Re: Panic on drm-devel v5.5.19.g20210909

From: Vladimir Kondratyev <wulf_at_FreeBSD.org>
Date: Sun, 12 Sep 2021 18:49:25 UTC
On 12.09.2021 03:47, Pete Wright via x11 wrote:
> 
> 
> On 9/11/21 9:53 AM, Pete Wright via freebsd-x11 wrote:
>>
>>
>> On 9/11/21 8:17 AM, Vladimir Kondratyev wrote:
>>> On 11.09.2021 02:53, Pete Wright via freebsd-x11 wrote:
>>>> hello,
>>>> i got the following panic after upgrading my current laptop to
>>>> drm-devel-v5.5.19.g20210909 (3df6598adf625a18a7c762b2d48b4c3dd7925838):
>>>>
>>>> panic: acquiring blockable sleep lock with spinlock or critical section
>>>> held (sleep mutex) linux_wq_exec @
>>>> /usr/home/pete/git/freebsd/sys/compat/linuxkpi/common/src/linux_work.c:105
>>>>
>>>> cpuid = 5
>>>> time = 1631315668
>>>> KDB: stack backtrace:
>>>> db_trace_self_wrapper() at db_trace_self_wrapper+0x2b/frame
>>>> 0xfffffe011d086060
>>>> vpanic() at vpanic+0x187/frame 0xfffffe011d0860c0
>>>> panic() at panic+0x43/frame 0xfffffe011d086120
>>>> witness_checkorder() at witness_checkorder+0xe78/frame
>>>> 0xfffffe011d0862f0
>>>> __mtx_lock_flags() at __mtx_lock_flags+0x94/frame 0xfffffe011d086340
>>>> linux_queue_work_on() at linux_queue_work_on+0x9a/frame
>>>> 0xfffffe011d086380
>>>> dma_fence_signal() at dma_fence_signal+0xc0/frame 0xfffffe011d0863d0
>>>> dma_resv_add_shared_fence() at dma_resv_add_shared_fence+0x96/frame
>>>> 0xfffffe011d086420
>>>> i915_vma_move_to_active() at i915_vma_move_to_active+0x7e/frame
>>>> 0xfffffe011d086460
>>>> i915_gem_do_execbuffer() at i915_gem_do_execbuffer+0x13d7/frame
>>>> 0xfffffe011d086670
>>>> i915_gem_execbuffer2_ioctl() at i915_gem_execbuffer2_ioctl+0x195/frame
>>>> 0xfffffe011d0866e0
>>>> drm_ioctl_kernel() at drm_ioctl_kernel+0x72/frame 0xfffffe011d086730
>>>> drm_ioctl() at drm_ioctl+0x2bf/frame 0xfffffe011d086820
>>>> linux_file_ioctl() at linux_file_ioctl+0x297/frame 0xfffffe011d086880
>>>> kern_ioctl() at kern_ioctl+0x202/frame 0xfffffe011d0868f0
>>>> sys_ioctl() at sys_ioctl+0x124/frame 0xfffffe011d0869c0
>>>> amd64_syscall() at amd64_syscall+0x12e/frame 0xfffffe011d086af0
>>>> fast_syscall_common() at fast_syscall_common+0xf8/frame
>>>> 0xfffffe011d086af0
>>>> --- syscall (54, FreeBSD ELF64, sys_ioctl), rip = 0x80f1ae3ca, rsp =
>>>> 0x7fffffffc8d8, rbp = 0x7fffffffc970 ---
>>>> KDB: enter: panic
>>>>
>>>>
>>>> I have a core and and additional info if that'll help debug this issue.
>>>> it happened three times while doing normal web+shell work. i've
>>>> reverted to the commit previous this one and it looks like the
>>>> system no
>>>> longer panic's.
>>>>
>>>> -pete
>>>>
>>> Try to revert this commit:
>>> https://github.com/freebsd/drm-kmod/commit/787ffa8c48ecd02310ce2ec27c4b84d1898201e5
>>>
>>>
>> Thanks for the heads up on that.  Just reverted on my system and it
>> seems better on my end now.  Let me know if there is any additional
>> debugging i can do to fix the original sleep lock issue.
>>
> 
> hrm - spoke too soon, just had the same panic with that commit
> reverted.  made it most of the day with several suspend/resume's as well
> as web browsing and watching videos.
> 
> -pete
> 

Could you make backtrace with kgdb? One from your mail is useless as kdb
skipped all functions declared as static.

just type

# kgdb /boot/kernel/kernel /var/crash/vmcore.last
(kgdb) bt

-- 
WBR
Vladimir Kondratyev