Re: panic: vm_fault failed: %lx error 1 (from arm64::data_abort)

[Just adding the missing Email address so it goes to the list.]

On Jan 5, 2023, at 19:56, Mark Millard <[email protected]> wrote:

Bjoern A. Zeeb <bzeeb-lists_at_lists.zabbadoz.net> wrote on
Date: Thu, 05 Jan 2023 22:57:34 UTC :

> On Thu, 5 Jan 2023, Bjoern A. Zeeb wrote:
> 
>> On Thu, 5 Jan 2023, Bjoern A. Zeeb wrote:
>> 
>> Hi,
>> 
>>> on an unattended console after updating the machine (previious builds were 
>>> Dec 23) did not come back up.
>>> I have a few last lines.
>>> 
>>> esr: 96000004
>>> panic: vm_fault failed: <addr> error 1
>>> cpuid = 0
>>> time = 1
>>> KDB: stack backtrace:
>>> ..
>>> data_abort()
>>> ..
>>> --- exception, esr 0x96000004
>>> thread_init()
>>> keg_alloc_slap()
>>> zone_import()
>>> cache_alloc()
>>> cace_alloc_retry()
>>> thread_alloc()
>>> fork()
>>> kproc_create()
>>> audit_worker_init()
>>> mi_startup()
>>> virtdone()
>> 
>> Follow-up, got a serial console hooked up and a kernel as of an hour ago
>> or so:
> 
> And as another data point: 6fd6a0e342fbfb8513ae56105cf0f85f55c6276e
> (Dec 23) does boot still just fine; did a rebuild with all the same
> local changes, same kernel modules loaded, ... same loader installed
> (not changed with the dowgrade), same firmware, ...
> 
> I'll try to bisect the next days unless someone can spot any other
> commit I may have missed which could cause this.

As a contrast, I've dd'd to microsd card media and booted:

FreeBSD-14.0-CURRENT-arm64-aarch64-RPI-20230101-231d75568f16-259905.img.xz <http://ftp3.freebsd.org/pub/FreeBSD/snapshots/ISO-IMAGES/14.0/FreeBSD-14.0-CURRENT-arm64-aarch64-RPI-20230101-231d75568f16-259905.img.xz>

231d75568f16 is from:

   • Sat, 31 Dec 2022
      . . .
       • git: 231d75568f16 - main - Move INVLPG to pmap_quick_enter_page() from pmap_quick_remove_page(). Konstantin Belousov

So: the last listed for 2022-Dec-31.

I've booted on a couple of RPi4B's ("C0T" and "B0T" 8 GiByte
ones as I remember). No boot crashes or such. You might want
to test if such crashes in your context. If it does not, then
something more specific to your environment is involved.

I sometimes do rough/partial kernel "bisect" via materials from:

https://artifact.ci.freebsd.org/snapshot/main/?C=M&O=D

without having to build. For one, if I get a replication then my
personal builds are not the source of whatever problem I'm
looking into at the time. Otherwise . . .


===
Mark Millard
marklmi at yahoo.com