[17599] in Athena Bugs
Re: sgi 8.3.25: IRIX panic
daemon@ATHENA.MIT.EDU (Robert A Basch)
Wed Mar 1 23:00:38 2000
Message-Id: <200003020400.XAA01540@aupair.mit.edu>
To: Robert A Basch <rbasch@MIT.EDU>
cc: John Hawkinson <jhawk@MIT.EDU>, bugs@MIT.EDU, sipb-staff@MIT.EDU
In-Reply-To: Your message of "Sun, 16 Jan 2000 15:58:35 EST."
<200001162058.PAA03215@aupair.mit.edu>
Date: Wed, 01 Mar 2000 23:00:09 -0500
From: Robert A Basch <rbasch@MIT.EDU>
There's finally some progress to report on the cause of the stack
overflow panic...
According to SGI, the problem occurs when the kernel exceeds its
allotted stack space in dumping a core (of netscape, in this case).
More stack is required when dumping core from an executable in AFS to
local disk, as calls into both xfs and afs code are stacked.
In fact, I have found that this can be easily reproduced, by starting
netscape while cd'ed to a directory on local disk, and doing kill -ABRT
to the process. (For some reason I haven't tracked down yet, this can
occur even when the coredump size limit is set to 0 -- apparently the
limit gets reset somehow).
SGI has now admitted the stack space problem, and an RFE is being
opened to deal with it.
Bob
