From: Walter Alejandro Iglesias <wai@roquesor.com>
Subject: Re: Blender crashes when rendering animation to ffmpeg video
To: Stuart Henderson <stu@spacehopper.org>
Cc: ports@openbsd.org
Date: Wed, 12 Nov 2025 11:27:35 +0100

On Wed, Nov 12, 2025 at 07:45:41AM +0000, Stuart Henderson wrote:
> ktrace produces a ktrace.out file in $PWD, kdump turns that into a textual 
> list of system calls
> 
> if you get an "illegal instruction" trap, there should be something in 
> there showing whether it's due to BTCFI or not
> 
> ktrace/kdump don't use debug symbols
> 

I used ktrace to find memory leaks in programs I wrote and in base
system applications, but I assumed it was necessary to compile with
debug symbols (I'm like the city dweller who visits the countryside and
looks for the gearshift on the horse.)  Anyways, kdump throws no output
in this case.

By the way, I don't know if this is useful.  While reproducing this
blender crash I always get this message in xconsole repeatedly:

  splassert: pool_do_put: want 0 have 1

Searching about it in the Internet I found this thread:

  https://misc.openbsd.narkive.com/FTP41nnt/splassert-pool-put-want-0-have-4

So, I tried the setting you point out in that thread, the 'sysctl
kern.splassert=2', then I reproduced the crash again and got this (also
repeatedly):

Starting stack trace...
pool_do_put(ffffffff82a6c580,fffffd862d2cc2d8) at pool_do_put+0x43
pool_put(ffffffff82a6c580,fffffd862d2cc2d8) at pool_put+0x79
i915_fence_release(ffff8000023e5b40) at i915_fence_release+0x4c
dma_fence_chain_release(ffff800002336a00) at dma_fence_chain_release+0x104
dma_fence_chain_walk(ffff8000024d2900) at dma_fence_chain_walk+0x1e5
dma_fence_chain_enable_signaling(ffff8000024d2900) at dma_fence_chain_enable_signaling+0x97
dma_fence_chain_timo(ffff8000024d2900) at dma_fence_chain_timo+0x21
softclock_process_tick_timeout(ffff8000024d2970,1) at softclock_process_tick_timeout+0x113
softclock(0) at softclock+0x10e
softintr_dispatch(0) at softintr_dispatch+0x101
dosoftint(0) at dosoftint+0x3f
Xsoftclock() at Xsoftclock+0x27
acpicpu_idle() at acpicpu_idle+0x228
sched_idle(ffffffff82a58ff0) at sched_idle+0x24b
end trace frame: 0x0, count: 243
End of stack trace.
splassert: pool_do_put: want 0 have 1


-- 
Walter