freebsd stable - Dec 2003 - Panic during kernel rebuild; file system left inconsistent

I've got a 4.9 machine that's acting as a router and is lightly loaded
otherwise. (There are a significant number of interrupts from network
interface cards at all times, but disk usage is normally quite light.) When I
typed

make depend; make; make install

to rebuild the kernel, the system had a panic during the "make" phase.
Upon
rebooting, I got a message saying that there was a "unexpected softupdates
inconsistency" in /usr. After running fsck (which reported a large number
of
problems), I found that all of the files in the lost+found directory were
(maybe not surprisingly) dependency files. I repeated the build, doing

nice make clean; sync; sleep 10; nice make depend; sync; ...

and this time the kernel built with no problems. I suspect (though I don't
have a backtrace) that heavily loading the file system and/or the ata driver
in the presence of many other interrupts can cause a panic which softupdates
is not prepared to handle. Have others seen this?

--Brett Glass

On Wed, Dec 03, 2003 at 09:13:46PM -0700, Brett Glass
wrote:
> I've got a 4.9 machine that's acting as a router and is lightly
loaded
> otherwise. (There are a significant number of interrupts from network
> interface cards at all times, but disk usage is normally quite light.) When
I
> typed
> 
> make depend; make; make install
> 
> to rebuild the kernel, the system had a panic during the "make"
phase. Upon
> rebooting, I got a message saying that there was a "unexpected
softupdates
> inconsistency" in /usr. After running fsck (which reported a large
number of
> problems), I found that all of the files in the lost+found directory were
> (maybe not surprisingly) dependency files. I repeated the build, doing
> 
> nice make clean; sync; sleep 10; nice make depend; sync; ...
> 
> and this time the kernel built with no problems. I suspect (though I
don't
> have a backtrace) that heavily loading the file system and/or the ata
driver
> in the presence of many other interrupts can cause a panic which
softupdates
> is not prepared to handle. Have others seen this?
I've seen the same message about softupdates, but it appeared after
unexpected
power problems, when computer turned off without syncs and umounts.

-ip