[Mesa3d-dev] [Regression] [bisected] r7xx rendering problems with master branch
Hello list, The environment: * Gentoo amd64; * kernel from the drm-radeon-testing branch of git://git.kernel.org/pub/scm/linux/kernel/git/airlied/drm-2.6 (latest HEAD); * lspci says about my card: "01:00.0 VGA compatible controller: ATI Technologies Inc RV790 [Radeon HD 4800 Series]"; * libdrm from master; * mesa from master; * radeon ddx from master; * Xorg server 1.7.4 RC 2. The testbed: the glsnake screensaver from xscreensaver, and crack-attack - both render badly with master. The symptoms are that the patterns rendered to the screen leave some traces behind them - ultimately, the screen turns completely white with the screensaver (glsnake is not the one with which that happens). I have bisected this rendering bug to commit 3728673bd1b974e54858fbab6ff62d3607b0d3f0. The comment says: Note that there's quite a few tests for ctx->Color.BlendEnabled != 0 in drivers, etc. Those tests can remain as-is since the mask will be 0 or ~0 unless GL_EXT_draw_buffers2 is enabled. And indeed, the r600_dri driver is such one, and more specifically, it seems, r7xx series (such as my card): f...@erwin ~/src/mesa $ git grep -n BlendEnabled src/mesa/drivers/dri/r600/ src/mesa/drivers/dri/r600/r700_state.c:454: if (RGBA_LOGICOP_ENABLED(ctx) || !ctx->Color.BlendEnabled) { Unfortunately, reverting this particular commit is nontrivial :( -- Francis Galiegue, fgalie...@gmail.com "It seems obvious [...] that at least some 'business intelligence' tools invest so much intelligence on the business side that they have nothing left for generating SQL queries" (Stéphane Faroult, in "The Art of SQL", ISBN 0-596-00894-5) -- This SF.Net email is sponsored by the Verizon Developer Community Take advantage of Verizon's best-in-class app development support A streamlined, 14 day to market process makes app distribution fast and easy Join now and get one step closer to millions of Verizon customers http://p.sf.net/sfu/verizon-dev2dev ___ Mesa3d-dev mailing list Mesa3d-dev@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/mesa3d-dev
Re: [Mesa3d-dev] [Regression] [bisected] r7xx rendering problems with master branch
On Sat, Jan 2, 2010 at 10:21 PM, Francis Galiegue wrote: [...] > (glsnake is not the one with which that happens). > Sorry, I meant "not the _only_ one" :/ -- Francis Galiegue, fgalie...@gmail.com "It seems obvious [...] that at least some 'business intelligence' tools invest so much intelligence on the business side that they have nothing left for generating SQL queries" (Stéphane Faroult, in "The Art of SQL", ISBN 0-596-00894-5) -- This SF.Net email is sponsored by the Verizon Developer Community Take advantage of Verizon's best-in-class app development support A streamlined, 14 day to market process makes app distribution fast and easy Join now and get one step closer to millions of Verizon customers http://p.sf.net/sfu/verizon-dev2dev ___ Mesa3d-dev mailing list Mesa3d-dev@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/mesa3d-dev
Re: [Mesa3d-dev] [Regression] [bisected] r7xx rendering problems with master branch - SOLVED! Sort of... (with patch)
On Sat, Jan 2, 2010 at 10:21 PM, Francis Galiegue wrote: > Hello list, > > The environment: > > * Gentoo amd64; > * kernel from the drm-radeon-testing branch of > git://git.kernel.org/pub/scm/linux/kernel/git/airlied/drm-2.6 (latest > HEAD); > * lspci says about my card: "01:00.0 VGA compatible controller: ATI > Technologies Inc RV790 [Radeon HD 4800 Series]"; > * libdrm from master; > * mesa from master; > * radeon ddx from master; > * Xorg server 1.7.4 RC 2. > > The testbed: the glsnake screensaver from xscreensaver, and > crack-attack - both render badly with master. The symptoms are that > the patterns rendered to the screen leave some traces behind them - > ultimately, the screen turns completely white with the screensaver > (glsnake is not the one with which that happens). > > I have bisected this rendering bug to commit > 3728673bd1b974e54858fbab6ff62d3607b0d3f0. The comment says: > > > Note that there's quite a few tests for ctx->Color.BlendEnabled != 0 in > drivers, etc. Those tests can remain as-is since the mask will be 0 or ~0 > unless GL_EXT_draw_buffers2 is enabled. > > Well, I have good news! First of all, on Freenode/#radeon, taiu, by his nick, reported that the problem also existed on Intel. Then he told me about commit 1677d5c0aebe8edb54aa3236294acede536013f3 in master (mesa: fix-up blend enable/disable code in _mesa_PopAttrib()), which changes file src/mesa/main/attrib.c. He told me to try and apply the same changes to src/mesa/drivers/common/meta.c. I did it and... It worked! Now glsnake and crack-attack render correctly. I also tested that it solves the rendenring on related screensavers (glmatrix is one). Patch to follow in a next mail. -- Francis Galiegue, fgalie...@gmail.com "It seems obvious [...] that at least some 'business intelligence' tools invest so much intelligence on the business side that they have nothing left for generating SQL queries" (Stéphane Faroult, in "The Art of SQL", ISBN 0-596-00894-5) -- This SF.Net email is sponsored by the Verizon Developer Community Take advantage of Verizon's best-in-class app development support A streamlined, 14 day to market process makes app distribution fast and easy Join now and get one step closer to millions of Verizon customers http://p.sf.net/sfu/verizon-dev2dev ___ Mesa3d-dev mailing list Mesa3d-dev@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/mesa3d-dev