| lists.openwall.net | lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC | |
|
Open Source and information security mailing list archives
| ||
|
Message-Id: <200906081615.45889.florian@openwrt.org>
Date: Mon, 8 Jun 2009 16:15:44 +0200
From: Florian Fainelli <florian@...nwrt.org>
To: matthieu castet <castet.matthieu@...e.fr>
Cc: Andrew Morton <akpm@...ux-foundation.org>, wim@...ana.be,
linux-kernel@...r.kernel.org, linux-mips@...ux-mips.org,
biblbroks@...ampro.rs
Subject: Re: add bcm47xx watchdog driver
Le Friday 05 June 2009 22:30:21 matthieu castet, vous avez écrit :
> Andrew Morton wrote:
> > On Thu, 04 Jun 2009 22:24:56 +0200
> >
> > matthieu castet <castet.matthieu@...e.fr> wrote:
> >> This add watchdog driver for broadcom 47xx device.
> >> It uses the ssb subsytem to access embeded watchdog device.
> >>
> >> Because the watchdog timeout is very short (about 2s), a soft timer is
> >> used to increase the watchdog period.
> >>
> >> Note : A patch for exporting the ssb_watchdog_timer_set will
> >> be submitted on next linux-mips merge. Without this patch it can't
> >> be build as a module.
It works very well on my Netgear WGT634U, thanks !
Tested-by: Florian Fainelli <florian@...nwrt.org>
> >>
> >>
> >> ...
> >>
> >> --- linux-2.6.orig/drivers/watchdog/Kconfig 2009-05-25
> >> 22:22:02.000000000 +0200 +++
> >> linux-2.6/drivers/watchdog/Kconfig 2009-05-25 22:26:06.000000000 +0200
> >> @@ -764,6 +764,12 @@
> >> help
> >> Hardware driver for the built-in watchdog timer on TXx9 MIPS SoCs.
> >>
> >> +config BCM47XX_WDT
> >> + tristate "Broadcom BCM47xx Watchdog Timer"
> >> + depends on BCM47XX
> >> + help
> >> + Hardware driver for the Broadcom BCM47xx Watchog Timer.
> >> +
> >
> > Please indent the kconfig body with a single tab character.
>
> Done
>
> >> ...
> >>
> >> +#define DRV_NAME "bcm47xx_wdt"
> >> +
> >> +#define WDT_DEFAULT_TIME 30 /* seconds */
> >> +#define WDT_MAX_TIME 256 /* seconds */
> >> +
> >> +static int wdt_time = WDT_DEFAULT_TIME;
> >> +static int nowayout = WATCHDOG_NOWAYOUT;
> >> +
> >> +module_param(wdt_time, int, 0);
> >> +MODULE_PARM_DESC(wdt_time, "Watchdog time in seconds. (default="
> >> + __MODULE_STRING(WDT_DEFAULT_TIME) ")");
> >> +
> >> +#ifdef CONFIG_WATCHDOG_NOWAYOUT
> >> +module_param(nowayout, int, 0);
> >> +MODULE_PARM_DESC(nowayout,
> >> + "Watchdog cannot be stopped once started (default="
> >> + __MODULE_STRING(WATCHDOG_NOWAYOUT) ")");
> >> +#endif
> >
> > hm, now what's happening with the third arg to module_param()?
>
> I don't understand what you mean.
> This thing is common in watchdog drivers. For example
> drivers/watchdog/at91rm9200_wdt.c does the same thing.
>
> >> +static struct platform_device *bcm47xx_wdt_platform_device;
> >> +
> >> +static unsigned long bcm47xx_wdt_busy;
> >> +static char expect_release;
> >> +static struct timer_list wdt_timer;
> >> +static atomic_t ticks;
> >> +
> >> +static inline void bcm47xx_wdt_hw_start(void)
> >> +{
> >> + /* this is 2,5s on 100Mhz clock and 2s on 133 Mhz */
> >> + ssb_watchdog_timer_set(&ssb_bcm47xx, 0xfffffff);
> >> +}
> >> +
> >> +static inline int bcm47xx_wdt_hw_stop(void)
> >> +{
> >> + return ssb_watchdog_timer_set(&ssb_bcm47xx, 0);
> >> +}
> >> +
> >> +static void bcm47xx_timer_tick(unsigned long unused)
> >> +{
> >> + if(!atomic_dec_and_test(&ticks)) {
> >
> > Please pass this patch (and all others) through scripts/checkpatch.pl
> > and review the resulting output.
>
> Done, everything is ok, expect a printk line over 80 characters.
>
> >> + bcm47xx_wdt_hw_start();
> >> + mod_timer(&wdt_timer, jiffies + HZ);
> >> + }
> >> + else {
> >> + printk(KERN_CRIT PFX "Watchdog will fire soon!!!.\n");
> >> + }
> >> +}
> >> +
> >> +static inline void bcm47xx_wdt_pet(void)
> >> +{
> >> + atomic_set(&ticks, wdt_time);
> >> +}
> >
> > What does "pet" stand for?
>
> A watchdog timer is a computer hardware timing device that triggers a
> system reset if the main program, due to some fault condition, such as a
> hang, neglects to regularly service the watchdog (writing a “service
> pulse” to it, also referred to as “petting the dog” [1]
>
> [1] http://en.wikipedia.org/wiki/Watchdog_timer
>
> But I can change the name if you want. Note that pet appear in
> drivers/watchdog/sb_wdog.c and drivers/watchdog/sbc_epx_c3.c
>
> >> +static void bcm47xx_wdt_start(void)
> >> +{
> >> + bcm47xx_wdt_pet();
> >> + bcm47xx_timer_tick(0);
> >> +}
> >> +
> >> +static void bcm47xx_wdt_pause(void)
> >> +{
> >> + del_timer(&wdt_timer);
> >
> > Should this be del_timer_sync()? The timer callback can still be
> > executing after del_timer() returns.
>
> Yes, changed to del_timer_sync()
>
> >> +static int bcm47xx_wdt_release(struct inode *inode, struct file *file)
> >> +{
> >> + if (expect_release == 42) {
> >> + bcm47xx_wdt_stop();
> >> + } else {
> >> + printk(KERN_CRIT DRV_NAME ": Unexpected close, not stopping
> >> watchdog!\n");
> >
> > Can this happen?
>
> yes : this is a common pattern in watchdog driver (check for example
> softdog) :
> - expect_release is in bss (set to 0)
> - we set expect_release to this magic value, only if we get a write with
> a special character and we are not in nowayout.
> - So for example doing a "cat /dev/watchdog" should go in this path.
>
> >> + bcm47xx_wdt_start();
> >> + }
> >> +
> >> + clear_bit(0, &bcm47xx_wdt_busy);
> >> + expect_release = 0;
> >> + return 0;
> >> +}
> >> +
>
> Thanks for the review.
>
> I attach a new version.
--
Best regards, Florian Fainelli
Email : florian@...nwrt.org
http://openwrt.org
-------------------------------
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
Powered by blists - more mailing lists