Re: [Evolution] Re: [Evolution-hackers] SIGSEGV on setting label - gdb output and analysis

[Mika Liljeberg <mika liljeberg welho com>]

On Mon, 2003-01-06 at 02:23, Not Zed wrote:
> Memory bugs are sometimes hard to find ... :-/
>
> e_thread_destroy doesn't need to lock because if anything still has a
> handle to the ethread at the time it is destroyed, its already a bug.

I figured as much, but the thread lock *is* acquired right after
emptying the message queue and I wasn't certain the code did not rely on
that to flush out the remaining clients as well.

I had another crash in a completely different place (syncing the index
file of my Sent folder to disk). One pointer on the block list had a
next pointer with the value 0x200. And here we had a pointer with the
value 0x100. Looks very suspicious to me. I don't suppose these numbers
mean anything to you? Could be a flags field, of some sort, I guess.

The annoying thing is that after compiling evolution-mail and some of
the relevant libs with -g, the mail client is now much more stable. When
compiled with -O2 it used to crash roughly once a day.

        MikaL