mq_timed... sem_timedwait, pthread_mutex_timed_lock

[Dmitriy Korovkin <dkorovkin at rambler dot ru>]

Jonathan, Nick and other nice guys,
As I promised, here is the patch that adds sem_timedwait, 
pthread_mutex_timed_lock functions and also Cyg_Mutex::lock() with 
timeout. It also slightly changes behavoir of mq_timedsend and 
mq_timedreceive so that they try to figure out why there was a timeout.
Patch was created over the existing CVS that doesn't include mq_timedsend 
and mq_timedreceive, so they are included again (I couldn't remove them 
because I've changed them). Sorry for inconvinience and thank you for your 
patience while you review my changes. And thank you very much for nice OS.

Regards,
-- 
Dmitriy

---------- Forwarded message ----------
Date: Wed, 15 Jan 2003 09:16:49 +0300 (MSK)
From: Dmitriy Korovkin <dkorovkin@rambler.ru>
To: Jonathan Larmour <jifl@ecoscentric.com>
Cc: Nick Garnett <nickg@ecoscentric.com>,  <ecos-patches@sources.redhat.com>
Subject: Re: mq_timedsend & mq_timedreceive

OK, I'll try to add the required stuff on the nearest weekend.
Regards,
-- 
Dmitriy

On Tue, 14 Jan 2003, Jonathan Larmour wrote:

> Date: Tue, 14 Jan 2003 07:10:12 +0000
> From: Jonathan Larmour <jifl@ecoscentric.com>
> To: Dmitriy Korovkin <dkorovkin@rambler.ru>
> Cc: Nick Garnett <nickg@ecoscentric.com>, ecos-patches@sources.redhat.com
> Subject: Re: mq_timedsend & mq_timedreceive
> 
> Dmitriy Korovkin wrote:
> > Nick, Jonathan
> > Does it mean I may do them?!  
> > Yesturday I found that I'd be glad to have sem_timedwait and 
> > pthread_mutextimedlock. I don't see a problem with implementing it but 
> > standards... Huh, actually asking around brought me that QNX guys included 
> > this into their manual pages and LynxOS guys seems to do this too. Why we 
> > shouldn't do this? At least, some things that are almost reqired by that 
> > ^&*@^$&*@ ISaGRAF PRO target.
> 
> Feel free. It's not a priority for us though so the ball is in your court 
> if you want it.
> 
> But do take care with your treatment of error conditions, for example I 
> had to add something to take care of invalid timeval nanosecond values 
> causing an EINVAL error. You should refer closely to the standard for 
> this. See <http://www.opengroup.org/onlinepubs/007904975/mindex.html>
> 
> Jifl
>

Attachment: mq_sem_mutex_timed.patch.bz2
Description: BZip2 compressed data