Mesa 17.2.3 Release Notes / October 19, 2017
Mesa 17.2.3 is a bug fix release which fixes bugs found since the 17.2.2 release.
Mesa 17.2.3 implements the OpenGL 4.5 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 4.5. OpenGL 4.5 is only available if requested at context creation because compatibility contexts are not supported.
SHA256 checksums
fb305eecfeec1fd771fdc96fff973c51871f7bd35fd2bd56cacc27b4b8823220 mesa-17.2.3.tar.gz a0b0ec8f7b24dd044d7ab30a8c7e6d3767521e245f88d4ed5dd93315dc56f837 mesa-17.2.3.tar.xz
New features
None
Bug fixes
- Bug 101832 - [PATCH][regression][bisect] Xorg fails to start after f50aa21456d82c8cb6fbaa565835f1acc1720a5d
- Bug 102852 - Scons: Support the new Scons 3.0.0
- Bug 102940 - Regression: Vulkan KMS rendering crashes since 17.2
Changes
Alex Smith (1):
- radv: Add R16G16B16A16_SNORM fast clear support
Bas Nieuwenhuizen (2):
- nir/spirv: Allow loop breaks in a switch body.
- radv: Only set the MTYPE flags on GFX9+.
Ben Crocker (4):
- gallivm: fix typo in debug_printf message
- gallivm: allow additional llc options
- gallivm/ppc64le: adjust VSX code generation control.
- gallivm/ppc64le: allow environmental control of Altivec code generation
Daniel Stone (2):
- egl/wayland: Check queryImage return for wl_buffer
- egl/wayland: Don't use dmabuf with no modifiers
Dave Airlie (2):
- radv: emit fmuladd instead of fma to llvm.
- radv: lower ffma in nir.
Emil Velikov (6):
- cherry-ignore: add "anv: Remove unreachable cases from isl_format_for_size"
- cherry-ignore: add "anv/wsi: Allocate enough memory for the entire image"
- swr/rast: do not crash on NULL strings returned by getenv
- wayland-drm: use a copy of the wayland_drm_callbacks struct
- eglmesaext: add forward declaration for struct wl_buffers
- Update version to 17.2.3
Eric Engestrom (1):
- scons: use python3-compatible print()
Ilia Mirkin (2):
- nv50/ir: fix 64-bit integer shifts
- nv50,nvc0: fix push hint logic in presence of a start offset
Jason Ekstrand (6):
- intel/compiler: Don't cmod propagate into a saturated operation
- intel/compiler: Don't propagate cmod into integer multiplies
- glsl/blob: Return false from ensure_can_read on overrun
- glsl/blob: Return false from grow_to_fit if we've ever failed
- nir/opcodes: Fix constant-folding of ufind_msb
- nir: Get rid of the variable on vote intrinsics
Juan A. Suarez Romero (1):
- docs: add sha256 checksums for 17.2.2
Józef Kucia (3):
- anv: Fix vkCmdFillBuffer()
- spirv: Fix SpvOpAtomicISub
- anv: Do not assert() on VK_ATTACHMENT_UNUSED
Leo Liu (3):
- st/va: use pipe transfer_map to map upload buffer
- st/vdpau: don't re-allocate interlaced buffer with packed YUV format
- st/va: don't re-allocate interlaced buffer with pakced format
Lionel Landwerlin (4):
- intel: compiler: vec4: add missing default 0 lod
- anv/cmd_buffer: fix push descriptors with set > 0
- anv/cmd_buffer: Reset state in cmd_buffer_destroy
- anv: bo_cache: allow importing a BO larger than needed
Marek Olšák (3):
- mesa: fix texture updates for ATI_fragment_shader
- st/mesa: don't use pipe_surface for passing information about EGLImage
- glsl_to_tgsi: fix instruction order for bindless textures
Nicolai Hähnle (14):
- st/glsl_to_tgsi: fix conditional assignments to packed shader outputs
- amd/common: fix build_cube_select
- radeonsi/gfx9: fix geometry shaders without output vertices
- util/queue: fix a race condition in the fence code
- glsl/lower_instruction: handle denorms and overflow in ldexp correctly
- radeonsi: move current_rast_prim to r600_common_context
- radeonsi: don't discard points and lines
- radeonsi: deduce rast_prim correctly for tessellation point mode
- radeonsi: fix maximum advertised point size / line width
- st/mesa: don't clobber glGetInternalformat* buffer for GL_NUM_SAMPLE_COUNTS
- st/glsl_to_tgsi: fix indirect access to 64-bit integer
- st/glsl_to_tgsi: fix a use-after-free in merge_two_dsts
- radeonsi: clamp depth comparison value only for fixed point formats
- radeonsi: clamp border colors for upgraded depth textures
Rob Clark (2):
- freedreno/a5xx: align height to GMEM
- freedreno/a5xx: fix missing restore state