On Tue, Apr 28, 2009 at 01:58:34PM +0300, Ville Syrjälä wrote: > On Tue, Apr 28, 2009 at 10:13:30AM +0200, Niels Roest wrote: > > Hi Ville, > > that's kinda odd, I am using this for some time now and I am experiencing > > no problem, e.g. with X11. > > What is your set-up? > > x86-32 laptop + fbdev + 2.6.29.1 kernel. I so far only updated this one > laptop with this version. There is something else fishy going on with > this kernel though since powertop reports >1000 wakeups per second on > an idle system but it doesn't list any reason for the excessive wakeups. > DirectFB is not involved with that since I even rmmoded fusion and it > still does that. > > I will try to update a few other systems with this stuff and see what > happens.
Well the same thing happened on another x86-32 laptop. This one didn't even have the powertop issue so we can rule that out as a contributing factor. The kernel was 2.6.29-rc2. I then tried it on my x86-64 dual core desktop and didn't see the problem there. The kernel on that one was 2.6.28 so I can't yet be sure if the difference is caused by the kernel version or 32bit vs. 64bit or UP vs. SMP. I will update the kernel on the 64 bit system tonight and see what happens. I am wondering how the code is supposed to work though. do_select() puts the task to INTERRUPTIBLE state and then calls the poll fop which calls wake_up_all() uncoditionally. AFAICS wake_up_all() ends up in try_to_wake_up() which should put the task back to RUNNING state. So how is it supposed to sleep at all? The code is rather hard to follow though so perhaps I missed something. -- Ville Syrjälä syrj...@sci.fi http://www.sci.fi/~syrjala/ _______________________________________________ directfb-dev mailing list directfb-dev@directfb.org http://mail.directfb.org/cgi-bin/mailman/listinfo/directfb-dev
