[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [Pan-users] [git 9c148a3] Reproducible segfault from new crc32 code
From: |
Duncan |
Subject: |
Re: [Pan-users] [git 9c148a3] Reproducible segfault from new crc32 code |
Date: |
Mon, 4 Feb 2013 06:23:47 +0000 (UTC) |
User-agent: |
Pan/0.140 (Chocolate Salty Balls; GIT 5133e23 /usr/src/portage/src/egit-src/pan2) |
Duncan posted on Mon, 04 Feb 2013 02:22:35 +0000 as excerpted:
> walt posted on Sun, 03 Feb 2013 11:58:59 -0800 as excerpted:
>
>> Hi Heinrich. I get a segfault (below) before pan even shows on my
>> screen.
>> I think I may have shut pan down with a jpeg displayed in the body
>> pane,
>> and now pan crashes while trying to reload that image during startup:
>>
>> Program received signal SIGSEGV
>>
>> Has anyone else tested the new crc32 code yet? I'll try starting pan
>> with a clean ~/.pan2 to see if I can make it crash again from a fresh
>> start...
>
> Something is indeed terribly wrong. I'm seeing a similar segfault on
> simple face: headers such as the ones sci-fi uses right here on this
> list, on messages I read before the update just fine.
Confirmed the problem as commit 9c148a3. Rolling back a single commit,
to 5133e2306, works great. Sci-fi's posts no longer trigger the
segfault. So it's definitely commit 9c148a3, introducing the new slices
algo.
Maybe it's a bitness/endianness thing? The old code specifically tested
for bitness (and endianness) and setup its types accordingly. I don't
see any of that in the new code, altho it's possible includes bring it in
or something.
You're on amd64 (as I am), correct? Will it work fine on 32-bit x86?
--
Duncan - List replies preferred. No HTML msgs.
"Every nonfree program has a lord, a master --
and if you use the program, he is your master." Richard Stallman