[R300 and other radeons] MergedFB lockups

[Benjamin Herrenschmidt]

On Sat, 2005-02-19 at 11:23 +0100, Philipp Klaus Krause wrote:
> Vladimir Dergachev schrieb:
> > 
> > I think I found the cause of lockups in VB mode - they were due to 
> > cursor updating function in radeon_mergedfb.c calling OUTREGP() which in 
> > turn called INREG.
> > 
> > When silken mouse is enabled this function could be called at any time 
> > which would do *bad* things when CP engine is active.
> > 
> > The fix of putting RADEONWaitForIdleMMIO() works fine on my setup.
> > 
> > I have *no* idea why this worked with immediate mode at all and why no 
> > issues were reported by R200 and Radeon users (well, I only looked 
> > through the mailing lists, perhaps there is something on bugzilla but I 
> > don't know how to use that efficiently)
> 
> Well my rv250 lockups occour only during mouse movement in fullscreen
> applications. But for month ago there were no lockups. The situation
> was slowly getting worse since. With current DRM and Mesa driver I
> get an immediate lockup in gl-117 when moving the mouse. If I use
> current DRM with an older Mesa it locks up after about a minute of mouse
> movement.

rv250 here, CVS HEAD X.org, Meas CVS Head DRI driver, current bk kernel
(well, actually, all of the above are about 5 days old, doesn't have
Vladimir latest fixes), I get very jerky display with g117 and broken
textures, ultimately it locks up after moving the mouse a little bit (at
the meny screen).

I can kill X tho, that works, so the chip/bus isn't totally dead (or it
recovers with an engine reset).

Ben.