguments. Consolidate software/hardware Tx/Rx ring setup in rtw_hwring_setup, rtw_swring_setup. Add a new constant, SA2400_OPMODE_DEFAULTS, for the bits that we *always* set in the SA2400 OPMODE register. Factor some code out into rtw_sa2400_calibrate. (Inspired by the Linux driver.) [] When the receiver goes into underrun/overflow state, call a new subroutine, rtw_kick() that stops the Rx/Tx processes, resets the chip, reinitializes the Tx/Rx rings, and restarts Rx/Tx processes. (Inspired by the Linux driver.) [] In rtw_intr_rx, check for too-short packets before calling ieee80211_find_rxnode. I believe this will prevent a repeat of the MCHK exception I saw once on macppc. [] Use seconds-elapased as well as microseconds-elapsed to set the next "due date" for the timeout interrupt. This keeps the driver from programming the timeout to expire too early. [] In rtw_intr, read RTW_ISR at most 10 times, then get out. If the interface is not enabled (RTW_F_ENABLED), then get out. [] In rtw_stop, get out if the interface is not enabled (RTW_F_ENABLED). Block IPL_NET interrupts. Don't read/write any registers if the interface is invalid (RTW_F_INVALID). [] Call rtw_stop in rtw_detach.
2004-12-12 09:37:59 +03:00
int i;
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
struct rtw_softc *sc = arg;
struct rtw_regs *regs = &sc->sc_regs;
u_int16_t isr;
/*
* If the interface isn't running, the interrupt couldn't
* possibly have come from us.
*/
Miscellaneous changes. Details below. Important changes flagged with []. Using the driver with my Linksys WPC11 ver. 4, it seems to be receiving packets for a change. The WPC11 ver. 4 has a Maxim RF section. My no-name rtw with Philips RF section still does not receive any packets. Keep access-level (analog params > config[0123] registers > none) in sc_access. Add rtw_set_access for changing the access level. Make rtw_continuous_tx_enable and other subroutines use rtw_set_access instead of rtw_config0123_enable and rtw_anaparm_enable. Factor part of the chip-reset code into rtw_chip_reset1. Change the 'struct foo (*bar)[N]'-style arguments to 'struct foo *bar'-style arguments. Consolidate software/hardware Tx/Rx ring setup in rtw_hwring_setup, rtw_swring_setup. Add a new constant, SA2400_OPMODE_DEFAULTS, for the bits that we *always* set in the SA2400 OPMODE register. Factor some code out into rtw_sa2400_calibrate. (Inspired by the Linux driver.) [] When the receiver goes into underrun/overflow state, call a new subroutine, rtw_kick() that stops the Rx/Tx processes, resets the chip, reinitializes the Tx/Rx rings, and restarts Rx/Tx processes. (Inspired by the Linux driver.) [] In rtw_intr_rx, check for too-short packets before calling ieee80211_find_rxnode. I believe this will prevent a repeat of the MCHK exception I saw once on macppc. [] Use seconds-elapased as well as microseconds-elapsed to set the next "due date" for the timeout interrupt. This keeps the driver from programming the timeout to expire too early. [] In rtw_intr, read RTW_ISR at most 10 times, then get out. If the interface is not enabled (RTW_F_ENABLED), then get out. [] In rtw_stop, get out if the interface is not enabled (RTW_F_ENABLED). Block IPL_NET interrupts. Don't read/write any registers if the interface is invalid (RTW_F_INVALID). [] Call rtw_stop in rtw_detach.
2004-12-12 09:37:59 +03:00
if ((sc->sc_flags & RTW_F_ENABLED) == 0 ||
(sc->sc_if.if_flags & IFF_RUNNING) == 0 ||
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
(sc->sc_dev.dv_flags & DVF_ACTIVE) == 0) {
Change rtw_debug from a debug level to a debug mask. Add a lot of debug flags. From Linux: handle an RTL8180 bug. Sometimes the NIC skips from the middle of the ring to the 0th rx descriptor. Now the driver resynchronizes. Handle a receive descriptor underrun or Rx FIFO overflow condition in the way that the Linux driver does. This kind of seems like overkill, but whatever. Protect rtw_ioctl with splnet(). Do not load a tx descriptor with a buffer shorter than 4 bytes. Handle a transmit timeout less disruptively.
2004-12-25 09:58:37 +03:00
RTW_DPRINTF(RTW_DEBUG_INTR, ("%s: stray interrupt\n", sc->sc_dev.dv_xname));
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
return (0);
}
Miscellaneous changes. Details below. Important changes flagged with []. Using the driver with my Linksys WPC11 ver. 4, it seems to be receiving packets for a change. The WPC11 ver. 4 has a Maxim RF section. My no-name rtw with Philips RF section still does not receive any packets. Keep access-level (analog params > config[0123] registers > none) in sc_access. Add rtw_set_access for changing the access level. Make rtw_continuous_tx_enable and other subroutines use rtw_set_access instead of rtw_config0123_enable and rtw_anaparm_enable. Factor part of the chip-reset code into rtw_chip_reset1. Change the 'struct foo (*bar)[N]'-style arguments to 'struct foo *bar'-style arguments. Consolidate software/hardware Tx/Rx ring setup in rtw_hwring_setup, rtw_swring_setup. Add a new constant, SA2400_OPMODE_DEFAULTS, for the bits that we *always* set in the SA2400 OPMODE register. Factor some code out into rtw_sa2400_calibrate. (Inspired by the Linux driver.) [] When the receiver goes into underrun/overflow state, call a new subroutine, rtw_kick() that stops the Rx/Tx processes, resets the chip, reinitializes the Tx/Rx rings, and restarts Rx/Tx processes. (Inspired by the Linux driver.) [] In rtw_intr_rx, check for too-short packets before calling ieee80211_find_rxnode. I believe this will prevent a repeat of the MCHK exception I saw once on macppc. [] Use seconds-elapased as well as microseconds-elapsed to set the next "due date" for the timeout interrupt. This keeps the driver from programming the timeout to expire too early. [] In rtw_intr, read RTW_ISR at most 10 times, then get out. If the interface is not enabled (RTW_F_ENABLED), then get out. [] In rtw_stop, get out if the interface is not enabled (RTW_F_ENABLED). Block IPL_NET interrupts. Don't read/write any registers if the interface is invalid (RTW_F_INVALID). [] Call rtw_stop in rtw_detach.
2004-12-12 09:37:59 +03:00
for (i = 0; i < 10; i++) {
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
isr = RTW_READ16(regs, RTW_ISR);
RTW_WRITE16(regs, RTW_ISR, isr);
Add some bus barriers. Remark where they are needed, e.g., read_eeprom. Change some printfs to RTW_DPRINTF2s to quiet rtw(4) when debugging is disabled.
2004-12-20 04:13:45 +03:00
RTW_WBR(regs, RTW_ISR, RTW_ISR);
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
if (sc->sc_intr_ack != NULL)
(*sc->sc_intr_ack)(regs);
if (isr == 0)
break;
#ifdef RTW_DEBUG
#define PRINTINTR(flag) do { \
if ((isr & flag) != 0) { \
printf("%s" #flag, delim); \
delim = ","; \
} \
} while (0)
Change rtw_debug from a debug level to a debug mask. Add a lot of debug flags. From Linux: handle an RTL8180 bug. Sometimes the NIC skips from the middle of the ring to the 0th rx descriptor. Now the driver resynchronizes. Handle a receive descriptor underrun or Rx FIFO overflow condition in the way that the Linux driver does. This kind of seems like overkill, but whatever. Protect rtw_ioctl with splnet(). Do not load a tx descriptor with a buffer shorter than 4 bytes. Handle a transmit timeout less disruptively.
2004-12-25 09:58:37 +03:00
if ((rtw_debug & RTW_DEBUG_INTR) != 0 && isr != 0) {
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
const char *delim = "<";
printf("%s: reg[ISR] = %x", sc->sc_dev.dv_xname, isr);
PRINTINTR(RTW_INTR_TXFOVW);
PRINTINTR(RTW_INTR_TIMEOUT);
PRINTINTR(RTW_INTR_BCNINT);
PRINTINTR(RTW_INTR_ATIMINT);
PRINTINTR(RTW_INTR_TBDER);
PRINTINTR(RTW_INTR_TBDOK);
PRINTINTR(RTW_INTR_THPDER);
PRINTINTR(RTW_INTR_THPDOK);
PRINTINTR(RTW_INTR_TNPDER);
PRINTINTR(RTW_INTR_TNPDOK);
PRINTINTR(RTW_INTR_RXFOVW);
PRINTINTR(RTW_INTR_RDU);
PRINTINTR(RTW_INTR_TLPDER);
PRINTINTR(RTW_INTR_TLPDOK);
PRINTINTR(RTW_INTR_RER);
PRINTINTR(RTW_INTR_ROK);
printf(">\n");
}
#undef PRINTINTR
#endif /* RTW_DEBUG */
if ((isr & RTW_INTR_RX) != 0)
rtw_intr_rx(sc, isr & RTW_INTR_RX);
if ((isr & RTW_INTR_TX) != 0)
rtw_intr_tx(sc, isr & RTW_INTR_TX);
if ((isr & RTW_INTR_BEACON) != 0)
rtw_intr_beacon(sc, isr & RTW_INTR_BEACON);
if ((isr & RTW_INTR_ATIMINT) != 0)
rtw_intr_atim(sc);
if ((isr & RTW_INTR_IOERROR) != 0)
rtw_intr_ioerror(sc, isr & RTW_INTR_IOERROR);
if ((isr & RTW_INTR_TIMEOUT) != 0)
rtw_intr_timeout(sc);
}
return 1;
}
Change rtw_debug from a debug level to a debug mask. Add a lot of debug flags. From Linux: handle an RTL8180 bug. Sometimes the NIC skips from the middle of the ring to the 0th rx descriptor. Now the driver resynchronizes. Handle a receive descriptor underrun or Rx FIFO overflow condition in the way that the Linux driver does. This kind of seems like overkill, but whatever. Protect rtw_ioctl with splnet(). Do not load a tx descriptor with a buffer shorter than 4 bytes. Handle a transmit timeout less disruptively.
2004-12-25 09:58:37 +03:00
/* Must be called at splnet. */
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
static void
rtw_stop(struct ifnet *ifp, int disable)
{
Change rtw_debug from a debug level to a debug mask. Add a lot of debug flags. From Linux: handle an RTL8180 bug. Sometimes the NIC skips from the middle of the ring to the 0th rx descriptor. Now the driver resynchronizes. Handle a receive descriptor underrun or Rx FIFO overflow condition in the way that the Linux driver does. This kind of seems like overkill, but whatever. Protect rtw_ioctl with splnet(). Do not load a tx descriptor with a buffer shorter than 4 bytes. Handle a transmit timeout less disruptively.
2004-12-25 09:58:37 +03:00
int pri;
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
struct rtw_softc *sc = (struct rtw_softc *)ifp->if_softc;
struct ieee80211com *ic = &sc->sc_ic;
struct rtw_regs *regs = &sc->sc_regs;
Miscellaneous changes. Details below. Important changes flagged with []. Using the driver with my Linksys WPC11 ver. 4, it seems to be receiving packets for a change. The WPC11 ver. 4 has a Maxim RF section. My no-name rtw with Philips RF section still does not receive any packets. Keep access-level (analog params > config[0123] registers > none) in sc_access. Add rtw_set_access for changing the access level. Make rtw_continuous_tx_enable and other subroutines use rtw_set_access instead of rtw_config0123_enable and rtw_anaparm_enable. Factor part of the chip-reset code into rtw_chip_reset1. Change the 'struct foo (*bar)[N]'-style arguments to 'struct foo *bar'-style arguments. Consolidate software/hardware Tx/Rx ring setup in rtw_hwring_setup, rtw_swring_setup. Add a new constant, SA2400_OPMODE_DEFAULTS, for the bits that we *always* set in the SA2400 OPMODE register. Factor some code out into rtw_sa2400_calibrate. (Inspired by the Linux driver.) [] When the receiver goes into underrun/overflow state, call a new subroutine, rtw_kick() that stops the Rx/Tx processes, resets the chip, reinitializes the Tx/Rx rings, and restarts Rx/Tx processes. (Inspired by the Linux driver.) [] In rtw_intr_rx, check for too-short packets before calling ieee80211_find_rxnode. I believe this will prevent a repeat of the MCHK exception I saw once on macppc. [] Use seconds-elapased as well as microseconds-elapsed to set the next "due date" for the timeout interrupt. This keeps the driver from programming the timeout to expire too early. [] In rtw_intr, read RTW_ISR at most 10 times, then get out. If the interface is not enabled (RTW_F_ENABLED), then get out. [] In rtw_stop, get out if the interface is not enabled (RTW_F_ENABLED). Block IPL_NET interrupts. Don't read/write any registers if the interface is invalid (RTW_F_INVALID). [] Call rtw_stop in rtw_detach.
2004-12-12 09:37:59 +03:00
if ((sc->sc_flags & RTW_F_ENABLED) == 0)
return;
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
rtw_suspend_ticks(sc);
ieee80211_new_state(ic, IEEE80211_S_INIT, -1);
Miscellaneous changes. Details below. Important changes flagged with []. Using the driver with my Linksys WPC11 ver. 4, it seems to be receiving packets for a change. The WPC11 ver. 4 has a Maxim RF section. My no-name rtw with Philips RF section still does not receive any packets. Keep access-level (analog params > config[0123] registers > none) in sc_access. Add rtw_set_access for changing the access level. Make rtw_continuous_tx_enable and other subroutines use rtw_set_access instead of rtw_config0123_enable and rtw_anaparm_enable. Factor part of the chip-reset code into rtw_chip_reset1. Change the 'struct foo (*bar)[N]'-style arguments to 'struct foo *bar'-style arguments. Consolidate software/hardware Tx/Rx ring setup in rtw_hwring_setup, rtw_swring_setup. Add a new constant, SA2400_OPMODE_DEFAULTS, for the bits that we *always* set in the SA2400 OPMODE register. Factor some code out into rtw_sa2400_calibrate. (Inspired by the Linux driver.) [] When the receiver goes into underrun/overflow state, call a new subroutine, rtw_kick() that stops the Rx/Tx processes, resets the chip, reinitializes the Tx/Rx rings, and restarts Rx/Tx processes. (Inspired by the Linux driver.) [] In rtw_intr_rx, check for too-short packets before calling ieee80211_find_rxnode. I believe this will prevent a repeat of the MCHK exception I saw once on macppc. [] Use seconds-elapased as well as microseconds-elapsed to set the next "due date" for the timeout interrupt. This keeps the driver from programming the timeout to expire too early. [] In rtw_intr, read RTW_ISR at most 10 times, then get out. If the interface is not enabled (RTW_F_ENABLED), then get out. [] In rtw_stop, get out if the interface is not enabled (RTW_F_ENABLED). Block IPL_NET interrupts. Don't read/write any registers if the interface is invalid (RTW_F_INVALID). [] Call rtw_stop in rtw_detach.
2004-12-12 09:37:59 +03:00
if ((sc->sc_flags & RTW_F_INVALID) == 0) {
/* Disable interrupts. */
RTW_WRITE16(regs, RTW_IMR, 0);
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
Add some bus barriers. Remark where they are needed, e.g., read_eeprom. Change some printfs to RTW_DPRINTF2s to quiet rtw(4) when debugging is disabled.
2004-12-20 04:13:45 +03:00
RTW_WBW(regs, RTW_TPPOLL, RTW_IMR);
Miscellaneous changes. Details below. Important changes flagged with []. Using the driver with my Linksys WPC11 ver. 4, it seems to be receiving packets for a change. The WPC11 ver. 4 has a Maxim RF section. My no-name rtw with Philips RF section still does not receive any packets. Keep access-level (analog params > config[0123] registers > none) in sc_access. Add rtw_set_access for changing the access level. Make rtw_continuous_tx_enable and other subroutines use rtw_set_access instead of rtw_config0123_enable and rtw_anaparm_enable. Factor part of the chip-reset code into rtw_chip_reset1. Change the 'struct foo (*bar)[N]'-style arguments to 'struct foo *bar'-style arguments. Consolidate software/hardware Tx/Rx ring setup in rtw_hwring_setup, rtw_swring_setup. Add a new constant, SA2400_OPMODE_DEFAULTS, for the bits that we *always* set in the SA2400 OPMODE register. Factor some code out into rtw_sa2400_calibrate. (Inspired by the Linux driver.) [] When the receiver goes into underrun/overflow state, call a new subroutine, rtw_kick() that stops the Rx/Tx processes, resets the chip, reinitializes the Tx/Rx rings, and restarts Rx/Tx processes. (Inspired by the Linux driver.) [] In rtw_intr_rx, check for too-short packets before calling ieee80211_find_rxnode. I believe this will prevent a repeat of the MCHK exception I saw once on macppc. [] Use seconds-elapased as well as microseconds-elapsed to set the next "due date" for the timeout interrupt. This keeps the driver from programming the timeout to expire too early. [] In rtw_intr, read RTW_ISR at most 10 times, then get out. If the interface is not enabled (RTW_F_ENABLED), then get out. [] In rtw_stop, get out if the interface is not enabled (RTW_F_ENABLED). Block IPL_NET interrupts. Don't read/write any registers if the interface is invalid (RTW_F_INVALID). [] Call rtw_stop in rtw_detach.
2004-12-12 09:37:59 +03:00
/* Stop the transmit and receive processes. First stop DMA,
* then disable receiver and transmitter.
*/
RTW_WRITE8(regs, RTW_TPPOLL,
RTW_TPPOLL_SBQ|RTW_TPPOLL_SHPQ|RTW_TPPOLL_SNPQ|
RTW_TPPOLL_SLPQ);
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
Add some bus barriers. Remark where they are needed, e.g., read_eeprom. Change some printfs to RTW_DPRINTF2s to quiet rtw(4) when debugging is disabled.
2004-12-20 04:13:45 +03:00
RTW_SYNC(regs, RTW_TPPOLL, RTW_IMR);
Miscellaneous changes. Details below. Important changes flagged with []. Using the driver with my Linksys WPC11 ver. 4, it seems to be receiving packets for a change. The WPC11 ver. 4 has a Maxim RF section. My no-name rtw with Philips RF section still does not receive any packets. Keep access-level (analog params > config[0123] registers > none) in sc_access. Add rtw_set_access for changing the access level. Make rtw_continuous_tx_enable and other subroutines use rtw_set_access instead of rtw_config0123_enable and rtw_anaparm_enable. Factor part of the chip-reset code into rtw_chip_reset1. Change the 'struct foo (*bar)[N]'-style arguments to 'struct foo *bar'-style arguments. Consolidate software/hardware Tx/Rx ring setup in rtw_hwring_setup, rtw_swring_setup. Add a new constant, SA2400_OPMODE_DEFAULTS, for the bits that we *always* set in the SA2400 OPMODE register. Factor some code out into rtw_sa2400_calibrate. (Inspired by the Linux driver.) [] When the receiver goes into underrun/overflow state, call a new subroutine, rtw_kick() that stops the Rx/Tx processes, resets the chip, reinitializes the Tx/Rx rings, and restarts Rx/Tx processes. (Inspired by the Linux driver.) [] In rtw_intr_rx, check for too-short packets before calling ieee80211_find_rxnode. I believe this will prevent a repeat of the MCHK exception I saw once on macppc. [] Use seconds-elapased as well as microseconds-elapsed to set the next "due date" for the timeout interrupt. This keeps the driver from programming the timeout to expire too early. [] In rtw_intr, read RTW_ISR at most 10 times, then get out. If the interface is not enabled (RTW_F_ENABLED), then get out. [] In rtw_stop, get out if the interface is not enabled (RTW_F_ENABLED). Block IPL_NET interrupts. Don't read/write any registers if the interface is invalid (RTW_F_INVALID). [] Call rtw_stop in rtw_detach.
2004-12-12 09:37:59 +03:00
rtw_io_enable(&sc->sc_regs, RTW_CR_RE|RTW_CR_TE, 0);
}
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
Add the transmit section of rtw(4): build up and tear down transmit descriptor rings, move packets from the transmit queues to the transmit rings, handle transmission-completed interrupts. My Linksys WPC11 ver. 4 with Maxim RF, generates interrupts like it is successfully transmitting packets. Clearly there are bugs: my G4 Powerbook locks up hard. I will debug tomorrow.
2004-12-19 11:19:25 +03:00
for (pri = 0; pri < RTW_NTXPRI; pri++) {
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
rtw_txsofts_release(sc->sc_dmat, &sc->sc_ic,
&sc->sc_txsoft_blk[pri]);
Add the transmit section of rtw(4): build up and tear down transmit descriptor rings, move packets from the transmit queues to the transmit rings, handle transmission-completed interrupts. My Linksys WPC11 ver. 4 with Maxim RF, generates interrupts like it is successfully transmitting packets. Clearly there are bugs: my G4 Powerbook locks up hard. I will debug tomorrow.
2004-12-19 11:19:25 +03:00
}
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
rtw_rxbufs_release(sc->sc_dmat, &sc->sc_rxsoft[0]);
If defined(RTW_DEBUG), provide a sysctl (hw.rtw.rxbufs_limit) for limiting the number of rx buffers an rtw may allocate. Use this sysctl to test the code that copes with buffer exhaustion. Allocate at most RTW_RXQLEN rx buffers, stopping at the sysctl limit. Record in sc_nrxdesc how many were allocated, and put the end-of-ring indication on sc_rxdesc[sc_nrxdesc - 1]. In rtw_init, if no rx buffers could be allocated, log a complaint, clear IFF_RUNNING, and exit with an error. Many changes to accomodate a short rx ring, mainly of the "add a rx-ring length argument" variety. XXX I really should consolidate all of the rx ring variables in one struct and pass that to the rx-ring subroutines. Bug fix: after calling MCLGET, use the (m->m_flags & M_EXT) idiom to check for success, instead of m != NULL. Bug fix: at the top of rtw_start, if IFF_RUNNING is not set, or IFF_OACTIVE is, get out.
2004-12-29 01:21:15 +03:00
if (disable)
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
rtw_disable(sc);
/* Mark the interface as not running. Cancel the watchdog timer. */
Change rtw_debug from a debug level to a debug mask. Add a lot of debug flags. From Linux: handle an RTL8180 bug. Sometimes the NIC skips from the middle of the ring to the 0th rx descriptor. Now the driver resynchronizes. Handle a receive descriptor underrun or Rx FIFO overflow condition in the way that the Linux driver does. This kind of seems like overkill, but whatever. Protect rtw_ioctl with splnet(). Do not load a tx descriptor with a buffer shorter than 4 bytes. Handle a transmit timeout less disruptively.
2004-12-25 09:58:37 +03:00
ifp->if_flags &= ~(IFF_RUNNING | IFF_OACTIVE);
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
ifp->if_timer = 0;
Miscellaneous changes. Details below. Important changes flagged with []. Using the driver with my Linksys WPC11 ver. 4, it seems to be receiving packets for a change. The WPC11 ver. 4 has a Maxim RF section. My no-name rtw with Philips RF section still does not receive any packets. Keep access-level (analog params > config[0123] registers > none) in sc_access. Add rtw_set_access for changing the access level. Make rtw_continuous_tx_enable and other subroutines use rtw_set_access instead of rtw_config0123_enable and rtw_anaparm_enable. Factor part of the chip-reset code into rtw_chip_reset1. Change the 'struct foo (*bar)[N]'-style arguments to 'struct foo *bar'-style arguments. Consolidate software/hardware Tx/Rx ring setup in rtw_hwring_setup, rtw_swring_setup. Add a new constant, SA2400_OPMODE_DEFAULTS, for the bits that we *always* set in the SA2400 OPMODE register. Factor some code out into rtw_sa2400_calibrate. (Inspired by the Linux driver.) [] When the receiver goes into underrun/overflow state, call a new subroutine, rtw_kick() that stops the Rx/Tx processes, resets the chip, reinitializes the Tx/Rx rings, and restarts Rx/Tx processes. (Inspired by the Linux driver.) [] In rtw_intr_rx, check for too-short packets before calling ieee80211_find_rxnode. I believe this will prevent a repeat of the MCHK exception I saw once on macppc. [] Use seconds-elapased as well as microseconds-elapsed to set the next "due date" for the timeout interrupt. This keeps the driver from programming the timeout to expire too early. [] In rtw_intr, read RTW_ISR at most 10 times, then get out. If the interface is not enabled (RTW_F_ENABLED), then get out. [] In rtw_stop, get out if the interface is not enabled (RTW_F_ENABLED). Block IPL_NET interrupts. Don't read/write any registers if the interface is invalid (RTW_F_INVALID). [] Call rtw_stop in rtw_detach.
2004-12-12 09:37:59 +03:00
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
return;
}
const char *
rtw_pwrstate_string(enum rtw_pwrstate power)
{
switch (power) {
case RTW_ON:
return "on";
case RTW_SLEEP:
return "sleep";
case RTW_OFF:
return "off";
default:
return "unknown";
}
}
Realtek sample sources are not consistent about the "Digital PHY" bit in the SROM. It seems as if it is set to 1 when the PHY is *analog*, not *digital*. Fix my sources. In rtw_intr_rx, use units of 500kb/s instead of 100kb/s for rate, to be consistent with net80211's expectations. Polish up some debugging ugly messages. Dump raw 802.11 packets if IFF_DEBUG|IFF_LINK2 and RTW_DEBUG is defined. Polish power-state (on/sleep/off) handling. Especially improve support for RFMD (totally untested) and Maxim. For Philips, take the Digital PHY property into account. Call the net80211 watchdog function from rtw_watchdog, so that we scan again if auth/assoc fails. Be a little more cautious about writing register[RTW_TPPOLL], since other drivers are.... Don't frob the high/low-priority queues right now, since I don't use them. Add rtw_join_bss which programs the card with the BSSID and other properties of a BSS. Use it on state transitions. Factor out rtw_set_nettype. Make rtw_recv_beacon call ieee80211_recv_mgmt instead of dropping beacons on the floor! TBD IBSS merges. Change some rtw_debug=2 printfs to rtw_debug=3 (RTW_DPRINTF3) printfs so the console doesn't get spammed so badly at rtw_debug=2. Change some debugging printfs to RTW_DPRINTFs. E.g., print the "RF programming method" only if debugging is enabled.
2004-12-21 02:05:41 +03:00
/* XXX For Maxim, I am using the RFMD settings gleaned from the
* reference driver, plus a magic Maxim "ON" value that comes from
* the Realtek document "Windows PG for Rtl8180."
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
*/
static void
rtw_maxim_pwrstate(struct rtw_regs *regs, enum rtw_pwrstate power,
Realtek sample sources are not consistent about the "Digital PHY" bit in the SROM. It seems as if it is set to 1 when the PHY is *analog*, not *digital*. Fix my sources. In rtw_intr_rx, use units of 500kb/s instead of 100kb/s for rate, to be consistent with net80211's expectations. Polish up some debugging ugly messages. Dump raw 802.11 packets if IFF_DEBUG|IFF_LINK2 and RTW_DEBUG is defined. Polish power-state (on/sleep/off) handling. Especially improve support for RFMD (totally untested) and Maxim. For Philips, take the Digital PHY property into account. Call the net80211 watchdog function from rtw_watchdog, so that we scan again if auth/assoc fails. Be a little more cautious about writing register[RTW_TPPOLL], since other drivers are.... Don't frob the high/low-priority queues right now, since I don't use them. Add rtw_join_bss which programs the card with the BSSID and other properties of a BSS. Use it on state transitions. Factor out rtw_set_nettype. Make rtw_recv_beacon call ieee80211_recv_mgmt instead of dropping beacons on the floor! TBD IBSS merges. Change some rtw_debug=2 printfs to rtw_debug=3 (RTW_DPRINTF3) printfs so the console doesn't get spammed so badly at rtw_debug=2. Change some debugging printfs to RTW_DPRINTFs. E.g., print the "RF programming method" only if debugging is enabled.
2004-12-21 02:05:41 +03:00
int before_rf, int digphy)
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
{
u_int32_t anaparm;
Realtek sample sources are not consistent about the "Digital PHY" bit in the SROM. It seems as if it is set to 1 when the PHY is *analog*, not *digital*. Fix my sources. In rtw_intr_rx, use units of 500kb/s instead of 100kb/s for rate, to be consistent with net80211's expectations. Polish up some debugging ugly messages. Dump raw 802.11 packets if IFF_DEBUG|IFF_LINK2 and RTW_DEBUG is defined. Polish power-state (on/sleep/off) handling. Especially improve support for RFMD (totally untested) and Maxim. For Philips, take the Digital PHY property into account. Call the net80211 watchdog function from rtw_watchdog, so that we scan again if auth/assoc fails. Be a little more cautious about writing register[RTW_TPPOLL], since other drivers are.... Don't frob the high/low-priority queues right now, since I don't use them. Add rtw_join_bss which programs the card with the BSSID and other properties of a BSS. Use it on state transitions. Factor out rtw_set_nettype. Make rtw_recv_beacon call ieee80211_recv_mgmt instead of dropping beacons on the floor! TBD IBSS merges. Change some rtw_debug=2 printfs to rtw_debug=3 (RTW_DPRINTF3) printfs so the console doesn't get spammed so badly at rtw_debug=2. Change some debugging printfs to RTW_DPRINTFs. E.g., print the "RF programming method" only if debugging is enabled.
2004-12-21 02:05:41 +03:00
anaparm = RTW_READ(regs, RTW_ANAPARM);
anaparm &= ~(RTW_ANAPARM_RFPOW_MASK | RTW_ANAPARM_TXDACOFF);
switch (power) {
case RTW_OFF:
if (before_rf)
return;
anaparm |= RTW_ANAPARM_RFPOW_MAXIM_OFF;
anaparm |= RTW_ANAPARM_TXDACOFF;
break;
case RTW_SLEEP:
if (!before_rf)
return;
anaparm |= RTW_ANAPARM_RFPOW_MAXIM_SLEEP;
anaparm |= RTW_ANAPARM_TXDACOFF;
break;
case RTW_ON:
if (!before_rf)
return;
anaparm |= RTW_ANAPARM_RFPOW_MAXIM_ON;
break;
}
Change rtw_debug from a debug level to a debug mask. Add a lot of debug flags. From Linux: handle an RTL8180 bug. Sometimes the NIC skips from the middle of the ring to the 0th rx descriptor. Now the driver resynchronizes. Handle a receive descriptor underrun or Rx FIFO overflow condition in the way that the Linux driver does. This kind of seems like overkill, but whatever. Protect rtw_ioctl with splnet(). Do not load a tx descriptor with a buffer shorter than 4 bytes. Handle a transmit timeout less disruptively.
2004-12-25 09:58:37 +03:00
RTW_DPRINTF(RTW_DEBUG_PWR,
("%s: power state %s, %s RF, reg[ANAPARM] <- %08x\n",
Realtek sample sources are not consistent about the "Digital PHY" bit in the SROM. It seems as if it is set to 1 when the PHY is *analog*, not *digital*. Fix my sources. In rtw_intr_rx, use units of 500kb/s instead of 100kb/s for rate, to be consistent with net80211's expectations. Polish up some debugging ugly messages. Dump raw 802.11 packets if IFF_DEBUG|IFF_LINK2 and RTW_DEBUG is defined. Polish power-state (on/sleep/off) handling. Especially improve support for RFMD (totally untested) and Maxim. For Philips, take the Digital PHY property into account. Call the net80211 watchdog function from rtw_watchdog, so that we scan again if auth/assoc fails. Be a little more cautious about writing register[RTW_TPPOLL], since other drivers are.... Don't frob the high/low-priority queues right now, since I don't use them. Add rtw_join_bss which programs the card with the BSSID and other properties of a BSS. Use it on state transitions. Factor out rtw_set_nettype. Make rtw_recv_beacon call ieee80211_recv_mgmt instead of dropping beacons on the floor! TBD IBSS merges. Change some rtw_debug=2 printfs to rtw_debug=3 (RTW_DPRINTF3) printfs so the console doesn't get spammed so badly at rtw_debug=2. Change some debugging printfs to RTW_DPRINTFs. E.g., print the "RF programming method" only if debugging is enabled.
2004-12-21 02:05:41 +03:00
__func__, rtw_pwrstate_string(power),
(before_rf) ? "before" : "after", anaparm));
RTW_WRITE(regs, RTW_ANAPARM, anaparm);
RTW_SYNC(regs, RTW_ANAPARM, RTW_ANAPARM);
}
/* XXX I am using the RFMD settings gleaned from the reference
* driver. They agree
*/
static void
rtw_rfmd_pwrstate(struct rtw_regs *regs, enum rtw_pwrstate power,
int before_rf, int digphy)
{
u_int32_t anaparm;
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
anaparm = RTW_READ(regs, RTW_ANAPARM);
Realtek sample sources are not consistent about the "Digital PHY" bit in the SROM. It seems as if it is set to 1 when the PHY is *analog*, not *digital*. Fix my sources. In rtw_intr_rx, use units of 500kb/s instead of 100kb/s for rate, to be consistent with net80211's expectations. Polish up some debugging ugly messages. Dump raw 802.11 packets if IFF_DEBUG|IFF_LINK2 and RTW_DEBUG is defined. Polish power-state (on/sleep/off) handling. Especially improve support for RFMD (totally untested) and Maxim. For Philips, take the Digital PHY property into account. Call the net80211 watchdog function from rtw_watchdog, so that we scan again if auth/assoc fails. Be a little more cautious about writing register[RTW_TPPOLL], since other drivers are.... Don't frob the high/low-priority queues right now, since I don't use them. Add rtw_join_bss which programs the card with the BSSID and other properties of a BSS. Use it on state transitions. Factor out rtw_set_nettype. Make rtw_recv_beacon call ieee80211_recv_mgmt instead of dropping beacons on the floor! TBD IBSS merges. Change some rtw_debug=2 printfs to rtw_debug=3 (RTW_DPRINTF3) printfs so the console doesn't get spammed so badly at rtw_debug=2. Change some debugging printfs to RTW_DPRINTFs. E.g., print the "RF programming method" only if debugging is enabled.
2004-12-21 02:05:41 +03:00
anaparm &= ~(RTW_ANAPARM_RFPOW_MASK | RTW_ANAPARM_TXDACOFF);
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
switch (power) {
case RTW_OFF:
if (before_rf)
return;
Realtek sample sources are not consistent about the "Digital PHY" bit in the SROM. It seems as if it is set to 1 when the PHY is *analog*, not *digital*. Fix my sources. In rtw_intr_rx, use units of 500kb/s instead of 100kb/s for rate, to be consistent with net80211's expectations. Polish up some debugging ugly messages. Dump raw 802.11 packets if IFF_DEBUG|IFF_LINK2 and RTW_DEBUG is defined. Polish power-state (on/sleep/off) handling. Especially improve support for RFMD (totally untested) and Maxim. For Philips, take the Digital PHY property into account. Call the net80211 watchdog function from rtw_watchdog, so that we scan again if auth/assoc fails. Be a little more cautious about writing register[RTW_TPPOLL], since other drivers are.... Don't frob the high/low-priority queues right now, since I don't use them. Add rtw_join_bss which programs the card with the BSSID and other properties of a BSS. Use it on state transitions. Factor out rtw_set_nettype. Make rtw_recv_beacon call ieee80211_recv_mgmt instead of dropping beacons on the floor! TBD IBSS merges. Change some rtw_debug=2 printfs to rtw_debug=3 (RTW_DPRINTF3) printfs so the console doesn't get spammed so badly at rtw_debug=2. Change some debugging printfs to RTW_DPRINTFs. E.g., print the "RF programming method" only if debugging is enabled.
2004-12-21 02:05:41 +03:00
anaparm |= RTW_ANAPARM_RFPOW_RFMD_OFF;
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
anaparm |= RTW_ANAPARM_TXDACOFF;
break;
case RTW_SLEEP:
if (!before_rf)
return;
Realtek sample sources are not consistent about the "Digital PHY" bit in the SROM. It seems as if it is set to 1 when the PHY is *analog*, not *digital*. Fix my sources. In rtw_intr_rx, use units of 500kb/s instead of 100kb/s for rate, to be consistent with net80211's expectations. Polish up some debugging ugly messages. Dump raw 802.11 packets if IFF_DEBUG|IFF_LINK2 and RTW_DEBUG is defined. Polish power-state (on/sleep/off) handling. Especially improve support for RFMD (totally untested) and Maxim. For Philips, take the Digital PHY property into account. Call the net80211 watchdog function from rtw_watchdog, so that we scan again if auth/assoc fails. Be a little more cautious about writing register[RTW_TPPOLL], since other drivers are.... Don't frob the high/low-priority queues right now, since I don't use them. Add rtw_join_bss which programs the card with the BSSID and other properties of a BSS. Use it on state transitions. Factor out rtw_set_nettype. Make rtw_recv_beacon call ieee80211_recv_mgmt instead of dropping beacons on the floor! TBD IBSS merges. Change some rtw_debug=2 printfs to rtw_debug=3 (RTW_DPRINTF3) printfs so the console doesn't get spammed so badly at rtw_debug=2. Change some debugging printfs to RTW_DPRINTFs. E.g., print the "RF programming method" only if debugging is enabled.
2004-12-21 02:05:41 +03:00
anaparm |= RTW_ANAPARM_RFPOW_RFMD_SLEEP;
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
anaparm |= RTW_ANAPARM_TXDACOFF;
break;
case RTW_ON:
if (!before_rf)
return;
Realtek sample sources are not consistent about the "Digital PHY" bit in the SROM. It seems as if it is set to 1 when the PHY is *analog*, not *digital*. Fix my sources. In rtw_intr_rx, use units of 500kb/s instead of 100kb/s for rate, to be consistent with net80211's expectations. Polish up some debugging ugly messages. Dump raw 802.11 packets if IFF_DEBUG|IFF_LINK2 and RTW_DEBUG is defined. Polish power-state (on/sleep/off) handling. Especially improve support for RFMD (totally untested) and Maxim. For Philips, take the Digital PHY property into account. Call the net80211 watchdog function from rtw_watchdog, so that we scan again if auth/assoc fails. Be a little more cautious about writing register[RTW_TPPOLL], since other drivers are.... Don't frob the high/low-priority queues right now, since I don't use them. Add rtw_join_bss which programs the card with the BSSID and other properties of a BSS. Use it on state transitions. Factor out rtw_set_nettype. Make rtw_recv_beacon call ieee80211_recv_mgmt instead of dropping beacons on the floor! TBD IBSS merges. Change some rtw_debug=2 printfs to rtw_debug=3 (RTW_DPRINTF3) printfs so the console doesn't get spammed so badly at rtw_debug=2. Change some debugging printfs to RTW_DPRINTFs. E.g., print the "RF programming method" only if debugging is enabled.
2004-12-21 02:05:41 +03:00
anaparm |= RTW_ANAPARM_RFPOW_RFMD_ON;
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
break;
}
Change rtw_debug from a debug level to a debug mask. Add a lot of debug flags. From Linux: handle an RTL8180 bug. Sometimes the NIC skips from the middle of the ring to the 0th rx descriptor. Now the driver resynchronizes. Handle a receive descriptor underrun or Rx FIFO overflow condition in the way that the Linux driver does. This kind of seems like overkill, but whatever. Protect rtw_ioctl with splnet(). Do not load a tx descriptor with a buffer shorter than 4 bytes. Handle a transmit timeout less disruptively.
2004-12-25 09:58:37 +03:00
RTW_DPRINTF(RTW_DEBUG_PWR,
("%s: power state %s, %s RF, reg[ANAPARM] <- %08x\n",
Realtek sample sources are not consistent about the "Digital PHY" bit in the SROM. It seems as if it is set to 1 when the PHY is *analog*, not *digital*. Fix my sources. In rtw_intr_rx, use units of 500kb/s instead of 100kb/s for rate, to be consistent with net80211's expectations. Polish up some debugging ugly messages. Dump raw 802.11 packets if IFF_DEBUG|IFF_LINK2 and RTW_DEBUG is defined. Polish power-state (on/sleep/off) handling. Especially improve support for RFMD (totally untested) and Maxim. For Philips, take the Digital PHY property into account. Call the net80211 watchdog function from rtw_watchdog, so that we scan again if auth/assoc fails. Be a little more cautious about writing register[RTW_TPPOLL], since other drivers are.... Don't frob the high/low-priority queues right now, since I don't use them. Add rtw_join_bss which programs the card with the BSSID and other properties of a BSS. Use it on state transitions. Factor out rtw_set_nettype. Make rtw_recv_beacon call ieee80211_recv_mgmt instead of dropping beacons on the floor! TBD IBSS merges. Change some rtw_debug=2 printfs to rtw_debug=3 (RTW_DPRINTF3) printfs so the console doesn't get spammed so badly at rtw_debug=2. Change some debugging printfs to RTW_DPRINTFs. E.g., print the "RF programming method" only if debugging is enabled.
2004-12-21 02:05:41 +03:00
__func__, rtw_pwrstate_string(power),
(before_rf) ? "before" : "after", anaparm));
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
RTW_WRITE(regs, RTW_ANAPARM, anaparm);
RTW_SYNC(regs, RTW_ANAPARM, RTW_ANAPARM);
}
static void
rtw_philips_pwrstate(struct rtw_regs *regs, enum rtw_pwrstate power,
Realtek sample sources are not consistent about the "Digital PHY" bit in the SROM. It seems as if it is set to 1 when the PHY is *analog*, not *digital*. Fix my sources. In rtw_intr_rx, use units of 500kb/s instead of 100kb/s for rate, to be consistent with net80211's expectations. Polish up some debugging ugly messages. Dump raw 802.11 packets if IFF_DEBUG|IFF_LINK2 and RTW_DEBUG is defined. Polish power-state (on/sleep/off) handling. Especially improve support for RFMD (totally untested) and Maxim. For Philips, take the Digital PHY property into account. Call the net80211 watchdog function from rtw_watchdog, so that we scan again if auth/assoc fails. Be a little more cautious about writing register[RTW_TPPOLL], since other drivers are.... Don't frob the high/low-priority queues right now, since I don't use them. Add rtw_join_bss which programs the card with the BSSID and other properties of a BSS. Use it on state transitions. Factor out rtw_set_nettype. Make rtw_recv_beacon call ieee80211_recv_mgmt instead of dropping beacons on the floor! TBD IBSS merges. Change some rtw_debug=2 printfs to rtw_debug=3 (RTW_DPRINTF3) printfs so the console doesn't get spammed so badly at rtw_debug=2. Change some debugging printfs to RTW_DPRINTFs. E.g., print the "RF programming method" only if debugging is enabled.
2004-12-21 02:05:41 +03:00
int before_rf, int digphy)
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
{
u_int32_t anaparm;
anaparm = RTW_READ(regs, RTW_ANAPARM);
Realtek sample sources are not consistent about the "Digital PHY" bit in the SROM. It seems as if it is set to 1 when the PHY is *analog*, not *digital*. Fix my sources. In rtw_intr_rx, use units of 500kb/s instead of 100kb/s for rate, to be consistent with net80211's expectations. Polish up some debugging ugly messages. Dump raw 802.11 packets if IFF_DEBUG|IFF_LINK2 and RTW_DEBUG is defined. Polish power-state (on/sleep/off) handling. Especially improve support for RFMD (totally untested) and Maxim. For Philips, take the Digital PHY property into account. Call the net80211 watchdog function from rtw_watchdog, so that we scan again if auth/assoc fails. Be a little more cautious about writing register[RTW_TPPOLL], since other drivers are.... Don't frob the high/low-priority queues right now, since I don't use them. Add rtw_join_bss which programs the card with the BSSID and other properties of a BSS. Use it on state transitions. Factor out rtw_set_nettype. Make rtw_recv_beacon call ieee80211_recv_mgmt instead of dropping beacons on the floor! TBD IBSS merges. Change some rtw_debug=2 printfs to rtw_debug=3 (RTW_DPRINTF3) printfs so the console doesn't get spammed so badly at rtw_debug=2. Change some debugging printfs to RTW_DPRINTFs. E.g., print the "RF programming method" only if debugging is enabled.
2004-12-21 02:05:41 +03:00
anaparm &= ~(RTW_ANAPARM_RFPOW_MASK | RTW_ANAPARM_TXDACOFF);
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
switch (power) {
case RTW_OFF:
if (before_rf)
return;
Realtek sample sources are not consistent about the "Digital PHY" bit in the SROM. It seems as if it is set to 1 when the PHY is *analog*, not *digital*. Fix my sources. In rtw_intr_rx, use units of 500kb/s instead of 100kb/s for rate, to be consistent with net80211's expectations. Polish up some debugging ugly messages. Dump raw 802.11 packets if IFF_DEBUG|IFF_LINK2 and RTW_DEBUG is defined. Polish power-state (on/sleep/off) handling. Especially improve support for RFMD (totally untested) and Maxim. For Philips, take the Digital PHY property into account. Call the net80211 watchdog function from rtw_watchdog, so that we scan again if auth/assoc fails. Be a little more cautious about writing register[RTW_TPPOLL], since other drivers are.... Don't frob the high/low-priority queues right now, since I don't use them. Add rtw_join_bss which programs the card with the BSSID and other properties of a BSS. Use it on state transitions. Factor out rtw_set_nettype. Make rtw_recv_beacon call ieee80211_recv_mgmt instead of dropping beacons on the floor! TBD IBSS merges. Change some rtw_debug=2 printfs to rtw_debug=3 (RTW_DPRINTF3) printfs so the console doesn't get spammed so badly at rtw_debug=2. Change some debugging printfs to RTW_DPRINTFs. E.g., print the "RF programming method" only if debugging is enabled.
2004-12-21 02:05:41 +03:00
anaparm |= RTW_ANAPARM_RFPOW_PHILIPS_OFF;
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
anaparm |= RTW_ANAPARM_TXDACOFF;
break;
case RTW_SLEEP:
if (!before_rf)
return;
Realtek sample sources are not consistent about the "Digital PHY" bit in the SROM. It seems as if it is set to 1 when the PHY is *analog*, not *digital*. Fix my sources. In rtw_intr_rx, use units of 500kb/s instead of 100kb/s for rate, to be consistent with net80211's expectations. Polish up some debugging ugly messages. Dump raw 802.11 packets if IFF_DEBUG|IFF_LINK2 and RTW_DEBUG is defined. Polish power-state (on/sleep/off) handling. Especially improve support for RFMD (totally untested) and Maxim. For Philips, take the Digital PHY property into account. Call the net80211 watchdog function from rtw_watchdog, so that we scan again if auth/assoc fails. Be a little more cautious about writing register[RTW_TPPOLL], since other drivers are.... Don't frob the high/low-priority queues right now, since I don't use them. Add rtw_join_bss which programs the card with the BSSID and other properties of a BSS. Use it on state transitions. Factor out rtw_set_nettype. Make rtw_recv_beacon call ieee80211_recv_mgmt instead of dropping beacons on the floor! TBD IBSS merges. Change some rtw_debug=2 printfs to rtw_debug=3 (RTW_DPRINTF3) printfs so the console doesn't get spammed so badly at rtw_debug=2. Change some debugging printfs to RTW_DPRINTFs. E.g., print the "RF programming method" only if debugging is enabled.
2004-12-21 02:05:41 +03:00
anaparm |= RTW_ANAPARM_RFPOW_PHILIPS_SLEEP;
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
anaparm |= RTW_ANAPARM_TXDACOFF;
break;
case RTW_ON:
if (!before_rf)
return;
Realtek sample sources are not consistent about the "Digital PHY" bit in the SROM. It seems as if it is set to 1 when the PHY is *analog*, not *digital*. Fix my sources. In rtw_intr_rx, use units of 500kb/s instead of 100kb/s for rate, to be consistent with net80211's expectations. Polish up some debugging ugly messages. Dump raw 802.11 packets if IFF_DEBUG|IFF_LINK2 and RTW_DEBUG is defined. Polish power-state (on/sleep/off) handling. Especially improve support for RFMD (totally untested) and Maxim. For Philips, take the Digital PHY property into account. Call the net80211 watchdog function from rtw_watchdog, so that we scan again if auth/assoc fails. Be a little more cautious about writing register[RTW_TPPOLL], since other drivers are.... Don't frob the high/low-priority queues right now, since I don't use them. Add rtw_join_bss which programs the card with the BSSID and other properties of a BSS. Use it on state transitions. Factor out rtw_set_nettype. Make rtw_recv_beacon call ieee80211_recv_mgmt instead of dropping beacons on the floor! TBD IBSS merges. Change some rtw_debug=2 printfs to rtw_debug=3 (RTW_DPRINTF3) printfs so the console doesn't get spammed so badly at rtw_debug=2. Change some debugging printfs to RTW_DPRINTFs. E.g., print the "RF programming method" only if debugging is enabled.
2004-12-21 02:05:41 +03:00
if (digphy) {
anaparm |= RTW_ANAPARM_RFPOW_DIG_PHILIPS_ON;
/* XXX guess */
anaparm |= RTW_ANAPARM_TXDACOFF;
} else
anaparm |= RTW_ANAPARM_RFPOW_ANA_PHILIPS_ON;
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
break;
}
Change rtw_debug from a debug level to a debug mask. Add a lot of debug flags. From Linux: handle an RTL8180 bug. Sometimes the NIC skips from the middle of the ring to the 0th rx descriptor. Now the driver resynchronizes. Handle a receive descriptor underrun or Rx FIFO overflow condition in the way that the Linux driver does. This kind of seems like overkill, but whatever. Protect rtw_ioctl with splnet(). Do not load a tx descriptor with a buffer shorter than 4 bytes. Handle a transmit timeout less disruptively.
2004-12-25 09:58:37 +03:00
RTW_DPRINTF(RTW_DEBUG_PWR,
("%s: power state %s, %s RF, reg[ANAPARM] <- %08x\n",
Realtek sample sources are not consistent about the "Digital PHY" bit in the SROM. It seems as if it is set to 1 when the PHY is *analog*, not *digital*. Fix my sources. In rtw_intr_rx, use units of 500kb/s instead of 100kb/s for rate, to be consistent with net80211's expectations. Polish up some debugging ugly messages. Dump raw 802.11 packets if IFF_DEBUG|IFF_LINK2 and RTW_DEBUG is defined. Polish power-state (on/sleep/off) handling. Especially improve support for RFMD (totally untested) and Maxim. For Philips, take the Digital PHY property into account. Call the net80211 watchdog function from rtw_watchdog, so that we scan again if auth/assoc fails. Be a little more cautious about writing register[RTW_TPPOLL], since other drivers are.... Don't frob the high/low-priority queues right now, since I don't use them. Add rtw_join_bss which programs the card with the BSSID and other properties of a BSS. Use it on state transitions. Factor out rtw_set_nettype. Make rtw_recv_beacon call ieee80211_recv_mgmt instead of dropping beacons on the floor! TBD IBSS merges. Change some rtw_debug=2 printfs to rtw_debug=3 (RTW_DPRINTF3) printfs so the console doesn't get spammed so badly at rtw_debug=2. Change some debugging printfs to RTW_DPRINTFs. E.g., print the "RF programming method" only if debugging is enabled.
2004-12-21 02:05:41 +03:00
__func__, rtw_pwrstate_string(power),
(before_rf) ? "before" : "after", anaparm));
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
RTW_WRITE(regs, RTW_ANAPARM, anaparm);
RTW_SYNC(regs, RTW_ANAPARM, RTW_ANAPARM);
}
static void
Realtek sample sources are not consistent about the "Digital PHY" bit in the SROM. It seems as if it is set to 1 when the PHY is *analog*, not *digital*. Fix my sources. In rtw_intr_rx, use units of 500kb/s instead of 100kb/s for rate, to be consistent with net80211's expectations. Polish up some debugging ugly messages. Dump raw 802.11 packets if IFF_DEBUG|IFF_LINK2 and RTW_DEBUG is defined. Polish power-state (on/sleep/off) handling. Especially improve support for RFMD (totally untested) and Maxim. For Philips, take the Digital PHY property into account. Call the net80211 watchdog function from rtw_watchdog, so that we scan again if auth/assoc fails. Be a little more cautious about writing register[RTW_TPPOLL], since other drivers are.... Don't frob the high/low-priority queues right now, since I don't use them. Add rtw_join_bss which programs the card with the BSSID and other properties of a BSS. Use it on state transitions. Factor out rtw_set_nettype. Make rtw_recv_beacon call ieee80211_recv_mgmt instead of dropping beacons on the floor! TBD IBSS merges. Change some rtw_debug=2 printfs to rtw_debug=3 (RTW_DPRINTF3) printfs so the console doesn't get spammed so badly at rtw_debug=2. Change some debugging printfs to RTW_DPRINTFs. E.g., print the "RF programming method" only if debugging is enabled.
2004-12-21 02:05:41 +03:00
rtw_pwrstate0(struct rtw_softc *sc, enum rtw_pwrstate power, int before_rf,
int digphy)
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
{
struct rtw_regs *regs = &sc->sc_regs;
Miscellaneous changes. Details below. Important changes flagged with []. Using the driver with my Linksys WPC11 ver. 4, it seems to be receiving packets for a change. The WPC11 ver. 4 has a Maxim RF section. My no-name rtw with Philips RF section still does not receive any packets. Keep access-level (analog params > config[0123] registers > none) in sc_access. Add rtw_set_access for changing the access level. Make rtw_continuous_tx_enable and other subroutines use rtw_set_access instead of rtw_config0123_enable and rtw_anaparm_enable. Factor part of the chip-reset code into rtw_chip_reset1. Change the 'struct foo (*bar)[N]'-style arguments to 'struct foo *bar'-style arguments. Consolidate software/hardware Tx/Rx ring setup in rtw_hwring_setup, rtw_swring_setup. Add a new constant, SA2400_OPMODE_DEFAULTS, for the bits that we *always* set in the SA2400 OPMODE register. Factor some code out into rtw_sa2400_calibrate. (Inspired by the Linux driver.) [] When the receiver goes into underrun/overflow state, call a new subroutine, rtw_kick() that stops the Rx/Tx processes, resets the chip, reinitializes the Tx/Rx rings, and restarts Rx/Tx processes. (Inspired by the Linux driver.) [] In rtw_intr_rx, check for too-short packets before calling ieee80211_find_rxnode. I believe this will prevent a repeat of the MCHK exception I saw once on macppc. [] Use seconds-elapased as well as microseconds-elapsed to set the next "due date" for the timeout interrupt. This keeps the driver from programming the timeout to expire too early. [] In rtw_intr, read RTW_ISR at most 10 times, then get out. If the interface is not enabled (RTW_F_ENABLED), then get out. [] In rtw_stop, get out if the interface is not enabled (RTW_F_ENABLED). Block IPL_NET interrupts. Don't read/write any registers if the interface is invalid (RTW_F_INVALID). [] Call rtw_stop in rtw_detach.
2004-12-12 09:37:59 +03:00
rtw_set_access(sc, RTW_ACCESS_ANAPARM);
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
Realtek sample sources are not consistent about the "Digital PHY" bit in the SROM. It seems as if it is set to 1 when the PHY is *analog*, not *digital*. Fix my sources. In rtw_intr_rx, use units of 500kb/s instead of 100kb/s for rate, to be consistent with net80211's expectations. Polish up some debugging ugly messages. Dump raw 802.11 packets if IFF_DEBUG|IFF_LINK2 and RTW_DEBUG is defined. Polish power-state (on/sleep/off) handling. Especially improve support for RFMD (totally untested) and Maxim. For Philips, take the Digital PHY property into account. Call the net80211 watchdog function from rtw_watchdog, so that we scan again if auth/assoc fails. Be a little more cautious about writing register[RTW_TPPOLL], since other drivers are.... Don't frob the high/low-priority queues right now, since I don't use them. Add rtw_join_bss which programs the card with the BSSID and other properties of a BSS. Use it on state transitions. Factor out rtw_set_nettype. Make rtw_recv_beacon call ieee80211_recv_mgmt instead of dropping beacons on the floor! TBD IBSS merges. Change some rtw_debug=2 printfs to rtw_debug=3 (RTW_DPRINTF3) printfs so the console doesn't get spammed so badly at rtw_debug=2. Change some debugging printfs to RTW_DPRINTFs. E.g., print the "RF programming method" only if debugging is enabled.
2004-12-21 02:05:41 +03:00
(*sc->sc_pwrstate_cb)(regs, power, before_rf, digphy);
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
Miscellaneous changes. Details below. Important changes flagged with []. Using the driver with my Linksys WPC11 ver. 4, it seems to be receiving packets for a change. The WPC11 ver. 4 has a Maxim RF section. My no-name rtw with Philips RF section still does not receive any packets. Keep access-level (analog params > config[0123] registers > none) in sc_access. Add rtw_set_access for changing the access level. Make rtw_continuous_tx_enable and other subroutines use rtw_set_access instead of rtw_config0123_enable and rtw_anaparm_enable. Factor part of the chip-reset code into rtw_chip_reset1. Change the 'struct foo (*bar)[N]'-style arguments to 'struct foo *bar'-style arguments. Consolidate software/hardware Tx/Rx ring setup in rtw_hwring_setup, rtw_swring_setup. Add a new constant, SA2400_OPMODE_DEFAULTS, for the bits that we *always* set in the SA2400 OPMODE register. Factor some code out into rtw_sa2400_calibrate. (Inspired by the Linux driver.) [] When the receiver goes into underrun/overflow state, call a new subroutine, rtw_kick() that stops the Rx/Tx processes, resets the chip, reinitializes the Tx/Rx rings, and restarts Rx/Tx processes. (Inspired by the Linux driver.) [] In rtw_intr_rx, check for too-short packets before calling ieee80211_find_rxnode. I believe this will prevent a repeat of the MCHK exception I saw once on macppc. [] Use seconds-elapased as well as microseconds-elapsed to set the next "due date" for the timeout interrupt. This keeps the driver from programming the timeout to expire too early. [] In rtw_intr, read RTW_ISR at most 10 times, then get out. If the interface is not enabled (RTW_F_ENABLED), then get out. [] In rtw_stop, get out if the interface is not enabled (RTW_F_ENABLED). Block IPL_NET interrupts. Don't read/write any registers if the interface is invalid (RTW_F_INVALID). [] Call rtw_stop in rtw_detach.
2004-12-12 09:37:59 +03:00
rtw_set_access(sc, RTW_ACCESS_NONE);
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
return;
}
static int
rtw_pwrstate(struct rtw_softc *sc, enum rtw_pwrstate power)
{
int rc;
Change rtw_debug from a debug level to a debug mask. Add a lot of debug flags. From Linux: handle an RTL8180 bug. Sometimes the NIC skips from the middle of the ring to the 0th rx descriptor. Now the driver resynchronizes. Handle a receive descriptor underrun or Rx FIFO overflow condition in the way that the Linux driver does. This kind of seems like overkill, but whatever. Protect rtw_ioctl with splnet(). Do not load a tx descriptor with a buffer shorter than 4 bytes. Handle a transmit timeout less disruptively.
2004-12-25 09:58:37 +03:00
RTW_DPRINTF(RTW_DEBUG_PWR,
("%s: %s->%s\n", __func__,
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
rtw_pwrstate_string(sc->sc_pwrstate), rtw_pwrstate_string(power)));
if (sc->sc_pwrstate == power)
return 0;
Realtek sample sources are not consistent about the "Digital PHY" bit in the SROM. It seems as if it is set to 1 when the PHY is *analog*, not *digital*. Fix my sources. In rtw_intr_rx, use units of 500kb/s instead of 100kb/s for rate, to be consistent with net80211's expectations. Polish up some debugging ugly messages. Dump raw 802.11 packets if IFF_DEBUG|IFF_LINK2 and RTW_DEBUG is defined. Polish power-state (on/sleep/off) handling. Especially improve support for RFMD (totally untested) and Maxim. For Philips, take the Digital PHY property into account. Call the net80211 watchdog function from rtw_watchdog, so that we scan again if auth/assoc fails. Be a little more cautious about writing register[RTW_TPPOLL], since other drivers are.... Don't frob the high/low-priority queues right now, since I don't use them. Add rtw_join_bss which programs the card with the BSSID and other properties of a BSS. Use it on state transitions. Factor out rtw_set_nettype. Make rtw_recv_beacon call ieee80211_recv_mgmt instead of dropping beacons on the floor! TBD IBSS merges. Change some rtw_debug=2 printfs to rtw_debug=3 (RTW_DPRINTF3) printfs so the console doesn't get spammed so badly at rtw_debug=2. Change some debugging printfs to RTW_DPRINTFs. E.g., print the "RF programming method" only if debugging is enabled.
2004-12-21 02:05:41 +03:00
rtw_pwrstate0(sc, power, 1, sc->sc_flags & RTW_F_DIGPHY);
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
rc = rtw_rf_pwrstate(sc->sc_rf, power);
Realtek sample sources are not consistent about the "Digital PHY" bit in the SROM. It seems as if it is set to 1 when the PHY is *analog*, not *digital*. Fix my sources. In rtw_intr_rx, use units of 500kb/s instead of 100kb/s for rate, to be consistent with net80211's expectations. Polish up some debugging ugly messages. Dump raw 802.11 packets if IFF_DEBUG|IFF_LINK2 and RTW_DEBUG is defined. Polish power-state (on/sleep/off) handling. Especially improve support for RFMD (totally untested) and Maxim. For Philips, take the Digital PHY property into account. Call the net80211 watchdog function from rtw_watchdog, so that we scan again if auth/assoc fails. Be a little more cautious about writing register[RTW_TPPOLL], since other drivers are.... Don't frob the high/low-priority queues right now, since I don't use them. Add rtw_join_bss which programs the card with the BSSID and other properties of a BSS. Use it on state transitions. Factor out rtw_set_nettype. Make rtw_recv_beacon call ieee80211_recv_mgmt instead of dropping beacons on the floor! TBD IBSS merges. Change some rtw_debug=2 printfs to rtw_debug=3 (RTW_DPRINTF3) printfs so the console doesn't get spammed so badly at rtw_debug=2. Change some debugging printfs to RTW_DPRINTFs. E.g., print the "RF programming method" only if debugging is enabled.
2004-12-21 02:05:41 +03:00
rtw_pwrstate0(sc, power, 0, sc->sc_flags & RTW_F_DIGPHY);
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
switch (power) {
case RTW_ON:
At last, I have rtw w/ Philips RF receiving packets. I added some sysctls to aid debugging: * hw.rtw.debug -- enable debugging * hw.rtw.flush_rfio -- Linux voodoo: possibly makes the MAC "flush" bits down the serial bus to the RF * hw.rtw.host_rfio: force the host to bang bits to the RF, instead of the MAC banging bits * hw.rtw.rfio_delay: after telling the MAC to bang bits to the RF front-end, delay rfio_delay microseconds. * hw.rtw.rfprog_fallback: there is this notion of the "RF programming method." I believe the choice influences the polarity/timing of the serial bus used to program the RF front-end. I know the correct choice for Intersil/RFMD/Philips front-ends, only. For all other front-ends, I "fallback" to rfprog_fallback. Make rtw_txdac_enable take an rtw_softc argument. I will probably revert this change. Add some Linux voodoo to rtw_continuous_tx_enable. I will probably revert this change. Important: add rtw_set_rfprog, which sets the correct RF programming method. This change and the following change are probably responsible for making the Philips RF work. Important: RTW_CONFIG1 is an 8-bit register, treat it that way! Important: RTW_BRSR is 16-bit, RTW_CRCOUNT, RTW_PHYDELAY, and RTW_MSR are 8-bit: treat them that way! Vastly simplify rtw_resume_ticks. Note to self: set the LED state to match the power state. Hedge against the possibility that RTW_MSR is protected as RTW_CONFIG[0123] are, meanwhile reworking that section of rtw_init a little. Add sc_anaparm, which isn't used, yet....
2004-12-13 03:48:02 +03:00
/* TBD set LEDs */
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
break;
case RTW_SLEEP:
/* TBD */
break;
case RTW_OFF:
/* TBD */
break;
}
if (rc == 0)
sc->sc_pwrstate = power;
else
sc->sc_pwrstate = RTW_OFF;
return rc;
}
static int
rtw_tune(struct rtw_softc *sc)
{
struct ieee80211com *ic = &sc->sc_ic;
u_int chan;
int rc;
int antdiv = sc->sc_flags & RTW_F_ANTDIV,
dflantb = sc->sc_flags & RTW_F_DFLANTB;
KASSERT(ic->ic_bss->ni_chan != NULL);
chan = ieee80211_chan2ieee(ic, ic->ic_bss->ni_chan);
if (chan == IEEE80211_CHAN_ANY)
panic("%s: chan == IEEE80211_CHAN_ANY\n", __func__);
if (chan == sc->sc_cur_chan) {
Change rtw_debug from a debug level to a debug mask. Add a lot of debug flags. From Linux: handle an RTL8180 bug. Sometimes the NIC skips from the middle of the ring to the 0th rx descriptor. Now the driver resynchronizes. Handle a receive descriptor underrun or Rx FIFO overflow condition in the way that the Linux driver does. This kind of seems like overkill, but whatever. Protect rtw_ioctl with splnet(). Do not load a tx descriptor with a buffer shorter than 4 bytes. Handle a transmit timeout less disruptively.
2004-12-25 09:58:37 +03:00
RTW_DPRINTF(RTW_DEBUG_TUNE,
("%s: already tuned chan #%d\n", __func__, chan));
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
return 0;
}
rtw_suspend_ticks(sc);
rtw_io_enable(&sc->sc_regs, RTW_CR_RE | RTW_CR_TE, 0);
/* TBD wait for Tx to complete */
KASSERT((sc->sc_flags & RTW_F_ENABLED) != 0);
if ((rc = rtw_phy_init(&sc->sc_regs, sc->sc_rf,
rtw_chan2txpower(&sc->sc_srom, ic, ic->ic_bss->ni_chan),
sc->sc_csthr, ic->ic_bss->ni_chan->ic_freq, antdiv,
dflantb, RTW_ON)) != 0) {
/* XXX condition on powersaving */
printf("%s: phy init failed\n", sc->sc_dev.dv_xname);
}
sc->sc_cur_chan = chan;
rtw_io_enable(&sc->sc_regs, RTW_CR_RE | RTW_CR_TE, 1);
rtw_resume_ticks(sc);
return rc;
}
void
rtw_disable(struct rtw_softc *sc)
{
int rc;
if ((sc->sc_flags & RTW_F_ENABLED) == 0)
return;
/* turn off PHY */
In rtw_detach, flag the rtw as "invalid" (RTW_F_INVALID). Do not try to power-down the RF section of an invalid rtw. The radiotap signal quality field is 16, not 8, bits wide. Convert the byte order.
2004-12-29 22:41:04 +03:00
if ((sc->sc_flags & RTW_F_INVALID) == 0 &&
(rc = rtw_pwrstate(sc, RTW_OFF)) != 0) {
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
printf("%s: failed to turn off PHY (%d)\n",
sc->sc_dev.dv_xname, rc);
In rtw_detach, flag the rtw as "invalid" (RTW_F_INVALID). Do not try to power-down the RF section of an invalid rtw. The radiotap signal quality field is 16, not 8, bits wide. Convert the byte order.
2004-12-29 22:41:04 +03:00
}
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
if (sc->sc_disable != NULL)
(*sc->sc_disable)(sc);
sc->sc_flags &= ~RTW_F_ENABLED;
}
int
rtw_enable(struct rtw_softc *sc)
{
if ((sc->sc_flags & RTW_F_ENABLED) == 0) {
if (sc->sc_enable != NULL && (*sc->sc_enable)(sc) != 0) {
printf("%s: device enable failed\n",
sc->sc_dev.dv_xname);
return (EIO);
}
sc->sc_flags |= RTW_F_ENABLED;
}
return (0);
}
static void
rtw_transmit_config(struct rtw_regs *regs)
{
u_int32_t tcr;
tcr = RTW_READ(regs, RTW_TCR);
Realtek sample sources are not consistent about the "Digital PHY" bit in the SROM. It seems as if it is set to 1 when the PHY is *analog*, not *digital*. Fix my sources. In rtw_intr_rx, use units of 500kb/s instead of 100kb/s for rate, to be consistent with net80211's expectations. Polish up some debugging ugly messages. Dump raw 802.11 packets if IFF_DEBUG|IFF_LINK2 and RTW_DEBUG is defined. Polish power-state (on/sleep/off) handling. Especially improve support for RFMD (totally untested) and Maxim. For Philips, take the Digital PHY property into account. Call the net80211 watchdog function from rtw_watchdog, so that we scan again if auth/assoc fails. Be a little more cautious about writing register[RTW_TPPOLL], since other drivers are.... Don't frob the high/low-priority queues right now, since I don't use them. Add rtw_join_bss which programs the card with the BSSID and other properties of a BSS. Use it on state transitions. Factor out rtw_set_nettype. Make rtw_recv_beacon call ieee80211_recv_mgmt instead of dropping beacons on the floor! TBD IBSS merges. Change some rtw_debug=2 printfs to rtw_debug=3 (RTW_DPRINTF3) printfs so the console doesn't get spammed so badly at rtw_debug=2. Change some debugging printfs to RTW_DPRINTFs. E.g., print the "RF programming method" only if debugging is enabled.
2004-12-21 02:05:41 +03:00
tcr |= RTW_TCR_CWMIN;
tcr &= ~RTW_TCR_MXDMA_MASK;
tcr |= RTW_TCR_MXDMA_256;
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
tcr |= RTW_TCR_SAT; /* send ACK as fast as possible */
tcr &= ~RTW_TCR_LBK_MASK;
tcr |= RTW_TCR_LBK_NORMAL; /* normal operating mode */
/* set short/long retry limits */
tcr &= ~(RTW_TCR_SRL_MASK|RTW_TCR_LRL_MASK);
Realtek sample sources are not consistent about the "Digital PHY" bit in the SROM. It seems as if it is set to 1 when the PHY is *analog*, not *digital*. Fix my sources. In rtw_intr_rx, use units of 500kb/s instead of 100kb/s for rate, to be consistent with net80211's expectations. Polish up some debugging ugly messages. Dump raw 802.11 packets if IFF_DEBUG|IFF_LINK2 and RTW_DEBUG is defined. Polish power-state (on/sleep/off) handling. Especially improve support for RFMD (totally untested) and Maxim. For Philips, take the Digital PHY property into account. Call the net80211 watchdog function from rtw_watchdog, so that we scan again if auth/assoc fails. Be a little more cautious about writing register[RTW_TPPOLL], since other drivers are.... Don't frob the high/low-priority queues right now, since I don't use them. Add rtw_join_bss which programs the card with the BSSID and other properties of a BSS. Use it on state transitions. Factor out rtw_set_nettype. Make rtw_recv_beacon call ieee80211_recv_mgmt instead of dropping beacons on the floor! TBD IBSS merges. Change some rtw_debug=2 printfs to rtw_debug=3 (RTW_DPRINTF3) printfs so the console doesn't get spammed so badly at rtw_debug=2. Change some debugging printfs to RTW_DPRINTFs. E.g., print the "RF programming method" only if debugging is enabled.
2004-12-21 02:05:41 +03:00
tcr |= LSHIFT(4, RTW_TCR_SRL_MASK) | LSHIFT(4, RTW_TCR_LRL_MASK);
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
Do not indicate to the NIC that the host has appended the CRC32 to Tx frames.
2004-12-23 08:50:24 +03:00
tcr &= ~RTW_TCR_CRC; /* NIC appends CRC32 */
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
RTW_WRITE(regs, RTW_TCR, tcr);
Add some bus barriers. Remark where they are needed, e.g., read_eeprom. Change some printfs to RTW_DPRINTF2s to quiet rtw(4) when debugging is disabled.
2004-12-20 04:13:45 +03:00
RTW_SYNC(regs, RTW_TCR, RTW_TCR);
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
}
static __inline void
rtw_enable_interrupts(struct rtw_softc *sc)
{
struct rtw_regs *regs = &sc->sc_regs;
sc->sc_inten = RTW_INTR_RX|RTW_INTR_TX|RTW_INTR_BEACON|RTW_INTR_ATIMINT;
sc->sc_inten |= RTW_INTR_IOERROR|RTW_INTR_TIMEOUT;
RTW_WRITE16(regs, RTW_IMR, sc->sc_inten);
Add some bus barriers. Remark where they are needed, e.g., read_eeprom. Change some printfs to RTW_DPRINTF2s to quiet rtw(4) when debugging is disabled.
2004-12-20 04:13:45 +03:00
RTW_WBW(regs, RTW_IMR, RTW_ISR);
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
RTW_WRITE16(regs, RTW_ISR, 0xffff);
Add some bus barriers. Remark where they are needed, e.g., read_eeprom. Change some printfs to RTW_DPRINTF2s to quiet rtw(4) when debugging is disabled.
2004-12-20 04:13:45 +03:00
RTW_SYNC(regs, RTW_IMR, RTW_ISR);
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
/* XXX necessary? */
if (sc->sc_intr_ack != NULL)
(*sc->sc_intr_ack)(regs);
}
Realtek sample sources are not consistent about the "Digital PHY" bit in the SROM. It seems as if it is set to 1 when the PHY is *analog*, not *digital*. Fix my sources. In rtw_intr_rx, use units of 500kb/s instead of 100kb/s for rate, to be consistent with net80211's expectations. Polish up some debugging ugly messages. Dump raw 802.11 packets if IFF_DEBUG|IFF_LINK2 and RTW_DEBUG is defined. Polish power-state (on/sleep/off) handling. Especially improve support for RFMD (totally untested) and Maxim. For Philips, take the Digital PHY property into account. Call the net80211 watchdog function from rtw_watchdog, so that we scan again if auth/assoc fails. Be a little more cautious about writing register[RTW_TPPOLL], since other drivers are.... Don't frob the high/low-priority queues right now, since I don't use them. Add rtw_join_bss which programs the card with the BSSID and other properties of a BSS. Use it on state transitions. Factor out rtw_set_nettype. Make rtw_recv_beacon call ieee80211_recv_mgmt instead of dropping beacons on the floor! TBD IBSS merges. Change some rtw_debug=2 printfs to rtw_debug=3 (RTW_DPRINTF3) printfs so the console doesn't get spammed so badly at rtw_debug=2. Change some debugging printfs to RTW_DPRINTFs. E.g., print the "RF programming method" only if debugging is enabled.
2004-12-21 02:05:41 +03:00
static void
rtw_set_nettype(struct rtw_softc *sc, enum ieee80211_opmode opmode)
{
uint8_t msr;
/* I'm guessing that MSR is protected as CONFIG[0123] are. */
rtw_set_access(sc, RTW_ACCESS_CONFIG);
msr = RTW_READ8(&sc->sc_regs, RTW_MSR) & ~RTW_MSR_NETYPE_MASK;
switch (opmode) {
case IEEE80211_M_AHDEMO:
case IEEE80211_M_IBSS:
msr |= RTW_MSR_NETYPE_ADHOC_OK;
break;
case IEEE80211_M_HOSTAP:
msr |= RTW_MSR_NETYPE_AP_OK;
break;
case IEEE80211_M_MONITOR:
/* XXX */
msr |= RTW_MSR_NETYPE_NOLINK;
break;
case IEEE80211_M_STA:
msr |= RTW_MSR_NETYPE_INFRA_OK;
break;
}
RTW_WRITE8(&sc->sc_regs, RTW_MSR, msr);
rtw_set_access(sc, RTW_ACCESS_NONE);
}
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
/* XXX is the endianness correct? test. */
#define rtw_calchash(addr) \
(ether_crc32_le((addr), IEEE80211_ADDR_LEN) & BITS(5, 0))
static void
rtw_pktfilt_load(struct rtw_softc *sc)
{
struct rtw_regs *regs = &sc->sc_regs;
struct ieee80211com *ic = &sc->sc_ic;
struct ethercom *ec = &ic->ic_ec;
struct ifnet *ifp = &sc->sc_ic.ic_if;
int hash;
u_int32_t hashes[2] = { 0, 0 };
struct ether_multi *enm;
struct ether_multistep step;
/* XXX might be necessary to stop Rx/Tx engines while setting filters */
Realtek sample sources are not consistent about the "Digital PHY" bit in the SROM. It seems as if it is set to 1 when the PHY is *analog*, not *digital*. Fix my sources. In rtw_intr_rx, use units of 500kb/s instead of 100kb/s for rate, to be consistent with net80211's expectations. Polish up some debugging ugly messages. Dump raw 802.11 packets if IFF_DEBUG|IFF_LINK2 and RTW_DEBUG is defined. Polish power-state (on/sleep/off) handling. Especially improve support for RFMD (totally untested) and Maxim. For Philips, take the Digital PHY property into account. Call the net80211 watchdog function from rtw_watchdog, so that we scan again if auth/assoc fails. Be a little more cautious about writing register[RTW_TPPOLL], since other drivers are.... Don't frob the high/low-priority queues right now, since I don't use them. Add rtw_join_bss which programs the card with the BSSID and other properties of a BSS. Use it on state transitions. Factor out rtw_set_nettype. Make rtw_recv_beacon call ieee80211_recv_mgmt instead of dropping beacons on the floor! TBD IBSS merges. Change some rtw_debug=2 printfs to rtw_debug=3 (RTW_DPRINTF3) printfs so the console doesn't get spammed so badly at rtw_debug=2. Change some debugging printfs to RTW_DPRINTFs. E.g., print the "RF programming method" only if debugging is enabled.
2004-12-21 02:05:41 +03:00
#define RTW_RCR_MONITOR (RTW_RCR_ACRC32|RTW_RCR_APM|RTW_RCR_AAP|RTW_RCR_AB|RTW_RCR_ACF | RTW_RCR_AICV | RTW_RCR_ACRC32)
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
if (ic->ic_opmode == IEEE80211_M_MONITOR)
sc->sc_rcr |= RTW_RCR_MONITOR;
else
sc->sc_rcr &= ~RTW_RCR_MONITOR;
/* XXX reference sources BEGIN */
Realtek sample sources are not consistent about the "Digital PHY" bit in the SROM. It seems as if it is set to 1 when the PHY is *analog*, not *digital*. Fix my sources. In rtw_intr_rx, use units of 500kb/s instead of 100kb/s for rate, to be consistent with net80211's expectations. Polish up some debugging ugly messages. Dump raw 802.11 packets if IFF_DEBUG|IFF_LINK2 and RTW_DEBUG is defined. Polish power-state (on/sleep/off) handling. Especially improve support for RFMD (totally untested) and Maxim. For Philips, take the Digital PHY property into account. Call the net80211 watchdog function from rtw_watchdog, so that we scan again if auth/assoc fails. Be a little more cautious about writing register[RTW_TPPOLL], since other drivers are.... Don't frob the high/low-priority queues right now, since I don't use them. Add rtw_join_bss which programs the card with the BSSID and other properties of a BSS. Use it on state transitions. Factor out rtw_set_nettype. Make rtw_recv_beacon call ieee80211_recv_mgmt instead of dropping beacons on the floor! TBD IBSS merges. Change some rtw_debug=2 printfs to rtw_debug=3 (RTW_DPRINTF3) printfs so the console doesn't get spammed so badly at rtw_debug=2. Change some debugging printfs to RTW_DPRINTFs. E.g., print the "RF programming method" only if debugging is enabled.
2004-12-21 02:05:41 +03:00
sc->sc_rcr |= RTW_RCR_ENMARP;
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
sc->sc_rcr |= RTW_RCR_AB | RTW_RCR_AM | RTW_RCR_APM;
#if 0
/* receive broadcasts in our BSS */
sc->sc_rcr |= RTW_RCR_ADD3;
#endif
/* XXX reference sources END */
/* receive pwrmgmt frames. */
sc->sc_rcr |= RTW_RCR_APWRMGT;
/* receive mgmt/ctrl/data frames. */
Realtek sample sources are not consistent about the "Digital PHY" bit in the SROM. It seems as if it is set to 1 when the PHY is *analog*, not *digital*. Fix my sources. In rtw_intr_rx, use units of 500kb/s instead of 100kb/s for rate, to be consistent with net80211's expectations. Polish up some debugging ugly messages. Dump raw 802.11 packets if IFF_DEBUG|IFF_LINK2 and RTW_DEBUG is defined. Polish power-state (on/sleep/off) handling. Especially improve support for RFMD (totally untested) and Maxim. For Philips, take the Digital PHY property into account. Call the net80211 watchdog function from rtw_watchdog, so that we scan again if auth/assoc fails. Be a little more cautious about writing register[RTW_TPPOLL], since other drivers are.... Don't frob the high/low-priority queues right now, since I don't use them. Add rtw_join_bss which programs the card with the BSSID and other properties of a BSS. Use it on state transitions. Factor out rtw_set_nettype. Make rtw_recv_beacon call ieee80211_recv_mgmt instead of dropping beacons on the floor! TBD IBSS merges. Change some rtw_debug=2 printfs to rtw_debug=3 (RTW_DPRINTF3) printfs so the console doesn't get spammed so badly at rtw_debug=2. Change some debugging printfs to RTW_DPRINTFs. E.g., print the "RF programming method" only if debugging is enabled.
2004-12-21 02:05:41 +03:00
sc->sc_rcr |= RTW_RCR_ADF | RTW_RCR_AMF;
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
/* initialize Rx DMA threshold, Tx DMA burst size */
sc->sc_rcr |= RTW_RCR_RXFTH_WHOLE | RTW_RCR_MXDMA_1024;
ifp->if_flags &= ~IFF_ALLMULTI;
if (ifp->if_flags & IFF_PROMISC) {
sc->sc_rcr |= RTW_RCR_AB; /* accept all broadcast */
allmulti:
ifp->if_flags |= IFF_ALLMULTI;
goto setit;
}
/*
* Program the 64-bit multicast hash filter.
*/
ETHER_FIRST_MULTI(step, ec, enm);
while (enm != NULL) {
/* XXX */
if (memcmp(enm->enm_addrlo, enm->enm_addrhi,
ETHER_ADDR_LEN) != 0)
goto allmulti;
hash = rtw_calchash(enm->enm_addrlo);
hashes[hash >> 5] |= 1 << (hash & 0x1f);
ETHER_NEXT_MULTI(step, enm);
}
if (ifp->if_flags & IFF_BROADCAST) {
hash = rtw_calchash(etherbroadcastaddr);
hashes[hash >> 5] |= 1 << (hash & 0x1f);
}
/* all bits set => hash is useless */
if (~(hashes[0] & hashes[1]) == 0)
goto allmulti;
setit:
if (ifp->if_flags & IFF_ALLMULTI)
sc->sc_rcr |= RTW_RCR_AM; /* accept all multicast */
if (ic->ic_state == IEEE80211_S_SCAN)
sc->sc_rcr |= RTW_RCR_AB; /* accept all broadcast */
hashes[0] = hashes[1] = 0xffffffff;
RTW_WRITE(regs, RTW_MAR0, hashes[0]);
RTW_WRITE(regs, RTW_MAR1, hashes[1]);
RTW_WRITE(regs, RTW_RCR, sc->sc_rcr);
RTW_SYNC(regs, RTW_MAR0, RTW_RCR); /* RTW_MAR0 < RTW_MAR1 < RTW_RCR */
Change rtw_debug from a debug level to a debug mask. Add a lot of debug flags. From Linux: handle an RTL8180 bug. Sometimes the NIC skips from the middle of the ring to the 0th rx descriptor. Now the driver resynchronizes. Handle a receive descriptor underrun or Rx FIFO overflow condition in the way that the Linux driver does. This kind of seems like overkill, but whatever. Protect rtw_ioctl with splnet(). Do not load a tx descriptor with a buffer shorter than 4 bytes. Handle a transmit timeout less disruptively.
2004-12-25 09:58:37 +03:00
DPRINTF(sc, RTW_DEBUG_PKTFILT,
("%s: RTW_MAR0 %08x RTW_MAR1 %08x RTW_RCR %08x\n",
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
sc->sc_dev.dv_xname, RTW_READ(regs, RTW_MAR0),
RTW_READ(regs, RTW_MAR1), RTW_READ(regs, RTW_RCR)));
return;
}
Change rtw_debug from a debug level to a debug mask. Add a lot of debug flags. From Linux: handle an RTL8180 bug. Sometimes the NIC skips from the middle of the ring to the 0th rx descriptor. Now the driver resynchronizes. Handle a receive descriptor underrun or Rx FIFO overflow condition in the way that the Linux driver does. This kind of seems like overkill, but whatever. Protect rtw_ioctl with splnet(). Do not load a tx descriptor with a buffer shorter than 4 bytes. Handle a transmit timeout less disruptively.
2004-12-25 09:58:37 +03:00
/* Must be called at splnet. */
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
static int
rtw_init(struct ifnet *ifp)
{
struct rtw_softc *sc = (struct rtw_softc *)ifp->if_softc;
struct ieee80211com *ic = &sc->sc_ic;
struct rtw_regs *regs = &sc->sc_regs;
At last, I have rtw w/ Philips RF receiving packets. I added some sysctls to aid debugging: * hw.rtw.debug -- enable debugging * hw.rtw.flush_rfio -- Linux voodoo: possibly makes the MAC "flush" bits down the serial bus to the RF * hw.rtw.host_rfio: force the host to bang bits to the RF, instead of the MAC banging bits * hw.rtw.rfio_delay: after telling the MAC to bang bits to the RF front-end, delay rfio_delay microseconds. * hw.rtw.rfprog_fallback: there is this notion of the "RF programming method." I believe the choice influences the polarity/timing of the serial bus used to program the RF front-end. I know the correct choice for Intersil/RFMD/Philips front-ends, only. For all other front-ends, I "fallback" to rfprog_fallback. Make rtw_txdac_enable take an rtw_softc argument. I will probably revert this change. Add some Linux voodoo to rtw_continuous_tx_enable. I will probably revert this change. Important: add rtw_set_rfprog, which sets the correct RF programming method. This change and the following change are probably responsible for making the Philips RF work. Important: RTW_CONFIG1 is an 8-bit register, treat it that way! Important: RTW_BRSR is 16-bit, RTW_CRCOUNT, RTW_PHYDELAY, and RTW_MSR are 8-bit: treat them that way! Vastly simplify rtw_resume_ticks. Note to self: set the LED state to match the power state. Hedge against the possibility that RTW_MSR is protected as RTW_CONFIG[0123] are, meanwhile reworking that section of rtw_init a little. Add sc_anaparm, which isn't used, yet....
2004-12-13 03:48:02 +03:00
int rc = 0;
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
if ((rc = rtw_enable(sc)) != 0)
goto out;
/* Cancel pending I/O and reset. */
rtw_stop(ifp, 0);
ic->ic_bss->ni_chan = ic->ic_ibss_chan;
Change rtw_debug from a debug level to a debug mask. Add a lot of debug flags. From Linux: handle an RTL8180 bug. Sometimes the NIC skips from the middle of the ring to the 0th rx descriptor. Now the driver resynchronizes. Handle a receive descriptor underrun or Rx FIFO overflow condition in the way that the Linux driver does. This kind of seems like overkill, but whatever. Protect rtw_ioctl with splnet(). Do not load a tx descriptor with a buffer shorter than 4 bytes. Handle a transmit timeout less disruptively.
2004-12-25 09:58:37 +03:00
DPRINTF(sc, RTW_DEBUG_TUNE, ("%s: channel %d freq %d flags 0x%04x\n",
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
__func__, ieee80211_chan2ieee(ic, ic->ic_bss->ni_chan),
ic->ic_bss->ni_chan->ic_freq, ic->ic_bss->ni_chan->ic_flags));
if ((rc = rtw_pwrstate(sc, RTW_OFF)) != 0)
goto out;
If defined(RTW_DEBUG), provide a sysctl (hw.rtw.rxbufs_limit) for limiting the number of rx buffers an rtw may allocate. Use this sysctl to test the code that copes with buffer exhaustion. Allocate at most RTW_RXQLEN rx buffers, stopping at the sysctl limit. Record in sc_nrxdesc how many were allocated, and put the end-of-ring indication on sc_rxdesc[sc_nrxdesc - 1]. In rtw_init, if no rx buffers could be allocated, log a complaint, clear IFF_RUNNING, and exit with an error. Many changes to accomodate a short rx ring, mainly of the "add a rx-ring length argument" variety. XXX I really should consolidate all of the rx ring variables in one struct and pass that to the rx-ring subroutines. Bug fix: after calling MCLGET, use the (m->m_flags & M_EXT) idiom to check for success, instead of m != NULL. Bug fix: at the top of rtw_start, if IFF_RUNNING is not set, or IFF_OACTIVE is, get out.
2004-12-29 01:21:15 +03:00
if ((rc = rtw_swring_setup(sc)) != 0)
goto out;
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
rtw_transmit_config(regs);
Miscellaneous changes. Details below. Important changes flagged with []. Using the driver with my Linksys WPC11 ver. 4, it seems to be receiving packets for a change. The WPC11 ver. 4 has a Maxim RF section. My no-name rtw with Philips RF section still does not receive any packets. Keep access-level (analog params > config[0123] registers > none) in sc_access. Add rtw_set_access for changing the access level. Make rtw_continuous_tx_enable and other subroutines use rtw_set_access instead of rtw_config0123_enable and rtw_anaparm_enable. Factor part of the chip-reset code into rtw_chip_reset1. Change the 'struct foo (*bar)[N]'-style arguments to 'struct foo *bar'-style arguments. Consolidate software/hardware Tx/Rx ring setup in rtw_hwring_setup, rtw_swring_setup. Add a new constant, SA2400_OPMODE_DEFAULTS, for the bits that we *always* set in the SA2400 OPMODE register. Factor some code out into rtw_sa2400_calibrate. (Inspired by the Linux driver.) [] When the receiver goes into underrun/overflow state, call a new subroutine, rtw_kick() that stops the Rx/Tx processes, resets the chip, reinitializes the Tx/Rx rings, and restarts Rx/Tx processes. (Inspired by the Linux driver.) [] In rtw_intr_rx, check for too-short packets before calling ieee80211_find_rxnode. I believe this will prevent a repeat of the MCHK exception I saw once on macppc. [] Use seconds-elapased as well as microseconds-elapsed to set the next "due date" for the timeout interrupt. This keeps the driver from programming the timeout to expire too early. [] In rtw_intr, read RTW_ISR at most 10 times, then get out. If the interface is not enabled (RTW_F_ENABLED), then get out. [] In rtw_stop, get out if the interface is not enabled (RTW_F_ENABLED). Block IPL_NET interrupts. Don't read/write any registers if the interface is invalid (RTW_F_INVALID). [] Call rtw_stop in rtw_detach.
2004-12-12 09:37:59 +03:00
rtw_set_access(sc, RTW_ACCESS_CONFIG);
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
At last, I have rtw w/ Philips RF receiving packets. I added some sysctls to aid debugging: * hw.rtw.debug -- enable debugging * hw.rtw.flush_rfio -- Linux voodoo: possibly makes the MAC "flush" bits down the serial bus to the RF * hw.rtw.host_rfio: force the host to bang bits to the RF, instead of the MAC banging bits * hw.rtw.rfio_delay: after telling the MAC to bang bits to the RF front-end, delay rfio_delay microseconds. * hw.rtw.rfprog_fallback: there is this notion of the "RF programming method." I believe the choice influences the polarity/timing of the serial bus used to program the RF front-end. I know the correct choice for Intersil/RFMD/Philips front-ends, only. For all other front-ends, I "fallback" to rfprog_fallback. Make rtw_txdac_enable take an rtw_softc argument. I will probably revert this change. Add some Linux voodoo to rtw_continuous_tx_enable. I will probably revert this change. Important: add rtw_set_rfprog, which sets the correct RF programming method. This change and the following change are probably responsible for making the Philips RF work. Important: RTW_CONFIG1 is an 8-bit register, treat it that way! Important: RTW_BRSR is 16-bit, RTW_CRCOUNT, RTW_PHYDELAY, and RTW_MSR are 8-bit: treat them that way! Vastly simplify rtw_resume_ticks. Note to self: set the LED state to match the power state. Hedge against the possibility that RTW_MSR is protected as RTW_CONFIG[0123] are, meanwhile reworking that section of rtw_init a little. Add sc_anaparm, which isn't used, yet....
2004-12-13 03:48:02 +03:00
RTW_WRITE8(regs, RTW_MSR, 0x0); /* no link */
Add some bus barriers. Remark where they are needed, e.g., read_eeprom. Change some printfs to RTW_DPRINTF2s to quiet rtw(4) when debugging is disabled.
2004-12-20 04:13:45 +03:00
RTW_WBW(regs, RTW_MSR, RTW_BRSR);
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
When checking if a split packet is completed, check the *last* segment, not the first. Otherwise we can overwrite parts of the TX ring that we shouldn't. Kluge the basic rate setting for now. Minor simplification to the dequeueing logic.
2004-12-27 12:40:18 +03:00
/* long PLCP header, 1Mb/2Mb basic rate */
RTW_WRITE16(regs, RTW_BRSR, RTW_BRSR_MBR8180_2MBPS);
Add some bus barriers. Remark where they are needed, e.g., read_eeprom. Change some printfs to RTW_DPRINTF2s to quiet rtw(4) when debugging is disabled.
2004-12-20 04:13:45 +03:00
RTW_SYNC(regs, RTW_BRSR, RTW_BRSR);
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
Miscellaneous changes. Details below. Important changes flagged with []. Using the driver with my Linksys WPC11 ver. 4, it seems to be receiving packets for a change. The WPC11 ver. 4 has a Maxim RF section. My no-name rtw with Philips RF section still does not receive any packets. Keep access-level (analog params > config[0123] registers > none) in sc_access. Add rtw_set_access for changing the access level. Make rtw_continuous_tx_enable and other subroutines use rtw_set_access instead of rtw_config0123_enable and rtw_anaparm_enable. Factor part of the chip-reset code into rtw_chip_reset1. Change the 'struct foo (*bar)[N]'-style arguments to 'struct foo *bar'-style arguments. Consolidate software/hardware Tx/Rx ring setup in rtw_hwring_setup, rtw_swring_setup. Add a new constant, SA2400_OPMODE_DEFAULTS, for the bits that we *always* set in the SA2400 OPMODE register. Factor some code out into rtw_sa2400_calibrate. (Inspired by the Linux driver.) [] When the receiver goes into underrun/overflow state, call a new subroutine, rtw_kick() that stops the Rx/Tx processes, resets the chip, reinitializes the Tx/Rx rings, and restarts Rx/Tx processes. (Inspired by the Linux driver.) [] In rtw_intr_rx, check for too-short packets before calling ieee80211_find_rxnode. I believe this will prevent a repeat of the MCHK exception I saw once on macppc. [] Use seconds-elapased as well as microseconds-elapsed to set the next "due date" for the timeout interrupt. This keeps the driver from programming the timeout to expire too early. [] In rtw_intr, read RTW_ISR at most 10 times, then get out. If the interface is not enabled (RTW_F_ENABLED), then get out. [] In rtw_stop, get out if the interface is not enabled (RTW_F_ENABLED). Block IPL_NET interrupts. Don't read/write any registers if the interface is invalid (RTW_F_INVALID). [] Call rtw_stop in rtw_detach.
2004-12-12 09:37:59 +03:00
rtw_set_access(sc, RTW_ACCESS_ANAPARM);
rtw_set_access(sc, RTW_ACCESS_NONE);
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
/* XXX from reference sources */
RTW_WRITE(regs, RTW_FEMR, 0xffff);
Add some bus barriers. Remark where they are needed, e.g., read_eeprom. Change some printfs to RTW_DPRINTF2s to quiet rtw(4) when debugging is disabled.
2004-12-20 04:13:45 +03:00
RTW_SYNC(regs, RTW_FEMR, RTW_FEMR);
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
At last, I have rtw w/ Philips RF receiving packets. I added some sysctls to aid debugging: * hw.rtw.debug -- enable debugging * hw.rtw.flush_rfio -- Linux voodoo: possibly makes the MAC "flush" bits down the serial bus to the RF * hw.rtw.host_rfio: force the host to bang bits to the RF, instead of the MAC banging bits * hw.rtw.rfio_delay: after telling the MAC to bang bits to the RF front-end, delay rfio_delay microseconds. * hw.rtw.rfprog_fallback: there is this notion of the "RF programming method." I believe the choice influences the polarity/timing of the serial bus used to program the RF front-end. I know the correct choice for Intersil/RFMD/Philips front-ends, only. For all other front-ends, I "fallback" to rfprog_fallback. Make rtw_txdac_enable take an rtw_softc argument. I will probably revert this change. Add some Linux voodoo to rtw_continuous_tx_enable. I will probably revert this change. Important: add rtw_set_rfprog, which sets the correct RF programming method. This change and the following change are probably responsible for making the Philips RF work. Important: RTW_CONFIG1 is an 8-bit register, treat it that way! Important: RTW_BRSR is 16-bit, RTW_CRCOUNT, RTW_PHYDELAY, and RTW_MSR are 8-bit: treat them that way! Vastly simplify rtw_resume_ticks. Note to self: set the LED state to match the power state. Hedge against the possibility that RTW_MSR is protected as RTW_CONFIG[0123] are, meanwhile reworking that section of rtw_init a little. Add sc_anaparm, which isn't used, yet....
2004-12-13 03:48:02 +03:00
rtw_set_rfprog(regs, sc->sc_rfchipid, sc->sc_dev.dv_xname);
RTW_WRITE8(regs, RTW_PHYDELAY, sc->sc_phydelay);
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
/* from Linux driver */
At last, I have rtw w/ Philips RF receiving packets. I added some sysctls to aid debugging: * hw.rtw.debug -- enable debugging * hw.rtw.flush_rfio -- Linux voodoo: possibly makes the MAC "flush" bits down the serial bus to the RF * hw.rtw.host_rfio: force the host to bang bits to the RF, instead of the MAC banging bits * hw.rtw.rfio_delay: after telling the MAC to bang bits to the RF front-end, delay rfio_delay microseconds. * hw.rtw.rfprog_fallback: there is this notion of the "RF programming method." I believe the choice influences the polarity/timing of the serial bus used to program the RF front-end. I know the correct choice for Intersil/RFMD/Philips front-ends, only. For all other front-ends, I "fallback" to rfprog_fallback. Make rtw_txdac_enable take an rtw_softc argument. I will probably revert this change. Add some Linux voodoo to rtw_continuous_tx_enable. I will probably revert this change. Important: add rtw_set_rfprog, which sets the correct RF programming method. This change and the following change are probably responsible for making the Philips RF work. Important: RTW_CONFIG1 is an 8-bit register, treat it that way! Important: RTW_BRSR is 16-bit, RTW_CRCOUNT, RTW_PHYDELAY, and RTW_MSR are 8-bit: treat them that way! Vastly simplify rtw_resume_ticks. Note to self: set the LED state to match the power state. Hedge against the possibility that RTW_MSR is protected as RTW_CONFIG[0123] are, meanwhile reworking that section of rtw_init a little. Add sc_anaparm, which isn't used, yet....
2004-12-13 03:48:02 +03:00
RTW_WRITE8(regs, RTW_CRCOUNT, RTW_CRCOUNT_MAGIC);
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
Add some bus barriers. Remark where they are needed, e.g., read_eeprom. Change some printfs to RTW_DPRINTF2s to quiet rtw(4) when debugging is disabled.
2004-12-20 04:13:45 +03:00
RTW_SYNC(regs, RTW_PHYDELAY, RTW_CRCOUNT);
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
rtw_enable_interrupts(sc);
rtw_pktfilt_load(sc);
Miscellaneous changes. Details below. Important changes flagged with []. Using the driver with my Linksys WPC11 ver. 4, it seems to be receiving packets for a change. The WPC11 ver. 4 has a Maxim RF section. My no-name rtw with Philips RF section still does not receive any packets. Keep access-level (analog params > config[0123] registers > none) in sc_access. Add rtw_set_access for changing the access level. Make rtw_continuous_tx_enable and other subroutines use rtw_set_access instead of rtw_config0123_enable and rtw_anaparm_enable. Factor part of the chip-reset code into rtw_chip_reset1. Change the 'struct foo (*bar)[N]'-style arguments to 'struct foo *bar'-style arguments. Consolidate software/hardware Tx/Rx ring setup in rtw_hwring_setup, rtw_swring_setup. Add a new constant, SA2400_OPMODE_DEFAULTS, for the bits that we *always* set in the SA2400 OPMODE register. Factor some code out into rtw_sa2400_calibrate. (Inspired by the Linux driver.) [] When the receiver goes into underrun/overflow state, call a new subroutine, rtw_kick() that stops the Rx/Tx processes, resets the chip, reinitializes the Tx/Rx rings, and restarts Rx/Tx processes. (Inspired by the Linux driver.) [] In rtw_intr_rx, check for too-short packets before calling ieee80211_find_rxnode. I believe this will prevent a repeat of the MCHK exception I saw once on macppc. [] Use seconds-elapased as well as microseconds-elapsed to set the next "due date" for the timeout interrupt. This keeps the driver from programming the timeout to expire too early. [] In rtw_intr, read RTW_ISR at most 10 times, then get out. If the interface is not enabled (RTW_F_ENABLED), then get out. [] In rtw_stop, get out if the interface is not enabled (RTW_F_ENABLED). Block IPL_NET interrupts. Don't read/write any registers if the interface is invalid (RTW_F_INVALID). [] Call rtw_stop in rtw_detach.
2004-12-12 09:37:59 +03:00
rtw_hwring_setup(sc);
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
rtw_io_enable(regs, RTW_CR_RE|RTW_CR_TE, 1);
ifp->if_flags |= IFF_RUNNING;
ic->ic_state = IEEE80211_S_INIT;
RTW_WRITE16(regs, RTW_BSSID16, 0x0);
RTW_WRITE(regs, RTW_BSSID32, 0x0);
rtw_resume_ticks(sc);
Realtek sample sources are not consistent about the "Digital PHY" bit in the SROM. It seems as if it is set to 1 when the PHY is *analog*, not *digital*. Fix my sources. In rtw_intr_rx, use units of 500kb/s instead of 100kb/s for rate, to be consistent with net80211's expectations. Polish up some debugging ugly messages. Dump raw 802.11 packets if IFF_DEBUG|IFF_LINK2 and RTW_DEBUG is defined. Polish power-state (on/sleep/off) handling. Especially improve support for RFMD (totally untested) and Maxim. For Philips, take the Digital PHY property into account. Call the net80211 watchdog function from rtw_watchdog, so that we scan again if auth/assoc fails. Be a little more cautious about writing register[RTW_TPPOLL], since other drivers are.... Don't frob the high/low-priority queues right now, since I don't use them. Add rtw_join_bss which programs the card with the BSSID and other properties of a BSS. Use it on state transitions. Factor out rtw_set_nettype. Make rtw_recv_beacon call ieee80211_recv_mgmt instead of dropping beacons on the floor! TBD IBSS merges. Change some rtw_debug=2 printfs to rtw_debug=3 (RTW_DPRINTF3) printfs so the console doesn't get spammed so badly at rtw_debug=2. Change some debugging printfs to RTW_DPRINTFs. E.g., print the "RF programming method" only if debugging is enabled.
2004-12-21 02:05:41 +03:00
rtw_set_nettype(sc, IEEE80211_M_MONITOR);
At last, I have rtw w/ Philips RF receiving packets. I added some sysctls to aid debugging: * hw.rtw.debug -- enable debugging * hw.rtw.flush_rfio -- Linux voodoo: possibly makes the MAC "flush" bits down the serial bus to the RF * hw.rtw.host_rfio: force the host to bang bits to the RF, instead of the MAC banging bits * hw.rtw.rfio_delay: after telling the MAC to bang bits to the RF front-end, delay rfio_delay microseconds. * hw.rtw.rfprog_fallback: there is this notion of the "RF programming method." I believe the choice influences the polarity/timing of the serial bus used to program the RF front-end. I know the correct choice for Intersil/RFMD/Philips front-ends, only. For all other front-ends, I "fallback" to rfprog_fallback. Make rtw_txdac_enable take an rtw_softc argument. I will probably revert this change. Add some Linux voodoo to rtw_continuous_tx_enable. I will probably revert this change. Important: add rtw_set_rfprog, which sets the correct RF programming method. This change and the following change are probably responsible for making the Philips RF work. Important: RTW_CONFIG1 is an 8-bit register, treat it that way! Important: RTW_BRSR is 16-bit, RTW_CRCOUNT, RTW_PHYDELAY, and RTW_MSR are 8-bit: treat them that way! Vastly simplify rtw_resume_ticks. Note to self: set the LED state to match the power state. Hedge against the possibility that RTW_MSR is protected as RTW_CONFIG[0123] are, meanwhile reworking that section of rtw_init a little. Add sc_anaparm, which isn't used, yet....
2004-12-13 03:48:02 +03:00
if (ic->ic_opmode == IEEE80211_M_MONITOR)
return ieee80211_new_state(ic, IEEE80211_S_RUN, -1);
else
return ieee80211_new_state(ic, IEEE80211_S_SCAN, -1);
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
out:
If defined(RTW_DEBUG), provide a sysctl (hw.rtw.rxbufs_limit) for limiting the number of rx buffers an rtw may allocate. Use this sysctl to test the code that copes with buffer exhaustion. Allocate at most RTW_RXQLEN rx buffers, stopping at the sysctl limit. Record in sc_nrxdesc how many were allocated, and put the end-of-ring indication on sc_rxdesc[sc_nrxdesc - 1]. In rtw_init, if no rx buffers could be allocated, log a complaint, clear IFF_RUNNING, and exit with an error. Many changes to accomodate a short rx ring, mainly of the "add a rx-ring length argument" variety. XXX I really should consolidate all of the rx ring variables in one struct and pass that to the rx-ring subroutines. Bug fix: after calling MCLGET, use the (m->m_flags & M_EXT) idiom to check for success, instead of m != NULL. Bug fix: at the top of rtw_start, if IFF_RUNNING is not set, or IFF_OACTIVE is, get out.
2004-12-29 01:21:15 +03:00
printf("%s: interface not running\n", sc->sc_dev.dv_xname);
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
return rc;
}
static int
rtw_ioctl(struct ifnet *ifp, u_long cmd, caddr_t data)
{
Change rtw_debug from a debug level to a debug mask. Add a lot of debug flags. From Linux: handle an RTL8180 bug. Sometimes the NIC skips from the middle of the ring to the 0th rx descriptor. Now the driver resynchronizes. Handle a receive descriptor underrun or Rx FIFO overflow condition in the way that the Linux driver does. This kind of seems like overkill, but whatever. Protect rtw_ioctl with splnet(). Do not load a tx descriptor with a buffer shorter than 4 bytes. Handle a transmit timeout less disruptively.
2004-12-25 09:58:37 +03:00
int rc = 0, s;
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
struct rtw_softc *sc = ifp->if_softc;
struct ifreq *ifr = (struct ifreq *)data;
Change rtw_debug from a debug level to a debug mask. Add a lot of debug flags. From Linux: handle an RTL8180 bug. Sometimes the NIC skips from the middle of the ring to the 0th rx descriptor. Now the driver resynchronizes. Handle a receive descriptor underrun or Rx FIFO overflow condition in the way that the Linux driver does. This kind of seems like overkill, but whatever. Protect rtw_ioctl with splnet(). Do not load a tx descriptor with a buffer shorter than 4 bytes. Handle a transmit timeout less disruptively.
2004-12-25 09:58:37 +03:00
s = splnet();
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
switch (cmd) {
case SIOCSIFFLAGS:
if ((ifp->if_flags & IFF_UP) != 0) {
Change rtw_debug from a debug level to a debug mask. Add a lot of debug flags. From Linux: handle an RTL8180 bug. Sometimes the NIC skips from the middle of the ring to the 0th rx descriptor. Now the driver resynchronizes. Handle a receive descriptor underrun or Rx FIFO overflow condition in the way that the Linux driver does. This kind of seems like overkill, but whatever. Protect rtw_ioctl with splnet(). Do not load a tx descriptor with a buffer shorter than 4 bytes. Handle a transmit timeout less disruptively.
2004-12-25 09:58:37 +03:00
if (0 && (sc->sc_flags & RTW_F_ENABLED) != 0) {
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
rtw_pktfilt_load(sc);
} else
rc = rtw_init(ifp);
#ifdef RTW_DEBUG
rtw_print_regs(&sc->sc_regs, ifp->if_xname, __func__);
#endif /* RTW_DEBUG */
} else if ((sc->sc_flags & RTW_F_ENABLED) != 0) {
#ifdef RTW_DEBUG
rtw_print_regs(&sc->sc_regs, ifp->if_xname, __func__);
#endif /* RTW_DEBUG */
rtw_stop(ifp, 1);
}
break;
case SIOCADDMULTI:
case SIOCDELMULTI:
if (cmd == SIOCADDMULTI)
rc = ether_addmulti(ifr, &sc->sc_ic.ic_ec);
else
rc = ether_delmulti(ifr, &sc->sc_ic.ic_ec);
if (rc == ENETRESET) {
When adding/deleting multicast addresses, only whack the address filter if the interface is marked RUNNING. Fixes kern/27678.
2004-10-30 22:08:34 +04:00
if (ifp->if_flags & IFF_RUNNING)
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
rtw_pktfilt_load(sc);
rc = 0;
}
break;
default:
if ((rc = ieee80211_ioctl(ifp, cmd, data)) == ENETRESET) {
if ((sc->sc_flags & RTW_F_ENABLED) != 0)
rc = rtw_init(ifp);
else
rc = 0;
}
break;
}
Change rtw_debug from a debug level to a debug mask. Add a lot of debug flags. From Linux: handle an RTL8180 bug. Sometimes the NIC skips from the middle of the ring to the 0th rx descriptor. Now the driver resynchronizes. Handle a receive descriptor underrun or Rx FIFO overflow condition in the way that the Linux driver does. This kind of seems like overkill, but whatever. Protect rtw_ioctl with splnet(). Do not load a tx descriptor with a buffer shorter than 4 bytes. Handle a transmit timeout less disruptively.
2004-12-25 09:58:37 +03:00
splx(s);
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
return rc;
}
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
/* Point *mp at the next 802.11 frame to transmit. Point *tsbp
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
* at the driver's selection of transmit control block for the packet.
*/
static __inline int
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
rtw_dequeue(struct ifnet *ifp, struct rtw_txsoft_blk **tsbp,
struct rtw_txdesc_blk **tdbp, struct mbuf **mp,
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
struct ieee80211_node **nip)
{
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
struct rtw_txsoft_blk *tsb;
struct rtw_txdesc_blk *tdb;
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
struct mbuf *m0;
struct rtw_softc *sc;
struct ieee80211com *ic;
sc = (struct rtw_softc *)ifp->if_softc;
Change rtw_debug from a debug level to a debug mask. Add a lot of debug flags. From Linux: handle an RTL8180 bug. Sometimes the NIC skips from the middle of the ring to the 0th rx descriptor. Now the driver resynchronizes. Handle a receive descriptor underrun or Rx FIFO overflow condition in the way that the Linux driver does. This kind of seems like overkill, but whatever. Protect rtw_ioctl with splnet(). Do not load a tx descriptor with a buffer shorter than 4 bytes. Handle a transmit timeout less disruptively.
2004-12-25 09:58:37 +03:00
DPRINTF(sc, RTW_DEBUG_XMIT,
("%s: enter %s\n", sc->sc_dev.dv_xname, __func__));
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
*mp = NULL;
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
tsb = &sc->sc_txsoft_blk[RTW_TXPRIMD];
tdb = &sc->sc_txdesc_blk[RTW_TXPRIMD];
Add the transmit section of rtw(4): build up and tear down transmit descriptor rings, move packets from the transmit queues to the transmit rings, handle transmission-completed interrupts. My Linksys WPC11 ver. 4 with Maxim RF, generates interrupts like it is successfully transmitting packets. Clearly there are bugs: my G4 Powerbook locks up hard. I will debug tomorrow.
2004-12-19 11:19:25 +03:00
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
if (SIMPLEQ_EMPTY(&tsb->tsb_freeq) || tdb->tdb_nfree == 0) {
Change rtw_debug from a debug level to a debug mask. Add a lot of debug flags. From Linux: handle an RTL8180 bug. Sometimes the NIC skips from the middle of the ring to the 0th rx descriptor. Now the driver resynchronizes. Handle a receive descriptor underrun or Rx FIFO overflow condition in the way that the Linux driver does. This kind of seems like overkill, but whatever. Protect rtw_ioctl with splnet(). Do not load a tx descriptor with a buffer shorter than 4 bytes. Handle a transmit timeout less disruptively.
2004-12-25 09:58:37 +03:00
DPRINTF(sc, RTW_DEBUG_XMIT,
("%s: out of descriptors\n", __func__));
Add the transmit section of rtw(4): build up and tear down transmit descriptor rings, move packets from the transmit queues to the transmit rings, handle transmission-completed interrupts. My Linksys WPC11 ver. 4 with Maxim RF, generates interrupts like it is successfully transmitting packets. Clearly there are bugs: my G4 Powerbook locks up hard. I will debug tomorrow.
2004-12-19 11:19:25 +03:00
ifp->if_flags |= IFF_OACTIVE;
return 0;
}
ic = &sc->sc_ic;
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
if (!IF_IS_EMPTY(&ic->ic_mgtq)) {
IF_DEQUEUE(&ic->ic_mgtq, m0);
*nip = (struct ieee80211_node *)m0->m_pkthdr.rcvif;
m0->m_pkthdr.rcvif = NULL;
Change rtw_debug from a debug level to a debug mask. Add a lot of debug flags. From Linux: handle an RTL8180 bug. Sometimes the NIC skips from the middle of the ring to the 0th rx descriptor. Now the driver resynchronizes. Handle a receive descriptor underrun or Rx FIFO overflow condition in the way that the Linux driver does. This kind of seems like overkill, but whatever. Protect rtw_ioctl with splnet(). Do not load a tx descriptor with a buffer shorter than 4 bytes. Handle a transmit timeout less disruptively.
2004-12-25 09:58:37 +03:00
DPRINTF(sc, RTW_DEBUG_XMIT,
("%s: dequeue mgt frame\n", __func__));
Add the transmit section of rtw(4): build up and tear down transmit descriptor rings, move packets from the transmit queues to the transmit rings, handle transmission-completed interrupts. My Linksys WPC11 ver. 4 with Maxim RF, generates interrupts like it is successfully transmitting packets. Clearly there are bugs: my G4 Powerbook locks up hard. I will debug tomorrow.
2004-12-19 11:19:25 +03:00
} else if (ic->ic_state != IEEE80211_S_RUN) {
Change rtw_debug from a debug level to a debug mask. Add a lot of debug flags. From Linux: handle an RTL8180 bug. Sometimes the NIC skips from the middle of the ring to the 0th rx descriptor. Now the driver resynchronizes. Handle a receive descriptor underrun or Rx FIFO overflow condition in the way that the Linux driver does. This kind of seems like overkill, but whatever. Protect rtw_ioctl with splnet(). Do not load a tx descriptor with a buffer shorter than 4 bytes. Handle a transmit timeout less disruptively.
2004-12-25 09:58:37 +03:00
DPRINTF(sc, RTW_DEBUG_XMIT, ("%s: not running\n", __func__));
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
return 0;
Add the transmit section of rtw(4): build up and tear down transmit descriptor rings, move packets from the transmit queues to the transmit rings, handle transmission-completed interrupts. My Linksys WPC11 ver. 4 with Maxim RF, generates interrupts like it is successfully transmitting packets. Clearly there are bugs: my G4 Powerbook locks up hard. I will debug tomorrow.
2004-12-19 11:19:25 +03:00
} else if (!IF_IS_EMPTY(&ic->ic_pwrsaveq)) {
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
IF_DEQUEUE(&ic->ic_pwrsaveq, m0);
*nip = (struct ieee80211_node *)m0->m_pkthdr.rcvif;
m0->m_pkthdr.rcvif = NULL;
Change rtw_debug from a debug level to a debug mask. Add a lot of debug flags. From Linux: handle an RTL8180 bug. Sometimes the NIC skips from the middle of the ring to the 0th rx descriptor. Now the driver resynchronizes. Handle a receive descriptor underrun or Rx FIFO overflow condition in the way that the Linux driver does. This kind of seems like overkill, but whatever. Protect rtw_ioctl with splnet(). Do not load a tx descriptor with a buffer shorter than 4 bytes. Handle a transmit timeout less disruptively.
2004-12-25 09:58:37 +03:00
DPRINTF(sc, RTW_DEBUG_XMIT,
("%s: dequeue pwrsave frame\n", __func__));
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
} else {
When checking if a split packet is completed, check the *last* segment, not the first. Otherwise we can overwrite parts of the TX ring that we shouldn't. Kluge the basic rate setting for now. Minor simplification to the dequeueing logic.
2004-12-27 12:40:18 +03:00
IFQ_DEQUEUE(&ifp->if_snd, m0);
Add the transmit section of rtw(4): build up and tear down transmit descriptor rings, move packets from the transmit queues to the transmit rings, handle transmission-completed interrupts. My Linksys WPC11 ver. 4 with Maxim RF, generates interrupts like it is successfully transmitting packets. Clearly there are bugs: my G4 Powerbook locks up hard. I will debug tomorrow.
2004-12-19 11:19:25 +03:00
if (m0 == NULL) {
Change rtw_debug from a debug level to a debug mask. Add a lot of debug flags. From Linux: handle an RTL8180 bug. Sometimes the NIC skips from the middle of the ring to the 0th rx descriptor. Now the driver resynchronizes. Handle a receive descriptor underrun or Rx FIFO overflow condition in the way that the Linux driver does. This kind of seems like overkill, but whatever. Protect rtw_ioctl with splnet(). Do not load a tx descriptor with a buffer shorter than 4 bytes. Handle a transmit timeout less disruptively.
2004-12-25 09:58:37 +03:00
DPRINTF(sc, RTW_DEBUG_XMIT,
("%s: no frame\n", __func__));
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
return 0;
Add the transmit section of rtw(4): build up and tear down transmit descriptor rings, move packets from the transmit queues to the transmit rings, handle transmission-completed interrupts. My Linksys WPC11 ver. 4 with Maxim RF, generates interrupts like it is successfully transmitting packets. Clearly there are bugs: my G4 Powerbook locks up hard. I will debug tomorrow.
2004-12-19 11:19:25 +03:00
}
Change rtw_debug from a debug level to a debug mask. Add a lot of debug flags. From Linux: handle an RTL8180 bug. Sometimes the NIC skips from the middle of the ring to the 0th rx descriptor. Now the driver resynchronizes. Handle a receive descriptor underrun or Rx FIFO overflow condition in the way that the Linux driver does. This kind of seems like overkill, but whatever. Protect rtw_ioctl with splnet(). Do not load a tx descriptor with a buffer shorter than 4 bytes. Handle a transmit timeout less disruptively.
2004-12-25 09:58:37 +03:00
DPRINTF(sc, RTW_DEBUG_XMIT,
("%s: dequeue data frame\n", __func__));
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
ifp->if_opackets++;
#if NBPFILTER > 0
if (ifp->if_bpf)
bpf_mtap(ifp->if_bpf, m0);
#endif
if ((m0 = ieee80211_encap(ifp, m0, nip)) == NULL) {
Change rtw_debug from a debug level to a debug mask. Add a lot of debug flags. From Linux: handle an RTL8180 bug. Sometimes the NIC skips from the middle of the ring to the 0th rx descriptor. Now the driver resynchronizes. Handle a receive descriptor underrun or Rx FIFO overflow condition in the way that the Linux driver does. This kind of seems like overkill, but whatever. Protect rtw_ioctl with splnet(). Do not load a tx descriptor with a buffer shorter than 4 bytes. Handle a transmit timeout less disruptively.
2004-12-25 09:58:37 +03:00
DPRINTF(sc, RTW_DEBUG_XMIT,
("%s: encap error\n", __func__));
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
ifp->if_oerrors++;
return -1;
}
}
Change rtw_debug from a debug level to a debug mask. Add a lot of debug flags. From Linux: handle an RTL8180 bug. Sometimes the NIC skips from the middle of the ring to the 0th rx descriptor. Now the driver resynchronizes. Handle a receive descriptor underrun or Rx FIFO overflow condition in the way that the Linux driver does. This kind of seems like overkill, but whatever. Protect rtw_ioctl with splnet(). Do not load a tx descriptor with a buffer shorter than 4 bytes. Handle a transmit timeout less disruptively.
2004-12-25 09:58:37 +03:00
DPRINTF(sc, RTW_DEBUG_XMIT, ("%s: leave\n", __func__));
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
*tsbp = tsb;
*tdbp = tdb;
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
*mp = m0;
return 0;
}
Change rtw_debug from a debug level to a debug mask. Add a lot of debug flags. From Linux: handle an RTL8180 bug. Sometimes the NIC skips from the middle of the ring to the 0th rx descriptor. Now the driver resynchronizes. Handle a receive descriptor underrun or Rx FIFO overflow condition in the way that the Linux driver does. This kind of seems like overkill, but whatever. Protect rtw_ioctl with splnet(). Do not load a tx descriptor with a buffer shorter than 4 bytes. Handle a transmit timeout less disruptively.
2004-12-25 09:58:37 +03:00
static int
rtw_seg_too_short(bus_dmamap_t dmamap)
{
int i;
for (i = 0; i < dmamap->dm_nsegs; i++) {
if (dmamap->dm_segs[i].ds_len < 4) {
printf("%s: segment too short\n", __func__);
return 1;
}
}
return 0;
}
Add the transmit section of rtw(4): build up and tear down transmit descriptor rings, move packets from the transmit queues to the transmit rings, handle transmission-completed interrupts. My Linksys WPC11 ver. 4 with Maxim RF, generates interrupts like it is successfully transmitting packets. Clearly there are bugs: my G4 Powerbook locks up hard. I will debug tomorrow.
2004-12-19 11:19:25 +03:00
/* TBD factor with atw_start */
static struct mbuf *
rtw_dmamap_load_txbuf(bus_dma_tag_t dmat, bus_dmamap_t dmam, struct mbuf *chain,
u_int ndescfree, short *ifflagsp, const char *dvname)
{
int first, rc;
struct mbuf *m, *m0;
m0 = chain;
/*
* Load the DMA map. Copy and try (once) again if the packet
* didn't fit in the alloted number of segments.
*/
for (first = 1;
((rc = bus_dmamap_load_mbuf(dmat, dmam, m0,
BUS_DMA_WRITE|BUS_DMA_NOWAIT)) != 0 ||
Change rtw_debug from a debug level to a debug mask. Add a lot of debug flags. From Linux: handle an RTL8180 bug. Sometimes the NIC skips from the middle of the ring to the 0th rx descriptor. Now the driver resynchronizes. Handle a receive descriptor underrun or Rx FIFO overflow condition in the way that the Linux driver does. This kind of seems like overkill, but whatever. Protect rtw_ioctl with splnet(). Do not load a tx descriptor with a buffer shorter than 4 bytes. Handle a transmit timeout less disruptively.
2004-12-25 09:58:37 +03:00
dmam->dm_nsegs > ndescfree || rtw_seg_too_short(dmam)) && first;
Add the transmit section of rtw(4): build up and tear down transmit descriptor rings, move packets from the transmit queues to the transmit rings, handle transmission-completed interrupts. My Linksys WPC11 ver. 4 with Maxim RF, generates interrupts like it is successfully transmitting packets. Clearly there are bugs: my G4 Powerbook locks up hard. I will debug tomorrow.
2004-12-19 11:19:25 +03:00
first = 0) {
if (rc == 0)
bus_dmamap_unload(dmat, dmam);
MGETHDR(m, M_DONTWAIT, MT_DATA);
if (m == NULL) {
printf("%s: unable to allocate Tx mbuf\n",
dvname);
break;
}
if (m0->m_pkthdr.len > MHLEN) {
MCLGET(m, M_DONTWAIT);
if ((m->m_flags & M_EXT) == 0) {
printf("%s: cannot allocate Tx cluster\n",
dvname);
m_freem(m);
break;
}
}
m_copydata(m0, 0, m0->m_pkthdr.len, mtod(m, caddr_t));
m->m_pkthdr.len = m->m_len = m0->m_pkthdr.len;
m_freem(m0);
m0 = m;
m = NULL;
}
if (rc != 0) {
printf("%s: cannot load Tx buffer, rc = %d\n", dvname, rc);
m_freem(m0);
return NULL;
Change rtw_debug from a debug level to a debug mask. Add a lot of debug flags. From Linux: handle an RTL8180 bug. Sometimes the NIC skips from the middle of the ring to the 0th rx descriptor. Now the driver resynchronizes. Handle a receive descriptor underrun or Rx FIFO overflow condition in the way that the Linux driver does. This kind of seems like overkill, but whatever. Protect rtw_ioctl with splnet(). Do not load a tx descriptor with a buffer shorter than 4 bytes. Handle a transmit timeout less disruptively.
2004-12-25 09:58:37 +03:00
} else if (rtw_seg_too_short(dmam)) {
printf("%s: cannot load Tx buffer, segment too short\n",
dvname);
bus_dmamap_unload(dmat, dmam);
m_freem(m0);
return NULL;
Add the transmit section of rtw(4): build up and tear down transmit descriptor rings, move packets from the transmit queues to the transmit rings, handle transmission-completed interrupts. My Linksys WPC11 ver. 4 with Maxim RF, generates interrupts like it is successfully transmitting packets. Clearly there are bugs: my G4 Powerbook locks up hard. I will debug tomorrow.
2004-12-19 11:19:25 +03:00
} else if (dmam->dm_nsegs > ndescfree) {
*ifflagsp |= IFF_OACTIVE;
bus_dmamap_unload(dmat, dmam);
m_freem(m0);
return NULL;
}
return m0;
}
Change rtw_debug from a debug level to a debug mask. Add a lot of debug flags. From Linux: handle an RTL8180 bug. Sometimes the NIC skips from the middle of the ring to the 0th rx descriptor. Now the driver resynchronizes. Handle a receive descriptor underrun or Rx FIFO overflow condition in the way that the Linux driver does. This kind of seems like overkill, but whatever. Protect rtw_ioctl with splnet(). Do not load a tx descriptor with a buffer shorter than 4 bytes. Handle a transmit timeout less disruptively.
2004-12-25 09:58:37 +03:00
#ifdef RTW_DEBUG
Add more comprehensive debug messages as we setup transmit descriptors. Also, dump packets if IFF_DEBUG|IFF_LINK2. Factor out a subroutine for printing transmit-descriptor debug messages.
2004-12-23 08:57:18 +03:00
static void
rtw_print_txdesc(struct rtw_softc *sc, const char *action,
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
struct rtw_txsoft *ts, struct rtw_txdesc_blk *tdb, int desc)
Add more comprehensive debug messages as we setup transmit descriptors. Also, dump packets if IFF_DEBUG|IFF_LINK2. Factor out a subroutine for printing transmit-descriptor debug messages.
2004-12-23 08:57:18 +03:00
{
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
struct rtw_txdesc *td = &tdb->tdb_desc[desc];
Change rtw_debug from a debug level to a debug mask. Add a lot of debug flags. From Linux: handle an RTL8180 bug. Sometimes the NIC skips from the middle of the ring to the 0th rx descriptor. Now the driver resynchronizes. Handle a receive descriptor underrun or Rx FIFO overflow condition in the way that the Linux driver does. This kind of seems like overkill, but whatever. Protect rtw_ioctl with splnet(). Do not load a tx descriptor with a buffer shorter than 4 bytes. Handle a transmit timeout less disruptively.
2004-12-25 09:58:37 +03:00
DPRINTF(sc, RTW_DEBUG_XMIT_DESC, ("%s: %p %s txdesc[%d] ctl0 %#08x "
Add more comprehensive debug messages as we setup transmit descriptors. Also, dump packets if IFF_DEBUG|IFF_LINK2. Factor out a subroutine for printing transmit-descriptor debug messages.
2004-12-23 08:57:18 +03:00
"ctl1 %#08x buf %#08x len %#08x\n",
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
sc->sc_dev.dv_xname, ts, action, desc,
le32toh(td->td_ctl0),
le32toh(td->td_ctl1), le32toh(td->td_buf),
le32toh(td->td_len)));
Add more comprehensive debug messages as we setup transmit descriptors. Also, dump packets if IFF_DEBUG|IFF_LINK2. Factor out a subroutine for printing transmit-descriptor debug messages.
2004-12-23 08:57:18 +03:00
}
Change rtw_debug from a debug level to a debug mask. Add a lot of debug flags. From Linux: handle an RTL8180 bug. Sometimes the NIC skips from the middle of the ring to the 0th rx descriptor. Now the driver resynchronizes. Handle a receive descriptor underrun or Rx FIFO overflow condition in the way that the Linux driver does. This kind of seems like overkill, but whatever. Protect rtw_ioctl with splnet(). Do not load a tx descriptor with a buffer shorter than 4 bytes. Handle a transmit timeout less disruptively.
2004-12-25 09:58:37 +03:00
#endif /* RTW_DEBUG */
Add more comprehensive debug messages as we setup transmit descriptors. Also, dump packets if IFF_DEBUG|IFF_LINK2. Factor out a subroutine for printing transmit-descriptor debug messages.
2004-12-23 08:57:18 +03:00
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
static void
rtw_start(struct ifnet *ifp)
{
Realtek sample sources are not consistent about the "Digital PHY" bit in the SROM. It seems as if it is set to 1 when the PHY is *analog*, not *digital*. Fix my sources. In rtw_intr_rx, use units of 500kb/s instead of 100kb/s for rate, to be consistent with net80211's expectations. Polish up some debugging ugly messages. Dump raw 802.11 packets if IFF_DEBUG|IFF_LINK2 and RTW_DEBUG is defined. Polish power-state (on/sleep/off) handling. Especially improve support for RFMD (totally untested) and Maxim. For Philips, take the Digital PHY property into account. Call the net80211 watchdog function from rtw_watchdog, so that we scan again if auth/assoc fails. Be a little more cautious about writing register[RTW_TPPOLL], since other drivers are.... Don't frob the high/low-priority queues right now, since I don't use them. Add rtw_join_bss which programs the card with the BSSID and other properties of a BSS. Use it on state transitions. Factor out rtw_set_nettype. Make rtw_recv_beacon call ieee80211_recv_mgmt instead of dropping beacons on the floor! TBD IBSS merges. Change some rtw_debug=2 printfs to rtw_debug=3 (RTW_DPRINTF3) printfs so the console doesn't get spammed so badly at rtw_debug=2. Change some debugging printfs to RTW_DPRINTFs. E.g., print the "RF programming method" only if debugging is enabled.
2004-12-21 02:05:41 +03:00
uint8_t tppoll;
Add the transmit section of rtw(4): build up and tear down transmit descriptor rings, move packets from the transmit queues to the transmit rings, handle transmission-completed interrupts. My Linksys WPC11 ver. 4 with Maxim RF, generates interrupts like it is successfully transmitting packets. Clearly there are bugs: my G4 Powerbook locks up hard. I will debug tomorrow.
2004-12-19 11:19:25 +03:00
int desc, i, lastdesc, npkt, rate;
Shorten variables, s/txctl/ctl/. This is rtw_start, after all: transmission is implied.
2004-12-23 08:52:27 +03:00
uint32_t proto_ctl0, ctl0, ctl1;
Add the transmit section of rtw(4): build up and tear down transmit descriptor rings, move packets from the transmit queues to the transmit rings, handle transmission-completed interrupts. My Linksys WPC11 ver. 4 with Maxim RF, generates interrupts like it is successfully transmitting packets. Clearly there are bugs: my G4 Powerbook locks up hard. I will debug tomorrow.
2004-12-19 11:19:25 +03:00
bus_dmamap_t dmamap;
struct ieee80211com *ic;
struct ieee80211_duration *d0;
struct ieee80211_frame *wh;
struct ieee80211_node *ni;
struct mbuf *m0;
struct rtw_softc *sc;
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
struct rtw_txsoft_blk *tsb;
struct rtw_txdesc_blk *tdb;
struct rtw_txsoft *ts;
struct rtw_txdesc *td;
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
sc = (struct rtw_softc *)ifp->if_softc;
Add the transmit section of rtw(4): build up and tear down transmit descriptor rings, move packets from the transmit queues to the transmit rings, handle transmission-completed interrupts. My Linksys WPC11 ver. 4 with Maxim RF, generates interrupts like it is successfully transmitting packets. Clearly there are bugs: my G4 Powerbook locks up hard. I will debug tomorrow.
2004-12-19 11:19:25 +03:00
ic = &sc->sc_ic;
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
Change rtw_debug from a debug level to a debug mask. Add a lot of debug flags. From Linux: handle an RTL8180 bug. Sometimes the NIC skips from the middle of the ring to the 0th rx descriptor. Now the driver resynchronizes. Handle a receive descriptor underrun or Rx FIFO overflow condition in the way that the Linux driver does. This kind of seems like overkill, but whatever. Protect rtw_ioctl with splnet(). Do not load a tx descriptor with a buffer shorter than 4 bytes. Handle a transmit timeout less disruptively.
2004-12-25 09:58:37 +03:00
DPRINTF(sc, RTW_DEBUG_XMIT,
("%s: enter %s\n", sc->sc_dev.dv_xname, __func__));
Add the transmit section of rtw(4): build up and tear down transmit descriptor rings, move packets from the transmit queues to the transmit rings, handle transmission-completed interrupts. My Linksys WPC11 ver. 4 with Maxim RF, generates interrupts like it is successfully transmitting packets. Clearly there are bugs: my G4 Powerbook locks up hard. I will debug tomorrow.
2004-12-19 11:19:25 +03:00
If defined(RTW_DEBUG), provide a sysctl (hw.rtw.rxbufs_limit) for limiting the number of rx buffers an rtw may allocate. Use this sysctl to test the code that copes with buffer exhaustion. Allocate at most RTW_RXQLEN rx buffers, stopping at the sysctl limit. Record in sc_nrxdesc how many were allocated, and put the end-of-ring indication on sc_rxdesc[sc_nrxdesc - 1]. In rtw_init, if no rx buffers could be allocated, log a complaint, clear IFF_RUNNING, and exit with an error. Many changes to accomodate a short rx ring, mainly of the "add a rx-ring length argument" variety. XXX I really should consolidate all of the rx ring variables in one struct and pass that to the rx-ring subroutines. Bug fix: after calling MCLGET, use the (m->m_flags & M_EXT) idiom to check for success, instead of m != NULL. Bug fix: at the top of rtw_start, if IFF_RUNNING is not set, or IFF_OACTIVE is, get out.
2004-12-29 01:21:15 +03:00
if ((ifp->if_flags & (IFF_RUNNING|IFF_OACTIVE)) != IFF_RUNNING)
goto out;
Add the transmit section of rtw(4): build up and tear down transmit descriptor rings, move packets from the transmit queues to the transmit rings, handle transmission-completed interrupts. My Linksys WPC11 ver. 4 with Maxim RF, generates interrupts like it is successfully transmitting packets. Clearly there are bugs: my G4 Powerbook locks up hard. I will debug tomorrow.
2004-12-19 11:19:25 +03:00
/* XXX do real rate control */
Shorten variables, s/txctl/ctl/. This is rtw_start, after all: transmission is implied.
2004-12-23 08:52:27 +03:00
proto_ctl0 = RTW_TXCTL0_RTSRATE_1MBPS;
Add the transmit section of rtw(4): build up and tear down transmit descriptor rings, move packets from the transmit queues to the transmit rings, handle transmission-completed interrupts. My Linksys WPC11 ver. 4 with Maxim RF, generates interrupts like it is successfully transmitting packets. Clearly there are bugs: my G4 Powerbook locks up hard. I will debug tomorrow.
2004-12-19 11:19:25 +03:00
switch (rate = MAX(2, ieee80211_get_rate(ic))) {
case 2:
Shorten variables, s/txctl/ctl/. This is rtw_start, after all: transmission is implied.
2004-12-23 08:52:27 +03:00
proto_ctl0 |= RTW_TXCTL0_RATE_1MBPS;
Add the transmit section of rtw(4): build up and tear down transmit descriptor rings, move packets from the transmit queues to the transmit rings, handle transmission-completed interrupts. My Linksys WPC11 ver. 4 with Maxim RF, generates interrupts like it is successfully transmitting packets. Clearly there are bugs: my G4 Powerbook locks up hard. I will debug tomorrow.
2004-12-19 11:19:25 +03:00
break;
case 4:
Shorten variables, s/txctl/ctl/. This is rtw_start, after all: transmission is implied.
2004-12-23 08:52:27 +03:00
proto_ctl0 |= RTW_TXCTL0_RATE_2MBPS;
Add the transmit section of rtw(4): build up and tear down transmit descriptor rings, move packets from the transmit queues to the transmit rings, handle transmission-completed interrupts. My Linksys WPC11 ver. 4 with Maxim RF, generates interrupts like it is successfully transmitting packets. Clearly there are bugs: my G4 Powerbook locks up hard. I will debug tomorrow.
2004-12-19 11:19:25 +03:00
break;
case 11:
Shorten variables, s/txctl/ctl/. This is rtw_start, after all: transmission is implied.
2004-12-23 08:52:27 +03:00
proto_ctl0 |= RTW_TXCTL0_RATE_5MBPS;
Add the transmit section of rtw(4): build up and tear down transmit descriptor rings, move packets from the transmit queues to the transmit rings, handle transmission-completed interrupts. My Linksys WPC11 ver. 4 with Maxim RF, generates interrupts like it is successfully transmitting packets. Clearly there are bugs: my G4 Powerbook locks up hard. I will debug tomorrow.
2004-12-19 11:19:25 +03:00
break;
case 22:
Shorten variables, s/txctl/ctl/. This is rtw_start, after all: transmission is implied.
2004-12-23 08:52:27 +03:00
proto_ctl0 |= RTW_TXCTL0_RATE_11MBPS;
Add the transmit section of rtw(4): build up and tear down transmit descriptor rings, move packets from the transmit queues to the transmit rings, handle transmission-completed interrupts. My Linksys WPC11 ver. 4 with Maxim RF, generates interrupts like it is successfully transmitting packets. Clearly there are bugs: my G4 Powerbook locks up hard. I will debug tomorrow.
2004-12-19 11:19:25 +03:00
break;
}
if ((ic->ic_flags & IEEE80211_F_SHPREAMBLE) != 0)
Shorten variables, s/txctl/ctl/. This is rtw_start, after all: transmission is implied.
2004-12-23 08:52:27 +03:00
proto_ctl0 |= RTW_TXCTL0_SPLCP;
Add the transmit section of rtw(4): build up and tear down transmit descriptor rings, move packets from the transmit queues to the transmit rings, handle transmission-completed interrupts. My Linksys WPC11 ver. 4 with Maxim RF, generates interrupts like it is successfully transmitting packets. Clearly there are bugs: my G4 Powerbook locks up hard. I will debug tomorrow.
2004-12-19 11:19:25 +03:00
for (;;) {
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
if (rtw_dequeue(ifp, &tsb, &tdb, &m0, &ni) == -1)
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
continue;
if (m0 == NULL)
break;
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
ts = SIMPLEQ_FIRST(&tsb->tsb_freeq);
Add the transmit section of rtw(4): build up and tear down transmit descriptor rings, move packets from the transmit queues to the transmit rings, handle transmission-completed interrupts. My Linksys WPC11 ver. 4 with Maxim RF, generates interrupts like it is successfully transmitting packets. Clearly there are bugs: my G4 Powerbook locks up hard. I will debug tomorrow.
2004-12-19 11:19:25 +03:00
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
dmamap = ts->ts_dmamap;
Add the transmit section of rtw(4): build up and tear down transmit descriptor rings, move packets from the transmit queues to the transmit rings, handle transmission-completed interrupts. My Linksys WPC11 ver. 4 with Maxim RF, generates interrupts like it is successfully transmitting packets. Clearly there are bugs: my G4 Powerbook locks up hard. I will debug tomorrow.
2004-12-19 11:19:25 +03:00
m0 = rtw_dmamap_load_txbuf(sc->sc_dmat, dmamap, m0,
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
tdb->tdb_nfree, &ifp->if_flags, sc->sc_dev.dv_xname);
Add the transmit section of rtw(4): build up and tear down transmit descriptor rings, move packets from the transmit queues to the transmit rings, handle transmission-completed interrupts. My Linksys WPC11 ver. 4 with Maxim RF, generates interrupts like it is successfully transmitting packets. Clearly there are bugs: my G4 Powerbook locks up hard. I will debug tomorrow.
2004-12-19 11:19:25 +03:00
if (m0 == NULL || dmamap->dm_nsegs == 0) {
Change rtw_debug from a debug level to a debug mask. Add a lot of debug flags. From Linux: handle an RTL8180 bug. Sometimes the NIC skips from the middle of the ring to the 0th rx descriptor. Now the driver resynchronizes. Handle a receive descriptor underrun or Rx FIFO overflow condition in the way that the Linux driver does. This kind of seems like overkill, but whatever. Protect rtw_ioctl with splnet(). Do not load a tx descriptor with a buffer shorter than 4 bytes. Handle a transmit timeout less disruptively.
2004-12-25 09:58:37 +03:00
DPRINTF(sc, RTW_DEBUG_XMIT,
("%s: fail dmamap load\n", __func__));
Add the transmit section of rtw(4): build up and tear down transmit descriptor rings, move packets from the transmit queues to the transmit rings, handle transmission-completed interrupts. My Linksys WPC11 ver. 4 with Maxim RF, generates interrupts like it is successfully transmitting packets. Clearly there are bugs: my G4 Powerbook locks up hard. I will debug tomorrow.
2004-12-19 11:19:25 +03:00
goto post_dequeue_err;
}
Add more comprehensive debug messages as we setup transmit descriptors. Also, dump packets if IFF_DEBUG|IFF_LINK2. Factor out a subroutine for printing transmit-descriptor debug messages.
2004-12-23 08:57:18 +03:00
#ifdef RTW_DEBUG
if ((sc->sc_if.if_flags & (IFF_DEBUG|IFF_LINK2)) ==
(IFF_DEBUG|IFF_LINK2)) {
ieee80211_dump_pkt(mtod(m0, uint8_t *),
(dmamap->dm_nsegs == 1) ? m0->m_pkthdr.len
: sizeof(wh),
rate, 0);
}
#endif /* RTW_DEBUG */
Shorten variables, s/txctl/ctl/. This is rtw_start, after all: transmission is implied.
2004-12-23 08:52:27 +03:00
ctl0 = proto_ctl0 |
Add the transmit section of rtw(4): build up and tear down transmit descriptor rings, move packets from the transmit queues to the transmit rings, handle transmission-completed interrupts. My Linksys WPC11 ver. 4 with Maxim RF, generates interrupts like it is successfully transmitting packets. Clearly there are bugs: my G4 Powerbook locks up hard. I will debug tomorrow.
2004-12-19 11:19:25 +03:00
LSHIFT(m0->m_pkthdr.len, RTW_TXCTL0_TPKTSIZE_MASK);
wh = mtod(m0, struct ieee80211_frame *);
Fix a bug in ieee80211_compute_duration: the 802.11 Duration field in an 802.11 unicast data packet is equal to the duration of the SIFS and Acknowledgement. That is, the amount of time reserved *after* the packet has finished transmitting. Change the arguments to ieee80211_compute_duration: pass the entire packet length, not just the payload length. Add a 'debug' argument to ieee80211_compute_duration and its helper subroutine, ieee80211_compute_duration1. If debug != 0, ieee80211_compute_duration printfs its arguments and several local variables. In rtw(4), load the 802.11 Duration field with the result from ieee80211_compute_duration.
2004-12-23 09:08:52 +03:00
if (ieee80211_compute_duration(wh, m0->m_pkthdr.len,
Add the transmit section of rtw(4): build up and tear down transmit descriptor rings, move packets from the transmit queues to the transmit rings, handle transmission-completed interrupts. My Linksys WPC11 ver. 4 with Maxim RF, generates interrupts like it is successfully transmitting packets. Clearly there are bugs: my G4 Powerbook locks up hard. I will debug tomorrow.
2004-12-19 11:19:25 +03:00
ic->ic_flags, ic->ic_fragthreshold,
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
rate, &ts->ts_d0, &ts->ts_dn, &npkt,
Fix a bug in ieee80211_compute_duration: the 802.11 Duration field in an 802.11 unicast data packet is equal to the duration of the SIFS and Acknowledgement. That is, the amount of time reserved *after* the packet has finished transmitting. Change the arguments to ieee80211_compute_duration: pass the entire packet length, not just the payload length. Add a 'debug' argument to ieee80211_compute_duration and its helper subroutine, ieee80211_compute_duration1. If debug != 0, ieee80211_compute_duration printfs its arguments and several local variables. In rtw(4), load the 802.11 Duration field with the result from ieee80211_compute_duration.
2004-12-23 09:08:52 +03:00
(sc->sc_if.if_flags & (IFF_DEBUG|IFF_LINK2)) ==
(IFF_DEBUG|IFF_LINK2)) == -1) {
Change rtw_debug from a debug level to a debug mask. Add a lot of debug flags. From Linux: handle an RTL8180 bug. Sometimes the NIC skips from the middle of the ring to the 0th rx descriptor. Now the driver resynchronizes. Handle a receive descriptor underrun or Rx FIFO overflow condition in the way that the Linux driver does. This kind of seems like overkill, but whatever. Protect rtw_ioctl with splnet(). Do not load a tx descriptor with a buffer shorter than 4 bytes. Handle a transmit timeout less disruptively.
2004-12-25 09:58:37 +03:00
DPRINTF(sc, RTW_DEBUG_XMIT,
("%s: fail compute duration\n", __func__));
Add the transmit section of rtw(4): build up and tear down transmit descriptor rings, move packets from the transmit queues to the transmit rings, handle transmission-completed interrupts. My Linksys WPC11 ver. 4 with Maxim RF, generates interrupts like it is successfully transmitting packets. Clearly there are bugs: my G4 Powerbook locks up hard. I will debug tomorrow.
2004-12-19 11:19:25 +03:00
goto post_load_err;
}
/* XXX >= ? */
if (m0->m_pkthdr.len > ic->ic_rtsthreshold)
Shorten variables, s/txctl/ctl/. This is rtw_start, after all: transmission is implied.
2004-12-23 08:52:27 +03:00
ctl0 |= RTW_TXCTL0_RTSEN;
Add the transmit section of rtw(4): build up and tear down transmit descriptor rings, move packets from the transmit queues to the transmit rings, handle transmission-completed interrupts. My Linksys WPC11 ver. 4 with Maxim RF, generates interrupts like it is successfully transmitting packets. Clearly there are bugs: my G4 Powerbook locks up hard. I will debug tomorrow.
2004-12-19 11:19:25 +03:00
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
d0 = &ts->ts_d0;
Add the transmit section of rtw(4): build up and tear down transmit descriptor rings, move packets from the transmit queues to the transmit rings, handle transmission-completed interrupts. My Linksys WPC11 ver. 4 with Maxim RF, generates interrupts like it is successfully transmitting packets. Clearly there are bugs: my G4 Powerbook locks up hard. I will debug tomorrow.
2004-12-19 11:19:25 +03:00
Oops, straggler from last commit: set the 802.11 Duration field with the result from ieee80211_compute_duration.
2004-12-23 09:12:43 +03:00
*(uint16_t*)wh->i_dur = htole16(d0->d_data_dur);
Shorten variables, s/txctl/ctl/. This is rtw_start, after all: transmission is implied.
2004-12-23 08:52:27 +03:00
ctl1 = LSHIFT(d0->d_plcp_len, RTW_TXCTL1_LENGTH_MASK) |
Add the transmit section of rtw(4): build up and tear down transmit descriptor rings, move packets from the transmit queues to the transmit rings, handle transmission-completed interrupts. My Linksys WPC11 ver. 4 with Maxim RF, generates interrupts like it is successfully transmitting packets. Clearly there are bugs: my G4 Powerbook locks up hard. I will debug tomorrow.
2004-12-19 11:19:25 +03:00
LSHIFT(d0->d_rts_dur, RTW_TXCTL1_RTSDUR_MASK);
Replace d_plcp_svc with d_residue. The latter is the number of whole empty/unused octets to fill out the data time slot. The value is constrained by math to 0 for <= 5.5Mb, 0-1 for 11Mb, and 0-2 for 22Mb. It is used to signal to the MAC that there is residue.
2004-12-27 04:51:49 +03:00
if (d0->d_residue)
Shorten variables, s/txctl/ctl/. This is rtw_start, after all: transmission is implied.
2004-12-23 08:52:27 +03:00
ctl1 |= RTW_TXCTL1_LENGEXT;
Add the transmit section of rtw(4): build up and tear down transmit descriptor rings, move packets from the transmit queues to the transmit rings, handle transmission-completed interrupts. My Linksys WPC11 ver. 4 with Maxim RF, generates interrupts like it is successfully transmitting packets. Clearly there are bugs: my G4 Powerbook locks up hard. I will debug tomorrow.
2004-12-19 11:19:25 +03:00
/* TBD fragmentation */
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
ts->ts_first = tdb->tdb_next;
Add the transmit section of rtw(4): build up and tear down transmit descriptor rings, move packets from the transmit queues to the transmit rings, handle transmission-completed interrupts. My Linksys WPC11 ver. 4 with Maxim RF, generates interrupts like it is successfully transmitting packets. Clearly there are bugs: my G4 Powerbook locks up hard. I will debug tomorrow.
2004-12-19 11:19:25 +03:00
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
rtw_txdescs_sync(tdb, ts->ts_first, dmamap->dm_nsegs,
Add the transmit section of rtw(4): build up and tear down transmit descriptor rings, move packets from the transmit queues to the transmit rings, handle transmission-completed interrupts. My Linksys WPC11 ver. 4 with Maxim RF, generates interrupts like it is successfully transmitting packets. Clearly there are bugs: my G4 Powerbook locks up hard. I will debug tomorrow.
2004-12-19 11:19:25 +03:00
BUS_DMASYNC_PREWRITE);
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
KASSERT(ts->ts_first < tdb->tdb_ndesc);
Change rtw_debug from a debug level to a debug mask. Add a lot of debug flags. From Linux: handle an RTL8180 bug. Sometimes the NIC skips from the middle of the ring to the 0th rx descriptor. Now the driver resynchronizes. Handle a receive descriptor underrun or Rx FIFO overflow condition in the way that the Linux driver does. This kind of seems like overkill, but whatever. Protect rtw_ioctl with splnet(). Do not load a tx descriptor with a buffer shorter than 4 bytes. Handle a transmit timeout less disruptively.
2004-12-25 09:58:37 +03:00
Add bpf(4) packet taps to rtw(4). Now you can tap DLT_EN10MB (Ethernet II), DLT_IEEE802_11 (802.11), and DLT_IEEE802_11_RADIO (radiotap + 802.11).
2004-12-29 01:30:07 +03:00
#if NBPFILTER > 0
if (ic->ic_rawbpf != NULL)
bpf_mtap((caddr_t)ic->ic_rawbpf, m0);
if (sc->sc_radiobpf != NULL) {
struct rtw_tx_radiotap_header *rt = &sc->sc_txtap;
rt->rt_flags = 0;
rt->rt_rate = rate;
rt->rt_chan_freq =
htole16(ic->ic_bss->ni_chan->ic_freq);
rt->rt_chan_flags =
htole16(ic->ic_bss->ni_chan->ic_flags);
bpf_mtap2(sc->sc_radiobpf, (caddr_t)rt,
sizeof(sc->sc_txtapu), m0);
}
#endif /* NPBFILTER > 0 */
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
for (i = 0, lastdesc = desc = ts->ts_first;
Add the transmit section of rtw(4): build up and tear down transmit descriptor rings, move packets from the transmit queues to the transmit rings, handle transmission-completed interrupts. My Linksys WPC11 ver. 4 with Maxim RF, generates interrupts like it is successfully transmitting packets. Clearly there are bugs: my G4 Powerbook locks up hard. I will debug tomorrow.
2004-12-19 11:19:25 +03:00
i < dmamap->dm_nsegs;
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
i++, desc = RTW_NEXT_IDX(tdb, desc)) {
Add the transmit section of rtw(4): build up and tear down transmit descriptor rings, move packets from the transmit queues to the transmit rings, handle transmission-completed interrupts. My Linksys WPC11 ver. 4 with Maxim RF, generates interrupts like it is successfully transmitting packets. Clearly there are bugs: my G4 Powerbook locks up hard. I will debug tomorrow.
2004-12-19 11:19:25 +03:00
if (dmamap->dm_segs[i].ds_len > RTW_TXLEN_LENGTH_MASK) {
Change rtw_debug from a debug level to a debug mask. Add a lot of debug flags. From Linux: handle an RTL8180 bug. Sometimes the NIC skips from the middle of the ring to the 0th rx descriptor. Now the driver resynchronizes. Handle a receive descriptor underrun or Rx FIFO overflow condition in the way that the Linux driver does. This kind of seems like overkill, but whatever. Protect rtw_ioctl with splnet(). Do not load a tx descriptor with a buffer shorter than 4 bytes. Handle a transmit timeout less disruptively.
2004-12-25 09:58:37 +03:00
DPRINTF(sc, RTW_DEBUG_XMIT_DESC,
("%s: seg too long\n", __func__));
Add the transmit section of rtw(4): build up and tear down transmit descriptor rings, move packets from the transmit queues to the transmit rings, handle transmission-completed interrupts. My Linksys WPC11 ver. 4 with Maxim RF, generates interrupts like it is successfully transmitting packets. Clearly there are bugs: my G4 Powerbook locks up hard. I will debug tomorrow.
2004-12-19 11:19:25 +03:00
goto post_load_err;
}
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
td = &tdb->tdb_desc[desc];
td->td_ctl0 = htole32(ctl0);
Add the transmit section of rtw(4): build up and tear down transmit descriptor rings, move packets from the transmit queues to the transmit rings, handle transmission-completed interrupts. My Linksys WPC11 ver. 4 with Maxim RF, generates interrupts like it is successfully transmitting packets. Clearly there are bugs: my G4 Powerbook locks up hard. I will debug tomorrow.
2004-12-19 11:19:25 +03:00
if (i != 0)
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
td->td_ctl0 |= htole32(RTW_TXCTL0_OWN);
td->td_ctl1 = htole32(ctl1);
td->td_buf = htole32(dmamap->dm_segs[i].ds_addr);
td->td_len = htole32(dmamap->dm_segs[i].ds_len);
Add the transmit section of rtw(4): build up and tear down transmit descriptor rings, move packets from the transmit queues to the transmit rings, handle transmission-completed interrupts. My Linksys WPC11 ver. 4 with Maxim RF, generates interrupts like it is successfully transmitting packets. Clearly there are bugs: my G4 Powerbook locks up hard. I will debug tomorrow.
2004-12-19 11:19:25 +03:00
lastdesc = desc;
Add more comprehensive debug messages as we setup transmit descriptors. Also, dump packets if IFF_DEBUG|IFF_LINK2. Factor out a subroutine for printing transmit-descriptor debug messages.
2004-12-23 08:57:18 +03:00
#ifdef RTW_DEBUG
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
rtw_print_txdesc(sc, "load", ts, tdb, desc);
Add more comprehensive debug messages as we setup transmit descriptors. Also, dump packets if IFF_DEBUG|IFF_LINK2. Factor out a subroutine for printing transmit-descriptor debug messages.
2004-12-23 08:57:18 +03:00
#endif /* RTW_DEBUG */
Add the transmit section of rtw(4): build up and tear down transmit descriptor rings, move packets from the transmit queues to the transmit rings, handle transmission-completed interrupts. My Linksys WPC11 ver. 4 with Maxim RF, generates interrupts like it is successfully transmitting packets. Clearly there are bugs: my G4 Powerbook locks up hard. I will debug tomorrow.
2004-12-19 11:19:25 +03:00
}
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
KASSERT(desc < tdb->tdb_ndesc);
Change rtw_debug from a debug level to a debug mask. Add a lot of debug flags. From Linux: handle an RTL8180 bug. Sometimes the NIC skips from the middle of the ring to the 0th rx descriptor. Now the driver resynchronizes. Handle a receive descriptor underrun or Rx FIFO overflow condition in the way that the Linux driver does. This kind of seems like overkill, but whatever. Protect rtw_ioctl with splnet(). Do not load a tx descriptor with a buffer shorter than 4 bytes. Handle a transmit timeout less disruptively.
2004-12-25 09:58:37 +03:00
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
ts->ts_ni = ni;
ts->ts_mbuf = m0;
ts->ts_last = lastdesc;
tdb->tdb_desc[ts->ts_last].td_ctl0 |= htole32(RTW_TXCTL0_LS);
tdb->tdb_desc[ts->ts_first].td_ctl0 |=
Add more comprehensive debug messages as we setup transmit descriptors. Also, dump packets if IFF_DEBUG|IFF_LINK2. Factor out a subroutine for printing transmit-descriptor debug messages.
2004-12-23 08:57:18 +03:00
htole32(RTW_TXCTL0_FS);
#ifdef RTW_DEBUG
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
rtw_print_txdesc(sc, "FS on", ts, tdb, ts->ts_first);
rtw_print_txdesc(sc, "LS on", ts, tdb, ts->ts_last);
Add more comprehensive debug messages as we setup transmit descriptors. Also, dump packets if IFF_DEBUG|IFF_LINK2. Factor out a subroutine for printing transmit-descriptor debug messages.
2004-12-23 08:57:18 +03:00
#endif /* RTW_DEBUG */
Add the transmit section of rtw(4): build up and tear down transmit descriptor rings, move packets from the transmit queues to the transmit rings, handle transmission-completed interrupts. My Linksys WPC11 ver. 4 with Maxim RF, generates interrupts like it is successfully transmitting packets. Clearly there are bugs: my G4 Powerbook locks up hard. I will debug tomorrow.
2004-12-19 11:19:25 +03:00
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
tdb->tdb_nfree -= dmamap->dm_nsegs;
tdb->tdb_next = desc;
Add the transmit section of rtw(4): build up and tear down transmit descriptor rings, move packets from the transmit queues to the transmit rings, handle transmission-completed interrupts. My Linksys WPC11 ver. 4 with Maxim RF, generates interrupts like it is successfully transmitting packets. Clearly there are bugs: my G4 Powerbook locks up hard. I will debug tomorrow.
2004-12-19 11:19:25 +03:00
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
rtw_txdescs_sync(tdb, ts->ts_first, dmamap->dm_nsegs,
Add the transmit section of rtw(4): build up and tear down transmit descriptor rings, move packets from the transmit queues to the transmit rings, handle transmission-completed interrupts. My Linksys WPC11 ver. 4 with Maxim RF, generates interrupts like it is successfully transmitting packets. Clearly there are bugs: my G4 Powerbook locks up hard. I will debug tomorrow.
2004-12-19 11:19:25 +03:00
BUS_DMASYNC_PREREAD|BUS_DMASYNC_PREWRITE);
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
tdb->tdb_desc[ts->ts_first].td_ctl0 |=
Add the transmit section of rtw(4): build up and tear down transmit descriptor rings, move packets from the transmit queues to the transmit rings, handle transmission-completed interrupts. My Linksys WPC11 ver. 4 with Maxim RF, generates interrupts like it is successfully transmitting packets. Clearly there are bugs: my G4 Powerbook locks up hard. I will debug tomorrow.
2004-12-19 11:19:25 +03:00
htole32(RTW_TXCTL0_OWN);
Add more comprehensive debug messages as we setup transmit descriptors. Also, dump packets if IFF_DEBUG|IFF_LINK2. Factor out a subroutine for printing transmit-descriptor debug messages.
2004-12-23 08:57:18 +03:00
#ifdef RTW_DEBUG
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
rtw_print_txdesc(sc, "OWN on", ts, tdb, ts->ts_first);
Add more comprehensive debug messages as we setup transmit descriptors. Also, dump packets if IFF_DEBUG|IFF_LINK2. Factor out a subroutine for printing transmit-descriptor debug messages.
2004-12-23 08:57:18 +03:00
#endif /* RTW_DEBUG */
Add the transmit section of rtw(4): build up and tear down transmit descriptor rings, move packets from the transmit queues to the transmit rings, handle transmission-completed interrupts. My Linksys WPC11 ver. 4 with Maxim RF, generates interrupts like it is successfully transmitting packets. Clearly there are bugs: my G4 Powerbook locks up hard. I will debug tomorrow.
2004-12-19 11:19:25 +03:00
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
rtw_txdescs_sync(tdb, ts->ts_first, 1,
Add the transmit section of rtw(4): build up and tear down transmit descriptor rings, move packets from the transmit queues to the transmit rings, handle transmission-completed interrupts. My Linksys WPC11 ver. 4 with Maxim RF, generates interrupts like it is successfully transmitting packets. Clearly there are bugs: my G4 Powerbook locks up hard. I will debug tomorrow.
2004-12-19 11:19:25 +03:00
BUS_DMASYNC_PREREAD|BUS_DMASYNC_PREWRITE);
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
SIMPLEQ_REMOVE_HEAD(&tsb->tsb_freeq, ts_q);
SIMPLEQ_INSERT_TAIL(&tsb->tsb_dirtyq, ts, ts_q);
Add the transmit section of rtw(4): build up and tear down transmit descriptor rings, move packets from the transmit queues to the transmit rings, handle transmission-completed interrupts. My Linksys WPC11 ver. 4 with Maxim RF, generates interrupts like it is successfully transmitting packets. Clearly there are bugs: my G4 Powerbook locks up hard. I will debug tomorrow.
2004-12-19 11:19:25 +03:00
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
tsb->tsb_tx_timer = 5;
Assert that I'm DMA-syncing a non-negative number of transmit descriptors. As I free dirty transmit buffers, move them to the free list, not back to the dirty list! Add a 5-second transmit timeout.
2004-12-20 03:16:21 +03:00
ifp->if_timer = 1;
Realtek sample sources are not consistent about the "Digital PHY" bit in the SROM. It seems as if it is set to 1 when the PHY is *analog*, not *digital*. Fix my sources. In rtw_intr_rx, use units of 500kb/s instead of 100kb/s for rate, to be consistent with net80211's expectations. Polish up some debugging ugly messages. Dump raw 802.11 packets if IFF_DEBUG|IFF_LINK2 and RTW_DEBUG is defined. Polish power-state (on/sleep/off) handling. Especially improve support for RFMD (totally untested) and Maxim. For Philips, take the Digital PHY property into account. Call the net80211 watchdog function from rtw_watchdog, so that we scan again if auth/assoc fails. Be a little more cautious about writing register[RTW_TPPOLL], since other drivers are.... Don't frob the high/low-priority queues right now, since I don't use them. Add rtw_join_bss which programs the card with the BSSID and other properties of a BSS. Use it on state transitions. Factor out rtw_set_nettype. Make rtw_recv_beacon call ieee80211_recv_mgmt instead of dropping beacons on the floor! TBD IBSS merges. Change some rtw_debug=2 printfs to rtw_debug=3 (RTW_DPRINTF3) printfs so the console doesn't get spammed so badly at rtw_debug=2. Change some debugging printfs to RTW_DPRINTFs. E.g., print the "RF programming method" only if debugging is enabled.
2004-12-21 02:05:41 +03:00
tppoll = RTW_READ8(&sc->sc_regs, RTW_TPPOLL);
/* TBD poke other queues. */
RTW_WRITE8(&sc->sc_regs, RTW_TPPOLL, tppoll | RTW_TPPOLL_NPQ);
Fix bug in last commit: use &sc->sc_regs, regs is not defined.
2004-12-20 04:28:24 +03:00
RTW_SYNC(&sc->sc_regs, RTW_TPPOLL, RTW_TPPOLL);
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
}
If defined(RTW_DEBUG), provide a sysctl (hw.rtw.rxbufs_limit) for limiting the number of rx buffers an rtw may allocate. Use this sysctl to test the code that copes with buffer exhaustion. Allocate at most RTW_RXQLEN rx buffers, stopping at the sysctl limit. Record in sc_nrxdesc how many were allocated, and put the end-of-ring indication on sc_rxdesc[sc_nrxdesc - 1]. In rtw_init, if no rx buffers could be allocated, log a complaint, clear IFF_RUNNING, and exit with an error. Many changes to accomodate a short rx ring, mainly of the "add a rx-ring length argument" variety. XXX I really should consolidate all of the rx ring variables in one struct and pass that to the rx-ring subroutines. Bug fix: after calling MCLGET, use the (m->m_flags & M_EXT) idiom to check for success, instead of m != NULL. Bug fix: at the top of rtw_start, if IFF_RUNNING is not set, or IFF_OACTIVE is, get out.
2004-12-29 01:21:15 +03:00
out:
Change rtw_debug from a debug level to a debug mask. Add a lot of debug flags. From Linux: handle an RTL8180 bug. Sometimes the NIC skips from the middle of the ring to the 0th rx descriptor. Now the driver resynchronizes. Handle a receive descriptor underrun or Rx FIFO overflow condition in the way that the Linux driver does. This kind of seems like overkill, but whatever. Protect rtw_ioctl with splnet(). Do not load a tx descriptor with a buffer shorter than 4 bytes. Handle a transmit timeout less disruptively.
2004-12-25 09:58:37 +03:00
DPRINTF(sc, RTW_DEBUG_XMIT, ("%s: leave\n", __func__));
Add the transmit section of rtw(4): build up and tear down transmit descriptor rings, move packets from the transmit queues to the transmit rings, handle transmission-completed interrupts. My Linksys WPC11 ver. 4 with Maxim RF, generates interrupts like it is successfully transmitting packets. Clearly there are bugs: my G4 Powerbook locks up hard. I will debug tomorrow.
2004-12-19 11:19:25 +03:00
return;
post_load_err:
bus_dmamap_unload(sc->sc_dmat, dmamap);
m_freem(m0);
post_dequeue_err:
ieee80211_release_node(&sc->sc_ic, ni);
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
return;
}
static void
rtw_watchdog(struct ifnet *ifp)
{
Add the transmit section of rtw(4): build up and tear down transmit descriptor rings, move packets from the transmit queues to the transmit rings, handle transmission-completed interrupts. My Linksys WPC11 ver. 4 with Maxim RF, generates interrupts like it is successfully transmitting packets. Clearly there are bugs: my G4 Powerbook locks up hard. I will debug tomorrow.
2004-12-19 11:19:25 +03:00
int pri;
struct rtw_softc *sc;
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
struct rtw_txsoft_blk *tsb;
Add the transmit section of rtw(4): build up and tear down transmit descriptor rings, move packets from the transmit queues to the transmit rings, handle transmission-completed interrupts. My Linksys WPC11 ver. 4 with Maxim RF, generates interrupts like it is successfully transmitting packets. Clearly there are bugs: my G4 Powerbook locks up hard. I will debug tomorrow.
2004-12-19 11:19:25 +03:00
sc = ifp->if_softc;
ifp->if_timer = 0;
if ((sc->sc_flags & RTW_F_ENABLED) == 0)
return;
for (pri = 0; pri < RTW_NTXPRI; pri++) {
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
tsb = &sc->sc_txsoft_blk[pri];
Add the transmit section of rtw(4): build up and tear down transmit descriptor rings, move packets from the transmit queues to the transmit rings, handle transmission-completed interrupts. My Linksys WPC11 ver. 4 with Maxim RF, generates interrupts like it is successfully transmitting packets. Clearly there are bugs: my G4 Powerbook locks up hard. I will debug tomorrow.
2004-12-19 11:19:25 +03:00
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
if (tsb->tsb_tx_timer == 0)
Add the transmit section of rtw(4): build up and tear down transmit descriptor rings, move packets from the transmit queues to the transmit rings, handle transmission-completed interrupts. My Linksys WPC11 ver. 4 with Maxim RF, generates interrupts like it is successfully transmitting packets. Clearly there are bugs: my G4 Powerbook locks up hard. I will debug tomorrow.
2004-12-19 11:19:25 +03:00
continue;
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
if (--tsb->tsb_tx_timer == 0) {
if (SIMPLEQ_EMPTY(&tsb->tsb_dirtyq))
Add the transmit section of rtw(4): build up and tear down transmit descriptor rings, move packets from the transmit queues to the transmit rings, handle transmission-completed interrupts. My Linksys WPC11 ver. 4 with Maxim RF, generates interrupts like it is successfully transmitting packets. Clearly there are bugs: my G4 Powerbook locks up hard. I will debug tomorrow.
2004-12-19 11:19:25 +03:00
continue;
printf("%s: transmit timeout, priority %d\n",
ifp->if_xname, pri);
ifp->if_oerrors++;
Simplify the reset we do on a transmit timeout. Really, this could be as simple as rtw_stop(, 0), rtw_init().
2004-12-27 22:49:16 +03:00
/* Stop Tx DMA, disable transmitter, clear
* Tx rings, and restart.
*/
RTW_WRITE8(&sc->sc_regs, RTW_TPPOLL, RTW_TPPOLL_SNPQ);
RTW_SYNC(&sc->sc_regs, RTW_TPPOLL, RTW_TPPOLL);
rtw_io_enable(&sc->sc_regs, RTW_CR_TE, 0);
Change rtw_debug from a debug level to a debug mask. Add a lot of debug flags. From Linux: handle an RTL8180 bug. Sometimes the NIC skips from the middle of the ring to the 0th rx descriptor. Now the driver resynchronizes. Handle a receive descriptor underrun or Rx FIFO overflow condition in the way that the Linux driver does. This kind of seems like overkill, but whatever. Protect rtw_ioctl with splnet(). Do not load a tx descriptor with a buffer shorter than 4 bytes. Handle a transmit timeout less disruptively.
2004-12-25 09:58:37 +03:00
rtw_txdescs_reset(sc);
Simplify the reset we do on a transmit timeout. Really, this could be as simple as rtw_stop(, 0), rtw_init().
2004-12-27 22:49:16 +03:00
rtw_io_enable(&sc->sc_regs, RTW_CR_TE, 1);
Add the transmit section of rtw(4): build up and tear down transmit descriptor rings, move packets from the transmit queues to the transmit rings, handle transmission-completed interrupts. My Linksys WPC11 ver. 4 with Maxim RF, generates interrupts like it is successfully transmitting packets. Clearly there are bugs: my G4 Powerbook locks up hard. I will debug tomorrow.
2004-12-19 11:19:25 +03:00
rtw_start(ifp);
} else
ifp->if_timer = 1;
}
Realtek sample sources are not consistent about the "Digital PHY" bit in the SROM. It seems as if it is set to 1 when the PHY is *analog*, not *digital*. Fix my sources. In rtw_intr_rx, use units of 500kb/s instead of 100kb/s for rate, to be consistent with net80211's expectations. Polish up some debugging ugly messages. Dump raw 802.11 packets if IFF_DEBUG|IFF_LINK2 and RTW_DEBUG is defined. Polish power-state (on/sleep/off) handling. Especially improve support for RFMD (totally untested) and Maxim. For Philips, take the Digital PHY property into account. Call the net80211 watchdog function from rtw_watchdog, so that we scan again if auth/assoc fails. Be a little more cautious about writing register[RTW_TPPOLL], since other drivers are.... Don't frob the high/low-priority queues right now, since I don't use them. Add rtw_join_bss which programs the card with the BSSID and other properties of a BSS. Use it on state transitions. Factor out rtw_set_nettype. Make rtw_recv_beacon call ieee80211_recv_mgmt instead of dropping beacons on the floor! TBD IBSS merges. Change some rtw_debug=2 printfs to rtw_debug=3 (RTW_DPRINTF3) printfs so the console doesn't get spammed so badly at rtw_debug=2. Change some debugging printfs to RTW_DPRINTFs. E.g., print the "RF programming method" only if debugging is enabled.
2004-12-21 02:05:41 +03:00
ieee80211_watchdog(ifp);
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
return;
}
static void
rtw_start_beacon(struct rtw_softc *sc, int enable)
{
/* TBD */
return;
}
static void
rtw_next_scan(void *arg)
{
struct ieee80211com *ic = arg;
int s;
/* don't call rtw_start w/o network interrupts blocked */
s = splnet();
if (ic->ic_state == IEEE80211_S_SCAN)
ieee80211_next_scan(ic);
splx(s);
}
Realtek sample sources are not consistent about the "Digital PHY" bit in the SROM. It seems as if it is set to 1 when the PHY is *analog*, not *digital*. Fix my sources. In rtw_intr_rx, use units of 500kb/s instead of 100kb/s for rate, to be consistent with net80211's expectations. Polish up some debugging ugly messages. Dump raw 802.11 packets if IFF_DEBUG|IFF_LINK2 and RTW_DEBUG is defined. Polish power-state (on/sleep/off) handling. Especially improve support for RFMD (totally untested) and Maxim. For Philips, take the Digital PHY property into account. Call the net80211 watchdog function from rtw_watchdog, so that we scan again if auth/assoc fails. Be a little more cautious about writing register[RTW_TPPOLL], since other drivers are.... Don't frob the high/low-priority queues right now, since I don't use them. Add rtw_join_bss which programs the card with the BSSID and other properties of a BSS. Use it on state transitions. Factor out rtw_set_nettype. Make rtw_recv_beacon call ieee80211_recv_mgmt instead of dropping beacons on the floor! TBD IBSS merges. Change some rtw_debug=2 printfs to rtw_debug=3 (RTW_DPRINTF3) printfs so the console doesn't get spammed so badly at rtw_debug=2. Change some debugging printfs to RTW_DPRINTFs. E.g., print the "RF programming method" only if debugging is enabled.
2004-12-21 02:05:41 +03:00
static void
rtw_join_bss(struct rtw_softc *sc, uint8_t *bssid, enum ieee80211_opmode opmode,
uint16_t intval0)
{
uint16_t bcnitv, intval;
int i;
struct rtw_regs *regs = &sc->sc_regs;
for (i = 0; i < IEEE80211_ADDR_LEN; i++)
RTW_WRITE8(regs, RTW_BSSID + i, bssid[i]);
RTW_SYNC(regs, RTW_BSSID16, RTW_BSSID32);
rtw_set_access(sc, RTW_ACCESS_CONFIG);
intval = MIN(intval0, PRESHIFT(RTW_BCNITV_BCNITV_MASK));
bcnitv = RTW_READ16(regs, RTW_BCNITV) & ~RTW_BCNITV_BCNITV_MASK;
bcnitv |= LSHIFT(intval, RTW_BCNITV_BCNITV_MASK);
RTW_WRITE16(regs, RTW_BCNITV, bcnitv);
/* magic from Linux */
RTW_WRITE16(regs, RTW_ATIMWND, LSHIFT(1, RTW_ATIMWND_ATIMWND));
RTW_WRITE16(regs, RTW_ATIMTRITV, LSHIFT(2, RTW_ATIMTRITV_ATIMTRITV));
rtw_set_nettype(sc, opmode);
rtw_set_access(sc, RTW_ACCESS_NONE);
/* TBD WEP */
RTW_WRITE8(regs, RTW_SCR, 0);
rtw_io_enable(regs, RTW_CR_RE | RTW_CR_TE, 1);
}
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
/* Synchronize the hardware state with the software state. */
static int
rtw_newstate(struct ieee80211com *ic, enum ieee80211_state nstate, int arg)
{
struct ifnet *ifp = &ic->ic_if;
struct rtw_softc *sc = ifp->if_softc;
enum ieee80211_state ostate;
int error;
ostate = ic->ic_state;
if (nstate == IEEE80211_S_INIT) {
callout_stop(&sc->sc_scan_ch);
sc->sc_cur_chan = IEEE80211_CHAN_ANY;
rtw_start_beacon(sc, 0);
return (*sc->sc_mtbl.mt_newstate)(ic, nstate, arg);
}
if (ostate == IEEE80211_S_INIT && nstate != IEEE80211_S_INIT)
rtw_pwrstate(sc, RTW_ON);
if ((error = rtw_tune(sc)) != 0)
return error;
switch (nstate) {
case IEEE80211_S_ASSOC:
Realtek sample sources are not consistent about the "Digital PHY" bit in the SROM. It seems as if it is set to 1 when the PHY is *analog*, not *digital*. Fix my sources. In rtw_intr_rx, use units of 500kb/s instead of 100kb/s for rate, to be consistent with net80211's expectations. Polish up some debugging ugly messages. Dump raw 802.11 packets if IFF_DEBUG|IFF_LINK2 and RTW_DEBUG is defined. Polish power-state (on/sleep/off) handling. Especially improve support for RFMD (totally untested) and Maxim. For Philips, take the Digital PHY property into account. Call the net80211 watchdog function from rtw_watchdog, so that we scan again if auth/assoc fails. Be a little more cautious about writing register[RTW_TPPOLL], since other drivers are.... Don't frob the high/low-priority queues right now, since I don't use them. Add rtw_join_bss which programs the card with the BSSID and other properties of a BSS. Use it on state transitions. Factor out rtw_set_nettype. Make rtw_recv_beacon call ieee80211_recv_mgmt instead of dropping beacons on the floor! TBD IBSS merges. Change some rtw_debug=2 printfs to rtw_debug=3 (RTW_DPRINTF3) printfs so the console doesn't get spammed so badly at rtw_debug=2. Change some debugging printfs to RTW_DPRINTFs. E.g., print the "RF programming method" only if debugging is enabled.
2004-12-21 02:05:41 +03:00
rtw_join_bss(sc, ic->ic_bss->ni_bssid, ic->ic_opmode,
ic->ic_bss->ni_intval);
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
break;
case IEEE80211_S_INIT:
panic("%s: unexpected state IEEE80211_S_INIT\n", __func__);
break;
case IEEE80211_S_SCAN:
Change rtw_debug from a debug level to a debug mask. Add a lot of debug flags. From Linux: handle an RTL8180 bug. Sometimes the NIC skips from the middle of the ring to the 0th rx descriptor. Now the driver resynchronizes. Handle a receive descriptor underrun or Rx FIFO overflow condition in the way that the Linux driver does. This kind of seems like overkill, but whatever. Protect rtw_ioctl with splnet(). Do not load a tx descriptor with a buffer shorter than 4 bytes. Handle a transmit timeout less disruptively.
2004-12-25 09:58:37 +03:00
if (ostate != IEEE80211_S_SCAN) {
(void)memset(ic->ic_bss->ni_bssid, 0,
IEEE80211_ADDR_LEN);
rtw_join_bss(sc, ic->ic_bss->ni_bssid, ic->ic_opmode,
ic->ic_bss->ni_intval);
}
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
callout_reset(&sc->sc_scan_ch, rtw_dwelltime * hz / 1000,
rtw_next_scan, ic);
break;
case IEEE80211_S_RUN:
if (ic->ic_opmode == IEEE80211_M_STA)
break;
/*FALLTHROUGH*/
case IEEE80211_S_AUTH:
#if 0
rtw_write_bcn_thresh(sc);
rtw_write_sup_rates(sc);
#endif
if (ic->ic_opmode == IEEE80211_M_AHDEMO ||
ic->ic_opmode == IEEE80211_M_MONITOR)
break;
Realtek sample sources are not consistent about the "Digital PHY" bit in the SROM. It seems as if it is set to 1 when the PHY is *analog*, not *digital*. Fix my sources. In rtw_intr_rx, use units of 500kb/s instead of 100kb/s for rate, to be consistent with net80211's expectations. Polish up some debugging ugly messages. Dump raw 802.11 packets if IFF_DEBUG|IFF_LINK2 and RTW_DEBUG is defined. Polish power-state (on/sleep/off) handling. Especially improve support for RFMD (totally untested) and Maxim. For Philips, take the Digital PHY property into account. Call the net80211 watchdog function from rtw_watchdog, so that we scan again if auth/assoc fails. Be a little more cautious about writing register[RTW_TPPOLL], since other drivers are.... Don't frob the high/low-priority queues right now, since I don't use them. Add rtw_join_bss which programs the card with the BSSID and other properties of a BSS. Use it on state transitions. Factor out rtw_set_nettype. Make rtw_recv_beacon call ieee80211_recv_mgmt instead of dropping beacons on the floor! TBD IBSS merges. Change some rtw_debug=2 printfs to rtw_debug=3 (RTW_DPRINTF3) printfs so the console doesn't get spammed so badly at rtw_debug=2. Change some debugging printfs to RTW_DPRINTFs. E.g., print the "RF programming method" only if debugging is enabled.
2004-12-21 02:05:41 +03:00
/* TBD set listen interval */
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
#if 0
rtw_tsf(sc);
#endif
break;
}
if (nstate != IEEE80211_S_SCAN)
callout_stop(&sc->sc_scan_ch);
if (nstate == IEEE80211_S_RUN &&
(ic->ic_opmode == IEEE80211_M_HOSTAP ||
ic->ic_opmode == IEEE80211_M_IBSS))
rtw_start_beacon(sc, 1);
else
rtw_start_beacon(sc, 0);
return (*sc->sc_mtbl.mt_newstate)(ic, nstate, arg);
}
static void
Realtek sample sources are not consistent about the "Digital PHY" bit in the SROM. It seems as if it is set to 1 when the PHY is *analog*, not *digital*. Fix my sources. In rtw_intr_rx, use units of 500kb/s instead of 100kb/s for rate, to be consistent with net80211's expectations. Polish up some debugging ugly messages. Dump raw 802.11 packets if IFF_DEBUG|IFF_LINK2 and RTW_DEBUG is defined. Polish power-state (on/sleep/off) handling. Especially improve support for RFMD (totally untested) and Maxim. For Philips, take the Digital PHY property into account. Call the net80211 watchdog function from rtw_watchdog, so that we scan again if auth/assoc fails. Be a little more cautious about writing register[RTW_TPPOLL], since other drivers are.... Don't frob the high/low-priority queues right now, since I don't use them. Add rtw_join_bss which programs the card with the BSSID and other properties of a BSS. Use it on state transitions. Factor out rtw_set_nettype. Make rtw_recv_beacon call ieee80211_recv_mgmt instead of dropping beacons on the floor! TBD IBSS merges. Change some rtw_debug=2 printfs to rtw_debug=3 (RTW_DPRINTF3) printfs so the console doesn't get spammed so badly at rtw_debug=2. Change some debugging printfs to RTW_DPRINTFs. E.g., print the "RF programming method" only if debugging is enabled.
2004-12-21 02:05:41 +03:00
rtw_recv_beacon(struct rtw_softc *sc, struct mbuf *m,
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
struct ieee80211_node *ni, int subtype, int rssi, u_int32_t rstamp)
{
Realtek sample sources are not consistent about the "Digital PHY" bit in the SROM. It seems as if it is set to 1 when the PHY is *analog*, not *digital*. Fix my sources. In rtw_intr_rx, use units of 500kb/s instead of 100kb/s for rate, to be consistent with net80211's expectations. Polish up some debugging ugly messages. Dump raw 802.11 packets if IFF_DEBUG|IFF_LINK2 and RTW_DEBUG is defined. Polish power-state (on/sleep/off) handling. Especially improve support for RFMD (totally untested) and Maxim. For Philips, take the Digital PHY property into account. Call the net80211 watchdog function from rtw_watchdog, so that we scan again if auth/assoc fails. Be a little more cautious about writing register[RTW_TPPOLL], since other drivers are.... Don't frob the high/low-priority queues right now, since I don't use them. Add rtw_join_bss which programs the card with the BSSID and other properties of a BSS. Use it on state transitions. Factor out rtw_set_nettype. Make rtw_recv_beacon call ieee80211_recv_mgmt instead of dropping beacons on the floor! TBD IBSS merges. Change some rtw_debug=2 printfs to rtw_debug=3 (RTW_DPRINTF3) printfs so the console doesn't get spammed so badly at rtw_debug=2. Change some debugging printfs to RTW_DPRINTFs. E.g., print the "RF programming method" only if debugging is enabled.
2004-12-21 02:05:41 +03:00
(*sc->sc_mtbl.mt_recv_mgmt)(&sc->sc_ic, m, ni, subtype, rssi, rstamp);
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
return;
}
static void
rtw_recv_mgmt(struct ieee80211com *ic, struct mbuf *m,
struct ieee80211_node *ni, int subtype, int rssi, u_int32_t rstamp)
{
struct rtw_softc *sc = (struct rtw_softc*)ic->ic_softc;
switch (subtype) {
case IEEE80211_FC0_SUBTYPE_PROBE_REQ:
/* do nothing: hardware answers probe request XXX */
break;
case IEEE80211_FC0_SUBTYPE_PROBE_RESP:
case IEEE80211_FC0_SUBTYPE_BEACON:
Realtek sample sources are not consistent about the "Digital PHY" bit in the SROM. It seems as if it is set to 1 when the PHY is *analog*, not *digital*. Fix my sources. In rtw_intr_rx, use units of 500kb/s instead of 100kb/s for rate, to be consistent with net80211's expectations. Polish up some debugging ugly messages. Dump raw 802.11 packets if IFF_DEBUG|IFF_LINK2 and RTW_DEBUG is defined. Polish power-state (on/sleep/off) handling. Especially improve support for RFMD (totally untested) and Maxim. For Philips, take the Digital PHY property into account. Call the net80211 watchdog function from rtw_watchdog, so that we scan again if auth/assoc fails. Be a little more cautious about writing register[RTW_TPPOLL], since other drivers are.... Don't frob the high/low-priority queues right now, since I don't use them. Add rtw_join_bss which programs the card with the BSSID and other properties of a BSS. Use it on state transitions. Factor out rtw_set_nettype. Make rtw_recv_beacon call ieee80211_recv_mgmt instead of dropping beacons on the floor! TBD IBSS merges. Change some rtw_debug=2 printfs to rtw_debug=3 (RTW_DPRINTF3) printfs so the console doesn't get spammed so badly at rtw_debug=2. Change some debugging printfs to RTW_DPRINTFs. E.g., print the "RF programming method" only if debugging is enabled.
2004-12-21 02:05:41 +03:00
rtw_recv_beacon(sc, m, ni, subtype, rssi, rstamp);
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
break;
default:
(*sc->sc_mtbl.mt_recv_mgmt)(ic, m, ni, subtype, rssi, rstamp);
break;
}
return;
}
static struct ieee80211_node *
rtw_node_alloc(struct ieee80211com *ic)
{
struct rtw_softc *sc = (struct rtw_softc *)ic->ic_if.if_softc;
struct ieee80211_node *ni = (*sc->sc_mtbl.mt_node_alloc)(ic);
Change rtw_debug from a debug level to a debug mask. Add a lot of debug flags. From Linux: handle an RTL8180 bug. Sometimes the NIC skips from the middle of the ring to the 0th rx descriptor. Now the driver resynchronizes. Handle a receive descriptor underrun or Rx FIFO overflow condition in the way that the Linux driver does. This kind of seems like overkill, but whatever. Protect rtw_ioctl with splnet(). Do not load a tx descriptor with a buffer shorter than 4 bytes. Handle a transmit timeout less disruptively.
2004-12-25 09:58:37 +03:00
DPRINTF(sc, RTW_DEBUG_NODE,
("%s: alloc node %p\n", sc->sc_dev.dv_xname, ni));
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
return ni;
}
static void
rtw_node_free(struct ieee80211com *ic, struct ieee80211_node *ni)
{
struct rtw_softc *sc = (struct rtw_softc *)ic->ic_if.if_softc;
Change rtw_debug from a debug level to a debug mask. Add a lot of debug flags. From Linux: handle an RTL8180 bug. Sometimes the NIC skips from the middle of the ring to the 0th rx descriptor. Now the driver resynchronizes. Handle a receive descriptor underrun or Rx FIFO overflow condition in the way that the Linux driver does. This kind of seems like overkill, but whatever. Protect rtw_ioctl with splnet(). Do not load a tx descriptor with a buffer shorter than 4 bytes. Handle a transmit timeout less disruptively.
2004-12-25 09:58:37 +03:00
DPRINTF(sc, RTW_DEBUG_NODE,
("%s: freeing node %p %s\n", sc->sc_dev.dv_xname, ni,
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
ether_sprintf(ni->ni_bssid)));
(*sc->sc_mtbl.mt_node_free)(ic, ni);
}
static int
rtw_media_change(struct ifnet *ifp)
{
int error;
error = ieee80211_media_change(ifp);
if (error == ENETRESET) {
if ((ifp->if_flags & (IFF_RUNNING|IFF_UP)) ==
(IFF_RUNNING|IFF_UP))
rtw_init(ifp); /* XXX lose error */
error = 0;
}
return error;
}
static void
rtw_media_status(struct ifnet *ifp, struct ifmediareq *imr)
{
struct rtw_softc *sc = ifp->if_softc;
if ((sc->sc_flags & RTW_F_ENABLED) == 0) {
imr->ifm_active = IFM_IEEE80211 | IFM_NONE;
imr->ifm_status = 0;
return;
}
ieee80211_media_status(ifp, imr);
}
void
rtw_power(int why, void *arg)
{
struct rtw_softc *sc = arg;
struct ifnet *ifp = &sc->sc_ic.ic_if;
int s;
Change rtw_debug from a debug level to a debug mask. Add a lot of debug flags. From Linux: handle an RTL8180 bug. Sometimes the NIC skips from the middle of the ring to the 0th rx descriptor. Now the driver resynchronizes. Handle a receive descriptor underrun or Rx FIFO overflow condition in the way that the Linux driver does. This kind of seems like overkill, but whatever. Protect rtw_ioctl with splnet(). Do not load a tx descriptor with a buffer shorter than 4 bytes. Handle a transmit timeout less disruptively.
2004-12-25 09:58:37 +03:00
DPRINTF(sc, RTW_DEBUG_PWR,
("%s: rtw_power(%d,)\n", sc->sc_dev.dv_xname, why));
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
s = splnet();
switch (why) {
case PWR_STANDBY:
/* XXX do nothing. */
break;
case PWR_SUSPEND:
rtw_stop(ifp, 0);
if (sc->sc_power != NULL)
(*sc->sc_power)(sc, why);
break;
case PWR_RESUME:
if (ifp->if_flags & IFF_UP) {
if (sc->sc_power != NULL)
(*sc->sc_power)(sc, why);
rtw_init(ifp);
}
break;
case PWR_SOFTSUSPEND:
case PWR_SOFTSTANDBY:
case PWR_SOFTRESUME:
break;
}
splx(s);
}
/* rtw_shutdown: make sure the interface is stopped at reboot time. */
void
rtw_shutdown(void *arg)
{
struct rtw_softc *sc = arg;
rtw_stop(&sc->sc_ic.ic_if, 1);
}
static __inline void
Get rid of the silly char (*dvname)[IFNAMSIZ] pattern. Replace it with const char *dvname. I'm not sure what I was thinking.
2004-12-20 03:28:02 +03:00
rtw_setifprops(struct ifnet *ifp, const char *dvname, void *softc)
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
{
Get rid of the silly char (*dvname)[IFNAMSIZ] pattern. Replace it with const char *dvname. I'm not sure what I was thinking.
2004-12-20 03:28:02 +03:00
(void)memcpy(ifp->if_xname, dvname, IFNAMSIZ);
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
ifp->if_softc = softc;
ifp->if_flags = IFF_SIMPLEX | IFF_BROADCAST | IFF_MULTICAST |
IFF_NOTRAILERS;
ifp->if_ioctl = rtw_ioctl;
ifp->if_start = rtw_start;
ifp->if_watchdog = rtw_watchdog;
ifp->if_init = rtw_init;
ifp->if_stop = rtw_stop;
}
static __inline void
rtw_set80211props(struct ieee80211com *ic)
{
int nrate;
ic->ic_phytype = IEEE80211_T_DS;
ic->ic_opmode = IEEE80211_M_STA;
ic->ic_caps = IEEE80211_C_PMGT | IEEE80211_C_IBSS |
IEEE80211_C_HOSTAP | IEEE80211_C_MONITOR | IEEE80211_C_WEP;
nrate = 0;
When we set up the supported rates table, mark 1Mb/s, 2Mb/s as basic rates.
2004-12-23 08:47:42 +03:00
ic->ic_sup_rates[IEEE80211_MODE_11B].rs_rates[nrate++] =
IEEE80211_RATE_BASIC | 2;
ic->ic_sup_rates[IEEE80211_MODE_11B].rs_rates[nrate++] =
IEEE80211_RATE_BASIC | 4;
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
ic->ic_sup_rates[IEEE80211_MODE_11B].rs_rates[nrate++] = 11;
ic->ic_sup_rates[IEEE80211_MODE_11B].rs_rates[nrate++] = 22;
ic->ic_sup_rates[IEEE80211_MODE_11B].rs_nrates = nrate;
}
static __inline void
rtw_set80211methods(struct rtw_mtbl *mtbl, struct ieee80211com *ic)
{
mtbl->mt_newstate = ic->ic_newstate;
ic->ic_newstate = rtw_newstate;
mtbl->mt_recv_mgmt = ic->ic_recv_mgmt;
ic->ic_recv_mgmt = rtw_recv_mgmt;
mtbl->mt_node_free = ic->ic_node_free;
ic->ic_node_free = rtw_node_free;
mtbl->mt_node_alloc = ic->ic_node_alloc;
ic->ic_node_alloc = rtw_node_alloc;
}
static __inline void
Get rid of the silly char (*dvname)[IFNAMSIZ] pattern. Replace it with const char *dvname. I'm not sure what I was thinking.
2004-12-20 03:28:02 +03:00
rtw_establish_hooks(struct rtw_hooks *hooks, const char *dvname,
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
void *arg)
{
/*
* Make sure the interface is shutdown during reboot.
*/
hooks->rh_shutdown = shutdownhook_establish(rtw_shutdown, arg);
if (hooks->rh_shutdown == NULL)
printf("%s: WARNING: unable to establish shutdown hook\n",
Get rid of the silly char (*dvname)[IFNAMSIZ] pattern. Replace it with const char *dvname. I'm not sure what I was thinking.
2004-12-20 03:28:02 +03:00
dvname);
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
/*
* Add a suspend hook to make sure we come back up after a
* resume.
*/
hooks->rh_power = powerhook_establish(rtw_power, arg);
if (hooks->rh_power == NULL)
printf("%s: WARNING: unable to establish power hook\n",
Get rid of the silly char (*dvname)[IFNAMSIZ] pattern. Replace it with const char *dvname. I'm not sure what I was thinking.
2004-12-20 03:28:02 +03:00
dvname);
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
}
static __inline void
Get rid of the silly char (*dvname)[IFNAMSIZ] pattern. Replace it with const char *dvname. I'm not sure what I was thinking.
2004-12-20 03:28:02 +03:00
rtw_disestablish_hooks(struct rtw_hooks *hooks, const char *dvname,
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
void *arg)
{
if (hooks->rh_shutdown != NULL)
shutdownhook_disestablish(hooks->rh_shutdown);
if (hooks->rh_power != NULL)
powerhook_disestablish(hooks->rh_power);
}
static __inline void
rtw_init_radiotap(struct rtw_softc *sc)
{
memset(&sc->sc_rxtapu, 0, sizeof(sc->sc_rxtapu));
Add bpf(4) packet taps to rtw(4). Now you can tap DLT_EN10MB (Ethernet II), DLT_IEEE802_11 (802.11), and DLT_IEEE802_11_RADIO (radiotap + 802.11).
2004-12-29 01:30:07 +03:00
sc->sc_rxtap.rr_ihdr.it_len = htole16(sizeof(sc->sc_rxtapu));
sc->sc_rxtap.rr_ihdr.it_present = htole32(RTW_RX_RADIOTAP_PRESENT);
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
memset(&sc->sc_txtapu, 0, sizeof(sc->sc_txtapu));
Add bpf(4) packet taps to rtw(4). Now you can tap DLT_EN10MB (Ethernet II), DLT_IEEE802_11 (802.11), and DLT_IEEE802_11_RADIO (radiotap + 802.11).
2004-12-29 01:30:07 +03:00
sc->sc_txtap.rt_ihdr.it_len = htole16(sizeof(sc->sc_txtapu));
sc->sc_txtap.rt_ihdr.it_present = htole32(RTW_TX_RADIOTAP_PRESENT);
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
}
static int
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
rtw_txsoft_blk_setup(struct rtw_txsoft_blk *tsb, u_int qlen)
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
{
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
SIMPLEQ_INIT(&tsb->tsb_dirtyq);
SIMPLEQ_INIT(&tsb->tsb_freeq);
tsb->tsb_ndesc = qlen;
tsb->tsb_desc = malloc(qlen * sizeof(*tsb->tsb_desc), M_DEVBUF,
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
M_NOWAIT);
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
if (tsb->tsb_desc == NULL)
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
return ENOMEM;
return 0;
}
static void
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
rtw_txsoft_blk_cleanup_all(struct rtw_softc *sc)
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
{
int pri;
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
struct rtw_txsoft_blk *tsb;
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
Change rtw_debug from a debug level to a debug mask. Add a lot of debug flags. From Linux: handle an RTL8180 bug. Sometimes the NIC skips from the middle of the ring to the 0th rx descriptor. Now the driver resynchronizes. Handle a receive descriptor underrun or Rx FIFO overflow condition in the way that the Linux driver does. This kind of seems like overkill, but whatever. Protect rtw_ioctl with splnet(). Do not load a tx descriptor with a buffer shorter than 4 bytes. Handle a transmit timeout less disruptively.
2004-12-25 09:58:37 +03:00
for (pri = 0; pri < RTW_NTXPRI; pri++) {
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
tsb = &sc->sc_txsoft_blk[pri];
free(tsb->tsb_desc, M_DEVBUF);
tsb->tsb_desc = NULL;
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
}
}
static int
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
rtw_txsoft_blk_setup_all(struct rtw_softc *sc)
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
{
int pri, rc = 0;
int qlen[RTW_NTXPRI] =
{RTW_TXQLENLO, RTW_TXQLENMD, RTW_TXQLENHI, RTW_TXQLENBCN};
Change rtw_debug from a debug level to a debug mask. Add a lot of debug flags. From Linux: handle an RTL8180 bug. Sometimes the NIC skips from the middle of the ring to the 0th rx descriptor. Now the driver resynchronizes. Handle a receive descriptor underrun or Rx FIFO overflow condition in the way that the Linux driver does. This kind of seems like overkill, but whatever. Protect rtw_ioctl with splnet(). Do not load a tx descriptor with a buffer shorter than 4 bytes. Handle a transmit timeout less disruptively.
2004-12-25 09:58:37 +03:00
for (pri = 0; pri < RTW_NTXPRI; pri++) {
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
rc = rtw_txsoft_blk_setup(&sc->sc_txsoft_blk[pri], qlen[pri]);
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
if (rc != 0)
break;
}
return rc;
}
static void
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
rtw_txdesc_blk_setup(struct rtw_txdesc_blk *tdb, struct rtw_txdesc *desc,
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
u_int ndesc, bus_addr_t ofs, bus_addr_t physbase)
{
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
tdb->tdb_ndesc = ndesc;
tdb->tdb_desc = desc;
tdb->tdb_physbase = physbase;
tdb->tdb_ofs = ofs;
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
(void)memset(tdb->tdb_desc, 0,
sizeof(tdb->tdb_desc[0]) * tdb->tdb_ndesc);
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
rtw_txdesc_blk_reset(tdb);
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
}
static void
rtw_txdesc_blk_setup_all(struct rtw_softc *sc)
{
rtw_txdesc_blk_setup(&sc->sc_txdesc_blk[RTW_TXPRILO],
&sc->sc_descs->hd_txlo[0], RTW_NTXDESCLO,
RTW_RING_OFFSET(hd_txlo), RTW_RING_BASE(sc, hd_txlo));
rtw_txdesc_blk_setup(&sc->sc_txdesc_blk[RTW_TXPRIMD],
&sc->sc_descs->hd_txmd[0], RTW_NTXDESCMD,
RTW_RING_OFFSET(hd_txmd), RTW_RING_BASE(sc, hd_txmd));
rtw_txdesc_blk_setup(&sc->sc_txdesc_blk[RTW_TXPRIHI],
&sc->sc_descs->hd_txhi[0], RTW_NTXDESCHI,
RTW_RING_OFFSET(hd_txhi), RTW_RING_BASE(sc, hd_txhi));
rtw_txdesc_blk_setup(&sc->sc_txdesc_blk[RTW_TXPRIBCN],
&sc->sc_descs->hd_bcn[0], RTW_NTXDESCBCN,
RTW_RING_OFFSET(hd_bcn), RTW_RING_BASE(sc, hd_bcn));
}
static struct rtw_rf *
rtw_rf_attach(struct rtw_softc *sc, enum rtw_rfchipid rfchipid,
rtw_rf_write_t rf_write, int digphy)
{
struct rtw_rf *rf;
switch (rfchipid) {
case RTW_RFCHIPID_MAXIM:
rf = rtw_max2820_create(&sc->sc_regs, rf_write, 0);
sc->sc_pwrstate_cb = rtw_maxim_pwrstate;
break;
case RTW_RFCHIPID_PHILIPS:
rf = rtw_sa2400_create(&sc->sc_regs, rf_write, digphy);
sc->sc_pwrstate_cb = rtw_philips_pwrstate;
break;
Realtek sample sources are not consistent about the "Digital PHY" bit in the SROM. It seems as if it is set to 1 when the PHY is *analog*, not *digital*. Fix my sources. In rtw_intr_rx, use units of 500kb/s instead of 100kb/s for rate, to be consistent with net80211's expectations. Polish up some debugging ugly messages. Dump raw 802.11 packets if IFF_DEBUG|IFF_LINK2 and RTW_DEBUG is defined. Polish power-state (on/sleep/off) handling. Especially improve support for RFMD (totally untested) and Maxim. For Philips, take the Digital PHY property into account. Call the net80211 watchdog function from rtw_watchdog, so that we scan again if auth/assoc fails. Be a little more cautious about writing register[RTW_TPPOLL], since other drivers are.... Don't frob the high/low-priority queues right now, since I don't use them. Add rtw_join_bss which programs the card with the BSSID and other properties of a BSS. Use it on state transitions. Factor out rtw_set_nettype. Make rtw_recv_beacon call ieee80211_recv_mgmt instead of dropping beacons on the floor! TBD IBSS merges. Change some rtw_debug=2 printfs to rtw_debug=3 (RTW_DPRINTF3) printfs so the console doesn't get spammed so badly at rtw_debug=2. Change some debugging printfs to RTW_DPRINTFs. E.g., print the "RF programming method" only if debugging is enabled.
2004-12-21 02:05:41 +03:00
case RTW_RFCHIPID_RFMD:
/* XXX RFMD has no RF constructor */
sc->sc_pwrstate_cb = rtw_rfmd_pwrstate;
/*FALLTHROUGH*/
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
default:
return NULL;
}
rf->rf_continuous_tx_cb =
(rtw_continuous_tx_cb_t)rtw_continuous_tx_enable;
rf->rf_continuous_tx_arg = (void *)sc;
return rf;
}
/* Revision C and later use a different PHY delay setting than
* revisions A and B.
*/
static u_int8_t
rtw_check_phydelay(struct rtw_regs *regs, u_int32_t rcr0)
{
#define REVAB (RTW_RCR_MXDMA_UNLIMITED | RTW_RCR_AICV)
#define REVC (REVAB | RTW_RCR_RXFTH_WHOLE)
u_int8_t phydelay = LSHIFT(0x6, RTW_PHYDELAY_PHYDELAY);
RTW_WRITE(regs, RTW_RCR, REVAB);
Add some bus barriers. Remark where they are needed, e.g., read_eeprom. Change some printfs to RTW_DPRINTF2s to quiet rtw(4) when debugging is disabled.
2004-12-20 04:13:45 +03:00
RTW_WBW(regs, RTW_RCR, RTW_RCR);
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
RTW_WRITE(regs, RTW_RCR, REVC);
RTW_WBR(regs, RTW_RCR, RTW_RCR);
if ((RTW_READ(regs, RTW_RCR) & REVC) == REVC)
phydelay |= RTW_PHYDELAY_REVC_MAGIC;
RTW_WRITE(regs, RTW_RCR, rcr0); /* restore RCR */
Add some bus barriers. Remark where they are needed, e.g., read_eeprom. Change some printfs to RTW_DPRINTF2s to quiet rtw(4) when debugging is disabled.
2004-12-20 04:13:45 +03:00
RTW_SYNC(regs, RTW_RCR, RTW_RCR);
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
return phydelay;
#undef REVC
}
void
rtw_attach(struct rtw_softc *sc)
{
rtw_rf_write_t rf_write;
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
struct rtw_txsoft_blk *tsb;
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
int pri, rc, vers;
NEXT_ATTACH_STATE(sc, DETACHED);
switch (RTW_READ(&sc->sc_regs, RTW_TCR) & RTW_TCR_HWVERID_MASK) {
case RTW_TCR_HWVERID_F:
vers = 'F';
rf_write = rtw_rf_hostwrite;
break;
case RTW_TCR_HWVERID_D:
vers = 'D';
At last, I have rtw w/ Philips RF receiving packets. I added some sysctls to aid debugging: * hw.rtw.debug -- enable debugging * hw.rtw.flush_rfio -- Linux voodoo: possibly makes the MAC "flush" bits down the serial bus to the RF * hw.rtw.host_rfio: force the host to bang bits to the RF, instead of the MAC banging bits * hw.rtw.rfio_delay: after telling the MAC to bang bits to the RF front-end, delay rfio_delay microseconds. * hw.rtw.rfprog_fallback: there is this notion of the "RF programming method." I believe the choice influences the polarity/timing of the serial bus used to program the RF front-end. I know the correct choice for Intersil/RFMD/Philips front-ends, only. For all other front-ends, I "fallback" to rfprog_fallback. Make rtw_txdac_enable take an rtw_softc argument. I will probably revert this change. Add some Linux voodoo to rtw_continuous_tx_enable. I will probably revert this change. Important: add rtw_set_rfprog, which sets the correct RF programming method. This change and the following change are probably responsible for making the Philips RF work. Important: RTW_CONFIG1 is an 8-bit register, treat it that way! Important: RTW_BRSR is 16-bit, RTW_CRCOUNT, RTW_PHYDELAY, and RTW_MSR are 8-bit: treat them that way! Vastly simplify rtw_resume_ticks. Note to self: set the LED state to match the power state. Hedge against the possibility that RTW_MSR is protected as RTW_CONFIG[0123] are, meanwhile reworking that section of rtw_init a little. Add sc_anaparm, which isn't used, yet....
2004-12-13 03:48:02 +03:00
if (rtw_host_rfio)
rf_write = rtw_rf_hostwrite;
else
rf_write = rtw_rf_macwrite;
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
break;
default:
vers = '?';
rf_write = rtw_rf_macwrite;
break;
}
printf("%s: hardware version %c\n", sc->sc_dev.dv_xname, vers);
rc = bus_dmamem_alloc(sc->sc_dmat, sizeof(struct rtw_descs),
RTW_DESC_ALIGNMENT, 0, &sc->sc_desc_segs, 1, &sc->sc_desc_nsegs,
0);
if (rc != 0) {
printf("%s: could not allocate hw descriptors, error %d\n",
sc->sc_dev.dv_xname, rc);
goto err;
}
NEXT_ATTACH_STATE(sc, FINISH_DESC_ALLOC);
rc = bus_dmamem_map(sc->sc_dmat, &sc->sc_desc_segs,
sc->sc_desc_nsegs, sizeof(struct rtw_descs),
(caddr_t*)&sc->sc_descs, BUS_DMA_COHERENT);
if (rc != 0) {
printf("%s: could not map hw descriptors, error %d\n",
sc->sc_dev.dv_xname, rc);
goto err;
}
NEXT_ATTACH_STATE(sc, FINISH_DESC_MAP);
rc = bus_dmamap_create(sc->sc_dmat, sizeof(struct rtw_descs), 1,
sizeof(struct rtw_descs), 0, 0, &sc->sc_desc_dmamap);
if (rc != 0) {
printf("%s: could not create DMA map for hw descriptors, "
"error %d\n", sc->sc_dev.dv_xname, rc);
goto err;
}
NEXT_ATTACH_STATE(sc, FINISH_DESCMAP_CREATE);
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
sc->sc_rxdesc_blk.rdb_dmat = sc->sc_dmat;
sc->sc_rxdesc_blk.rdb_dmamap = sc->sc_desc_dmamap;
Consolidate variables related to the rx ring in sc_rxdesc_blk, which is a struct rtw_rxdesc_blk. Put a copy of the DMA tag and the DMA map into the rx- and tx-ring blocks so that I don't have to pass them to subroutines all of the time.
2004-12-29 03:21:37 +03:00
for (pri = 0; pri < RTW_NTXPRI; pri++) {
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
sc->sc_txdesc_blk[pri].tdb_dmat = sc->sc_dmat;
sc->sc_txdesc_blk[pri].tdb_dmamap = sc->sc_desc_dmamap;
Consolidate variables related to the rx ring in sc_rxdesc_blk, which is a struct rtw_rxdesc_blk. Put a copy of the DMA tag and the DMA map into the rx- and tx-ring blocks so that I don't have to pass them to subroutines all of the time.
2004-12-29 03:21:37 +03:00
}
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
rc = bus_dmamap_load(sc->sc_dmat, sc->sc_desc_dmamap, sc->sc_descs,
sizeof(struct rtw_descs), NULL, 0);
if (rc != 0) {
printf("%s: could not load DMA map for hw descriptors, "
"error %d\n", sc->sc_dev.dv_xname, rc);
goto err;
}
NEXT_ATTACH_STATE(sc, FINISH_DESCMAP_LOAD);
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
if (rtw_txsoft_blk_setup_all(sc) != 0)
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
goto err;
NEXT_ATTACH_STATE(sc, FINISH_TXCTLBLK_SETUP);
rtw_txdesc_blk_setup_all(sc);
NEXT_ATTACH_STATE(sc, FINISH_TXDESCBLK_SETUP);
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
sc->sc_rxdesc_blk.rdb_desc = &sc->sc_descs->hd_rx[0];
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
for (pri = 0; pri < RTW_NTXPRI; pri++) {
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
tsb = &sc->sc_txsoft_blk[pri];
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
if ((rc = rtw_txdesc_dmamaps_create(sc->sc_dmat,
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
&tsb->tsb_desc[0], tsb->tsb_ndesc)) != 0) {
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
printf("%s: could not load DMA map for "
"hw tx descriptors, error %d\n",
sc->sc_dev.dv_xname, rc);
goto err;
}
}
NEXT_ATTACH_STATE(sc, FINISH_TXMAPS_CREATE);
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
if ((rc = rtw_rxdesc_dmamaps_create(sc->sc_dmat, &sc->sc_rxsoft[0],
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
RTW_RXQLEN)) != 0) {
printf("%s: could not load DMA map for hw rx descriptors, "
"error %d\n", sc->sc_dev.dv_xname, rc);
goto err;
}
NEXT_ATTACH_STATE(sc, FINISH_RXMAPS_CREATE);
/* Reset the chip to a known state. */
if (rtw_reset(sc) != 0)
goto err;
NEXT_ATTACH_STATE(sc, FINISH_RESET);
sc->sc_rcr = RTW_READ(&sc->sc_regs, RTW_RCR);
if ((sc->sc_rcr & RTW_RCR_9356SEL) != 0)
sc->sc_flags |= RTW_F_9356SROM;
if (rtw_srom_read(&sc->sc_regs, sc->sc_flags, &sc->sc_srom,
Get rid of the silly char (*dvname)[IFNAMSIZ] pattern. Replace it with const char *dvname. I'm not sure what I was thinking.
2004-12-20 03:28:02 +03:00
sc->sc_dev.dv_xname) != 0)
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
goto err;
NEXT_ATTACH_STATE(sc, FINISH_READ_SROM);
if (rtw_srom_parse(&sc->sc_srom, &sc->sc_flags, &sc->sc_csthr,
&sc->sc_rfchipid, &sc->sc_rcr, &sc->sc_locale,
Get rid of the silly char (*dvname)[IFNAMSIZ] pattern. Replace it with const char *dvname. I'm not sure what I was thinking.
2004-12-20 03:28:02 +03:00
sc->sc_dev.dv_xname) != 0) {
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
printf("%s: attach failed, malformed serial ROM\n",
sc->sc_dev.dv_xname);
goto err;
}
Realtek sample sources are not consistent about the "Digital PHY" bit in the SROM. It seems as if it is set to 1 when the PHY is *analog*, not *digital*. Fix my sources. In rtw_intr_rx, use units of 500kb/s instead of 100kb/s for rate, to be consistent with net80211's expectations. Polish up some debugging ugly messages. Dump raw 802.11 packets if IFF_DEBUG|IFF_LINK2 and RTW_DEBUG is defined. Polish power-state (on/sleep/off) handling. Especially improve support for RFMD (totally untested) and Maxim. For Philips, take the Digital PHY property into account. Call the net80211 watchdog function from rtw_watchdog, so that we scan again if auth/assoc fails. Be a little more cautious about writing register[RTW_TPPOLL], since other drivers are.... Don't frob the high/low-priority queues right now, since I don't use them. Add rtw_join_bss which programs the card with the BSSID and other properties of a BSS. Use it on state transitions. Factor out rtw_set_nettype. Make rtw_recv_beacon call ieee80211_recv_mgmt instead of dropping beacons on the floor! TBD IBSS merges. Change some rtw_debug=2 printfs to rtw_debug=3 (RTW_DPRINTF3) printfs so the console doesn't get spammed so badly at rtw_debug=2. Change some debugging printfs to RTW_DPRINTFs. E.g., print the "RF programming method" only if debugging is enabled.
2004-12-21 02:05:41 +03:00
printf("%s: %s PHY\n", sc->sc_dev.dv_xname,
((sc->sc_flags & RTW_F_DIGPHY) != 0) ? "digital" : "analog");
printf("%s: CS threshold %u\n", sc->sc_dev.dv_xname, sc->sc_csthr);
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
NEXT_ATTACH_STATE(sc, FINISH_PARSE_SROM);
sc->sc_rf = rtw_rf_attach(sc, sc->sc_rfchipid, rf_write,
sc->sc_flags & RTW_F_DIGPHY);
if (sc->sc_rf == NULL) {
printf("%s: attach failed, could not attach RF\n",
sc->sc_dev.dv_xname);
goto err;
}
NEXT_ATTACH_STATE(sc, FINISH_RF_ATTACH);
sc->sc_phydelay = rtw_check_phydelay(&sc->sc_regs, sc->sc_rcr);
Change rtw_debug from a debug level to a debug mask. Add a lot of debug flags. From Linux: handle an RTL8180 bug. Sometimes the NIC skips from the middle of the ring to the 0th rx descriptor. Now the driver resynchronizes. Handle a receive descriptor underrun or Rx FIFO overflow condition in the way that the Linux driver does. This kind of seems like overkill, but whatever. Protect rtw_ioctl with splnet(). Do not load a tx descriptor with a buffer shorter than 4 bytes. Handle a transmit timeout less disruptively.
2004-12-25 09:58:37 +03:00
RTW_DPRINTF(RTW_DEBUG_ATTACH,
("%s: PHY delay %d\n", sc->sc_dev.dv_xname, sc->sc_phydelay));
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
if (sc->sc_locale == RTW_LOCALE_UNKNOWN)
rtw_identify_country(&sc->sc_regs, &sc->sc_locale,
Get rid of the silly char (*dvname)[IFNAMSIZ] pattern. Replace it with const char *dvname. I'm not sure what I was thinking.
2004-12-20 03:28:02 +03:00
sc->sc_dev.dv_xname);
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
rtw_init_channels(sc->sc_locale, &sc->sc_ic.ic_channels,
Get rid of the silly char (*dvname)[IFNAMSIZ] pattern. Replace it with const char *dvname. I'm not sure what I was thinking.
2004-12-20 03:28:02 +03:00
sc->sc_dev.dv_xname);
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
if (rtw_identify_sta(&sc->sc_regs, &sc->sc_ic.ic_myaddr,
Get rid of the silly char (*dvname)[IFNAMSIZ] pattern. Replace it with const char *dvname. I'm not sure what I was thinking.
2004-12-20 03:28:02 +03:00
sc->sc_dev.dv_xname) != 0)
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
goto err;
NEXT_ATTACH_STATE(sc, FINISH_ID_STA);
Get rid of the silly char (*dvname)[IFNAMSIZ] pattern. Replace it with const char *dvname. I'm not sure what I was thinking.
2004-12-20 03:28:02 +03:00
rtw_setifprops(&sc->sc_if, sc->sc_dev.dv_xname, (void*)sc);
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
IFQ_SET_READY(&sc->sc_if.if_snd);
rtw_set80211props(&sc->sc_ic);
/*
* Call MI attach routines.
*/
if_attach(&sc->sc_if);
ieee80211_ifattach(&sc->sc_if);
rtw_set80211methods(&sc->sc_mtbl, &sc->sc_ic);
/* possibly we should fill in our own sc_send_prresp, since
* the RTL8180 is probably sending probe responses in ad hoc
* mode.
*/
/* complete initialization */
ieee80211_media_init(&sc->sc_if, rtw_media_change, rtw_media_status);
callout_init(&sc->sc_scan_ch);
Add bpf(4) packet taps to rtw(4). Now you can tap DLT_EN10MB (Ethernet II), DLT_IEEE802_11 (802.11), and DLT_IEEE802_11_RADIO (radiotap + 802.11).
2004-12-29 01:30:07 +03:00
rtw_init_radiotap(sc);
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
#if NBPFILTER > 0
bpfattach2(&sc->sc_if, DLT_IEEE802_11_RADIO,
sizeof(struct ieee80211_frame) + 64, &sc->sc_radiobpf);
#endif
Get rid of the silly char (*dvname)[IFNAMSIZ] pattern. Replace it with const char *dvname. I'm not sure what I was thinking.
2004-12-20 03:28:02 +03:00
rtw_establish_hooks(&sc->sc_hooks, sc->sc_dev.dv_xname, (void*)sc);
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
NEXT_ATTACH_STATE(sc, FINISHED);
return;
err:
rtw_detach(sc);
return;
}
int
rtw_detach(struct rtw_softc *sc)
{
int pri;
In rtw_detach, flag the rtw as "invalid" (RTW_F_INVALID). Do not try to power-down the RF section of an invalid rtw. The radiotap signal quality field is 16, not 8, bits wide. Convert the byte order.
2004-12-29 22:41:04 +03:00
sc->sc_flags |= RTW_F_INVALID;
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
switch (sc->sc_attach_state) {
case FINISHED:
Miscellaneous changes. Details below. Important changes flagged with []. Using the driver with my Linksys WPC11 ver. 4, it seems to be receiving packets for a change. The WPC11 ver. 4 has a Maxim RF section. My no-name rtw with Philips RF section still does not receive any packets. Keep access-level (analog params > config[0123] registers > none) in sc_access. Add rtw_set_access for changing the access level. Make rtw_continuous_tx_enable and other subroutines use rtw_set_access instead of rtw_config0123_enable and rtw_anaparm_enable. Factor part of the chip-reset code into rtw_chip_reset1. Change the 'struct foo (*bar)[N]'-style arguments to 'struct foo *bar'-style arguments. Consolidate software/hardware Tx/Rx ring setup in rtw_hwring_setup, rtw_swring_setup. Add a new constant, SA2400_OPMODE_DEFAULTS, for the bits that we *always* set in the SA2400 OPMODE register. Factor some code out into rtw_sa2400_calibrate. (Inspired by the Linux driver.) [] When the receiver goes into underrun/overflow state, call a new subroutine, rtw_kick() that stops the Rx/Tx processes, resets the chip, reinitializes the Tx/Rx rings, and restarts Rx/Tx processes. (Inspired by the Linux driver.) [] In rtw_intr_rx, check for too-short packets before calling ieee80211_find_rxnode. I believe this will prevent a repeat of the MCHK exception I saw once on macppc. [] Use seconds-elapased as well as microseconds-elapsed to set the next "due date" for the timeout interrupt. This keeps the driver from programming the timeout to expire too early. [] In rtw_intr, read RTW_ISR at most 10 times, then get out. If the interface is not enabled (RTW_F_ENABLED), then get out. [] In rtw_stop, get out if the interface is not enabled (RTW_F_ENABLED). Block IPL_NET interrupts. Don't read/write any registers if the interface is invalid (RTW_F_INVALID). [] Call rtw_stop in rtw_detach.
2004-12-12 09:37:59 +03:00
rtw_stop(&sc->sc_if, 1);
Get rid of the silly char (*dvname)[IFNAMSIZ] pattern. Replace it with const char *dvname. I'm not sure what I was thinking.
2004-12-20 03:28:02 +03:00
rtw_disestablish_hooks(&sc->sc_hooks, sc->sc_dev.dv_xname,
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
(void*)sc);
callout_stop(&sc->sc_scan_ch);
ieee80211_ifdetach(&sc->sc_if);
if_detach(&sc->sc_if);
break;
case FINISH_ID_STA:
case FINISH_RF_ATTACH:
rtw_rf_destroy(sc->sc_rf);
sc->sc_rf = NULL;
/*FALLTHROUGH*/
case FINISH_PARSE_SROM:
case FINISH_READ_SROM:
rtw_srom_free(&sc->sc_srom);
/*FALLTHROUGH*/
case FINISH_RESET:
case FINISH_RXMAPS_CREATE:
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
rtw_rxdesc_dmamaps_destroy(sc->sc_dmat, &sc->sc_rxsoft[0],
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
RTW_RXQLEN);
/*FALLTHROUGH*/
case FINISH_TXMAPS_CREATE:
for (pri = 0; pri < RTW_NTXPRI; pri++) {
rtw_txdesc_dmamaps_destroy(sc->sc_dmat,
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
sc->sc_txsoft_blk[pri].tsb_desc,
sc->sc_txsoft_blk[pri].tsb_ndesc);
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
}
/*FALLTHROUGH*/
case FINISH_TXDESCBLK_SETUP:
case FINISH_TXCTLBLK_SETUP:
I like the tlp/atw-style names for software descriptors, txsoft, better than txctl. Change from rtw_txctl/rtw_rxctl to rtw_txsoft/rtw_rxsoft. Change the descriptor blocks' names to match: rtw_txctl_blk becomes rtw_txsoft_blk. Change the member-name prefixes for both software and hardware descriptors.
2004-12-29 04:06:52 +03:00
rtw_txsoft_blk_cleanup_all(sc);
Add work-in-progress driver rtw(4) for Realtek RTL8180 MAC/baseband. Add register definitions for Maxim MAX2820 and Philips SA2400 radio front-ends.
2004-09-26 06:29:15 +04:00
/*FALLTHROUGH*/
case FINISH_DESCMAP_LOAD:
bus_dmamap_unload(sc->sc_dmat, sc->sc_desc_dmamap);
/*FALLTHROUGH*/
case FINISH_DESCMAP_CREATE:
bus_dmamap_destroy(sc->sc_dmat, sc->sc_desc_dmamap);
/*FALLTHROUGH*/
case FINISH_DESC_MAP:
bus_dmamem_unmap(sc->sc_dmat, (caddr_t)sc->sc_descs,
sizeof(struct rtw_descs));
/*FALLTHROUGH*/
case FINISH_DESC_ALLOC:
bus_dmamem_free(sc->sc_dmat, &sc->sc_desc_segs,
sc->sc_desc_nsegs);
/*FALLTHROUGH*/
case DETACHED:
NEXT_ATTACH_STATE(sc, DETACHED);
break;
}
return 0;
}
int
rtw_activate(struct device *self, enum devact act)
{
struct rtw_softc *sc = (struct rtw_softc *)self;
int rc = 0, s;
s = splnet();
switch (act) {
case DVACT_ACTIVATE:
rc = EOPNOTSUPP;
break;
case DVACT_DEACTIVATE:
if_deactivate(&sc->sc_ic.ic_if);
break;
}
splx(s);
return rc;
}