Branch :
| Author | Commit | Date | CI | Message |
|---|---|---|---|---|
| 939203af | 2025-07-22 07:05:18 | radeonsi: always lower alu bit sizes From Georg Lehmann 3bc31dccee87af718707d89527afa92ade3568ff in Mesa main | ||
| 599faf75 | 2025-07-22 06:56:57 | Revert 25.0 version of "radeonsi: always lower alu bit sizes" 2d525d5e3401555d97fe827f8db5bfed8887dc33 in Mesa 25.0 33b5d8b2ec26904ec1c331dd64ee04e351fd8e0e in Mesa main The backport caused Xorg to abort with 'Unsupported opcode: v1: :v[2] = v_lshlrev_b16 8, :v[2' on a machine with PITCAIRN radeondrm when running startx. Problem reported by Avon Robertson on misc@. discussed upstream: https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35150 | ||
| 08fc30ef | 2025-06-24 14:57:03 | With LLVM 19 setting attributes to enable BTI and PAC at the module level no longer has the desired effect. Instead we have to set attributes on each function in the module. Fixes software rendering with LLVM on machines with BTI. ok jsg@, mbuhl@ | ||
| 452696b8 | 2025-06-05 14:17:56 | Merge Mesa 25.0.7 | ||
| 67d6f117 | 2025-06-05 11:23:11 | Import Mesa 25.0.7 | ||
| ac809dfd | 2025-01-16 15:27:05 | unbreak build with llvm-19+ by pulling the following upstream commits 5a9c052ba78e5d1b0020b8f8320d211d2205a958 amd: Include missing llvm IR header Module.h fa9cd89a85b904615ebc11da609445b5b751e68d Update lp_bld_misc.cpp to support llvm-19+. a8eed9cca26b1c4b6526be7d042adec6703a30ae gallium: Don't pass avx512er and avx512pf features on LLVM 19 ok jsg@ | ||
| 277d4fbe | 2024-07-25 09:04:06 | Pass no-verify-fixpoint option to instcombine in LLVM 18 From Nikita Popov 99f0449987bec1f82cd42a06f40bb4a863a37792 in mainline Mesa | ||
| d230847f | 2024-04-03 21:47:30 | Enable PAC in addition to BTI on arm64 such that JIT code matches the default branch protection provided by our base compiler. ok jsg@ | ||
| f3c1e4e9 | 2024-04-02 10:41:59 | Merge Mesa 23.3.6 | ||
| 874d2ba3 | 2024-04-02 09:25:31 | Import Mesa 23.3.6 | ||
| d127cc35 | 2024-03-18 00:15:10 | Tell LLVM to generate code with BTI instructions. "looks ok" jsg@, ok deraadt@ | ||
| 0eabb5c6 | 2023-12-01 07:56:04 | r600/sfn: Don't try to re-use iterators when the set is made empty From Gert Wollny c13de0509c43f9b9764dc939aa64fe70c6a80870 in mainline Mesa fixes games/xonotic crash on r600 reported by edd@ on bugs@ | ||
| 3c324595 | 2023-11-02 06:00:03 | don't warn when file descriptor equality is unknown linux uses kcmp(2) with KCMP_FILE, we have no equivalent | ||
| 4f0d8474 | 2023-11-02 04:53:12 | Merge Mesa 23.1.9 | ||
| 60059fee | 2023-11-02 04:31:01 | Import Mesa 23.1.9 | ||
| 5f7ec4d4 | 2023-04-06 10:23:42 | Merge Mesa 22.3.7 | ||
| 175bb703 | 2023-04-06 10:08:20 | Import Mesa 22.3.7 | ||
| 687ad782 | 2023-03-10 04:15:06 | revert upstream Mesa changes to remove non-TLS paths https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17814 https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17213 Emulated TLS isn't enough when many shared libraries are involved. This caused problems with some ports: kitty, calibre, qutebrowser. reported by tb@ and others | ||
| b5d82fed | 2023-01-28 08:56:19 | Merge Mesa 22.3.4 | ||
| 5642995e | 2023-01-28 08:09:28 | Import Mesa 22.3.4 | ||
| 796f4ae4 | 2022-09-13 10:04:23 | iris: check i915 features after hw gen when running recent Mesa on i855 (gen 2) without amber drivers: error: Kernel is too old for Iris. Consider upgrading to kernel v4.16. libGL error: glx: failed to create dri3 screen libGL error: failed to load driver: iris error: Kernel is too old for Iris. Consider upgrading to kernel v4.16. libGL error: glx: failed to create dri2 screen libGL error: failed to load driver: iris move the i915 feature check to after the hardware generation check which results in: MESA: warning: Driver does not support the 0x3582 PCI ID. libGL error: glx: failed to create dri3 screen libGL error: failed to load driver: iris MESA: warning: Driver does not support the 0x3582 PCI ID. libGL error: glx: failed to create dri2 screen libGL error: failed to load driver: iris | ||
| 2e058c3c | 2022-09-02 05:46:23 | Merge Mesa 22.1.7 | ||
| 4968d392 | 2022-09-02 05:12:48 | Import Mesa 22.1.7 | ||
| f4f27ace | 2022-04-21 04:37:17 | Merge Mesa 21.3.8 | ||
| 9c5449b6 | 2022-04-21 04:19:02 | Import Mesa 21.3.8 | ||
| 75c71730 | 2022-02-24 02:29:36 | Merge Mesa 21.3.7 | ||
| 50b4ee55 | 2022-02-24 01:51:18 | Import Mesa 21.3.7 | ||
| 3cd30267 | 2021-12-09 00:21:07 | backport from Mesa upstream to fix i386 build with llvm 13 gallivm: fix FTBFS on i386 with LLVM >= 13, StackAlignmentOverride is gone 3a2d317b996f57647da23de7876142be4b9b71f3 gallivm: add new wrapper around Module::setOverrideStackAlignment() c1b4c64a28d9fc093229eab91a3a7fc4cb4fe29a https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11940 ok sthen@ | ||
| 3715a2f9 | 2021-09-09 09:59:12 | Merge Mesa 21.1.8 | ||
| bf4ec162 | 2021-09-09 09:42:50 | Import Mesa 21.1.8 | ||
| aa17e02d | 2021-07-22 10:50:23 | Merge Mesa 21.1.5 | ||
| f849497a | 2021-07-22 09:57:37 | Import Mesa 21.1.5 | ||
| d494b905 | 2021-05-25 05:53:44 | remove 'si_init_perfcounters: max_sh_per_se' warning In drm from linux >= 5.9 renoir no longer uses the gpu info firmware and some reported hardware information changes. 'drm/amdgpu: use IP discovery table for renoir' c1cf79ca5ced46ea6a7843cbe98836e4d51c6c52 Remove warning which is known to be harmless https://gitlab.freedesktop.org/drm/amd/-/issues/1332 and was removed with mesa 20.2 in 'radeonsi/gfx10: implement most performance counters' 38a4b861459b02401d3ff71670218506e7acf019 | ||
| 9052cbbf | 2021-03-11 08:21:03 | gallivm: fix half to float conversions with llvm 11 From Roland Scheidegger 51a82ec3e437d1d2dc4c688578640d25b3e7f0a2 in mainline Mesa | ||
| 9f99a932 | 2021-03-11 05:14:48 | ac: unify denorm setting enforcement From Marek Olsak 56cc10bd27b24d513de88bf7fa94a6c8f43e348f in mainline Mesa | ||
| 5546265a | 2021-03-09 09:24:08 | gallium/gallivm: remove unused header include for newer LLVM From Jan Zielinski 52aa730d07618513d6c055618069b2f4680974cc in mainline Mesa Encountered by naddy@ when doing a clang 11 build. This commit suggested by patrick@ | ||
| cf3c0c9e | 2021-01-18 01:47:11 | gallivm: Add missing header for powf. From Vinson Lee ff8daa013621019f1606dc0c188b16f1ce34fea7 in mainline Mesa fixes clang 11 build mortimer@ had almost the same diff | ||
| 41a55d03 | 2020-09-22 02:08:55 | Merge Mesa 20.0.8 With Mesa 20.1 even after the kernel change to do wbinvd on all cpus sthen@ reported that hard hangs still occurred on his Haswell system with inteldrm. Mark Kane also reported seeing hangs on Ivy Bridge on bugs@. Some systems/workloads seem to be more prone to triggering this than others as I have not seen any hangs on Ivy Bridge and the only hangs I saw on Haswell when running piglit went away with the wbinvd change. It seems something is wrong with drm memory attributes or coherency in the kernel and newer Mesa versions expect behaviour we don't have. | ||
| 5bc2fd43 | 2020-09-22 01:28:34 | Import Mesa 20.0.8 | ||
| 1c86a178 | 2020-09-03 07:17:20 | Merge Mesa 20.1.7 | ||
| 7fa385c8 | 2020-09-03 06:40:00 | Import Mesa 20.1.7 | ||
| 6cd42bcb | 2020-08-26 06:02:31 | Merge Mesa 20.1.6 | ||
| 241b1c13 | 2020-08-26 05:26:37 | Import Mesa 20.1.6 | ||
| 587f130e | 2020-08-01 03:51:31 | Fix Mesa build with clang 10 on mips64. On mips64, the compiler does not allow use of non-zero argument with __builtin_frame_address(). However, the returned frame address is only used when PIPE_ARCH_X86 is defined. The compile error can be avoided by making #ifdef PIPE_ARCH_X86 cover the getting of frame address too. The argument checking of __builtin_frame_address() has been present as a debug assert in clang 8. In clang 10, there is a proper runtime check for the argument. This is why the build has not failed before. OK jsg@ | ||
| cb3c6b92 | 2020-06-18 11:45:13 | llvmpipe: use LLVM version string instead of re-computing it From Eric Engestrom 7f5ef97a07d4054efb96f0d644344644023af82c in Mesa git fixes llvmpipe renderer string showing "LLVM 16.0" with LLVM 10 | ||
| d37d5f3d | 2020-01-22 03:34:13 | fix some things which got broken in merge | ||
| c0a649c0 | 2020-01-22 03:07:41 | add files generated by python scripts which stopped being shipped with distfiles when upstream removed autotools | ||
| 75490c6c | 2020-01-22 03:02:59 | add back autotools files removed upstream | ||
| 2796e1e6 | 2020-01-22 02:49:17 | Merge Mesa 19.2.8 | ||
| 18af3bca | 2020-01-22 02:07:15 | Import Mesa 19.2.8 | ||
| 76ee08e8 | 2019-11-20 03:13:04 | Call shmget() with permission 0600 instead of 0777 From Brian Paul 02c3dad0f3b4d26e0faa5cc51d06bc50d693dcdc in mainline Mesa "A security advisory (TALOS-2019-0857/CVE-2019-5068) found that creating shared memory regions with permission mode 0777 could allow any user to access that memory. Several Mesa drivers use shared- memory XImages to implement back buffers for improved performance. This path changes the shmget() calls to use 0600 (user r/w). Tested with legacy Xlib driver and llvmpipe." | ||
| ff40cb71 | 2019-07-01 08:36:47 | Merge Mesa 19.0.8 | ||
| b1b22c2a | 2019-07-01 07:44:02 | Import Mesa 19.0.8 | ||
| 1c9ad579 | 2019-05-23 05:33:19 | Merge Mesa 19.0.5 | ||
| 1dab445d | 2019-05-23 04:22:53 | Import Mesa 19.0.5 | ||
| 702ae9a5 | 2019-03-19 10:59:02 | Merge Mesa 18.3.5 | ||
| 34c49fbb | 2019-03-19 09:37:42 | Import Mesa 18.3.5 | ||
| b9f320ec | 2019-02-19 04:24:00 | Merge Mesa 18.3.4 | ||
| ae7ee137 | 2019-02-19 03:25:14 | Import Mesa 18.3.4 | ||
| 338bc0a2 | 2019-01-29 11:52:04 | Merge Mesa 18.3.2 | ||
| fecc0d40 | 2019-01-29 10:45:31 | Import Mesa 18.3.2 | ||
| 19f2c52c | 2018-10-23 06:35:32 | Merge Mesa 17.3.9 Mesa 18.x needs an ld with build-id for at least the intel code Mesa 18.2 assumes linux only memfd syscalls in intel code Tested by matthieu@, kettenis@ and myself on a variety of hardware and architectures. ok kettenis@ | ||
| 587354b6 | 2018-10-23 05:40:45 | Import Mesa 17.3.9 | ||
| 90d807dc | 2018-03-02 05:25:17 | use clock_gettime() path on PIPE_OS_BSD in os_time_get_nano() committed upstream as 7bea40e56652a1ded4374d92fb340b454fbac475 clock_nanosleep() isn't available yet so the usleep() path stays for os_time_sleep() | ||
| 646d14d2 | 2018-01-08 05:41:20 | Revert to Mesa 13.0.6 again. Corruption has again been reported on Intel hardware running Xorg with the modesetting driver (which uses OpenGL based acceleration instead of SNA acceleration the intel driver defaults to). Reported in various forms on Sandy Bridge (X220), Ivy Bridge (X230) and Haswell (X240). Confirmed to not occur with the intel driver but the xserver was changed to default to the modesetting driver on >= gen4 hardware (except Ironlake). One means of triggering this is to open a large pdf with xpdf on an idle machine and highlight a section of the document. There have been reports of gpu hangs on gen4 intel hardware (T500 with GM45, X61 with 965GM) when starting Xorg as well. | ||
| 702572c9 | 2017-12-31 07:12:12 | Merge Mesa 17.2.8 | ||
| cc122a5e | 2017-12-31 06:17:38 | Import Mesa 17.2.8 | ||
| ad2ad70a | 2017-08-26 16:59:17 | Revert to Mesa 13.0.6 to hopefully address rendering issues a handful of people have reported with xpdf/fvwm on ivy bridge with modesetting driver. | ||
| 36c1bd02 | 2017-08-14 09:57:57 | Merge Mesa 17.1.6 | ||
| 6526d331 | 2017-08-14 09:30:06 | Import Mesa 17.1.6 | ||
| fa6c2100 | 2017-03-25 00:29:27 | Merge Mesa 13.0.6 | ||
| 480cde33 | 2017-03-25 00:05:01 | Import Mesa 13.0.6 | ||
| f9668bb9 | 2017-02-26 12:28:55 | Merge Mesa 13.0.5 | ||
| ba80afc3 | 2017-02-26 12:06:04 | Import Mesa 13.0.5 | ||
| 6d4ad6fa | 2017-01-19 05:38:52 | Import Mesa 13.0.3 | ||
| 2047dff5 | 2016-12-11 10:03:01 | disable additional rules to regen sources when python is not found After a cvs import timestamps change which triggers rules that attempt to call python to regenerate files. When python is not found this generates empty files that override those from the distfile, breaking the build. When building inside xenocara python is never found as the pkg paths are not searched. | ||
| e2d98075 | 2016-12-11 08:53:19 | Merge Mesa 13.0.2 | ||
| a53e30e0 | 2016-12-11 08:25:29 | Import Mesa 13.0.2 | ||
| fd18c20e | 2016-10-11 22:14:30 | regen | ||
| 4159a76a | 2016-07-25 20:12:06 | Disable the code that allocates W|X memory. There is fallback code that gets used if allocating W|X memory fails, which is probably a bit slower. However, that is much better than commit a W^X violation which currently gets you killed. ok jca@ | ||
| 363dd58b | 2016-05-29 10:40:19 | Merge Mesa 11.2.2 | ||
| f8d9379c | 2016-05-29 10:11:54 | Import Mesa 11.2.2 | ||
| f83fcc6f | 2016-01-24 00:59:28 | Import Mesa 11.0.9 | ||
| a06be69d | 2015-12-23 13:43:05 | Merge Mesa 11.0.8 | ||
| 9efbf61f | 2015-12-23 13:18:51 | Import Mesa 11.0.8 This seems to fix some of the problems with clutter/gnome reported to occur on r600 with 11.0.6 | ||
| 987bf76a | 2015-11-22 07:14:29 | regen | ||
| a8524181 | 2015-11-22 06:57:53 | Disable the bison/lex/python rules to regenerate sources that were included in the tarball. cvs import seems to have changed timestamps of files which caused these rules to trigger and the build to fail. | ||
| 0ecb5590 | 2015-11-22 03:01:23 | don't try to symlink files that don't exist | ||
| 1fe54f21 | 2015-11-22 02:36:47 | import Mesa 11.0.6 |