Re: system/1823: fsck segfaults in single user mode

[Bitt Faulk <wfaulk@beaglebros.com>]

The following reply was made to PR system/1823; it has been noted by GNATS.

From: Bitt Faulk <wfaulk@beaglebros.com>
To: Theo de Raadt <deraadt@cvs.openbsd.org>
Cc: gnats@openbsd.org
Subject: Re: system/1823: fsck segfaults in single user mode 
Date: Tue, 15 May 2001 00:30:56 -0400 (Eastern Daylight Time)

 Turns out the problem has to do with the amount of memory available on the
 system.  fsck_ffs allocates large amounts of memory.  Unfortunately, the
 debugger slows that process down so much that I was unable to get through
 it.  The first of the enormous callocs happens at setup.c:348 (this is on
 2.8 RELEASE).  I never got past that one.  However, when I added swap
 space, the fsck_ffs proceeded fine.  So it seems that there's some bad
 error checking there somewhere or the kernel just spits up or whatever.
 
 BTW, that machine has 16MB physical memory.  That filesystem is 58GB.
 
 Regardless, it's definitely a lack-of-memory issue.  It'd be nice if it
 would at least error out in a more informative manner.  Even better would
 be to *try* to activate a swap device.  I can see problems there, too, but
 it *might* not hurt to try.  As it is, I'm unable to get my wife to simply
 press the reset button when the machine wedges (which is fairly often due
 to that 16MB of memory :-), because it can't fsck that filesystem.  I
 could leave it out of the fstab and mount it after the fact, but that
 seems inappropriate.
 
 -Bitt