Re: GSoC ticket #1069 (thread scheduler) application

On 2007-04-12 at 02:19:13 [+0200], André Braga <[email protected]> wrote:
> On 4/11/07, Ingo Weinhold <bonefish-CFLBMwTPW48UNGrzBIF7/[email protected]> wrote:
> > Please note that: (1) There are usually not that many threads waiting for 
> > a
> > single semaphore anyway. (2) The most likely semaphore release operation 
> > is
> > with count 1, i.e. only the frontmost thread in the semaphore's queue 
> > will be
> > woken up. (3) Even in cases were all waiting threads are woken up (e.g.
> > delete_sem()) some info for each one needs to be updated, so the 
> > operation is
> > "O(n)" anyway.
> 
> I might have a hidden ace on my sleeves... Who knows ;)

Oohhh. :-)

> Could you please explain what info must be updated on every thread?

According to src/system/kernel/sem.c (I'd recommend to have a look, since you 
might modify a few things there, anyway) it's thread::state in any case and 
fields of thread::sem -- definitely count and potentially also acquire_status.

CU, Ingo

-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys-and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV