RE: Repeatable crash on Solaris, with recipe
-----Original Message-----
From: Stephen J. Turnbull [mailto:stephen＠xemacs.org]
Sent: Monday, May 20, 2002 8:20 PM
To: Horning, Jim
Cc: 'Raymond Toy'; XEmacs Developers; Benjamin P. Wing
Subject: Re: Repeatable crash on Solaris, with recipe
>>>>> "Jim" == Jim Horning <Horning> writes:
Jim> Tested your recipe on 21.4.8 built from the tarballs with Sun
Jim> Workshop 6 update 1 on Solaris 2.7 results in an (apparently)
Sun CC?
Yes. I'm not sure my terminology is entirely correct, it's not my area of
expertise. But in looking at Sun's READMES in the install tree, that is
what I found.
Jim> hung editor. It's not consuming any noticeable quantity of
Jim> CPU cycles and I can interrupt it with C-g and break out of
Jim> whatever it thinks is going on.
Try it in the debugger, put it in that state, and see where it's hung?
For Ray, crashing under the debugger ought to get a plausible trace; I
suspect you're not really in zero_out_command_line_status_vars ().
Maybe this weekend, I think I remember someone sending out a "How-to debug
XEmacs" message on the list not to long ago that I could try and follow.
It's not really a problem for me, I was just curious because I had recently
built 21.4.8 coupled with the similarity between Ray's OS and mine and tried
his recipe. Since my results differed noticeably from his and it was not
correct operation, I offered them up as another data-point.
Jim> Hmmm, for me it seems to hang on the C-x h, actually.
Jim> This seems to start with only 3236 lines of your test pattern
Jim> in the buffer.
262116 characters? Doesn't ring a bell for me. This is very sudden,
right? Normality until that size, then wham, hangs?
Correct, or at least down to plus/minus a lines worth of characters. At
3235 lines, C-w and other operations I tried after a C-x h worked, at 3236
lines it hung as stated.
Jim> I didn't crash at all while narrowing down the buffer size.
By "narrowing down" you mean repeated experiments to discover the
smallest buffer that hangs? Did you restart XEmacs or use the same
instance?
Yes to the first. I repeatedly tried C-x h until I had bounded where the
problem occurred to the nearest line of characters. This was all done on
the same instance of XEmacs, I did not restart each time since C-g broke out
of the "hang" and the editor seemed to work normally afterward.
>>>>> From: Raymond Toy [mailto:toy＠rtp.ericsson.se]
Ray> I found a repeatable crash on Solaris with the following
Ray> recipe. This is 21.5-b6 built from a full tarball using gcc
Ray> 3.0.4 on Solaris 2.7.
FWIW I don't see either of these behaviors with XEmacs 21.4.7 or
21.4.9-work-in-progress (it's an old build, at least 50 hours ;-) and
!= 21.4.8 nor 21.4.9-rc1).
Linux 2.2.20 glibc 2.2.5 gcc 2.95.4
Honestly, I have no idea where to point the finger, since it doesn't
seem to be either Motif or EsounD ;-)
--
Institute of Policy and Planning Sciences
http://turnbull.sk.tsukuba.ac.jp
University of Tsukuba Tennodai 1-1-1
Tsukuba 305-8573 JAPAN
My nostalgia for Icon makes me forget about any of the bad
things. I don't
have much nostalgia for Perl, so its faults I remember.
Scott Gilbert c.l.py