Re: CURRENT (after commit b1f72dd2a9a5e69a310b9e76b55f16cff433b8da): silently crashing when make -jXX buildworld

On Sat, 21 Mar 2026 04:41:42 -0700
David Wolfskill <david@catwhisker.org> wrote:

> On Sat, Mar 21, 2026 at 11:58:39AM +0100, Herbert J. Skuhra wrote:
> > On Sat, 21 Mar 2026, at 09:30, A FreeBSD User wrote:
> > > Hello,
> > >
> > > after upgrading sources and recompiling averything from scratch after 
> > > make cleanworld, host is
> > > silently rebooting (not trace of a crash or anything suitable to 
> > > present).
> > ...
> > I guess this is caused by e90950627327de9226b052851b36f341bc72b746.
> > /var/log/messages shows:
> > Mar 21 11:38:08 srv03 kernel: Fatal trap 9: general protection fault while in kernel mode
> > Mar 21 11:38:08 srv03 kernel: 
> > Mar 21 11:38:08 srv03 syslogd: last message repeated 4 times
> > Mar 21 11:38:08 srv03 kernel: cpuid = 5; apic id = 05
> > Mar 21 11:38:08 srv03 kernel: 
> > Mar 21 11:38:08 srv03 kernel: Fatal trap 9: general protection fault while in kernel mode
> > Mar 21 11:38:08 srv03 syslogd: last message repeated 1 times
> > ....
> 
> What I find (after the update to main-n284608-b1f72dd2a9a5, rebooting, &
> crashing) in messages appears to be a little different:
> 
> ...
> Mar 21 11:27:02 freebeast kernel: Security policy loaded: MAC/ntpd (mac_ntpd)
> Mar 21 11:27:02 freebeast kernel: GEOM_ELI: Device ada0s4b.eli created.
> Mar 21 11:27:02 freebeast kernel: GEOM_ELI: Encryption: AES-XTS 128
> Mar 21 11:27:02 freebeast kernel: GEOM_ELI:     Crypto: accelerated software
> Mar 21 11:27:35 freebeast kernel: Kernel page fault with the following non-sleepable locks held:
> Mar 21 11:27:35 freebeast kernel: exclusive sleep mutex sigacts (sigacts) r = 0 (0xfffff8209f0a3ca8) locked @ /usr/src/sys/kern/kern_sig.c:311
> Mar 21 11:27:35 freebeast kernel: exclusive sleep mutex process lock (process lock) r = 0 (0xfffffe04de855138) locked @ /usr/src/sys/kern/kern_sig.c:308
> Mar 21 11:27:35 freebeast kernel: stack backtrace:
> Mar 21 11:27:35 freebeast kernel: #0 0xffffffff80c4407c at witness_debugger+0x6c
> Mar 21 11:27:35 freebeast kernel: #1 0xffffffff80c45989 at witness_warn+0x4c9
> Mar 21 11:27:35 freebeast kernel: #2 0xffffffff8112acec at trap_pfault+0x8c
> Mar 21 11:27:35 freebeast kernel: #3 0xffffffff810f9ae8 at calltrap+0x8
> Mar 21 11:31:19 freebeast syslogd: kernel boot file is /boot/kernel.old/kernel
> Mar 21 11:31:19 freebeast kernel: ---<<BOOT>>---
> ....
> 
> The machine did reboot without my intervention (and as may be seen
> above, I chose kernel.old so I could poke around).  Unfortunately,
> there was no crash dump, and (as far as I could tell on the serial
> console), there was no drop to ddb.
> 
> Peace,
> david
> -- 
> David H. Wolfskill                              david@catwhisker.org
> 
> See https://www.catwhisker.org/~david/publickey.gpg for my public key.

Just back to my daily driver, stable/15.

For me, bitten when I attempt to rebuild kmod ports after installworld.
Instanlty hard-reboots without any core nor crash dump when actual
rebuild starts.

(I'm not using PORTS_MODULES variable, but building manually
 as part of upgrades after installworld and following etcupdate.)


Bitten at: commit 8cc1c0f35ec8d5a3edb4ee1ede962a3c1f6baef0

Fixes by:
  Rolling back /usr/obj (separate dataset on ZFS) to
  commit 33345b8834859cf1ad7f1256f3a34b4a7c83e8e8
  with commit 14b9955e57cc28b61e785165b9effcbe620edb46 reverted,

  Checking out commit 67b918e03475e780854a43fe4eb7c8e95f4deb29

  buildworld, buildkernel, ... as usual

Regards.


-- 
Tomoaki AOKI    <junchoon@dec.sakura.ne.jp>