1Mesa 20.0.8 Release Notes / 2020-06-11 2====================================== 3 4Mesa 20.0.8 is a bug fix release which fixes bugs found since the 20.0.7 5release. 6 7Mesa 20.0.8 implements the OpenGL 4.6 API, but the version reported by 8glGetString(GL_VERSION) or glGetIntegerv(GL_MAJOR_VERSION) / 9glGetIntegerv(GL_MINOR_VERSION) depends on the particular driver being 10used. Some drivers don't support all the features required in OpenGL 114.6. OpenGL 4.6 is **only** available if requested at context creation. 12Compatibility contexts may report a lower version depending on each 13driver. 14 15Mesa 20.0.8 implements the Vulkan 1.2 API, but the version reported by 16the apiVersion property of the VkPhysicalDeviceProperties struct depends 17on the particular driver being used. 18 19SHA256 checksum 20--------------- 21 22:: 23 24 6cf0c010df89680f9b2bc6432ff01400031795e39bceda7535fa00af06740b6c mesa-20.0.8.tar.xz 25 26New features 27------------ 28 29- VK_GOOGLE_user_type on ANV and RADV. 30 31Bug fixes 32--------- 33 34- iris/i965: possible regression in 20.0.5 due to changes in buffer 35 manager sharing across screens (firefox/mozilla#1634213) 36- [RADV] - Doom Eternal (782330) & Metro Exodus (412020) - Title 37 requires 'RADV_DEBUG=zerovram' to eliminate colorful graphical 38 aberrations. 39- [RADV] - Doom Eternal (782330) & Metro Exodus (412020) - Title 40 requires 'RADV_DEBUG=zerovram' to eliminate colorful graphical 41 aberrations. 42- NIR validation failed after glsl to nir, before function inline, 43 wrong {src,dst}->type ? 44- Mesa 20.0.7 / 20.1.0-rc4 regression, extremally long shader 45 compilation time in NIR 46- Mesa-git build fails on Fedora Rawhide 47- iris/i965: possible regression in 20.0.5 due to changes in buffer 48 manager sharing across screens (firefox/mozilla#1634213) 49- Incorrect \_NetBSD_\_ macro inside execmem.c 50- Possible invalid sizeof in device.c 51- mesa trunk master vulkan overlay-layer meson.build warning empty 52 configuration_data() object 53- 20.0.7: mesa still is not ready to gcc 10 default settings 54- [Gen9/icl] [Bisected] [Regression] 55 dEQP-GLES3.functional.shaders.loops.short_circuit.do_while_fragment 56 fail 57- Reproduceable i915 gpu hang Intel Iris Plus Graphics (Ice Lake 8x8 58 GT2) 59- Double lock in fbobject.c 60- [bisected] Steam crashes when newest Iris built with LTO 61- freedreno: glamor issue with x11 desktops 62- Deadlock in anv_timelines_wait() 63 64Changes 65------- 66 67Bas Nieuwenhuizen (3): 68 69- radv/winsys: Remove extra sizeof multiply. 70- radv: Handle failing to create .cache dir. 71- radv: Provide a better error for permission issues with priorities. 72 73D Scott Phillips (1): 74 75- anv/gen11+: Disable object level preemption 76 77Danylo Piliaiev (6): 78 79- anv: Translate relative timeout to absolute when calling 80 anv_timelines_wait 81- anv: Fix deadlock in anv_timelines_wait 82- meson: Disable GCC's dead store elimination for memory zeroing custom 83 new 84- mesa: Fix double-lock of Shared->FrameBuffers and usage of wrong 85 mutex 86- intel/fs: Work around dual-source blending hangs in combination with 87 SIMD16 88- glsl: inline functions with unsupported return type before converting 89 to nir 90 91Dave Airlie (1): 92 93- llvmpipe: compute shaders work better with all the threads. 94 95Dylan Baker (10): 96 97- docs/relnotes Add sha256 sums to 20.0.7 98- .pick_status.json: Update to ceae09da156309327d7ba6f4a59d3a2e9b8837d9 99- .pick_status.json: Update to a887ad7c84e14fdad7907037a39e9fee9d504bf3 100- .pick_status.json: Update to 4504d6374dbe2aa40af519c16765457bcbf81b84 101- .pick_status.json: Update to f0c102c075f8ac76629bb34619187262ccc3e9d8 102- tests: Make tests aware of meson test wrapper 103- .pick_status.json: Update to e58112bc08f99861ac634ede8db0f98cd497fc14 104- radonsi/si_state.c: retab 105- .pick_status.json: Update to 0795241dde1507e0c6a3f9ef07c281ad4f2acf7b 106- vulkan-overlay/meson: use install_data instead of configure_file 107 108Eric Engestrom (3): 109 110- tree-wide: fix deprecated GitLab URLs 111- glapi: remove deprecated .getchildren() that has been replace with an 112 iterator 113- intel: fix gen_sort_tags.py 114 115Erik Faye-Lund (2): 116 117- zink: use general-layout when blitting to/from same resource 118- nir: reuse existing psiz-variable 119 120Gert Wollny (1): 121 122- nir: lower_tex: Don't normalize coordinates for TXF with RECT 123 124Ian Romanick (1): 125 126- anv/tests: Don't rely on assert or changing NDEBUG in tests 127 128Ilia Mirkin (1): 129 130- nouveau: allow invalidating coherent/persistent buffer backings 131 132Jan Palus (1): 133 134- targets/opencl: fix build against LLVM>=10 with Polly support 135 136Jason Ekstrand (6): 137 138- anv:gpu_memcpy: Emit 3DSTATE_VF_INDEXING on Gen8+ 139- nir/lower_double_ops: Rework the if (progress) tree 140- nir/opt_deref: Report progress if we remove a deref 141- nir/copy_prop_vars: Record progress in more places 142- intel/vec4: Stomp the return type of RESINFO to UINT32 143- intel/fs: Fix unused texture coordinate zeroing on Gen4-5 144 145Jonathan Marek (1): 146 147- freedreno/a6xx: use nonbinning VS when GS is used 148 149Joshua Ashton (1): 150 151- radeonsi: Use TRUNC_COORD on samplers 152 153Lionel Landwerlin (4): 154 155- iris: fix BO destruction in error path 156- i965: don't forget to set screen on duped image 157- i965: fix export of GEM handles 158- iris: fix export of GEM handles 159 160Lucas Stach (1): 161 162- etnaviv: retarget transfer to render resource when necessary 163 164Marek Olšák (2): 165 166- radeonsi: don't expose 16xAA on chips with 1 RB due to an occlusion 167 query issue 168- radeonsi: add a hack to disable TRUNC_COORD for shadow samplers 169 170Marek Vasut (1): 171 172- etnaviv: Disable seamless cube map on GC880 173 174Michel Dänzer (1): 175 176- util: Change os_same_file_description return type from bool to int 177 178Nataraj Deshpande (1): 179 180- dri_util: Update internal_format to GL_RGB8 for 181 MESA_FORMAT_R8G8B8X8_UNORM 182 183Neha Bhende (1): 184 185- util: Initialize pipe_shader_state for passthrough and transform 186 shaders 187 188Pierre-Eric Pelloux-Prayer (1): 189 190- omx: fix build with gcc 10 191 192Rhys Perry (4): 193 194- nir: fix lowering to scratch with boolean access 195- aco: fix interaction with 3f branch workaround and p_constaddr 196- aco: check instruction format before waiting for a previous SMEM 197 store 198- aco: preserve more fields when combining additions into SMEM 199 200Rob Clark (1): 201 202- freedreno: clear last_fence after resource tracking 203 204Samuel Pitoiset (4): 205 206- spirv,radv,anv: implement no-op VK_GOOGLE_user_type 207- nir/lower_explicit_io: fix NON_UNIFORM access for UBO loads 208- radv: enable zero VRAM for Doom Eternal 209- radv: enable zero VRAM for all VKD3D (DX12->VK) games 210 211Timothy Arceri (3): 212 213- glsl: stop cascading errors if process_parameters() fails 214- radv: fix regression with builtin cache 215- glsl: fix potential slow compile times for GLSLOptimizeConservatively 216 217Vinson Lee (4): 218 219- zink: Check fopen result. 220- r300g: Remove extra printf format specifiers. 221- vdpau: Fix wrong calloc sizeof argument. 222- mesa: Fix NetBSD compiler macro. 223 224Yevhenii Kolesnikov (1): 225 226- intel/compiler: fix cmod propagation optimisations 227