kern/59694: Kernel panic in X after a few minutes

[bsiegert%netbsd.org@localhost]

>Number:         59694
>Category:       kern
>Synopsis:       Kernel panic in X after a few minutes
>Confidential:   no
>Severity:       critical
>Priority:       medium
>Responsible:    kern-bug-people
>State:          open
>Class:          sw-bug
>Submitter-Id:   net
>Arrival-Date:   Mon Oct 06 09:30:00 +0000 2025
>Originator:     Benny Siegert
>Release:        NetBSD 11.0_BETA
>Organization:
The NetBSD Foundation
>Environment:
System: NetBSD takai.bentsukun.ch 11.0_BETA NetBSD 11.0_BETA (GENERIC) #0: Wed Sep 24 12:39:13 UTC 2025 mkrepro%mkrepro.NetBSD.org@localhost:/usr/src/sys/arch/amd64/compile/GENERIC amd64
Architecture: x86_64
Machine: amd64
>Description:

I have a PC with an unsupported NVidia card and a modern Intel CPU that
exposes 28 cores. There is a 4K monitor connected to it. I run X without
graphics acceleration. It's slow but useable (barely).

[     1.050560] genfb0 at pci1 dev 0 function 0: NVIDIA product 2786 (rev. 0xa1)
[     1.050560] genfb0: framebuffer at 0x4000000000, size 3840x2160, depth 32, stride 16384
[     1.050560] genfb0: shadow framebuffer enabled, size 34560 KB
[     1.050560] wsdisplay0 at genfb0 kbdmux 1: console (default, vt100 emulation)
[     1.050560] wsmux1: connecting to wsdisplay0
[     1.050560] drm at genfb0 not configured

Since upgrading to 11_BETA, doing graphics-intensive operations like,
you know, scrolling text in xterm when compiling software leads to
"hiccups", where the screen stays half-updated for a second or two. This
gets worse progressively, until eventually the system crashes:

[   179.004874] cpu24: found cpu25 heart stopped beating after 16 seconds
[   179.004874] panic: cpu25[305 softclk/25]: heart stopped beating
[   179.004874] cpu25: Begin traceback...
[   179.004874] vpanic() at netbsd:vpanic+0x171
[   179.004874] panic() at netbsd:panic+0x3c
[   179.004874] defibrillator() at netbsd:defibrillator+0x6b
[   179.004874] ipi_msg_cpu_handler() at netbsd:ipi_msg_cpu_handler+0x68
[   179.004874] ipi_cpu_handler() at netbsd:ipi_cpu_handler+0xa2
[   179.004874] x86_ipi_handler() at netbsd:x86_ipi_handler+0x71
[   179.004874] Xresume_lapic_ipi() at netbsd:Xresume_lapic_ipi+0x18
[   179.004874] --- interrupt ---
[   179.004874] mutex_spin_enter() at netbsd:mutex_spin_enter+0x38
[   179.004874] itimer_arm_real() at netbsd:itimer_arm_real+0x33
[   179.004874] itimer_callout() at netbsd:itimer_callout+0xad
[   179.004874] callout_softclock() at netbsd:callout_softclock+0xd9
[   179.004874] softint_dispatch() at netbsd:softint_dispatch+0x112
[   179.004874] cpu25: End traceback...

[   179.004874] dumping to dev 168,2 (offset=201709087, size=8340796):
[   179.004874] dump 

I put the dump and the kernel under
http://www.netbsd.org/~bsiegert/crash/ in case it could be useful.

NOTE: Previously, I ran this system on NetBSD 10.99.12, where it ran
perfectly stable.

>How-To-Repeat:

Start X.
Run "pkgrrxx -u" in a terminal.
Observe a crash after a few minutes.

>Fix:

?