[En-Nut-Discussion] NutSemTryWait Bug

[Matthias Ringwald]

Hi "thezachariasgroup"

On 16.06.2008, at 16:23, ennut at thezachariasgroup.com wrote:

>
>  I think I found a problem in NutSemTryWait in NUT\OS\semaphore.c
>
> Original Code:
>
>     int NutSemTryWait(SEM * sem) {
>         if (sem->value < 0)
>             return -1;
>         else
>             NutSemWait(sem);
>         return 0;
>     }
>
> Suggested code:
>
>
>     int NutSemTryWait(SEM * sem) {
>         if (sem->value <= 0)
>             return -1;
>         else
>             NutSemWait(sem);
>         return 0;
>     }
>
>   If the sem->value was initialized to 1, and a routine has grabbed
> it, sem->value would be zero.
> NutSemTryWait should return a -1 under these conditions since the sem
> is not available.
> However sem->value is not less than zero and the code calls
> NutSemWait and waits for the sem to become free.
> This is not the intent of the routine as I understand it.
>
>   Thoughts?

Yeah. You're right. Thanks for spotting this, I'll fix it in CVS HEAD  
now.

Regards
Matthias