lkml.org 
[lkml]   [2022]   [Sep]   [20]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
SubjectRe: [PATCH 1/1] riscv: enable Docker requirements in defconfig
From
On 9/20/22 13:25, Palmer Dabbelt wrote:
> On Sat, 13 Aug 2022 08:58:51 PDT (-0700), linux@roeck-us.net wrote:
>> On Wed, Jun 08, 2022 at 02:02:51AM +0200, Heinrich Schuchardt wrote:
>>> The defconfig kernel should be able to run Docker.
>>>
>>> Enable the missing settings according to [1].
>>> make savedefconfig eliminates CONFIG_STACKTRACE
>>> which is enabled by default.
>>>
>>> Many of the settings are also needed to run a defconfig kernel
>>> on default distro installations or to run snaps.
>>>
>>> [1] https://github.com/moby/moby/blob/master/contrib/check-config.sh
>>>
>>> Signed-off-by: Heinrich Schuchardt <heinrich.schuchardt@canonical.com>
>>
>> This patch results in (or exposes) the following boot warning with all
>> riscv32 and riscv64 qemu boots.
>>
>> [   12.278014] DEBUG_LOCKS_WARN_ON(!lockdep_hardirqs_enabled())
>> [   12.278218] WARNING: CPU: 0 PID: 140 at kernel/locking/lockdep.c:5510 check_flags+0xe2/0x1c2
>> [   12.278446] Modules linked in:
>> [   12.278623] CPU: 0 PID: 140 Comm: S01syslogd Tainted: G                 N 5.19.0-14184-g69dac8e431af #1
>> [   12.278791] Hardware name: riscv-virtio,qemu (DT)
>> [   12.278931] epc : check_flags+0xe2/0x1c2
>> [   12.279024]  ra : check_flags+0xe2/0x1c2
>> [   12.279092] epc : ffffffff80a8ea50 ra : ffffffff80a8ea50 sp : ff200000107f3cb0
>> [   12.279180]  gp : ffffffff819fec50 tp : ff600000049895c0 t0 : ffffffff81833520
>> [   12.279287]  t1 : 0000000000000001 t2 : 2d2d2d2d2d2d2d2d s0 : ff200000107f3cd0
>> [   12.279395]  s1 : ffffffff81a00130 a0 : 0000000000000030 a1 : ffffffff818925e8
>> [   12.279502]  a2 : 0000000000000010 a3 : fffffffffffffffe a4 : 0000000000000000
>> [   12.279608]  a5 : 0000000000000000 a6 : ffffffff80066d80 a7 : 0000000000000038
>> [   12.279714]  s2 : ffffffff81a024e0 s3 : ffffffff818935b0 s4 : ff200000107f3e28
>> [   12.279825]  s5 : ffffffff80e27708 s6 : 0000000200000022 s7 : ffffffffffffffff
>> [   12.279932]  s8 : ffffffff8249da18 s9 : 0000000000001000 s10: 0000000000000001
>> [   12.280037]  s11: 0000000000000001 t3 : ffffffff824740af t4 : ffffffff824740af
>> [   12.280142]  t5 : ffffffff824740b0 t6 : ff200000107f3aa8
>> [   12.280227] status: 0000000200000100 badaddr: 0000000000000000 cause: 0000000000000003
>> [   12.280382] [<ffffffff80a8eba8>] lock_is_held_type+0x78/0x14a
>> [   12.280505] [<ffffffff8003d648>] __might_resched+0x26/0x232
>> [   12.280608] [<ffffffff8003d892>] __might_sleep+0x3e/0x66
>> [   12.280690] [<ffffffff8002262c>] get_signal+0xa6/0x8f6
>> [   12.280771] [<ffffffff800051da>] do_notify_resume+0x68/0x418
>> [   12.280863] [<ffffffff80003af4>] ret_from_exception+0x0/0x10
>> [   12.280978] irq event stamp: 2900
>> [   12.281041] hardirqs last  enabled at (2899): [<ffffffff80a9761c>] _raw_spin_unlock_irqrestore+0x54/0x62
>> [   12.281176] hardirqs last disabled at (2900): [<ffffffff8000812c>] __trace_hardirqs_off+0xc/0x14
>> [   12.281303] softirqs last  enabled at (1860): [<ffffffff80a98092>] __do_softirq+0x3e2/0x51c
>> [   12.281435] softirqs last disabled at (1843): [<ffffffff80016fc8>] __irq_exit_rcu+0xb4/0xdc
>> [   12.281564] ---[ end trace 0000000000000000 ]---
>> [   12.281735] possible reason: unannotated irqs-on.
>> [   12.281815] irq event stamp: 2900
>> [   12.281875] hardirqs last  enabled at (2899): [<ffffffff80a9761c>] _raw_spin_unlock_irqrestore+0x54/0x62
>> [   12.282004] hardirqs last disabled at (2900): [<ffffffff8000812c>] __trace_hardirqs_off+0xc/0x14
>> [   12.282125] softirqs last  enabled at (1860): [<ffffffff80a98092>] __do_softirq+0x3e2/0x51c
>> [   12.282242] softirqs last disabled at (1843): [<ffffffff80016fc8>] __irq_exit_rcu+0xb4/0xdc
>>
>> Reverting this patch fixes the problem. Bisect log attached.
>
> Sorry for having missed this report, but I don't see this now and don't remember having seen it before.  Are you still seeing the issue?
>

Yes and no. The backtrace is only seen if CONFIG_CGROUP_FREEZER is enabled.
I disabled it in my boot tests, so I don't see it anymore.

Guenter

\
 
 \ /
  Last update: 2022-09-20 23:57    [W:0.045 / U:0.288 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site