NetBSD/sys/dev/pci/if_jme.c

2162 lines
65 KiB
C

/* $NetBSD: if_jme.c,v 1.8 2009/03/18 17:06:49 cegger Exp $ */
/*
* Copyright (c) 2008 Manuel Bouyer. All rights reserved.
*
* 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. All advertising materials mentioning features or use of this software
* must display the following acknowledgement:
* This product includes software developed by Manuel Bouyer.
* 4. The name of the author may not be used to endorse or promote products
* derived from this software without specific prior written permission.
*
* THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``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 THE AUTHOR 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.
*/
/*-
* Copyright (c) 2008, Pyun YongHyeon <yongari@FreeBSD.org>
* All rights reserved.
*
* 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 unmodified, 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.
*
* THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``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 THE AUTHOR OR CONTRIBUTORS 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.
*/
/*
* Driver for JMicron Technologies JMC250 (Giganbit) and JMC260 (Fast)
* Ethernet Controllers.
*/
#include <sys/cdefs.h>
__KERNEL_RCSID(0, "$NetBSD: if_jme.c,v 1.8 2009/03/18 17:06:49 cegger Exp $");
#include <sys/param.h>
#include <sys/systm.h>
#include <sys/mbuf.h>
#include <sys/protosw.h>
#include <sys/socket.h>
#include <sys/ioctl.h>
#include <sys/errno.h>
#include <sys/malloc.h>
#include <sys/kernel.h>
#include <sys/proc.h> /* only for declaration of wakeup() used by vm.h */
#include <sys/device.h>
#include <sys/syslog.h>
#include <sys/sysctl.h>
#include <net/if.h>
#if defined(SIOCSIFMEDIA)
#include <net/if_media.h>
#endif
#include <net/if_types.h>
#include <net/if_dl.h>
#include <net/route.h>
#include <net/netisr.h>
#include "bpfilter.h"
#if NBPFILTER > 0
#include <net/bpf.h>
#include <net/bpfdesc.h>
#endif
#include "rnd.h"
#if NRND > 0
#include <sys/rnd.h>
#endif
#ifdef INET
#include <netinet/in.h>
#include <netinet/in_systm.h>
#include <netinet/in_var.h>
#include <netinet/ip.h>
#include <netinet/tcp.h>
#endif
#include <net/if_ether.h>
#include <uvm/uvm_extern.h>
#if defined(INET)
#include <netinet/if_inarp.h>
#endif
#include <sys/bus.h>
#include <sys/intr.h>
#include <dev/pci/pcireg.h>
#include <dev/pci/pcivar.h>
#include <dev/pci/pcidevs.h>
#include <dev/pci/if_jmereg.h>
#include <dev/mii/mii.h>
#include <dev/mii/miivar.h>
struct jme_product_desc {
u_int32_t jme_product;
const char *jme_desc;
};
/* number of entries in transmit and receive rings */
#define JME_NBUFS (PAGE_SIZE / sizeof(struct jme_desc))
#define JME_DESC_INC(x, y) ((x) = ((x) + 1) % (y))
/* Water mark to kick reclaiming Tx buffers. */
#define JME_TX_DESC_HIWAT (JME_NBUFS - (((JME_NBUFS) * 3) / 10))
struct jme_softc {
device_t jme_dev; /* base device */
bus_space_tag_t jme_bt_mac;
bus_space_handle_t jme_bh_mac; /* Mac registers */
bus_space_tag_t jme_bt_phy;
bus_space_handle_t jme_bh_phy; /* PHY registers */
bus_space_tag_t jme_bt_misc;
bus_space_handle_t jme_bh_misc; /* Misc registers */
bus_dma_tag_t jme_dmatag;
bus_dma_segment_t jme_txseg; /* transmit ring seg */
bus_dmamap_t jme_txmap; /* transmit ring DMA map */
struct jme_desc* jme_txring; /* transmit ring */
bus_dmamap_t jme_txmbufm[JME_NBUFS]; /* transmit mbufs DMA map */
struct mbuf *jme_txmbuf[JME_NBUFS]; /* mbufs being transmitted */
int jme_tx_cons; /* transmit ring consumer */
int jme_tx_prod; /* transmit ring producer */
int jme_tx_cnt; /* transmit ring active count */
bus_dma_segment_t jme_rxseg; /* receive ring seg */
bus_dmamap_t jme_rxmap; /* receive ring DMA map */
struct jme_desc* jme_rxring; /* receive ring */
bus_dmamap_t jme_rxmbufm[JME_NBUFS]; /* receive mbufs DMA map */
struct mbuf *jme_rxmbuf[JME_NBUFS]; /* mbufs being received */
int jme_rx_cons; /* receive ring consumer */
int jme_rx_prod; /* receive ring producer */
void* jme_ih; /* our interrupt */
struct ethercom jme_ec;
struct callout jme_tick_ch; /* tick callout */
u_int8_t jme_enaddr[ETHER_ADDR_LEN];/* hardware address */
u_int8_t jme_phyaddr; /* address of integrated phy */
u_int8_t jme_chip_rev; /* chip revision */
u_int8_t jme_rev; /* PCI revision */
mii_data_t jme_mii; /* mii bus */
u_int32_t jme_flags; /* device features, see below */
uint32_t jme_txcsr; /* TX config register */
uint32_t jme_rxcsr; /* RX config register */
#if NRND > 0
rndsource_element_t rnd_source;
#endif
/* interrupt coalition parameters */
struct sysctllog *jme_clog;
int jme_intrxto; /* interrupt RX timeout */
int jme_intrxct; /* interrupt RX packets counter */
int jme_inttxto; /* interrupt TX timeout */
int jme_inttxct; /* interrupt TX packets counter */
};
#define JME_FLAG_FPGA 0x0001 /* FPGA version */
#define JME_FLAG_GIGA 0x0002 /* giga Ethernet capable */
#define jme_if jme_ec.ec_if
#define jme_bpf jme_if.if_bpf
typedef struct jme_softc jme_softc_t;
typedef u_long ioctl_cmd_t;
static int jme_pci_match(device_t, cfdata_t, void *);
static void jme_pci_attach(device_t, device_t, void *);
static void jme_intr_rx(jme_softc_t *);
static int jme_intr(void *);
static int jme_ifioctl(struct ifnet *, ioctl_cmd_t, void *);
static int jme_mediachange(struct ifnet *);
static void jme_ifwatchdog(struct ifnet *);
static void jme_shutdown(void *);
static void jme_txeof(struct jme_softc *);
static void jme_ifstart(struct ifnet *);
static void jme_reset(jme_softc_t *);
static int jme_ifinit(struct ifnet *);
static int jme_init(struct ifnet *, int);
static void jme_stop(struct ifnet *, int);
// static void jme_restart(void *);
static void jme_ticks(void *);
static void jme_mac_config(jme_softc_t *);
static void jme_set_filter(jme_softc_t *);
int jme_mii_read(device_t, int, int);
void jme_mii_write(device_t, int, int, int);
void jme_statchg(device_t);
static int jme_eeprom_read_byte(struct jme_softc *, uint8_t, uint8_t *);
static int jme_eeprom_macaddr(struct jme_softc *);
#define JME_TIMEOUT 1000
#define JME_PHY_TIMEOUT 1000
#define JME_EEPROM_TIMEOUT 1000
static int jme_sysctl_intrxto(SYSCTLFN_PROTO);
static int jme_sysctl_intrxct(SYSCTLFN_PROTO);
static int jme_sysctl_inttxto(SYSCTLFN_PROTO);
static int jme_sysctl_inttxct(SYSCTLFN_PROTO);
static int jme_root_num;
CFATTACH_DECL_NEW(jme, sizeof(jme_softc_t),
jme_pci_match, jme_pci_attach, NULL, NULL);
static const struct jme_product_desc jme_products[] = {
{ PCI_PRODUCT_JMICRON_JMC250,
"JMicron JMC250 Gigabit Ethernet Controller" },
{ PCI_PRODUCT_JMICRON_JMC260,
"JMicron JMC260 Gigabit Ethernet Controller" },
{ 0, NULL },
};
static const struct jme_product_desc *jme_lookup_product(uint32_t);
static const struct jme_product_desc *
jme_lookup_product(uint32_t id)
{
const struct jme_product_desc *jp;
for (jp = jme_products ; jp->jme_desc != NULL; jp++)
if (PCI_PRODUCT(id) == jp->jme_product)
return jp;
return NULL;
}
static int
jme_pci_match(device_t parent, cfdata_t cf, void *aux)
{
struct pci_attach_args *pa = (struct pci_attach_args *)aux;
if (PCI_VENDOR(pa->pa_id) != PCI_VENDOR_JMICRON)
return 0;
if (jme_lookup_product(pa->pa_id) != NULL)
return 1;
return 0;
}
static void
jme_pci_attach(device_t parent, device_t self, void *aux)
{
jme_softc_t *sc = device_private(self);
struct pci_attach_args * const pa = (struct pci_attach_args *)aux;
const struct jme_product_desc *jp;
struct ifnet * const ifp = &sc->jme_if;
bus_space_tag_t iot1, iot2, memt;
bus_space_handle_t ioh1, ioh2, memh;
bus_size_t size, size2;
pci_intr_handle_t intrhandle;
const char *intrstr;
pcireg_t csr;
int nsegs, i;
const struct sysctlnode *node;
int jme_nodenum;
sc->jme_dev = self;
aprint_normal("\n");
callout_init(&sc->jme_tick_ch, 0);
jp = jme_lookup_product(pa->pa_id);
if (jp == NULL)
panic("jme_pci_attach: impossible");
if (jp->jme_product == PCI_PRODUCT_JMICRON_JMC250)
sc->jme_flags = JME_FLAG_GIGA;
/*
* Map the card space. Try Mem first.
*/
if (pci_mapreg_map(pa, JME_PCI_BAR0,
PCI_MAPREG_TYPE_MEM | PCI_MAPREG_MEM_TYPE_32BIT,
0, &memt, &memh, NULL, &size) == 0) {
sc->jme_bt_mac = memt;
sc->jme_bh_mac = memh;
sc->jme_bt_phy = memt;
if (bus_space_subregion(memt, memh, JME_PHY_EEPROM_BASE_MEMOFF,
JME_PHY_EEPROM_SIZE, &sc->jme_bh_phy) != 0) {
aprint_error_dev(self, "can't subregion PHY space\n");
bus_space_unmap(memt, memh, size);
return;
}
sc->jme_bt_misc = memt;
if (bus_space_subregion(memt, memh, JME_MISC_BASE_MEMOFF,
JME_MISC_SIZE, &sc->jme_bh_misc) != 0) {
aprint_error_dev(self, "can't subregion misc space\n");
bus_space_unmap(memt, memh, size);
return;
}
} else {
if (pci_mapreg_map(pa, JME_PCI_BAR1, PCI_MAPREG_TYPE_IO,
0, &iot1, &ioh1, NULL, &size) != 0) {
aprint_error_dev(self, "can't map I/O space 1\n");
return;
}
sc->jme_bt_mac = iot1;
sc->jme_bh_mac = ioh1;
if (pci_mapreg_map(pa, JME_PCI_BAR2, PCI_MAPREG_TYPE_IO,
0, &iot2, &ioh2, NULL, &size2) != 0) {
aprint_error_dev(self, "can't map I/O space 2\n");
bus_space_unmap(iot1, ioh1, size);
return;
}
sc->jme_bt_phy = iot2;
sc->jme_bh_phy = ioh2;
sc->jme_bt_misc = iot2;
if (bus_space_subregion(iot2, ioh2, JME_MISC_BASE_IOOFF,
JME_MISC_SIZE, &sc->jme_bh_misc) != 0) {
aprint_error_dev(self, "can't subregion misc space\n");
bus_space_unmap(iot1, ioh1, size);
bus_space_unmap(iot2, ioh2, size2);
return;
}
}
if (pci_dma64_available(pa))
sc->jme_dmatag = pa->pa_dmat64;
else
sc->jme_dmatag = pa->pa_dmat;
/* Enable the device. */
csr = pci_conf_read(pa->pa_pc, pa->pa_tag, PCI_COMMAND_STATUS_REG);
pci_conf_write(pa->pa_pc, pa->pa_tag, PCI_COMMAND_STATUS_REG,
csr | PCI_COMMAND_MASTER_ENABLE);
aprint_normal_dev(self, "%s\n", jp->jme_desc);
sc->jme_rev = PCI_REVISION(pa->pa_class);
csr = bus_space_read_4(sc->jme_bt_misc, sc->jme_bh_misc, JME_CHIPMODE);
if (((csr & CHIPMODE_FPGA_REV_MASK) >> CHIPMODE_FPGA_REV_SHIFT) !=
CHIPMODE_NOT_FPGA)
sc->jme_flags |= JME_FLAG_FPGA;
sc->jme_chip_rev = (csr & CHIPMODE_REV_MASK) >> CHIPMODE_REV_SHIFT;
aprint_verbose_dev(self, "PCI device revision : 0x%x, Chip revision: "
"0x%x", sc->jme_rev, sc->jme_chip_rev);
if (sc->jme_flags & JME_FLAG_FPGA)
aprint_verbose(" FPGA revision: 0x%x",
(csr & CHIPMODE_FPGA_REV_MASK) >> CHIPMODE_FPGA_REV_SHIFT);
aprint_verbose("\n");
/*
* Save PHY address.
* Integrated JR0211 has fixed PHY address whereas FPGA version
* requires PHY probing to get correct PHY address.
*/
if ((sc->jme_flags & JME_FLAG_FPGA) == 0) {
sc->jme_phyaddr =
bus_space_read_4(sc->jme_bt_misc, sc->jme_bh_misc,
JME_GPREG0) & GPREG0_PHY_ADDR_MASK;
} else
sc->jme_phyaddr = 0;
jme_reset(sc);
/* read mac addr */
if (jme_eeprom_macaddr(sc)) {
aprint_error_dev(self, "error reading Ethernet address\n");
/* return; */
}
aprint_normal_dev(self, "Ethernet address %s\n",
ether_sprintf(sc->jme_enaddr));
/* Map and establish interrupts */
if (pci_intr_map(pa, &intrhandle)) {
aprint_error_dev(self, "couldn't map interrupt\n");
return;
}
intrstr = pci_intr_string(pa->pa_pc, intrhandle);
sc->jme_if.if_softc = sc;
sc->jme_ih = pci_intr_establish(pa->pa_pc, intrhandle, IPL_NET,
jme_intr, sc);
if (sc->jme_ih == NULL) {
aprint_error_dev(self, "couldn't establish interrupt");
if (intrstr != NULL)
aprint_error(" at %s", intrstr);
aprint_error("\n");
return;
}
aprint_normal_dev(self, "interrupting at %s\n", intrstr);
/* allocate and map DMA-safe memory for transmit ring */
if (bus_dmamem_alloc(sc->jme_dmatag, PAGE_SIZE, 0, PAGE_SIZE,
&sc->jme_txseg, 1, &nsegs, BUS_DMA_NOWAIT) != 0 ||
bus_dmamem_map(sc->jme_dmatag, &sc->jme_txseg,
nsegs, PAGE_SIZE, (void **)&sc->jme_txring,
BUS_DMA_NOWAIT | BUS_DMA_COHERENT) != 0 ||
bus_dmamap_create(sc->jme_dmatag, PAGE_SIZE, 1, PAGE_SIZE, 0,
BUS_DMA_NOWAIT | BUS_DMA_ALLOCNOW, &sc->jme_txmap) != 0 ||
bus_dmamap_load(sc->jme_dmatag, sc->jme_txmap, sc->jme_txring,
PAGE_SIZE, NULL, BUS_DMA_NOWAIT) != 0) {
aprint_error_dev(self, "can't allocate DMA memory TX ring\n");
return;
}
/* allocate and map DMA-safe memory for receive ring */
if (bus_dmamem_alloc(sc->jme_dmatag, PAGE_SIZE, 0, PAGE_SIZE,
&sc->jme_rxseg, 1, &nsegs, BUS_DMA_NOWAIT) != 0 ||
bus_dmamem_map(sc->jme_dmatag, &sc->jme_rxseg,
nsegs, PAGE_SIZE, (void **)&sc->jme_rxring,
BUS_DMA_NOWAIT | BUS_DMA_COHERENT) != 0 ||
bus_dmamap_create(sc->jme_dmatag, PAGE_SIZE, 1, PAGE_SIZE, 0,
BUS_DMA_NOWAIT | BUS_DMA_ALLOCNOW, &sc->jme_rxmap) != 0 ||
bus_dmamap_load(sc->jme_dmatag, sc->jme_rxmap, sc->jme_rxring,
PAGE_SIZE, NULL, BUS_DMA_NOWAIT) != 0) {
aprint_error_dev(self, "can't allocate DMA memory RX ring\n");
return;
}
for (i = 0; i < JME_NBUFS; i++) {
sc->jme_txmbuf[i] = sc->jme_rxmbuf[i] = NULL;
if (bus_dmamap_create(sc->jme_dmatag, JME_MAX_TX_LEN,
JME_NBUFS, JME_MAX_TX_LEN, 0, BUS_DMA_NOWAIT,
&sc->jme_txmbufm[i]) != 0) {
aprint_error_dev(self, "can't allocate DMA TX map\n");
return;
}
if (bus_dmamap_create(sc->jme_dmatag, JME_MAX_RX_LEN,
1, JME_MAX_RX_LEN, 0, BUS_DMA_NOWAIT,
&sc->jme_rxmbufm[i]) != 0) {
aprint_error_dev(self, "can't allocate DMA RX map\n");
return;
}
}
/*
* Add shutdown hook so that DMA is disabled prior to reboot.
*/
(void)shutdownhook_establish(jme_shutdown, ifp);
/*
* Initialize our media structures and probe the MII.
*
* Note that we don't care about the media instance. We
* are expecting to have multiple PHYs on the 10/100 cards,
* and on those cards we exclude the internal PHY from providing
* 10baseT. By ignoring the instance, it allows us to not have
* to specify it on the command line when switching media.
*/
sc->jme_mii.mii_ifp = ifp;
sc->jme_mii.mii_readreg = jme_mii_read;
sc->jme_mii.mii_writereg = jme_mii_write;
sc->jme_mii.mii_statchg = jme_statchg;
sc->jme_ec.ec_mii = &sc->jme_mii;
ifmedia_init(&sc->jme_mii.mii_media, IFM_IMASK, jme_mediachange,
ether_mediastatus);
mii_attach(self, &sc->jme_mii, 0xffffffff, MII_PHY_ANY,
MII_OFFSET_ANY, 0);
if (LIST_FIRST(&sc->jme_mii.mii_phys) == NULL) {
ifmedia_add(&sc->jme_mii.mii_media, IFM_ETHER|IFM_NONE, 0, NULL);
ifmedia_set(&sc->jme_mii.mii_media, IFM_ETHER|IFM_NONE);
} else
ifmedia_set(&sc->jme_mii.mii_media, IFM_ETHER|IFM_AUTO);
/*
* We can support 802.1Q VLAN-sized frames.
*/
sc->jme_ec.ec_capabilities |=
ETHERCAP_VLAN_MTU | ETHERCAP_VLAN_HWTAGGING;
if (sc->jme_flags & JME_FLAG_GIGA)
sc->jme_ec.ec_capabilities |= ETHERCAP_JUMBO_MTU;
strlcpy(ifp->if_xname, device_xname(self), IFNAMSIZ);
ifp->if_flags = IFF_BROADCAST|IFF_SIMPLEX|IFF_NOTRAILERS|IFF_MULTICAST;
ifp->if_ioctl = jme_ifioctl;
ifp->if_start = jme_ifstart;
ifp->if_watchdog = jme_ifwatchdog;
ifp->if_init = jme_ifinit;
ifp->if_stop = jme_stop;
ifp->if_timer = 0;
ifp->if_capabilities |=
IFCAP_CSUM_IPv4_Tx | IFCAP_CSUM_IPv4_Rx |
IFCAP_CSUM_TCPv4_Tx | IFCAP_CSUM_TCPv4_Rx |
IFCAP_CSUM_UDPv4_Tx | IFCAP_CSUM_UDPv4_Rx |
IFCAP_CSUM_TCPv6_Tx | /* IFCAP_CSUM_TCPv6_Rx | hardware bug */
IFCAP_CSUM_UDPv6_Tx | /* IFCAP_CSUM_UDPv6_Rx | hardware bug */
IFCAP_TSOv4 | IFCAP_TSOv6;
IFQ_SET_READY(&ifp->if_snd);
if_attach(ifp);
ether_ifattach(&(sc)->jme_if, (sc)->jme_enaddr);
#if NRND > 0
rnd_attach_source(&sc->rnd_source, device_xname(self),
RND_TYPE_NET, 0);
#endif
sc->jme_intrxto = PCCRX_COAL_TO_DEFAULT;
sc->jme_intrxct = PCCRX_COAL_PKT_DEFAULT;
sc->jme_inttxto = PCCTX_COAL_TO_DEFAULT;
sc->jme_inttxct = PCCTX_COAL_PKT_DEFAULT;
if (sysctl_createv(&sc->jme_clog, 0, NULL, &node,
0, CTLTYPE_NODE, device_xname(sc->jme_dev),
SYSCTL_DESCR("jme per-controller controls"),
NULL, 0, NULL, 0, CTL_HW, jme_root_num, CTL_CREATE,
CTL_EOL) != 0) {
aprint_normal_dev(sc->jme_dev, "couldn't create sysctl node\n");
return;
}
jme_nodenum = node->sysctl_num;
/* interrupt moderation sysctls */
if (sysctl_createv(&sc->jme_clog, 0, NULL, &node,
CTLFLAG_READWRITE,
CTLTYPE_INT, "int_rxto",
SYSCTL_DESCR("jme RX interrupt moderation timer"),
jme_sysctl_intrxto, 0, sc,
0, CTL_HW, jme_root_num, jme_nodenum, CTL_CREATE,
CTL_EOL) != 0) {
aprint_normal_dev(sc->jme_dev,
"couldn't create int_rxto sysctl node\n");
}
if (sysctl_createv(&sc->jme_clog, 0, NULL, &node,
CTLFLAG_READWRITE,
CTLTYPE_INT, "int_rxct",
SYSCTL_DESCR("jme RX interrupt moderation packet counter"),
jme_sysctl_intrxct, 0, sc,
0, CTL_HW, jme_root_num, jme_nodenum, CTL_CREATE,
CTL_EOL) != 0) {
aprint_normal_dev(sc->jme_dev,
"couldn't create int_rxct sysctl node\n");
}
if (sysctl_createv(&sc->jme_clog, 0, NULL, &node,
CTLFLAG_READWRITE,
CTLTYPE_INT, "int_txto",
SYSCTL_DESCR("jme TX interrupt moderation timer"),
jme_sysctl_inttxto, 0, sc,
0, CTL_HW, jme_root_num, jme_nodenum, CTL_CREATE,
CTL_EOL) != 0) {
aprint_normal_dev(sc->jme_dev,
"couldn't create int_txto sysctl node\n");
}
if (sysctl_createv(&sc->jme_clog, 0, NULL, &node,
CTLFLAG_READWRITE,
CTLTYPE_INT, "int_txct",
SYSCTL_DESCR("jme TX interrupt moderation packet counter"),
jme_sysctl_inttxct, 0, sc,
0, CTL_HW, jme_root_num, jme_nodenum, CTL_CREATE,
CTL_EOL) != 0) {
aprint_normal_dev(sc->jme_dev,
"couldn't create int_txct sysctl node\n");
}
}
static void
jme_stop_rx(jme_softc_t *sc)
{
uint32_t reg;
int i;
reg = bus_space_read_4(sc->jme_bt_mac, sc->jme_bh_mac, JME_RXCSR);
if ((reg & RXCSR_RX_ENB) == 0)
return;
reg &= ~RXCSR_RX_ENB;
bus_space_write_4(sc->jme_bt_mac, sc->jme_bh_mac, JME_RXCSR, reg);
for (i = JME_TIMEOUT / 10; i > 0; i--) {
DELAY(10);
if ((bus_space_read_4(sc->jme_bt_mac, sc->jme_bh_mac,
JME_RXCSR) & RXCSR_RX_ENB) == 0)
break;
}
if (i == 0)
aprint_error_dev(sc->jme_dev, "stopping recevier timeout!\n");
}
static void
jme_stop_tx(jme_softc_t *sc)
{
uint32_t reg;
int i;
reg = bus_space_read_4(sc->jme_bt_mac, sc->jme_bh_mac, JME_TXCSR);
if ((reg & TXCSR_TX_ENB) == 0)
return;
reg &= ~TXCSR_TX_ENB;
bus_space_write_4(sc->jme_bt_mac, sc->jme_bh_mac, JME_TXCSR, reg);
for (i = JME_TIMEOUT / 10; i > 0; i--) {
DELAY(10);
if ((bus_space_read_4(sc->jme_bt_mac, sc->jme_bh_mac,
JME_TXCSR) & TXCSR_TX_ENB) == 0)
break;
}
if (i == 0)
aprint_error_dev(sc->jme_dev,
"stopping transmitter timeout!\n");
}
static void
jme_reset(jme_softc_t *sc)
{
bus_space_write_4(sc->jme_bt_mac, sc->jme_bh_mac, JME_GHC, GHC_RESET);
DELAY(10);
bus_space_write_4(sc->jme_bt_mac, sc->jme_bh_mac, JME_GHC, 0);
}
static void
jme_shutdown(void *v)
{
jme_stop(v, 1);
}
static void
jme_stop(struct ifnet *ifp, int disable)
{
jme_softc_t *sc = ifp->if_softc;
int i;
/* Stop receiver, transmitter. */
jme_stop_rx(sc);
jme_stop_tx(sc);
/* free receive mbufs */
for (i = 0; i < JME_NBUFS; i++) {
if (sc->jme_rxmbuf[i]) {
bus_dmamap_unload(sc->jme_dmatag, sc->jme_rxmbufm[i]);
m_freem(sc->jme_rxmbuf[i]);
}
sc->jme_rxmbuf[i] = NULL;
}
/* process completed transmits */
jme_txeof(sc);
/* free abort pending transmits */
for (i = 0; i < JME_NBUFS; i++) {
if (sc->jme_txmbuf[i]) {
bus_dmamap_unload(sc->jme_dmatag, sc->jme_txmbufm[i]);
m_freem(sc->jme_txmbuf[i]);
sc->jme_txmbuf[i] = NULL;
}
}
ifp->if_flags &= ~(IFF_RUNNING | IFF_OACTIVE);
ifp->if_timer = 0;
}
#if 0
static void
jme_restart(void *v)
{
jme_init(v);
}
#endif
static int
jme_add_rxbuf(jme_softc_t *sc, struct mbuf *m)
{
int error;
bus_dmamap_t map;
int i = sc->jme_rx_prod;
if (sc->jme_rxmbuf[i] != NULL) {
aprint_error_dev(sc->jme_dev,
"mbuf already here: rxprod %d rxcons %d\n",
sc->jme_rx_prod, sc->jme_rx_cons);
if (m)
m_freem(m);
return EINVAL;
}
if (m == NULL) {
sc->jme_rxmbuf[i] = NULL;
MGETHDR(m, M_DONTWAIT, MT_DATA);
if (m == NULL)
return (ENOBUFS);
MCLGET(m, M_DONTWAIT);
if ((m->m_flags & M_EXT) == 0) {
m_freem(m);
return (ENOBUFS);
}
}
map = sc->jme_rxmbufm[i];
m->m_len = m->m_pkthdr.len = m->m_ext.ext_size;
error = bus_dmamap_load_mbuf(sc->jme_dmatag, map, m,
BUS_DMA_READ|BUS_DMA_NOWAIT);
if (error) {
sc->jme_rxmbuf[i] = NULL;
aprint_error_dev(sc->jme_dev,
"unable to load rx DMA map %d, error = %d\n",
i, error);
m_freem(m);
return (error);
}
bus_dmamap_sync(sc->jme_dmatag, map, 0, map->dm_mapsize,
BUS_DMASYNC_PREREAD);
sc->jme_rxmbuf[i] = m;
sc->jme_rxring[i].buflen = htole32(map->dm_segs[0].ds_len);
sc->jme_rxring[i].addr_lo =
htole32(JME_ADDR_LO(map->dm_segs[0].ds_addr));
sc->jme_rxring[i].addr_hi =
htole32(JME_ADDR_HI(map->dm_segs[0].ds_addr));
sc->jme_rxring[i].flags =
htole32(JME_RD_OWN | JME_RD_INTR | JME_RD_64BIT);
bus_dmamap_sync(sc->jme_dmatag, sc->jme_rxmap,
i * sizeof(struct jme_desc), sizeof(struct jme_desc),
BUS_DMASYNC_PREREAD | BUS_DMASYNC_PREWRITE);
JME_DESC_INC(sc->jme_rx_prod, JME_NBUFS);
return (0);
}
static int
jme_ifinit(struct ifnet *ifp)
{
return jme_init(ifp, 1);
}
static int
jme_init(struct ifnet *ifp, int do_ifinit)
{
jme_softc_t *sc = ifp->if_softc;
int i, s;
uint8_t eaddr[ETHER_ADDR_LEN];
uint32_t reg;
s = splnet();
/* cancel any pending IO */
jme_stop(ifp, 1);
jme_reset(sc);
if ((sc->jme_if.if_flags & IFF_UP) == 0) {
splx(s);
return 0;
}
/* allocate receive ring */
sc->jme_rx_prod = 0;
for (i = 0; i < JME_NBUFS; i++) {
if (jme_add_rxbuf(sc, NULL) < 0) {
aprint_error_dev(sc->jme_dev,
"can't allocate rx mbuf\n");
for (i--; i >= 0; i--) {
bus_dmamap_unload(sc->jme_dmatag,
sc->jme_rxmbufm[i]);
m_freem(sc->jme_rxmbuf[i]);
sc->jme_rxmbuf[i] = NULL;
}
splx(s);
return ENOMEM;
}
}
/* init TX ring */
memset(sc->jme_txring, 0, JME_NBUFS * sizeof(struct jme_desc));
bus_dmamap_sync(sc->jme_dmatag, sc->jme_txmap,
0, JME_NBUFS * sizeof(struct jme_desc),
BUS_DMASYNC_PREREAD | BUS_DMASYNC_PREWRITE);
for (i = 0; i < JME_NBUFS; i++)
sc->jme_txmbuf[i] = NULL;
sc->jme_tx_cons = sc->jme_tx_prod = sc->jme_tx_cnt = 0;
/* Reprogram the station address. */
memcpy( eaddr, CLLADDR(ifp->if_sadl), ETHER_ADDR_LEN);
bus_space_write_4(sc->jme_bt_mac, sc->jme_bh_mac, JME_PAR0,
eaddr[3] << 24 | eaddr[2] << 16 | eaddr[1] << 8 | eaddr[0]);
bus_space_write_4(sc->jme_bt_mac, sc->jme_bh_mac,
JME_PAR1, eaddr[5] << 8 | eaddr[4]);
/*
* Configure Tx queue.
* Tx priority queue weight value : 0
* Tx FIFO threshold for processing next packet : 16QW
* Maximum Tx DMA length : 512
* Allow Tx DMA burst.
*/
sc->jme_txcsr = TXCSR_TXQ_N_SEL(TXCSR_TXQ0);
sc->jme_txcsr |= TXCSR_TXQ_WEIGHT(TXCSR_TXQ_WEIGHT_MIN);
sc->jme_txcsr |= TXCSR_FIFO_THRESH_16QW;
sc->jme_txcsr |= TXCSR_DMA_SIZE_512;
sc->jme_txcsr |= TXCSR_DMA_BURST;
bus_space_write_4(sc->jme_bt_mac, sc->jme_bh_mac,
JME_TXCSR, sc->jme_txcsr);
/* Set Tx descriptor counter. */
bus_space_write_4(sc->jme_bt_mac, sc->jme_bh_mac,
JME_TXQDC, JME_NBUFS);
/* Set Tx ring address to the hardware. */
bus_space_write_4(sc->jme_bt_mac, sc->jme_bh_mac, JME_TXDBA_HI,
JME_ADDR_HI(sc->jme_txmap->dm_segs[0].ds_addr));
bus_space_write_4(sc->jme_bt_mac, sc->jme_bh_mac, JME_TXDBA_LO,
JME_ADDR_LO(sc->jme_txmap->dm_segs[0].ds_addr));
/* Configure TxMAC parameters. */
bus_space_write_4(sc->jme_bt_mac, sc->jme_bh_mac, JME_TXMAC,
TXMAC_IFG1_DEFAULT | TXMAC_IFG2_DEFAULT | TXMAC_IFG_ENB |
TXMAC_THRESH_1_PKT | TXMAC_CRC_ENB | TXMAC_PAD_ENB);
/*
* Configure Rx queue.
* FIFO full threshold for transmitting Tx pause packet : 128T
* FIFO threshold for processing next packet : 128QW
* Rx queue 0 select
* Max Rx DMA length : 128
* Rx descriptor retry : 32
* Rx descriptor retry time gap : 256ns
* Don't receive runt/bad frame.
*/
sc->jme_rxcsr = RXCSR_FIFO_FTHRESH_128T;
/*
* Since Rx FIFO size is 4K bytes, receiving frames larger
* than 4K bytes will suffer from Rx FIFO overruns. So
* decrease FIFO threshold to reduce the FIFO overruns for
* frames larger than 4000 bytes.
* For best performance of standard MTU sized frames use
* maximum allowable FIFO threshold, 128QW.
*/
if ((ifp->if_mtu + ETHER_HDR_LEN + ETHER_VLAN_ENCAP_LEN +
ETHER_CRC_LEN) > JME_RX_FIFO_SIZE)
sc->jme_rxcsr |= RXCSR_FIFO_THRESH_16QW;
else
sc->jme_rxcsr |= RXCSR_FIFO_THRESH_128QW;
sc->jme_rxcsr |= RXCSR_DMA_SIZE_128 | RXCSR_RXQ_N_SEL(RXCSR_RXQ0);
sc->jme_rxcsr |= RXCSR_DESC_RT_CNT(RXCSR_DESC_RT_CNT_DEFAULT);
sc->jme_rxcsr |= RXCSR_DESC_RT_GAP_256 & RXCSR_DESC_RT_GAP_MASK;
bus_space_write_4(sc->jme_bt_mac, sc->jme_bh_mac,
JME_RXCSR, sc->jme_rxcsr);
/* Set Rx descriptor counter. */
bus_space_write_4(sc->jme_bt_mac, sc->jme_bh_mac,
JME_RXQDC, JME_NBUFS);
/* Set Rx ring address to the hardware. */
bus_space_write_4(sc->jme_bt_mac, sc->jme_bh_mac, JME_RXDBA_HI,
JME_ADDR_HI(sc->jme_rxmap->dm_segs[0].ds_addr));
bus_space_write_4(sc->jme_bt_mac, sc->jme_bh_mac, JME_RXDBA_LO,
JME_ADDR_LO(sc->jme_rxmap->dm_segs[0].ds_addr));
/* Clear receive filter. */
bus_space_write_4(sc->jme_bt_mac, sc->jme_bh_mac, JME_RXMAC, 0);
/* Set up the receive filter. */
jme_set_filter(sc);
/*
* Disable all WOL bits as WOL can interfere normal Rx
* operation. Also clear WOL detection status bits.
*/
reg = bus_space_read_4(sc->jme_bt_mac, sc->jme_bh_mac, JME_PMCS);
reg &= ~PMCS_WOL_ENB_MASK;
bus_space_write_4(sc->jme_bt_mac, sc->jme_bh_mac, JME_PMCS, reg);
reg = bus_space_read_4(sc->jme_bt_mac, sc->jme_bh_mac, JME_RXMAC);
/*
* Pad 10bytes right before received frame. This will greatly
* help Rx performance on strict-alignment architectures as
* it does not need to copy the frame to align the payload.
*/
reg |= RXMAC_PAD_10BYTES;
if ((ifp->if_capenable &
(IFCAP_CSUM_IPv4_Rx|IFCAP_CSUM_TCPv4_Rx|IFCAP_CSUM_UDPv4_Rx|
IFCAP_CSUM_TCPv6_Rx|IFCAP_CSUM_UDPv6_Rx)) != 0)
reg |= RXMAC_CSUM_ENB;
reg |= RXMAC_VLAN_ENB; /* enable hardware vlan */
bus_space_write_4(sc->jme_bt_mac, sc->jme_bh_mac, JME_RXMAC, reg);
/* Configure general purpose reg0 */
reg = bus_space_read_4(sc->jme_bt_misc, sc->jme_bh_misc, JME_GPREG0);
reg &= ~GPREG0_PCC_UNIT_MASK;
/* Set PCC timer resolution to micro-seconds unit. */
reg |= GPREG0_PCC_UNIT_US;
/*
* Disable all shadow register posting as we have to read
* JME_INTR_STATUS register in jme_int_task. Also it seems
* that it's hard to synchronize interrupt status between
* hardware and software with shadow posting due to
* requirements of bus_dmamap_sync(9).
*/
reg |= GPREG0_SH_POST_DW7_DIS | GPREG0_SH_POST_DW6_DIS |
GPREG0_SH_POST_DW5_DIS | GPREG0_SH_POST_DW4_DIS |
GPREG0_SH_POST_DW3_DIS | GPREG0_SH_POST_DW2_DIS |
GPREG0_SH_POST_DW1_DIS | GPREG0_SH_POST_DW0_DIS;
/* Disable posting of DW0. */
reg &= ~GPREG0_POST_DW0_ENB;
/* Clear PME message. */
reg &= ~GPREG0_PME_ENB;
/* Set PHY address. */
reg &= ~GPREG0_PHY_ADDR_MASK;
reg |= sc->jme_phyaddr;
bus_space_write_4(sc->jme_bt_misc, sc->jme_bh_misc, JME_GPREG0, reg);
/* Configure Tx queue 0 packet completion coalescing. */
reg = (sc->jme_inttxto << PCCTX_COAL_TO_SHIFT) & PCCTX_COAL_TO_MASK;
reg |= (sc->jme_inttxct << PCCTX_COAL_PKT_SHIFT) & PCCTX_COAL_PKT_MASK;
reg |= PCCTX_COAL_TXQ0;
bus_space_write_4(sc->jme_bt_misc, sc->jme_bh_misc, JME_PCCTX, reg);
/* Configure Rx queue 0 packet completion coalescing. */
reg = (sc->jme_intrxto << PCCRX_COAL_TO_SHIFT) & PCCRX_COAL_TO_MASK;
reg |= (sc->jme_intrxct << PCCRX_COAL_PKT_SHIFT) & PCCRX_COAL_PKT_MASK;
bus_space_write_4(sc->jme_bt_misc, sc->jme_bh_misc, JME_PCCRX0, reg);
/* Disable Timers */
bus_space_write_4(sc->jme_bt_misc, sc->jme_bh_misc, JME_TMCSR, 0);
bus_space_write_4(sc->jme_bt_misc, sc->jme_bh_misc, JME_TIMER1, 0);
bus_space_write_4(sc->jme_bt_misc, sc->jme_bh_misc, JME_TIMER2, 0);
/* Configure retry transmit period, retry limit value. */
bus_space_write_4(sc->jme_bt_mac, sc->jme_bh_mac, JME_TXTRHD,
((TXTRHD_RT_PERIOD_DEFAULT << TXTRHD_RT_PERIOD_SHIFT) &
TXTRHD_RT_PERIOD_MASK) |
((TXTRHD_RT_LIMIT_DEFAULT << TXTRHD_RT_LIMIT_SHIFT) &
TXTRHD_RT_LIMIT_SHIFT));
/* Disable RSS. */
bus_space_write_4(sc->jme_bt_misc, sc->jme_bh_misc,
JME_RSSC, RSSC_DIS_RSS);
/* Initialize the interrupt mask. */
bus_space_write_4(sc->jme_bt_misc, sc->jme_bh_misc,
JME_INTR_MASK_SET, JME_INTRS_ENABLE);
bus_space_write_4(sc->jme_bt_misc, sc->jme_bh_misc,
JME_INTR_STATUS, 0xFFFFFFFF);
/* set media, if not already handling a media change */
if (do_ifinit) {
int error;
if ((error = mii_mediachg(&sc->jme_mii)) == ENXIO)
error = 0;
else if (error != 0) {
aprint_error_dev(sc->jme_dev, "could not set media\n");
return error;
}
}
/* Program MAC with resolved speed/duplex/flow-control. */
jme_mac_config(sc);
/* Start receiver/transmitter. */
sc->jme_rx_cons = 0;
bus_space_write_4(sc->jme_bt_mac, sc->jme_bh_mac, JME_RXCSR,
sc->jme_rxcsr | RXCSR_RX_ENB | RXCSR_RXQ_START);
bus_space_write_4(sc->jme_bt_mac, sc->jme_bh_mac, JME_TXCSR,
sc->jme_txcsr | TXCSR_TX_ENB);
/* start ticks calls */
callout_reset(&sc->jme_tick_ch, hz, jme_ticks, sc);
sc->jme_if.if_flags |= IFF_RUNNING;
sc->jme_if.if_flags &= ~IFF_OACTIVE;
splx(s);
return 0;
}
int
jme_mii_read(device_t self, int phy, int reg)
{
struct jme_softc *sc = device_private(self);
int val, i;
/* For FPGA version, PHY address 0 should be ignored. */
if ((sc->jme_flags & JME_FLAG_FPGA) != 0) {
if (phy == 0)
return (0);
} else {
if (sc->jme_phyaddr != phy)
return (0);
}
bus_space_write_4(sc->jme_bt_mac, sc->jme_bh_mac, JME_SMI,
SMI_OP_READ | SMI_OP_EXECUTE |
SMI_PHY_ADDR(phy) | SMI_REG_ADDR(reg));
for (i = JME_PHY_TIMEOUT / 10; i > 0; i--) {
delay(10);
if (((val = bus_space_read_4(sc->jme_bt_mac, sc->jme_bh_mac,
JME_SMI)) & SMI_OP_EXECUTE) == 0)
break;
}
if (i == 0) {
aprint_error_dev(sc->jme_dev, "phy read timeout : %d\n", reg);
return (0);
}
return ((val & SMI_DATA_MASK) >> SMI_DATA_SHIFT);
}
void
jme_mii_write(device_t self, int phy, int reg, int val)
{
struct jme_softc *sc = device_private(self);
int i;
/* For FPGA version, PHY address 0 should be ignored. */
if ((sc->jme_flags & JME_FLAG_FPGA) != 0) {
if (phy == 0)
return;
} else {
if (sc->jme_phyaddr != phy)
return;
}
bus_space_write_4(sc->jme_bt_mac, sc->jme_bh_mac, JME_SMI,
SMI_OP_WRITE | SMI_OP_EXECUTE |
((val << SMI_DATA_SHIFT) & SMI_DATA_MASK) |
SMI_PHY_ADDR(phy) | SMI_REG_ADDR(reg));
for (i = JME_PHY_TIMEOUT / 10; i > 0; i--) {
delay(10);
if (((val = bus_space_read_4(sc->jme_bt_mac, sc->jme_bh_mac,
JME_SMI)) & SMI_OP_EXECUTE) == 0)
break;
}
if (i == 0)
aprint_error_dev(sc->jme_dev, "phy write timeout : %d\n", reg);
return;
}
void
jme_statchg(device_t self)
{
jme_softc_t *sc = device_private(self);
struct ifnet *ifp = &sc->jme_if;
if ((ifp->if_flags & (IFF_UP|IFF_RUNNING)) == (IFF_UP|IFF_RUNNING))
jme_init(ifp, 0);
}
static void
jme_intr_rx(jme_softc_t *sc) {
struct mbuf *m, *mhead;
struct ifnet *ifp = &sc->jme_if;
uint32_t flags, buflen;
int i, ipackets, nsegs, seg, error;
struct jme_desc *desc;
bus_dmamap_sync(sc->jme_dmatag, sc->jme_rxmap, 0,
sizeof(struct jme_desc) * JME_NBUFS,
BUS_DMASYNC_POSTREAD | BUS_DMASYNC_POSTWRITE);
#ifdef JMEDEBUG_RX
printf("rxintr sc->jme_rx_cons %d flags 0x%x\n",
sc->jme_rx_cons, le32toh(sc->jme_rxring[sc->jme_rx_cons].flags));
#endif
ipackets = 0;
while((le32toh(sc->jme_rxring[ sc->jme_rx_cons].flags) & JME_RD_OWN)
== 0) {
i = sc->jme_rx_cons;
desc = &sc->jme_rxring[i];
#ifdef JMEDEBUG_RX
printf("rxintr i %d flags 0x%x buflen 0x%x\n",
i, le32toh(desc->flags), le32toh(desc->buflen));
#endif
if ((le32toh(desc->buflen) & JME_RD_VALID) == 0)
break;
bus_dmamap_sync(sc->jme_dmatag, sc->jme_rxmbufm[i], 0,
sc->jme_rxmbufm[i]->dm_mapsize, BUS_DMASYNC_POSTREAD);
bus_dmamap_unload(sc->jme_dmatag, sc->jme_rxmbufm[i]);
buflen = le32toh(desc->buflen);
nsegs = JME_RX_NSEGS(buflen);
flags = le32toh(desc->flags);
if ((buflen & JME_RX_ERR_STAT) != 0 ||
JME_RX_BYTES(buflen) < sizeof(struct ether_header) ||
JME_RX_BYTES(buflen) >
(ifp->if_mtu + ETHER_HDR_LEN + JME_RX_PAD_BYTES)) {
#ifdef JMEDEBUG_RX
printf("rx error flags 0x%x buflen 0x%x\n",
flags, buflen);
#endif
ifp->if_ierrors++;
/* reuse the mbufs */
for (seg = 0; seg < nsegs; seg++) {
m = sc->jme_rxmbuf[i];
sc->jme_rxmbuf[i] = NULL;
if ((error = jme_add_rxbuf(sc, m)) != 0)
aprint_error_dev(sc->jme_dev,
"can't reuse mbuf: %d\n", error);
JME_DESC_INC(sc->jme_rx_cons, JME_NBUFS);
i = sc->jme_rx_cons;
}
continue;
}
/* receive this packet */
mhead = m = sc->jme_rxmbuf[i];
sc->jme_rxmbuf[i] = NULL;
/* add a new buffer to chain */
if (jme_add_rxbuf(sc, NULL) == ENOBUFS) {
for (seg = 0; seg < nsegs; seg++) {
m = sc->jme_rxmbuf[i];
sc->jme_rxmbuf[i] = NULL;
if ((error = jme_add_rxbuf(sc, m)) != 0)
aprint_error_dev(sc->jme_dev,
"can't reuse mbuf: %d\n", error);
JME_DESC_INC(sc->jme_rx_cons, JME_NBUFS);
i = sc->jme_rx_cons;
}
ifp->if_ierrors++;
continue;
}
/* build mbuf chain: head, then remaining segments */
m->m_pkthdr.rcvif = ifp;
m->m_pkthdr.len = JME_RX_BYTES(buflen) - JME_RX_PAD_BYTES;
m->m_len = (nsegs > 1) ? (MCLBYTES - JME_RX_PAD_BYTES) :
m->m_pkthdr.len;
m->m_data = m->m_ext.ext_buf + JME_RX_PAD_BYTES;
JME_DESC_INC(sc->jme_rx_cons, JME_NBUFS);
for (seg = 1; seg < nsegs; seg++) {
i = sc->jme_rx_cons;
m = sc->jme_rxmbuf[i];
sc->jme_rxmbuf[i] = NULL;
(void)jme_add_rxbuf(sc, NULL);
m->m_flags &= ~M_PKTHDR;
m_cat(mhead, m);
JME_DESC_INC(sc->jme_rx_cons, JME_NBUFS);
}
/* and adjust last mbuf's size */
if (nsegs > 1) {
m->m_len =
JME_RX_BYTES(buflen) - (MCLBYTES * (nsegs - 1));
}
ifp->if_ipackets++;
ipackets++;
#if NBPFILTER > 0
if (ifp->if_bpf)
bpf_mtap(ifp->if_bpf, mhead);
#endif /* NBPFILTER > 0 */
if ((ifp->if_capenable & IFCAP_CSUM_IPv4_Rx) &&
(flags & JME_RD_IPV4)) {
mhead->m_pkthdr.csum_flags |= M_CSUM_IPv4;
if (!(flags & JME_RD_IPCSUM))
mhead->m_pkthdr.csum_flags |= M_CSUM_IPv4_BAD;
}
if ((ifp->if_capenable & IFCAP_CSUM_TCPv4_Rx) &&
(flags & JME_RD_TCPV4) == JME_RD_TCPV4) {
mhead->m_pkthdr.csum_flags |= M_CSUM_TCPv4;
if (!(flags & JME_RD_TCPCSUM))
mhead->m_pkthdr.csum_flags |=
M_CSUM_TCP_UDP_BAD;
}
if ((ifp->if_capenable & IFCAP_CSUM_UDPv4_Rx) &&
(flags & JME_RD_UDPV4) == JME_RD_UDPV4) {
mhead->m_pkthdr.csum_flags |= M_CSUM_UDPv4;
if (!(flags & JME_RD_UDPCSUM))
mhead->m_pkthdr.csum_flags |=
M_CSUM_TCP_UDP_BAD;
}
if ((ifp->if_capenable & IFCAP_CSUM_TCPv6_Rx) &&
(flags & JME_RD_TCPV6) == JME_RD_TCPV6) {
mhead->m_pkthdr.csum_flags |= M_CSUM_TCPv6;
if (!(flags & JME_RD_TCPCSUM))
mhead->m_pkthdr.csum_flags |=
M_CSUM_TCP_UDP_BAD;
}
if ((ifp->if_capenable & IFCAP_CSUM_UDPv6_Rx) &&
(flags & JME_RD_UDPV6) == JME_RD_UDPV6) {
m->m_pkthdr.csum_flags |= M_CSUM_UDPv6;
if (!(flags & JME_RD_UDPCSUM))
mhead->m_pkthdr.csum_flags |=
M_CSUM_TCP_UDP_BAD;
}
if (flags & JME_RD_VLAN_TAG) {
/* pass to vlan_input() */
VLAN_INPUT_TAG(ifp, mhead,
(flags & JME_RD_VLAN_MASK), continue);
}
(*ifp->if_input)(ifp, mhead);
}
#if NRND > 0
if (ipackets && RND_ENABLED(&sc->rnd_source))
rnd_add_uint32(&sc->rnd_source, ipackets);
#endif /* NRND > 0 */
}
static int
jme_intr(void *v)
{
jme_softc_t *sc = v;
uint32_t istatus;
istatus = bus_space_read_4(sc->jme_bt_misc, sc->jme_bh_misc,
JME_INTR_STATUS);
if (istatus == 0 || istatus == 0xFFFFFFFF)
return 0;
/* Disable interrupts. */
bus_space_write_4(sc->jme_bt_misc, sc->jme_bh_misc,
JME_INTR_MASK_CLR, 0xFFFFFFFF);
again:
/* and update istatus */
istatus = bus_space_read_4(sc->jme_bt_misc, sc->jme_bh_misc,
JME_INTR_STATUS);
if ((istatus & JME_INTRS_CHECK) == 0)
goto done;
/* Reset PCC counter/timer and Ack interrupts. */
if ((istatus & (INTR_TXQ_COMP | INTR_TXQ_COAL | INTR_TXQ_COAL_TO)) != 0)
istatus |= INTR_TXQ_COAL | INTR_TXQ_COAL_TO | INTR_TXQ_COMP;
if ((istatus & (INTR_RXQ_COMP | INTR_RXQ_COAL | INTR_RXQ_COAL_TO)) != 0)
istatus |= INTR_RXQ_COAL | INTR_RXQ_COAL_TO | INTR_RXQ_COMP;
bus_space_write_4(sc->jme_bt_misc, sc->jme_bh_misc,
JME_INTR_STATUS, istatus);
if ((sc->jme_if.if_flags & IFF_RUNNING) == 0)
goto done;
#ifdef JMEDEBUG_RX
printf("jme_intr 0x%x RXCS 0x%x RXDBA 0x%x 0x%x RXQDC 0x%x RXNDA 0x%x RXMCS 0x%x\n", istatus,
bus_space_read_4(sc->jme_bt_mac, sc->jme_bh_mac, JME_RXCSR),
bus_space_read_4(sc->jme_bt_mac, sc->jme_bh_mac, JME_RXDBA_LO),
bus_space_read_4(sc->jme_bt_mac, sc->jme_bh_mac, JME_RXDBA_HI),
bus_space_read_4(sc->jme_bt_mac, sc->jme_bh_mac, JME_RXQDC),
bus_space_read_4(sc->jme_bt_mac, sc->jme_bh_mac, JME_RXNDA),
bus_space_read_4(sc->jme_bt_mac, sc->jme_bh_mac, JME_RXMAC));
printf("jme_intr RXUMA 0x%x 0x%x RXMCHT 0x%x 0x%x GHC 0x%x\n",
bus_space_read_4(sc->jme_bt_mac, sc->jme_bh_mac, JME_PAR0),
bus_space_read_4(sc->jme_bt_mac, sc->jme_bh_mac, JME_PAR1),
bus_space_read_4(sc->jme_bt_mac, sc->jme_bh_mac, JME_MAR0),
bus_space_read_4(sc->jme_bt_mac, sc->jme_bh_mac, JME_MAR1),
bus_space_read_4(sc->jme_bt_mac, sc->jme_bh_mac, JME_GHC));
#endif
if ((istatus & (INTR_RXQ_COMP | INTR_RXQ_COAL | INTR_RXQ_COAL_TO)) != 0)
jme_intr_rx(sc);
if ((istatus & INTR_RXQ_DESC_EMPTY) != 0) {
/*
* Notify hardware availability of new Rx
* buffers.
* Reading RXCSR takes very long time under
* heavy load so cache RXCSR value and writes
* the ORed value with the kick command to
* the RXCSR. This saves one register access
* cycle.
*/
sc->jme_rx_cons = 0;
bus_space_write_4(sc->jme_bt_mac, sc->jme_bh_mac,
JME_RXCSR,
sc->jme_rxcsr | RXCSR_RX_ENB | RXCSR_RXQ_START);
}
if ((istatus & (INTR_TXQ_COMP | INTR_TXQ_COAL | INTR_TXQ_COAL_TO)) != 0)
jme_ifstart(&sc->jme_if);
goto again;
done:
/* enable interrupts. */
bus_space_write_4(sc->jme_bt_misc, sc->jme_bh_misc,
JME_INTR_MASK_SET, JME_INTRS_ENABLE);
return 1;
}
static int
jme_ifioctl(struct ifnet *ifp, unsigned long cmd, void *data)
{
struct jme_softc *sc = ifp->if_softc;
int s, error;
struct ifreq *ifr;
struct ifcapreq *ifcr;
s = splnet();
/*
* we can't support at the same time jumbo frames and
* TX checksums offload/TSO
*/
switch(cmd) {
case SIOCSIFMTU:
ifr = data;
if (ifr->ifr_mtu > JME_TX_FIFO_SIZE &&
(ifp->if_capenable & (
IFCAP_CSUM_IPv4_Tx|IFCAP_CSUM_TCPv4_Tx|IFCAP_CSUM_UDPv4_Tx|
IFCAP_CSUM_TCPv6_Tx|IFCAP_CSUM_UDPv6_Tx|
IFCAP_TSOv4|IFCAP_TSOv6)) != 0) {
splx(s);
return EINVAL;
}
break;
case SIOCSIFCAP:
ifcr = data;
if (ifp->if_mtu > JME_TX_FIFO_SIZE &&
(ifcr->ifcr_capenable & (
IFCAP_CSUM_IPv4_Tx|IFCAP_CSUM_TCPv4_Tx|IFCAP_CSUM_UDPv4_Tx|
IFCAP_CSUM_TCPv6_Tx|IFCAP_CSUM_UDPv6_Tx|
IFCAP_TSOv4|IFCAP_TSOv6)) != 0) {
splx(s);
return EINVAL;
}
break;
}
error = ether_ioctl(ifp, cmd, data);
if (error == ENETRESET && (ifp->if_flags & IFF_RUNNING)) {
if (cmd == SIOCADDMULTI || cmd == SIOCDELMULTI) {
jme_set_filter(sc);
error = 0;
} else {
error = jme_init(ifp, 0);
}
}
splx(s);
return error;
}
static int
jme_encap(struct jme_softc *sc, struct mbuf **m_head)
{
struct jme_desc *txd;
struct jme_desc *desc;
struct mbuf *m;
struct m_tag *mtag;
int error, i, prod, headdsc, nsegs;
uint32_t cflags, tso_segsz;
if (((*m_head)->m_pkthdr.csum_flags & (M_CSUM_TSOv4|M_CSUM_TSOv6)) != 0){
/*
* Due to the adherence to NDIS specification JMC250
* assumes upper stack computed TCP pseudo checksum
* without including payload length. This breaks
* checksum offload for TSO case so recompute TCP
* pseudo checksum for JMC250. Hopefully this wouldn't
* be much burden on modern CPUs.
*/
bool v4 = ((*m_head)->m_pkthdr.csum_flags & M_CSUM_TSOv4) != 0;
int iphl = v4 ?
M_CSUM_DATA_IPv4_IPHL((*m_head)->m_pkthdr.csum_data) :
M_CSUM_DATA_IPv6_HL((*m_head)->m_pkthdr.csum_data);
/*
* note: we support vlan offloading, so we should never have
* a ETHERTYPE_VLAN packet here - so ETHER_HDR_LEN is always
* right.
*/
int hlen = ETHER_HDR_LEN + iphl;
if (__predict_false((*m_head)->m_len <
(hlen + sizeof(struct tcphdr)))) {
/*
* TCP/IP headers are not in the first mbuf; we need
* to do this the slow and painful way. Let's just
* hope this doesn't happen very often.
*/
struct tcphdr th;
m_copydata((*m_head), hlen, sizeof(th), &th);
if (v4) {
struct ip ip;
m_copydata((*m_head), ETHER_HDR_LEN,
sizeof(ip), &ip);
ip.ip_len = 0;
m_copyback((*m_head),
ETHER_HDR_LEN + offsetof(struct ip, ip_len),
sizeof(ip.ip_len), &ip.ip_len);
th.th_sum = in_cksum_phdr(ip.ip_src.s_addr,
ip.ip_dst.s_addr, htons(IPPROTO_TCP));
} else {
#if INET6
struct ip6_hdr ip6;
m_copydata((*m_head), ETHER_HDR_LEN,
sizeof(ip6), &ip6);
ip6.ip6_plen = 0;
m_copyback((*m_head), ETHER_HDR_LEN +
offsetof(struct ip6_hdr, ip6_plen),
sizeof(ip6.ip6_plen), &ip6.ip6_plen);
th.th_sum = in6_cksum_phdr(&ip6.ip6_src,
&ip6.ip6_dst, 0, htonl(IPPROTO_TCP));
#endif /* INET6 */
}
m_copyback((*m_head),
hlen + offsetof(struct tcphdr, th_sum),
sizeof(th.th_sum), &th.th_sum);
hlen += th.th_off << 2;
} else {
/*
* TCP/IP headers are in the first mbuf; we can do
* this the easy way.
*/
struct tcphdr *th;
if (v4) {
struct ip *ip =
(void *)(mtod((*m_head), char *) +
ETHER_HDR_LEN);
th = (void *)(mtod((*m_head), char *) + hlen);
ip->ip_len = 0;
th->th_sum = in_cksum_phdr(ip->ip_src.s_addr,
ip->ip_dst.s_addr, htons(IPPROTO_TCP));
} else {
#if INET6
struct ip6_hdr *ip6 =
(void *)(mtod((*m_head), char *) +
ETHER_HDR_LEN);
th = (void *)(mtod((*m_head), char *) + hlen);
ip6->ip6_plen = 0;
th->th_sum = in6_cksum_phdr(&ip6->ip6_src,
&ip6->ip6_dst, 0, htonl(IPPROTO_TCP));
#endif /* INET6 */
}
hlen += th->th_off << 2;
}
}
prod = sc->jme_tx_prod;
txd = &sc->jme_txring[prod];
error = bus_dmamap_load_mbuf(sc->jme_dmatag, sc->jme_txmbufm[prod],
*m_head, BUS_DMA_WRITE);
if (error) {
if (error == EFBIG) {
log(LOG_ERR, "%s: Tx packet consumes too many "
"DMA segments, dropping...\n",
device_xname(sc->jme_dev));
m_freem(*m_head);
m_head = NULL;
}
return (error);
}
/*
* Check descriptor overrun. Leave one free descriptor.
* Since we always use 64bit address mode for transmitting,
* each Tx request requires one more dummy descriptor.
*/
nsegs = sc->jme_txmbufm[prod]->dm_nsegs;
#ifdef JMEDEBUG_TX
printf("jme_encap prod %d nsegs %d jme_tx_cnt %d\n", prod, nsegs, sc->jme_tx_cnt);
#endif
if (sc->jme_tx_cnt + nsegs + 1 > JME_NBUFS - 1) {
bus_dmamap_unload(sc->jme_dmatag, sc->jme_txmbufm[prod]);
return (ENOBUFS);
}
bus_dmamap_sync(sc->jme_dmatag, sc->jme_txmbufm[prod],
0, sc->jme_txmbufm[prod]->dm_mapsize, BUS_DMASYNC_PREWRITE);
m = *m_head;
cflags = 0;
tso_segsz = 0;
/* Configure checksum offload and TSO. */
if ((m->m_pkthdr.csum_flags & (M_CSUM_TSOv4|M_CSUM_TSOv6)) != 0) {
tso_segsz = (uint32_t)m->m_pkthdr.segsz << JME_TD_MSS_SHIFT;
cflags |= JME_TD_TSO;
} else {
if ((m->m_pkthdr.csum_flags & M_CSUM_IPv4) != 0)
cflags |= JME_TD_IPCSUM;
if ((m->m_pkthdr.csum_flags & (M_CSUM_TCPv4|M_CSUM_TCPv6)) != 0)
cflags |= JME_TD_TCPCSUM;
if ((m->m_pkthdr.csum_flags & (M_CSUM_UDPv4|M_CSUM_UDPv6)) != 0)
cflags |= JME_TD_UDPCSUM;
}
/* Configure VLAN. */
if ((mtag = VLAN_OUTPUT_TAG(&sc->jme_ec, m)) != NULL) {
cflags |= (VLAN_TAG_VALUE(mtag) & JME_TD_VLAN_MASK);
cflags |= JME_TD_VLAN_TAG;
}
desc = &sc->jme_txring[prod];
desc->flags = htole32(cflags);
desc->buflen = htole32(tso_segsz);
desc->addr_hi = htole32(m->m_pkthdr.len);
desc->addr_lo = 0;
headdsc = prod;
sc->jme_tx_cnt++;
JME_DESC_INC(prod, JME_NBUFS);
for (i = 0; i < nsegs; i++) {
desc = &sc->jme_txring[prod];
desc->flags = htole32(JME_TD_OWN | JME_TD_64BIT);
desc->buflen =
htole32(sc->jme_txmbufm[headdsc]->dm_segs[i].ds_len);
desc->addr_hi = htole32(
JME_ADDR_HI(sc->jme_txmbufm[headdsc]->dm_segs[i].ds_addr));
desc->addr_lo = htole32(
JME_ADDR_LO(sc->jme_txmbufm[headdsc]->dm_segs[i].ds_addr));
bus_dmamap_sync(sc->jme_dmatag, sc->jme_txmap,
prod * sizeof(struct jme_desc), sizeof(struct jme_desc),
BUS_DMASYNC_PREREAD | BUS_DMASYNC_PREWRITE);
sc->jme_txmbuf[prod] = NULL;
sc->jme_tx_cnt++;
JME_DESC_INC(prod, JME_NBUFS);
}
/* Update producer index. */
sc->jme_tx_prod = prod;
#ifdef JMEDEBUG_TX
printf("jme_encap prod now %d\n", sc->jme_tx_prod);
#endif
/*
* Finally request interrupt and give the first descriptor
* owenership to hardware.
*/
desc = &sc->jme_txring[headdsc];
desc->flags |= htole32(JME_TD_OWN | JME_TD_INTR);
bus_dmamap_sync(sc->jme_dmatag, sc->jme_txmap,
headdsc * sizeof(struct jme_desc), sizeof(struct jme_desc),
BUS_DMASYNC_PREREAD | BUS_DMASYNC_PREWRITE);
sc->jme_txmbuf[headdsc] = m;
return (0);
}
static void
jme_txeof(struct jme_softc *sc)
{
struct ifnet *ifp;
struct jme_desc *desc;
uint32_t status;
int cons, cons0, nsegs, seg;
ifp = &sc->jme_if;
#ifdef JMEDEBUG_TX
printf("jme_txeof cons %d prod %d\n",
sc->jme_tx_cons, sc->jme_tx_prod);
printf("jme_txeof JME_TXCSR 0x%x JME_TXDBA_LO 0x%x JME_TXDBA_HI 0x%x "
"JME_TXQDC 0x%x JME_TXNDA 0x%x JME_TXMAC 0x%x JME_TXPFC 0x%x "
"JME_TXTRHD 0x%x\n",
bus_space_read_4(sc->jme_bt_mac, sc->jme_bh_mac, JME_TXCSR),
bus_space_read_4(sc->jme_bt_mac, sc->jme_bh_mac, JME_TXDBA_LO),
bus_space_read_4(sc->jme_bt_mac, sc->jme_bh_mac, JME_TXDBA_HI),
bus_space_read_4(sc->jme_bt_mac, sc->jme_bh_mac, JME_TXQDC),
bus_space_read_4(sc->jme_bt_mac, sc->jme_bh_mac, JME_TXNDA),
bus_space_read_4(sc->jme_bt_mac, sc->jme_bh_mac, JME_TXMAC),
bus_space_read_4(sc->jme_bt_mac, sc->jme_bh_mac, JME_TXPFC),
bus_space_read_4(sc->jme_bt_mac, sc->jme_bh_mac, JME_TXTRHD));
for (cons = sc->jme_tx_cons; cons != sc->jme_tx_prod; ) {
desc = &sc->jme_txring[cons];
printf("ring[%d] 0x%x 0x%x 0x%x 0x%x\n", cons,
desc->flags, desc->buflen, desc->addr_hi, desc->addr_lo);
JME_DESC_INC(cons, JME_NBUFS);
}
#endif
cons = sc->jme_tx_cons;
if (cons == sc->jme_tx_prod)
return;
/*
* Go through our Tx list and free mbufs for those
* frames which have been transmitted.
*/
for (; cons != sc->jme_tx_prod;) {
bus_dmamap_sync(sc->jme_dmatag, sc->jme_txmap,
cons * sizeof(struct jme_desc), sizeof(struct jme_desc),
BUS_DMASYNC_POSTREAD | BUS_DMASYNC_POSTWRITE);
desc = &sc->jme_txring[cons];
status = le32toh(desc->flags);
#ifdef JMEDEBUG_TX
printf("jme_txeof %i status 0x%x nsegs %d\n", cons, status,
sc->jme_txmbufm[cons]->dm_nsegs);
#endif
if (status & JME_TD_OWN)
break;
if ((status & (JME_TD_TMOUT | JME_TD_RETRY_EXP)) != 0)
ifp->if_oerrors++;
else {
ifp->if_opackets++;
if ((status & JME_TD_COLLISION) != 0)
ifp->if_collisions +=
le32toh(desc->buflen) &
JME_TD_BUF_LEN_MASK;
}
/*
* Only the first descriptor of multi-descriptor
* transmission is updated so driver have to skip entire
* chained buffers for the transmiited frame. In other
* words, JME_TD_OWN bit is valid only at the first
* descriptor of a multi-descriptor transmission.
*/
nsegs = sc->jme_txmbufm[cons]->dm_nsegs;
cons0 = cons;
JME_DESC_INC(cons, JME_NBUFS);
for (seg = 1; seg < nsegs + 1; seg++) {
bus_dmamap_sync(sc->jme_dmatag, sc->jme_txmap,
cons * sizeof(struct jme_desc),
sizeof(struct jme_desc),
BUS_DMASYNC_POSTREAD | BUS_DMASYNC_POSTWRITE);
sc->jme_txring[cons].flags = 0;
JME_DESC_INC(cons, JME_NBUFS);
}
/* Reclaim transferred mbufs. */
bus_dmamap_sync(sc->jme_dmatag, sc->jme_txmbufm[cons0],
0, sc->jme_txmbufm[cons0]->dm_mapsize,
BUS_DMASYNC_POSTWRITE);
bus_dmamap_unload(sc->jme_dmatag, sc->jme_txmbufm[cons0]);
KASSERT(sc->jme_txmbuf[cons0] != NULL);
m_freem(sc->jme_txmbuf[cons0]);
sc->jme_txmbuf[cons0] = NULL;
sc->jme_tx_cnt -= nsegs + 1;
KASSERT(sc->jme_tx_cnt >= 0);
sc->jme_if.if_flags &= ~IFF_OACTIVE;
}
sc->jme_tx_cons = cons;
/* Unarm watchog timer when there is no pending descriptors in queue. */
if (sc->jme_tx_cnt == 0)
ifp->if_timer = 0;
#ifdef JMEDEBUG_TX
printf("jme_txeof jme_tx_cnt %d\n", sc->jme_tx_cnt);
#endif
}
static void
jme_ifstart(struct ifnet *ifp)
{
jme_softc_t *sc = ifp->if_softc;
struct mbuf *mb_head;
int enq;
/*
* check if we can free some desc.
* Clear TX interrupt status to reset TX coalescing counters.
*/
bus_space_write_4(sc->jme_bt_misc, sc->jme_bh_misc,
JME_INTR_STATUS, INTR_TXQ_COMP);
jme_txeof(sc);
if ((sc->jme_if.if_flags & (IFF_RUNNING|IFF_OACTIVE)) != IFF_RUNNING)
return;
for (enq = 0;; enq++) {
nexttx:
/* Grab a paquet for output */
IFQ_DEQUEUE(&ifp->if_snd, mb_head);
if (mb_head == NULL) {
#ifdef JMEDEBUG_TX
printf("%s: nothing to send\n", __func__);
#endif
break;
}
/* try to add this mbuf to the TX ring */
if (jme_encap(sc, &mb_head)) {
if (mb_head == NULL) {
ifp->if_oerrors++;
/* packet dropped, try next one */
goto nexttx;
}
/* resource shortage, try again later */
IF_PREPEND(&ifp->if_snd, mb_head);
ifp->if_flags |= IFF_OACTIVE;
break;
}
#if NBPFILTER > 0
/* Pass packet to bpf if there is a listener */
if (ifp->if_bpf)
bpf_mtap(ifp->if_bpf, mb_head);
#endif
}
#ifdef JMEDEBUG_TX
printf("jme_ifstart enq %d\n", enq);
#endif
if (enq) {
/*
* Set a 5 second timer just in case we don't hear from
* the card again.
*/
ifp->if_timer = 5;
/*
* Reading TXCSR takes very long time under heavy load
* so cache TXCSR value and writes the ORed value with
* the kick command to the TXCSR. This saves one register
* access cycle.
*/
bus_space_write_4(sc->jme_bt_mac, sc->jme_bh_mac, JME_TXCSR,
sc->jme_txcsr | TXCSR_TX_ENB | TXCSR_TXQ_N_START(TXCSR_TXQ0));
#ifdef JMEDEBUG_TX
printf("jme_ifstart JME_TXCSR 0x%x JME_TXDBA_LO 0x%x JME_TXDBA_HI 0x%x "
"JME_TXQDC 0x%x JME_TXNDA 0x%x JME_TXMAC 0x%x JME_TXPFC 0x%x "
"JME_TXTRHD 0x%x\n",
bus_space_read_4(sc->jme_bt_mac, sc->jme_bh_mac, JME_TXCSR),
bus_space_read_4(sc->jme_bt_mac, sc->jme_bh_mac, JME_TXDBA_LO),
bus_space_read_4(sc->jme_bt_mac, sc->jme_bh_mac, JME_TXDBA_HI),
bus_space_read_4(sc->jme_bt_mac, sc->jme_bh_mac, JME_TXQDC),
bus_space_read_4(sc->jme_bt_mac, sc->jme_bh_mac, JME_TXNDA),
bus_space_read_4(sc->jme_bt_mac, sc->jme_bh_mac, JME_TXMAC),
bus_space_read_4(sc->jme_bt_mac, sc->jme_bh_mac, JME_TXPFC),
bus_space_read_4(sc->jme_bt_mac, sc->jme_bh_mac, JME_TXTRHD));
#endif
}
}
static void
jme_ifwatchdog(struct ifnet *ifp)
{
jme_softc_t *sc = ifp->if_softc;
if ((ifp->if_flags & IFF_RUNNING) == 0)
return;
printf("%s: device timeout\n", device_xname(sc->jme_dev));
ifp->if_oerrors++;
jme_init(ifp, 0);
}
static int
jme_mediachange(struct ifnet *ifp)
{
int error;
jme_softc_t *sc = ifp->if_softc;
if ((error = mii_mediachg(&sc->jme_mii)) == ENXIO)
error = 0;
else if (error != 0) {
aprint_error_dev(sc->jme_dev, "could not set media\n");
return error;
}
return 0;
}
static void
jme_ticks(void *v)
{
jme_softc_t *sc = v;
int s = splnet();
/* Tick the MII. */
mii_tick(&sc->jme_mii);
/* every seconds */
callout_reset(&sc->jme_tick_ch, hz, jme_ticks, sc);
splx(s);
}
static void
jme_mac_config(jme_softc_t *sc)
{
uint32_t ghc, gpreg, rxmac, txmac, txpause;
struct mii_data *mii = &sc->jme_mii;
ghc = 0;
rxmac = bus_space_read_4(sc->jme_bt_mac, sc->jme_bh_mac, JME_RXMAC);
rxmac &= ~RXMAC_FC_ENB;
txmac = bus_space_read_4(sc->jme_bt_mac, sc->jme_bh_mac, JME_TXMAC);
txmac &= ~(TXMAC_CARRIER_EXT | TXMAC_FRAME_BURST);
txpause = bus_space_read_4(sc->jme_bt_mac, sc->jme_bh_mac, JME_TXPFC);
txpause &= ~TXPFC_PAUSE_ENB;
if (mii->mii_media_active & IFM_FDX) {
ghc |= GHC_FULL_DUPLEX;
rxmac &= ~RXMAC_COLL_DET_ENB;
txmac &= ~(TXMAC_COLL_ENB | TXMAC_CARRIER_SENSE |
TXMAC_BACKOFF | TXMAC_CARRIER_EXT |
TXMAC_FRAME_BURST);
/* Disable retry transmit timer/retry limit. */
bus_space_write_4(sc->jme_bt_mac, sc->jme_bh_mac, JME_TXTRHD,
bus_space_read_4(sc->jme_bt_mac, sc->jme_bh_mac, JME_TXTRHD)
& ~(TXTRHD_RT_PERIOD_ENB | TXTRHD_RT_LIMIT_ENB));
} else {
rxmac |= RXMAC_COLL_DET_ENB;
txmac |= TXMAC_COLL_ENB | TXMAC_CARRIER_SENSE | TXMAC_BACKOFF;
/* Enable retry transmit timer/retry limit. */
bus_space_write_4(sc->jme_bt_mac, sc->jme_bh_mac, JME_TXTRHD,
bus_space_read_4(sc->jme_bt_mac, sc->jme_bh_mac, JME_TXTRHD) | TXTRHD_RT_PERIOD_ENB | TXTRHD_RT_LIMIT_ENB);
}
/* Reprogram Tx/Rx MACs with resolved speed/duplex. */
switch (IFM_SUBTYPE(mii->mii_media_active)) {
case IFM_10_T:
ghc |= GHC_SPEED_10 | GHC_CLKSRC_10_100;
break;
case IFM_100_TX:
ghc |= GHC_SPEED_100 | GHC_CLKSRC_10_100;
break;
case IFM_1000_T:
ghc |= GHC_SPEED_1000 | GHC_CLKSRC_1000;
if ((IFM_OPTIONS(mii->mii_media_active) & IFM_FDX) == 0)
txmac |= TXMAC_CARRIER_EXT | TXMAC_FRAME_BURST;
break;
default:
break;
}
if ((sc->jme_flags & JME_FLAG_GIGA) &&
sc->jme_chip_rev == DEVICEREVID_JMC250_A2) {
/*
* Workaround occasional packet loss issue of JMC250 A2
* when it runs on half-duplex media.
*/
#ifdef JMEDEBUG
printf("JME250 A2 workaround\n");
#endif
gpreg = bus_space_read_4(sc->jme_bt_misc, sc->jme_bh_misc,
JME_GPREG1);
if ((IFM_OPTIONS(mii->mii_media_active) & IFM_FDX) != 0)
gpreg &= ~GPREG1_HDPX_FIX;
else
gpreg |= GPREG1_HDPX_FIX;
bus_space_write_4(sc->jme_bt_misc, sc->jme_bh_misc,
JME_GPREG1, gpreg);
/* Workaround CRC errors at 100Mbps on JMC250 A2. */
if (IFM_SUBTYPE(mii->mii_media_active) == IFM_100_TX) {
/* Extend interface FIFO depth. */
jme_mii_write(sc->jme_dev, sc->jme_phyaddr,
0x1B, 0x0000);
} else {
/* Select default interface FIFO depth. */
jme_mii_write(sc->jme_dev, sc->jme_phyaddr,
0x1B, 0x0004);
}
}
bus_space_write_4(sc->jme_bt_mac, sc->jme_bh_mac, JME_GHC, ghc);
bus_space_write_4(sc->jme_bt_mac, sc->jme_bh_mac, JME_RXMAC, rxmac);
bus_space_write_4(sc->jme_bt_mac, sc->jme_bh_mac, JME_TXMAC, txmac);
bus_space_write_4(sc->jme_bt_mac, sc->jme_bh_mac, JME_TXPFC, txpause);
}
static void
jme_set_filter(jme_softc_t *sc)
{
struct ifnet *ifp = &sc->jme_if;
struct ether_multistep step;
struct ether_multi *enm;
uint32_t hash[2] = {0, 0};
int i;
uint32_t rxcfg;
rxcfg = bus_space_read_4(sc->jme_bt_mac, sc->jme_bh_mac, JME_RXMAC);
rxcfg &= ~ (RXMAC_BROADCAST | RXMAC_PROMISC | RXMAC_MULTICAST |
RXMAC_ALLMULTI);
/* Always accept frames destined to our station address. */
rxcfg |= RXMAC_UNICAST;
if ((ifp->if_flags & IFF_BROADCAST) != 0)
rxcfg |= RXMAC_BROADCAST;
if ((ifp->if_flags & (IFF_PROMISC | IFF_ALLMULTI)) != 0) {
if ((ifp->if_flags & IFF_PROMISC) != 0)
rxcfg |= RXMAC_PROMISC;
if ((ifp->if_flags & IFF_ALLMULTI) != 0)
rxcfg |= RXMAC_ALLMULTI;
bus_space_write_4(sc->jme_bt_mac, sc->jme_bh_mac,
JME_MAR0, 0xFFFFFFFF);
bus_space_write_4(sc->jme_bt_mac, sc->jme_bh_mac,
JME_MAR1, 0xFFFFFFFF);
bus_space_write_4(sc->jme_bt_mac, sc->jme_bh_mac,
JME_RXMAC, rxcfg);
return;
}
/*
* Set up the multicast address filter by passing all multicast
* addresses through a CRC generator, and then using the low-order
* 6 bits as an index into the 64 bit multicast hash table. The
* high order bits select the register, while the rest of the bits
* select the bit within the register.
*/
rxcfg |= RXMAC_MULTICAST;
memset(hash, 0, sizeof(hash));
ETHER_FIRST_MULTI(step, &sc->jme_ec, enm);
while (enm != NULL) {
#ifdef JEMDBUG
printf("%s: addrs %s %s\n", __func__,
ether_sprintf(enm->enm_addrlo),
ether_sprintf(enm->enm_addrhi));
#endif
if (memcmp(enm->enm_addrlo, enm->enm_addrhi, 6) == 0) {
i = ether_crc32_be(enm->enm_addrlo, 6);
/* Just want the 6 least significant bits. */
i &= 0x3f;
hash[i / 32] |= 1 << (i%32);
} else {
hash[0] = hash[1] = 0xffffffff;
sc->jme_if.if_flags |= IFF_ALLMULTI;
break;
}
ETHER_NEXT_MULTI(step, enm);
}
#ifdef JMEDEBUG
printf("%s: hash1 %x has2 %x\n", __func__, hash[0], hash[1]);
#endif
bus_space_write_4(sc->jme_bt_mac, sc->jme_bh_mac, JME_MAR0, hash[0]);
bus_space_write_4(sc->jme_bt_mac, sc->jme_bh_mac, JME_MAR1, hash[1]);
bus_space_write_4(sc->jme_bt_mac, sc->jme_bh_mac, JME_RXMAC, rxcfg);
}
#if 0
static int
jme_multicast_hash(uint8_t *a)
{
int hash;
#define DA(addr,bit) (addr[5 - (bit / 8)] & (1 << (bit % 8)))
#define xor8(a,b,c,d,e,f,g,h) \
(((a != 0) + (b != 0) + (c != 0) + (d != 0) + \
(e != 0) + (f != 0) + (g != 0) + (h != 0)) & 1)
hash = xor8(DA(a,0), DA(a, 6), DA(a,12), DA(a,18), DA(a,24), DA(a,30),
DA(a,36), DA(a,42));
hash |= xor8(DA(a,1), DA(a, 7), DA(a,13), DA(a,19), DA(a,25), DA(a,31),
DA(a,37), DA(a,43)) << 1;
hash |= xor8(DA(a,2), DA(a, 8), DA(a,14), DA(a,20), DA(a,26), DA(a,32),
DA(a,38), DA(a,44)) << 2;
hash |= xor8(DA(a,3), DA(a, 9), DA(a,15), DA(a,21), DA(a,27), DA(a,33),
DA(a,39), DA(a,45)) << 3;
hash |= xor8(DA(a,4), DA(a,10), DA(a,16), DA(a,22), DA(a,28), DA(a,34),
DA(a,40), DA(a,46)) << 4;
hash |= xor8(DA(a,5), DA(a,11), DA(a,17), DA(a,23), DA(a,29), DA(a,35),
DA(a,41), DA(a,47)) << 5;
return hash;
}
#endif
static int
jme_eeprom_read_byte(struct jme_softc *sc, uint8_t addr, uint8_t *val)
{
uint32_t reg;
int i;
*val = 0;
for (i = JME_EEPROM_TIMEOUT / 10; i > 0; i--) {
reg = bus_space_read_4(sc->jme_bt_phy, sc->jme_bh_phy,
JME_SMBCSR);
if ((reg & SMBCSR_HW_BUSY_MASK) == SMBCSR_HW_IDLE)
break;
delay(10);
}
if (i == 0) {
aprint_error_dev(sc->jme_dev, "EEPROM idle timeout!\n");
return (ETIMEDOUT);
}
reg = ((uint32_t)addr << SMBINTF_ADDR_SHIFT) & SMBINTF_ADDR_MASK;
bus_space_write_4(sc->jme_bt_phy, sc->jme_bh_phy,
JME_SMBINTF, reg | SMBINTF_RD | SMBINTF_CMD_TRIGGER);
for (i = JME_EEPROM_TIMEOUT / 10; i > 0; i--) {
delay(10);
reg = bus_space_read_4(sc->jme_bt_phy, sc->jme_bh_phy,
JME_SMBINTF);
if ((reg & SMBINTF_CMD_TRIGGER) == 0)
break;
}
if (i == 0) {
aprint_error_dev(sc->jme_dev, "EEPROM read timeout!\n");
return (ETIMEDOUT);
}
reg = bus_space_read_4(sc->jme_bt_phy, sc->jme_bh_phy, JME_SMBINTF);
*val = (reg & SMBINTF_RD_DATA_MASK) >> SMBINTF_RD_DATA_SHIFT;
return (0);
}
static int
jme_eeprom_macaddr(struct jme_softc *sc)
{
uint8_t eaddr[ETHER_ADDR_LEN];
uint8_t fup, reg, val;
uint32_t offset;
int match;
offset = 0;
if (jme_eeprom_read_byte(sc, offset++, &fup) != 0 ||
fup != JME_EEPROM_SIG0)
return (ENOENT);
if (jme_eeprom_read_byte(sc, offset++, &fup) != 0 ||
fup != JME_EEPROM_SIG1)
return (ENOENT);
match = 0;
do {
if (jme_eeprom_read_byte(sc, offset, &fup) != 0)
break;
if (JME_EEPROM_MKDESC(JME_EEPROM_FUNC0, JME_EEPROM_PAGE_BAR1)
== (fup & (JME_EEPROM_FUNC_MASK|JME_EEPROM_PAGE_MASK))) {
if (jme_eeprom_read_byte(sc, offset + 1, &reg) != 0)
break;
if (reg >= JME_PAR0 &&
reg < JME_PAR0 + ETHER_ADDR_LEN) {
if (jme_eeprom_read_byte(sc, offset + 2,
&val) != 0)
break;
eaddr[reg - JME_PAR0] = val;
match++;
}
}
if (fup & JME_EEPROM_DESC_END)
break;
/* Try next eeprom descriptor. */
offset += JME_EEPROM_DESC_BYTES;
} while (match != ETHER_ADDR_LEN && offset < JME_EEPROM_END);
if (match == ETHER_ADDR_LEN) {
memcpy( sc->jme_enaddr, eaddr, ETHER_ADDR_LEN);
return (0);
}
return (ENOENT);
}
/*
* Set up sysctl(3) MIB, hw.jme.* - Individual controllers will be
* set up in jme_pci_attach()
*/
SYSCTL_SETUP(sysctl_jme, "sysctl jme subtree setup")
{
int rc;
const struct sysctlnode *node;
if ((rc = sysctl_createv(clog, 0, NULL, NULL,
0, CTLTYPE_NODE, "hw", NULL,
NULL, 0, NULL, 0, CTL_HW, CTL_EOL)) != 0) {
goto err;
}
if ((rc = sysctl_createv(clog, 0, NULL, &node,
0, CTLTYPE_NODE, "jme",
SYSCTL_DESCR("jme interface controls"),
NULL, 0, NULL, 0, CTL_HW, CTL_CREATE, CTL_EOL)) != 0) {
goto err;
}
jme_root_num = node->sysctl_num;
return;
err:
aprint_error("%s: syctl_createv failed (rc = %d)\n", __func__, rc);
}
static int
jme_sysctl_intrxto(SYSCTLFN_ARGS)
{
int error, t;
struct sysctlnode node;
struct jme_softc *sc;
uint32_t reg;
node = *rnode;
sc = node.sysctl_data;
t = sc->jme_intrxto;
node.sysctl_data = &t;
error = sysctl_lookup(SYSCTLFN_CALL(&node));
if (error || newp == NULL)
return error;
if (t < PCCRX_COAL_TO_MIN || t > PCCRX_COAL_TO_MAX)
return EINVAL;
/*
* update the softc with sysctl-changed value, and mark
* for hardware update
*/
sc->jme_intrxto = t;
/* Configure Rx queue 0 packet completion coalescing. */
reg = (sc->jme_intrxto << PCCRX_COAL_TO_SHIFT) & PCCRX_COAL_TO_MASK;
reg |= (sc->jme_intrxct << PCCRX_COAL_PKT_SHIFT) & PCCRX_COAL_PKT_MASK;
bus_space_write_4(sc->jme_bt_misc, sc->jme_bh_misc, JME_PCCRX0, reg);
return 0;
}
static int
jme_sysctl_intrxct(SYSCTLFN_ARGS)
{
int error, t;
struct sysctlnode node;
struct jme_softc *sc;
uint32_t reg;
node = *rnode;
sc = node.sysctl_data;
t = sc->jme_intrxct;
node.sysctl_data = &t;
error = sysctl_lookup(SYSCTLFN_CALL(&node));
if (error || newp == NULL)
return error;
if (t < PCCRX_COAL_PKT_MIN || t > PCCRX_COAL_PKT_MAX)
return EINVAL;
/*
* update the softc with sysctl-changed value, and mark
* for hardware update
*/
sc->jme_intrxct = t;
/* Configure Rx queue 0 packet completion coalescing. */
reg = (sc->jme_intrxto << PCCRX_COAL_TO_SHIFT) & PCCRX_COAL_TO_MASK;
reg |= (sc->jme_intrxct << PCCRX_COAL_PKT_SHIFT) & PCCRX_COAL_PKT_MASK;
bus_space_write_4(sc->jme_bt_misc, sc->jme_bh_misc, JME_PCCRX0, reg);
return 0;
}
static int
jme_sysctl_inttxto(SYSCTLFN_ARGS)
{
int error, t;
struct sysctlnode node;
struct jme_softc *sc;
uint32_t reg;
node = *rnode;
sc = node.sysctl_data;
t = sc->jme_inttxto;
node.sysctl_data = &t;
error = sysctl_lookup(SYSCTLFN_CALL(&node));
if (error || newp == NULL)
return error;
if (t < PCCTX_COAL_TO_MIN || t > PCCTX_COAL_TO_MAX)
return EINVAL;
/*
* update the softc with sysctl-changed value, and mark
* for hardware update
*/
sc->jme_inttxto = t;
/* Configure Tx queue 0 packet completion coalescing. */
reg = (sc->jme_inttxto << PCCTX_COAL_TO_SHIFT) & PCCTX_COAL_TO_MASK;
reg |= (sc->jme_inttxct << PCCTX_COAL_PKT_SHIFT) & PCCTX_COAL_PKT_MASK;
reg |= PCCTX_COAL_TXQ0;
bus_space_write_4(sc->jme_bt_misc, sc->jme_bh_misc, JME_PCCTX, reg);
return 0;
}
static int
jme_sysctl_inttxct(SYSCTLFN_ARGS)
{
int error, t;
struct sysctlnode node;
struct jme_softc *sc;
uint32_t reg;
node = *rnode;
sc = node.sysctl_data;
t = sc->jme_inttxct;
node.sysctl_data = &t;
error = sysctl_lookup(SYSCTLFN_CALL(&node));
if (error || newp == NULL)
return error;
if (t < PCCTX_COAL_PKT_MIN || t > PCCTX_COAL_PKT_MAX)
return EINVAL;
/*
* update the softc with sysctl-changed value, and mark
* for hardware update
*/
sc->jme_inttxct = t;
/* Configure Tx queue 0 packet completion coalescing. */
reg = (sc->jme_inttxto << PCCTX_COAL_TO_SHIFT) & PCCTX_COAL_TO_MASK;
reg |= (sc->jme_inttxct << PCCTX_COAL_PKT_SHIFT) & PCCTX_COAL_PKT_MASK;
reg |= PCCTX_COAL_TXQ0;
bus_space_write_4(sc->jme_bt_misc, sc->jme_bh_misc, JME_PCCTX, reg);
return 0;
}