ipc, sockets and windows sp2

[Vincent Dedun]

Corinna Vinschen wrote :

>
>Thanks again for the testcase.  It helped to track down the problem which
>was a result of my previous check in.  It should be solved in CVS now.
>
>Since you're building from CVS anyway, I don't create another snapshot
>for now.  We're that close to 1.5.14 anyway...
>  
>
Thanks again for having solved this so quickly.

However, semaphores still doesn't work properly.
There is no more problem with semop not waiting, but with quick 
semaphores locking unlocking.

I attach a new testcase, which is the same as previous one, except each 
child task will lock the semaphore, wait 100ms, then release the 
semaphore and die. Each time a child dies, a new one is created to keep 
10 running childs.

On osx, you will see forever locking/unlocking of semaphore written on 
output.
On lastest cygwin, you will see some locking/unlocking (about 11 or 12 
lock/unlock pair for me), then every child keep waiting for locking, and 
the program comes in dead-lock.

I think that semop doesn't get awaken properly when semaphore value 
change quickly.

Same conditions (windows/cygwin version), as previous mail apply.

Kraken
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: fork-ipc-sem2.c
URL: <http://cygwin.com/pipermail/cygwin/attachments/20050404/e22c8d7d/attachment.c>
-------------- next part --------------
--
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple
Problem reports:       http://cygwin.com/problems.html
Documentation:         http://cygwin.com/docs.html
FAQ:                   http://cygwin.com/faq/