[Xenomai-help] sem_timedwait return value

[Xenomai-help] sem_timedwait return value

[Jeff Webb]

[-- Attachment #1: Type: text/plain, Size: 663 bytes --]

In my userspace POSIX program, I get a '-1' return value from sem_timedwait, even when the function succeeds.  When the function succeeds, errno is '-1' as well.  I am using this workaround at the moment:

      result = sem_timedwait(&cas_sem, &wake_time);
      if (result < 0 && errno > 0) /* errno > 0 to work around xenomai bug */
        {
          print_msg(WARN_MSG, "sem_wait returned with error code %d\n",
                    errno);
          continue;
        }

I am using xenomai 2.2.4, but I think the problem exists in other branches as well.  I believe the attached patch against the trunk should fix the problem.  Do you agree?

Thanks,

Jeff

[-- Attachment #2: sem_timedwait.patch --]
[-- Type: text/x-patch, Size: 516 bytes --]

Index: ksrc/skins/posix/syscall.c
===================================================================
--- ksrc/skins/posix/syscall.c	(revision 1911)
+++ ksrc/skins/posix/syscall.c	(working copy)
@@ -492,7 +492,7 @@
 			    &ts,
 			    (void __user *)__xn_reg_arg2(regs), sizeof(ts));
 
-	return sem_timedwait(&sm.native_sem, &ts) == 0 ? : -thread_get_errno();
+	return sem_timedwait(&sm.native_sem, &ts) == 0 ? 0 : -thread_get_errno();
 }
 
 static int __sem_trywait(struct task_struct *curr, struct pt_regs *regs)

Re: [Xenomai-help] sem_timedwait return value

[Philippe Gerum]

On Fri, 2006-12-01 at 14:12 -0600, Jeff Webb wrote:
> In my userspace POSIX program, I get a '-1' return value from sem_timedwait, even when the function succeeds.  When the function succeeds, errno is '-1' as well.  I am using this workaround at the moment:
> 
>       result = sem_timedwait(&cas_sem, &wake_time);
>       if (result < 0 && errno > 0) /* errno > 0 to work around xenomai bug */
>         {
>           print_msg(WARN_MSG, "sem_wait returned with error code %d\n",
>                     errno);
>           continue;
>         }
> 
> I am using xenomai 2.2.4, but I think the problem exists in other branches as well.  I believe the attached patch against the trunk should fix the problem.  Do you agree?
> 

Oops, yes. This bug belongs to me. Applied, thanks.

> Thanks,
> 
> Jeff
> _______________________________________________
> Xenomai-help mailing list
> Xenomai-help@domain.hid
> https://mail.gna.org/listinfo/xenomai-help
-- 
Philippe.