On Fri, Feb 15, 2008 at 9:10 AM, Fabien MAHOT <[EMAIL PROTECTED]> wrote: > OK, we can see that there is the same problem. T1 thread prempts Main > Thread during the pthread_create of T2 with Xenomai and not with Linux > standard. I would like to know why. > > I don't that there is a difference beetween the posix Linux pthread_create > and the posix Xenomai pthread_create ? > > If pthread_create suspends a short time its caller, why we don t have the > same behaviour in the two cases ? > > In my opinion, there is a problem with Xenomai, no ? > > I hope that you could help me. thank you
Could you try the attached patch ? It should give you the illusion that pthread_create caller is not suspended if your glibc has none of the bugs we worked around. That said, I really think that you should not rely on this behaviour of pthread_create, because it is not garanteed by the posix standard, besides, it will not work with broken glibcs. And in fact, a usual rule is not to rely on scheduler priorities for mutual exclusion, but rather to make it explicit by using mutex or semaphores. -- Gilles Chanteperdrix
xeno-posix-fix-pthread_create.diff
Description: Binary data
_______________________________________________ Xenomai-help mailing list Xenomai-help@gna.org https://mail.gna.org/listinfo/xenomai-help
