While others have produced an ebuild for the ATI catalyst 7.11 drivers that will compile in bug 200081, I believe that this ebuild is a better choice for the new binary drivers from ATI. I will try to explain why: /Name change Though ati-drivers-8.42.3 was compatible with xorg-server-1.4, that version is unlikely to ever be stabled. This therefore presents us with a unique opprtunity to make a "clean break" with the old name and versioning scheme. Any temporary fix to the versioning scheme would be only a band-aid. New xorg-servers will depend on ati-catalyst and the upgrade to ati-catalyst will happen as part of the upgrade to the new xorg-server. /Bug fixes This ebuild solves: bug 197378 (x11-drivers/ati-drivers depend on libXrandr) An easy addition of one line. bug 199146 (x11-drivers/ati-drivers support for debug output) The other ebuilds purport to solve this, but don't. They're missing a dir in their sed. Also, this ebuild uses the same coding as the rest of the ebuild to solve it. Tested and working. bug 197470 (x11-drivers/ati-drivers-8.40.4: ati-powermode.sh doesn't work) Tested patch from bug. Working. patch: ati-powermode.sh.who.patch bug 196684 (x11-drivers/ati-drivers resume broken with 2.6.23 due to renamed config option) I have made a patch that actually solves this problems for kernels < 2.6.23 too. Works for me, but not tested, since I have a non-SMP system. patch: ati-catalyst-smp-sleep-fix.patch Fix wannabe-bug where a QA warning would appear about TEXTRELS in glesx.so. /New features Working warnings patch. Split up the warnings patch into two. patch: ati-catalyst-firegl_smp_func_parameter_wrap-warning.patch Fixes "'firegl_smp_func_parameter_wrap' defined but not used" in a way that works and doesn't disable features on SMP systems. patch: ati-catalyst-x86_64-warnings.patch Fixes warnings on x86_64. Dropped the chunk with "firegl_interrupt_poll_wrap" because it was causing warnings on x86 systems, also seemed incorrect. New (very experimental) fireglhack use flag that enables users of FireGL cards to use the ati-catalyst drivers. Since we know that there were FireGL beta-testers who tried the drivers and since phoronix FireGL users have reported succes using them, this will probably be a worst-case scenario of a bad user experience. Nevertheless, apart from enabling the use flag, users will have to pass BREAKMYCOMPUTER="yes" to the emerge command to succesfully emerge ati-catalyst with that use flag, like so: BREAKMYCOMPUTER="yes" emerge ati-catalyst Reproducible: Always
Created attachment 136821 [details] ati-catalyst.tar.bz2 A .tar.bz2 archive of the entire thing. Broken-out to follow.
Created attachment 136822 [details, diff] ati.diff diff to ati-drivers-8.42.3
Created attachment 136824 [details] ati-catalyst-7.11.ebuild Ebuild as a whole.
Created attachment 136825 [details] ati-powermode.sh.who.patch This fixes bug 197470 goes in ${FILESDIR}
Created attachment 136826 [details] ati-catalyst-smp-sleep-fix.patch Fixes bug 196684 goes in ${FILESDIR}/${PV}
Created attachment 136827 [details] ati-catalyst-firegl_smp_func_parameter_wrap-warning.patch Fixes 'firegl_smp_func_parameter_wrap' defined but not used goes in ${FILESDIR}/${PV}
Created attachment 136829 [details] ati-catalyst-x86_64-warnings.patch Fixes warnings on x86_64 goes in ${FILESDIR}/${PV}
Created attachment 136830 [details] ati-catalyst-firegl-pci-id-hack-hack-hack.patch Allows FireGL users to try out the driver unofficially. goes in ${FILESDIR}/${PV}
(In reply to comment #6) > Created an attachment (id=136827) [edit] > ati-catalyst-firegl_smp_func_parameter_wrap-warning.patch > > Fixes 'firegl_smp_func_parameter_wrap' defined but not used > goes in ${FILESDIR}/${PV} > Is this only to fix a warning? Jakub could you mark this as being a dup of 200081?
(In reply to comment #9) > (In reply to comment #6) > > Created an attachment (id=136827) [edit] > > ati-catalyst-firegl_smp_func_parameter_wrap-warning.patch > > > > Fixes 'firegl_smp_func_parameter_wrap' defined but not used > > goes in ${FILESDIR}/${PV} > > > > Is this only to fix a warning? Actually it's to fix the warnings patch from bug 187015 which has made it so far with all the other ebuilds. It can be dropped with no bad consequences, but since people have seemed to want those patches till now, I thought it'd be better to fix it, so functionality isn't sacrificed on CONFIG_SMP systems. > Jakub could you mark this as being a dup of 200081? Jakub isn't in the CC list. If anything, perhaps mark bug 200081 as a duplicate of this? That of course depends on what naming scheme is chosen. THe ebuild in 200081 is still buggy, by the way.
(In reply to comment #10) > > Jakub could you mark this as being a dup of 200081? > > Jakub isn't in the CC list. If anything, perhaps mark bug 200081 as a duplicate > of this? > That of course depends on what naming scheme is chosen. Don't care, all I want is not to split the discussion. > THe ebuild in 200081 is still buggy, by the way. > Why?
8.433 is in the tree
