On Fri, Jan 17, 2014 at 06:00:13PM +0100, Daniel Vetter wrote: > On Fri, Jan 17, 2014 at 11:44:33AM +0200, ville.syrjala@xxxxxxxxxxxxxxx wrote: > > From: Ville Syrjälä <ville.syrjala@xxxxxxxxxxxxxxx> > > > > I'm interested in underruns so having the totally off is not good. After > > disabling underruns, re-enable them after 2 seconds. I just added one > > timer for this, even though we should have one for each PCH and CPU, > > or maybe even per pipe/transcoder, but then we should track underrun > > disable also per pipe/transcoder. > > > > Signed-off-by: Ville Syrjälä <ville.syrjala@xxxxxxxxxxxxxxx> > > The kernel has some neat ratelimiting stuff in ratelimit.h. You've looked > into that? Open coding timer code always freaks me out a bit because of > the bazillion ways you can screw up jiffy handling ;-) Nope didn't really look. printk_timed_ratelimit() would seem like a reasonable thing, except on PCH platforms we're also limiting the interrupts themself, not just the printks. So using prink ratelimiting can't give us quite the same thing. -- Ville Syrjälä Intel OTC _______________________________________________ Intel-gfx mailing list Intel-gfx@xxxxxxxxxxxxxxxxxxxxx http://lists.freedesktop.org/mailman/listinfo/intel-gfx
