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

1618 lines
38 KiB
C
Raw Normal View History

usbnet: Delete the core lock from the API. Init/stop and ioctl happen under IFNET_LOCK. Multicast updates only happen after init and before stop. Core lock is no longer a relevant part of the API. Internally, it serves essentially just to lock out asynchronous mii activity during init/stop.
2022-03-03 08:54:52 +03:00
/* $NetBSD: usbnet.c,v 1.84 2022/03/03 05:54:52 riastradh Exp $ */
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
/*
* Copyright (c) 2019 Matthew R. Green
* 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.
*
* 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.
*/
/*
don't set dv_private. it's a bad pattern and only didn't trip kmem_free() lossage because struct usbnet is at the start of the softc. for now, enforce this as part of the ABI. catch up urndis with tx_prepare checking buffer length, and also add an assert to usbnet_start_locked() to match.
2019-08-12 02:55:43 +03:00
* Common code shared between USB network drivers.
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
*/
#include <sys/cdefs.h>
usbnet: Delete the core lock from the API. Init/stop and ioctl happen under IFNET_LOCK. Multicast updates only happen after init and before stop. Core lock is no longer a relevant part of the API. Internally, it serves essentially just to lock out asynchronous mii activity during init/stop.
2022-03-03 08:54:52 +03:00
__KERNEL_RCSID(0, "$NetBSD: usbnet.c,v 1.84 2022/03/03 05:54:52 riastradh Exp $");
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
#include <sys/param.h>
#include <sys/kernel.h>
#include <sys/kmem.h>
#include <sys/module.h>
- remove unused uncd_rx_cnt - add USBNETHIST_CALLARGSN and use it in many places - add more history logs - add a log if watchdog abort leaves uncd_tx_cnt non zero - add a unique (32-bit) number for a each usbnet device and use it for logging - explicitly check for full tx ring in usbnet_start_locked() and return early, which avoids setting the 5 second timer and triggering a watchdog - reset uncd_tx_prod and uncd_tx_cnt to zero in stop
2019-08-23 07:29:28 +03:00
#include <sys/atomic.h>
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
#include <dev/usb/usbnet.h>
- adjust usbnet interface to allow usbd_open_pipe_intr(), from the new comment: * if un_intr_buf is not NULL, use usbd_open_pipe_intr() not * usbd_open_pipe() for USBNET_ENDPT_INTR, with this buffer, * size, and interval. the standard handling is in usbnet.c, with a callback to deal with the interrupt it self. not fully tested, designed for if_aue.c and a few others not yet converted. - make usbhist for usbnet.c work, thanks paulg - usbnet_init_rx_tx() clears out all allocations upon failure now - add usbnet_ec() to get a pointer to the struct ethercom - add usbnet_{lock,unlock,owned}*() to lock/unlock the various locks and *owned*() for asserting welcome 9.99.3!
2019-08-04 11:59:13 +03:00
#include <dev/usb/usbhist.h>
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
reduce the scope of struct usbnet: - move a large number of members internal to usbnet.c's new "struct usbnet_private". - provide accessors for a few of these - move struct usbnet_cdata into usbnet.c as well, but move bufsz, list count, and xfer flags back out into struct usbnet, and have them set as part of the setup efore usbnet_attach() - split the intr pipe parts into their own structure - move all the main usbnet*lock* code into usbnet.c too usbnet_attach() goes down to 2 args, and the inputs needed are now the full contents of 'struct usbnet' besides the driver owned 'un_flags' and usbnet owned 'un_pri'. welcome netbsd 9.99.6.
2019-08-10 05:17:36 +03:00
struct usbnet_cdata {
struct usbnet_chain *uncd_tx_chain;
struct usbnet_chain *uncd_rx_chain;
int uncd_tx_prod;
int uncd_tx_cnt;
};
struct usbnet_private {
/*
Define and implement a locking protocol for the ifmedia / mii layers: - MP-safe drivers provide a mutex to ifmedia that is used to serialize access to media-related structures / hardware regsiters. Converted drivers use the new ifmedia_init_with_lock() function for this. The new name is provided to ease the transition. - Un-converted drivers continue to call ifmedia_init(), which will supply a compatibility lock to be used instead. Several media-related entry points must be aware of this compatibility lock, and are able to acquire it recursively a limited number of times, if needed. This is a SPIN mutex with priority IPL_NET. - This same lock is used to serialize access to PHY registers and other MII-related data structures. The PHY drivers are modified to acquire and release the lock, as needed, and assert the lock is held as a diagnostic aid. The "usbnet" framework has had an overhaul of its internal locking protocols to fit in with the media / mii changes, and the drivers adapted. USB wifi drivers have been changed to provide their own adaptive mutex to the ifmedia later via a new ieee80211_media_init_with_lock() function. This is required because the USB drivers need an adaptive mutex. Besised "usbnet", a few other drivers are converted: vmx, wm, ixgbe / ixv. mcx also now calls ifmedia_init_with_lock() because it needs to also use an adaptive mutex. The mcx driver still needs to be fully converted to NET_MPSAFE.
2020-03-16 02:04:50 +03:00
* - unp_core_lock protects most of this structure, the public one,
* and the MII / media data.
reduce the scope of struct usbnet: - move a large number of members internal to usbnet.c's new "struct usbnet_private". - provide accessors for a few of these - move struct usbnet_cdata into usbnet.c as well, but move bufsz, list count, and xfer flags back out into struct usbnet, and have them set as part of the setup efore usbnet_attach() - split the intr pipe parts into their own structure - move all the main usbnet*lock* code into usbnet.c too usbnet_attach() goes down to 2 args, and the inputs needed are now the full contents of 'struct usbnet' besides the driver owned 'un_flags' and usbnet owned 'un_pri'. welcome netbsd 9.99.6.
2019-08-10 05:17:36 +03:00
* - unp_rxlock protects the rx path and its data
* - unp_txlock protects the tx path and its data
document usbnet_private locking. minor knf.
2019-09-09 10:20:16 +03:00
*
* the lock ordering is:
Define and implement a locking protocol for the ifmedia / mii layers: - MP-safe drivers provide a mutex to ifmedia that is used to serialize access to media-related structures / hardware regsiters. Converted drivers use the new ifmedia_init_with_lock() function for this. The new name is provided to ease the transition. - Un-converted drivers continue to call ifmedia_init(), which will supply a compatibility lock to be used instead. Several media-related entry points must be aware of this compatibility lock, and are able to acquire it recursively a limited number of times, if needed. This is a SPIN mutex with priority IPL_NET. - This same lock is used to serialize access to PHY registers and other MII-related data structures. The PHY drivers are modified to acquire and release the lock, as needed, and assert the lock is held as a diagnostic aid. The "usbnet" framework has had an overhaul of its internal locking protocols to fit in with the media / mii changes, and the drivers adapted. USB wifi drivers have been changed to provide their own adaptive mutex to the ifmedia later via a new ieee80211_media_init_with_lock() function. This is required because the USB drivers need an adaptive mutex. Besised "usbnet", a few other drivers are converted: vmx, wm, ixgbe / ixv. mcx also now calls ifmedia_init_with_lock() because it needs to also use an adaptive mutex. The mcx driver still needs to be fully converted to NET_MPSAFE.
2020-03-16 02:04:50 +03:00
* ifnet lock -> unp_core_lock -> unp_rxlock -> unp_txlock
usbnet: Apply hardware multicast filter updates synchronously again. To make this work: 1. Do it only under a new lock, unp_mcastlock. This lock lives at IPL_SOFTCLOCK so it can be taken from network stack callouts. It is forbidden to acquire the usbnet core lock under unp_mcastlock. 2. Do it only after usbnet_init_rx_tx and before usbnet_stop; if issued at any other time, drop the update on the floor. 3. Make usbnet_init_rx_tx apply any pending multicast filter updates under the lock before setting the flag that allows SIOCADDMULTI or SIOCDELMULTI to apply the updates. 4. Remove core lock asserts from various drivers' register access routines. This is necessary because the multicast filter updates are done with register reads/writes, but _cannot_ take the core lock when the caller holds softnet_lock. This now programs the hardware multicast filter redundantly in many drivers which already explicitly call *_uno_mcast from the *_uno_init routines. This is probably harmless, but it will likely be better to remove the explicit calls.
2022-03-03 08:53:23 +03:00
* -> unp_mcastlock
Define and implement a locking protocol for the ifmedia / mii layers: - MP-safe drivers provide a mutex to ifmedia that is used to serialize access to media-related structures / hardware regsiters. Converted drivers use the new ifmedia_init_with_lock() function for this. The new name is provided to ease the transition. - Un-converted drivers continue to call ifmedia_init(), which will supply a compatibility lock to be used instead. Several media-related entry points must be aware of this compatibility lock, and are able to acquire it recursively a limited number of times, if needed. This is a SPIN mutex with priority IPL_NET. - This same lock is used to serialize access to PHY registers and other MII-related data structures. The PHY drivers are modified to acquire and release the lock, as needed, and assert the lock is held as a diagnostic aid. The "usbnet" framework has had an overhaul of its internal locking protocols to fit in with the media / mii changes, and the drivers adapted. USB wifi drivers have been changed to provide their own adaptive mutex to the ifmedia later via a new ieee80211_media_init_with_lock() function. This is required because the USB drivers need an adaptive mutex. Besised "usbnet", a few other drivers are converted: vmx, wm, ixgbe / ixv. mcx also now calls ifmedia_init_with_lock() because it needs to also use an adaptive mutex. The mcx driver still needs to be fully converted to NET_MPSAFE.
2020-03-16 02:04:50 +03:00
* - ifnet lock is not needed for unp_core_lock, but if ifnet lock is
add a note about ifnet lock vs usbnet locks.
2019-11-06 10:30:59 +03:00
* involved, it must be taken first
reduce the scope of struct usbnet: - move a large number of members internal to usbnet.c's new "struct usbnet_private". - provide accessors for a few of these - move struct usbnet_cdata into usbnet.c as well, but move bufsz, list count, and xfer flags back out into struct usbnet, and have them set as part of the setup efore usbnet_attach() - split the intr pipe parts into their own structure - move all the main usbnet*lock* code into usbnet.c too usbnet_attach() goes down to 2 args, and the inputs needed are now the full contents of 'struct usbnet' besides the driver owned 'un_flags' and usbnet owned 'un_pri'. welcome netbsd 9.99.6.
2019-08-10 05:17:36 +03:00
*/
Define and implement a locking protocol for the ifmedia / mii layers: - MP-safe drivers provide a mutex to ifmedia that is used to serialize access to media-related structures / hardware regsiters. Converted drivers use the new ifmedia_init_with_lock() function for this. The new name is provided to ease the transition. - Un-converted drivers continue to call ifmedia_init(), which will supply a compatibility lock to be used instead. Several media-related entry points must be aware of this compatibility lock, and are able to acquire it recursively a limited number of times, if needed. This is a SPIN mutex with priority IPL_NET. - This same lock is used to serialize access to PHY registers and other MII-related data structures. The PHY drivers are modified to acquire and release the lock, as needed, and assert the lock is held as a diagnostic aid. The "usbnet" framework has had an overhaul of its internal locking protocols to fit in with the media / mii changes, and the drivers adapted. USB wifi drivers have been changed to provide their own adaptive mutex to the ifmedia later via a new ieee80211_media_init_with_lock() function. This is required because the USB drivers need an adaptive mutex. Besised "usbnet", a few other drivers are converted: vmx, wm, ixgbe / ixv. mcx also now calls ifmedia_init_with_lock() because it needs to also use an adaptive mutex. The mcx driver still needs to be fully converted to NET_MPSAFE.
2020-03-16 02:04:50 +03:00
kmutex_t unp_core_lock;
reduce the scope of struct usbnet: - move a large number of members internal to usbnet.c's new "struct usbnet_private". - provide accessors for a few of these - move struct usbnet_cdata into usbnet.c as well, but move bufsz, list count, and xfer flags back out into struct usbnet, and have them set as part of the setup efore usbnet_attach() - split the intr pipe parts into their own structure - move all the main usbnet*lock* code into usbnet.c too usbnet_attach() goes down to 2 args, and the inputs needed are now the full contents of 'struct usbnet' besides the driver owned 'un_flags' and usbnet owned 'un_pri'. welcome netbsd 9.99.6.
2019-08-10 05:17:36 +03:00
kmutex_t unp_rxlock;
kmutex_t unp_txlock;
usbnet: Apply hardware multicast filter updates synchronously again. To make this work: 1. Do it only under a new lock, unp_mcastlock. This lock lives at IPL_SOFTCLOCK so it can be taken from network stack callouts. It is forbidden to acquire the usbnet core lock under unp_mcastlock. 2. Do it only after usbnet_init_rx_tx and before usbnet_stop; if issued at any other time, drop the update on the floor. 3. Make usbnet_init_rx_tx apply any pending multicast filter updates under the lock before setting the flag that allows SIOCADDMULTI or SIOCDELMULTI to apply the updates. 4. Remove core lock asserts from various drivers' register access routines. This is necessary because the multicast filter updates are done with register reads/writes, but _cannot_ take the core lock when the caller holds softnet_lock. This now programs the hardware multicast filter redundantly in many drivers which already explicitly call *_uno_mcast from the *_uno_init routines. This is probably harmless, but it will likely be better to remove the explicit calls.
2022-03-03 08:53:23 +03:00
kmutex_t unp_mcastlock;
bool unp_mcastactive;
reduce the scope of struct usbnet: - move a large number of members internal to usbnet.c's new "struct usbnet_private". - provide accessors for a few of these - move struct usbnet_cdata into usbnet.c as well, but move bufsz, list count, and xfer flags back out into struct usbnet, and have them set as part of the setup efore usbnet_attach() - split the intr pipe parts into their own structure - move all the main usbnet*lock* code into usbnet.c too usbnet_attach() goes down to 2 args, and the inputs needed are now the full contents of 'struct usbnet' besides the driver owned 'un_flags' and usbnet owned 'un_pri'. welcome netbsd 9.99.6.
2019-08-10 05:17:36 +03:00
struct usbnet_cdata unp_cdata;
struct ethercom unp_ec;
struct mii_data unp_mii;
struct usb_task unp_ticktask;
struct callout unp_stat_ch;
struct usbd_pipe *unp_ep[USBNET_ENDPT_MAX];
usbnet: Use atomic_load/store_relaxed for unp_dying. This way we don't need to hold the core lock to avoid upsetting sanitizers (which probably find the current code upsetting), and we can use it to exit early from timeout loops that run under the core lock (which is probably not necessary for them to do anyway, but let's worry about that later).
2022-03-03 08:50:39 +03:00
volatile bool unp_dying;
reduce the scope of struct usbnet: - move a large number of members internal to usbnet.c's new "struct usbnet_private". - provide accessors for a few of these - move struct usbnet_cdata into usbnet.c as well, but move bufsz, list count, and xfer flags back out into struct usbnet, and have them set as part of the setup efore usbnet_attach() - split the intr pipe parts into their own structure - move all the main usbnet*lock* code into usbnet.c too usbnet_attach() goes down to 2 args, and the inputs needed are now the full contents of 'struct usbnet' besides the driver owned 'un_flags' and usbnet owned 'un_pri'. welcome netbsd 9.99.6.
2019-08-10 05:17:36 +03:00
bool unp_stopping;
bool unp_attached;
usbnet: Ensure ifp->if_softc is initialized _before_ publishing ifp. Otherwise other parts of the system might start using ifp the moment we if_register it, and trip over null if_softc.
2022-03-03 08:47:14 +03:00
bool unp_ifp_attached;
reduce the scope of struct usbnet: - move a large number of members internal to usbnet.c's new "struct usbnet_private". - provide accessors for a few of these - move struct usbnet_cdata into usbnet.c as well, but move bufsz, list count, and xfer flags back out into struct usbnet, and have them set as part of the setup efore usbnet_attach() - split the intr pipe parts into their own structure - move all the main usbnet*lock* code into usbnet.c too usbnet_attach() goes down to 2 args, and the inputs needed are now the full contents of 'struct usbnet' besides the driver owned 'un_flags' and usbnet owned 'un_pri'. welcome netbsd 9.99.6.
2019-08-10 05:17:36 +03:00
bool unp_link;
int unp_timer;
if_flags is neither int nor short. It's unsigned short.
2019-09-13 10:55:05 +03:00
unsigned short unp_if_flags;
- remove unused uncd_rx_cnt - add USBNETHIST_CALLARGSN and use it in many places - add more history logs - add a log if watchdog abort leaves uncd_tx_cnt non zero - add a unique (32-bit) number for a each usbnet device and use it for logging - explicitly check for full tx ring in usbnet_start_locked() and return early, which avoids setting the 5 second timer and triggering a watchdog - reset uncd_tx_prod and uncd_tx_cnt to zero in stop
2019-08-23 07:29:28 +03:00
unsigned unp_number;
reduce the scope of struct usbnet: - move a large number of members internal to usbnet.c's new "struct usbnet_private". - provide accessors for a few of these - move struct usbnet_cdata into usbnet.c as well, but move bufsz, list count, and xfer flags back out into struct usbnet, and have them set as part of the setup efore usbnet_attach() - split the intr pipe parts into their own structure - move all the main usbnet*lock* code into usbnet.c too usbnet_attach() goes down to 2 args, and the inputs needed are now the full contents of 'struct usbnet' besides the driver owned 'un_flags' and usbnet owned 'un_pri'. welcome netbsd 9.99.6.
2019-08-10 05:17:36 +03:00
krndsource_t unp_rndsrc;
struct timeval unp_rx_notice;
struct timeval unp_tx_notice;
struct timeval unp_intr_notice;
};
#define un_cdata(un) (&(un)->un_pri->unp_cdata)
- remove unused uncd_rx_cnt - add USBNETHIST_CALLARGSN and use it in many places - add more history logs - add a log if watchdog abort leaves uncd_tx_cnt non zero - add a unique (32-bit) number for a each usbnet device and use it for logging - explicitly check for full tx ring in usbnet_start_locked() and return early, which avoids setting the 5 second timer and triggering a watchdog - reset uncd_tx_prod and uncd_tx_cnt to zero in stop
2019-08-23 07:29:28 +03:00
volatile unsigned usbnet_number;
usbnet: Make the tx/rx locks private to usbnet.c. Suffice it for the drivers to know that uno_tx_prepare and uno_rx_loop have exclusive access to the chain, and, for tx, exclusive access to the mbuf.
2022-03-03 08:52:20 +03:00
static void usbnet_isowned_rx(struct usbnet *);
static void usbnet_isowned_tx(struct usbnet *);
usbnet: Delete the core lock from the API. Init/stop and ioctl happen under IFNET_LOCK. Multicast updates only happen after init and before stop. Core lock is no longer a relevant part of the API. Internally, it serves essentially just to lock out asynchronous mii activity during init/stop.
2022-03-03 08:54:52 +03:00
static kmutex_t *
usbnet_mutex_core(struct usbnet *un)
{
return &un->un_pri->unp_core_lock;
}
static __inline__ void
usbnet_isowned_core(struct usbnet *un)
{
KASSERT(mutex_owned(usbnet_mutex_core(un)));
}
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
static int usbnet_modcmd(modcmd_t, void *);
couple of minor API updates: - change the read/write register callbacks to have the same phy/reg order as the MII code. - add "mii_flags" param to usbnet_attach_ifp(). axe(4) wants it. also: - add usbnet debug code, sysctl node support - remove commented DPRINTF()s accidentally left in place - add usbnet_softc() - reorder some attach code to be consistent - re-add USBD_FORCE_SHORT_XFER for axen rx chain ride 9.99.2 bump.
2019-08-01 02:47:16 +03:00
#ifdef USB_DEBUG
#ifndef USBNET_DEBUG
#define usbnetdebug 0
#else
- use CALLARGS vs CALLED for better usbhist - turn off usbnetdebug default - log for all entry/exit points of usbnet_pipe_intr() - in usbnet_start_locked() track whether any packet has been transmitted for setting the timer. avoids spurious "watchdog timeouts" - in usbnet_stop() use callout_halt() vs callout_halt, and also stop the usb task. fixes crash of usbtask after the phy has detached. - add a little more defensive checking in the tick task, and add some high-log-level logs. - in usbnet_detach() move the call to usbnet_stop_ifp() above the calls to callout/usbtask stopping. - set ec_mii and unp_pri to NULL when freeing their data
2019-09-08 21:59:32 +03:00
static int usbnetdebug = 0;
- adjust usbnet interface to allow usbd_open_pipe_intr(), from the new comment: * if un_intr_buf is not NULL, use usbd_open_pipe_intr() not * usbd_open_pipe() for USBNET_ENDPT_INTR, with this buffer, * size, and interval. the standard handling is in usbnet.c, with a callback to deal with the interrupt it self. not fully tested, designed for if_aue.c and a few others not yet converted. - make usbhist for usbnet.c work, thanks paulg - usbnet_init_rx_tx() clears out all allocations upon failure now - add usbnet_ec() to get a pointer to the struct ethercom - add usbnet_{lock,unlock,owned}*() to lock/unlock the various locks and *owned*() for asserting welcome 9.99.3!
2019-08-04 11:59:13 +03:00
couple of minor API updates: - change the read/write register callbacks to have the same phy/reg order as the MII code. - add "mii_flags" param to usbnet_attach_ifp(). axe(4) wants it. also: - add usbnet debug code, sysctl node support - remove commented DPRINTF()s accidentally left in place - add usbnet_softc() - reorder some attach code to be consistent - re-add USBD_FORCE_SHORT_XFER for axen rx chain ride 9.99.2 bump.
2019-08-01 02:47:16 +03:00
SYSCTL_SETUP(sysctl_hw_usbnet_setup, "sysctl hw.usbnet setup")
{
int err;
const struct sysctlnode *rnode;
const struct sysctlnode *cnode;
err = sysctl_createv(clog, 0, NULL, &rnode,
CTLFLAG_PERMANENT, CTLTYPE_NODE, "usbnet",
SYSCTL_DESCR("usbnet global controls"),
NULL, 0, NULL, 0, CTL_HW, CTL_CREATE, CTL_EOL);
if (err)
goto fail;
/* control debugging printfs */
err = sysctl_createv(clog, 0, &rnode, &cnode,
CTLFLAG_PERMANENT | CTLFLAG_READWRITE, CTLTYPE_INT,
"debug", SYSCTL_DESCR("Enable debugging output"),
NULL, 0, &usbnetdebug, sizeof(usbnetdebug), CTL_CREATE, CTL_EOL);
if (err)
goto fail;
return;
fail:
aprint_error("%s: sysctl_createv failed (err = %d)\n", __func__, err);
}
#endif /* USBNET_DEBUG */
#endif /* USB_DEBUG */
#define DPRINTF(FMT,A,B,C,D) USBHIST_LOGN(usbnetdebug,1,FMT,A,B,C,D)
#define DPRINTFN(N,FMT,A,B,C,D) USBHIST_LOGN(usbnetdebug,N,FMT,A,B,C,D)
#define USBNETHIST_FUNC() USBHIST_FUNC()
#define USBNETHIST_CALLED(name) USBHIST_CALLED(usbnetdebug)
add USBHIST_CALLARGS() frontend to KERNHIST_CALLARGS(). add USBNETHIST_CALLARGS() frontend to USBHIST_CALLARGS(). use both in read/write reg, instead of aprint. use %jx and (uintptr_t) and fix the 32 bit debug build.
2019-08-19 09:35:14 +03:00
#define USBNETHIST_CALLARGS(FMT,A,B,C,D) \
USBHIST_CALLARGS(usbnetdebug,FMT,A,B,C,D)
- remove unused uncd_rx_cnt - add USBNETHIST_CALLARGSN and use it in many places - add more history logs - add a log if watchdog abort leaves uncd_tx_cnt non zero - add a unique (32-bit) number for a each usbnet device and use it for logging - explicitly check for full tx ring in usbnet_start_locked() and return early, which avoids setting the 5 second timer and triggering a watchdog - reset uncd_tx_prod and uncd_tx_cnt to zero in stop
2019-08-23 07:29:28 +03:00
#define USBNETHIST_CALLARGSN(N,FMT,A,B,C,D) \
USBHIST_CALLARGSN(usbnetdebug,N,FMT,A,B,C,D)
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
update usbnet some: - move rx/tx xfer flags into usbnet_cdata - move the callbacks into usbnet_ops structure - move rx/tx xfer flags arguments from usbnet_init_rx_tx() and move them all into usbnet_attach() arguments - s/miibus/mii/ in some places for consistency other clean up: - create wrapper functions for callbacks, move knowledge about special handling (OK to be missing, error eating) there. - use cdata pointer if already available - provide some more macros (will be real functions later) for accessing usbnet members, use existing ones more bump kernel version.
2019-08-09 04:17:33 +03:00
/* Callback vectors. */
static void
uno_stop(struct usbnet *un, struct ifnet *ifp, int disable)
{
usbnet: Assert IFNET_LOCKED in usbnet_init_rx_tx, usbnet_stop. Exception: urndis(4) abuses this API to start this logic before the ifp is actually initialized. So for the sake of urndis(4), until sense can be beaten into it, allow the !unp_ifp_attached case to run without IFNET_LOCK.
2022-03-03 08:47:36 +03:00
KASSERTMSG(!un->un_pri->unp_ifp_attached || IFNET_LOCKED(ifp),
"%s", ifp->if_xname);
Define and implement a locking protocol for the ifmedia / mii layers: - MP-safe drivers provide a mutex to ifmedia that is used to serialize access to media-related structures / hardware regsiters. Converted drivers use the new ifmedia_init_with_lock() function for this. The new name is provided to ease the transition. - Un-converted drivers continue to call ifmedia_init(), which will supply a compatibility lock to be used instead. Several media-related entry points must be aware of this compatibility lock, and are able to acquire it recursively a limited number of times, if needed. This is a SPIN mutex with priority IPL_NET. - This same lock is used to serialize access to PHY registers and other MII-related data structures. The PHY drivers are modified to acquire and release the lock, as needed, and assert the lock is held as a diagnostic aid. The "usbnet" framework has had an overhaul of its internal locking protocols to fit in with the media / mii changes, and the drivers adapted. USB wifi drivers have been changed to provide their own adaptive mutex to the ifmedia later via a new ieee80211_media_init_with_lock() function. This is required because the USB drivers need an adaptive mutex. Besised "usbnet", a few other drivers are converted: vmx, wm, ixgbe / ixv. mcx also now calls ifmedia_init_with_lock() because it needs to also use an adaptive mutex. The mcx driver still needs to be fully converted to NET_MPSAFE.
2020-03-16 02:04:50 +03:00
usbnet_isowned_core(un);
update usbnet some: - move rx/tx xfer flags into usbnet_cdata - move the callbacks into usbnet_ops structure - move rx/tx xfer flags arguments from usbnet_init_rx_tx() and move them all into usbnet_attach() arguments - s/miibus/mii/ in some places for consistency other clean up: - create wrapper functions for callbacks, move knowledge about special handling (OK to be missing, error eating) there. - use cdata pointer if already available - provide some more macros (will be real functions later) for accessing usbnet members, use existing ones more bump kernel version.
2019-08-09 04:17:33 +03:00
if (un->un_ops->uno_stop)
(*un->un_ops->uno_stop)(ifp, disable);
}
static int
uno_ioctl(struct usbnet *un, struct ifnet *ifp, u_long cmd, void *data)
{
usbnet: Assert ioctl locking.
2022-03-03 08:50:12 +03:00
usbnet: Split multicast filter reprogramming into separate operation.
2022-03-03 08:51:06 +03:00
KASSERTMSG(cmd != SIOCADDMULTI, "%s", ifp->if_xname);
KASSERTMSG(cmd != SIOCDELMULTI, "%s", ifp->if_xname);
KASSERTMSG(IFNET_LOCKED(ifp), "%s", ifp->if_xname);
usbnet: Assert ioctl locking.
2022-03-03 08:50:12 +03:00
update usbnet some: - move rx/tx xfer flags into usbnet_cdata - move the callbacks into usbnet_ops structure - move rx/tx xfer flags arguments from usbnet_init_rx_tx() and move them all into usbnet_attach() arguments - s/miibus/mii/ in some places for consistency other clean up: - create wrapper functions for callbacks, move knowledge about special handling (OK to be missing, error eating) there. - use cdata pointer if already available - provide some more macros (will be real functions later) for accessing usbnet members, use existing ones more bump kernel version.
2019-08-09 04:17:33 +03:00
if (un->un_ops->uno_ioctl)
return (*un->un_ops->uno_ioctl)(ifp, cmd, data);
return 0;
}
static int
uno_override_ioctl(struct usbnet *un, struct ifnet *ifp, u_long cmd, void *data)
{
usbnet: Assert ioctl locking.
2022-03-03 08:50:12 +03:00
switch (cmd) {
case SIOCADDMULTI:
case SIOCDELMULTI:
break;
default:
KASSERTMSG(IFNET_LOCKED(ifp), "%s", ifp->if_xname);
}
update usbnet some: - move rx/tx xfer flags into usbnet_cdata - move the callbacks into usbnet_ops structure - move rx/tx xfer flags arguments from usbnet_init_rx_tx() and move them all into usbnet_attach() arguments - s/miibus/mii/ in some places for consistency other clean up: - create wrapper functions for callbacks, move knowledge about special handling (OK to be missing, error eating) there. - use cdata pointer if already available - provide some more macros (will be real functions later) for accessing usbnet members, use existing ones more bump kernel version.
2019-08-09 04:17:33 +03:00
return (*un->un_ops->uno_override_ioctl)(ifp, cmd, data);
}
static int
uno_init(struct usbnet *un, struct ifnet *ifp)
{
usbnet: Impart blame on whose ifnet is unlocked in uno_init.
2022-03-03 08:50:05 +03:00
KASSERTMSG(IFNET_LOCKED(ifp), "%s", ifp->if_xname);
update usbnet some: - move rx/tx xfer flags into usbnet_cdata - move the callbacks into usbnet_ops structure - move rx/tx xfer flags arguments from usbnet_init_rx_tx() and move them all into usbnet_attach() arguments - s/miibus/mii/ in some places for consistency other clean up: - create wrapper functions for callbacks, move knowledge about special handling (OK to be missing, error eating) there. - use cdata pointer if already available - provide some more macros (will be real functions later) for accessing usbnet members, use existing ones more bump kernel version.
2019-08-09 04:17:33 +03:00
return (*un->un_ops->uno_init)(ifp);
}
static int
uno_read_reg(struct usbnet *un, int phy, int reg, uint16_t *val)
{
Define and implement a locking protocol for the ifmedia / mii layers: - MP-safe drivers provide a mutex to ifmedia that is used to serialize access to media-related structures / hardware regsiters. Converted drivers use the new ifmedia_init_with_lock() function for this. The new name is provided to ease the transition. - Un-converted drivers continue to call ifmedia_init(), which will supply a compatibility lock to be used instead. Several media-related entry points must be aware of this compatibility lock, and are able to acquire it recursively a limited number of times, if needed. This is a SPIN mutex with priority IPL_NET. - This same lock is used to serialize access to PHY registers and other MII-related data structures. The PHY drivers are modified to acquire and release the lock, as needed, and assert the lock is held as a diagnostic aid. The "usbnet" framework has had an overhaul of its internal locking protocols to fit in with the media / mii changes, and the drivers adapted. USB wifi drivers have been changed to provide their own adaptive mutex to the ifmedia later via a new ieee80211_media_init_with_lock() function. This is required because the USB drivers need an adaptive mutex. Besised "usbnet", a few other drivers are converted: vmx, wm, ixgbe / ixv. mcx also now calls ifmedia_init_with_lock() because it needs to also use an adaptive mutex. The mcx driver still needs to be fully converted to NET_MPSAFE.
2020-03-16 02:04:50 +03:00
usbnet_isowned_core(un);
update usbnet some: - move rx/tx xfer flags into usbnet_cdata - move the callbacks into usbnet_ops structure - move rx/tx xfer flags arguments from usbnet_init_rx_tx() and move them all into usbnet_attach() arguments - s/miibus/mii/ in some places for consistency other clean up: - create wrapper functions for callbacks, move knowledge about special handling (OK to be missing, error eating) there. - use cdata pointer if already available - provide some more macros (will be real functions later) for accessing usbnet members, use existing ones more bump kernel version.
2019-08-09 04:17:33 +03:00
return (*un->un_ops->uno_read_reg)(un, phy, reg, val);
}
static int
uno_write_reg(struct usbnet *un, int phy, int reg, uint16_t val)
{
Define and implement a locking protocol for the ifmedia / mii layers: - MP-safe drivers provide a mutex to ifmedia that is used to serialize access to media-related structures / hardware regsiters. Converted drivers use the new ifmedia_init_with_lock() function for this. The new name is provided to ease the transition. - Un-converted drivers continue to call ifmedia_init(), which will supply a compatibility lock to be used instead. Several media-related entry points must be aware of this compatibility lock, and are able to acquire it recursively a limited number of times, if needed. This is a SPIN mutex with priority IPL_NET. - This same lock is used to serialize access to PHY registers and other MII-related data structures. The PHY drivers are modified to acquire and release the lock, as needed, and assert the lock is held as a diagnostic aid. The "usbnet" framework has had an overhaul of its internal locking protocols to fit in with the media / mii changes, and the drivers adapted. USB wifi drivers have been changed to provide their own adaptive mutex to the ifmedia later via a new ieee80211_media_init_with_lock() function. This is required because the USB drivers need an adaptive mutex. Besised "usbnet", a few other drivers are converted: vmx, wm, ixgbe / ixv. mcx also now calls ifmedia_init_with_lock() because it needs to also use an adaptive mutex. The mcx driver still needs to be fully converted to NET_MPSAFE.
2020-03-16 02:04:50 +03:00
usbnet_isowned_core(un);
update usbnet some: - move rx/tx xfer flags into usbnet_cdata - move the callbacks into usbnet_ops structure - move rx/tx xfer flags arguments from usbnet_init_rx_tx() and move them all into usbnet_attach() arguments - s/miibus/mii/ in some places for consistency other clean up: - create wrapper functions for callbacks, move knowledge about special handling (OK to be missing, error eating) there. - use cdata pointer if already available - provide some more macros (will be real functions later) for accessing usbnet members, use existing ones more bump kernel version.
2019-08-09 04:17:33 +03:00
return (*un->un_ops->uno_write_reg)(un, phy, reg, val);
}
static void
uno_mii_statchg(struct usbnet *un, struct ifnet *ifp)
{
Define and implement a locking protocol for the ifmedia / mii layers: - MP-safe drivers provide a mutex to ifmedia that is used to serialize access to media-related structures / hardware regsiters. Converted drivers use the new ifmedia_init_with_lock() function for this. The new name is provided to ease the transition. - Un-converted drivers continue to call ifmedia_init(), which will supply a compatibility lock to be used instead. Several media-related entry points must be aware of this compatibility lock, and are able to acquire it recursively a limited number of times, if needed. This is a SPIN mutex with priority IPL_NET. - This same lock is used to serialize access to PHY registers and other MII-related data structures. The PHY drivers are modified to acquire and release the lock, as needed, and assert the lock is held as a diagnostic aid. The "usbnet" framework has had an overhaul of its internal locking protocols to fit in with the media / mii changes, and the drivers adapted. USB wifi drivers have been changed to provide their own adaptive mutex to the ifmedia later via a new ieee80211_media_init_with_lock() function. This is required because the USB drivers need an adaptive mutex. Besised "usbnet", a few other drivers are converted: vmx, wm, ixgbe / ixv. mcx also now calls ifmedia_init_with_lock() because it needs to also use an adaptive mutex. The mcx driver still needs to be fully converted to NET_MPSAFE.
2020-03-16 02:04:50 +03:00
usbnet_isowned_core(un);
update usbnet some: - move rx/tx xfer flags into usbnet_cdata - move the callbacks into usbnet_ops structure - move rx/tx xfer flags arguments from usbnet_init_rx_tx() and move them all into usbnet_attach() arguments - s/miibus/mii/ in some places for consistency other clean up: - create wrapper functions for callbacks, move knowledge about special handling (OK to be missing, error eating) there. - use cdata pointer if already available - provide some more macros (will be real functions later) for accessing usbnet members, use existing ones more bump kernel version.
2019-08-09 04:17:33 +03:00
(*un->un_ops->uno_statchg)(ifp);
}
static unsigned
uno_tx_prepare(struct usbnet *un, struct mbuf *m, struct usbnet_chain *c)
{
Define and implement a locking protocol for the ifmedia / mii layers: - MP-safe drivers provide a mutex to ifmedia that is used to serialize access to media-related structures / hardware regsiters. Converted drivers use the new ifmedia_init_with_lock() function for this. The new name is provided to ease the transition. - Un-converted drivers continue to call ifmedia_init(), which will supply a compatibility lock to be used instead. Several media-related entry points must be aware of this compatibility lock, and are able to acquire it recursively a limited number of times, if needed. This is a SPIN mutex with priority IPL_NET. - This same lock is used to serialize access to PHY registers and other MII-related data structures. The PHY drivers are modified to acquire and release the lock, as needed, and assert the lock is held as a diagnostic aid. The "usbnet" framework has had an overhaul of its internal locking protocols to fit in with the media / mii changes, and the drivers adapted. USB wifi drivers have been changed to provide their own adaptive mutex to the ifmedia later via a new ieee80211_media_init_with_lock() function. This is required because the USB drivers need an adaptive mutex. Besised "usbnet", a few other drivers are converted: vmx, wm, ixgbe / ixv. mcx also now calls ifmedia_init_with_lock() because it needs to also use an adaptive mutex. The mcx driver still needs to be fully converted to NET_MPSAFE.
2020-03-16 02:04:50 +03:00
usbnet_isowned_tx(un);
update usbnet some: - move rx/tx xfer flags into usbnet_cdata - move the callbacks into usbnet_ops structure - move rx/tx xfer flags arguments from usbnet_init_rx_tx() and move them all into usbnet_attach() arguments - s/miibus/mii/ in some places for consistency other clean up: - create wrapper functions for callbacks, move knowledge about special handling (OK to be missing, error eating) there. - use cdata pointer if already available - provide some more macros (will be real functions later) for accessing usbnet members, use existing ones more bump kernel version.
2019-08-09 04:17:33 +03:00
return (*un->un_ops->uno_tx_prepare)(un, m, c);
}
static void
- usbnet_rx_loop_cb's usbd_xfer parameter is never used and available in the usbnet_chain if needed. remove it - usbnet media status change already set link to false, don't repeat this in every driver - don't clear link in stop, nothing was re-enabling it for non-MII - add optional uno_tick_cb(struct usbnet *un) that is called from the usbnet tick timer - remove wrong debug sysctl prototype rx_loop and timer are kernel versions changes, but hopefully this is the last one for usbnet. working with 3 more drivers now (cue, mue and url), leaving only aue, kue, upl and umb undone (aue may work with previously supported devices, mine doesn't work with our driver, kue and upl have patches for testing and umb is undone.)
2019-08-15 08:52:23 +03:00
uno_rx_loop(struct usbnet *un, struct usbnet_chain *c, uint32_t total_len)
{
Define and implement a locking protocol for the ifmedia / mii layers: - MP-safe drivers provide a mutex to ifmedia that is used to serialize access to media-related structures / hardware regsiters. Converted drivers use the new ifmedia_init_with_lock() function for this. The new name is provided to ease the transition. - Un-converted drivers continue to call ifmedia_init(), which will supply a compatibility lock to be used instead. Several media-related entry points must be aware of this compatibility lock, and are able to acquire it recursively a limited number of times, if needed. This is a SPIN mutex with priority IPL_NET. - This same lock is used to serialize access to PHY registers and other MII-related data structures. The PHY drivers are modified to acquire and release the lock, as needed, and assert the lock is held as a diagnostic aid. The "usbnet" framework has had an overhaul of its internal locking protocols to fit in with the media / mii changes, and the drivers adapted. USB wifi drivers have been changed to provide their own adaptive mutex to the ifmedia later via a new ieee80211_media_init_with_lock() function. This is required because the USB drivers need an adaptive mutex. Besised "usbnet", a few other drivers are converted: vmx, wm, ixgbe / ixv. mcx also now calls ifmedia_init_with_lock() because it needs to also use an adaptive mutex. The mcx driver still needs to be fully converted to NET_MPSAFE.
2020-03-16 02:04:50 +03:00
usbnet_isowned_rx(un);
- usbnet_rx_loop_cb's usbd_xfer parameter is never used and available in the usbnet_chain if needed. remove it - usbnet media status change already set link to false, don't repeat this in every driver - don't clear link in stop, nothing was re-enabling it for non-MII - add optional uno_tick_cb(struct usbnet *un) that is called from the usbnet tick timer - remove wrong debug sysctl prototype rx_loop and timer are kernel versions changes, but hopefully this is the last one for usbnet. working with 3 more drivers now (cue, mue and url), leaving only aue, kue, upl and umb undone (aue may work with previously supported devices, mine doesn't work with our driver, kue and upl have patches for testing and umb is undone.)
2019-08-15 08:52:23 +03:00
(*un->un_ops->uno_rx_loop)(un, c, total_len);
}
static void
uno_tick(struct usbnet *un)
update usbnet some: - move rx/tx xfer flags into usbnet_cdata - move the callbacks into usbnet_ops structure - move rx/tx xfer flags arguments from usbnet_init_rx_tx() and move them all into usbnet_attach() arguments - s/miibus/mii/ in some places for consistency other clean up: - create wrapper functions for callbacks, move knowledge about special handling (OK to be missing, error eating) there. - use cdata pointer if already available - provide some more macros (will be real functions later) for accessing usbnet members, use existing ones more bump kernel version.
2019-08-09 04:17:33 +03:00
{
- usbnet_rx_loop_cb's usbd_xfer parameter is never used and available in the usbnet_chain if needed. remove it - usbnet media status change already set link to false, don't repeat this in every driver - don't clear link in stop, nothing was re-enabling it for non-MII - add optional uno_tick_cb(struct usbnet *un) that is called from the usbnet tick timer - remove wrong debug sysctl prototype rx_loop and timer are kernel versions changes, but hopefully this is the last one for usbnet. working with 3 more drivers now (cue, mue and url), leaving only aue, kue, upl and umb undone (aue may work with previously supported devices, mine doesn't work with our driver, kue and upl have patches for testing and umb is undone.)
2019-08-15 08:52:23 +03:00
if (un->un_ops->uno_tick)
(*un->un_ops->uno_tick)(un);
update usbnet some: - move rx/tx xfer flags into usbnet_cdata - move the callbacks into usbnet_ops structure - move rx/tx xfer flags arguments from usbnet_init_rx_tx() and move them all into usbnet_attach() arguments - s/miibus/mii/ in some places for consistency other clean up: - create wrapper functions for callbacks, move knowledge about special handling (OK to be missing, error eating) there. - use cdata pointer if already available - provide some more macros (will be real functions later) for accessing usbnet members, use existing ones more bump kernel version.
2019-08-09 04:17:33 +03:00
}
static void
uno_intr(struct usbnet *un, usbd_status status)
{
if (un->un_ops->uno_intr)
(*un->un_ops->uno_intr)(un, status);
}
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
/* Interrupt handling. */
static struct mbuf *
internal to usbnet: add buflen param to usbnet_newbuf(). use this to skip allocating an mbuf cluster for small packets (ported from kue(4).) remove usbnet_rx_start_pipes()'s always usbnet_rxeof argument.
2019-08-16 11:38:21 +03:00
usbnet_newbuf(size_t buflen)
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
{
struct mbuf *m;
usbnet: Reject buflen>MCLBYTES in usbnet_newbuf.
2020-08-28 20:05:32 +03:00
if (buflen > MCLBYTES)
return NULL;
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
MGETHDR(m, M_DONTWAIT, MT_DATA);
if (m == NULL)
return NULL;
internal to usbnet: add buflen param to usbnet_newbuf(). use this to skip allocating an mbuf cluster for small packets (ported from kue(4).) remove usbnet_rx_start_pipes()'s always usbnet_rxeof argument.
2019-08-16 11:38:21 +03:00
if (buflen > MHLEN - ETHER_ALIGN) {
MCLGET(m, M_DONTWAIT);
if (!(m->m_flags & M_EXT)) {
m_freem(m);
return NULL;
}
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
}
m_adj(m, ETHER_ALIGN);
internal to usbnet: add buflen param to usbnet_newbuf(). use this to skip allocating an mbuf cluster for small packets (ported from kue(4).) remove usbnet_rx_start_pipes()'s always usbnet_rxeof argument.
2019-08-16 11:38:21 +03:00
m->m_len = m->m_pkthdr.len = buflen;
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
return m;
}
/*
* usbnet_rxeof() is designed to be the done callback for rx completion.
* it provides generic setup and finalisation, calls a different usbnet
* rx_loop callback in the middle, which can use usbnet_enqueue() to
extend usbnet to cope with if_upl, if_smsc, and if_umb needs: - usbnet_enqueue() can set mbuf flags and csum_data - usbnet_input() for non-ethernet based devices (upl, umb) - allow a complete override for ioctl() - remove converted list -- we have compiling and/or working patches for all the devices except for umb(4), will be merged as testing happens hopefully this is the last ABI change, though it may end up being extended for additional smsc(4) support. hello for real netbsd 9.99.3!
2019-08-06 03:19:57 +03:00
* enqueue a packet for higher levels (or usbnet_input() if previously
* using if_input() path.)
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
*/
void
usbnet_enqueue(struct usbnet * const un, uint8_t *buf, size_t buflen,
extend usbnet to cope with if_upl, if_smsc, and if_umb needs: - usbnet_enqueue() can set mbuf flags and csum_data - usbnet_input() for non-ethernet based devices (upl, umb) - allow a complete override for ioctl() - remove converted list -- we have compiling and/or working patches for all the devices except for umb(4), will be merged as testing happens hopefully this is the last ABI change, though it may end up being extended for additional smsc(4) support. hello for real netbsd 9.99.3!
2019-08-06 03:19:57 +03:00
int csum_flags, uint32_t csum_data, int mbuf_flags)
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
{
- remove unused uncd_rx_cnt - add USBNETHIST_CALLARGSN and use it in many places - add more history logs - add a log if watchdog abort leaves uncd_tx_cnt non zero - add a unique (32-bit) number for a each usbnet device and use it for logging - explicitly check for full tx ring in usbnet_start_locked() and return early, which avoids setting the 5 second timer and triggering a watchdog - reset uncd_tx_prod and uncd_tx_cnt to zero in stop
2019-08-23 07:29:28 +03:00
USBNETHIST_FUNC();
reduce the scope of struct usbnet: - move a large number of members internal to usbnet.c's new "struct usbnet_private". - provide accessors for a few of these - move struct usbnet_cdata into usbnet.c as well, but move bufsz, list count, and xfer flags back out into struct usbnet, and have them set as part of the setup efore usbnet_attach() - split the intr pipe parts into their own structure - move all the main usbnet*lock* code into usbnet.c too usbnet_attach() goes down to 2 args, and the inputs needed are now the full contents of 'struct usbnet' besides the driver owned 'un_flags' and usbnet owned 'un_pri'. welcome netbsd 9.99.6.
2019-08-10 05:17:36 +03:00
struct ifnet * const ifp = usbnet_ifp(un);
- remove unused uncd_rx_cnt - add USBNETHIST_CALLARGSN and use it in many places - add more history logs - add a log if watchdog abort leaves uncd_tx_cnt non zero - add a unique (32-bit) number for a each usbnet device and use it for logging - explicitly check for full tx ring in usbnet_start_locked() and return early, which avoids setting the 5 second timer and triggering a watchdog - reset uncd_tx_prod and uncd_tx_cnt to zero in stop
2019-08-23 07:29:28 +03:00
struct usbnet_private * const unp __unused = un->un_pri;
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
struct mbuf *m;
PR/55068: sc.dying: Fix printf formats: - no %s/%p for kernel log - 0x% -> %# - always %j for kernel log
2020-03-13 21:17:40 +03:00
USBNETHIST_CALLARGSN(5, "%jd: enter: len=%ju csf %#jx mbf %#jx",
- remove unused uncd_rx_cnt - add USBNETHIST_CALLARGSN and use it in many places - add more history logs - add a log if watchdog abort leaves uncd_tx_cnt non zero - add a unique (32-bit) number for a each usbnet device and use it for logging - explicitly check for full tx ring in usbnet_start_locked() and return early, which avoids setting the 5 second timer and triggering a watchdog - reset uncd_tx_prod and uncd_tx_cnt to zero in stop
2019-08-23 07:29:28 +03:00
unp->unp_number, buflen, csum_flags, mbuf_flags);
- use usbnet_isowned*() more - more const
2019-08-11 04:31:19 +03:00
usbnet_isowned_rx(un);
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
internal to usbnet: add buflen param to usbnet_newbuf(). use this to skip allocating an mbuf cluster for small packets (ported from kue(4).) remove usbnet_rx_start_pipes()'s always usbnet_rxeof argument.
2019-08-16 11:38:21 +03:00
m = usbnet_newbuf(buflen);
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
if (m == NULL) {
PR/55068: sc.dying: Fix printf formats: - no %s/%p for kernel log - 0x% -> %# - always %j for kernel log
2020-03-13 21:17:40 +03:00
DPRINTF("%jd: no memory", unp->unp_number, 0, 0, 0);
Adopt <net/if_stats.h>.
2020-01-29 09:35:28 +03:00
if_statinc(ifp, if_ierrors);
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
return;
}
m_set_rcvif(m, ifp);
extend usbnet to cope with if_upl, if_smsc, and if_umb needs: - usbnet_enqueue() can set mbuf flags and csum_data - usbnet_input() for non-ethernet based devices (upl, umb) - allow a complete override for ioctl() - remove converted list -- we have compiling and/or working patches for all the devices except for umb(4), will be merged as testing happens hopefully this is the last ABI change, though it may end up being extended for additional smsc(4) support. hello for real netbsd 9.99.3!
2019-08-06 03:19:57 +03:00
m->m_pkthdr.csum_flags = csum_flags;
m->m_pkthdr.csum_data = csum_data;
m->m_flags |= mbuf_flags;
internal to usbnet: add buflen param to usbnet_newbuf(). use this to skip allocating an mbuf cluster for small packets (ported from kue(4).) remove usbnet_rx_start_pipes()'s always usbnet_rxeof argument.
2019-08-16 11:38:21 +03:00
memcpy(mtod(m, uint8_t *), buf, buflen);
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
/* push the packet up */
if_percpuq_enqueue(ifp->if_percpuq, m);
}
extend usbnet to cope with if_upl, if_smsc, and if_umb needs: - usbnet_enqueue() can set mbuf flags and csum_data - usbnet_input() for non-ethernet based devices (upl, umb) - allow a complete override for ioctl() - remove converted list -- we have compiling and/or working patches for all the devices except for umb(4), will be merged as testing happens hopefully this is the last ABI change, though it may end up being extended for additional smsc(4) support. hello for real netbsd 9.99.3!
2019-08-06 03:19:57 +03:00
void
usbnet_input(struct usbnet * const un, uint8_t *buf, size_t buflen)
{
- remove unused uncd_rx_cnt - add USBNETHIST_CALLARGSN and use it in many places - add more history logs - add a log if watchdog abort leaves uncd_tx_cnt non zero - add a unique (32-bit) number for a each usbnet device and use it for logging - explicitly check for full tx ring in usbnet_start_locked() and return early, which avoids setting the 5 second timer and triggering a watchdog - reset uncd_tx_prod and uncd_tx_cnt to zero in stop
2019-08-23 07:29:28 +03:00
USBNETHIST_FUNC();
extend usbnet to cope with if_upl, if_smsc, and if_umb needs: - usbnet_enqueue() can set mbuf flags and csum_data - usbnet_input() for non-ethernet based devices (upl, umb) - allow a complete override for ioctl() - remove converted list -- we have compiling and/or working patches for all the devices except for umb(4), will be merged as testing happens hopefully this is the last ABI change, though it may end up being extended for additional smsc(4) support. hello for real netbsd 9.99.3!
2019-08-06 03:19:57 +03:00
struct ifnet * const ifp = usbnet_ifp(un);
- remove unused uncd_rx_cnt - add USBNETHIST_CALLARGSN and use it in many places - add more history logs - add a log if watchdog abort leaves uncd_tx_cnt non zero - add a unique (32-bit) number for a each usbnet device and use it for logging - explicitly check for full tx ring in usbnet_start_locked() and return early, which avoids setting the 5 second timer and triggering a watchdog - reset uncd_tx_prod and uncd_tx_cnt to zero in stop
2019-08-23 07:29:28 +03:00
struct usbnet_private * const unp __unused = un->un_pri;
extend usbnet to cope with if_upl, if_smsc, and if_umb needs: - usbnet_enqueue() can set mbuf flags and csum_data - usbnet_input() for non-ethernet based devices (upl, umb) - allow a complete override for ioctl() - remove converted list -- we have compiling and/or working patches for all the devices except for umb(4), will be merged as testing happens hopefully this is the last ABI change, though it may end up being extended for additional smsc(4) support. hello for real netbsd 9.99.3!
2019-08-06 03:19:57 +03:00
struct mbuf *m;
PR/55068: sc.dying: Fix printf formats: - no %s/%p for kernel log - 0x% -> %# - always %j for kernel log
2020-03-13 21:17:40 +03:00
USBNETHIST_CALLARGSN(5, "%jd: enter: buf %#jx len %ju",
- remove unused uncd_rx_cnt - add USBNETHIST_CALLARGSN and use it in many places - add more history logs - add a log if watchdog abort leaves uncd_tx_cnt non zero - add a unique (32-bit) number for a each usbnet device and use it for logging - explicitly check for full tx ring in usbnet_start_locked() and return early, which avoids setting the 5 second timer and triggering a watchdog - reset uncd_tx_prod and uncd_tx_cnt to zero in stop
2019-08-23 07:29:28 +03:00
unp->unp_number, (uintptr_t)buf, buflen, 0);
- use usbnet_isowned*() more - more const
2019-08-11 04:31:19 +03:00
usbnet_isowned_rx(un);
extend usbnet to cope with if_upl, if_smsc, and if_umb needs: - usbnet_enqueue() can set mbuf flags and csum_data - usbnet_input() for non-ethernet based devices (upl, umb) - allow a complete override for ioctl() - remove converted list -- we have compiling and/or working patches for all the devices except for umb(4), will be merged as testing happens hopefully this is the last ABI change, though it may end up being extended for additional smsc(4) support. hello for real netbsd 9.99.3!
2019-08-06 03:19:57 +03:00
internal to usbnet: add buflen param to usbnet_newbuf(). use this to skip allocating an mbuf cluster for small packets (ported from kue(4).) remove usbnet_rx_start_pipes()'s always usbnet_rxeof argument.
2019-08-16 11:38:21 +03:00
m = usbnet_newbuf(buflen);
extend usbnet to cope with if_upl, if_smsc, and if_umb needs: - usbnet_enqueue() can set mbuf flags and csum_data - usbnet_input() for non-ethernet based devices (upl, umb) - allow a complete override for ioctl() - remove converted list -- we have compiling and/or working patches for all the devices except for umb(4), will be merged as testing happens hopefully this is the last ABI change, though it may end up being extended for additional smsc(4) support. hello for real netbsd 9.99.3!
2019-08-06 03:19:57 +03:00
if (m == NULL) {
Adopt <net/if_stats.h>.
2020-01-29 09:35:28 +03:00
if_statinc(ifp, if_ierrors);
extend usbnet to cope with if_upl, if_smsc, and if_umb needs: - usbnet_enqueue() can set mbuf flags and csum_data - usbnet_input() for non-ethernet based devices (upl, umb) - allow a complete override for ioctl() - remove converted list -- we have compiling and/or working patches for all the devices except for umb(4), will be merged as testing happens hopefully this is the last ABI change, though it may end up being extended for additional smsc(4) support. hello for real netbsd 9.99.3!
2019-08-06 03:19:57 +03:00
return;
}
m_set_rcvif(m, ifp);
memcpy(mtod(m, char *), buf, buflen);
/* push the packet up */
if_input(ifp, m);
}
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
/*
* A frame has been uploaded: pass the resulting mbuf chain up to
* the higher level protocols.
*/
static void
- adjust usbnet interface to allow usbd_open_pipe_intr(), from the new comment: * if un_intr_buf is not NULL, use usbd_open_pipe_intr() not * usbd_open_pipe() for USBNET_ENDPT_INTR, with this buffer, * size, and interval. the standard handling is in usbnet.c, with a callback to deal with the interrupt it self. not fully tested, designed for if_aue.c and a few others not yet converted. - make usbhist for usbnet.c work, thanks paulg - usbnet_init_rx_tx() clears out all allocations upon failure now - add usbnet_ec() to get a pointer to the struct ethercom - add usbnet_{lock,unlock,owned}*() to lock/unlock the various locks and *owned*() for asserting welcome 9.99.3!
2019-08-04 11:59:13 +03:00
usbnet_rxeof(struct usbd_xfer *xfer, void *priv, usbd_status status)
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
{
- remove unused uncd_rx_cnt - add USBNETHIST_CALLARGSN and use it in many places - add more history logs - add a log if watchdog abort leaves uncd_tx_cnt non zero - add a unique (32-bit) number for a each usbnet device and use it for logging - explicitly check for full tx ring in usbnet_start_locked() and return early, which avoids setting the 5 second timer and triggering a watchdog - reset uncd_tx_prod and uncd_tx_cnt to zero in stop
2019-08-23 07:29:28 +03:00
USBNETHIST_FUNC();
reduce the scope of struct usbnet: - move a large number of members internal to usbnet.c's new "struct usbnet_private". - provide accessors for a few of these - move struct usbnet_cdata into usbnet.c as well, but move bufsz, list count, and xfer flags back out into struct usbnet, and have them set as part of the setup efore usbnet_attach() - split the intr pipe parts into their own structure - move all the main usbnet*lock* code into usbnet.c too usbnet_attach() goes down to 2 args, and the inputs needed are now the full contents of 'struct usbnet' besides the driver owned 'un_flags' and usbnet owned 'un_pri'. welcome netbsd 9.99.6.
2019-08-10 05:17:36 +03:00
struct usbnet_chain * const c = priv;
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
struct usbnet * const un = c->unc_un;
reduce the scope of struct usbnet: - move a large number of members internal to usbnet.c's new "struct usbnet_private". - provide accessors for a few of these - move struct usbnet_cdata into usbnet.c as well, but move bufsz, list count, and xfer flags back out into struct usbnet, and have them set as part of the setup efore usbnet_attach() - split the intr pipe parts into their own structure - move all the main usbnet*lock* code into usbnet.c too usbnet_attach() goes down to 2 args, and the inputs needed are now the full contents of 'struct usbnet' besides the driver owned 'un_flags' and usbnet owned 'un_pri'. welcome netbsd 9.99.6.
2019-08-10 05:17:36 +03:00
struct usbnet_private * const unp = un->un_pri;
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
uint32_t total_len;
PR/55068: sc.dying: Fix printf formats: - no %s/%p for kernel log - 0x% -> %# - always %j for kernel log
2020-03-13 21:17:40 +03:00
USBNETHIST_CALLARGSN(5, "%jd: enter: status %#jx xfer %#jx",
- remove unused uncd_rx_cnt - add USBNETHIST_CALLARGSN and use it in many places - add more history logs - add a log if watchdog abort leaves uncd_tx_cnt non zero - add a unique (32-bit) number for a each usbnet device and use it for logging - explicitly check for full tx ring in usbnet_start_locked() and return early, which avoids setting the 5 second timer and triggering a watchdog - reset uncd_tx_prod and uncd_tx_cnt to zero in stop
2019-08-23 07:29:28 +03:00
unp->unp_number, status, (uintptr_t)xfer, 0);
reduce the scope of struct usbnet: - move a large number of members internal to usbnet.c's new "struct usbnet_private". - provide accessors for a few of these - move struct usbnet_cdata into usbnet.c as well, but move bufsz, list count, and xfer flags back out into struct usbnet, and have them set as part of the setup efore usbnet_attach() - split the intr pipe parts into their own structure - move all the main usbnet*lock* code into usbnet.c too usbnet_attach() goes down to 2 args, and the inputs needed are now the full contents of 'struct usbnet' besides the driver owned 'un_flags' and usbnet owned 'un_pri'. welcome netbsd 9.99.6.
2019-08-10 05:17:36 +03:00
mutex_enter(&unp->unp_rxlock);
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
usbnet: Use atomic_load/store_relaxed for unp_dying. This way we don't need to hold the core lock to avoid upsetting sanitizers (which probably find the current code upsetting), and we can use it to exit early from timeout loops that run under the core lock (which is probably not necessary for them to do anyway, but let's worry about that later).
2022-03-03 08:50:39 +03:00
if (usbnet_isdying(un) || unp->unp_stopping ||
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
status == USBD_INVAL || status == USBD_NOT_STARTED ||
usbnet: Don't check if_flags for IFF_RUNNING in usbnet_rxeof. This can only run after we start the pipes in usbnet_init_rx_tx, and before we abort the pipes in usbnet_stop, during which time if_flags & IFF_RUNNING is stably set.
2022-03-03 08:47:43 +03:00
status == USBD_CANCELLED)
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
goto out;
if (status != USBD_NORMAL_COMPLETION) {
reduce the scope of struct usbnet: - move a large number of members internal to usbnet.c's new "struct usbnet_private". - provide accessors for a few of these - move struct usbnet_cdata into usbnet.c as well, but move bufsz, list count, and xfer flags back out into struct usbnet, and have them set as part of the setup efore usbnet_attach() - split the intr pipe parts into their own structure - move all the main usbnet*lock* code into usbnet.c too usbnet_attach() goes down to 2 args, and the inputs needed are now the full contents of 'struct usbnet' besides the driver owned 'un_flags' and usbnet owned 'un_pri'. welcome netbsd 9.99.6.
2019-08-10 05:17:36 +03:00
if (usbd_ratecheck(&unp->unp_rx_notice))
reduce aprint_error(9) abuse
2021-03-01 20:41:00 +03:00
device_printf(un->un_dev, "usb errors on rx: %s\n",
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
usbd_errstr(status));
if (status == USBD_STALLED)
reduce the scope of struct usbnet: - move a large number of members internal to usbnet.c's new "struct usbnet_private". - provide accessors for a few of these - move struct usbnet_cdata into usbnet.c as well, but move bufsz, list count, and xfer flags back out into struct usbnet, and have them set as part of the setup efore usbnet_attach() - split the intr pipe parts into their own structure - move all the main usbnet*lock* code into usbnet.c too usbnet_attach() goes down to 2 args, and the inputs needed are now the full contents of 'struct usbnet' besides the driver owned 'un_flags' and usbnet owned 'un_pri'. welcome netbsd 9.99.6.
2019-08-10 05:17:36 +03:00
usbd_clear_endpoint_stall_async(unp->unp_ep[USBNET_ENDPT_RX]);
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
goto done;
}
usbd_get_xfer_status(xfer, NULL, NULL, &total_len, NULL);
reduce the scope of struct usbnet: - move a large number of members internal to usbnet.c's new "struct usbnet_private". - provide accessors for a few of these - move struct usbnet_cdata into usbnet.c as well, but move bufsz, list count, and xfer flags back out into struct usbnet, and have them set as part of the setup efore usbnet_attach() - split the intr pipe parts into their own structure - move all the main usbnet*lock* code into usbnet.c too usbnet_attach() goes down to 2 args, and the inputs needed are now the full contents of 'struct usbnet' besides the driver owned 'un_flags' and usbnet owned 'un_pri'. welcome netbsd 9.99.6.
2019-08-10 05:17:36 +03:00
if (total_len > un->un_rx_bufsz) {
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
aprint_error_dev(un->un_dev,
"rxeof: too large transfer (%u > %u)\n",
reduce the scope of struct usbnet: - move a large number of members internal to usbnet.c's new "struct usbnet_private". - provide accessors for a few of these - move struct usbnet_cdata into usbnet.c as well, but move bufsz, list count, and xfer flags back out into struct usbnet, and have them set as part of the setup efore usbnet_attach() - split the intr pipe parts into their own structure - move all the main usbnet*lock* code into usbnet.c too usbnet_attach() goes down to 2 args, and the inputs needed are now the full contents of 'struct usbnet' besides the driver owned 'un_flags' and usbnet owned 'un_pri'. welcome netbsd 9.99.6.
2019-08-10 05:17:36 +03:00
total_len, un->un_rx_bufsz);
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
goto done;
}
- usbnet_rx_loop_cb's usbd_xfer parameter is never used and available in the usbnet_chain if needed. remove it - usbnet media status change already set link to false, don't repeat this in every driver - don't clear link in stop, nothing was re-enabling it for non-MII - add optional uno_tick_cb(struct usbnet *un) that is called from the usbnet tick timer - remove wrong debug sysctl prototype rx_loop and timer are kernel versions changes, but hopefully this is the last one for usbnet. working with 3 more drivers now (cue, mue and url), leaving only aue, kue, upl and umb undone (aue may work with previously supported devices, mine doesn't work with our driver, kue and upl have patches for testing and umb is undone.)
2019-08-15 08:52:23 +03:00
uno_rx_loop(un, c, total_len);
- use usbnet_isowned*() more - more const
2019-08-11 04:31:19 +03:00
usbnet_isowned_rx(un);
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
done:
usbnet: Use atomic_load/store_relaxed for unp_dying. This way we don't need to hold the core lock to avoid upsetting sanitizers (which probably find the current code upsetting), and we can use it to exit early from timeout loops that run under the core lock (which is probably not necessary for them to do anyway, but let's worry about that later).
2022-03-03 08:50:39 +03:00
if (usbnet_isdying(un) || unp->unp_stopping)
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
goto out;
reduce the scope of struct usbnet: - move a large number of members internal to usbnet.c's new "struct usbnet_private". - provide accessors for a few of these - move struct usbnet_cdata into usbnet.c as well, but move bufsz, list count, and xfer flags back out into struct usbnet, and have them set as part of the setup efore usbnet_attach() - split the intr pipe parts into their own structure - move all the main usbnet*lock* code into usbnet.c too usbnet_attach() goes down to 2 args, and the inputs needed are now the full contents of 'struct usbnet' besides the driver owned 'un_flags' and usbnet owned 'un_pri'. welcome netbsd 9.99.6.
2019-08-10 05:17:36 +03:00
mutex_exit(&unp->unp_rxlock);
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
/* Setup new transfer. */
reduce the scope of struct usbnet: - move a large number of members internal to usbnet.c's new "struct usbnet_private". - provide accessors for a few of these - move struct usbnet_cdata into usbnet.c as well, but move bufsz, list count, and xfer flags back out into struct usbnet, and have them set as part of the setup efore usbnet_attach() - split the intr pipe parts into their own structure - move all the main usbnet*lock* code into usbnet.c too usbnet_attach() goes down to 2 args, and the inputs needed are now the full contents of 'struct usbnet' besides the driver owned 'un_flags' and usbnet owned 'un_pri'. welcome netbsd 9.99.6.
2019-08-10 05:17:36 +03:00
usbd_setup_xfer(xfer, c, c->unc_buf, un->un_rx_bufsz,
un->un_rx_xfer_flags, USBD_NO_TIMEOUT, usbnet_rxeof);
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
usbd_transfer(xfer);
return;
out:
reduce the scope of struct usbnet: - move a large number of members internal to usbnet.c's new "struct usbnet_private". - provide accessors for a few of these - move struct usbnet_cdata into usbnet.c as well, but move bufsz, list count, and xfer flags back out into struct usbnet, and have them set as part of the setup efore usbnet_attach() - split the intr pipe parts into their own structure - move all the main usbnet*lock* code into usbnet.c too usbnet_attach() goes down to 2 args, and the inputs needed are now the full contents of 'struct usbnet' besides the driver owned 'un_flags' and usbnet owned 'un_pri'. welcome netbsd 9.99.6.
2019-08-10 05:17:36 +03:00
mutex_exit(&unp->unp_rxlock);
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
}
static void
- adjust usbnet interface to allow usbd_open_pipe_intr(), from the new comment: * if un_intr_buf is not NULL, use usbd_open_pipe_intr() not * usbd_open_pipe() for USBNET_ENDPT_INTR, with this buffer, * size, and interval. the standard handling is in usbnet.c, with a callback to deal with the interrupt it self. not fully tested, designed for if_aue.c and a few others not yet converted. - make usbhist for usbnet.c work, thanks paulg - usbnet_init_rx_tx() clears out all allocations upon failure now - add usbnet_ec() to get a pointer to the struct ethercom - add usbnet_{lock,unlock,owned}*() to lock/unlock the various locks and *owned*() for asserting welcome 9.99.3!
2019-08-04 11:59:13 +03:00
usbnet_txeof(struct usbd_xfer *xfer, void *priv, usbd_status status)
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
{
- adjust usbnet interface to allow usbd_open_pipe_intr(), from the new comment: * if un_intr_buf is not NULL, use usbd_open_pipe_intr() not * usbd_open_pipe() for USBNET_ENDPT_INTR, with this buffer, * size, and interval. the standard handling is in usbnet.c, with a callback to deal with the interrupt it self. not fully tested, designed for if_aue.c and a few others not yet converted. - make usbhist for usbnet.c work, thanks paulg - usbnet_init_rx_tx() clears out all allocations upon failure now - add usbnet_ec() to get a pointer to the struct ethercom - add usbnet_{lock,unlock,owned}*() to lock/unlock the various locks and *owned*() for asserting welcome 9.99.3!
2019-08-04 11:59:13 +03:00
USBNETHIST_FUNC(); USBNETHIST_CALLED();
reduce the scope of struct usbnet: - move a large number of members internal to usbnet.c's new "struct usbnet_private". - provide accessors for a few of these - move struct usbnet_cdata into usbnet.c as well, but move bufsz, list count, and xfer flags back out into struct usbnet, and have them set as part of the setup efore usbnet_attach() - split the intr pipe parts into their own structure - move all the main usbnet*lock* code into usbnet.c too usbnet_attach() goes down to 2 args, and the inputs needed are now the full contents of 'struct usbnet' besides the driver owned 'un_flags' and usbnet owned 'un_pri'. welcome netbsd 9.99.6.
2019-08-10 05:17:36 +03:00
struct usbnet_chain * const c = priv;
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
struct usbnet * const un = c->unc_un;
reduce the scope of struct usbnet: - move a large number of members internal to usbnet.c's new "struct usbnet_private". - provide accessors for a few of these - move struct usbnet_cdata into usbnet.c as well, but move bufsz, list count, and xfer flags back out into struct usbnet, and have them set as part of the setup efore usbnet_attach() - split the intr pipe parts into their own structure - move all the main usbnet*lock* code into usbnet.c too usbnet_attach() goes down to 2 args, and the inputs needed are now the full contents of 'struct usbnet' besides the driver owned 'un_flags' and usbnet owned 'un_pri'. welcome netbsd 9.99.6.
2019-08-10 05:17:36 +03:00
struct usbnet_cdata * const cd = un_cdata(un);
struct usbnet_private * const unp = un->un_pri;
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
struct ifnet * const ifp = usbnet_ifp(un);
PR/55068: sc.dying: Fix printf formats: - no %s/%p for kernel log - 0x% -> %# - always %j for kernel log
2020-03-13 21:17:40 +03:00
USBNETHIST_CALLARGSN(5, "%jd: enter: status %#jx xfer %#jx",
- remove unused uncd_rx_cnt - add USBNETHIST_CALLARGSN and use it in many places - add more history logs - add a log if watchdog abort leaves uncd_tx_cnt non zero - add a unique (32-bit) number for a each usbnet device and use it for logging - explicitly check for full tx ring in usbnet_start_locked() and return early, which avoids setting the 5 second timer and triggering a watchdog - reset uncd_tx_prod and uncd_tx_cnt to zero in stop
2019-08-23 07:29:28 +03:00
unp->unp_number, status, (uintptr_t)xfer, 0);
reduce the scope of struct usbnet: - move a large number of members internal to usbnet.c's new "struct usbnet_private". - provide accessors for a few of these - move struct usbnet_cdata into usbnet.c as well, but move bufsz, list count, and xfer flags back out into struct usbnet, and have them set as part of the setup efore usbnet_attach() - split the intr pipe parts into their own structure - move all the main usbnet*lock* code into usbnet.c too usbnet_attach() goes down to 2 args, and the inputs needed are now the full contents of 'struct usbnet' besides the driver owned 'un_flags' and usbnet owned 'un_pri'. welcome netbsd 9.99.6.
2019-08-10 05:17:36 +03:00
mutex_enter(&unp->unp_txlock);
usbnet: Use atomic_load/store_relaxed for unp_dying. This way we don't need to hold the core lock to avoid upsetting sanitizers (which probably find the current code upsetting), and we can use it to exit early from timeout loops that run under the core lock (which is probably not necessary for them to do anyway, but let's worry about that later).
2022-03-03 08:50:39 +03:00
if (unp->unp_stopping || usbnet_isdying(un)) {
reduce the scope of struct usbnet: - move a large number of members internal to usbnet.c's new "struct usbnet_private". - provide accessors for a few of these - move struct usbnet_cdata into usbnet.c as well, but move bufsz, list count, and xfer flags back out into struct usbnet, and have them set as part of the setup efore usbnet_attach() - split the intr pipe parts into their own structure - move all the main usbnet*lock* code into usbnet.c too usbnet_attach() goes down to 2 args, and the inputs needed are now the full contents of 'struct usbnet' besides the driver owned 'un_flags' and usbnet owned 'un_pri'. welcome netbsd 9.99.6.
2019-08-10 05:17:36 +03:00
mutex_exit(&unp->unp_txlock);
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
return;
}
KASSERT(cd->uncd_tx_cnt > 0);
cd->uncd_tx_cnt--;
reduce the scope of struct usbnet: - move a large number of members internal to usbnet.c's new "struct usbnet_private". - provide accessors for a few of these - move struct usbnet_cdata into usbnet.c as well, but move bufsz, list count, and xfer flags back out into struct usbnet, and have them set as part of the setup efore usbnet_attach() - split the intr pipe parts into their own structure - move all the main usbnet*lock* code into usbnet.c too usbnet_attach() goes down to 2 args, and the inputs needed are now the full contents of 'struct usbnet' besides the driver owned 'un_flags' and usbnet owned 'un_pri'. welcome netbsd 9.99.6.
2019-08-10 05:17:36 +03:00
unp->unp_timer = 0;
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
switch (status) {
case USBD_NOT_STARTED:
case USBD_CANCELLED:
break;
case USBD_NORMAL_COMPLETION:
Adopt <net/if_stats.h>.
2020-01-29 09:35:28 +03:00
if_statinc(ifp, if_opackets);
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
break;
default:
Adopt <net/if_stats.h>.
2020-01-29 09:35:28 +03:00
if_statinc(ifp, if_oerrors);
reduce the scope of struct usbnet: - move a large number of members internal to usbnet.c's new "struct usbnet_private". - provide accessors for a few of these - move struct usbnet_cdata into usbnet.c as well, but move bufsz, list count, and xfer flags back out into struct usbnet, and have them set as part of the setup efore usbnet_attach() - split the intr pipe parts into their own structure - move all the main usbnet*lock* code into usbnet.c too usbnet_attach() goes down to 2 args, and the inputs needed are now the full contents of 'struct usbnet' besides the driver owned 'un_flags' and usbnet owned 'un_pri'. welcome netbsd 9.99.6.
2019-08-10 05:17:36 +03:00
if (usbd_ratecheck(&unp->unp_tx_notice))
reduce aprint_error(9) abuse
2021-03-01 20:41:00 +03:00
device_printf(un->un_dev, "usb error on tx: %s\n",
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
usbd_errstr(status));
if (status == USBD_STALLED)
reduce the scope of struct usbnet: - move a large number of members internal to usbnet.c's new "struct usbnet_private". - provide accessors for a few of these - move struct usbnet_cdata into usbnet.c as well, but move bufsz, list count, and xfer flags back out into struct usbnet, and have them set as part of the setup efore usbnet_attach() - split the intr pipe parts into their own structure - move all the main usbnet*lock* code into usbnet.c too usbnet_attach() goes down to 2 args, and the inputs needed are now the full contents of 'struct usbnet' besides the driver owned 'un_flags' and usbnet owned 'un_pri'. welcome netbsd 9.99.6.
2019-08-10 05:17:36 +03:00
usbd_clear_endpoint_stall_async(unp->unp_ep[USBNET_ENDPT_TX]);
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
break;
}
reduce the scope of struct usbnet: - move a large number of members internal to usbnet.c's new "struct usbnet_private". - provide accessors for a few of these - move struct usbnet_cdata into usbnet.c as well, but move bufsz, list count, and xfer flags back out into struct usbnet, and have them set as part of the setup efore usbnet_attach() - split the intr pipe parts into their own structure - move all the main usbnet*lock* code into usbnet.c too usbnet_attach() goes down to 2 args, and the inputs needed are now the full contents of 'struct usbnet' besides the driver owned 'un_flags' and usbnet owned 'un_pri'. welcome netbsd 9.99.6.
2019-08-10 05:17:36 +03:00
mutex_exit(&unp->unp_txlock);
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
if (status == USBD_NORMAL_COMPLETION && !IFQ_IS_EMPTY(&ifp->if_snd))
(*ifp->if_start)(ifp);
}
- adjust usbnet interface to allow usbd_open_pipe_intr(), from the new comment: * if un_intr_buf is not NULL, use usbd_open_pipe_intr() not * usbd_open_pipe() for USBNET_ENDPT_INTR, with this buffer, * size, and interval. the standard handling is in usbnet.c, with a callback to deal with the interrupt it self. not fully tested, designed for if_aue.c and a few others not yet converted. - make usbhist for usbnet.c work, thanks paulg - usbnet_init_rx_tx() clears out all allocations upon failure now - add usbnet_ec() to get a pointer to the struct ethercom - add usbnet_{lock,unlock,owned}*() to lock/unlock the various locks and *owned*() for asserting welcome 9.99.3!
2019-08-04 11:59:13 +03:00
static void
reduce the scope of struct usbnet: - move a large number of members internal to usbnet.c's new "struct usbnet_private". - provide accessors for a few of these - move struct usbnet_cdata into usbnet.c as well, but move bufsz, list count, and xfer flags back out into struct usbnet, and have them set as part of the setup efore usbnet_attach() - split the intr pipe parts into their own structure - move all the main usbnet*lock* code into usbnet.c too usbnet_attach() goes down to 2 args, and the inputs needed are now the full contents of 'struct usbnet' besides the driver owned 'un_flags' and usbnet owned 'un_pri'. welcome netbsd 9.99.6.
2019-08-10 05:17:36 +03:00
usbnet_pipe_intr(struct usbd_xfer *xfer, void *priv, usbd_status status)
- adjust usbnet interface to allow usbd_open_pipe_intr(), from the new comment: * if un_intr_buf is not NULL, use usbd_open_pipe_intr() not * usbd_open_pipe() for USBNET_ENDPT_INTR, with this buffer, * size, and interval. the standard handling is in usbnet.c, with a callback to deal with the interrupt it self. not fully tested, designed for if_aue.c and a few others not yet converted. - make usbhist for usbnet.c work, thanks paulg - usbnet_init_rx_tx() clears out all allocations upon failure now - add usbnet_ec() to get a pointer to the struct ethercom - add usbnet_{lock,unlock,owned}*() to lock/unlock the various locks and *owned*() for asserting welcome 9.99.3!
2019-08-04 11:59:13 +03:00
{
- use CALLARGS vs CALLED for better usbhist - turn off usbnetdebug default - log for all entry/exit points of usbnet_pipe_intr() - in usbnet_start_locked() track whether any packet has been transmitted for setting the timer. avoids spurious "watchdog timeouts" - in usbnet_stop() use callout_halt() vs callout_halt, and also stop the usb task. fixes crash of usbtask after the phy has detached. - add a little more defensive checking in the tick task, and add some high-log-level logs. - in usbnet_detach() move the call to usbnet_stop_ifp() above the calls to callout/usbtask stopping. - set ec_mii and unp_pri to NULL when freeing their data
2019-09-08 21:59:32 +03:00
USBNETHIST_FUNC();
reduce the scope of struct usbnet: - move a large number of members internal to usbnet.c's new "struct usbnet_private". - provide accessors for a few of these - move struct usbnet_cdata into usbnet.c as well, but move bufsz, list count, and xfer flags back out into struct usbnet, and have them set as part of the setup efore usbnet_attach() - split the intr pipe parts into their own structure - move all the main usbnet*lock* code into usbnet.c too usbnet_attach() goes down to 2 args, and the inputs needed are now the full contents of 'struct usbnet' besides the driver owned 'un_flags' and usbnet owned 'un_pri'. welcome netbsd 9.99.6.
2019-08-10 05:17:36 +03:00
struct usbnet * const un = priv;
struct usbnet_private * const unp = un->un_pri;
struct usbnet_intr * const uni = un->un_intr;
- adjust usbnet interface to allow usbd_open_pipe_intr(), from the new comment: * if un_intr_buf is not NULL, use usbd_open_pipe_intr() not * usbd_open_pipe() for USBNET_ENDPT_INTR, with this buffer, * size, and interval. the standard handling is in usbnet.c, with a callback to deal with the interrupt it self. not fully tested, designed for if_aue.c and a few others not yet converted. - make usbhist for usbnet.c work, thanks paulg - usbnet_init_rx_tx() clears out all allocations upon failure now - add usbnet_ec() to get a pointer to the struct ethercom - add usbnet_{lock,unlock,owned}*() to lock/unlock the various locks and *owned*() for asserting welcome 9.99.3!
2019-08-04 11:59:13 +03:00
usbnet: Use atomic_load/store_relaxed for unp_dying. This way we don't need to hold the core lock to avoid upsetting sanitizers (which probably find the current code upsetting), and we can use it to exit early from timeout loops that run under the core lock (which is probably not necessary for them to do anyway, but let's worry about that later).
2022-03-03 08:50:39 +03:00
if (uni == NULL || usbnet_isdying(un) || unp->unp_stopping ||
- adjust usbnet interface to allow usbd_open_pipe_intr(), from the new comment: * if un_intr_buf is not NULL, use usbd_open_pipe_intr() not * usbd_open_pipe() for USBNET_ENDPT_INTR, with this buffer, * size, and interval. the standard handling is in usbnet.c, with a callback to deal with the interrupt it self. not fully tested, designed for if_aue.c and a few others not yet converted. - make usbhist for usbnet.c work, thanks paulg - usbnet_init_rx_tx() clears out all allocations upon failure now - add usbnet_ec() to get a pointer to the struct ethercom - add usbnet_{lock,unlock,owned}*() to lock/unlock the various locks and *owned*() for asserting welcome 9.99.3!
2019-08-04 11:59:13 +03:00
status == USBD_INVAL || status == USBD_NOT_STARTED ||
usbnet: Don't check if_flags for IFF_RUNNING in usbnet_pipe_intr. The one user of this interface in tree, aue(4), doesn't care -- if_statinc is safe whether IFF_RUNNING or not.
2022-03-03 08:47:50 +03:00
status == USBD_CANCELLED) {
PR/55068: sc.dying: Fix printf formats: - no %s/%p for kernel log - 0x% -> %# - always %j for kernel log
2020-03-13 21:17:40 +03:00
USBNETHIST_CALLARGS("%jd: uni %#jx d/s %#jx status %#jx",
- use CALLARGS vs CALLED for better usbhist - turn off usbnetdebug default - log for all entry/exit points of usbnet_pipe_intr() - in usbnet_start_locked() track whether any packet has been transmitted for setting the timer. avoids spurious "watchdog timeouts" - in usbnet_stop() use callout_halt() vs callout_halt, and also stop the usb task. fixes crash of usbtask after the phy has detached. - add a little more defensive checking in the tick task, and add some high-log-level logs. - in usbnet_detach() move the call to usbnet_stop_ifp() above the calls to callout/usbtask stopping. - set ec_mii and unp_pri to NULL when freeing their data
2019-09-08 21:59:32 +03:00
unp->unp_number, (uintptr_t)uni,
usbnet: Use atomic_load/store_relaxed for unp_dying. This way we don't need to hold the core lock to avoid upsetting sanitizers (which probably find the current code upsetting), and we can use it to exit early from timeout loops that run under the core lock (which is probably not necessary for them to do anyway, but let's worry about that later).
2022-03-03 08:50:39 +03:00
(usbnet_isdying(un) << 8) | unp->unp_stopping, status);
- adjust usbnet interface to allow usbd_open_pipe_intr(), from the new comment: * if un_intr_buf is not NULL, use usbd_open_pipe_intr() not * usbd_open_pipe() for USBNET_ENDPT_INTR, with this buffer, * size, and interval. the standard handling is in usbnet.c, with a callback to deal with the interrupt it self. not fully tested, designed for if_aue.c and a few others not yet converted. - make usbhist for usbnet.c work, thanks paulg - usbnet_init_rx_tx() clears out all allocations upon failure now - add usbnet_ec() to get a pointer to the struct ethercom - add usbnet_{lock,unlock,owned}*() to lock/unlock the various locks and *owned*() for asserting welcome 9.99.3!
2019-08-04 11:59:13 +03:00
return;
- use CALLARGS vs CALLED for better usbhist - turn off usbnetdebug default - log for all entry/exit points of usbnet_pipe_intr() - in usbnet_start_locked() track whether any packet has been transmitted for setting the timer. avoids spurious "watchdog timeouts" - in usbnet_stop() use callout_halt() vs callout_halt, and also stop the usb task. fixes crash of usbtask after the phy has detached. - add a little more defensive checking in the tick task, and add some high-log-level logs. - in usbnet_detach() move the call to usbnet_stop_ifp() above the calls to callout/usbtask stopping. - set ec_mii and unp_pri to NULL when freeing their data
2019-09-08 21:59:32 +03:00
}
- adjust usbnet interface to allow usbd_open_pipe_intr(), from the new comment: * if un_intr_buf is not NULL, use usbd_open_pipe_intr() not * usbd_open_pipe() for USBNET_ENDPT_INTR, with this buffer, * size, and interval. the standard handling is in usbnet.c, with a callback to deal with the interrupt it self. not fully tested, designed for if_aue.c and a few others not yet converted. - make usbhist for usbnet.c work, thanks paulg - usbnet_init_rx_tx() clears out all allocations upon failure now - add usbnet_ec() to get a pointer to the struct ethercom - add usbnet_{lock,unlock,owned}*() to lock/unlock the various locks and *owned*() for asserting welcome 9.99.3!
2019-08-04 11:59:13 +03:00
if (status != USBD_NORMAL_COMPLETION) {
reduce the scope of struct usbnet: - move a large number of members internal to usbnet.c's new "struct usbnet_private". - provide accessors for a few of these - move struct usbnet_cdata into usbnet.c as well, but move bufsz, list count, and xfer flags back out into struct usbnet, and have them set as part of the setup efore usbnet_attach() - split the intr pipe parts into their own structure - move all the main usbnet*lock* code into usbnet.c too usbnet_attach() goes down to 2 args, and the inputs needed are now the full contents of 'struct usbnet' besides the driver owned 'un_flags' and usbnet owned 'un_pri'. welcome netbsd 9.99.6.
2019-08-10 05:17:36 +03:00
if (usbd_ratecheck(&unp->unp_intr_notice)) {
- adjust usbnet interface to allow usbd_open_pipe_intr(), from the new comment: * if un_intr_buf is not NULL, use usbd_open_pipe_intr() not * usbd_open_pipe() for USBNET_ENDPT_INTR, with this buffer, * size, and interval. the standard handling is in usbnet.c, with a callback to deal with the interrupt it self. not fully tested, designed for if_aue.c and a few others not yet converted. - make usbhist for usbnet.c work, thanks paulg - usbnet_init_rx_tx() clears out all allocations upon failure now - add usbnet_ec() to get a pointer to the struct ethercom - add usbnet_{lock,unlock,owned}*() to lock/unlock the various locks and *owned*() for asserting welcome 9.99.3!
2019-08-04 11:59:13 +03:00
aprint_error_dev(un->un_dev, "usb error on intr: %s\n",
usbd_errstr(status));
}
if (status == USBD_STALLED)
reduce the scope of struct usbnet: - move a large number of members internal to usbnet.c's new "struct usbnet_private". - provide accessors for a few of these - move struct usbnet_cdata into usbnet.c as well, but move bufsz, list count, and xfer flags back out into struct usbnet, and have them set as part of the setup efore usbnet_attach() - split the intr pipe parts into their own structure - move all the main usbnet*lock* code into usbnet.c too usbnet_attach() goes down to 2 args, and the inputs needed are now the full contents of 'struct usbnet' besides the driver owned 'un_flags' and usbnet owned 'un_pri'. welcome netbsd 9.99.6.
2019-08-10 05:17:36 +03:00
usbd_clear_endpoint_stall_async(unp->unp_ep[USBNET_ENDPT_INTR]);
PR/55068: sc.dying: Fix printf formats: - no %s/%p for kernel log - 0x% -> %# - always %j for kernel log
2020-03-13 21:17:40 +03:00
USBNETHIST_CALLARGS("%jd: not normal status %#jx",
- use CALLARGS vs CALLED for better usbhist - turn off usbnetdebug default - log for all entry/exit points of usbnet_pipe_intr() - in usbnet_start_locked() track whether any packet has been transmitted for setting the timer. avoids spurious "watchdog timeouts" - in usbnet_stop() use callout_halt() vs callout_halt, and also stop the usb task. fixes crash of usbtask after the phy has detached. - add a little more defensive checking in the tick task, and add some high-log-level logs. - in usbnet_detach() move the call to usbnet_stop_ifp() above the calls to callout/usbtask stopping. - set ec_mii and unp_pri to NULL when freeing their data
2019-09-08 21:59:32 +03:00
unp->unp_number, status, 0, 0);
- adjust usbnet interface to allow usbd_open_pipe_intr(), from the new comment: * if un_intr_buf is not NULL, use usbd_open_pipe_intr() not * usbd_open_pipe() for USBNET_ENDPT_INTR, with this buffer, * size, and interval. the standard handling is in usbnet.c, with a callback to deal with the interrupt it self. not fully tested, designed for if_aue.c and a few others not yet converted. - make usbhist for usbnet.c work, thanks paulg - usbnet_init_rx_tx() clears out all allocations upon failure now - add usbnet_ec() to get a pointer to the struct ethercom - add usbnet_{lock,unlock,owned}*() to lock/unlock the various locks and *owned*() for asserting welcome 9.99.3!
2019-08-04 11:59:13 +03:00
return;
}
update usbnet some: - move rx/tx xfer flags into usbnet_cdata - move the callbacks into usbnet_ops structure - move rx/tx xfer flags arguments from usbnet_init_rx_tx() and move them all into usbnet_attach() arguments - s/miibus/mii/ in some places for consistency other clean up: - create wrapper functions for callbacks, move knowledge about special handling (OK to be missing, error eating) there. - use cdata pointer if already available - provide some more macros (will be real functions later) for accessing usbnet members, use existing ones more bump kernel version.
2019-08-09 04:17:33 +03:00
uno_intr(un, status);
- adjust usbnet interface to allow usbd_open_pipe_intr(), from the new comment: * if un_intr_buf is not NULL, use usbd_open_pipe_intr() not * usbd_open_pipe() for USBNET_ENDPT_INTR, with this buffer, * size, and interval. the standard handling is in usbnet.c, with a callback to deal with the interrupt it self. not fully tested, designed for if_aue.c and a few others not yet converted. - make usbhist for usbnet.c work, thanks paulg - usbnet_init_rx_tx() clears out all allocations upon failure now - add usbnet_ec() to get a pointer to the struct ethercom - add usbnet_{lock,unlock,owned}*() to lock/unlock the various locks and *owned*() for asserting welcome 9.99.3!
2019-08-04 11:59:13 +03:00
}
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
static void
usbnet_start_locked(struct ifnet *ifp)
{
- use CALLARGS vs CALLED for better usbhist - turn off usbnetdebug default - log for all entry/exit points of usbnet_pipe_intr() - in usbnet_start_locked() track whether any packet has been transmitted for setting the timer. avoids spurious "watchdog timeouts" - in usbnet_stop() use callout_halt() vs callout_halt, and also stop the usb task. fixes crash of usbtask after the phy has detached. - add a little more defensive checking in the tick task, and add some high-log-level logs. - in usbnet_detach() move the call to usbnet_stop_ifp() above the calls to callout/usbtask stopping. - set ec_mii and unp_pri to NULL when freeing their data
2019-09-08 21:59:32 +03:00
USBNETHIST_FUNC();
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
struct usbnet * const un = ifp->if_softc;
reduce the scope of struct usbnet: - move a large number of members internal to usbnet.c's new "struct usbnet_private". - provide accessors for a few of these - move struct usbnet_cdata into usbnet.c as well, but move bufsz, list count, and xfer flags back out into struct usbnet, and have them set as part of the setup efore usbnet_attach() - split the intr pipe parts into their own structure - move all the main usbnet*lock* code into usbnet.c too usbnet_attach() goes down to 2 args, and the inputs needed are now the full contents of 'struct usbnet' besides the driver owned 'un_flags' and usbnet owned 'un_pri'. welcome netbsd 9.99.6.
2019-08-10 05:17:36 +03:00
struct usbnet_cdata * const cd = un_cdata(un);
struct usbnet_private * const unp = un->un_pri;
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
struct mbuf *m;
unsigned length;
- use CALLARGS vs CALLED for better usbhist - turn off usbnetdebug default - log for all entry/exit points of usbnet_pipe_intr() - in usbnet_start_locked() track whether any packet has been transmitted for setting the timer. avoids spurious "watchdog timeouts" - in usbnet_stop() use callout_halt() vs callout_halt, and also stop the usb task. fixes crash of usbtask after the phy has detached. - add a little more defensive checking in the tick task, and add some high-log-level logs. - in usbnet_detach() move the call to usbnet_stop_ifp() above the calls to callout/usbtask stopping. - set ec_mii and unp_pri to NULL when freeing their data
2019-09-08 21:59:32 +03:00
bool done_transmit = false;
Sample # of TX packets as entropy source. For RX packets, individual drivers need to be modified.
2021-04-25 08:15:20 +03:00
int idx, count;
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
PR/55068: sc.dying: Fix printf formats: - no %s/%p for kernel log - 0x% -> %# - always %j for kernel log
2020-03-13 21:17:40 +03:00
USBNETHIST_CALLARGS("%jd: tx_cnt %jd list_cnt %jd link %jd",
- use CALLARGS vs CALLED for better usbhist - turn off usbnetdebug default - log for all entry/exit points of usbnet_pipe_intr() - in usbnet_start_locked() track whether any packet has been transmitted for setting the timer. avoids spurious "watchdog timeouts" - in usbnet_stop() use callout_halt() vs callout_halt, and also stop the usb task. fixes crash of usbtask after the phy has detached. - add a little more defensive checking in the tick task, and add some high-log-level logs. - in usbnet_detach() move the call to usbnet_stop_ifp() above the calls to callout/usbtask stopping. - set ec_mii and unp_pri to NULL when freeing their data
2019-09-08 21:59:32 +03:00
unp->unp_number, cd->uncd_tx_cnt, un->un_tx_list_cnt,
unp->unp_link);
- use usbnet_isowned*() more - more const
2019-08-11 04:31:19 +03:00
usbnet_isowned_tx(un);
reduce the scope of struct usbnet: - move a large number of members internal to usbnet.c's new "struct usbnet_private". - provide accessors for a few of these - move struct usbnet_cdata into usbnet.c as well, but move bufsz, list count, and xfer flags back out into struct usbnet, and have them set as part of the setup efore usbnet_attach() - split the intr pipe parts into their own structure - move all the main usbnet*lock* code into usbnet.c too usbnet_attach() goes down to 2 args, and the inputs needed are now the full contents of 'struct usbnet' besides the driver owned 'un_flags' and usbnet owned 'un_pri'. welcome netbsd 9.99.6.
2019-08-10 05:17:36 +03:00
KASSERT(cd->uncd_tx_cnt <= un->un_tx_list_cnt);
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
- usbnet_rx_loop_cb's usbd_xfer parameter is never used and available in the usbnet_chain if needed. remove it - usbnet media status change already set link to false, don't repeat this in every driver - don't clear link in stop, nothing was re-enabling it for non-MII - add optional uno_tick_cb(struct usbnet *un) that is called from the usbnet tick timer - remove wrong debug sysctl prototype rx_loop and timer are kernel versions changes, but hopefully this is the last one for usbnet. working with 3 more drivers now (cue, mue and url), leaving only aue, kue, upl and umb undone (aue may work with previously supported devices, mine doesn't work with our driver, kue and upl have patches for testing and umb is undone.)
2019-08-15 08:52:23 +03:00
if (!unp->unp_link || (ifp->if_flags & IFF_RUNNING) == 0) {
fix more broken kernhist formats (now I got them all).
2020-03-14 06:01:36 +03:00
DPRINTF("start called no link (%jx) or running (flags %jx)",
- usbnet_rx_loop_cb's usbd_xfer parameter is never used and available in the usbnet_chain if needed. remove it - usbnet media status change already set link to false, don't repeat this in every driver - don't clear link in stop, nothing was re-enabling it for non-MII - add optional uno_tick_cb(struct usbnet *un) that is called from the usbnet tick timer - remove wrong debug sysctl prototype rx_loop and timer are kernel versions changes, but hopefully this is the last one for usbnet. working with 3 more drivers now (cue, mue and url), leaving only aue, kue, upl and umb undone (aue may work with previously supported devices, mine doesn't work with our driver, kue and upl have patches for testing and umb is undone.)
2019-08-15 08:52:23 +03:00
unp->unp_link, ifp->if_flags, 0, 0);
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
return;
- usbnet_rx_loop_cb's usbd_xfer parameter is never used and available in the usbnet_chain if needed. remove it - usbnet media status change already set link to false, don't repeat this in every driver - don't clear link in stop, nothing was re-enabling it for non-MII - add optional uno_tick_cb(struct usbnet *un) that is called from the usbnet tick timer - remove wrong debug sysctl prototype rx_loop and timer are kernel versions changes, but hopefully this is the last one for usbnet. working with 3 more drivers now (cue, mue and url), leaving only aue, kue, upl and umb undone (aue may work with previously supported devices, mine doesn't work with our driver, kue and upl have patches for testing and umb is undone.)
2019-08-15 08:52:23 +03:00
}
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
- remove unused uncd_rx_cnt - add USBNETHIST_CALLARGSN and use it in many places - add more history logs - add a log if watchdog abort leaves uncd_tx_cnt non zero - add a unique (32-bit) number for a each usbnet device and use it for logging - explicitly check for full tx ring in usbnet_start_locked() and return early, which avoids setting the 5 second timer and triggering a watchdog - reset uncd_tx_prod and uncd_tx_cnt to zero in stop
2019-08-23 07:29:28 +03:00
if (cd->uncd_tx_cnt == un->un_tx_list_cnt) {
PR/55068: sc.dying: Fix printf formats: - no %s/%p for kernel log - 0x% -> %# - always %j for kernel log
2020-03-13 21:17:40 +03:00
DPRINTF("start called, tx busy (%#jx == %#jx)",
- remove unused uncd_rx_cnt - add USBNETHIST_CALLARGSN and use it in many places - add more history logs - add a log if watchdog abort leaves uncd_tx_cnt non zero - add a unique (32-bit) number for a each usbnet device and use it for logging - explicitly check for full tx ring in usbnet_start_locked() and return early, which avoids setting the 5 second timer and triggering a watchdog - reset uncd_tx_prod and uncd_tx_cnt to zero in stop
2019-08-23 07:29:28 +03:00
cd->uncd_tx_cnt, un->un_tx_list_cnt, 0, 0);
return;
}
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
idx = cd->uncd_tx_prod;
Sample # of TX packets as entropy source. For RX packets, individual drivers need to be modified.
2021-04-25 08:15:20 +03:00
count = 0;
reduce the scope of struct usbnet: - move a large number of members internal to usbnet.c's new "struct usbnet_private". - provide accessors for a few of these - move struct usbnet_cdata into usbnet.c as well, but move bufsz, list count, and xfer flags back out into struct usbnet, and have them set as part of the setup efore usbnet_attach() - split the intr pipe parts into their own structure - move all the main usbnet*lock* code into usbnet.c too usbnet_attach() goes down to 2 args, and the inputs needed are now the full contents of 'struct usbnet' besides the driver owned 'un_flags' and usbnet owned 'un_pri'. welcome netbsd 9.99.6.
2019-08-10 05:17:36 +03:00
while (cd->uncd_tx_cnt < un->un_tx_list_cnt) {
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
IFQ_POLL(&ifp->if_snd, m);
- remove unused uncd_rx_cnt - add USBNETHIST_CALLARGSN and use it in many places - add more history logs - add a log if watchdog abort leaves uncd_tx_cnt non zero - add a unique (32-bit) number for a each usbnet device and use it for logging - explicitly check for full tx ring in usbnet_start_locked() and return early, which avoids setting the 5 second timer and triggering a watchdog - reset uncd_tx_prod and uncd_tx_cnt to zero in stop
2019-08-23 07:29:28 +03:00
if (m == NULL) {
DPRINTF("start called, queue empty", 0, 0, 0, 0);
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
break;
- remove unused uncd_rx_cnt - add USBNETHIST_CALLARGSN and use it in many places - add more history logs - add a log if watchdog abort leaves uncd_tx_cnt non zero - add a unique (32-bit) number for a each usbnet device and use it for logging - explicitly check for full tx ring in usbnet_start_locked() and return early, which avoids setting the 5 second timer and triggering a watchdog - reset uncd_tx_prod and uncd_tx_cnt to zero in stop
2019-08-23 07:29:28 +03:00
}
don't set dv_private. it's a bad pattern and only didn't trip kmem_free() lossage because struct usbnet is at the start of the softc. for now, enforce this as part of the ABI. catch up urndis with tx_prepare checking buffer length, and also add an assert to usbnet_start_locked() to match.
2019-08-12 02:55:43 +03:00
KASSERT(m->m_pkthdr.len <= un->un_tx_bufsz);
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
update usbnet some: - move rx/tx xfer flags into usbnet_cdata - move the callbacks into usbnet_ops structure - move rx/tx xfer flags arguments from usbnet_init_rx_tx() and move them all into usbnet_attach() arguments - s/miibus/mii/ in some places for consistency other clean up: - create wrapper functions for callbacks, move knowledge about special handling (OK to be missing, error eating) there. - use cdata pointer if already available - provide some more macros (will be real functions later) for accessing usbnet members, use existing ones more bump kernel version.
2019-08-09 04:17:33 +03:00
struct usbnet_chain *c = &cd->uncd_tx_chain[idx];
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
update usbnet some: - move rx/tx xfer flags into usbnet_cdata - move the callbacks into usbnet_ops structure - move rx/tx xfer flags arguments from usbnet_init_rx_tx() and move them all into usbnet_attach() arguments - s/miibus/mii/ in some places for consistency other clean up: - create wrapper functions for callbacks, move knowledge about special handling (OK to be missing, error eating) there. - use cdata pointer if already available - provide some more macros (will be real functions later) for accessing usbnet members, use existing ones more bump kernel version.
2019-08-09 04:17:33 +03:00
length = uno_tx_prepare(un, m, c);
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
if (length == 0) {
- remove unused uncd_rx_cnt - add USBNETHIST_CALLARGSN and use it in many places - add more history logs - add a log if watchdog abort leaves uncd_tx_cnt non zero - add a unique (32-bit) number for a each usbnet device and use it for logging - explicitly check for full tx ring in usbnet_start_locked() and return early, which avoids setting the 5 second timer and triggering a watchdog - reset uncd_tx_prod and uncd_tx_cnt to zero in stop
2019-08-23 07:29:28 +03:00
DPRINTF("uno_tx_prepare gave zero length", 0, 0, 0, 0);
Adopt <net/if_stats.h>.
2020-01-29 09:35:28 +03:00
if_statinc(ifp, if_oerrors);
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
break;
}
if (__predict_false(c->unc_xfer == NULL)) {
- remove unused uncd_rx_cnt - add USBNETHIST_CALLARGSN and use it in many places - add more history logs - add a log if watchdog abort leaves uncd_tx_cnt non zero - add a unique (32-bit) number for a each usbnet device and use it for logging - explicitly check for full tx ring in usbnet_start_locked() and return early, which avoids setting the 5 second timer and triggering a watchdog - reset uncd_tx_prod and uncd_tx_cnt to zero in stop
2019-08-23 07:29:28 +03:00
DPRINTF("unc_xfer is NULL", 0, 0, 0, 0);
Adopt <net/if_stats.h>.
2020-01-29 09:35:28 +03:00
if_statinc(ifp, if_oerrors);
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
break;
}
usbd_setup_xfer(c->unc_xfer, c, c->unc_buf, length,
reduce the scope of struct usbnet: - move a large number of members internal to usbnet.c's new "struct usbnet_private". - provide accessors for a few of these - move struct usbnet_cdata into usbnet.c as well, but move bufsz, list count, and xfer flags back out into struct usbnet, and have them set as part of the setup efore usbnet_attach() - split the intr pipe parts into their own structure - move all the main usbnet*lock* code into usbnet.c too usbnet_attach() goes down to 2 args, and the inputs needed are now the full contents of 'struct usbnet' besides the driver owned 'un_flags' and usbnet owned 'un_pri'. welcome netbsd 9.99.6.
2019-08-10 05:17:36 +03:00
un->un_tx_xfer_flags, 10000, usbnet_txeof);
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
/* Transmit */
usbd_status err = usbd_transfer(c->unc_xfer);
if (err != USBD_IN_PROGRESS) {
fix more broken kernhist formats (now I got them all).
2020-03-14 06:01:36 +03:00
DPRINTF("usbd_transfer on %#jx for %ju bytes: %jd",
- remove unused uncd_rx_cnt - add USBNETHIST_CALLARGSN and use it in many places - add more history logs - add a log if watchdog abort leaves uncd_tx_cnt non zero - add a unique (32-bit) number for a each usbnet device and use it for logging - explicitly check for full tx ring in usbnet_start_locked() and return early, which avoids setting the 5 second timer and triggering a watchdog - reset uncd_tx_prod and uncd_tx_cnt to zero in stop
2019-08-23 07:29:28 +03:00
(uintptr_t)c->unc_buf, length, err, 0);
Adopt <net/if_stats.h>.
2020-01-29 09:35:28 +03:00
if_statinc(ifp, if_oerrors);
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
break;
}
- use CALLARGS vs CALLED for better usbhist - turn off usbnetdebug default - log for all entry/exit points of usbnet_pipe_intr() - in usbnet_start_locked() track whether any packet has been transmitted for setting the timer. avoids spurious "watchdog timeouts" - in usbnet_stop() use callout_halt() vs callout_halt, and also stop the usb task. fixes crash of usbtask after the phy has detached. - add a little more defensive checking in the tick task, and add some high-log-level logs. - in usbnet_detach() move the call to usbnet_stop_ifp() above the calls to callout/usbtask stopping. - set ec_mii and unp_pri to NULL when freeing their data
2019-09-08 21:59:32 +03:00
done_transmit = true;
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
IFQ_DEQUEUE(&ifp->if_snd, m);
/*
* If there's a BPF listener, bounce a copy of this frame
* to him.
*/
bpf_mtap(ifp, m, BPF_D_OUT);
m_freem(m);
reduce the scope of struct usbnet: - move a large number of members internal to usbnet.c's new "struct usbnet_private". - provide accessors for a few of these - move struct usbnet_cdata into usbnet.c as well, but move bufsz, list count, and xfer flags back out into struct usbnet, and have them set as part of the setup efore usbnet_attach() - split the intr pipe parts into their own structure - move all the main usbnet*lock* code into usbnet.c too usbnet_attach() goes down to 2 args, and the inputs needed are now the full contents of 'struct usbnet' besides the driver owned 'un_flags' and usbnet owned 'un_pri'. welcome netbsd 9.99.6.
2019-08-10 05:17:36 +03:00
idx = (idx + 1) % un->un_tx_list_cnt;
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
cd->uncd_tx_cnt++;
Sample # of TX packets as entropy source. For RX packets, individual drivers need to be modified.
2021-04-25 08:15:20 +03:00
count++;
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
}
cd->uncd_tx_prod = idx;
fix more broken kernhist formats (now I got them all).
2020-03-14 06:01:36 +03:00
DPRINTF("finished with start; tx_cnt %jd list_cnt %jd link %jd",
- use CALLARGS vs CALLED for better usbhist - turn off usbnetdebug default - log for all entry/exit points of usbnet_pipe_intr() - in usbnet_start_locked() track whether any packet has been transmitted for setting the timer. avoids spurious "watchdog timeouts" - in usbnet_stop() use callout_halt() vs callout_halt, and also stop the usb task. fixes crash of usbtask after the phy has detached. - add a little more defensive checking in the tick task, and add some high-log-level logs. - in usbnet_detach() move the call to usbnet_stop_ifp() above the calls to callout/usbtask stopping. - set ec_mii and unp_pri to NULL when freeing their data
2019-09-08 21:59:32 +03:00
cd->uncd_tx_cnt, un->un_tx_list_cnt, unp->unp_link, 0);
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
/*
* Set a timeout in case the chip goes out to lunch.
*/
- use CALLARGS vs CALLED for better usbhist - turn off usbnetdebug default - log for all entry/exit points of usbnet_pipe_intr() - in usbnet_start_locked() track whether any packet has been transmitted for setting the timer. avoids spurious "watchdog timeouts" - in usbnet_stop() use callout_halt() vs callout_halt, and also stop the usb task. fixes crash of usbtask after the phy has detached. - add a little more defensive checking in the tick task, and add some high-log-level logs. - in usbnet_detach() move the call to usbnet_stop_ifp() above the calls to callout/usbtask stopping. - set ec_mii and unp_pri to NULL when freeing their data
2019-09-08 21:59:32 +03:00
if (done_transmit)
unp->unp_timer = 5;
Sample # of TX packets as entropy source. For RX packets, individual drivers need to be modified.
2021-04-25 08:15:20 +03:00
if (count != 0)
rnd_add_uint32(&unp->unp_rndsrc, count);
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
}
static void
Define and implement a locking protocol for the ifmedia / mii layers: - MP-safe drivers provide a mutex to ifmedia that is used to serialize access to media-related structures / hardware regsiters. Converted drivers use the new ifmedia_init_with_lock() function for this. The new name is provided to ease the transition. - Un-converted drivers continue to call ifmedia_init(), which will supply a compatibility lock to be used instead. Several media-related entry points must be aware of this compatibility lock, and are able to acquire it recursively a limited number of times, if needed. This is a SPIN mutex with priority IPL_NET. - This same lock is used to serialize access to PHY registers and other MII-related data structures. The PHY drivers are modified to acquire and release the lock, as needed, and assert the lock is held as a diagnostic aid. The "usbnet" framework has had an overhaul of its internal locking protocols to fit in with the media / mii changes, and the drivers adapted. USB wifi drivers have been changed to provide their own adaptive mutex to the ifmedia later via a new ieee80211_media_init_with_lock() function. This is required because the USB drivers need an adaptive mutex. Besised "usbnet", a few other drivers are converted: vmx, wm, ixgbe / ixv. mcx also now calls ifmedia_init_with_lock() because it needs to also use an adaptive mutex. The mcx driver still needs to be fully converted to NET_MPSAFE.
2020-03-16 02:04:50 +03:00
usbnet_if_start(struct ifnet *ifp)
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
{
struct usbnet * const un = ifp->if_softc;
reduce the scope of struct usbnet: - move a large number of members internal to usbnet.c's new "struct usbnet_private". - provide accessors for a few of these - move struct usbnet_cdata into usbnet.c as well, but move bufsz, list count, and xfer flags back out into struct usbnet, and have them set as part of the setup efore usbnet_attach() - split the intr pipe parts into their own structure - move all the main usbnet*lock* code into usbnet.c too usbnet_attach() goes down to 2 args, and the inputs needed are now the full contents of 'struct usbnet' besides the driver owned 'un_flags' and usbnet owned 'un_pri'. welcome netbsd 9.99.6.
2019-08-10 05:17:36 +03:00
struct usbnet_private * const unp = un->un_pri;
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
- use CALLARGS vs CALLED for better usbhist - turn off usbnetdebug default - log for all entry/exit points of usbnet_pipe_intr() - in usbnet_start_locked() track whether any packet has been transmitted for setting the timer. avoids spurious "watchdog timeouts" - in usbnet_stop() use callout_halt() vs callout_halt, and also stop the usb task. fixes crash of usbtask after the phy has detached. - add a little more defensive checking in the tick task, and add some high-log-level logs. - in usbnet_detach() move the call to usbnet_stop_ifp() above the calls to callout/usbtask stopping. - set ec_mii and unp_pri to NULL when freeing their data
2019-09-08 21:59:32 +03:00
USBNETHIST_FUNC();
PR/55068: sc.dying: Fix printf formats: - no %s/%p for kernel log - 0x% -> %# - always %j for kernel log
2020-03-13 21:17:40 +03:00
USBNETHIST_CALLARGS("%jd: stopping %jd",
- use CALLARGS vs CALLED for better usbhist - turn off usbnetdebug default - log for all entry/exit points of usbnet_pipe_intr() - in usbnet_start_locked() track whether any packet has been transmitted for setting the timer. avoids spurious "watchdog timeouts" - in usbnet_stop() use callout_halt() vs callout_halt, and also stop the usb task. fixes crash of usbtask after the phy has detached. - add a little more defensive checking in the tick task, and add some high-log-level logs. - in usbnet_detach() move the call to usbnet_stop_ifp() above the calls to callout/usbtask stopping. - set ec_mii and unp_pri to NULL when freeing their data
2019-09-08 21:59:32 +03:00
unp->unp_number, unp->unp_stopping, 0, 0);
reduce the scope of struct usbnet: - move a large number of members internal to usbnet.c's new "struct usbnet_private". - provide accessors for a few of these - move struct usbnet_cdata into usbnet.c as well, but move bufsz, list count, and xfer flags back out into struct usbnet, and have them set as part of the setup efore usbnet_attach() - split the intr pipe parts into their own structure - move all the main usbnet*lock* code into usbnet.c too usbnet_attach() goes down to 2 args, and the inputs needed are now the full contents of 'struct usbnet' besides the driver owned 'un_flags' and usbnet owned 'un_pri'. welcome netbsd 9.99.6.
2019-08-10 05:17:36 +03:00
mutex_enter(&unp->unp_txlock);
if (!unp->unp_stopping)
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
usbnet_start_locked(ifp);
reduce the scope of struct usbnet: - move a large number of members internal to usbnet.c's new "struct usbnet_private". - provide accessors for a few of these - move struct usbnet_cdata into usbnet.c as well, but move bufsz, list count, and xfer flags back out into struct usbnet, and have them set as part of the setup efore usbnet_attach() - split the intr pipe parts into their own structure - move all the main usbnet*lock* code into usbnet.c too usbnet_attach() goes down to 2 args, and the inputs needed are now the full contents of 'struct usbnet' besides the driver owned 'un_flags' and usbnet owned 'un_pri'. welcome netbsd 9.99.6.
2019-08-10 05:17:36 +03:00
mutex_exit(&unp->unp_txlock);
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
}
/*
* Chain management.
*
* RX and TX are identical. Keep them that way.
*/
/* Start of common RX functions */
static size_t
- use usbnet_isowned*() more - more const
2019-08-11 04:31:19 +03:00
usbnet_rx_list_size(struct usbnet_cdata * const cd, struct usbnet * const un)
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
{
reduce the scope of struct usbnet: - move a large number of members internal to usbnet.c's new "struct usbnet_private". - provide accessors for a few of these - move struct usbnet_cdata into usbnet.c as well, but move bufsz, list count, and xfer flags back out into struct usbnet, and have them set as part of the setup efore usbnet_attach() - split the intr pipe parts into their own structure - move all the main usbnet*lock* code into usbnet.c too usbnet_attach() goes down to 2 args, and the inputs needed are now the full contents of 'struct usbnet' besides the driver owned 'un_flags' and usbnet owned 'un_pri'. welcome netbsd 9.99.6.
2019-08-10 05:17:36 +03:00
return sizeof(*cd->uncd_rx_chain) * un->un_rx_list_cnt;
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
}
static void
- use usbnet_isowned*() more - more const
2019-08-11 04:31:19 +03:00
usbnet_rx_list_alloc(struct usbnet * const un)
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
{
reduce the scope of struct usbnet: - move a large number of members internal to usbnet.c's new "struct usbnet_private". - provide accessors for a few of these - move struct usbnet_cdata into usbnet.c as well, but move bufsz, list count, and xfer flags back out into struct usbnet, and have them set as part of the setup efore usbnet_attach() - split the intr pipe parts into their own structure - move all the main usbnet*lock* code into usbnet.c too usbnet_attach() goes down to 2 args, and the inputs needed are now the full contents of 'struct usbnet' besides the driver owned 'un_flags' and usbnet owned 'un_pri'. welcome netbsd 9.99.6.
2019-08-10 05:17:36 +03:00
struct usbnet_cdata * const cd = un_cdata(un);
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
reduce the scope of struct usbnet: - move a large number of members internal to usbnet.c's new "struct usbnet_private". - provide accessors for a few of these - move struct usbnet_cdata into usbnet.c as well, but move bufsz, list count, and xfer flags back out into struct usbnet, and have them set as part of the setup efore usbnet_attach() - split the intr pipe parts into their own structure - move all the main usbnet*lock* code into usbnet.c too usbnet_attach() goes down to 2 args, and the inputs needed are now the full contents of 'struct usbnet' besides the driver owned 'un_flags' and usbnet owned 'un_pri'. welcome netbsd 9.99.6.
2019-08-10 05:17:36 +03:00
cd->uncd_rx_chain = kmem_zalloc(usbnet_rx_list_size(cd, un), KM_SLEEP);
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
}
static void
- use usbnet_isowned*() more - more const
2019-08-11 04:31:19 +03:00
usbnet_rx_list_free(struct usbnet * const un)
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
{
reduce the scope of struct usbnet: - move a large number of members internal to usbnet.c's new "struct usbnet_private". - provide accessors for a few of these - move struct usbnet_cdata into usbnet.c as well, but move bufsz, list count, and xfer flags back out into struct usbnet, and have them set as part of the setup efore usbnet_attach() - split the intr pipe parts into their own structure - move all the main usbnet*lock* code into usbnet.c too usbnet_attach() goes down to 2 args, and the inputs needed are now the full contents of 'struct usbnet' besides the driver owned 'un_flags' and usbnet owned 'un_pri'. welcome netbsd 9.99.6.
2019-08-10 05:17:36 +03:00
struct usbnet_cdata * const cd = un_cdata(un);
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
if (cd->uncd_rx_chain) {
reduce the scope of struct usbnet: - move a large number of members internal to usbnet.c's new "struct usbnet_private". - provide accessors for a few of these - move struct usbnet_cdata into usbnet.c as well, but move bufsz, list count, and xfer flags back out into struct usbnet, and have them set as part of the setup efore usbnet_attach() - split the intr pipe parts into their own structure - move all the main usbnet*lock* code into usbnet.c too usbnet_attach() goes down to 2 args, and the inputs needed are now the full contents of 'struct usbnet' besides the driver owned 'un_flags' and usbnet owned 'un_pri'. welcome netbsd 9.99.6.
2019-08-10 05:17:36 +03:00
kmem_free(cd->uncd_rx_chain, usbnet_rx_list_size(cd, un));
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
cd->uncd_rx_chain = NULL;
}
}
static int
- use usbnet_isowned*() more - more const
2019-08-11 04:31:19 +03:00
usbnet_rx_list_init(struct usbnet * const un)
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
{
reduce the scope of struct usbnet: - move a large number of members internal to usbnet.c's new "struct usbnet_private". - provide accessors for a few of these - move struct usbnet_cdata into usbnet.c as well, but move bufsz, list count, and xfer flags back out into struct usbnet, and have them set as part of the setup efore usbnet_attach() - split the intr pipe parts into their own structure - move all the main usbnet*lock* code into usbnet.c too usbnet_attach() goes down to 2 args, and the inputs needed are now the full contents of 'struct usbnet' besides the driver owned 'un_flags' and usbnet owned 'un_pri'. welcome netbsd 9.99.6.
2019-08-10 05:17:36 +03:00
struct usbnet_cdata * const cd = un_cdata(un);
struct usbnet_private * const unp = un->un_pri;
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
reduce the scope of struct usbnet: - move a large number of members internal to usbnet.c's new "struct usbnet_private". - provide accessors for a few of these - move struct usbnet_cdata into usbnet.c as well, but move bufsz, list count, and xfer flags back out into struct usbnet, and have them set as part of the setup efore usbnet_attach() - split the intr pipe parts into their own structure - move all the main usbnet*lock* code into usbnet.c too usbnet_attach() goes down to 2 args, and the inputs needed are now the full contents of 'struct usbnet' besides the driver owned 'un_flags' and usbnet owned 'un_pri'. welcome netbsd 9.99.6.
2019-08-10 05:17:36 +03:00
for (size_t i = 0; i < un->un_rx_list_cnt; i++) {
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
struct usbnet_chain *c = &cd->uncd_rx_chain[i];
c->unc_un = un;
if (c->unc_xfer == NULL) {
reduce the scope of struct usbnet: - move a large number of members internal to usbnet.c's new "struct usbnet_private". - provide accessors for a few of these - move struct usbnet_cdata into usbnet.c as well, but move bufsz, list count, and xfer flags back out into struct usbnet, and have them set as part of the setup efore usbnet_attach() - split the intr pipe parts into their own structure - move all the main usbnet*lock* code into usbnet.c too usbnet_attach() goes down to 2 args, and the inputs needed are now the full contents of 'struct usbnet' besides the driver owned 'un_flags' and usbnet owned 'un_pri'. welcome netbsd 9.99.6.
2019-08-10 05:17:36 +03:00
int err = usbd_create_xfer(unp->unp_ep[USBNET_ENDPT_RX],
un->un_rx_bufsz, un->un_rx_xfer_flags, 0,
update usbnet some: - move rx/tx xfer flags into usbnet_cdata - move the callbacks into usbnet_ops structure - move rx/tx xfer flags arguments from usbnet_init_rx_tx() and move them all into usbnet_attach() arguments - s/miibus/mii/ in some places for consistency other clean up: - create wrapper functions for callbacks, move knowledge about special handling (OK to be missing, error eating) there. - use cdata pointer if already available - provide some more macros (will be real functions later) for accessing usbnet members, use existing ones more bump kernel version.
2019-08-09 04:17:33 +03:00
&c->unc_xfer);
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
if (err)
return err;
c->unc_buf = usbd_get_buffer(c->unc_xfer);
}
}
return 0;
}
static void
- use usbnet_isowned*() more - more const
2019-08-11 04:31:19 +03:00
usbnet_rx_list_fini(struct usbnet * const un)
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
{
reduce the scope of struct usbnet: - move a large number of members internal to usbnet.c's new "struct usbnet_private". - provide accessors for a few of these - move struct usbnet_cdata into usbnet.c as well, but move bufsz, list count, and xfer flags back out into struct usbnet, and have them set as part of the setup efore usbnet_attach() - split the intr pipe parts into their own structure - move all the main usbnet*lock* code into usbnet.c too usbnet_attach() goes down to 2 args, and the inputs needed are now the full contents of 'struct usbnet' besides the driver owned 'un_flags' and usbnet owned 'un_pri'. welcome netbsd 9.99.6.
2019-08-10 05:17:36 +03:00
struct usbnet_cdata * const cd = un_cdata(un);
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
reduce the scope of struct usbnet: - move a large number of members internal to usbnet.c's new "struct usbnet_private". - provide accessors for a few of these - move struct usbnet_cdata into usbnet.c as well, but move bufsz, list count, and xfer flags back out into struct usbnet, and have them set as part of the setup efore usbnet_attach() - split the intr pipe parts into their own structure - move all the main usbnet*lock* code into usbnet.c too usbnet_attach() goes down to 2 args, and the inputs needed are now the full contents of 'struct usbnet' besides the driver owned 'un_flags' and usbnet owned 'un_pri'. welcome netbsd 9.99.6.
2019-08-10 05:17:36 +03:00
for (size_t i = 0; i < un->un_rx_list_cnt; i++) {
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
struct usbnet_chain *c = &cd->uncd_rx_chain[i];
if (c->unc_xfer != NULL) {
usbd_destroy_xfer(c->unc_xfer);
c->unc_xfer = NULL;
c->unc_buf = NULL;
}
}
}
/* End of common RX functions */
static void
internal to usbnet: add buflen param to usbnet_newbuf(). use this to skip allocating an mbuf cluster for small packets (ported from kue(4).) remove usbnet_rx_start_pipes()'s always usbnet_rxeof argument.
2019-08-16 11:38:21 +03:00
usbnet_rx_start_pipes(struct usbnet * const un)
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
{
reduce the scope of struct usbnet: - move a large number of members internal to usbnet.c's new "struct usbnet_private". - provide accessors for a few of these - move struct usbnet_cdata into usbnet.c as well, but move bufsz, list count, and xfer flags back out into struct usbnet, and have them set as part of the setup efore usbnet_attach() - split the intr pipe parts into their own structure - move all the main usbnet*lock* code into usbnet.c too usbnet_attach() goes down to 2 args, and the inputs needed are now the full contents of 'struct usbnet' besides the driver owned 'un_flags' and usbnet owned 'un_pri'. welcome netbsd 9.99.6.
2019-08-10 05:17:36 +03:00
struct usbnet_cdata * const cd = un_cdata(un);
struct usbnet_private * const unp = un->un_pri;
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
reduce the scope of struct usbnet: - move a large number of members internal to usbnet.c's new "struct usbnet_private". - provide accessors for a few of these - move struct usbnet_cdata into usbnet.c as well, but move bufsz, list count, and xfer flags back out into struct usbnet, and have them set as part of the setup efore usbnet_attach() - split the intr pipe parts into their own structure - move all the main usbnet*lock* code into usbnet.c too usbnet_attach() goes down to 2 args, and the inputs needed are now the full contents of 'struct usbnet' besides the driver owned 'un_flags' and usbnet owned 'un_pri'. welcome netbsd 9.99.6.
2019-08-10 05:17:36 +03:00
mutex_enter(&unp->unp_rxlock);
mutex_enter(&unp->unp_txlock);
unp->unp_stopping = false;
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
reduce the scope of struct usbnet: - move a large number of members internal to usbnet.c's new "struct usbnet_private". - provide accessors for a few of these - move struct usbnet_cdata into usbnet.c as well, but move bufsz, list count, and xfer flags back out into struct usbnet, and have them set as part of the setup efore usbnet_attach() - split the intr pipe parts into their own structure - move all the main usbnet*lock* code into usbnet.c too usbnet_attach() goes down to 2 args, and the inputs needed are now the full contents of 'struct usbnet' besides the driver owned 'un_flags' and usbnet owned 'un_pri'. welcome netbsd 9.99.6.
2019-08-10 05:17:36 +03:00
for (size_t i = 0; i < un->un_rx_list_cnt; i++) {
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
struct usbnet_chain *c = &cd->uncd_rx_chain[i];
reduce the scope of struct usbnet: - move a large number of members internal to usbnet.c's new "struct usbnet_private". - provide accessors for a few of these - move struct usbnet_cdata into usbnet.c as well, but move bufsz, list count, and xfer flags back out into struct usbnet, and have them set as part of the setup efore usbnet_attach() - split the intr pipe parts into their own structure - move all the main usbnet*lock* code into usbnet.c too usbnet_attach() goes down to 2 args, and the inputs needed are now the full contents of 'struct usbnet' besides the driver owned 'un_flags' and usbnet owned 'un_pri'. welcome netbsd 9.99.6.
2019-08-10 05:17:36 +03:00
usbd_setup_xfer(c->unc_xfer, c, c->unc_buf, un->un_rx_bufsz,
internal to usbnet: add buflen param to usbnet_newbuf(). use this to skip allocating an mbuf cluster for small packets (ported from kue(4).) remove usbnet_rx_start_pipes()'s always usbnet_rxeof argument.
2019-08-16 11:38:21 +03:00
un->un_rx_xfer_flags, USBD_NO_TIMEOUT, usbnet_rxeof);
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
usbd_transfer(c->unc_xfer);
Trailing white space
2019-08-03 18:58:14 +03:00
}
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
reduce the scope of struct usbnet: - move a large number of members internal to usbnet.c's new "struct usbnet_private". - provide accessors for a few of these - move struct usbnet_cdata into usbnet.c as well, but move bufsz, list count, and xfer flags back out into struct usbnet, and have them set as part of the setup efore usbnet_attach() - split the intr pipe parts into their own structure - move all the main usbnet*lock* code into usbnet.c too usbnet_attach() goes down to 2 args, and the inputs needed are now the full contents of 'struct usbnet' besides the driver owned 'un_flags' and usbnet owned 'un_pri'. welcome netbsd 9.99.6.
2019-08-10 05:17:36 +03:00
mutex_exit(&unp->unp_txlock);
mutex_exit(&unp->unp_rxlock);
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
}
/* Start of common TX functions */
static size_t
- use usbnet_isowned*() more - more const
2019-08-11 04:31:19 +03:00
usbnet_tx_list_size(struct usbnet_cdata * const cd, struct usbnet * const un)
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
{
reduce the scope of struct usbnet: - move a large number of members internal to usbnet.c's new "struct usbnet_private". - provide accessors for a few of these - move struct usbnet_cdata into usbnet.c as well, but move bufsz, list count, and xfer flags back out into struct usbnet, and have them set as part of the setup efore usbnet_attach() - split the intr pipe parts into their own structure - move all the main usbnet*lock* code into usbnet.c too usbnet_attach() goes down to 2 args, and the inputs needed are now the full contents of 'struct usbnet' besides the driver owned 'un_flags' and usbnet owned 'un_pri'. welcome netbsd 9.99.6.
2019-08-10 05:17:36 +03:00
return sizeof(*cd->uncd_tx_chain) * un->un_tx_list_cnt;
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
}
static void
- use usbnet_isowned*() more - more const
2019-08-11 04:31:19 +03:00
usbnet_tx_list_alloc(struct usbnet * const un)
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
{
reduce the scope of struct usbnet: - move a large number of members internal to usbnet.c's new "struct usbnet_private". - provide accessors for a few of these - move struct usbnet_cdata into usbnet.c as well, but move bufsz, list count, and xfer flags back out into struct usbnet, and have them set as part of the setup efore usbnet_attach() - split the intr pipe parts into their own structure - move all the main usbnet*lock* code into usbnet.c too usbnet_attach() goes down to 2 args, and the inputs needed are now the full contents of 'struct usbnet' besides the driver owned 'un_flags' and usbnet owned 'un_pri'. welcome netbsd 9.99.6.
2019-08-10 05:17:36 +03:00
struct usbnet_cdata * const cd = un_cdata(un);
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
reduce the scope of struct usbnet: - move a large number of members internal to usbnet.c's new "struct usbnet_private". - provide accessors for a few of these - move struct usbnet_cdata into usbnet.c as well, but move bufsz, list count, and xfer flags back out into struct usbnet, and have them set as part of the setup efore usbnet_attach() - split the intr pipe parts into their own structure - move all the main usbnet*lock* code into usbnet.c too usbnet_attach() goes down to 2 args, and the inputs needed are now the full contents of 'struct usbnet' besides the driver owned 'un_flags' and usbnet owned 'un_pri'. welcome netbsd 9.99.6.
2019-08-10 05:17:36 +03:00
cd->uncd_tx_chain = kmem_zalloc(usbnet_tx_list_size(cd, un), KM_SLEEP);
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
}
static void
- use usbnet_isowned*() more - more const
2019-08-11 04:31:19 +03:00
usbnet_tx_list_free(struct usbnet * const un)
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
{
reduce the scope of struct usbnet: - move a large number of members internal to usbnet.c's new "struct usbnet_private". - provide accessors for a few of these - move struct usbnet_cdata into usbnet.c as well, but move bufsz, list count, and xfer flags back out into struct usbnet, and have them set as part of the setup efore usbnet_attach() - split the intr pipe parts into their own structure - move all the main usbnet*lock* code into usbnet.c too usbnet_attach() goes down to 2 args, and the inputs needed are now the full contents of 'struct usbnet' besides the driver owned 'un_flags' and usbnet owned 'un_pri'. welcome netbsd 9.99.6.
2019-08-10 05:17:36 +03:00
struct usbnet_cdata * const cd = un_cdata(un);
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
if (cd->uncd_tx_chain) {
reduce the scope of struct usbnet: - move a large number of members internal to usbnet.c's new "struct usbnet_private". - provide accessors for a few of these - move struct usbnet_cdata into usbnet.c as well, but move bufsz, list count, and xfer flags back out into struct usbnet, and have them set as part of the setup efore usbnet_attach() - split the intr pipe parts into their own structure - move all the main usbnet*lock* code into usbnet.c too usbnet_attach() goes down to 2 args, and the inputs needed are now the full contents of 'struct usbnet' besides the driver owned 'un_flags' and usbnet owned 'un_pri'. welcome netbsd 9.99.6.
2019-08-10 05:17:36 +03:00
kmem_free(cd->uncd_tx_chain, usbnet_tx_list_size(cd, un));
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
cd->uncd_tx_chain = NULL;
}
}
static int
- use usbnet_isowned*() more - more const
2019-08-11 04:31:19 +03:00
usbnet_tx_list_init(struct usbnet * const un)
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
{
reduce the scope of struct usbnet: - move a large number of members internal to usbnet.c's new "struct usbnet_private". - provide accessors for a few of these - move struct usbnet_cdata into usbnet.c as well, but move bufsz, list count, and xfer flags back out into struct usbnet, and have them set as part of the setup efore usbnet_attach() - split the intr pipe parts into their own structure - move all the main usbnet*lock* code into usbnet.c too usbnet_attach() goes down to 2 args, and the inputs needed are now the full contents of 'struct usbnet' besides the driver owned 'un_flags' and usbnet owned 'un_pri'. welcome netbsd 9.99.6.
2019-08-10 05:17:36 +03:00
struct usbnet_cdata * const cd = un_cdata(un);
struct usbnet_private * const unp = un->un_pri;
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
reduce the scope of struct usbnet: - move a large number of members internal to usbnet.c's new "struct usbnet_private". - provide accessors for a few of these - move struct usbnet_cdata into usbnet.c as well, but move bufsz, list count, and xfer flags back out into struct usbnet, and have them set as part of the setup efore usbnet_attach() - split the intr pipe parts into their own structure - move all the main usbnet*lock* code into usbnet.c too usbnet_attach() goes down to 2 args, and the inputs needed are now the full contents of 'struct usbnet' besides the driver owned 'un_flags' and usbnet owned 'un_pri'. welcome netbsd 9.99.6.
2019-08-10 05:17:36 +03:00
for (size_t i = 0; i < un->un_tx_list_cnt; i++) {
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
struct usbnet_chain *c = &cd->uncd_tx_chain[i];
c->unc_un = un;
if (c->unc_xfer == NULL) {
reduce the scope of struct usbnet: - move a large number of members internal to usbnet.c's new "struct usbnet_private". - provide accessors for a few of these - move struct usbnet_cdata into usbnet.c as well, but move bufsz, list count, and xfer flags back out into struct usbnet, and have them set as part of the setup efore usbnet_attach() - split the intr pipe parts into their own structure - move all the main usbnet*lock* code into usbnet.c too usbnet_attach() goes down to 2 args, and the inputs needed are now the full contents of 'struct usbnet' besides the driver owned 'un_flags' and usbnet owned 'un_pri'. welcome netbsd 9.99.6.
2019-08-10 05:17:36 +03:00
int err = usbd_create_xfer(unp->unp_ep[USBNET_ENDPT_TX],
un->un_tx_bufsz, un->un_tx_xfer_flags, 0,
update usbnet some: - move rx/tx xfer flags into usbnet_cdata - move the callbacks into usbnet_ops structure - move rx/tx xfer flags arguments from usbnet_init_rx_tx() and move them all into usbnet_attach() arguments - s/miibus/mii/ in some places for consistency other clean up: - create wrapper functions for callbacks, move knowledge about special handling (OK to be missing, error eating) there. - use cdata pointer if already available - provide some more macros (will be real functions later) for accessing usbnet members, use existing ones more bump kernel version.
2019-08-09 04:17:33 +03:00
&c->unc_xfer);
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
if (err)
return err;
c->unc_buf = usbd_get_buffer(c->unc_xfer);
}
}
return 0;
}
static void
- use usbnet_isowned*() more - more const
2019-08-11 04:31:19 +03:00
usbnet_tx_list_fini(struct usbnet * const un)
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
{
reduce the scope of struct usbnet: - move a large number of members internal to usbnet.c's new "struct usbnet_private". - provide accessors for a few of these - move struct usbnet_cdata into usbnet.c as well, but move bufsz, list count, and xfer flags back out into struct usbnet, and have them set as part of the setup efore usbnet_attach() - split the intr pipe parts into their own structure - move all the main usbnet*lock* code into usbnet.c too usbnet_attach() goes down to 2 args, and the inputs needed are now the full contents of 'struct usbnet' besides the driver owned 'un_flags' and usbnet owned 'un_pri'. welcome netbsd 9.99.6.
2019-08-10 05:17:36 +03:00
struct usbnet_cdata * const cd = un_cdata(un);
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
reduce the scope of struct usbnet: - move a large number of members internal to usbnet.c's new "struct usbnet_private". - provide accessors for a few of these - move struct usbnet_cdata into usbnet.c as well, but move bufsz, list count, and xfer flags back out into struct usbnet, and have them set as part of the setup efore usbnet_attach() - split the intr pipe parts into their own structure - move all the main usbnet*lock* code into usbnet.c too usbnet_attach() goes down to 2 args, and the inputs needed are now the full contents of 'struct usbnet' besides the driver owned 'un_flags' and usbnet owned 'un_pri'. welcome netbsd 9.99.6.
2019-08-10 05:17:36 +03:00
for (size_t i = 0; i < un->un_tx_list_cnt; i++) {
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
struct usbnet_chain *c = &cd->uncd_tx_chain[i];
if (c->unc_xfer != NULL) {
usbd_destroy_xfer(c->unc_xfer);
c->unc_xfer = NULL;
c->unc_buf = NULL;
}
}
- remove unused uncd_rx_cnt - add USBNETHIST_CALLARGSN and use it in many places - add more history logs - add a log if watchdog abort leaves uncd_tx_cnt non zero - add a unique (32-bit) number for a each usbnet device and use it for logging - explicitly check for full tx ring in usbnet_start_locked() and return early, which avoids setting the 5 second timer and triggering a watchdog - reset uncd_tx_prod and uncd_tx_cnt to zero in stop
2019-08-23 07:29:28 +03:00
cd->uncd_tx_prod = cd->uncd_tx_cnt = 0;
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
}
/* End of common TX functions */
/* Endpoint pipe management. */
static void
- use usbnet_isowned*() more - more const
2019-08-11 04:31:19 +03:00
usbnet_ep_close_pipes(struct usbnet * const un)
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
{
reduce the scope of struct usbnet: - move a large number of members internal to usbnet.c's new "struct usbnet_private". - provide accessors for a few of these - move struct usbnet_cdata into usbnet.c as well, but move bufsz, list count, and xfer flags back out into struct usbnet, and have them set as part of the setup efore usbnet_attach() - split the intr pipe parts into their own structure - move all the main usbnet*lock* code into usbnet.c too usbnet_attach() goes down to 2 args, and the inputs needed are now the full contents of 'struct usbnet' besides the driver owned 'un_flags' and usbnet owned 'un_pri'. welcome netbsd 9.99.6.
2019-08-10 05:17:36 +03:00
struct usbnet_private * const unp = un->un_pri;
for (size_t i = 0; i < __arraycount(unp->unp_ep); i++) {
if (unp->unp_ep[i] == NULL)
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
continue;
reduce the scope of struct usbnet: - move a large number of members internal to usbnet.c's new "struct usbnet_private". - provide accessors for a few of these - move struct usbnet_cdata into usbnet.c as well, but move bufsz, list count, and xfer flags back out into struct usbnet, and have them set as part of the setup efore usbnet_attach() - split the intr pipe parts into their own structure - move all the main usbnet*lock* code into usbnet.c too usbnet_attach() goes down to 2 args, and the inputs needed are now the full contents of 'struct usbnet' besides the driver owned 'un_flags' and usbnet owned 'un_pri'. welcome netbsd 9.99.6.
2019-08-10 05:17:36 +03:00
usbd_status err = usbd_close_pipe(unp->unp_ep[i]);
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
if (err)
aprint_error_dev(un->un_dev, "close pipe %zu: %s\n", i,
usbd_errstr(err));
reduce the scope of struct usbnet: - move a large number of members internal to usbnet.c's new "struct usbnet_private". - provide accessors for a few of these - move struct usbnet_cdata into usbnet.c as well, but move bufsz, list count, and xfer flags back out into struct usbnet, and have them set as part of the setup efore usbnet_attach() - split the intr pipe parts into their own structure - move all the main usbnet*lock* code into usbnet.c too usbnet_attach() goes down to 2 args, and the inputs needed are now the full contents of 'struct usbnet' besides the driver owned 'un_flags' and usbnet owned 'un_pri'. welcome netbsd 9.99.6.
2019-08-10 05:17:36 +03:00
unp->unp_ep[i] = NULL;
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
}
}
static usbd_status
- use usbnet_isowned*() more - more const
2019-08-11 04:31:19 +03:00
usbnet_ep_open_pipes(struct usbnet * const un)
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
{
reduce the scope of struct usbnet: - move a large number of members internal to usbnet.c's new "struct usbnet_private". - provide accessors for a few of these - move struct usbnet_cdata into usbnet.c as well, but move bufsz, list count, and xfer flags back out into struct usbnet, and have them set as part of the setup efore usbnet_attach() - split the intr pipe parts into their own structure - move all the main usbnet*lock* code into usbnet.c too usbnet_attach() goes down to 2 args, and the inputs needed are now the full contents of 'struct usbnet' besides the driver owned 'un_flags' and usbnet owned 'un_pri'. welcome netbsd 9.99.6.
2019-08-10 05:17:36 +03:00
struct usbnet_intr * const uni = un->un_intr;
struct usbnet_private * const unp = un->un_pri;
for (size_t i = 0; i < __arraycount(unp->unp_ep); i++) {
- adjust usbnet interface to allow usbd_open_pipe_intr(), from the new comment: * if un_intr_buf is not NULL, use usbd_open_pipe_intr() not * usbd_open_pipe() for USBNET_ENDPT_INTR, with this buffer, * size, and interval. the standard handling is in usbnet.c, with a callback to deal with the interrupt it self. not fully tested, designed for if_aue.c and a few others not yet converted. - make usbhist for usbnet.c work, thanks paulg - usbnet_init_rx_tx() clears out all allocations upon failure now - add usbnet_ec() to get a pointer to the struct ethercom - add usbnet_{lock,unlock,owned}*() to lock/unlock the various locks and *owned*() for asserting welcome 9.99.3!
2019-08-04 11:59:13 +03:00
usbd_status err;
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
if (un->un_ed[i] == 0)
continue;
- adjust usbnet interface to allow usbd_open_pipe_intr(), from the new comment: * if un_intr_buf is not NULL, use usbd_open_pipe_intr() not * usbd_open_pipe() for USBNET_ENDPT_INTR, with this buffer, * size, and interval. the standard handling is in usbnet.c, with a callback to deal with the interrupt it self. not fully tested, designed for if_aue.c and a few others not yet converted. - make usbhist for usbnet.c work, thanks paulg - usbnet_init_rx_tx() clears out all allocations upon failure now - add usbnet_ec() to get a pointer to the struct ethercom - add usbnet_{lock,unlock,owned}*() to lock/unlock the various locks and *owned*() for asserting welcome 9.99.3!
2019-08-04 11:59:13 +03:00
reduce the scope of struct usbnet: - move a large number of members internal to usbnet.c's new "struct usbnet_private". - provide accessors for a few of these - move struct usbnet_cdata into usbnet.c as well, but move bufsz, list count, and xfer flags back out into struct usbnet, and have them set as part of the setup efore usbnet_attach() - split the intr pipe parts into their own structure - move all the main usbnet*lock* code into usbnet.c too usbnet_attach() goes down to 2 args, and the inputs needed are now the full contents of 'struct usbnet' besides the driver owned 'un_flags' and usbnet owned 'un_pri'. welcome netbsd 9.99.6.
2019-08-10 05:17:36 +03:00
if (i == USBNET_ENDPT_INTR && uni) {
- adjust usbnet interface to allow usbd_open_pipe_intr(), from the new comment: * if un_intr_buf is not NULL, use usbd_open_pipe_intr() not * usbd_open_pipe() for USBNET_ENDPT_INTR, with this buffer, * size, and interval. the standard handling is in usbnet.c, with a callback to deal with the interrupt it self. not fully tested, designed for if_aue.c and a few others not yet converted. - make usbhist for usbnet.c work, thanks paulg - usbnet_init_rx_tx() clears out all allocations upon failure now - add usbnet_ec() to get a pointer to the struct ethercom - add usbnet_{lock,unlock,owned}*() to lock/unlock the various locks and *owned*() for asserting welcome 9.99.3!
2019-08-04 11:59:13 +03:00
err = usbd_open_pipe_intr(un->un_iface, un->un_ed[i],
reduce the scope of struct usbnet: - move a large number of members internal to usbnet.c's new "struct usbnet_private". - provide accessors for a few of these - move struct usbnet_cdata into usbnet.c as well, but move bufsz, list count, and xfer flags back out into struct usbnet, and have them set as part of the setup efore usbnet_attach() - split the intr pipe parts into their own structure - move all the main usbnet*lock* code into usbnet.c too usbnet_attach() goes down to 2 args, and the inputs needed are now the full contents of 'struct usbnet' besides the driver owned 'un_flags' and usbnet owned 'un_pri'. welcome netbsd 9.99.6.
2019-08-10 05:17:36 +03:00
USBD_EXCLUSIVE_USE | USBD_MPSAFE, &unp->unp_ep[i], un,
uni->uni_buf, uni->uni_bufsz, usbnet_pipe_intr,
uni->uni_interval);
- adjust usbnet interface to allow usbd_open_pipe_intr(), from the new comment: * if un_intr_buf is not NULL, use usbd_open_pipe_intr() not * usbd_open_pipe() for USBNET_ENDPT_INTR, with this buffer, * size, and interval. the standard handling is in usbnet.c, with a callback to deal with the interrupt it self. not fully tested, designed for if_aue.c and a few others not yet converted. - make usbhist for usbnet.c work, thanks paulg - usbnet_init_rx_tx() clears out all allocations upon failure now - add usbnet_ec() to get a pointer to the struct ethercom - add usbnet_{lock,unlock,owned}*() to lock/unlock the various locks and *owned*() for asserting welcome 9.99.3!
2019-08-04 11:59:13 +03:00
} else {
err = usbd_open_pipe(un->un_iface, un->un_ed[i],
reduce the scope of struct usbnet: - move a large number of members internal to usbnet.c's new "struct usbnet_private". - provide accessors for a few of these - move struct usbnet_cdata into usbnet.c as well, but move bufsz, list count, and xfer flags back out into struct usbnet, and have them set as part of the setup efore usbnet_attach() - split the intr pipe parts into their own structure - move all the main usbnet*lock* code into usbnet.c too usbnet_attach() goes down to 2 args, and the inputs needed are now the full contents of 'struct usbnet' besides the driver owned 'un_flags' and usbnet owned 'un_pri'. welcome netbsd 9.99.6.
2019-08-10 05:17:36 +03:00
USBD_EXCLUSIVE_USE | USBD_MPSAFE, &unp->unp_ep[i]);
- adjust usbnet interface to allow usbd_open_pipe_intr(), from the new comment: * if un_intr_buf is not NULL, use usbd_open_pipe_intr() not * usbd_open_pipe() for USBNET_ENDPT_INTR, with this buffer, * size, and interval. the standard handling is in usbnet.c, with a callback to deal with the interrupt it self. not fully tested, designed for if_aue.c and a few others not yet converted. - make usbhist for usbnet.c work, thanks paulg - usbnet_init_rx_tx() clears out all allocations upon failure now - add usbnet_ec() to get a pointer to the struct ethercom - add usbnet_{lock,unlock,owned}*() to lock/unlock the various locks and *owned*() for asserting welcome 9.99.3!
2019-08-04 11:59:13 +03:00
}
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
if (err) {
usbnet_ep_close_pipes(un);
return err;
}
}
return USBD_NORMAL_COMPLETION;
}
static usbd_status
- use usbnet_isowned*() more - more const
2019-08-11 04:31:19 +03:00
usbnet_ep_stop_pipes(struct usbnet * const un)
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
{
reduce the scope of struct usbnet: - move a large number of members internal to usbnet.c's new "struct usbnet_private". - provide accessors for a few of these - move struct usbnet_cdata into usbnet.c as well, but move bufsz, list count, and xfer flags back out into struct usbnet, and have them set as part of the setup efore usbnet_attach() - split the intr pipe parts into their own structure - move all the main usbnet*lock* code into usbnet.c too usbnet_attach() goes down to 2 args, and the inputs needed are now the full contents of 'struct usbnet' besides the driver owned 'un_flags' and usbnet owned 'un_pri'. welcome netbsd 9.99.6.
2019-08-10 05:17:36 +03:00
struct usbnet_private * const unp = un->un_pri;
update usbnet slight: - drivers that want to use if_input() will also set _if_input. for now, avoid attaching a per-cpu queue for them. use if_initialize() and if_register(). - when stopping pipes, don't give up after the first failure, but keep the first failure error for return and keep going - if 0 a KASSERT() in usbnet_init_rx_tx(). there's a path via if_mcast_op() that can have the ifnet unlocked today.. - in usbnet_watchdog(), abort the pipe instead of faking tx completion. avoids issues with devices with more than one tx descriptor, as well as avoiding triggering usb asserts. with these, upl(4) port to usbnet(9) now works. (would be a version bump, but upl(4) and the unported umb(4) are the only consumers that would care.)
2019-08-18 12:29:38 +03:00
usbd_status err = USBD_NORMAL_COMPLETION;
reduce the scope of struct usbnet: - move a large number of members internal to usbnet.c's new "struct usbnet_private". - provide accessors for a few of these - move struct usbnet_cdata into usbnet.c as well, but move bufsz, list count, and xfer flags back out into struct usbnet, and have them set as part of the setup efore usbnet_attach() - split the intr pipe parts into their own structure - move all the main usbnet*lock* code into usbnet.c too usbnet_attach() goes down to 2 args, and the inputs needed are now the full contents of 'struct usbnet' besides the driver owned 'un_flags' and usbnet owned 'un_pri'. welcome netbsd 9.99.6.
2019-08-10 05:17:36 +03:00
for (size_t i = 0; i < __arraycount(unp->unp_ep); i++) {
if (unp->unp_ep[i] == NULL)
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
continue;
update usbnet slight: - drivers that want to use if_input() will also set _if_input. for now, avoid attaching a per-cpu queue for them. use if_initialize() and if_register(). - when stopping pipes, don't give up after the first failure, but keep the first failure error for return and keep going - if 0 a KASSERT() in usbnet_init_rx_tx(). there's a path via if_mcast_op() that can have the ifnet unlocked today.. - in usbnet_watchdog(), abort the pipe instead of faking tx completion. avoids issues with devices with more than one tx descriptor, as well as avoiding triggering usb asserts. with these, upl(4) port to usbnet(9) now works. (would be a version bump, but upl(4) and the unported umb(4) are the only consumers that would care.)
2019-08-18 12:29:38 +03:00
usbd_status err2 = usbd_abort_pipe(unp->unp_ep[i]);
if (err == USBD_NORMAL_COMPLETION && err2)
fix a think-o in the previous: don't return when saving the first error. from skrll@.
2019-08-18 12:46:58 +03:00
err = err2;
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
}
update usbnet slight: - drivers that want to use if_input() will also set _if_input. for now, avoid attaching a per-cpu queue for them. use if_initialize() and if_register(). - when stopping pipes, don't give up after the first failure, but keep the first failure error for return and keep going - if 0 a KASSERT() in usbnet_init_rx_tx(). there's a path via if_mcast_op() that can have the ifnet unlocked today.. - in usbnet_watchdog(), abort the pipe instead of faking tx completion. avoids issues with devices with more than one tx descriptor, as well as avoiding triggering usb asserts. with these, upl(4) port to usbnet(9) now works. (would be a version bump, but upl(4) and the unported umb(4) are the only consumers that would care.)
2019-08-18 12:29:38 +03:00
return err;
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
}
int
update usbnet some: - move rx/tx xfer flags into usbnet_cdata - move the callbacks into usbnet_ops structure - move rx/tx xfer flags arguments from usbnet_init_rx_tx() and move them all into usbnet_attach() arguments - s/miibus/mii/ in some places for consistency other clean up: - create wrapper functions for callbacks, move knowledge about special handling (OK to be missing, error eating) there. - use cdata pointer if already available - provide some more macros (will be real functions later) for accessing usbnet members, use existing ones more bump kernel version.
2019-08-09 04:17:33 +03:00
usbnet_init_rx_tx(struct usbnet * const un)
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
{
- adjust usbnet interface to allow usbd_open_pipe_intr(), from the new comment: * if un_intr_buf is not NULL, use usbd_open_pipe_intr() not * usbd_open_pipe() for USBNET_ENDPT_INTR, with this buffer, * size, and interval. the standard handling is in usbnet.c, with a callback to deal with the interrupt it self. not fully tested, designed for if_aue.c and a few others not yet converted. - make usbhist for usbnet.c work, thanks paulg - usbnet_init_rx_tx() clears out all allocations upon failure now - add usbnet_ec() to get a pointer to the struct ethercom - add usbnet_{lock,unlock,owned}*() to lock/unlock the various locks and *owned*() for asserting welcome 9.99.3!
2019-08-04 11:59:13 +03:00
USBNETHIST_FUNC(); USBNETHIST_CALLED();
reduce the scope of struct usbnet: - move a large number of members internal to usbnet.c's new "struct usbnet_private". - provide accessors for a few of these - move struct usbnet_cdata into usbnet.c as well, but move bufsz, list count, and xfer flags back out into struct usbnet, and have them set as part of the setup efore usbnet_attach() - split the intr pipe parts into their own structure - move all the main usbnet*lock* code into usbnet.c too usbnet_attach() goes down to 2 args, and the inputs needed are now the full contents of 'struct usbnet' besides the driver owned 'un_flags' and usbnet owned 'un_pri'. welcome netbsd 9.99.6.
2019-08-10 05:17:36 +03:00
struct usbnet_private * const unp = un->un_pri;
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
struct ifnet * const ifp = usbnet_ifp(un);
usbd_status err;
- adjust usbnet interface to allow usbd_open_pipe_intr(), from the new comment: * if un_intr_buf is not NULL, use usbd_open_pipe_intr() not * usbd_open_pipe() for USBNET_ENDPT_INTR, with this buffer, * size, and interval. the standard handling is in usbnet.c, with a callback to deal with the interrupt it self. not fully tested, designed for if_aue.c and a few others not yet converted. - make usbhist for usbnet.c work, thanks paulg - usbnet_init_rx_tx() clears out all allocations upon failure now - add usbnet_ec() to get a pointer to the struct ethercom - add usbnet_{lock,unlock,owned}*() to lock/unlock the various locks and *owned*() for asserting welcome 9.99.3!
2019-08-04 11:59:13 +03:00
int error = 0;
usbnet: Assert IFNET_LOCKED in usbnet_init_rx_tx, usbnet_stop. Exception: urndis(4) abuses this API to start this logic before the ifp is actually initialized. So for the sake of urndis(4), until sense can be beaten into it, allow the !unp_ifp_attached case to run without IFNET_LOCK.
2022-03-03 08:47:36 +03:00
KASSERTMSG(!unp->unp_ifp_attached || IFNET_LOCKED(ifp),
"%s", ifp->if_xname);
Define and implement a locking protocol for the ifmedia / mii layers: - MP-safe drivers provide a mutex to ifmedia that is used to serialize access to media-related structures / hardware regsiters. Converted drivers use the new ifmedia_init_with_lock() function for this. The new name is provided to ease the transition. - Un-converted drivers continue to call ifmedia_init(), which will supply a compatibility lock to be used instead. Several media-related entry points must be aware of this compatibility lock, and are able to acquire it recursively a limited number of times, if needed. This is a SPIN mutex with priority IPL_NET. - This same lock is used to serialize access to PHY registers and other MII-related data structures. The PHY drivers are modified to acquire and release the lock, as needed, and assert the lock is held as a diagnostic aid. The "usbnet" framework has had an overhaul of its internal locking protocols to fit in with the media / mii changes, and the drivers adapted. USB wifi drivers have been changed to provide their own adaptive mutex to the ifmedia later via a new ieee80211_media_init_with_lock() function. This is required because the USB drivers need an adaptive mutex. Besised "usbnet", a few other drivers are converted: vmx, wm, ixgbe / ixv. mcx also now calls ifmedia_init_with_lock() because it needs to also use an adaptive mutex. The mcx driver still needs to be fully converted to NET_MPSAFE.
2020-03-16 02:04:50 +03:00
usbnet_isowned_core(un);
- adjust usbnet interface to allow usbd_open_pipe_intr(), from the new comment: * if un_intr_buf is not NULL, use usbd_open_pipe_intr() not * usbd_open_pipe() for USBNET_ENDPT_INTR, with this buffer, * size, and interval. the standard handling is in usbnet.c, with a callback to deal with the interrupt it self. not fully tested, designed for if_aue.c and a few others not yet converted. - make usbhist for usbnet.c work, thanks paulg - usbnet_init_rx_tx() clears out all allocations upon failure now - add usbnet_ec() to get a pointer to the struct ethercom - add usbnet_{lock,unlock,owned}*() to lock/unlock the various locks and *owned*() for asserting welcome 9.99.3!
2019-08-04 11:59:13 +03:00
usbnet: Use atomic_load/store_relaxed for unp_dying. This way we don't need to hold the core lock to avoid upsetting sanitizers (which probably find the current code upsetting), and we can use it to exit early from timeout loops that run under the core lock (which is probably not necessary for them to do anyway, but let's worry about that later).
2022-03-03 08:50:39 +03:00
if (usbnet_isdying(un)) {
- adjust usbnet interface to allow usbd_open_pipe_intr(), from the new comment: * if un_intr_buf is not NULL, use usbd_open_pipe_intr() not * usbd_open_pipe() for USBNET_ENDPT_INTR, with this buffer, * size, and interval. the standard handling is in usbnet.c, with a callback to deal with the interrupt it self. not fully tested, designed for if_aue.c and a few others not yet converted. - make usbhist for usbnet.c work, thanks paulg - usbnet_init_rx_tx() clears out all allocations upon failure now - add usbnet_ec() to get a pointer to the struct ethercom - add usbnet_{lock,unlock,owned}*() to lock/unlock the various locks and *owned*() for asserting welcome 9.99.3!
2019-08-04 11:59:13 +03:00
return EIO;
}
Define and implement a locking protocol for the ifmedia / mii layers: - MP-safe drivers provide a mutex to ifmedia that is used to serialize access to media-related structures / hardware regsiters. Converted drivers use the new ifmedia_init_with_lock() function for this. The new name is provided to ease the transition. - Un-converted drivers continue to call ifmedia_init(), which will supply a compatibility lock to be used instead. Several media-related entry points must be aware of this compatibility lock, and are able to acquire it recursively a limited number of times, if needed. This is a SPIN mutex with priority IPL_NET. - This same lock is used to serialize access to PHY registers and other MII-related data structures. The PHY drivers are modified to acquire and release the lock, as needed, and assert the lock is held as a diagnostic aid. The "usbnet" framework has had an overhaul of its internal locking protocols to fit in with the media / mii changes, and the drivers adapted. USB wifi drivers have been changed to provide their own adaptive mutex to the ifmedia later via a new ieee80211_media_init_with_lock() function. This is required because the USB drivers need an adaptive mutex. Besised "usbnet", a few other drivers are converted: vmx, wm, ixgbe / ixv. mcx also now calls ifmedia_init_with_lock() because it needs to also use an adaptive mutex. The mcx driver still needs to be fully converted to NET_MPSAFE.
2020-03-16 02:04:50 +03:00
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
/* Open RX and TX pipes. */
err = usbnet_ep_open_pipes(un);
if (err) {
aprint_error_dev(un->un_dev, "open rx/tx pipes failed: %s\n",
usbd_errstr(err));
- adjust usbnet interface to allow usbd_open_pipe_intr(), from the new comment: * if un_intr_buf is not NULL, use usbd_open_pipe_intr() not * usbd_open_pipe() for USBNET_ENDPT_INTR, with this buffer, * size, and interval. the standard handling is in usbnet.c, with a callback to deal with the interrupt it self. not fully tested, designed for if_aue.c and a few others not yet converted. - make usbhist for usbnet.c work, thanks paulg - usbnet_init_rx_tx() clears out all allocations upon failure now - add usbnet_ec() to get a pointer to the struct ethercom - add usbnet_{lock,unlock,owned}*() to lock/unlock the various locks and *owned*() for asserting welcome 9.99.3!
2019-08-04 11:59:13 +03:00
error = EIO;
goto out;
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
}
/* Init RX ring. */
update usbnet some: - move rx/tx xfer flags into usbnet_cdata - move the callbacks into usbnet_ops structure - move rx/tx xfer flags arguments from usbnet_init_rx_tx() and move them all into usbnet_attach() arguments - s/miibus/mii/ in some places for consistency other clean up: - create wrapper functions for callbacks, move knowledge about special handling (OK to be missing, error eating) there. - use cdata pointer if already available - provide some more macros (will be real functions later) for accessing usbnet members, use existing ones more bump kernel version.
2019-08-09 04:17:33 +03:00
if (usbnet_rx_list_init(un)) {
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
aprint_error_dev(un->un_dev, "rx list init failed\n");
- adjust usbnet interface to allow usbd_open_pipe_intr(), from the new comment: * if un_intr_buf is not NULL, use usbd_open_pipe_intr() not * usbd_open_pipe() for USBNET_ENDPT_INTR, with this buffer, * size, and interval. the standard handling is in usbnet.c, with a callback to deal with the interrupt it self. not fully tested, designed for if_aue.c and a few others not yet converted. - make usbhist for usbnet.c work, thanks paulg - usbnet_init_rx_tx() clears out all allocations upon failure now - add usbnet_ec() to get a pointer to the struct ethercom - add usbnet_{lock,unlock,owned}*() to lock/unlock the various locks and *owned*() for asserting welcome 9.99.3!
2019-08-04 11:59:13 +03:00
error = ENOBUFS;
goto out;
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
}
/* Init TX ring. */
update usbnet some: - move rx/tx xfer flags into usbnet_cdata - move the callbacks into usbnet_ops structure - move rx/tx xfer flags arguments from usbnet_init_rx_tx() and move them all into usbnet_attach() arguments - s/miibus/mii/ in some places for consistency other clean up: - create wrapper functions for callbacks, move knowledge about special handling (OK to be missing, error eating) there. - use cdata pointer if already available - provide some more macros (will be real functions later) for accessing usbnet members, use existing ones more bump kernel version.
2019-08-09 04:17:33 +03:00
if (usbnet_tx_list_init(un)) {
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
aprint_error_dev(un->un_dev, "tx list init failed\n");
- adjust usbnet interface to allow usbd_open_pipe_intr(), from the new comment: * if un_intr_buf is not NULL, use usbd_open_pipe_intr() not * usbd_open_pipe() for USBNET_ENDPT_INTR, with this buffer, * size, and interval. the standard handling is in usbnet.c, with a callback to deal with the interrupt it self. not fully tested, designed for if_aue.c and a few others not yet converted. - make usbhist for usbnet.c work, thanks paulg - usbnet_init_rx_tx() clears out all allocations upon failure now - add usbnet_ec() to get a pointer to the struct ethercom - add usbnet_{lock,unlock,owned}*() to lock/unlock the various locks and *owned*() for asserting welcome 9.99.3!
2019-08-04 11:59:13 +03:00
error = ENOBUFS;
goto out;
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
}
/* Indicate we are up and running. */
usbnet: Uncomment and fix assertion for ifp->if_flags |= IFF_RUNNING. We always hold IFNET_LOCK for ioctls that end up here -- the ones that don't hold it are only SIOCADDMULTI/SIOCDELMULTI, which don't end up here. However, urndis(4) throws a spanner in the works by doing weird device initialization.
2022-03-03 08:49:14 +03:00
/* XXX urndis calls usbnet_init_rx_tx before usbnet_attach_ifp. */
KASSERTMSG(!unp->unp_ifp_attached || IFNET_LOCKED(ifp),
"%s", ifp->if_xname);
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
ifp->if_flags |= IFF_RUNNING;
usbnet: Apply hardware multicast filter updates synchronously again. To make this work: 1. Do it only under a new lock, unp_mcastlock. This lock lives at IPL_SOFTCLOCK so it can be taken from network stack callouts. It is forbidden to acquire the usbnet core lock under unp_mcastlock. 2. Do it only after usbnet_init_rx_tx and before usbnet_stop; if issued at any other time, drop the update on the floor. 3. Make usbnet_init_rx_tx apply any pending multicast filter updates under the lock before setting the flag that allows SIOCADDMULTI or SIOCDELMULTI to apply the updates. 4. Remove core lock asserts from various drivers' register access routines. This is necessary because the multicast filter updates are done with register reads/writes, but _cannot_ take the core lock when the caller holds softnet_lock. This now programs the hardware multicast filter redundantly in many drivers which already explicitly call *_uno_mcast from the *_uno_init routines. This is probably harmless, but it will likely be better to remove the explicit calls.
2022-03-03 08:53:23 +03:00
/*
* If the hardware has a multicast filter, program it and then
* allow updates to it while we're running.
*/
if (un->un_ops->uno_mcast) {
mutex_enter(&unp->unp_mcastlock);
(*un->un_ops->uno_mcast)(ifp);
unp->unp_mcastactive = true;
mutex_exit(&unp->unp_mcastlock);
}
usbnet: Set and clear IFF_RUNNING slightly earlier and later. - Set IFF_RUNNING before any calls to usbnet_rxeof are possible. - Don't clear IFF_RUNNING until all transfers have been aborted.
2022-03-03 08:46:58 +03:00
/* Start up the receive pipe(s). */
usbnet_rx_start_pipes(un);
reduce the scope of struct usbnet: - move a large number of members internal to usbnet.c's new "struct usbnet_private". - provide accessors for a few of these - move struct usbnet_cdata into usbnet.c as well, but move bufsz, list count, and xfer flags back out into struct usbnet, and have them set as part of the setup efore usbnet_attach() - split the intr pipe parts into their own structure - move all the main usbnet*lock* code into usbnet.c too usbnet_attach() goes down to 2 args, and the inputs needed are now the full contents of 'struct usbnet' besides the driver owned 'un_flags' and usbnet owned 'un_pri'. welcome netbsd 9.99.6.
2019-08-10 05:17:36 +03:00
callout_schedule(&unp->unp_stat_ch, hz);
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
- adjust usbnet interface to allow usbd_open_pipe_intr(), from the new comment: * if un_intr_buf is not NULL, use usbd_open_pipe_intr() not * usbd_open_pipe() for USBNET_ENDPT_INTR, with this buffer, * size, and interval. the standard handling is in usbnet.c, with a callback to deal with the interrupt it self. not fully tested, designed for if_aue.c and a few others not yet converted. - make usbhist for usbnet.c work, thanks paulg - usbnet_init_rx_tx() clears out all allocations upon failure now - add usbnet_ec() to get a pointer to the struct ethercom - add usbnet_{lock,unlock,owned}*() to lock/unlock the various locks and *owned*() for asserting welcome 9.99.3!
2019-08-04 11:59:13 +03:00
out:
if (error) {
usbnet_rx_list_fini(un);
usbnet_tx_list_fini(un);
usbnet_ep_close_pipes(un);
}
Define and implement a locking protocol for the ifmedia / mii layers: - MP-safe drivers provide a mutex to ifmedia that is used to serialize access to media-related structures / hardware regsiters. Converted drivers use the new ifmedia_init_with_lock() function for this. The new name is provided to ease the transition. - Un-converted drivers continue to call ifmedia_init(), which will supply a compatibility lock to be used instead. Several media-related entry points must be aware of this compatibility lock, and are able to acquire it recursively a limited number of times, if needed. This is a SPIN mutex with priority IPL_NET. - This same lock is used to serialize access to PHY registers and other MII-related data structures. The PHY drivers are modified to acquire and release the lock, as needed, and assert the lock is held as a diagnostic aid. The "usbnet" framework has had an overhaul of its internal locking protocols to fit in with the media / mii changes, and the drivers adapted. USB wifi drivers have been changed to provide their own adaptive mutex to the ifmedia later via a new ieee80211_media_init_with_lock() function. This is required because the USB drivers need an adaptive mutex. Besised "usbnet", a few other drivers are converted: vmx, wm, ixgbe / ixv. mcx also now calls ifmedia_init_with_lock() because it needs to also use an adaptive mutex. The mcx driver still needs to be fully converted to NET_MPSAFE.
2020-03-16 02:04:50 +03:00
usbnet_isowned_core(un);
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
- adjust usbnet interface to allow usbd_open_pipe_intr(), from the new comment: * if un_intr_buf is not NULL, use usbd_open_pipe_intr() not * usbd_open_pipe() for USBNET_ENDPT_INTR, with this buffer, * size, and interval. the standard handling is in usbnet.c, with a callback to deal with the interrupt it self. not fully tested, designed for if_aue.c and a few others not yet converted. - make usbhist for usbnet.c work, thanks paulg - usbnet_init_rx_tx() clears out all allocations upon failure now - add usbnet_ec() to get a pointer to the struct ethercom - add usbnet_{lock,unlock,owned}*() to lock/unlock the various locks and *owned*() for asserting welcome 9.99.3!
2019-08-04 11:59:13 +03:00
return error;
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
}
Define and implement a locking protocol for the ifmedia / mii layers: - MP-safe drivers provide a mutex to ifmedia that is used to serialize access to media-related structures / hardware regsiters. Converted drivers use the new ifmedia_init_with_lock() function for this. The new name is provided to ease the transition. - Un-converted drivers continue to call ifmedia_init(), which will supply a compatibility lock to be used instead. Several media-related entry points must be aware of this compatibility lock, and are able to acquire it recursively a limited number of times, if needed. This is a SPIN mutex with priority IPL_NET. - This same lock is used to serialize access to PHY registers and other MII-related data structures. The PHY drivers are modified to acquire and release the lock, as needed, and assert the lock is held as a diagnostic aid. The "usbnet" framework has had an overhaul of its internal locking protocols to fit in with the media / mii changes, and the drivers adapted. USB wifi drivers have been changed to provide their own adaptive mutex to the ifmedia later via a new ieee80211_media_init_with_lock() function. This is required because the USB drivers need an adaptive mutex. Besised "usbnet", a few other drivers are converted: vmx, wm, ixgbe / ixv. mcx also now calls ifmedia_init_with_lock() because it needs to also use an adaptive mutex. The mcx driver still needs to be fully converted to NET_MPSAFE.
2020-03-16 02:04:50 +03:00
/* MII management. */
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
usbnet: Make usbnet_mii_readreg/writereg/statchg private to usbnet.c. No drivers need to use these.
2022-03-03 08:54:28 +03:00
static int
update usbnet some: - move rx/tx xfer flags into usbnet_cdata - move the callbacks into usbnet_ops structure - move rx/tx xfer flags arguments from usbnet_init_rx_tx() and move them all into usbnet_attach() arguments - s/miibus/mii/ in some places for consistency other clean up: - create wrapper functions for callbacks, move knowledge about special handling (OK to be missing, error eating) there. - use cdata pointer if already available - provide some more macros (will be real functions later) for accessing usbnet members, use existing ones more bump kernel version.
2019-08-09 04:17:33 +03:00
usbnet_mii_readreg(device_t dev, int phy, int reg, uint16_t *val)
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
{
add USBHIST_CALLARGS() frontend to KERNHIST_CALLARGS(). add USBNETHIST_CALLARGS() frontend to USBHIST_CALLARGS(). use both in read/write reg, instead of aprint. use %jx and (uintptr_t) and fix the 32 bit debug build.
2019-08-19 09:35:14 +03:00
USBNETHIST_FUNC();
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
struct usbnet * const un = device_private(dev);
couple more changes to usbnet(9): - MII read/write reg return int instead of usbd_status (requested by skrll) - usbnet_attach_ifp(9) changes arg, two mii-specific flags are placed by a pointer to new struct usbnet_mii. if not NULL, then attach an MII to this interface like previous have_mii parameter. use this to allow ure(4) to properly pass PHY location to mii_attach(). welcome netbsd 9.99.10.
2019-08-20 09:37:06 +03:00
int err;
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
Define and implement a locking protocol for the ifmedia / mii layers: - MP-safe drivers provide a mutex to ifmedia that is used to serialize access to media-related structures / hardware regsiters. Converted drivers use the new ifmedia_init_with_lock() function for this. The new name is provided to ease the transition. - Un-converted drivers continue to call ifmedia_init(), which will supply a compatibility lock to be used instead. Several media-related entry points must be aware of this compatibility lock, and are able to acquire it recursively a limited number of times, if needed. This is a SPIN mutex with priority IPL_NET. - This same lock is used to serialize access to PHY registers and other MII-related data structures. The PHY drivers are modified to acquire and release the lock, as needed, and assert the lock is held as a diagnostic aid. The "usbnet" framework has had an overhaul of its internal locking protocols to fit in with the media / mii changes, and the drivers adapted. USB wifi drivers have been changed to provide their own adaptive mutex to the ifmedia later via a new ieee80211_media_init_with_lock() function. This is required because the USB drivers need an adaptive mutex. Besised "usbnet", a few other drivers are converted: vmx, wm, ixgbe / ixv. mcx also now calls ifmedia_init_with_lock() because it needs to also use an adaptive mutex. The mcx driver still needs to be fully converted to NET_MPSAFE.
2020-03-16 02:04:50 +03:00
/* MII layer ensures core_lock is held. */
usbnet_isowned_core(un);
usbnet: Use atomic_load/store_relaxed for unp_dying. This way we don't need to hold the core lock to avoid upsetting sanitizers (which probably find the current code upsetting), and we can use it to exit early from timeout loops that run under the core lock (which is probably not necessary for them to do anyway, but let's worry about that later).
2022-03-03 08:50:39 +03:00
if (usbnet_isdying(un)) {
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
return EIO;
}
update usbnet some: - move rx/tx xfer flags into usbnet_cdata - move the callbacks into usbnet_ops structure - move rx/tx xfer flags arguments from usbnet_init_rx_tx() and move them all into usbnet_attach() arguments - s/miibus/mii/ in some places for consistency other clean up: - create wrapper functions for callbacks, move knowledge about special handling (OK to be missing, error eating) there. - use cdata pointer if already available - provide some more macros (will be real functions later) for accessing usbnet members, use existing ones more bump kernel version.
2019-08-09 04:17:33 +03:00
err = uno_read_reg(un, phy, reg, val);
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
if (err) {
PR/55068: sc.dying: Fix printf formats: - no %s/%p for kernel log - 0x% -> %# - always %j for kernel log
2020-03-13 21:17:40 +03:00
USBNETHIST_CALLARGS("%jd: read PHY failed: %jd",
usbnet: Use atomic_load/store_relaxed for unp_dying. This way we don't need to hold the core lock to avoid upsetting sanitizers (which probably find the current code upsetting), and we can use it to exit early from timeout loops that run under the core lock (which is probably not necessary for them to do anyway, but let's worry about that later).
2022-03-03 08:50:39 +03:00
un->un_pri->unp_number, err, 0, 0);
couple more changes to usbnet(9): - MII read/write reg return int instead of usbd_status (requested by skrll) - usbnet_attach_ifp(9) changes arg, two mii-specific flags are placed by a pointer to new struct usbnet_mii. if not NULL, then attach an MII to this interface like previous have_mii parameter. use this to allow ure(4) to properly pass PHY location to mii_attach(). welcome netbsd 9.99.10.
2019-08-20 09:37:06 +03:00
return err;
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
}
return 0;
}
usbnet: Make usbnet_mii_readreg/writereg/statchg private to usbnet.c. No drivers need to use these.
2022-03-03 08:54:28 +03:00
static int
update usbnet some: - move rx/tx xfer flags into usbnet_cdata - move the callbacks into usbnet_ops structure - move rx/tx xfer flags arguments from usbnet_init_rx_tx() and move them all into usbnet_attach() arguments - s/miibus/mii/ in some places for consistency other clean up: - create wrapper functions for callbacks, move knowledge about special handling (OK to be missing, error eating) there. - use cdata pointer if already available - provide some more macros (will be real functions later) for accessing usbnet members, use existing ones more bump kernel version.
2019-08-09 04:17:33 +03:00
usbnet_mii_writereg(device_t dev, int phy, int reg, uint16_t val)
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
{
add USBHIST_CALLARGS() frontend to KERNHIST_CALLARGS(). add USBNETHIST_CALLARGS() frontend to USBHIST_CALLARGS(). use both in read/write reg, instead of aprint. use %jx and (uintptr_t) and fix the 32 bit debug build.
2019-08-19 09:35:14 +03:00
USBNETHIST_FUNC();
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
struct usbnet * const un = device_private(dev);
couple more changes to usbnet(9): - MII read/write reg return int instead of usbd_status (requested by skrll) - usbnet_attach_ifp(9) changes arg, two mii-specific flags are placed by a pointer to new struct usbnet_mii. if not NULL, then attach an MII to this interface like previous have_mii parameter. use this to allow ure(4) to properly pass PHY location to mii_attach(). welcome netbsd 9.99.10.
2019-08-20 09:37:06 +03:00
int err;
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
Define and implement a locking protocol for the ifmedia / mii layers: - MP-safe drivers provide a mutex to ifmedia that is used to serialize access to media-related structures / hardware regsiters. Converted drivers use the new ifmedia_init_with_lock() function for this. The new name is provided to ease the transition. - Un-converted drivers continue to call ifmedia_init(), which will supply a compatibility lock to be used instead. Several media-related entry points must be aware of this compatibility lock, and are able to acquire it recursively a limited number of times, if needed. This is a SPIN mutex with priority IPL_NET. - This same lock is used to serialize access to PHY registers and other MII-related data structures. The PHY drivers are modified to acquire and release the lock, as needed, and assert the lock is held as a diagnostic aid. The "usbnet" framework has had an overhaul of its internal locking protocols to fit in with the media / mii changes, and the drivers adapted. USB wifi drivers have been changed to provide their own adaptive mutex to the ifmedia later via a new ieee80211_media_init_with_lock() function. This is required because the USB drivers need an adaptive mutex. Besised "usbnet", a few other drivers are converted: vmx, wm, ixgbe / ixv. mcx also now calls ifmedia_init_with_lock() because it needs to also use an adaptive mutex. The mcx driver still needs to be fully converted to NET_MPSAFE.
2020-03-16 02:04:50 +03:00
/* MII layer ensures core_lock is held. */
usbnet_isowned_core(un);
usbnet: Use atomic_load/store_relaxed for unp_dying. This way we don't need to hold the core lock to avoid upsetting sanitizers (which probably find the current code upsetting), and we can use it to exit early from timeout loops that run under the core lock (which is probably not necessary for them to do anyway, but let's worry about that later).
2022-03-03 08:50:39 +03:00
if (usbnet_isdying(un)) {
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
return EIO;
}
update usbnet some: - move rx/tx xfer flags into usbnet_cdata - move the callbacks into usbnet_ops structure - move rx/tx xfer flags arguments from usbnet_init_rx_tx() and move them all into usbnet_attach() arguments - s/miibus/mii/ in some places for consistency other clean up: - create wrapper functions for callbacks, move knowledge about special handling (OK to be missing, error eating) there. - use cdata pointer if already available - provide some more macros (will be real functions later) for accessing usbnet members, use existing ones more bump kernel version.
2019-08-09 04:17:33 +03:00
err = uno_write_reg(un, phy, reg, val);
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
if (err) {
PR/55068: sc.dying: Fix printf formats: - no %s/%p for kernel log - 0x% -> %# - always %j for kernel log
2020-03-13 21:17:40 +03:00
USBNETHIST_CALLARGS("%jd: write PHY failed: %jd",
usbnet: Use atomic_load/store_relaxed for unp_dying. This way we don't need to hold the core lock to avoid upsetting sanitizers (which probably find the current code upsetting), and we can use it to exit early from timeout loops that run under the core lock (which is probably not necessary for them to do anyway, but let's worry about that later).
2022-03-03 08:50:39 +03:00
un->un_pri->unp_number, err, 0, 0);
couple more changes to usbnet(9): - MII read/write reg return int instead of usbd_status (requested by skrll) - usbnet_attach_ifp(9) changes arg, two mii-specific flags are placed by a pointer to new struct usbnet_mii. if not NULL, then attach an MII to this interface like previous have_mii parameter. use this to allow ure(4) to properly pass PHY location to mii_attach(). welcome netbsd 9.99.10.
2019-08-20 09:37:06 +03:00
return err;
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
}
return 0;
}
usbnet: Make usbnet_mii_readreg/writereg/statchg private to usbnet.c. No drivers need to use these.
2022-03-03 08:54:28 +03:00
static void
update usbnet some: - move rx/tx xfer flags into usbnet_cdata - move the callbacks into usbnet_ops structure - move rx/tx xfer flags arguments from usbnet_init_rx_tx() and move them all into usbnet_attach() arguments - s/miibus/mii/ in some places for consistency other clean up: - create wrapper functions for callbacks, move knowledge about special handling (OK to be missing, error eating) there. - use cdata pointer if already available - provide some more macros (will be real functions later) for accessing usbnet members, use existing ones more bump kernel version.
2019-08-09 04:17:33 +03:00
usbnet_mii_statchg(struct ifnet *ifp)
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
{
- adjust usbnet interface to allow usbd_open_pipe_intr(), from the new comment: * if un_intr_buf is not NULL, use usbd_open_pipe_intr() not * usbd_open_pipe() for USBNET_ENDPT_INTR, with this buffer, * size, and interval. the standard handling is in usbnet.c, with a callback to deal with the interrupt it self. not fully tested, designed for if_aue.c and a few others not yet converted. - make usbhist for usbnet.c work, thanks paulg - usbnet_init_rx_tx() clears out all allocations upon failure now - add usbnet_ec() to get a pointer to the struct ethercom - add usbnet_{lock,unlock,owned}*() to lock/unlock the various locks and *owned*() for asserting welcome 9.99.3!
2019-08-04 11:59:13 +03:00
USBNETHIST_FUNC(); USBNETHIST_CALLED();
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
struct usbnet * const un = ifp->if_softc;
Define and implement a locking protocol for the ifmedia / mii layers: - MP-safe drivers provide a mutex to ifmedia that is used to serialize access to media-related structures / hardware regsiters. Converted drivers use the new ifmedia_init_with_lock() function for this. The new name is provided to ease the transition. - Un-converted drivers continue to call ifmedia_init(), which will supply a compatibility lock to be used instead. Several media-related entry points must be aware of this compatibility lock, and are able to acquire it recursively a limited number of times, if needed. This is a SPIN mutex with priority IPL_NET. - This same lock is used to serialize access to PHY registers and other MII-related data structures. The PHY drivers are modified to acquire and release the lock, as needed, and assert the lock is held as a diagnostic aid. The "usbnet" framework has had an overhaul of its internal locking protocols to fit in with the media / mii changes, and the drivers adapted. USB wifi drivers have been changed to provide their own adaptive mutex to the ifmedia later via a new ieee80211_media_init_with_lock() function. This is required because the USB drivers need an adaptive mutex. Besised "usbnet", a few other drivers are converted: vmx, wm, ixgbe / ixv. mcx also now calls ifmedia_init_with_lock() because it needs to also use an adaptive mutex. The mcx driver still needs to be fully converted to NET_MPSAFE.
2020-03-16 02:04:50 +03:00
/* MII layer ensures core_lock is held. */
usbnet_isowned_core(un);
update usbnet some: - move rx/tx xfer flags into usbnet_cdata - move the callbacks into usbnet_ops structure - move rx/tx xfer flags arguments from usbnet_init_rx_tx() and move them all into usbnet_attach() arguments - s/miibus/mii/ in some places for consistency other clean up: - create wrapper functions for callbacks, move knowledge about special handling (OK to be missing, error eating) there. - use cdata pointer if already available - provide some more macros (will be real functions later) for accessing usbnet members, use existing ones more bump kernel version.
2019-08-09 04:17:33 +03:00
uno_mii_statchg(un, ifp);
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
}
static int
usbnet_media_upd(struct ifnet *ifp)
{
- adjust usbnet interface to allow usbd_open_pipe_intr(), from the new comment: * if un_intr_buf is not NULL, use usbd_open_pipe_intr() not * usbd_open_pipe() for USBNET_ENDPT_INTR, with this buffer, * size, and interval. the standard handling is in usbnet.c, with a callback to deal with the interrupt it self. not fully tested, designed for if_aue.c and a few others not yet converted. - make usbhist for usbnet.c work, thanks paulg - usbnet_init_rx_tx() clears out all allocations upon failure now - add usbnet_ec() to get a pointer to the struct ethercom - add usbnet_{lock,unlock,owned}*() to lock/unlock the various locks and *owned*() for asserting welcome 9.99.3!
2019-08-04 11:59:13 +03:00
USBNETHIST_FUNC(); USBNETHIST_CALLED();
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
struct usbnet * const un = ifp->if_softc;
reduce the scope of struct usbnet: - move a large number of members internal to usbnet.c's new "struct usbnet_private". - provide accessors for a few of these - move struct usbnet_cdata into usbnet.c as well, but move bufsz, list count, and xfer flags back out into struct usbnet, and have them set as part of the setup efore usbnet_attach() - split the intr pipe parts into their own structure - move all the main usbnet*lock* code into usbnet.c too usbnet_attach() goes down to 2 args, and the inputs needed are now the full contents of 'struct usbnet' besides the driver owned 'un_flags' and usbnet owned 'un_pri'. welcome netbsd 9.99.6.
2019-08-10 05:17:36 +03:00
struct usbnet_private * const unp = un->un_pri;
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
struct mii_data * const mii = usbnet_mii(un);
Define and implement a locking protocol for the ifmedia / mii layers: - MP-safe drivers provide a mutex to ifmedia that is used to serialize access to media-related structures / hardware regsiters. Converted drivers use the new ifmedia_init_with_lock() function for this. The new name is provided to ease the transition. - Un-converted drivers continue to call ifmedia_init(), which will supply a compatibility lock to be used instead. Several media-related entry points must be aware of this compatibility lock, and are able to acquire it recursively a limited number of times, if needed. This is a SPIN mutex with priority IPL_NET. - This same lock is used to serialize access to PHY registers and other MII-related data structures. The PHY drivers are modified to acquire and release the lock, as needed, and assert the lock is held as a diagnostic aid. The "usbnet" framework has had an overhaul of its internal locking protocols to fit in with the media / mii changes, and the drivers adapted. USB wifi drivers have been changed to provide their own adaptive mutex to the ifmedia later via a new ieee80211_media_init_with_lock() function. This is required because the USB drivers need an adaptive mutex. Besised "usbnet", a few other drivers are converted: vmx, wm, ixgbe / ixv. mcx also now calls ifmedia_init_with_lock() because it needs to also use an adaptive mutex. The mcx driver still needs to be fully converted to NET_MPSAFE.
2020-03-16 02:04:50 +03:00
/* ifmedia layer ensures core_lock is held. */
usbnet_isowned_core(un);
usbnet: Assert IFNET_LOCKED in usbnet_media_upd. This ensures, if the device is being initialized or stopped, usbnet_media_upd will not run until it's done, so the reset sequence has exclusive access to the device registers used by mii.
2022-03-03 08:48:30 +03:00
/* ifmedia changes only with IFNET_LOCK held. */
KASSERTMSG(IFNET_LOCKED(ifp), "%s", ifp->if_xname);
usbnet: Use atomic_load/store_relaxed for unp_dying. This way we don't need to hold the core lock to avoid upsetting sanitizers (which probably find the current code upsetting), and we can use it to exit early from timeout loops that run under the core lock (which is probably not necessary for them to do anyway, but let's worry about that later).
2022-03-03 08:50:39 +03:00
if (usbnet_isdying(un))
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
return EIO;
reduce the scope of struct usbnet: - move a large number of members internal to usbnet.c's new "struct usbnet_private". - provide accessors for a few of these - move struct usbnet_cdata into usbnet.c as well, but move bufsz, list count, and xfer flags back out into struct usbnet, and have them set as part of the setup efore usbnet_attach() - split the intr pipe parts into their own structure - move all the main usbnet*lock* code into usbnet.c too usbnet_attach() goes down to 2 args, and the inputs needed are now the full contents of 'struct usbnet' besides the driver owned 'un_flags' and usbnet owned 'un_pri'. welcome netbsd 9.99.6.
2019-08-10 05:17:36 +03:00
unp->unp_link = false;
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
if (mii->mii_instance) {
struct mii_softc *miisc;
LIST_FOREACH(miisc, &mii->mii_phys, mii_list)
mii_phy_reset(miisc);
}
return ether_mediachange(ifp);
}
/* ioctl */
static int
usbnet_ifflags_cb(struct ethercom *ec)
{
- adjust usbnet interface to allow usbd_open_pipe_intr(), from the new comment: * if un_intr_buf is not NULL, use usbd_open_pipe_intr() not * usbd_open_pipe() for USBNET_ENDPT_INTR, with this buffer, * size, and interval. the standard handling is in usbnet.c, with a callback to deal with the interrupt it self. not fully tested, designed for if_aue.c and a few others not yet converted. - make usbhist for usbnet.c work, thanks paulg - usbnet_init_rx_tx() clears out all allocations upon failure now - add usbnet_ec() to get a pointer to the struct ethercom - add usbnet_{lock,unlock,owned}*() to lock/unlock the various locks and *owned*() for asserting welcome 9.99.3!
2019-08-04 11:59:13 +03:00
USBNETHIST_FUNC(); USBNETHIST_CALLED();
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
struct ifnet *ifp = &ec->ec_if;
struct usbnet *un = ifp->if_softc;
reduce the scope of struct usbnet: - move a large number of members internal to usbnet.c's new "struct usbnet_private". - provide accessors for a few of these - move struct usbnet_cdata into usbnet.c as well, but move bufsz, list count, and xfer flags back out into struct usbnet, and have them set as part of the setup efore usbnet_attach() - split the intr pipe parts into their own structure - move all the main usbnet*lock* code into usbnet.c too usbnet_attach() goes down to 2 args, and the inputs needed are now the full contents of 'struct usbnet' besides the driver owned 'un_flags' and usbnet owned 'un_pri'. welcome netbsd 9.99.6.
2019-08-10 05:17:36 +03:00
struct usbnet_private * const unp = un->un_pri;
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
int rv = 0;
usbnet: Assert IFNET_LOCKED on if_flags change callbacks. - if_init - if_stop - ethersubr(9) ifflags_cb
2022-03-03 08:47:28 +03:00
KASSERTMSG(IFNET_LOCKED(ifp), "%s", ifp->if_xname);
if_flags is neither int nor short. It's unsigned short.
2019-09-13 10:55:05 +03:00
const u_short changed = ifp->if_flags ^ unp->unp_if_flags;
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
if ((changed & ~(IFF_CANTCHANGE | IFF_DEBUG)) == 0) {
reduce the scope of struct usbnet: - move a large number of members internal to usbnet.c's new "struct usbnet_private". - provide accessors for a few of these - move struct usbnet_cdata into usbnet.c as well, but move bufsz, list count, and xfer flags back out into struct usbnet, and have them set as part of the setup efore usbnet_attach() - split the intr pipe parts into their own structure - move all the main usbnet*lock* code into usbnet.c too usbnet_attach() goes down to 2 args, and the inputs needed are now the full contents of 'struct usbnet' besides the driver owned 'un_flags' and usbnet owned 'un_pri'. welcome netbsd 9.99.6.
2019-08-10 05:17:36 +03:00
unp->unp_if_flags = ifp->if_flags;
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
if ((changed & IFF_PROMISC) != 0)
rv = ENETRESET;
} else {
rv = ENETRESET;
}
return rv;
}
static int
Define and implement a locking protocol for the ifmedia / mii layers: - MP-safe drivers provide a mutex to ifmedia that is used to serialize access to media-related structures / hardware regsiters. Converted drivers use the new ifmedia_init_with_lock() function for this. The new name is provided to ease the transition. - Un-converted drivers continue to call ifmedia_init(), which will supply a compatibility lock to be used instead. Several media-related entry points must be aware of this compatibility lock, and are able to acquire it recursively a limited number of times, if needed. This is a SPIN mutex with priority IPL_NET. - This same lock is used to serialize access to PHY registers and other MII-related data structures. The PHY drivers are modified to acquire and release the lock, as needed, and assert the lock is held as a diagnostic aid. The "usbnet" framework has had an overhaul of its internal locking protocols to fit in with the media / mii changes, and the drivers adapted. USB wifi drivers have been changed to provide their own adaptive mutex to the ifmedia later via a new ieee80211_media_init_with_lock() function. This is required because the USB drivers need an adaptive mutex. Besised "usbnet", a few other drivers are converted: vmx, wm, ixgbe / ixv. mcx also now calls ifmedia_init_with_lock() because it needs to also use an adaptive mutex. The mcx driver still needs to be fully converted to NET_MPSAFE.
2020-03-16 02:04:50 +03:00
usbnet_if_ioctl(struct ifnet *ifp, u_long cmd, void *data)
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
{
- remove unused uncd_rx_cnt - add USBNETHIST_CALLARGSN and use it in many places - add more history logs - add a log if watchdog abort leaves uncd_tx_cnt non zero - add a unique (32-bit) number for a each usbnet device and use it for logging - explicitly check for full tx ring in usbnet_start_locked() and return early, which avoids setting the 5 second timer and triggering a watchdog - reset uncd_tx_prod and uncd_tx_cnt to zero in stop
2019-08-23 07:29:28 +03:00
USBNETHIST_FUNC();
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
struct usbnet * const un = ifp->if_softc;
- remove unused uncd_rx_cnt - add USBNETHIST_CALLARGSN and use it in many places - add more history logs - add a log if watchdog abort leaves uncd_tx_cnt non zero - add a unique (32-bit) number for a each usbnet device and use it for logging - explicitly check for full tx ring in usbnet_start_locked() and return early, which avoids setting the 5 second timer and triggering a watchdog - reset uncd_tx_prod and uncd_tx_cnt to zero in stop
2019-08-23 07:29:28 +03:00
struct usbnet_private * const unp __unused = un->un_pri;
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
int error;
PR/55068: sc.dying: Fix printf formats: - no %s/%p for kernel log - 0x% -> %# - always %j for kernel log
2020-03-13 21:17:40 +03:00
USBNETHIST_CALLARGSN(11, "%jd: enter %#jx data %#jx",
- remove unused uncd_rx_cnt - add USBNETHIST_CALLARGSN and use it in many places - add more history logs - add a log if watchdog abort leaves uncd_tx_cnt non zero - add a unique (32-bit) number for a each usbnet device and use it for logging - explicitly check for full tx ring in usbnet_start_locked() and return early, which avoids setting the 5 second timer and triggering a watchdog - reset uncd_tx_prod and uncd_tx_cnt to zero in stop
2019-08-23 07:29:28 +03:00
unp->unp_number, cmd, (uintptr_t)data, 0);
update usbnet some: - move rx/tx xfer flags into usbnet_cdata - move the callbacks into usbnet_ops structure - move rx/tx xfer flags arguments from usbnet_init_rx_tx() and move them all into usbnet_attach() arguments - s/miibus/mii/ in some places for consistency other clean up: - create wrapper functions for callbacks, move knowledge about special handling (OK to be missing, error eating) there. - use cdata pointer if already available - provide some more macros (will be real functions later) for accessing usbnet members, use existing ones more bump kernel version.
2019-08-09 04:17:33 +03:00
if (un->un_ops->uno_override_ioctl)
return uno_override_ioctl(un, ifp, cmd, data);
extend usbnet to cope with if_upl, if_smsc, and if_umb needs: - usbnet_enqueue() can set mbuf flags and csum_data - usbnet_input() for non-ethernet based devices (upl, umb) - allow a complete override for ioctl() - remove converted list -- we have compiling and/or working patches for all the devices except for umb(4), will be merged as testing happens hopefully this is the last ABI change, though it may end up being extended for additional smsc(4) support. hello for real netbsd 9.99.3!
2019-08-06 03:19:57 +03:00
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
error = ether_ioctl(ifp, cmd, data);
usbnet: Defer hardware multicast filter updates to USB task. Breaks deadlock: - usbnet_detach holds usbnet lock, awaits kpause in ure_reset - callout holds softclock `lock' (sequential softints, blocks kpause wakeup), awaits softnet_lock in tcp_timer_keep, frag6_fasttimo, &c. - soclose holds softnet_lock, awaits usbnet lock in SIOCDELMULTI This change breaks the deadlock by not passing the SIOCADDMULTI or SIOCDELMULTI ioctl synchronously to the driver, which typically takes the usbnet lock. With this change, the ethernet layer still maintains the list of multicast addresses synchronously, but we defer the driver logic that updates the hardware multicast filter to an asynchronous USB task without softnet_lock held. This doesn't cause exactly the same ioctl to be sent to the driver -- usbnet just sends SIOCDELMULTI with an all-zero struct ifreq, and might drop some ioctls if issued in quick succession. This is OK because none of the drivers actually distinguish between SIOCADDMULTI and SIOCDELMULTI, or examine the argument; the drivers just commit whatever multicast addresses are listed in the ethercom. Other than the different ioctl submitted, there is no change to the ABI or locking scheme of usbnet, so this is safe to pull up to netbsd-9. This means we unfortunately can't guarantee that if a process issues SIOCADDMULTI and then sendto, the multicast filter update will be done by the time of the sendto -- and, more importantly, the packets received in reply to it. But failing to guarantee that is better than deadlocking! Later changes on HEAD will restore the synchronous multicast filter updates with much more extensive ABI changes and API simplifications in usbnet(9). Proposed on tech-net: https://mail-index.netbsd.org/tech-net/2021/12/30/msg008164.html XXX pullup-9
2022-01-30 00:37:07 +03:00
if (error == ENETRESET) {
switch (cmd) {
case SIOCADDMULTI:
case SIOCDELMULTI:
usbnet: Apply hardware multicast filter updates synchronously again. To make this work: 1. Do it only under a new lock, unp_mcastlock. This lock lives at IPL_SOFTCLOCK so it can be taken from network stack callouts. It is forbidden to acquire the usbnet core lock under unp_mcastlock. 2. Do it only after usbnet_init_rx_tx and before usbnet_stop; if issued at any other time, drop the update on the floor. 3. Make usbnet_init_rx_tx apply any pending multicast filter updates under the lock before setting the flag that allows SIOCADDMULTI or SIOCDELMULTI to apply the updates. 4. Remove core lock asserts from various drivers' register access routines. This is necessary because the multicast filter updates are done with register reads/writes, but _cannot_ take the core lock when the caller holds softnet_lock. This now programs the hardware multicast filter redundantly in many drivers which already explicitly call *_uno_mcast from the *_uno_init routines. This is probably harmless, but it will likely be better to remove the explicit calls.
2022-03-03 08:53:23 +03:00
/*
* If there's a hardware multicast filter, and
* it has been programmed by usbnet_init_rx_tx
* and is active, update it now. Otherwise,
* drop the update on the floor -- it will be
* observed by usbnet_init_rx_tx next time we
* bring the interface up.
*/
if (un->un_ops->uno_mcast) {
mutex_enter(&unp->unp_mcastlock);
if (unp->unp_mcastactive)
(*un->un_ops->uno_mcast)(ifp);
mutex_exit(&unp->unp_mcastlock);
}
usbnet: Defer hardware multicast filter updates to USB task. Breaks deadlock: - usbnet_detach holds usbnet lock, awaits kpause in ure_reset - callout holds softclock `lock' (sequential softints, blocks kpause wakeup), awaits softnet_lock in tcp_timer_keep, frag6_fasttimo, &c. - soclose holds softnet_lock, awaits usbnet lock in SIOCDELMULTI This change breaks the deadlock by not passing the SIOCADDMULTI or SIOCDELMULTI ioctl synchronously to the driver, which typically takes the usbnet lock. With this change, the ethernet layer still maintains the list of multicast addresses synchronously, but we defer the driver logic that updates the hardware multicast filter to an asynchronous USB task without softnet_lock held. This doesn't cause exactly the same ioctl to be sent to the driver -- usbnet just sends SIOCDELMULTI with an all-zero struct ifreq, and might drop some ioctls if issued in quick succession. This is OK because none of the drivers actually distinguish between SIOCADDMULTI and SIOCDELMULTI, or examine the argument; the drivers just commit whatever multicast addresses are listed in the ethercom. Other than the different ioctl submitted, there is no change to the ABI or locking scheme of usbnet, so this is safe to pull up to netbsd-9. This means we unfortunately can't guarantee that if a process issues SIOCADDMULTI and then sendto, the multicast filter update will be done by the time of the sendto -- and, more importantly, the packets received in reply to it. But failing to guarantee that is better than deadlocking! Later changes on HEAD will restore the synchronous multicast filter updates with much more extensive ABI changes and API simplifications in usbnet(9). Proposed on tech-net: https://mail-index.netbsd.org/tech-net/2021/12/30/msg008164.html XXX pullup-9
2022-01-30 00:37:07 +03:00
error = 0;
break;
default:
error = uno_ioctl(un, ifp, cmd, data);
}
}
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
return error;
}
/*
* Generic stop network function:
* - mark as stopping
* - call DD routine to stop the device
* - turn off running, timer, statchg callout, link
* - stop transfers
* - free RX and TX resources
* - close pipes
*
* usbnet_stop() is exported for drivers to use, expects lock held.
*
Define and implement a locking protocol for the ifmedia / mii layers: - MP-safe drivers provide a mutex to ifmedia that is used to serialize access to media-related structures / hardware regsiters. Converted drivers use the new ifmedia_init_with_lock() function for this. The new name is provided to ease the transition. - Un-converted drivers continue to call ifmedia_init(), which will supply a compatibility lock to be used instead. Several media-related entry points must be aware of this compatibility lock, and are able to acquire it recursively a limited number of times, if needed. This is a SPIN mutex with priority IPL_NET. - This same lock is used to serialize access to PHY registers and other MII-related data structures. The PHY drivers are modified to acquire and release the lock, as needed, and assert the lock is held as a diagnostic aid. The "usbnet" framework has had an overhaul of its internal locking protocols to fit in with the media / mii changes, and the drivers adapted. USB wifi drivers have been changed to provide their own adaptive mutex to the ifmedia later via a new ieee80211_media_init_with_lock() function. This is required because the USB drivers need an adaptive mutex. Besised "usbnet", a few other drivers are converted: vmx, wm, ixgbe / ixv. mcx also now calls ifmedia_init_with_lock() because it needs to also use an adaptive mutex. The mcx driver still needs to be fully converted to NET_MPSAFE.
2020-03-16 02:04:50 +03:00
* usbnet_if_stop() is for the if_stop handler.
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
*/
void
usbnet_stop(struct usbnet *un, struct ifnet *ifp, int disable)
{
reduce the scope of struct usbnet: - move a large number of members internal to usbnet.c's new "struct usbnet_private". - provide accessors for a few of these - move struct usbnet_cdata into usbnet.c as well, but move bufsz, list count, and xfer flags back out into struct usbnet, and have them set as part of the setup efore usbnet_attach() - split the intr pipe parts into their own structure - move all the main usbnet*lock* code into usbnet.c too usbnet_attach() goes down to 2 args, and the inputs needed are now the full contents of 'struct usbnet' besides the driver owned 'un_flags' and usbnet owned 'un_pri'. welcome netbsd 9.99.6.
2019-08-10 05:17:36 +03:00
struct usbnet_private * const unp = un->un_pri;
- adjust usbnet interface to allow usbd_open_pipe_intr(), from the new comment: * if un_intr_buf is not NULL, use usbd_open_pipe_intr() not * usbd_open_pipe() for USBNET_ENDPT_INTR, with this buffer, * size, and interval. the standard handling is in usbnet.c, with a callback to deal with the interrupt it self. not fully tested, designed for if_aue.c and a few others not yet converted. - make usbhist for usbnet.c work, thanks paulg - usbnet_init_rx_tx() clears out all allocations upon failure now - add usbnet_ec() to get a pointer to the struct ethercom - add usbnet_{lock,unlock,owned}*() to lock/unlock the various locks and *owned*() for asserting welcome 9.99.3!
2019-08-04 11:59:13 +03:00
USBNETHIST_FUNC(); USBNETHIST_CALLED();
usbnet: Assert IFNET_LOCKED in usbnet_init_rx_tx, usbnet_stop. Exception: urndis(4) abuses this API to start this logic before the ifp is actually initialized. So for the sake of urndis(4), until sense can be beaten into it, allow the !unp_ifp_attached case to run without IFNET_LOCK.
2022-03-03 08:47:36 +03:00
KASSERTMSG(!unp->unp_ifp_attached || IFNET_LOCKED(ifp),
"%s", ifp->if_xname);
Define and implement a locking protocol for the ifmedia / mii layers: - MP-safe drivers provide a mutex to ifmedia that is used to serialize access to media-related structures / hardware regsiters. Converted drivers use the new ifmedia_init_with_lock() function for this. The new name is provided to ease the transition. - Un-converted drivers continue to call ifmedia_init(), which will supply a compatibility lock to be used instead. Several media-related entry points must be aware of this compatibility lock, and are able to acquire it recursively a limited number of times, if needed. This is a SPIN mutex with priority IPL_NET. - This same lock is used to serialize access to PHY registers and other MII-related data structures. The PHY drivers are modified to acquire and release the lock, as needed, and assert the lock is held as a diagnostic aid. The "usbnet" framework has had an overhaul of its internal locking protocols to fit in with the media / mii changes, and the drivers adapted. USB wifi drivers have been changed to provide their own adaptive mutex to the ifmedia later via a new ieee80211_media_init_with_lock() function. This is required because the USB drivers need an adaptive mutex. Besised "usbnet", a few other drivers are converted: vmx, wm, ixgbe / ixv. mcx also now calls ifmedia_init_with_lock() because it needs to also use an adaptive mutex. The mcx driver still needs to be fully converted to NET_MPSAFE.
2020-03-16 02:04:50 +03:00
usbnet_isowned_core(un);
usbnet: Apply hardware multicast filter updates synchronously again. To make this work: 1. Do it only under a new lock, unp_mcastlock. This lock lives at IPL_SOFTCLOCK so it can be taken from network stack callouts. It is forbidden to acquire the usbnet core lock under unp_mcastlock. 2. Do it only after usbnet_init_rx_tx and before usbnet_stop; if issued at any other time, drop the update on the floor. 3. Make usbnet_init_rx_tx apply any pending multicast filter updates under the lock before setting the flag that allows SIOCADDMULTI or SIOCDELMULTI to apply the updates. 4. Remove core lock asserts from various drivers' register access routines. This is necessary because the multicast filter updates are done with register reads/writes, but _cannot_ take the core lock when the caller holds softnet_lock. This now programs the hardware multicast filter redundantly in many drivers which already explicitly call *_uno_mcast from the *_uno_init routines. This is probably harmless, but it will likely be better to remove the explicit calls.
2022-03-03 08:53:23 +03:00
/*
* For drivers with hardware multicast filter update callbacks:
* Prevent concurrent access to the hardware registers by
* multicast filter updates, which happens without IFNET_LOCK
* or the usbnet core lock.
*/
if (un->un_ops->uno_mcast) {
mutex_enter(&unp->unp_mcastlock);
unp->unp_mcastactive = false;
mutex_exit(&unp->unp_mcastlock);
}
usbnet: Clear watchdog timer before stopping hardware. No need to take the lock again -- which might not be necessary because the callout and task have completed, but let's obviate the need to think about that.
2022-03-03 08:49:37 +03:00
/*
* Prevent new activity (rescheduling ticks, xfers, &c.) and
* clear the watchdog timer.
*/
reduce the scope of struct usbnet: - move a large number of members internal to usbnet.c's new "struct usbnet_private". - provide accessors for a few of these - move struct usbnet_cdata into usbnet.c as well, but move bufsz, list count, and xfer flags back out into struct usbnet, and have them set as part of the setup efore usbnet_attach() - split the intr pipe parts into their own structure - move all the main usbnet*lock* code into usbnet.c too usbnet_attach() goes down to 2 args, and the inputs needed are now the full contents of 'struct usbnet' besides the driver owned 'un_flags' and usbnet owned 'un_pri'. welcome netbsd 9.99.6.
2019-08-10 05:17:36 +03:00
mutex_enter(&unp->unp_rxlock);
mutex_enter(&unp->unp_txlock);
unp->unp_stopping = true;
usbnet: Clear watchdog timer before stopping hardware. No need to take the lock again -- which might not be necessary because the callout and task have completed, but let's obviate the need to think about that.
2022-03-03 08:49:37 +03:00
unp->unp_timer = 0;
reduce the scope of struct usbnet: - move a large number of members internal to usbnet.c's new "struct usbnet_private". - provide accessors for a few of these - move struct usbnet_cdata into usbnet.c as well, but move bufsz, list count, and xfer flags back out into struct usbnet, and have them set as part of the setup efore usbnet_attach() - split the intr pipe parts into their own structure - move all the main usbnet*lock* code into usbnet.c too usbnet_attach() goes down to 2 args, and the inputs needed are now the full contents of 'struct usbnet' besides the driver owned 'un_flags' and usbnet owned 'un_pri'. welcome netbsd 9.99.6.
2019-08-10 05:17:36 +03:00
mutex_exit(&unp->unp_txlock);
mutex_exit(&unp->unp_rxlock);
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
usbnet: Fix ordering of actions in usbnet_stop. Make sure all software activity is quiescent (callouts and tasks, including ifmedia and mii callbacks -- anything that might trigger register access) before asking the driver to stop the hardware. This way, the driver uno_stop routine is guaranteed exclusive access to the registers. This will also enable us to simplify the callouts and tasks so they don't have to check the software state -- to be done in a separate commit.
2022-03-03 08:48:22 +03:00
/*
* Stop the timer first, then the task -- if the timer was
* already firing, we stop the task or wait for it complete
* only after if last fired. Setting unp_stopping prevents the
* timer task from being scheduled again.
*/
callout_halt(&unp->unp_stat_ch, &unp->unp_core_lock);
usb_rem_task_wait(un->un_udev, &unp->unp_ticktask, USB_TASKQ_DRIVER,
&unp->unp_core_lock);
/*
* Now that the software is quiescent, ask the driver to stop
* the hardware. The driver's uno_stop routine now has
* exclusive access to any registers that might previously have
* been used by to ifmedia, mii, or ioctl callbacks.
usbnet: Don't waste time calling uno_stop if device is detaching. The hardware is most likely gone, so trying to write to its registers (and, in some cases, wait until a timeout for a device to reset) is a waste of time. Even if it was detached only in software with drvctl, reattaching it will reset the device anyway.
2022-03-03 08:49:58 +03:00
*
* Don't bother if the device is being detached, though -- if
* it's been unplugged then there's no point in trying to touch
* the registers.
usbnet: Fix ordering of actions in usbnet_stop. Make sure all software activity is quiescent (callouts and tasks, including ifmedia and mii callbacks -- anything that might trigger register access) before asking the driver to stop the hardware. This way, the driver uno_stop routine is guaranteed exclusive access to the registers. This will also enable us to simplify the callouts and tasks so they don't have to check the software state -- to be done in a separate commit.
2022-03-03 08:48:22 +03:00
*/
usbnet: Use atomic_load/store_relaxed for unp_dying. This way we don't need to hold the core lock to avoid upsetting sanitizers (which probably find the current code upsetting), and we can use it to exit early from timeout loops that run under the core lock (which is probably not necessary for them to do anyway, but let's worry about that later).
2022-03-03 08:50:39 +03:00
if (!usbnet_isdying(un))
usbnet: Don't waste time calling uno_stop if device is detaching. The hardware is most likely gone, so trying to write to its registers (and, in some cases, wait until a timeout for a device to reset) is a waste of time. Even if it was detached only in software with drvctl, reattaching it will reset the device anyway.
2022-03-03 08:49:58 +03:00
uno_stop(un, ifp, disable);
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
/* Stop transfers. */
usbnet_ep_stop_pipes(un);
/* Free RX/TX resources. */
usbnet_rx_list_fini(un);
usbnet_tx_list_fini(un);
/* Close pipes. */
usbnet_ep_close_pipes(un);
Define and implement a locking protocol for the ifmedia / mii layers: - MP-safe drivers provide a mutex to ifmedia that is used to serialize access to media-related structures / hardware regsiters. Converted drivers use the new ifmedia_init_with_lock() function for this. The new name is provided to ease the transition. - Un-converted drivers continue to call ifmedia_init(), which will supply a compatibility lock to be used instead. Several media-related entry points must be aware of this compatibility lock, and are able to acquire it recursively a limited number of times, if needed. This is a SPIN mutex with priority IPL_NET. - This same lock is used to serialize access to PHY registers and other MII-related data structures. The PHY drivers are modified to acquire and release the lock, as needed, and assert the lock is held as a diagnostic aid. The "usbnet" framework has had an overhaul of its internal locking protocols to fit in with the media / mii changes, and the drivers adapted. USB wifi drivers have been changed to provide their own adaptive mutex to the ifmedia later via a new ieee80211_media_init_with_lock() function. This is required because the USB drivers need an adaptive mutex. Besised "usbnet", a few other drivers are converted: vmx, wm, ixgbe / ixv. mcx also now calls ifmedia_init_with_lock() because it needs to also use an adaptive mutex. The mcx driver still needs to be fully converted to NET_MPSAFE.
2020-03-16 02:04:50 +03:00
usbnet: Assert IFNET_LOCKED in usbnet_init_rx_tx, usbnet_stop. Exception: urndis(4) abuses this API to start this logic before the ifp is actually initialized. So for the sake of urndis(4), until sense can be beaten into it, allow the !unp_ifp_attached case to run without IFNET_LOCK.
2022-03-03 08:47:36 +03:00
/* Everything is quesced now. */
KASSERTMSG(!unp->unp_ifp_attached || IFNET_LOCKED(ifp),
"%s", ifp->if_xname);
usbnet: Set and clear IFF_RUNNING slightly earlier and later. - Set IFF_RUNNING before any calls to usbnet_rxeof are possible. - Don't clear IFF_RUNNING until all transfers have been aborted.
2022-03-03 08:46:58 +03:00
ifp->if_flags &= ~IFF_RUNNING;
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
}
static void
Define and implement a locking protocol for the ifmedia / mii layers: - MP-safe drivers provide a mutex to ifmedia that is used to serialize access to media-related structures / hardware regsiters. Converted drivers use the new ifmedia_init_with_lock() function for this. The new name is provided to ease the transition. - Un-converted drivers continue to call ifmedia_init(), which will supply a compatibility lock to be used instead. Several media-related entry points must be aware of this compatibility lock, and are able to acquire it recursively a limited number of times, if needed. This is a SPIN mutex with priority IPL_NET. - This same lock is used to serialize access to PHY registers and other MII-related data structures. The PHY drivers are modified to acquire and release the lock, as needed, and assert the lock is held as a diagnostic aid. The "usbnet" framework has had an overhaul of its internal locking protocols to fit in with the media / mii changes, and the drivers adapted. USB wifi drivers have been changed to provide their own adaptive mutex to the ifmedia later via a new ieee80211_media_init_with_lock() function. This is required because the USB drivers need an adaptive mutex. Besised "usbnet", a few other drivers are converted: vmx, wm, ixgbe / ixv. mcx also now calls ifmedia_init_with_lock() because it needs to also use an adaptive mutex. The mcx driver still needs to be fully converted to NET_MPSAFE.
2020-03-16 02:04:50 +03:00
usbnet_if_stop(struct ifnet *ifp, int disable)
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
{
struct usbnet * const un = ifp->if_softc;
reduce the scope of struct usbnet: - move a large number of members internal to usbnet.c's new "struct usbnet_private". - provide accessors for a few of these - move struct usbnet_cdata into usbnet.c as well, but move bufsz, list count, and xfer flags back out into struct usbnet, and have them set as part of the setup efore usbnet_attach() - split the intr pipe parts into their own structure - move all the main usbnet*lock* code into usbnet.c too usbnet_attach() goes down to 2 args, and the inputs needed are now the full contents of 'struct usbnet' besides the driver owned 'un_flags' and usbnet owned 'un_pri'. welcome netbsd 9.99.6.
2019-08-10 05:17:36 +03:00
struct usbnet_private * const unp = un->un_pri;
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
usbnet: Assert IFNET_LOCKED on if_flags change callbacks. - if_init - if_stop - ethersubr(9) ifflags_cb
2022-03-03 08:47:28 +03:00
KASSERTMSG(IFNET_LOCKED(ifp), "%s", ifp->if_xname);
Define and implement a locking protocol for the ifmedia / mii layers: - MP-safe drivers provide a mutex to ifmedia that is used to serialize access to media-related structures / hardware regsiters. Converted drivers use the new ifmedia_init_with_lock() function for this. The new name is provided to ease the transition. - Un-converted drivers continue to call ifmedia_init(), which will supply a compatibility lock to be used instead. Several media-related entry points must be aware of this compatibility lock, and are able to acquire it recursively a limited number of times, if needed. This is a SPIN mutex with priority IPL_NET. - This same lock is used to serialize access to PHY registers and other MII-related data structures. The PHY drivers are modified to acquire and release the lock, as needed, and assert the lock is held as a diagnostic aid. The "usbnet" framework has had an overhaul of its internal locking protocols to fit in with the media / mii changes, and the drivers adapted. USB wifi drivers have been changed to provide their own adaptive mutex to the ifmedia later via a new ieee80211_media_init_with_lock() function. This is required because the USB drivers need an adaptive mutex. Besised "usbnet", a few other drivers are converted: vmx, wm, ixgbe / ixv. mcx also now calls ifmedia_init_with_lock() because it needs to also use an adaptive mutex. The mcx driver still needs to be fully converted to NET_MPSAFE.
2020-03-16 02:04:50 +03:00
mutex_enter(&unp->unp_core_lock);
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
usbnet_stop(un, ifp, disable);
Define and implement a locking protocol for the ifmedia / mii layers: - MP-safe drivers provide a mutex to ifmedia that is used to serialize access to media-related structures / hardware regsiters. Converted drivers use the new ifmedia_init_with_lock() function for this. The new name is provided to ease the transition. - Un-converted drivers continue to call ifmedia_init(), which will supply a compatibility lock to be used instead. Several media-related entry points must be aware of this compatibility lock, and are able to acquire it recursively a limited number of times, if needed. This is a SPIN mutex with priority IPL_NET. - This same lock is used to serialize access to PHY registers and other MII-related data structures. The PHY drivers are modified to acquire and release the lock, as needed, and assert the lock is held as a diagnostic aid. The "usbnet" framework has had an overhaul of its internal locking protocols to fit in with the media / mii changes, and the drivers adapted. USB wifi drivers have been changed to provide their own adaptive mutex to the ifmedia later via a new ieee80211_media_init_with_lock() function. This is required because the USB drivers need an adaptive mutex. Besised "usbnet", a few other drivers are converted: vmx, wm, ixgbe / ixv. mcx also now calls ifmedia_init_with_lock() because it needs to also use an adaptive mutex. The mcx driver still needs to be fully converted to NET_MPSAFE.
2020-03-16 02:04:50 +03:00
mutex_exit(&unp->unp_core_lock);
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
}
/*
* Generic tick task function.
*
* usbnet_tick() is triggered from a callout, and triggers a call to
* usbnet_tick_task() from the usb_task subsystem.
*/
static void
usbnet_tick(void *arg)
{
- use CALLARGS vs CALLED for better usbhist - turn off usbnetdebug default - log for all entry/exit points of usbnet_pipe_intr() - in usbnet_start_locked() track whether any packet has been transmitted for setting the timer. avoids spurious "watchdog timeouts" - in usbnet_stop() use callout_halt() vs callout_halt, and also stop the usb task. fixes crash of usbtask after the phy has detached. - add a little more defensive checking in the tick task, and add some high-log-level logs. - in usbnet_detach() move the call to usbnet_stop_ifp() above the calls to callout/usbtask stopping. - set ec_mii and unp_pri to NULL when freeing their data
2019-09-08 21:59:32 +03:00
USBNETHIST_FUNC();
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
struct usbnet * const un = arg;
reduce the scope of struct usbnet: - move a large number of members internal to usbnet.c's new "struct usbnet_private". - provide accessors for a few of these - move struct usbnet_cdata into usbnet.c as well, but move bufsz, list count, and xfer flags back out into struct usbnet, and have them set as part of the setup efore usbnet_attach() - split the intr pipe parts into their own structure - move all the main usbnet*lock* code into usbnet.c too usbnet_attach() goes down to 2 args, and the inputs needed are now the full contents of 'struct usbnet' besides the driver owned 'un_flags' and usbnet owned 'un_pri'. welcome netbsd 9.99.6.
2019-08-10 05:17:36 +03:00
struct usbnet_private * const unp = un->un_pri;
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
PR/55068: sc.dying: Fix printf formats: - no %s/%p for kernel log - 0x% -> %# - always %j for kernel log
2020-03-13 21:17:40 +03:00
USBNETHIST_CALLARGSN(10, "%jd: enter", unp->unp_number, 0, 0, 0);
- use CALLARGS vs CALLED for better usbhist - turn off usbnetdebug default - log for all entry/exit points of usbnet_pipe_intr() - in usbnet_start_locked() track whether any packet has been transmitted for setting the timer. avoids spurious "watchdog timeouts" - in usbnet_stop() use callout_halt() vs callout_halt, and also stop the usb task. fixes crash of usbtask after the phy has detached. - add a little more defensive checking in the tick task, and add some high-log-level logs. - in usbnet_detach() move the call to usbnet_stop_ifp() above the calls to callout/usbtask stopping. - set ec_mii and unp_pri to NULL when freeing their data
2019-09-08 21:59:32 +03:00
usbnet: Omit needless tests in usbnet_tick. It's harmless for us to schedule the tick task even if unp_dying or unp_stopping is set by now, because usbnet_stop will just wait for it to finish anyway, and the callout can't be scheduled again until the interface is done stopping and is brought back up again. No need for unp == NULL test -- un->un_pri is initialized well before this callout can be scheduled, and is nulled out only at the end of usbnet_detach, at which point we have already halted this callout.
2022-03-03 08:49:22 +03:00
/* Perform periodic stuff in process context */
usb_add_task(un->un_udev, &unp->unp_ticktask, USB_TASKQ_DRIVER);
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
}
static void
usbnet_watchdog(struct ifnet *ifp)
{
- remove unused uncd_rx_cnt - add USBNETHIST_CALLARGSN and use it in many places - add more history logs - add a log if watchdog abort leaves uncd_tx_cnt non zero - add a unique (32-bit) number for a each usbnet device and use it for logging - explicitly check for full tx ring in usbnet_start_locked() and return early, which avoids setting the 5 second timer and triggering a watchdog - reset uncd_tx_prod and uncd_tx_cnt to zero in stop
2019-08-23 07:29:28 +03:00
USBNETHIST_FUNC(); USBNETHIST_CALLED();
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
struct usbnet * const un = ifp->if_softc;
update usbnet slight: - drivers that want to use if_input() will also set _if_input. for now, avoid attaching a per-cpu queue for them. use if_initialize() and if_register(). - when stopping pipes, don't give up after the first failure, but keep the first failure error for return and keep going - if 0 a KASSERT() in usbnet_init_rx_tx(). there's a path via if_mcast_op() that can have the ifnet unlocked today.. - in usbnet_watchdog(), abort the pipe instead of faking tx completion. avoids issues with devices with more than one tx descriptor, as well as avoiding triggering usb asserts. with these, upl(4) port to usbnet(9) now works. (would be a version bump, but upl(4) and the unported umb(4) are the only consumers that would care.)
2019-08-18 12:29:38 +03:00
struct usbnet_private * const unp = un->un_pri;
reduce the scope of struct usbnet: - move a large number of members internal to usbnet.c's new "struct usbnet_private". - provide accessors for a few of these - move struct usbnet_cdata into usbnet.c as well, but move bufsz, list count, and xfer flags back out into struct usbnet, and have them set as part of the setup efore usbnet_attach() - split the intr pipe parts into their own structure - move all the main usbnet*lock* code into usbnet.c too usbnet_attach() goes down to 2 args, and the inputs needed are now the full contents of 'struct usbnet' besides the driver owned 'un_flags' and usbnet owned 'un_pri'. welcome netbsd 9.99.6.
2019-08-10 05:17:36 +03:00
struct usbnet_cdata * const cd = un_cdata(un);
update usbnet slight: - drivers that want to use if_input() will also set _if_input. for now, avoid attaching a per-cpu queue for them. use if_initialize() and if_register(). - when stopping pipes, don't give up after the first failure, but keep the first failure error for return and keep going - if 0 a KASSERT() in usbnet_init_rx_tx(). there's a path via if_mcast_op() that can have the ifnet unlocked today.. - in usbnet_watchdog(), abort the pipe instead of faking tx completion. avoids issues with devices with more than one tx descriptor, as well as avoiding triggering usb asserts. with these, upl(4) port to usbnet(9) now works. (would be a version bump, but upl(4) and the unported umb(4) are the only consumers that would care.)
2019-08-18 12:29:38 +03:00
usbd_status err;
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
Adopt <net/if_stats.h>.
2020-01-29 09:35:28 +03:00
if_statinc(ifp, if_oerrors);
reduce aprint_error(9) abuse
2021-03-01 20:41:00 +03:00
device_printf(un->un_dev, "watchdog timeout\n");
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
if (cd->uncd_tx_cnt > 0) {
fix more broken kernhist formats (now I got them all).
2020-03-14 06:01:36 +03:00
DPRINTF("uncd_tx_cnt=%ju non zero, aborting pipe", 0, 0, 0, 0);
update usbnet slight: - drivers that want to use if_input() will also set _if_input. for now, avoid attaching a per-cpu queue for them. use if_initialize() and if_register(). - when stopping pipes, don't give up after the first failure, but keep the first failure error for return and keep going - if 0 a KASSERT() in usbnet_init_rx_tx(). there's a path via if_mcast_op() that can have the ifnet unlocked today.. - in usbnet_watchdog(), abort the pipe instead of faking tx completion. avoids issues with devices with more than one tx descriptor, as well as avoiding triggering usb asserts. with these, upl(4) port to usbnet(9) now works. (would be a version bump, but upl(4) and the unported umb(4) are the only consumers that would care.)
2019-08-18 12:29:38 +03:00
err = usbd_abort_pipe(unp->unp_ep[USBNET_ENDPT_TX]);
move the check against un_phyno from usbnet back into the drivers that do this (axe, axen, mue, smsc, ure.) it made mii scanning only work for phy 0, and aue needs it for at least one device. fix smsc to return usbd_status not -1 on failure. XXX smsc was writing to '*val' even in error cases, it does not now. remove a double call to IFQ_SET_READY() (noticed by chuq). avoid unlock+instant relock by using usbnet_lock_mii_un_locked().
2019-08-19 10:33:37 +03:00
if (err)
reduce aprint_error(9) abuse
2021-03-01 20:41:00 +03:00
device_printf(un->un_dev, "pipe abort failed: %s\n",
move the check against un_phyno from usbnet back into the drivers that do this (axe, axen, mue, smsc, ure.) it made mii scanning only work for phy 0, and aue needs it for at least one device. fix smsc to return usbd_status not -1 on failure. XXX smsc was writing to '*val' even in error cases, it does not now. remove a double call to IFQ_SET_READY() (noticed by chuq). avoid unlock+instant relock by using usbnet_lock_mii_un_locked().
2019-08-19 10:33:37 +03:00
usbd_errstr(err));
- remove unused uncd_rx_cnt - add USBNETHIST_CALLARGSN and use it in many places - add more history logs - add a log if watchdog abort leaves uncd_tx_cnt non zero - add a unique (32-bit) number for a each usbnet device and use it for logging - explicitly check for full tx ring in usbnet_start_locked() and return early, which avoids setting the 5 second timer and triggering a watchdog - reset uncd_tx_prod and uncd_tx_cnt to zero in stop
2019-08-23 07:29:28 +03:00
if (cd->uncd_tx_cnt != 0)
fix more broken kernhist formats (now I got them all).
2020-03-14 06:01:36 +03:00
DPRINTF("uncd_tx_cnt now %ju", cd->uncd_tx_cnt, 0, 0, 0);
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
}
if (!IFQ_IS_EMPTY(&ifp->if_snd))
(*ifp->if_start)(ifp);
}
static void
usbnet_tick_task(void *arg)
{
- use CALLARGS vs CALLED for better usbhist - turn off usbnetdebug default - log for all entry/exit points of usbnet_pipe_intr() - in usbnet_start_locked() track whether any packet has been transmitted for setting the timer. avoids spurious "watchdog timeouts" - in usbnet_stop() use callout_halt() vs callout_halt, and also stop the usb task. fixes crash of usbtask after the phy has detached. - add a little more defensive checking in the tick task, and add some high-log-level logs. - in usbnet_detach() move the call to usbnet_stop_ifp() above the calls to callout/usbtask stopping. - set ec_mii and unp_pri to NULL when freeing their data
2019-09-08 21:59:32 +03:00
USBNETHIST_FUNC();
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
struct usbnet * const un = arg;
reduce the scope of struct usbnet: - move a large number of members internal to usbnet.c's new "struct usbnet_private". - provide accessors for a few of these - move struct usbnet_cdata into usbnet.c as well, but move bufsz, list count, and xfer flags back out into struct usbnet, and have them set as part of the setup efore usbnet_attach() - split the intr pipe parts into their own structure - move all the main usbnet*lock* code into usbnet.c too usbnet_attach() goes down to 2 args, and the inputs needed are now the full contents of 'struct usbnet' besides the driver owned 'un_flags' and usbnet owned 'un_pri'. welcome netbsd 9.99.6.
2019-08-10 05:17:36 +03:00
struct usbnet_private * const unp = un->un_pri;
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
struct ifnet * const ifp = usbnet_ifp(un);
struct mii_data * const mii = usbnet_mii(un);
usbnet: Omit needless locking/busying/testing in usbnet_tick_task. usbnet_stop waits for the task to complete before resetting the hardware, and usbnet_detach waits for usbnet_stop to complete before destroying anything, so there's no need for any of this.
2022-03-03 08:49:29 +03:00
USBNETHIST_CALLARGSN(8, "%jd: enter", unp->unp_number, 0, 0, 0);
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
usbnet: Ensure access to unp_timer is protected by unp_txlock.
2022-03-03 08:47:21 +03:00
mutex_enter(&unp->unp_txlock);
const bool timeout = unp->unp_timer != 0 && --unp->unp_timer == 0;
mutex_exit(&unp->unp_txlock);
if (timeout)
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
usbnet_watchdog(ifp);
PR/55068: sc.dying: Fix printf formats: - no %s/%p for kernel log - 0x% -> %# - always %j for kernel log
2020-03-13 21:17:40 +03:00
DPRINTFN(8, "mii %#jx ifp %#jx", (uintptr_t)mii, (uintptr_t)ifp, 0, 0);
Don't assume mii is not null here. Some drivers like urndis don't use mii, so they always have mii == NULL. ok riastradh. fixes PR kern/54762
2019-12-14 18:40:43 +03:00
if (mii) {
Define and implement a locking protocol for the ifmedia / mii layers: - MP-safe drivers provide a mutex to ifmedia that is used to serialize access to media-related structures / hardware regsiters. Converted drivers use the new ifmedia_init_with_lock() function for this. The new name is provided to ease the transition. - Un-converted drivers continue to call ifmedia_init(), which will supply a compatibility lock to be used instead. Several media-related entry points must be aware of this compatibility lock, and are able to acquire it recursively a limited number of times, if needed. This is a SPIN mutex with priority IPL_NET. - This same lock is used to serialize access to PHY registers and other MII-related data structures. The PHY drivers are modified to acquire and release the lock, as needed, and assert the lock is held as a diagnostic aid. The "usbnet" framework has had an overhaul of its internal locking protocols to fit in with the media / mii changes, and the drivers adapted. USB wifi drivers have been changed to provide their own adaptive mutex to the ifmedia later via a new ieee80211_media_init_with_lock() function. This is required because the USB drivers need an adaptive mutex. Besised "usbnet", a few other drivers are converted: vmx, wm, ixgbe / ixv. mcx also now calls ifmedia_init_with_lock() because it needs to also use an adaptive mutex. The mcx driver still needs to be fully converted to NET_MPSAFE.
2020-03-16 02:04:50 +03:00
mutex_enter(&unp->unp_core_lock);
Don't assume mii is not null here. Some drivers like urndis don't use mii, so they always have mii == NULL. ok riastradh. fixes PR kern/54762
2019-12-14 18:40:43 +03:00
mii_tick(mii);
if (!unp->unp_link)
(*mii->mii_statchg)(ifp);
Define and implement a locking protocol for the ifmedia / mii layers: - MP-safe drivers provide a mutex to ifmedia that is used to serialize access to media-related structures / hardware regsiters. Converted drivers use the new ifmedia_init_with_lock() function for this. The new name is provided to ease the transition. - Un-converted drivers continue to call ifmedia_init(), which will supply a compatibility lock to be used instead. Several media-related entry points must be aware of this compatibility lock, and are able to acquire it recursively a limited number of times, if needed. This is a SPIN mutex with priority IPL_NET. - This same lock is used to serialize access to PHY registers and other MII-related data structures. The PHY drivers are modified to acquire and release the lock, as needed, and assert the lock is held as a diagnostic aid. The "usbnet" framework has had an overhaul of its internal locking protocols to fit in with the media / mii changes, and the drivers adapted. USB wifi drivers have been changed to provide their own adaptive mutex to the ifmedia later via a new ieee80211_media_init_with_lock() function. This is required because the USB drivers need an adaptive mutex. Besised "usbnet", a few other drivers are converted: vmx, wm, ixgbe / ixv. mcx also now calls ifmedia_init_with_lock() because it needs to also use an adaptive mutex. The mcx driver still needs to be fully converted to NET_MPSAFE.
2020-03-16 02:04:50 +03:00
mutex_exit(&unp->unp_core_lock);
Don't assume mii is not null here. Some drivers like urndis don't use mii, so they always have mii == NULL. ok riastradh. fixes PR kern/54762
2019-12-14 18:40:43 +03:00
}
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
- usbnet_rx_loop_cb's usbd_xfer parameter is never used and available in the usbnet_chain if needed. remove it - usbnet media status change already set link to false, don't repeat this in every driver - don't clear link in stop, nothing was re-enabling it for non-MII - add optional uno_tick_cb(struct usbnet *un) that is called from the usbnet tick timer - remove wrong debug sysctl prototype rx_loop and timer are kernel versions changes, but hopefully this is the last one for usbnet. working with 3 more drivers now (cue, mue and url), leaving only aue, kue, upl and umb undone (aue may work with previously supported devices, mine doesn't work with our driver, kue and upl have patches for testing and umb is undone.)
2019-08-15 08:52:23 +03:00
/* Call driver if requested. */
uno_tick(un);
Define and implement a locking protocol for the ifmedia / mii layers: - MP-safe drivers provide a mutex to ifmedia that is used to serialize access to media-related structures / hardware regsiters. Converted drivers use the new ifmedia_init_with_lock() function for this. The new name is provided to ease the transition. - Un-converted drivers continue to call ifmedia_init(), which will supply a compatibility lock to be used instead. Several media-related entry points must be aware of this compatibility lock, and are able to acquire it recursively a limited number of times, if needed. This is a SPIN mutex with priority IPL_NET. - This same lock is used to serialize access to PHY registers and other MII-related data structures. The PHY drivers are modified to acquire and release the lock, as needed, and assert the lock is held as a diagnostic aid. The "usbnet" framework has had an overhaul of its internal locking protocols to fit in with the media / mii changes, and the drivers adapted. USB wifi drivers have been changed to provide their own adaptive mutex to the ifmedia later via a new ieee80211_media_init_with_lock() function. This is required because the USB drivers need an adaptive mutex. Besised "usbnet", a few other drivers are converted: vmx, wm, ixgbe / ixv. mcx also now calls ifmedia_init_with_lock() because it needs to also use an adaptive mutex. The mcx driver still needs to be fully converted to NET_MPSAFE.
2020-03-16 02:04:50 +03:00
mutex_enter(&unp->unp_core_lock);
usbnet: Use atomic_load/store_relaxed for unp_dying. This way we don't need to hold the core lock to avoid upsetting sanitizers (which probably find the current code upsetting), and we can use it to exit early from timeout loops that run under the core lock (which is probably not necessary for them to do anyway, but let's worry about that later).
2022-03-03 08:50:39 +03:00
if (!unp->unp_stopping && !usbnet_isdying(un))
reduce the scope of struct usbnet: - move a large number of members internal to usbnet.c's new "struct usbnet_private". - provide accessors for a few of these - move struct usbnet_cdata into usbnet.c as well, but move bufsz, list count, and xfer flags back out into struct usbnet, and have them set as part of the setup efore usbnet_attach() - split the intr pipe parts into their own structure - move all the main usbnet*lock* code into usbnet.c too usbnet_attach() goes down to 2 args, and the inputs needed are now the full contents of 'struct usbnet' besides the driver owned 'un_flags' and usbnet owned 'un_pri'. welcome netbsd 9.99.6.
2019-08-10 05:17:36 +03:00
callout_schedule(&unp->unp_stat_ch, hz);
Define and implement a locking protocol for the ifmedia / mii layers: - MP-safe drivers provide a mutex to ifmedia that is used to serialize access to media-related structures / hardware regsiters. Converted drivers use the new ifmedia_init_with_lock() function for this. The new name is provided to ease the transition. - Un-converted drivers continue to call ifmedia_init(), which will supply a compatibility lock to be used instead. Several media-related entry points must be aware of this compatibility lock, and are able to acquire it recursively a limited number of times, if needed. This is a SPIN mutex with priority IPL_NET. - This same lock is used to serialize access to PHY registers and other MII-related data structures. The PHY drivers are modified to acquire and release the lock, as needed, and assert the lock is held as a diagnostic aid. The "usbnet" framework has had an overhaul of its internal locking protocols to fit in with the media / mii changes, and the drivers adapted. USB wifi drivers have been changed to provide their own adaptive mutex to the ifmedia later via a new ieee80211_media_init_with_lock() function. This is required because the USB drivers need an adaptive mutex. Besised "usbnet", a few other drivers are converted: vmx, wm, ixgbe / ixv. mcx also now calls ifmedia_init_with_lock() because it needs to also use an adaptive mutex. The mcx driver still needs to be fully converted to NET_MPSAFE.
2020-03-16 02:04:50 +03:00
mutex_exit(&unp->unp_core_lock);
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
}
static int
Define and implement a locking protocol for the ifmedia / mii layers: - MP-safe drivers provide a mutex to ifmedia that is used to serialize access to media-related structures / hardware regsiters. Converted drivers use the new ifmedia_init_with_lock() function for this. The new name is provided to ease the transition. - Un-converted drivers continue to call ifmedia_init(), which will supply a compatibility lock to be used instead. Several media-related entry points must be aware of this compatibility lock, and are able to acquire it recursively a limited number of times, if needed. This is a SPIN mutex with priority IPL_NET. - This same lock is used to serialize access to PHY registers and other MII-related data structures. The PHY drivers are modified to acquire and release the lock, as needed, and assert the lock is held as a diagnostic aid. The "usbnet" framework has had an overhaul of its internal locking protocols to fit in with the media / mii changes, and the drivers adapted. USB wifi drivers have been changed to provide their own adaptive mutex to the ifmedia later via a new ieee80211_media_init_with_lock() function. This is required because the USB drivers need an adaptive mutex. Besised "usbnet", a few other drivers are converted: vmx, wm, ixgbe / ixv. mcx also now calls ifmedia_init_with_lock() because it needs to also use an adaptive mutex. The mcx driver still needs to be fully converted to NET_MPSAFE.
2020-03-16 02:04:50 +03:00
usbnet_if_init(struct ifnet *ifp)
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
{
- adjust usbnet interface to allow usbd_open_pipe_intr(), from the new comment: * if un_intr_buf is not NULL, use usbd_open_pipe_intr() not * usbd_open_pipe() for USBNET_ENDPT_INTR, with this buffer, * size, and interval. the standard handling is in usbnet.c, with a callback to deal with the interrupt it self. not fully tested, designed for if_aue.c and a few others not yet converted. - make usbhist for usbnet.c work, thanks paulg - usbnet_init_rx_tx() clears out all allocations upon failure now - add usbnet_ec() to get a pointer to the struct ethercom - add usbnet_{lock,unlock,owned}*() to lock/unlock the various locks and *owned*() for asserting welcome 9.99.3!
2019-08-04 11:59:13 +03:00
USBNETHIST_FUNC(); USBNETHIST_CALLED();
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
struct usbnet * const un = ifp->if_softc;
usbnet: Enter uno_init with the core lock held. This reduces code in all drivers except urndis(4) and aue(4). However, it's still safe for urndis to drop the core lock because the ifnet is locked, and the ifnet lock covers the DOWN->UP (uno_init) and UP->DOWN (uno_stop) transitions.
2022-03-03 08:50:22 +03:00
int error;
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
usbnet: Assert IFNET_LOCKED on if_flags change callbacks. - if_init - if_stop - ethersubr(9) ifflags_cb
2022-03-03 08:47:28 +03:00
KASSERTMSG(IFNET_LOCKED(ifp), "%s", ifp->if_xname);
usbnet: Refuse to bring interfaces back up once dying. Make this happen uniformly across all usbnet drivers, not on a per-driver basis. This ensures new activity on the interface can't happen by the time we have stopped existing activity and waited for it to complete.
2022-03-03 08:48:37 +03:00
/*
* Prevent anyone from bringing the interface back up once
* we're detaching.
*/
usbnet: Omit needless locking around usbnet_isdying. Now that is tested and set with atomic_load/store, there is no need to hold the lock -- which means we can set it while the core lock is held during, e.g., a reset sequence, and use that to interrupt the sequence so it doesn't get stuck waiting to time out when the device is physically removed.
2022-03-03 08:50:47 +03:00
if (usbnet_isdying(un))
usbnet: Refuse to bring interfaces back up once dying. Make this happen uniformly across all usbnet drivers, not on a per-driver basis. This ensures new activity on the interface can't happen by the time we have stopped existing activity and waited for it to complete.
2022-03-03 08:48:37 +03:00
return EIO;
usbnet: Enter uno_init with the core lock held. This reduces code in all drivers except urndis(4) and aue(4). However, it's still safe for urndis to drop the core lock because the ifnet is locked, and the ifnet lock covers the DOWN->UP (uno_init) and UP->DOWN (uno_stop) transitions.
2022-03-03 08:50:22 +03:00
mutex_enter(&un->un_pri->unp_core_lock);
error = uno_init(un, ifp);
mutex_exit(&un->un_pri->unp_core_lock);
return error;
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
}
- use usbnet_isowned*() more - more const
2019-08-11 04:31:19 +03:00
reduce the scope of struct usbnet: - move a large number of members internal to usbnet.c's new "struct usbnet_private". - provide accessors for a few of these - move struct usbnet_cdata into usbnet.c as well, but move bufsz, list count, and xfer flags back out into struct usbnet, and have them set as part of the setup efore usbnet_attach() - split the intr pipe parts into their own structure - move all the main usbnet*lock* code into usbnet.c too usbnet_attach() goes down to 2 args, and the inputs needed are now the full contents of 'struct usbnet' besides the driver owned 'un_flags' and usbnet owned 'un_pri'. welcome netbsd 9.99.6.
2019-08-10 05:17:36 +03:00
/* Various accessors. */
void
usbnet_set_link(struct usbnet *un, bool link)
{
un->un_pri->unp_link = link;
}
struct ifnet *
usbnet_ifp(struct usbnet *un)
{
return &un->un_pri->unp_ec.ec_if;
}
struct ethercom *
usbnet_ec(struct usbnet *un)
{
return &un->un_pri->unp_ec;
}
struct mii_data *
usbnet_mii(struct usbnet *un)
{
return un->un_pri->unp_ec.ec_mii;
}
krndsource_t *
usbnet_rndsrc(struct usbnet *un)
{
return &un->un_pri->unp_rndsrc;
}
void *
usbnet_softc(struct usbnet *un)
{
return un->un_sc;
}
bool
usbnet_havelink(struct usbnet *un)
{
return un->un_pri->unp_link;
}
bool
usbnet_isdying(struct usbnet *un)
{
usbnet: Use atomic_load/store_relaxed for unp_dying. This way we don't need to hold the core lock to avoid upsetting sanitizers (which probably find the current code upsetting), and we can use it to exit early from timeout loops that run under the core lock (which is probably not necessary for them to do anyway, but let's worry about that later).
2022-03-03 08:50:39 +03:00
return atomic_load_relaxed(&un->un_pri->unp_dying);
reduce the scope of struct usbnet: - move a large number of members internal to usbnet.c's new "struct usbnet_private". - provide accessors for a few of these - move struct usbnet_cdata into usbnet.c as well, but move bufsz, list count, and xfer flags back out into struct usbnet, and have them set as part of the setup efore usbnet_attach() - split the intr pipe parts into their own structure - move all the main usbnet*lock* code into usbnet.c too usbnet_attach() goes down to 2 args, and the inputs needed are now the full contents of 'struct usbnet' besides the driver owned 'un_flags' and usbnet owned 'un_pri'. welcome netbsd 9.99.6.
2019-08-10 05:17:36 +03:00
}
/* Locking. */
usbnet: Make the tx/rx locks private to usbnet.c. Suffice it for the drivers to know that uno_tx_prepare and uno_rx_loop have exclusive access to the chain, and, for tx, exclusive access to the mbuf.
2022-03-03 08:52:20 +03:00
static void
usbnet_isowned_rx(struct usbnet *un)
reduce the scope of struct usbnet: - move a large number of members internal to usbnet.c's new "struct usbnet_private". - provide accessors for a few of these - move struct usbnet_cdata into usbnet.c as well, but move bufsz, list count, and xfer flags back out into struct usbnet, and have them set as part of the setup efore usbnet_attach() - split the intr pipe parts into their own structure - move all the main usbnet*lock* code into usbnet.c too usbnet_attach() goes down to 2 args, and the inputs needed are now the full contents of 'struct usbnet' besides the driver owned 'un_flags' and usbnet owned 'un_pri'. welcome netbsd 9.99.6.
2019-08-10 05:17:36 +03:00
{
usbnet: Make the tx/rx locks private to usbnet.c. Suffice it for the drivers to know that uno_tx_prepare and uno_rx_loop have exclusive access to the chain, and, for tx, exclusive access to the mbuf.
2022-03-03 08:52:20 +03:00
KASSERT(mutex_owned(&un->un_pri->unp_rxlock));
reduce the scope of struct usbnet: - move a large number of members internal to usbnet.c's new "struct usbnet_private". - provide accessors for a few of these - move struct usbnet_cdata into usbnet.c as well, but move bufsz, list count, and xfer flags back out into struct usbnet, and have them set as part of the setup efore usbnet_attach() - split the intr pipe parts into their own structure - move all the main usbnet*lock* code into usbnet.c too usbnet_attach() goes down to 2 args, and the inputs needed are now the full contents of 'struct usbnet' besides the driver owned 'un_flags' and usbnet owned 'un_pri'. welcome netbsd 9.99.6.
2019-08-10 05:17:36 +03:00
}
usbnet: Make the tx/rx locks private to usbnet.c. Suffice it for the drivers to know that uno_tx_prepare and uno_rx_loop have exclusive access to the chain, and, for tx, exclusive access to the mbuf.
2022-03-03 08:52:20 +03:00
static void
usbnet_isowned_tx(struct usbnet *un)
reduce the scope of struct usbnet: - move a large number of members internal to usbnet.c's new "struct usbnet_private". - provide accessors for a few of these - move struct usbnet_cdata into usbnet.c as well, but move bufsz, list count, and xfer flags back out into struct usbnet, and have them set as part of the setup efore usbnet_attach() - split the intr pipe parts into their own structure - move all the main usbnet*lock* code into usbnet.c too usbnet_attach() goes down to 2 args, and the inputs needed are now the full contents of 'struct usbnet' besides the driver owned 'un_flags' and usbnet owned 'un_pri'. welcome netbsd 9.99.6.
2019-08-10 05:17:36 +03:00
{
usbnet: Make the tx/rx locks private to usbnet.c. Suffice it for the drivers to know that uno_tx_prepare and uno_rx_loop have exclusive access to the chain, and, for tx, exclusive access to the mbuf.
2022-03-03 08:52:20 +03:00
KASSERT(mutex_owned(&un->un_pri->unp_txlock));
reduce the scope of struct usbnet: - move a large number of members internal to usbnet.c's new "struct usbnet_private". - provide accessors for a few of these - move struct usbnet_cdata into usbnet.c as well, but move bufsz, list count, and xfer flags back out into struct usbnet, and have them set as part of the setup efore usbnet_attach() - split the intr pipe parts into their own structure - move all the main usbnet*lock* code into usbnet.c too usbnet_attach() goes down to 2 args, and the inputs needed are now the full contents of 'struct usbnet' besides the driver owned 'un_flags' and usbnet owned 'un_pri'. welcome netbsd 9.99.6.
2019-08-10 05:17:36 +03:00
}
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
/* Autoconf management. */
extend usbnet to cope with if_upl, if_smsc, and if_umb needs: - usbnet_enqueue() can set mbuf flags and csum_data - usbnet_input() for non-ethernet based devices (upl, umb) - allow a complete override for ioctl() - remove converted list -- we have compiling and/or working patches for all the devices except for umb(4), will be merged as testing happens hopefully this is the last ABI change, though it may end up being extended for additional smsc(4) support. hello for real netbsd 9.99.3!
2019-08-06 03:19:57 +03:00
static bool
- use usbnet_isowned*() more - more const
2019-08-11 04:31:19 +03:00
usbnet_empty_eaddr(struct usbnet * const un)
extend usbnet to cope with if_upl, if_smsc, and if_umb needs: - usbnet_enqueue() can set mbuf flags and csum_data - usbnet_input() for non-ethernet based devices (upl, umb) - allow a complete override for ioctl() - remove converted list -- we have compiling and/or working patches for all the devices except for umb(4), will be merged as testing happens hopefully this is the last ABI change, though it may end up being extended for additional smsc(4) support. hello for real netbsd 9.99.3!
2019-08-06 03:19:57 +03:00
{
return (un->un_eaddr[0] == 0 && un->un_eaddr[1] == 0 &&
un->un_eaddr[2] == 0 && un->un_eaddr[3] == 0 &&
un->un_eaddr[4] == 0 && un->un_eaddr[5] == 0);
}
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
/*
* usbnet_attach() and usbnet_attach_ifp() perform setup of the relevant
* 'usbnet'. The first is enough to enable device access (eg, endpoints
* are connected and commands can be sent), and the second connects the
* device to the system networking.
*
* Always call usbnet_detach(), even if usbnet_attach_ifp() is skippped.
* Also usable as driver detach directly.
extend usbnet to cope with if_upl, if_smsc, and if_umb needs: - usbnet_enqueue() can set mbuf flags and csum_data - usbnet_input() for non-ethernet based devices (upl, umb) - allow a complete override for ioctl() - remove converted list -- we have compiling and/or working patches for all the devices except for umb(4), will be merged as testing happens hopefully this is the last ABI change, though it may end up being extended for additional smsc(4) support. hello for real netbsd 9.99.3!
2019-08-06 03:19:57 +03:00
*
* To skip ethernet configuration (eg, point-to-point), make sure that
* the un_eaddr[] is fully zero.
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
*/
reduce the scope of struct usbnet: - move a large number of members internal to usbnet.c's new "struct usbnet_private". - provide accessors for a few of these - move struct usbnet_cdata into usbnet.c as well, but move bufsz, list count, and xfer flags back out into struct usbnet, and have them set as part of the setup efore usbnet_attach() - split the intr pipe parts into their own structure - move all the main usbnet*lock* code into usbnet.c too usbnet_attach() goes down to 2 args, and the inputs needed are now the full contents of 'struct usbnet' besides the driver owned 'un_flags' and usbnet owned 'un_pri'. welcome netbsd 9.99.6.
2019-08-10 05:17:36 +03:00
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
void
usbnet_attach(struct usbnet *un,
reduce the scope of struct usbnet: - move a large number of members internal to usbnet.c's new "struct usbnet_private". - provide accessors for a few of these - move struct usbnet_cdata into usbnet.c as well, but move bufsz, list count, and xfer flags back out into struct usbnet, and have them set as part of the setup efore usbnet_attach() - split the intr pipe parts into their own structure - move all the main usbnet*lock* code into usbnet.c too usbnet_attach() goes down to 2 args, and the inputs needed are now the full contents of 'struct usbnet' besides the driver owned 'un_flags' and usbnet owned 'un_pri'. welcome netbsd 9.99.6.
2019-08-10 05:17:36 +03:00
const char *detname) /* detach cv name */
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
{
- adjust usbnet interface to allow usbd_open_pipe_intr(), from the new comment: * if un_intr_buf is not NULL, use usbd_open_pipe_intr() not * usbd_open_pipe() for USBNET_ENDPT_INTR, with this buffer, * size, and interval. the standard handling is in usbnet.c, with a callback to deal with the interrupt it self. not fully tested, designed for if_aue.c and a few others not yet converted. - make usbhist for usbnet.c work, thanks paulg - usbnet_init_rx_tx() clears out all allocations upon failure now - add usbnet_ec() to get a pointer to the struct ethercom - add usbnet_{lock,unlock,owned}*() to lock/unlock the various locks and *owned*() for asserting welcome 9.99.3!
2019-08-04 11:59:13 +03:00
USBNETHIST_FUNC(); USBNETHIST_CALLED();
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
update usbnet some: - move rx/tx xfer flags into usbnet_cdata - move the callbacks into usbnet_ops structure - move rx/tx xfer flags arguments from usbnet_init_rx_tx() and move them all into usbnet_attach() arguments - s/miibus/mii/ in some places for consistency other clean up: - create wrapper functions for callbacks, move knowledge about special handling (OK to be missing, error eating) there. - use cdata pointer if already available - provide some more macros (will be real functions later) for accessing usbnet members, use existing ones more bump kernel version.
2019-08-09 04:17:33 +03:00
/* Required inputs. */
KASSERT(un->un_ops->uno_tx_prepare);
KASSERT(un->un_ops->uno_rx_loop);
KASSERT(un->un_ops->uno_init);
reduce the scope of struct usbnet: - move a large number of members internal to usbnet.c's new "struct usbnet_private". - provide accessors for a few of these - move struct usbnet_cdata into usbnet.c as well, but move bufsz, list count, and xfer flags back out into struct usbnet, and have them set as part of the setup efore usbnet_attach() - split the intr pipe parts into their own structure - move all the main usbnet*lock* code into usbnet.c too usbnet_attach() goes down to 2 args, and the inputs needed are now the full contents of 'struct usbnet' besides the driver owned 'un_flags' and usbnet owned 'un_pri'. welcome netbsd 9.99.6.
2019-08-10 05:17:36 +03:00
KASSERT(un->un_rx_bufsz);
KASSERT(un->un_tx_bufsz);
KASSERT(un->un_rx_list_cnt);
KASSERT(un->un_tx_list_cnt);
don't set dv_private. it's a bad pattern and only didn't trip kmem_free() lossage because struct usbnet is at the start of the softc. for now, enforce this as part of the ABI. catch up urndis with tx_prepare checking buffer length, and also add an assert to usbnet_start_locked() to match.
2019-08-12 02:55:43 +03:00
/* Unfortunate fact. */
KASSERT(un == device_private(un->un_dev));
reduce the scope of struct usbnet: - move a large number of members internal to usbnet.c's new "struct usbnet_private". - provide accessors for a few of these - move struct usbnet_cdata into usbnet.c as well, but move bufsz, list count, and xfer flags back out into struct usbnet, and have them set as part of the setup efore usbnet_attach() - split the intr pipe parts into their own structure - move all the main usbnet*lock* code into usbnet.c too usbnet_attach() goes down to 2 args, and the inputs needed are now the full contents of 'struct usbnet' besides the driver owned 'un_flags' and usbnet owned 'un_pri'. welcome netbsd 9.99.6.
2019-08-10 05:17:36 +03:00
un->un_pri = kmem_zalloc(sizeof(*un->un_pri), KM_SLEEP);
struct usbnet_private * const unp = un->un_pri;
usbnet: Defer hardware multicast filter updates to USB task. Breaks deadlock: - usbnet_detach holds usbnet lock, awaits kpause in ure_reset - callout holds softclock `lock' (sequential softints, blocks kpause wakeup), awaits softnet_lock in tcp_timer_keep, frag6_fasttimo, &c. - soclose holds softnet_lock, awaits usbnet lock in SIOCDELMULTI This change breaks the deadlock by not passing the SIOCADDMULTI or SIOCDELMULTI ioctl synchronously to the driver, which typically takes the usbnet lock. With this change, the ethernet layer still maintains the list of multicast addresses synchronously, but we defer the driver logic that updates the hardware multicast filter to an asynchronous USB task without softnet_lock held. This doesn't cause exactly the same ioctl to be sent to the driver -- usbnet just sends SIOCDELMULTI with an all-zero struct ifreq, and might drop some ioctls if issued in quick succession. This is OK because none of the drivers actually distinguish between SIOCADDMULTI and SIOCDELMULTI, or examine the argument; the drivers just commit whatever multicast addresses are listed in the ethercom. Other than the different ioctl submitted, there is no change to the ABI or locking scheme of usbnet, so this is safe to pull up to netbsd-9. This means we unfortunately can't guarantee that if a process issues SIOCADDMULTI and then sendto, the multicast filter update will be done by the time of the sendto -- and, more importantly, the packets received in reply to it. But failing to guarantee that is better than deadlocking! Later changes on HEAD will restore the synchronous multicast filter updates with much more extensive ABI changes and API simplifications in usbnet(9). Proposed on tech-net: https://mail-index.netbsd.org/tech-net/2021/12/30/msg008164.html XXX pullup-9
2022-01-30 00:37:07 +03:00
usb_init_task(&unp->unp_ticktask, usbnet_tick_task, un,
USB_TASKQ_MPSAFE);
reduce the scope of struct usbnet: - move a large number of members internal to usbnet.c's new "struct usbnet_private". - provide accessors for a few of these - move struct usbnet_cdata into usbnet.c as well, but move bufsz, list count, and xfer flags back out into struct usbnet, and have them set as part of the setup efore usbnet_attach() - split the intr pipe parts into their own structure - move all the main usbnet*lock* code into usbnet.c too usbnet_attach() goes down to 2 args, and the inputs needed are now the full contents of 'struct usbnet' besides the driver owned 'un_flags' and usbnet owned 'un_pri'. welcome netbsd 9.99.6.
2019-08-10 05:17:36 +03:00
callout_init(&unp->unp_stat_ch, CALLOUT_MPSAFE);
callout_setfunc(&unp->unp_stat_ch, usbnet_tick, un);
mutex_init(&unp->unp_txlock, MUTEX_DEFAULT, IPL_SOFTUSB);
mutex_init(&unp->unp_rxlock, MUTEX_DEFAULT, IPL_SOFTUSB);
Define and implement a locking protocol for the ifmedia / mii layers: - MP-safe drivers provide a mutex to ifmedia that is used to serialize access to media-related structures / hardware regsiters. Converted drivers use the new ifmedia_init_with_lock() function for this. The new name is provided to ease the transition. - Un-converted drivers continue to call ifmedia_init(), which will supply a compatibility lock to be used instead. Several media-related entry points must be aware of this compatibility lock, and are able to acquire it recursively a limited number of times, if needed. This is a SPIN mutex with priority IPL_NET. - This same lock is used to serialize access to PHY registers and other MII-related data structures. The PHY drivers are modified to acquire and release the lock, as needed, and assert the lock is held as a diagnostic aid. The "usbnet" framework has had an overhaul of its internal locking protocols to fit in with the media / mii changes, and the drivers adapted. USB wifi drivers have been changed to provide their own adaptive mutex to the ifmedia later via a new ieee80211_media_init_with_lock() function. This is required because the USB drivers need an adaptive mutex. Besised "usbnet", a few other drivers are converted: vmx, wm, ixgbe / ixv. mcx also now calls ifmedia_init_with_lock() because it needs to also use an adaptive mutex. The mcx driver still needs to be fully converted to NET_MPSAFE.
2020-03-16 02:04:50 +03:00
mutex_init(&unp->unp_core_lock, MUTEX_DEFAULT, IPL_NONE);
usbnet: Apply hardware multicast filter updates synchronously again. To make this work: 1. Do it only under a new lock, unp_mcastlock. This lock lives at IPL_SOFTCLOCK so it can be taken from network stack callouts. It is forbidden to acquire the usbnet core lock under unp_mcastlock. 2. Do it only after usbnet_init_rx_tx and before usbnet_stop; if issued at any other time, drop the update on the floor. 3. Make usbnet_init_rx_tx apply any pending multicast filter updates under the lock before setting the flag that allows SIOCADDMULTI or SIOCDELMULTI to apply the updates. 4. Remove core lock asserts from various drivers' register access routines. This is necessary because the multicast filter updates are done with register reads/writes, but _cannot_ take the core lock when the caller holds softnet_lock. This now programs the hardware multicast filter redundantly in many drivers which already explicitly call *_uno_mcast from the *_uno_init routines. This is probably harmless, but it will likely be better to remove the explicit calls.
2022-03-03 08:53:23 +03:00
mutex_init(&unp->unp_mcastlock, MUTEX_DEFAULT, IPL_SOFTCLOCK);
reduce the scope of struct usbnet: - move a large number of members internal to usbnet.c's new "struct usbnet_private". - provide accessors for a few of these - move struct usbnet_cdata into usbnet.c as well, but move bufsz, list count, and xfer flags back out into struct usbnet, and have them set as part of the setup efore usbnet_attach() - split the intr pipe parts into their own structure - move all the main usbnet*lock* code into usbnet.c too usbnet_attach() goes down to 2 args, and the inputs needed are now the full contents of 'struct usbnet' besides the driver owned 'un_flags' and usbnet owned 'un_pri'. welcome netbsd 9.99.6.
2019-08-10 05:17:36 +03:00
rnd_attach_source(&unp->unp_rndsrc, device_xname(un->un_dev),
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
RND_TYPE_NET, RND_FLAG_DEFAULT);
update usbnet some: - move rx/tx xfer flags into usbnet_cdata - move the callbacks into usbnet_ops structure - move rx/tx xfer flags arguments from usbnet_init_rx_tx() and move them all into usbnet_attach() arguments - s/miibus/mii/ in some places for consistency other clean up: - create wrapper functions for callbacks, move knowledge about special handling (OK to be missing, error eating) there. - use cdata pointer if already available - provide some more macros (will be real functions later) for accessing usbnet members, use existing ones more bump kernel version.
2019-08-09 04:17:33 +03:00
usbnet_rx_list_alloc(un);
usbnet_tx_list_alloc(un);
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
- remove unused uncd_rx_cnt - add USBNETHIST_CALLARGSN and use it in many places - add more history logs - add a log if watchdog abort leaves uncd_tx_cnt non zero - add a unique (32-bit) number for a each usbnet device and use it for logging - explicitly check for full tx ring in usbnet_start_locked() and return early, which avoids setting the 5 second timer and triggering a watchdog - reset uncd_tx_prod and uncd_tx_cnt to zero in stop
2019-08-23 07:29:28 +03:00
unp->unp_number = atomic_inc_uint_nv(&usbnet_number);
reduce the scope of struct usbnet: - move a large number of members internal to usbnet.c's new "struct usbnet_private". - provide accessors for a few of these - move struct usbnet_cdata into usbnet.c as well, but move bufsz, list count, and xfer flags back out into struct usbnet, and have them set as part of the setup efore usbnet_attach() - split the intr pipe parts into their own structure - move all the main usbnet*lock* code into usbnet.c too usbnet_attach() goes down to 2 args, and the inputs needed are now the full contents of 'struct usbnet' besides the driver owned 'un_flags' and usbnet owned 'un_pri'. welcome netbsd 9.99.6.
2019-08-10 05:17:36 +03:00
unp->unp_attached = true;
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
}
static void
couple more changes to usbnet(9): - MII read/write reg return int instead of usbd_status (requested by skrll) - usbnet_attach_ifp(9) changes arg, two mii-specific flags are placed by a pointer to new struct usbnet_mii. if not NULL, then attach an MII to this interface like previous have_mii parameter. use this to allow ure(4) to properly pass PHY location to mii_attach(). welcome netbsd 9.99.10.
2019-08-20 09:37:06 +03:00
usbnet_attach_mii(struct usbnet *un, const struct usbnet_mii *unm)
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
{
- adjust usbnet interface to allow usbd_open_pipe_intr(), from the new comment: * if un_intr_buf is not NULL, use usbd_open_pipe_intr() not * usbd_open_pipe() for USBNET_ENDPT_INTR, with this buffer, * size, and interval. the standard handling is in usbnet.c, with a callback to deal with the interrupt it self. not fully tested, designed for if_aue.c and a few others not yet converted. - make usbhist for usbnet.c work, thanks paulg - usbnet_init_rx_tx() clears out all allocations upon failure now - add usbnet_ec() to get a pointer to the struct ethercom - add usbnet_{lock,unlock,owned}*() to lock/unlock the various locks and *owned*() for asserting welcome 9.99.3!
2019-08-04 11:59:13 +03:00
USBNETHIST_FUNC(); USBNETHIST_CALLED();
reduce the scope of struct usbnet: - move a large number of members internal to usbnet.c's new "struct usbnet_private". - provide accessors for a few of these - move struct usbnet_cdata into usbnet.c as well, but move bufsz, list count, and xfer flags back out into struct usbnet, and have them set as part of the setup efore usbnet_attach() - split the intr pipe parts into their own structure - move all the main usbnet*lock* code into usbnet.c too usbnet_attach() goes down to 2 args, and the inputs needed are now the full contents of 'struct usbnet' besides the driver owned 'un_flags' and usbnet owned 'un_pri'. welcome netbsd 9.99.6.
2019-08-10 05:17:36 +03:00
struct usbnet_private * const unp = un->un_pri;
struct mii_data * const mii = &unp->unp_mii;
struct ifnet * const ifp = usbnet_ifp(un);
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
update usbnet some: - move rx/tx xfer flags into usbnet_cdata - move the callbacks into usbnet_ops structure - move rx/tx xfer flags arguments from usbnet_init_rx_tx() and move them all into usbnet_attach() arguments - s/miibus/mii/ in some places for consistency other clean up: - create wrapper functions for callbacks, move knowledge about special handling (OK to be missing, error eating) there. - use cdata pointer if already available - provide some more macros (will be real functions later) for accessing usbnet members, use existing ones more bump kernel version.
2019-08-09 04:17:33 +03:00
KASSERT(un->un_ops->uno_read_reg);
KASSERT(un->un_ops->uno_write_reg);
KASSERT(un->un_ops->uno_statchg);
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
mii->mii_ifp = ifp;
update usbnet some: - move rx/tx xfer flags into usbnet_cdata - move the callbacks into usbnet_ops structure - move rx/tx xfer flags arguments from usbnet_init_rx_tx() and move them all into usbnet_attach() arguments - s/miibus/mii/ in some places for consistency other clean up: - create wrapper functions for callbacks, move knowledge about special handling (OK to be missing, error eating) there. - use cdata pointer if already available - provide some more macros (will be real functions later) for accessing usbnet members, use existing ones more bump kernel version.
2019-08-09 04:17:33 +03:00
mii->mii_readreg = usbnet_mii_readreg;
mii->mii_writereg = usbnet_mii_writereg;
mii->mii_statchg = usbnet_mii_statchg;
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
mii->mii_flags = MIIF_AUTOTSLEEP;
reduce the scope of struct usbnet: - move a large number of members internal to usbnet.c's new "struct usbnet_private". - provide accessors for a few of these - move struct usbnet_cdata into usbnet.c as well, but move bufsz, list count, and xfer flags back out into struct usbnet, and have them set as part of the setup efore usbnet_attach() - split the intr pipe parts into their own structure - move all the main usbnet*lock* code into usbnet.c too usbnet_attach() goes down to 2 args, and the inputs needed are now the full contents of 'struct usbnet' besides the driver owned 'un_flags' and usbnet owned 'un_pri'. welcome netbsd 9.99.6.
2019-08-10 05:17:36 +03:00
usbnet_ec(un)->ec_mii = mii;
Define and implement a locking protocol for the ifmedia / mii layers: - MP-safe drivers provide a mutex to ifmedia that is used to serialize access to media-related structures / hardware regsiters. Converted drivers use the new ifmedia_init_with_lock() function for this. The new name is provided to ease the transition. - Un-converted drivers continue to call ifmedia_init(), which will supply a compatibility lock to be used instead. Several media-related entry points must be aware of this compatibility lock, and are able to acquire it recursively a limited number of times, if needed. This is a SPIN mutex with priority IPL_NET. - This same lock is used to serialize access to PHY registers and other MII-related data structures. The PHY drivers are modified to acquire and release the lock, as needed, and assert the lock is held as a diagnostic aid. The "usbnet" framework has had an overhaul of its internal locking protocols to fit in with the media / mii changes, and the drivers adapted. USB wifi drivers have been changed to provide their own adaptive mutex to the ifmedia later via a new ieee80211_media_init_with_lock() function. This is required because the USB drivers need an adaptive mutex. Besised "usbnet", a few other drivers are converted: vmx, wm, ixgbe / ixv. mcx also now calls ifmedia_init_with_lock() because it needs to also use an adaptive mutex. The mcx driver still needs to be fully converted to NET_MPSAFE.
2020-03-16 02:04:50 +03:00
ifmedia_init_with_lock(&mii->mii_media, 0,
usbnet_media_upd, ether_mediastatus, usbnet_mutex_core(un));
couple more changes to usbnet(9): - MII read/write reg return int instead of usbd_status (requested by skrll) - usbnet_attach_ifp(9) changes arg, two mii-specific flags are placed by a pointer to new struct usbnet_mii. if not NULL, then attach an MII to this interface like previous have_mii parameter. use this to allow ure(4) to properly pass PHY location to mii_attach(). welcome netbsd 9.99.10.
2019-08-20 09:37:06 +03:00
mii_attach(un->un_dev, mii, unm->un_mii_capmask, unm->un_mii_phyloc,
unm->un_mii_offset, unm->un_mii_flags);
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
if (LIST_FIRST(&mii->mii_phys) == NULL) {
ifmedia_add(&mii->mii_media, IFM_ETHER | IFM_NONE, 0, NULL);
ifmedia_set(&mii->mii_media, IFM_ETHER | IFM_NONE);
} else
ifmedia_set(&mii->mii_media, IFM_ETHER | IFM_AUTO);
}
void
usbnet_attach_ifp(struct usbnet *un,
unsigned if_flags, /* additional if_flags */
couple of minor API updates: - change the read/write register callbacks to have the same phy/reg order as the MII code. - add "mii_flags" param to usbnet_attach_ifp(). axe(4) wants it. also: - add usbnet debug code, sysctl node support - remove commented DPRINTF()s accidentally left in place - add usbnet_softc() - reorder some attach code to be consistent - re-add USBD_FORCE_SHORT_XFER for axen rx chain ride 9.99.2 bump.
2019-08-01 02:47:16 +03:00
unsigned if_extflags, /* additional if_extflags */
couple more changes to usbnet(9): - MII read/write reg return int instead of usbd_status (requested by skrll) - usbnet_attach_ifp(9) changes arg, two mii-specific flags are placed by a pointer to new struct usbnet_mii. if not NULL, then attach an MII to this interface like previous have_mii parameter. use this to allow ure(4) to properly pass PHY location to mii_attach(). welcome netbsd 9.99.10.
2019-08-20 09:37:06 +03:00
const struct usbnet_mii *unm) /* additional mii_attach flags */
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
{
- adjust usbnet interface to allow usbd_open_pipe_intr(), from the new comment: * if un_intr_buf is not NULL, use usbd_open_pipe_intr() not * usbd_open_pipe() for USBNET_ENDPT_INTR, with this buffer, * size, and interval. the standard handling is in usbnet.c, with a callback to deal with the interrupt it self. not fully tested, designed for if_aue.c and a few others not yet converted. - make usbhist for usbnet.c work, thanks paulg - usbnet_init_rx_tx() clears out all allocations upon failure now - add usbnet_ec() to get a pointer to the struct ethercom - add usbnet_{lock,unlock,owned}*() to lock/unlock the various locks and *owned*() for asserting welcome 9.99.3!
2019-08-04 11:59:13 +03:00
USBNETHIST_FUNC(); USBNETHIST_CALLED();
reduce the scope of struct usbnet: - move a large number of members internal to usbnet.c's new "struct usbnet_private". - provide accessors for a few of these - move struct usbnet_cdata into usbnet.c as well, but move bufsz, list count, and xfer flags back out into struct usbnet, and have them set as part of the setup efore usbnet_attach() - split the intr pipe parts into their own structure - move all the main usbnet*lock* code into usbnet.c too usbnet_attach() goes down to 2 args, and the inputs needed are now the full contents of 'struct usbnet' besides the driver owned 'un_flags' and usbnet owned 'un_pri'. welcome netbsd 9.99.6.
2019-08-10 05:17:36 +03:00
struct usbnet_private * const unp = un->un_pri;
struct ifnet * const ifp = usbnet_ifp(un);
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
reduce the scope of struct usbnet: - move a large number of members internal to usbnet.c's new "struct usbnet_private". - provide accessors for a few of these - move struct usbnet_cdata into usbnet.c as well, but move bufsz, list count, and xfer flags back out into struct usbnet, and have them set as part of the setup efore usbnet_attach() - split the intr pipe parts into their own structure - move all the main usbnet*lock* code into usbnet.c too usbnet_attach() goes down to 2 args, and the inputs needed are now the full contents of 'struct usbnet' besides the driver owned 'un_flags' and usbnet owned 'un_pri'. welcome netbsd 9.99.6.
2019-08-10 05:17:36 +03:00
KASSERT(unp->unp_attached);
usbnet: Ensure ifp->if_softc is initialized _before_ publishing ifp. Otherwise other parts of the system might start using ifp the moment we if_register it, and trip over null if_softc.
2022-03-03 08:47:14 +03:00
KASSERT(!unp->unp_ifp_attached);
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
usbnet: Ensure ifp->if_softc is initialized _before_ publishing ifp. Otherwise other parts of the system might start using ifp the moment we if_register it, and trip over null if_softc.
2022-03-03 08:47:14 +03:00
ifp->if_softc = un;
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
strlcpy(ifp->if_xname, device_xname(un->un_dev), IFNAMSIZ);
ifp->if_flags = if_flags;
ifp->if_extflags = IFEF_MPSAFE | if_extflags;
Define and implement a locking protocol for the ifmedia / mii layers: - MP-safe drivers provide a mutex to ifmedia that is used to serialize access to media-related structures / hardware regsiters. Converted drivers use the new ifmedia_init_with_lock() function for this. The new name is provided to ease the transition. - Un-converted drivers continue to call ifmedia_init(), which will supply a compatibility lock to be used instead. Several media-related entry points must be aware of this compatibility lock, and are able to acquire it recursively a limited number of times, if needed. This is a SPIN mutex with priority IPL_NET. - This same lock is used to serialize access to PHY registers and other MII-related data structures. The PHY drivers are modified to acquire and release the lock, as needed, and assert the lock is held as a diagnostic aid. The "usbnet" framework has had an overhaul of its internal locking protocols to fit in with the media / mii changes, and the drivers adapted. USB wifi drivers have been changed to provide their own adaptive mutex to the ifmedia later via a new ieee80211_media_init_with_lock() function. This is required because the USB drivers need an adaptive mutex. Besised "usbnet", a few other drivers are converted: vmx, wm, ixgbe / ixv. mcx also now calls ifmedia_init_with_lock() because it needs to also use an adaptive mutex. The mcx driver still needs to be fully converted to NET_MPSAFE.
2020-03-16 02:04:50 +03:00
ifp->if_ioctl = usbnet_if_ioctl;
ifp->if_start = usbnet_if_start;
ifp->if_init = usbnet_if_init;
ifp->if_stop = usbnet_if_stop;
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
couple more changes to usbnet(9): - MII read/write reg return int instead of usbd_status (requested by skrll) - usbnet_attach_ifp(9) changes arg, two mii-specific flags are placed by a pointer to new struct usbnet_mii. if not NULL, then attach an MII to this interface like previous have_mii parameter. use this to allow ure(4) to properly pass PHY location to mii_attach(). welcome netbsd 9.99.10.
2019-08-20 09:37:06 +03:00
if (unm)
usbnet_attach_mii(un, unm);
- adjust usbnet interface to allow usbd_open_pipe_intr(), from the new comment: * if un_intr_buf is not NULL, use usbd_open_pipe_intr() not * usbd_open_pipe() for USBNET_ENDPT_INTR, with this buffer, * size, and interval. the standard handling is in usbnet.c, with a callback to deal with the interrupt it self. not fully tested, designed for if_aue.c and a few others not yet converted. - make usbhist for usbnet.c work, thanks paulg - usbnet_init_rx_tx() clears out all allocations upon failure now - add usbnet_ec() to get a pointer to the struct ethercom - add usbnet_{lock,unlock,owned}*() to lock/unlock the various locks and *owned*() for asserting welcome 9.99.3!
2019-08-04 11:59:13 +03:00
else
reduce the scope of struct usbnet: - move a large number of members internal to usbnet.c's new "struct usbnet_private". - provide accessors for a few of these - move struct usbnet_cdata into usbnet.c as well, but move bufsz, list count, and xfer flags back out into struct usbnet, and have them set as part of the setup efore usbnet_attach() - split the intr pipe parts into their own structure - move all the main usbnet*lock* code into usbnet.c too usbnet_attach() goes down to 2 args, and the inputs needed are now the full contents of 'struct usbnet' besides the driver owned 'un_flags' and usbnet owned 'un_pri'. welcome netbsd 9.99.6.
2019-08-10 05:17:36 +03:00
unp->unp_link = true;
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
/* Attach the interface. */
if_attach and if_initialize cannot fail, don't test return value These were originally made failable back in 2017 when if_initialize allocated a softint in every interface for link state changes, so that it could fail gracefully instead of panicking: https://mail-index.NetBSD.org/source-changes/2017/10/23/msg089053.html However, this spawned many seldom- or never-tested error branches, which are risky to have around. And that softint in every interface has since been replaced by a single global workqueue, because link state changes require thread context but not low latency or high throughput: https://mail-index.NetBSD.org/source-changes/2020/02/06/msg113759.html So there is no longer any reason for if_initialize to fail. (The subroutine if_stats_init can't fail because percpu_alloc can't fail either.) There is a snag: the softint_establish in if_percpuq_create could fail, potentially leading to bad consequences later on trying to use the softint. This change doesn't introduce any new bugs because of the snag -- if_percpuq_attach was already broken. However, the snag can be better addressed without spawning error branches, either by using a single softint or making softints less scarce. (Separate commit will change the signatures of if_attach and if_initialize to return void, scheduled to ride whatever is the next convenient kernel bump.) Patch and testing on amd64 and evbmips64-eb by maya@; commit message soliloquy, and compile-testing on evbppc/i386/earmv7hf, by me.
2021-06-16 03:21:17 +03:00
if_initialize(ifp);
update usbnet slight: - drivers that want to use if_input() will also set _if_input. for now, avoid attaching a per-cpu queue for them. use if_initialize() and if_register(). - when stopping pipes, don't give up after the first failure, but keep the first failure error for return and keep going - if 0 a KASSERT() in usbnet_init_rx_tx(). there's a path via if_mcast_op() that can have the ifnet unlocked today.. - in usbnet_watchdog(), abort the pipe instead of faking tx completion. avoids issues with devices with more than one tx descriptor, as well as avoiding triggering usb asserts. with these, upl(4) port to usbnet(9) now works. (would be a version bump, but upl(4) and the unported umb(4) are the only consumers that would care.)
2019-08-18 12:29:38 +03:00
if (ifp->_if_input == NULL)
ifp->if_percpuq = if_percpuq_create(ifp);
if_register(ifp);
usbnet: Ensure ifp->if_softc is initialized _before_ publishing ifp. Otherwise other parts of the system might start using ifp the moment we if_register it, and trip over null if_softc.
2022-03-03 08:47:14 +03:00
unp->unp_ifp_attached = true;
extend usbnet to cope with if_upl, if_smsc, and if_umb needs: - usbnet_enqueue() can set mbuf flags and csum_data - usbnet_input() for non-ethernet based devices (upl, umb) - allow a complete override for ioctl() - remove converted list -- we have compiling and/or working patches for all the devices except for umb(4), will be merged as testing happens hopefully this is the last ABI change, though it may end up being extended for additional smsc(4) support. hello for real netbsd 9.99.3!
2019-08-06 03:19:57 +03:00
/*
* If ethernet address is all zero, skip ether_ifattach() and
* instead attach bpf here..
*/
if (!usbnet_empty_eaddr(un)) {
reduce the scope of struct usbnet: - move a large number of members internal to usbnet.c's new "struct usbnet_private". - provide accessors for a few of these - move struct usbnet_cdata into usbnet.c as well, but move bufsz, list count, and xfer flags back out into struct usbnet, and have them set as part of the setup efore usbnet_attach() - split the intr pipe parts into their own structure - move all the main usbnet*lock* code into usbnet.c too usbnet_attach() goes down to 2 args, and the inputs needed are now the full contents of 'struct usbnet' besides the driver owned 'un_flags' and usbnet owned 'un_pri'. welcome netbsd 9.99.6.
2019-08-10 05:17:36 +03:00
ether_set_ifflags_cb(&unp->unp_ec, usbnet_ifflags_cb);
for ethernet usb network devices, print the mac addr in usbnet.
2019-08-06 04:42:22 +03:00
aprint_normal_dev(un->un_dev, "Ethernet address %s\n",
ether_sprintf(un->un_eaddr));
extend usbnet to cope with if_upl, if_smsc, and if_umb needs: - usbnet_enqueue() can set mbuf flags and csum_data - usbnet_input() for non-ethernet based devices (upl, umb) - allow a complete override for ioctl() - remove converted list -- we have compiling and/or working patches for all the devices except for umb(4), will be merged as testing happens hopefully this is the last ABI change, though it may end up being extended for additional smsc(4) support. hello for real netbsd 9.99.3!
2019-08-06 03:19:57 +03:00
ether_ifattach(ifp, un->un_eaddr);
} else {
if_alloc_sadl(ifp);
bpf_attach(ifp, DLT_RAW, 0);
}
fix two issues, found by maya@ while testing urndis: - avoid an KASSERT() in usbnet_rx_tx_init() when called early - move the pmf and usb driver event calls out of the mii specific code, now suspend is able to work on them again. urndis port to usbnet works with additional ipv6 lossage that may indicate it doesn't work (not commited yet.)
2019-08-07 04:47:18 +03:00
update usbnet slight: - drivers that want to use if_input() will also set _if_input. for now, avoid attaching a per-cpu queue for them. use if_initialize() and if_register(). - when stopping pipes, don't give up after the first failure, but keep the first failure error for return and keep going - if 0 a KASSERT() in usbnet_init_rx_tx(). there's a path via if_mcast_op() that can have the ifnet unlocked today.. - in usbnet_watchdog(), abort the pipe instead of faking tx completion. avoids issues with devices with more than one tx descriptor, as well as avoiding triggering usb asserts. with these, upl(4) port to usbnet(9) now works. (would be a version bump, but upl(4) and the unported umb(4) are the only consumers that would care.)
2019-08-18 12:29:38 +03:00
/* Now ready, and attached. */
IFQ_SET_READY(&ifp->if_snd);
fix two issues, found by maya@ while testing urndis: - avoid an KASSERT() in usbnet_rx_tx_init() when called early - move the pmf and usb driver event calls out of the mii specific code, now suspend is able to work on them again. urndis port to usbnet works with additional ipv6 lossage that may indicate it doesn't work (not commited yet.)
2019-08-07 04:47:18 +03:00
usbd_add_drv_event(USB_EVENT_DRIVER_ATTACH, un->un_udev, un->un_dev);
if (!pmf_device_register(un->un_dev, NULL, NULL))
aprint_error_dev(un->un_dev, "couldn't establish power handler\n");
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
}
int
usbnet_detach(device_t self, int flags)
{
- adjust usbnet interface to allow usbd_open_pipe_intr(), from the new comment: * if un_intr_buf is not NULL, use usbd_open_pipe_intr() not * usbd_open_pipe() for USBNET_ENDPT_INTR, with this buffer, * size, and interval. the standard handling is in usbnet.c, with a callback to deal with the interrupt it self. not fully tested, designed for if_aue.c and a few others not yet converted. - make usbhist for usbnet.c work, thanks paulg - usbnet_init_rx_tx() clears out all allocations upon failure now - add usbnet_ec() to get a pointer to the struct ethercom - add usbnet_{lock,unlock,owned}*() to lock/unlock the various locks and *owned*() for asserting welcome 9.99.3!
2019-08-04 11:59:13 +03:00
USBNETHIST_FUNC(); USBNETHIST_CALLED();
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
struct usbnet * const un = device_private(self);
in usbnet_detach(), check both that the private pointer has been allocated and that unp_attached is true before trying to access anything else. fixes a detach problem noticed by maxv when, eg, the call to usbd_set_config_no() fails and attach bails early.
2019-08-28 09:07:21 +03:00
struct usbnet_private * const unp = un->un_pri;
/* Detached before attached finished, so just bail out. */
if (unp == NULL || !unp->unp_attached)
return 0;
reduce the scope of struct usbnet: - move a large number of members internal to usbnet.c's new "struct usbnet_private". - provide accessors for a few of these - move struct usbnet_cdata into usbnet.c as well, but move bufsz, list count, and xfer flags back out into struct usbnet, and have them set as part of the setup efore usbnet_attach() - split the intr pipe parts into their own structure - move all the main usbnet*lock* code into usbnet.c too usbnet_attach() goes down to 2 args, and the inputs needed are now the full contents of 'struct usbnet' besides the driver owned 'un_flags' and usbnet owned 'un_pri'. welcome netbsd 9.99.6.
2019-08-10 05:17:36 +03:00
struct ifnet * const ifp = usbnet_ifp(un);
struct mii_data * const mii = usbnet_mii(un);
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
usbnet: Refuse to bring interfaces back up once dying. Make this happen uniformly across all usbnet drivers, not on a per-driver basis. This ensures new activity on the interface can't happen by the time we have stopped existing activity and waited for it to complete.
2022-03-03 08:48:37 +03:00
/*
* Prevent new activity. After we stop the interface, it
* cannot be brought back up.
*/
usbnet: Use atomic_load/store_relaxed for unp_dying. This way we don't need to hold the core lock to avoid upsetting sanitizers (which probably find the current code upsetting), and we can use it to exit early from timeout loops that run under the core lock (which is probably not necessary for them to do anyway, but let's worry about that later).
2022-03-03 08:50:39 +03:00
atomic_store_relaxed(&unp->unp_dying, true);
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
usbnet: Refuse to bring interfaces back up once dying. Make this happen uniformly across all usbnet drivers, not on a per-driver basis. This ensures new activity on the interface can't happen by the time we have stopped existing activity and waited for it to complete.
2022-03-03 08:48:37 +03:00
/*
* If we're still running on the network, stop and wait for all
* asynchronous activity to finish.
usbnet: Avoid IFNET_LOCK on detach if we never attached the ifp.
2022-03-03 08:49:44 +03:00
*
* If usbnet_attach_ifp never ran, IFNET_LOCK won't work, but
* no activity is possible, so just skip this part.
usbnet: Refuse to bring interfaces back up once dying. Make this happen uniformly across all usbnet drivers, not on a per-driver basis. This ensures new activity on the interface can't happen by the time we have stopped existing activity and waited for it to complete.
2022-03-03 08:48:37 +03:00
*/
usbnet: Avoid IFNET_LOCK on detach if we never attached the ifp.
2022-03-03 08:49:44 +03:00
if (unp->unp_ifp_attached) {
IFNET_LOCK(ifp);
if (ifp->if_flags & IFF_RUNNING) {
usbnet_if_stop(ifp, 1);
}
IFNET_UNLOCK(ifp);
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
}
usbnet: Omit needless callout_halt and usb_rem_task_wait. The callout and tasks cannot be pending at this point -- it is a bug if usbnet_if_stop failed to quiesce everything, so turn these into KASSERTs.
2022-03-03 08:48:45 +03:00
/*
* The callout and tick task can't be scheduled anew at this
* point, and usbnet_if_stop has waited for them to complete.
*/
KASSERT(!callout_pending(&unp->unp_stat_ch));
KASSERT(!usb_task_pending(un->un_udev, &unp->unp_ticktask));
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
if (mii) {
mii_detach(mii, MII_PHY_ANY, MII_OFFSET_ANY);
Use ifmedia_fini().
2020-02-04 08:44:14 +03:00
ifmedia_fini(&mii->mii_media);
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
}
usbnet: Ensure ifp->if_softc is initialized _before_ publishing ifp. Otherwise other parts of the system might start using ifp the moment we if_register it, and trip over null if_softc.
2022-03-03 08:47:14 +03:00
if (unp->unp_ifp_attached) {
extend usbnet to cope with if_upl, if_smsc, and if_umb needs: - usbnet_enqueue() can set mbuf flags and csum_data - usbnet_input() for non-ethernet based devices (upl, umb) - allow a complete override for ioctl() - remove converted list -- we have compiling and/or working patches for all the devices except for umb(4), will be merged as testing happens hopefully this is the last ABI change, though it may end up being extended for additional smsc(4) support. hello for real netbsd 9.99.3!
2019-08-06 03:19:57 +03:00
if (!usbnet_empty_eaddr(un))
ether_ifdetach(ifp);
else
bpf_detach(ifp);
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
if_detach(ifp);
}
Fix order of nulling un->un_pri->unp_ec.ec_mii. Can't null it until after if_detach prevents further use. While here, fix conditionals in usbnet_tick_task to use the unp_dying flag, not the nullness of mii (or of ifp, which never null because it's an embedded member).
2019-12-03 08:01:45 +03:00
usbnet_ec(un)->ec_mii = NULL;
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
usbnet: Detach interface and mii before waiting for refcnt to drain. All outstanding software activity under usbnet's control -- which is all that participates in the refcnting -- should be quiesced by stopping and detaching everything.
2022-03-03 08:48:52 +03:00
usbnet_rx_list_free(un);
usbnet_tx_list_free(un);
rnd_detach_source(&unp->unp_rndsrc);
usbnet: Apply hardware multicast filter updates synchronously again. To make this work: 1. Do it only under a new lock, unp_mcastlock. This lock lives at IPL_SOFTCLOCK so it can be taken from network stack callouts. It is forbidden to acquire the usbnet core lock under unp_mcastlock. 2. Do it only after usbnet_init_rx_tx and before usbnet_stop; if issued at any other time, drop the update on the floor. 3. Make usbnet_init_rx_tx apply any pending multicast filter updates under the lock before setting the flag that allows SIOCADDMULTI or SIOCDELMULTI to apply the updates. 4. Remove core lock asserts from various drivers' register access routines. This is necessary because the multicast filter updates are done with register reads/writes, but _cannot_ take the core lock when the caller holds softnet_lock. This now programs the hardware multicast filter redundantly in many drivers which already explicitly call *_uno_mcast from the *_uno_init routines. This is probably harmless, but it will likely be better to remove the explicit calls.
2022-03-03 08:53:23 +03:00
mutex_destroy(&unp->unp_mcastlock);
Define and implement a locking protocol for the ifmedia / mii layers: - MP-safe drivers provide a mutex to ifmedia that is used to serialize access to media-related structures / hardware regsiters. Converted drivers use the new ifmedia_init_with_lock() function for this. The new name is provided to ease the transition. - Un-converted drivers continue to call ifmedia_init(), which will supply a compatibility lock to be used instead. Several media-related entry points must be aware of this compatibility lock, and are able to acquire it recursively a limited number of times, if needed. This is a SPIN mutex with priority IPL_NET. - This same lock is used to serialize access to PHY registers and other MII-related data structures. The PHY drivers are modified to acquire and release the lock, as needed, and assert the lock is held as a diagnostic aid. The "usbnet" framework has had an overhaul of its internal locking protocols to fit in with the media / mii changes, and the drivers adapted. USB wifi drivers have been changed to provide their own adaptive mutex to the ifmedia later via a new ieee80211_media_init_with_lock() function. This is required because the USB drivers need an adaptive mutex. Besised "usbnet", a few other drivers are converted: vmx, wm, ixgbe / ixv. mcx also now calls ifmedia_init_with_lock() because it needs to also use an adaptive mutex. The mcx driver still needs to be fully converted to NET_MPSAFE.
2020-03-16 02:04:50 +03:00
mutex_destroy(&unp->unp_core_lock);
reduce the scope of struct usbnet: - move a large number of members internal to usbnet.c's new "struct usbnet_private". - provide accessors for a few of these - move struct usbnet_cdata into usbnet.c as well, but move bufsz, list count, and xfer flags back out into struct usbnet, and have them set as part of the setup efore usbnet_attach() - split the intr pipe parts into their own structure - move all the main usbnet*lock* code into usbnet.c too usbnet_attach() goes down to 2 args, and the inputs needed are now the full contents of 'struct usbnet' besides the driver owned 'un_flags' and usbnet owned 'un_pri'. welcome netbsd 9.99.6.
2019-08-10 05:17:36 +03:00
mutex_destroy(&unp->unp_rxlock);
mutex_destroy(&unp->unp_txlock);
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
usbnet: Make detach order reverse attach order, for unp_stat_ch. No functional change intended.
2022-03-03 08:49:00 +03:00
callout_destroy(&unp->unp_stat_ch);
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
pmf_device_deregister(un->un_dev);
usbnet: Don't issue a detach event if we never issued an attach one.
2022-03-03 08:49:07 +03:00
/*
* Notify userland that we're going away, if we arrived in the
* first place.
*/
if (unp->unp_ifp_attached) {
usbd_add_drv_event(USB_EVENT_DRIVER_DETACH, un->un_udev,
un->un_dev);
}
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
reduce the scope of struct usbnet: - move a large number of members internal to usbnet.c's new "struct usbnet_private". - provide accessors for a few of these - move struct usbnet_cdata into usbnet.c as well, but move bufsz, list count, and xfer flags back out into struct usbnet, and have them set as part of the setup efore usbnet_attach() - split the intr pipe parts into their own structure - move all the main usbnet*lock* code into usbnet.c too usbnet_attach() goes down to 2 args, and the inputs needed are now the full contents of 'struct usbnet' besides the driver owned 'un_flags' and usbnet owned 'un_pri'. welcome netbsd 9.99.6.
2019-08-10 05:17:36 +03:00
kmem_free(unp, sizeof(*unp));
- use CALLARGS vs CALLED for better usbhist - turn off usbnetdebug default - log for all entry/exit points of usbnet_pipe_intr() - in usbnet_start_locked() track whether any packet has been transmitted for setting the timer. avoids spurious "watchdog timeouts" - in usbnet_stop() use callout_halt() vs callout_halt, and also stop the usb task. fixes crash of usbtask after the phy has detached. - add a little more defensive checking in the tick task, and add some high-log-level logs. - in usbnet_detach() move the call to usbnet_stop_ifp() above the calls to callout/usbtask stopping. - set ec_mii and unp_pri to NULL when freeing their data
2019-09-08 21:59:32 +03:00
un->un_pri = NULL;
reduce the scope of struct usbnet: - move a large number of members internal to usbnet.c's new "struct usbnet_private". - provide accessors for a few of these - move struct usbnet_cdata into usbnet.c as well, but move bufsz, list count, and xfer flags back out into struct usbnet, and have them set as part of the setup efore usbnet_attach() - split the intr pipe parts into their own structure - move all the main usbnet*lock* code into usbnet.c too usbnet_attach() goes down to 2 args, and the inputs needed are now the full contents of 'struct usbnet' besides the driver owned 'un_flags' and usbnet owned 'un_pri'. welcome netbsd 9.99.6.
2019-08-10 05:17:36 +03:00
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
return 0;
}
int
usbnet_activate(device_t self, devact_t act)
{
- adjust usbnet interface to allow usbd_open_pipe_intr(), from the new comment: * if un_intr_buf is not NULL, use usbd_open_pipe_intr() not * usbd_open_pipe() for USBNET_ENDPT_INTR, with this buffer, * size, and interval. the standard handling is in usbnet.c, with a callback to deal with the interrupt it self. not fully tested, designed for if_aue.c and a few others not yet converted. - make usbhist for usbnet.c work, thanks paulg - usbnet_init_rx_tx() clears out all allocations upon failure now - add usbnet_ec() to get a pointer to the struct ethercom - add usbnet_{lock,unlock,owned}*() to lock/unlock the various locks and *owned*() for asserting welcome 9.99.3!
2019-08-04 11:59:13 +03:00
USBNETHIST_FUNC(); USBNETHIST_CALLED();
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
struct usbnet * const un = device_private(self);
reduce the scope of struct usbnet: - move a large number of members internal to usbnet.c's new "struct usbnet_private". - provide accessors for a few of these - move struct usbnet_cdata into usbnet.c as well, but move bufsz, list count, and xfer flags back out into struct usbnet, and have them set as part of the setup efore usbnet_attach() - split the intr pipe parts into their own structure - move all the main usbnet*lock* code into usbnet.c too usbnet_attach() goes down to 2 args, and the inputs needed are now the full contents of 'struct usbnet' besides the driver owned 'un_flags' and usbnet owned 'un_pri'. welcome netbsd 9.99.6.
2019-08-10 05:17:36 +03:00
struct usbnet_private * const unp = un->un_pri;
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
struct ifnet * const ifp = usbnet_ifp(un);
switch (act) {
case DVACT_DEACTIVATE:
if_deactivate(ifp);
usbnet: Use atomic_load/store_relaxed for unp_dying. This way we don't need to hold the core lock to avoid upsetting sanitizers (which probably find the current code upsetting), and we can use it to exit early from timeout loops that run under the core lock (which is probably not necessary for them to do anyway, but let's worry about that later).
2022-03-03 08:50:39 +03:00
atomic_store_relaxed(&unp->unp_dying, true);
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
reduce the scope of struct usbnet: - move a large number of members internal to usbnet.c's new "struct usbnet_private". - provide accessors for a few of these - move struct usbnet_cdata into usbnet.c as well, but move bufsz, list count, and xfer flags back out into struct usbnet, and have them set as part of the setup efore usbnet_attach() - split the intr pipe parts into their own structure - move all the main usbnet*lock* code into usbnet.c too usbnet_attach() goes down to 2 args, and the inputs needed are now the full contents of 'struct usbnet' besides the driver owned 'un_flags' and usbnet owned 'un_pri'. welcome netbsd 9.99.6.
2019-08-10 05:17:36 +03:00
mutex_enter(&unp->unp_rxlock);
mutex_enter(&unp->unp_txlock);
unp->unp_stopping = true;
mutex_exit(&unp->unp_txlock);
mutex_exit(&unp->unp_rxlock);
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
return 0;
default:
return EOPNOTSUPP;
}
}
MODULE(MODULE_CLASS_MISC, usbnet, NULL);
static int
usbnet_modcmd(modcmd_t cmd, void *arg)
{
switch (cmd) {
case MODULE_CMD_INIT:
- adjust usbnet interface to allow usbd_open_pipe_intr(), from the new comment: * if un_intr_buf is not NULL, use usbd_open_pipe_intr() not * usbd_open_pipe() for USBNET_ENDPT_INTR, with this buffer, * size, and interval. the standard handling is in usbnet.c, with a callback to deal with the interrupt it self. not fully tested, designed for if_aue.c and a few others not yet converted. - make usbhist for usbnet.c work, thanks paulg - usbnet_init_rx_tx() clears out all allocations upon failure now - add usbnet_ec() to get a pointer to the struct ethercom - add usbnet_{lock,unlock,owned}*() to lock/unlock the various locks and *owned*() for asserting welcome 9.99.3!
2019-08-04 11:59:13 +03:00
return 0;
introduce a library of common code / backends to share code between USB ethernet drivers. usbnet.h introduces a new set of APIs to provide common solutions for these driver features: - USB endpoint pipe handling - rx and tx chain handling - generic handlers or support for several struct ifnet callbacks - MII bus locking - interrupt handling - partial autoconf handling: much of attach, and detach/activate can use common versions directly. currently, only axen(4) and cdce(4) are converted. the reductions in these drivers are quite significant: if_cdce.c is reduced from 1000 lines to 320 lines, and if_axen is reduced from 1902 lines to 1021 lines. add a "usbnet" module and make the if_axen module depend upon it.
2019-07-31 12:13:16 +03:00
case MODULE_CMD_FINI:
return 0;
case MODULE_CMD_STAT:
case MODULE_CMD_AUTOUNLOAD:
default:
return ENOTTY;
}
}