RE: [lwip-devel] [bug #1902] Timeouts and semaphores/mailboxes are tootightly integrated

[Taranowski, Thomas \(SWCOE\)]

I'm all in for the decoupling, by switching to the sys_arch_mbox_fetch routine. 
 

I don't like the system tick hook idea at all (although it would make the code 
simpler).  The current design is very efficient, in terms of context switch 
overhead, with the way it does the timeout handling inline, and is a feature 
that is unique to lwIP.  On my OS, context switches are fairly expensive, so 
handling the timeouts inline result in a significant performance gain.  In 
addition, mucking around with the system tick handler often has an affect on 
other parts of the system, which adds to the complexity of integration.

> -----Original Message-----
> From: address@hidden
> [mailto:address@hidden On
> Behalf Of Simon Goldschmidt
> Sent: Friday, March 09, 2007 4:01 AM
> To: Simon Goldschmidt; Frédéric Bernon; Atte Kojo; Dmitry Potapov; lwip-
> address@hidden
> Subject: [lwip-devel] [bug #1902] Timeouts and semaphores/mailboxes are
> tootightly integrated
> 
> 
> Follow-up Comment #6, bug #1902 (project lwip):
> 
> >sending a message per each tick is too expensive...
> 
> definitively, yes
> 
> > So, perhaps, we should replace sys_mbox_fetch with sys_arch_mbox_fetch
> in
> api_lib.c
> 
> That's good for me. But how do we get more comments? I'd like to hear from
> more users than the 4 we've had so far.
> 
>     _______________________________________________________
> 
> Reply to this item at:
> 
>   <http://savannah.nongnu.org/bugs/?1902>
> 
> _______________________________________________
>   Nachricht geschickt von/durch Savannah
>   http://savannah.nongnu.org/
> 
> 
> 
> _______________________________________________
> lwip-devel mailing list
> address@hidden
> http://lists.nongnu.org/mailman/listinfo/lwip-devel