[m-rev.] diff: Fix bug #233

[Julien Fischer]

On Fri, 25 Nov 2011, Paul Bone wrote:

> On Fri, Nov 25, 2011 at 12:49:10AM +1100, Julien Fischer wrote:
>>
>>
>> On Thu, 24 Nov 2011, Paul Bone wrote:
>>
>>> Fix an error where the runtime system aborted with "sem_post: Invalid
>>> argument".
>>>
>>> runtime/mercury_context.c:
>>>   The error message as important, since sem_timedwait actually raised the
>>>   error.  When this occurred the error was incorrectly reported as being
>>>   raised by the sem_post syscall.
>>>
>>>   Use >= rather than > when testing if the nanoseconds field of a timeval
>>>   structure for sem_timedwait should overflow into the seconds field.
>>
>> Does this also need to go on to the release branch?
>>
>
> That's up to you (as release manager),

I know that - what I wanted to know was does the problem that this patch
fixes also exist on the release branch?

> if you think it's a serious enough bug
> or the parallel low-level C grades are important enough.

The parallel low-level C grades should work in the 11.07 release,
granted that the auto parallisation stuff won't but users can still
request parallism manually.

> I've made this
> seperate to my other patches in case you want to apply it to the release
> branch.
>
> It may not apply cleanly, if it doesn't (becase sem_timedwait is not used) then
> the bug doesn't exist on the release branch.

Ok, thanks - I'll take a look at it.

Julien.
--------------------------------------------------------------------------
mercury-reviews mailing list
Post messages to:       mercury-reviews at csse.unimelb.edu.au
Administrative Queries: owner-mercury-reviews at csse.unimelb.edu.au
Subscriptions:          mercury-reviews-request at csse.unimelb.edu.au
--------------------------------------------------------------------------