Mesa 10.2.7 Release Notes / September 06, 2014

Mesa 10.2.7 is a bug fix release which fixes bugs found since the 10.2.6 release.

Mesa 10.2.7 implements the OpenGL 3.3 API, but the version reported by glGetString(GL_VERSION) or glGetIntegerv(GL_MAJOR_VERSION) / glGetIntegerv(GL_MINOR_VERSION) depends on the particular driver being used. Some drivers don’t support all the features required in OpenGL 3.3. OpenGL 3.3 is only available if requested at context creation because compatibility contexts are not supported.

SHA256 checksums

cb67dfaabf88acba29aa2cf0dd58ee17b21ebf9594f8d1226c41794da8de3e9d  MesaLib-10.2.7.tar.gz
27b958063a4c002071f14ed45c7d2a1ee52cd85e4ac8876e8a1c273495a7d43f  MesaLib-10.2.7.tar.bz2

New features


Bug fixes

This list is likely incomplete.

  • Bug 36193 - [i965] brw_eu_emit.c:182: validate_reg: Assertion `execsize >= width’ failed.

  • Bug 66184 - src/mesa/state_tracker/st_glsl_to_tgsi.cpp:3216:simplify_cmp: Assertion `inst->dst.index < 4096’ failed.

  • Bug 70441 - [Gen4-5 clip] Piglit spec_OpenGL_1.1_polygon-offset hits (execsize >= width) assertion

  • Bug 76188 - EGL_EXT_image_dma_buf_import fd ownership is incorrect

  • Bug 76789 - [radeonsi] si_descriptors.c requires -std=gnu99 or -fms-extensions

  • Bug 82139 - [r600g, bisected] multiple ubo piglit regressions

  • Bug 82255 - [VP2] Chroma planes are vertically stretched during VDPAU playback

  • Bug 82671 - [r600g-evergreen][compute]Empty kernel execution causes crash

  • Bug 82709 - OpenCL not working on radeon hainan

  • Bug 82814 - glDrawBuffers(0, NULL) segfaults in _mesa_drawbuffers

  • Bug 83079 - [NVC0] Dota 2 (Linux native and Wine) crash with Nouveau Drivers

  • Bug 83355 - FTBFS: src/mesa/program/program_lexer.l:122:64: error: unknown type name ‘YYSTYPE’


Adam Jackson (1):

  • radeonsi: Don’t use anonymous struct trick in atom tracking

Alex Deucher (2):

  • radeonsi: add new CIK pci ids

  • radeonsi: add new SI pci ids

Andreas Boll (1):

  • winsys/radeon: fix nop packet padding for hawaii

Anuj Phogat (1):

  • i965: Bail on vec4 copy propagation for scratch writes with source modifiers

Brian Paul (1):

  • mesa: fix NULL pointer deref bug in _mesa_drawbuffers()

Carl Worth (2):

  • docs: Add sha256 sums for the 10.2.6 release

  • Makefile: Switch from md5sums to sha256sums

Dave Airlie (1):

  • i965: add missing parens in vec4 visitor

Emil Velikov (17):

  • bail out if building gallium_gbm without gallium_egl

  • android: gallium/nouveau: fix include folders, link against libstlport

  • android: egl/main: fixup the nouveau build

  • automake: gallium/freedreno: drop spurious include dirs

  • android: gallium/freedreno: add preliminary build

  • android: egl/main: add/enable freedreno

  • android: gallium/auxiliary: drop log2/log2f redefitions

  • android: drop HAL_PIXEL_FORMAT_RGBA_{5551,4444}

  • android: glsl: the stlport over the limited Android STL

  • android: dri/i915: do not build an ‘empty’ driver

  • cherry-ignore: remove patch that lacking previous dependencies

  • cherry-ignore: PIPE_SHADER_CAP_MAX_CONST_BUFFER_SIZE is not it 10.2

  • cherry-ignore: drop whitespace fix

  • cherry-ignore: reject a15088338eb

  • Require explicit “10.2” for nominating stable patches

  • mesa: fix make tarballs

  • Update VERSION to 10.2.7

Ian Romanick (1):

  • mesa: Handle uninitialized textures like other textures in get_tex_level_parameter_image

Ilia Mirkin (9):

  • nouveau: make sure to invalidate any vbo state as well

  • nouveau: don’t keep stale pointer to free’d data

  • nvc0/ir: avoid infinite recursion when finding first uses of tex

  • nv50: zero out unbound samplers

  • nvc0: don’t make 1d staging textures linear

  • nv50/ir: avoid creating instructions that can’t be emitted

  • nv50: set the miptree address when clearing bo’s in vp2 init

  • nv50: mt address may not be the underlying bo’s start address

  • nv50: attach the buffer bo to the miptree structures

Jan Vesely (1):

  • gallivm: Fix build with latest LLVM

José Fonseca (1):

  • mesa: Move declaration to top of block.

Kenneth Graunke (3):

  • i965/vec4: Set NoMask for GS_OPCODE_SET_VERTEX_COUNT on Gen8+.

  • i965/vec4: Respect ir->force_writemask_all in Gen8 code generation.

  • i965/clip: Fix brw_clip_unfilled.c/compute_offset’s assembly.

Marek Olšák (3):

  • r600g: fix constant buffer fetches

  • radeonsi: save scissor state and sample mask for u_blitter

  • glsl_to_tgsi: allocate and enlarge arrays for temporaries on demand

Paulo Sergio Travaglia (2):

  • android: gallium/radeon: attempt to fix the android build

  • android: egl/main: resolve radeon linking issues

Pekka Paalanen (1):

  • egl_dri2: fix EXT_image_dma_buf_import fds

Robert Bragg (1):

  • meta: save and restore swizzle for _GenerateMipmap

Tom Stellard (7):

  • radeon/compute: Fix reported values for MAX_GLOBAL_SIZE and MAX_MEM_ALLOC_SIZE

  • radeonsi/compute: Update reference counts for buffers in si_set_global_binding()

  • radeonsi/compute: Call si_pm4_free_state() after emitting compute state

  • clover: Flush the command queue in clReleaseCommandQueue()

  • radeon: Add work-around for missing Hainan support in clang < 3.6 v2

  • pipe-loader: Fix memory leak v2

  • r600g/compute: Don’t initialize vertex_buffer_state masks to 0x2

Vinson Lee (1):

  • gallivm: Fix build with LLVM >= 3.6 r215967.