* Michel Dänzer wrote: > On Thu, 2006-09-07 at 10:48 +0200, Thierry Reding wrote: > > > > 20_dont-backfill-bg-none.diff: > > Not applied upstream, but probably just an optimization. > > [...] > > > 26_offscreen-pixmaps.diff: > > Not applied upstream. It seems to disables all offscreen pixmaps. Option > > "XAANoOffscreenPixmaps" works just as well, but there may be more involved. > > The patch only disables offscreen pixmaps when the > GLX_EXT_texture_from_pixmap extension is first used, so it's probably an > improvement for people switching between GLX compositing managers and > other means of presentation. However, I wonder if this works with EXA. > It doesn't seem to check that XAA is actually active, and I suspect it's > basically luck if the XAA functions don't crash when it's not. This > needs clarifying before the patch can be applied. I would be willing to do some testing, but I'm unsure as to where to start. In addition, I only have an Intel 855GM card to test on, and IIRC, EXA for the i810 driver wasn't very stable last time I checked. > > 28_no-move-damage.diff: > > Not applied upstream, probably just an optimization. Kristian comments > > about why this patch is useful. > > Can you check with him why he hasn't pushed it upstream yet? Last I > heard it had some semantic issues IIRC. The same may be true about patch > 20 I suspect. I will do that. > > In summary, patches 25 and 27 are essential to make the xserver build. > > ... with a current Mesa snapshot? Yes, I'm currently running a recent version from Kristian's mirrored Mesa in git. > > Patch 23 is also more or less necessary, because otherwise X is unusable after > > switching to the console and back. > > Indeed. Without this patch, AIGLX should definitely be disabled by > default for etch. With this patch (or whatever ends up getting applied > upstream), we could leave it enabled by default for now and wait for bug > reports once it's in sid or even testing. I can only say that the patch works fine for me, but I am uncertain as to why it hasn't been included upstream. I will ask Kristian about it. > > Finally I think it'd be safe to also add patches 21, 22 and 24 since they have > > already been applied to upstream git, and especially 21 dramatically increases > > rendering speed. > > I agree. These patches should have very low if any impact outside of > AIGLX. > > > The only problem remaining is that the version of Mesa currently in > > experimental is not recent enough for compiz to work correctly. > > I think we should go with the 6.5.1 release anyway. That was going to be next on my checklist. Try and build with Mesa 6.5.1, but I think Shawn already tried that and it still needed some patching to work correctly, but I'll double-check. I am going to be away on holiday for a week from tomorrow, and it's unlikely I'll get any further testing done, but I'll get back to it as soon as I'm back (which should be around the 16th). - Thierry P.S.: no need to Cc: me, I'm subscribed to the list.
Attachment:
signature.asc
Description: Digital signature