subj Reproducible: Always
just tried make ebuild for this version. renamed one from x11-overlay, added fetch restriction, changed SRC_URI. and then followed requirements from release notes. x11 started. everything seems to work.
I'm on 4.0.5, I needed these patches in addition to those already in the ebuild: kolasa-3.19-get_cpu_var.patch kolasa_4.0-cr4-strn.patch kolasa_4.1_remove-IRQF_DISABLED.patch fglrx_gpl_symbol.patch I fetched them from here: https://aur.archlinux.org/packages/catalyst/?comments=all That gpl patch might be an issue that makes this uninstallable on >=4.0-kernels without cheating though...
(In reply to Mads from comment #2) > That gpl patch might be an issue that makes this uninstallable on > >=4.0-kernels without cheating though... yep and we can't add it to the tree for that reason, which limits ati-drivers currently to kernels <4.0.
Sorry for not being responsive here, as far as I understood this is older release branch that is less useful than the latest ebuild in portage, I guess this ebuild could just be updated and then 15.1 could be bumped to higher version to allow people still to get the best available code. Does this release bring any value beyond ati-drivers-15.1 ?
*** Bug 552596 has been marked as a duplicate of this bug. ***
With this release suspend to ram finally works. With previous I was unable to resume from suspend.
http://www.gentooforum.de/artikel/22061/linux-4-0-5-ati-drivers-cpu-tlbstate-problem.html may be the solution to the license problem. 15.5 does not need the license trickery until kernel's modpost calls EXPORT_SYMBOL_GPL(cpu_tlbstate) instead of EXPORT_SYMBOL(cpu_tlbstate) patching /usr/src/linux/arch/x86/mm/init.c allows the ebuild install to run. not knowing what parts of the amd 15.5 download need to be in the portage distfile I user patched ati-drivers-15.1 (fglrx-core-15.200). see bug 548118.
New Features: AMD PowerXpress support for laptops equipped with Intel 6th generation (Skylake) CPUs Linux Platform Atomics & SVM Fine Grain Buffer support for Carrizo APUs Multi-Device support for OpenCL 2.0 Website still says only kernel <4.0 support
If bug #553244 got fixed, people could manually inject fglrx_gpl_symbol.patch if needed.
submitted https://bugzilla.kernel.org/show_bug.cgi?id=101281 hoping kernel devs will fix the license problem.
Pushed to x11 overlay, though I did not get it to work immediately. Also, if you have HSA hardware, please test for regressions, one of the hsa drivers was either merged to something or removed.
Created attachment 406430 [details] results of emerge of ati-drivers-15.7 emerge of 15.7 from overlay failed. apparent regressions in amd's code. see the attachment.
Created attachment 406432 [details, diff] remove IRQF_DISABLED, makes it compile on 4.1++ Thanks! When compiling the ebuild using gentoo-sources-4.1.1, I run into missing IRQF_DISABLED compile errors. Yet, kolasa_4.1_remove-IRQF_DISABLED.patch doesn't apply for me, therefore I created the attached patch. And of course, I need: fglrx_gpl_symbol.patch kolasa_4.0-cr4-strn.patch (both from the AUR link).
Not sure if it's news, since I didn't use ati-drivers for a long time, but I don't need to use any gdm-hack on this driver version.
(In reply to Olaf Leidinger from comment #14) > Not sure if it's news, since I didn't use ati-drivers for a long time, but I > don't need to use any gdm-hack on this driver version. Yes it is, thanks for noting.
*ati-drivers-15.7 (10 Jul 2015) 10 Jul 2015; Manuel Rüger <mrueg@gentoo.org> +ati-drivers-15.7.ebuild: Version bump. Adds epatch_user support. Proxy commit for Emil Karlson. Fixes bugs #551458 #553244.
it's not buildable with any kernel declared as supported in release notes. * Failed Patch: kernel-4-compat.patch great work. thanks!
false alarm. my bad
(In reply to Rion from comment #17) > it's not buildable with any kernel declared as supported in release notes. > * Failed Patch: kernel-4-compat.patch > > great work. thanks! (In reply to Rion from comment #18) > false alarm. my bad I don't see the patch included in our cvs repository. Remember that, if you apply user patches, you apply them on your own risk and without support. If there are additional issues with the package itself please file another bug. Thanks!