Fwd: Re: Alpine help, please

[Rares Aioanei <bsdlisten%gmail.com@localhost>]

Hello, everyone, just wanted to let everyone know the outcome of my 
Alpine problem with
a big Gmail IMAP mailbox (segfaults after a while). Maybe this will help 
someone else. Thanks again to Jean-Yves for the help.

On Wed, 09 Nov 2011 11:30:55 +0200, Rares Aioanei wrote:
>  On 11/09/2011 01:16 AM, Jean-Yves Migeon wrote:
> >  On 08.11.2011 14:49, Rares Aioanei wrote:
> > >  On 11/08/2011 02:56 PM, Jean-Yves Migeon wrote:
> > > >  On Tue, 08 Nov 2011 14:13:53 +0200, Rares Aioanei wrote:
> > > > >  On 11/08/2011 03:03 AM, Jean-Yves Migeon wrote:
> > > > > >  On 07.11.2011 00:11, Rares Aioanei wrote:
> > > > > > >  Hi list,
> > > > > > >
> > > > > > >  I compiled Alpine (pkgsrc-2011Q3) for my 5-STABLE system and it
> > > > > > >  works
> > > > > > >  OK, but when I tried to open my inbox on a rather large IMAP
> > > > > > >  Gmail
> > > > > > >  account, it segfaults. On Free and OpenBSD Alpine works with
> > > > > > >  the same
> > > > > > >  account with no problems. I tried compiling using pthreads but
> > > > > > >  the
> > > > > > >  outcome is the same: it stays a while on "Sorting messages"
> > > > > > >  then
> > > > > > >  segfaults leaving a .core file in ~ . Running GDB shows
> > > > > > >  problems
> > > > > > >  coming
> > > > > > >  from nanosleep(), telling me that it's indeed a threads
> > > > > > >  problem.
> > > > > > >  Anyone
> > > > > > >  else having the same issue or should I file a bug?
> > > > > > >
> > > > > > >  Thank you in advance.
> > > > > >
> > > > > >  Hmmm I am not quite sure that nanosleep is the culprit here. Any
> > > > > >  idea how much memory does it try to eat?
> > > > >  Right before the crash, I ran top and it showed ~450 MB.
> > > >
> > > >  Does it look the same on other OS where it does not crash? You may
> > > >  hit
> > > >  a ulimit that prevents further allocations. The output of ulimit
> > > >  -a
> > > >  for your session would be interesting too.
> > >  ulimit -a:
> > >
> > >  data 262144
> > >  stack 2048
> > >  lockedmem 672500
> > >  nofiles(descriptors) 128
> > >  processes 160
> > >  resident-set 2017500
> > >  , the rest being set to unlimited. The machine has 2GB RAM, if that
> > >  matters, and lots of swap/disk space. The OpenBSD machine is a
> > >  ThinkPad
> > >  T22 with 256 MB RAM, and on it alpine eats less, around 170 MB, as
> > >  shown
> > >  by top.
> >
> >  Nothing unusual there, except maybe the stack size (you can raise it
> >  with ulimit -s though).
> >
> >  Is it possible to get a backtrace from gdb please, without thread
> >  support compile in?
>  Just as I thought! Setting the stack size to 4096 did the trick.
>  Thanks, Jean-Yves.

You're welcome

Could you please send a message on the ML? You probably are not in this
situation. NetBSD has a rather small stack size by default, and some
applications choke on it (especially for PIE with ASLR).

--
Jean-Yves Migeon
jeanyves.migeon%free.fr@localhost