| 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: <51266897.9070308@linaro.org>
Date: Thu, 21 Feb 2013 10:33:59 -0800
From: John Stultz <john.stultz@...aro.org>
To: Kukjin Kim <kgene.kim@...sung.com>,
Andrzej Pietrasiewicz <andrzej.p@...sung.com>
CC: Sachin Kamat <sachin.kamat@...aro.org>,
Tushar Behera <tushar.behera@...aro.org>,
lkml <linux-kernel@...r.kernel.org>,
Sebastian Andrzej Siewior <bigeasy@...utronix.de>
Subject: Re: Origen board hang with functionfs
On 02/20/2013 06:01 PM, John Stultz wrote:
> Hey Kukjin, Andrzej,
> I recently started playing around with functionfs, and have
> noticed some strange behavior with my origen board.
>
> If I enable the FunctionFS gadget driver, I see the board hang at boot
> here:
>
> [ 2.360000] USB Mass Storage support registered.
> [ 2.365000] s3c-hsotg s3c-hsotg: regs f0040000, irq 103
> [ 2.375000] s3c-hsotg s3c-hsotg: EPs:15
> [ 2.380000] s3c-hsotg s3c-hsotg: dedicated fifos
> [ 2.385000] g_ffs: file system registered
> [ 2.390000] mousedev: PS/2 mouse device common for all mice
> [ 2.395000] s5p-fimc exynos4-fimc.0: start latency exceeded, new
> value 792 ns
> [ 2.400000] s5p-fimc exynos4-fimc.0: state restore latency
> exceeded, new vals
> [ 2.410000] s5p-fimc exynos4-fimc.0: stop latency exceeded, new
> value 667 ns
> [ 2.420000] s5p-fimc exynos4-fimc.0: state save latency exceeded,
> new value s
> [ 2.425000] s5p-fimc exynos4-fimc.1: start latency exceeded, new
> value 625 ns
> [ 2.435000] s5p-fimc exynos4-fimc.1: state restore latency
> exceeded, new vals
> [ 2.440000] s5p-fimc exynos4-fimc.1: stop latency exceeded, new
> value 750 ns
> [ 2.450000] s5p-fimc exynos4-fimc.1: start latency exceeded, new
> value 667 ns
> [ 2.455000] s5p-fimc exynos4-fimc.1: state save latency exceeded,
> new value s
>
>
> And it gets no further. Trying to send a sysrq-t (or anything after a
> break) doesn't seem to do anything either.
>
> Trying to debug this a bit more, I added "initcall_debug
> ignore_loglevel" to the boot arguments, but strangely, that allows it
> to boot and everything seems to be fine. I added a number of printks
> to the g_ffs driver, but it doesn't seem like the hang is happening in
> that code (though de-configuring it does avoid the issue - so maybe
> not?).
So I also enabled lockdep to see if it might show a locking problem even
when initcall_debug avoids the race. Unfortunately nothing shows up at
boot, but when I try to use FunctionFS I got the following lockdep
splat, which then triggered a warning and then finally an oops all in
succession.
The core lockdep parts are below, and the full log is attached. (Sorry
its cropped at 80 lines, I wasn't logging my screen session at the time,
so its all copy-pasted. Let me know if you'd need me to recreate the
problem.)
[ 109.290000] =========================================================
[ 109.295000] [ INFO: possible irq lock inversion dependency detected ]
[ 109.305000] 3.8.0-dirty #199 Not tainted
[ 109.310000] ---------------------------------------------------------
[ 109.315000] swapper/0/0 just changed the state of lock:
[ 109.320000] (&(&hsotg->lock)->rlock){-.....}, at: [<c029d35c>]
s3c_hsotg_ir8
[ 109.325000] but this lock took another, HARDIRQ-unsafe lock in the past:
[ 109.335000] (clocks_mutex){+.+...}
and interrupts could create inverse lock ordering between them.
[ 109.345000]
[ 109.345000] other info that might help us debug this:
[ 109.350000] Possible interrupt unsafe locking scenario:
[ 109.350000]
[ 109.355000] CPU0 CPU1
[ 109.360000] ---- ----
[ 109.365000] lock(clocks_mutex);
[ 109.370000] local_irq_disable();
[ 109.375000] lock(&(&hsotg->lock)->rlock);
[ 109.380000] lock(clocks_mutex);
[ 109.390000] <Interrupt>
[ 109.390000] lock(&(&hsotg->lock)->rlock);
[ 109.395000]
[ 109.395000] *** DEADLOCK ***
[ 109.395000]
[ 109.400000] no locks held by swapper/0/0.
[ 109.405000]
[ 109.405000] the shortest dependencies between 2nd lock and 1st lock:
[ 109.410000] -> (clocks_mutex){+.+...} ops: 221 {
[ 109.415000] HARDIRQ-ON-W at:
[ 109.420000] [<c0064b08>] __lock_acquire+0x560/0xa94
[ 109.425000] [<c0066684>] lock_acquire+0x60/0x74
[ 109.435000] [<c045b010>]
mutex_lock_nested+0x54/0x3a8
[ 109.440000] [<c0310888>] clk_get_sys+0x24/0xf4
[ 109.445000] [<c059994c>]
exynos_pm_drvinit+0x2c/0x64
[ 109.455000] [<c05918e0>] do_one_initcall+0x50/0x164
[ 109.460000] [<c0591b0c>]
kernel_init_freeable+0x118/0xc
[ 109.470000] [<c0451494>] kernel_init+0x8/0xe4
[ 109.475000] [<c000e1b0>] ret_from_fork+0x14/0x24
[ 109.480000] SOFTIRQ-ON-W at:
[ 109.485000] [<c0064b2c>] __lock_acquire+0x584/0xa94
[ 109.490000] [<c0066684>] lock_acquire+0x60/0x74
[ 109.495000] [<c045b010>]
mutex_lock_nested+0x54/0x3a8
[ 109.505000] [<c0310888>] clk_get_sys+0x24/0xf4
[ 109.510000] [<c059994c>]
exynos_pm_drvinit+0x2c/0x64
[ 109.515000] [<c05918e0>] do_one_initcall+0x50/0x164
[ 109.525000] [<c0591b0c>]
kernel_init_freeable+0x118/0xc
[ 109.530000] [<c0451494>] kernel_init+0x8/0xe4
[ 109.535000] [<c000e1b0>] ret_from_fork+0x14/0x24
[ 109.545000] INITIAL USE at:
[ 109.545000] [<c0064840>] __lock_acquire+0x298/0xa94
[ 109.555000] [<c0066684>] lock_acquire+0x60/0x74
[ 109.560000] [<c045b010>]
mutex_lock_nested+0x54/0x3a8
[ 109.565000] [<c0310998>] clkdev_add+0x1c/0x3c
[ 109.575000] [<c001fcd4>]
s3c24xx_register_clock+0x34/0x0
[ 109.580000] [<c059b3b0>]
s3c24xx_register_baseclocks+0x4
[ 109.590000] [<c0598d1c>]
exynos4_init_clocks+0x1c/0x80
[ 109.595000] [<c059ae00>]
s3c24xx_init_clocks+0x44/0x54
[ 109.600000] [<c059a848>] origen_map_io+0x1c/0x34
[ 109.610000] [<c0597d6c>] paging_init+0x5ec/0x798
[ 109.615000] [<c0594c00>] setup_arch+0x3c8/0x65c
[ 109.620000] [<c0591588>] start_kernel+0x7c/0x384
[ 109.625000] [<40008078>] 0x40008078
[ 109.630000] }
View attachment "g_ffs-s3c-clock-potential-deadlock.txt" of type "text/plain" (15683 bytes)
Powered by blists - more mailing lists