Hang on latest kubuntu linux when built with icc

Hang on latest kubuntu linux when built with icc

[Vin Shelton]

Greetings -

On my new kubuntu box - uname -a reports:

Linux samwise 2.6.15-23-386 #1 PREEMPT Tue May 23 13:49:40 UTC 2006
i686 GNU/Linux

when I build the latest CVS sources with icc, the Intel C compiler,
the shell eventually hangs after running a command but before
returning to the prompt.  It usually takes only 1-5 commands before
this happens.

Here's an strace of zsh -f:

write(10, "\r\n", 2)                    = 2
alarm(0)                                = 0
ioctl(10, SNDCTL_TMR_TIMEBASE or TCGETS, {B38400 opost isig -icanon
-echo ...}) = 0
ioctl(10, SNDCTL_TMR_STOP or TCSETSW, {B38400 opost isig icanon echo ...}) = 0
ioctl(10, SNDCTL_TMR_TIMEBASE or TCGETS, {B38400 opost isig icanon
echo ...}) = 0
time(NULL)                              = 1149988783
rt_sigprocmask(SIG_BLOCK, [CHLD], [], 8) = 0
rt_sigprocmask(SIG_BLOCK, [CHLD], [CHLD], 8) = 0
pipe([3, 4])                            = 0
gettimeofday({1149988783, 119463}, {240, 0}) = 0
clone(child_stack=0,
flags=CLONE_CHILD_CLEARTID|CLONE_CHILD_SETTID|SIGCHLD,
child_tidptr=0xb7b8c708) = 18370
close(4)                                = 0
read(3, "", 1)                          = 0
close(3)                                = 0
rt_sigprocmask(SIG_BLOCK, [CHLD], [CHLD], 8) = 0
rt_sigprocmask(SIG_SETMASK, ~[HUP CHLD RTMIN RT_1], [CHLD], 8) = 0
--- SIGCHLD (Child exited) @ 0 (0) ---
rt_sigprocmask(SIG_BLOCK, ~[RTMIN RT_1], ~[HUP KILL STOP RTMIN RT_1], 8) = 0
rt_sigprocmask(SIG_SETMASK, ~[HUP KILL STOP RTMIN RT_1], ~[KILL STOP
RTMIN RT_1], 8) = 0
wait4(-1, [{WIFEXITED(s) && WEXITSTATUS(s) == 0}], WNOHANG|WSTOPPED,
{ru_utime={0, 999}, ru_stime={0, 999}, ...}) = 18370
gettimeofday({1149988783, 123302}, {240, 0}) = 0
ioctl(10, SNDCTL_TMR_TIMEBASE or TCGETS, {B38400 opost isig icanon
echo ...}) = 0
wait4(-1, 0xbfd3ac30, WNOHANG|WSTOPPED, 0xbfd3abd4) = -1 ECHILD (No
child processes)
sigreturn()                             = ? (mask now ~[HUP KILL CHLD
STOP RTMIN])
pause()                                 = ? ERESTARTNOHAND (To be restarted)
--- SIGHUP (Hangup) @ 0 (0) ---
rt_sigprocmask(SIG_BLOCK, ~[RTMIN RT_1], ~[KILL CHLD STOP RTMIN RT_1], 8) = 0
rt_sigprocmask(SIG_SETMASK, ~[KILL CHLD STOP RTMIN RT_1], ~[KILL STOP
RTMIN RT_1], 8) = 0
exit_group(1)                           = ?


The SIGHUP is where I killed the shell from another window.  The shell
hangs at the pause just before there.  Is there a race condition on
handling the signals?

Let me know what other info I can provide.

  - Vin
-- 
Whoever you are, no matter how lonely,
the world offers itself to your imagination,
calls to you like the wild geese, harsh and exciting--
over and over announcing your place
in the family of things.			Mary Oliver

Re: Hang on latest kubuntu linux when built with icc

[Peter Stephenson]

"Vin Shelton" wrote:
> when I build the latest CVS sources with icc, the Intel C compiler,
> the shell eventually hangs after running a command but before
> returning to the prompt.  It usually takes only 1-5 commands before
> this happens.
> 
> Here's an strace of zsh -f:
>... 
> wait4(-1, 0xbfd3ac30, WNOHANG|WSTOPPED, 0xbfd3abd4) = -1 ECHILD (No
> child processes)
> sigreturn()                             = ? (mask now ~[HUP KILL CHLD
> STOP RTMIN])
> pause()                                 = ? ERESTARTNOHAND (To be restarted)
> --- SIGHUP (Hangup) @ 0 (0) ---
>...
> The SIGHUP is where I killed the shell from another window.

Unfortunately I'm not a signal expert and it really needs someone who
does this sort of thing regularly.

The pause() looks a bit fishy.  There are two explicit occurrences in
the code:  I'll assume it's one of these for now.  What strace
reports might be only distantly related to the source code; still, my
Fedora Core 4 with 2.6.14 says it's calling rt_sigsuspend(), which
sounds much more plausible.

One call is here:

#ifdef BROKEN_POSIX_SIGSUSPEND
    sigprocmask(SIG_SETMASK, &set, &oset);
    pause();
    sigprocmask(SIG_SETMASK, &oset, NULL);
#else /* not BROKEN_POSIX_SIGSUSPEND */
    ret = sigsuspend(&set);
#endif /* BROKEN_POSIX_SIGSUSPEND */

and the other is right down at the tail end where we're virtually
implementing signal suspension from scratch.

I don't think you should be getting either of those... certainly not the
second, but probably not BROKEN_POSIX_SIGSUSPEND either.  Could you
check if that's defined?  If it is, try undefining it; the test may need
tweaking.

-- 
Peter Stephenson <pws@csr.com>                  Software Engineer
CSR PLC, Churchill House, Cambridge Business Park, Cowley Road
Cambridge, CB4 0WZ, UK                          Tel: +44 (0)1223 692070


To access the latest news from CSR copy this link into a web browser:  http://www.csr.com/email_sig.php

Re: Hang on latest kubuntu linux when built with icc

[Vin Shelton]

On 6/12/06, Peter Stephenson <pws@csr.com> wrote:
> "Vin Shelton" wrote:
> > when I build the latest CVS sources with icc, the Intel C compiler,
> > the shell eventually hangs after running a command but before
> > returning to the prompt.  It usually takes only 1-5 commands before
> > this happens.
> >
> > Here's an strace of zsh -f:
> >...
> > wait4(-1, 0xbfd3ac30, WNOHANG|WSTOPPED, 0xbfd3abd4) = -1 ECHILD (No
> > child processes)
> > sigreturn()                             = ? (mask now ~[HUP KILL CHLD
> > STOP RTMIN])
> > pause()                                 = ? ERESTARTNOHAND (To be restarted)
> > --- SIGHUP (Hangup) @ 0 (0) ---
> >...
> > The SIGHUP is where I killed the shell from another window.
>
> Unfortunately I'm not a signal expert and it really needs someone who
> does this sort of thing regularly.
>
> The pause() looks a bit fishy.  There are two explicit occurrences in
> the code:  I'll assume it's one of these for now.  What strace
> reports might be only distantly related to the source code; still, my
> Fedora Core 4 with 2.6.14 says it's calling rt_sigsuspend(), which
> sounds much more plausible.
>
> One call is here:
>
> #ifdef BROKEN_POSIX_SIGSUSPEND
>     sigprocmask(SIG_SETMASK, &set, &oset);
>     pause();
>     sigprocmask(SIG_SETMASK, &oset, NULL);
> #else /* not BROKEN_POSIX_SIGSUSPEND */
>     ret = sigsuspend(&set);
> #endif /* BROKEN_POSIX_SIGSUSPEND */
>
> and the other is right down at the tail end where we're virtually
> implementing signal suspension from scratch.
>
> I don't think you should be getting either of those... certainly not the
> second, but probably not BROKEN_POSIX_SIGSUSPEND either.  Could you
> check if that's defined?  If it is, try undefining it; the test may need
> tweaking.

Peter,

Thanks for your help.  I'm slightly embarrassed to admit that the
problem was that my system was misconfigured - because I had not added
/opt/intel/compiler/9.0/lib to /etc/ld.so.conf, a dynamic lib could
not be found.  I fixed this in the executable by setting
LD_LIBRARY_PATH before running Src/zsh, by I didn't think that this
would affect the configure tests, too.  Once I added
/opt/intel/compiler/9.0/lib to /etc/ld.so.conf and re-ran ldconfig,
everything worked fine. Previously, BROKEN POSIX_SIGSUSPEND was set,
and no it is not. Zsh is now working fine.

Thanks,
  Vin


-- 
Whoever you are, no matter how lonely,
the world offers itself to your imagination,
calls to you like the wild geese, harsh and exciting--
over and over announcing your place
in the family of things.			Mary Oliver