On Wed, 12 Jun 2013 18:56:05 +0200 Cedric Blancher wrote:
>On 12 June 2013 15:05, Simon Toedt <[email protected]> wrote:
>> On Wed, Jun 12, 2013 at 1:46 PM, Glenn Fowler <[email protected]> wrote:
>>> the ast-ksh 2013-06-11 alpha source has been posted to
>>>         http://www.research.att.com/sw/download/alpha/
>>> its still a work in progress
>>> dgk and I are corrdinating the new vmalloc implementation vs signals
>>> in addition to the work dgk is doing in ksh vs signals
>>
>> On Debian Linux signals.sh keeps hanging. An example:
 ...
>> The parent process hangs in the memory allocator:
>> #0  0x00007eff3039c570 in __nanosleep_nocancel () at
>> ../sysdeps/unix/syscall-template.S:81
>> #1  0x00007eff306fde67 in tvsleep (tv=0x0, rv=0x0) at
>> /home/simon/kshbuild/src/lib/libast/tm/tvsleep.c:63
>> #2  0x00007eff3076746f in asorelax (nsec=262144) at
>> /home/simon/kshbuild/src/lib/libast/aso/asorelax.c:46
>> #3  0x00007eff3076741e in asolock (lock=0x7eff301e303c, key=20130501,
>> type=4) at /home/simon/kshbuild/src/lib/libast/aso/asolock.c:40
>> #4  0x00007eff3076ebf9 in dballoc (vm=0x7eff31b22dc0, size=144,
>> local=0) at /home/simon/kshbuild/src/lib/libast/vmalloc/vmdebug.c:333
>> #5  0x00007eff30768793 in _ast_malloc (size=144) at
 ...
>Solaris 11.1 too hangs:
> fffffd7ffbcaa44a nanosleep (fffffd7fffdfd058, fffffd7fffdfd048)
> fffffd7ffba6a050 tvsleep () + 40
> fffffd7ffbb07cac asorelax () + 3c
> fffffd7ffbb07c2b asolock () + bb
> fffffd7ffbb11f16 dballoc () + 96
> fffffd7ffbb0968c _ast_malloc () + fc
 ...

I've been working on the main allocator
the debug allocator dballoc() is next
they both have windows of code that can hang when a signal handler
calls malloc or free, the dballoc windows are definitely bigger at the moment

>The funny thing is the trap nesting, i.e. SIGCONT immediately followed
>by a SIGRTMIN+7.
>
>WTF does the SIGCONT come from? Glenn?

SIGCONT may be an artifact of debugging/tracing?

_______________________________________________
ast-developers mailing list
[email protected]
http://lists.research.att.com/mailman/listinfo/ast-developers

Reply via email to