On Fri, Jun 4, 2010 at 10:14 AM, Otto Moerbeek <[email protected]> wrote: > On Fri, Jun 04, 2010 at 09:51:16AM +0200, Markus Hennecke wrote: > >> Am 03.06.2010 23:52, schrieb Ted Unangst: >> >On Thu, Jun 3, 2010 at 5:21 PM, Christopher Zimmermann >> ><[email protected]> wrote: >> >>Are you sure this is a problem in kernel? Christopher Linn and I only >> >>experience this problem with gtk2 apps. How could the kernel know wether the >> >>current focus is on a gtk2 window? >> >Because the kernel is responsible for saving and restoring FPU state, >> >which is pretty high on the list of suspected trouble makers. >> Which would explain why the problem goes away if mouse acceleration >> is disabled via xset. >> >> Kind regards, >> Markus > > The problem only becomes less frequent with xset m 1. >
Thanks to the efforts of kettenis@, deraadt@ and guenter@, the problem has been identifier and fixed by this commit: > CVSROOT: /cvs > Module name: src > Changes by: [email protected] 2010/06/04 23:55:56 > > Modified files: > sys/arch/amd64/amd64: machdep.c > Log message: > Pass signal handlers a more clean FPU state (ie. the right modes) and > properly restore it in sigreturn. Lots of deep digging by matthieu, > otto, guenther, kettenis and I.. and I am certain I forgot some other > people. > ok kettenis otto matthieu -- Matthieu Herrb
