[Mesa-dev] Mesa 13.0.2 release candidate

Thu Nov 24 17:25:26 UTC 2016

Hello list,

The candidate for the Mesa 13.0.2 is now available. Currently we have:
 - 49 queued
 - 4 nominated (outstanding)
 - and 1 rejected patch(es)

With this series we have - fixes for vc4, i965 and radeon drivers. In addition
to that PCI IDs for Geminilake have been added to the i965 driver,

The respective Vulkan drivers have seen multiple improvements some of which
include improved smoketesting and addressed memory leaks.

Races during _mesa_HashWalk() (while using glDeleteFramebuffers alongside
glTexImage2D) and "#version 0" in GLSL programs have been addressed.

BSD and Hurd users should be above to build the latest code as we no longer
use PATH_MAX.

Take a look at section "Mesa stable queue" for more information.

Testing reports/general approval
--------------------------------
Any testing reports (or general approval of the state of the branch) will be
greatly appreciated.

The plan is to have 13.0.2 this Saturday (25th of November), around or
shortly after 17:00 GMT.

If you have any questions or suggestions - be that about the current patch
queue or otherwise, please go ahead.

Trivial merge conflicts
-----------------------
commit 9d5c3fc12b05d944508ef4e3b1f2ddc4f23c0a82
Author: Jason Ekstrand <jason.ekstrand at intel.com>

    i965/gs: Allow primitive id to be a system value

    (cherry picked from commit a5e88e66e633aaeb587b274d80e21cd46c8ee2cb)

commit 8dbdbc21910a6d37c381535186f9e728fff8690d
Author: Jason Ekstrand <jason.ekstrand at intel.com>

    anv: Handle null in all destructors

    (cherry picked from commit 49f08ad77f51cc344e4bfe60ba9f8d9fccfbd753)

commit 1809f17bda56d4f9d6385f63a9c4a5df890e3cad
Author: Kenneth Graunke <kenneth at whitecape.org>

    mesa: Drop PATH_MAX usage.

    (cherry picked from commit 9bfee7047b70cb0aa026ca9536465762f96cb2b1)

Cheers,
Emil

Mesa stable queue
-----------------

Nominated (4)
=============

Dave Airlie (1):
      2de85eb radv: fix texturesamples to handle single sample case

Jason Ekstrand (2):
      e73d136 vulkan/wsi/x11: Implement FIFO mode.

Note: temporary on hold.
Commit seems to be a feature and provides no clear indication about the bugs
it addresses. Jason, is this really applicable for stable ?

      054e48e anv/cmd_buffer: Re-emit MEDIA_CURBE_LOAD when CS push
constants are dirty

Note: temporary on hold.
Depends on the refactoring d33e2ad67c3 (anv: Move INTERFACE_DESCRIPTOR_DATA
setup to the pipeline) which in itself depends on another refactoring (cleanup)
commit 623e1e06d8c and likely others. Jason, any input ?

Kevin Strasser (1):
      932bb3f vulkan/wsi: Add a thread-safe queue implementation
Requirement for "Implement FIFO mode above" above.

Queued (49)
===========

Ben Widawsky (3):
      i965: Add some APL and KBL SKU strings
      i965: Reorder PCI ID list to match release order
      i965/glk: Add basic Geminilake support

Dave Airlie (7):
      radv: fix texturesamples to handle single sample case
      wsi: fix VK_INCOMPLETE for vkGetSwapchainImagesKHR
      radv: don't crash on null swapchain destroy.
      ac/nir/llvm: fix channel in texture gather lowering code.
      radv: make sure to flush input attachments correctly.
      radv: fix image view creation for depth and stencil only
      radv: spir-v allows texture size query with and without lod.

Eduardo Lima Mitev (2):
      vulkan/wsi/x11: Fix behavior of vkGetPhysicalDeviceSurfaceFormatsKHR
      vulkan/wsi/x11: Fix behavior of vkGetPhysicalDeviceSurfacePresentModesKHR

Eduardo, I've picked these two since they are perfectly reasonable
stable material.
Let me know if you think that we should keep or drop them.

Emil Velikov (3):
      docs: add sha256 checksums for 13.0.1
      cherry-ignore: add reverted LLVM_LIBDIR patch
      anv: fix enumeration of properties

Eric Anholt (3):
      vc4: Don't abort when a shader compile fails.
      vc4: Clamp the shadow comparison value.
      vc4: Fix register class handling of DDX/DDY arguments.

Gwan-gyeong Mun (2):
      util/disk_cache: close a previously opened handle in disk_cache_put (v2)
      anv: Fix unintentional integer overflow in anv_CreateDmaBufImageINTEL

Iago Toral Quiroga (1):
      anv/format: handle unsupported formats properly

Ian Romanick (2):
      glcpp: Handle '#version 0' and other invalid values
      glsl: Parse 0 as a preprocessor INTCONSTANT

Jason Ekstrand (14):
      anv/gen8: Stall when needed in Cmd(Set|Reset)Event
      anv/wsi: Set the fence to signaled in AcquireNextImageKHR
      anv: Rework fences
      vulkan/wsi/wayland: Include pthread.h
      vulkan/wsi/wayland: Clean up some error handling paths
      vulkan/wsi: Report the correct min/maxImageCount
      i965/gs: Allow primitive id to be a system value
      anv: Handle null in all destructors
      anv/fence: Handle ANV_FENCE_CREATE_SIGNALED_BIT
      nir/spirv: Fix handling of gl_PrimitiveId
      anv/blorp: Ignore clears for attachments first used as resolve
destinations
      anv: Implement a depth stall restriction on gen7
      anv/cmd_buffer: Handle running out of binding tables in compute shaders
      anv/cmd_buffer: Emit a CS stall before setting a CS pipeline

Jordan Justen (2):
      isl: Fix height calculation in isl_msaa_interleaved_scale_px_to_sa
      i965/hsw: Set integer mode in sampling state for stencil texturing

Kenneth Graunke (4):
      intel: Set min_ds_entries on Broxton.
      i965: Fix compute shader crash.
      mesa: Drop PATH_MAX usage.
      i965: Fix GS push inputs with enhanced layouts.

Lionel Landwerlin (1):
      anv: fix multi level clears with VK_REMAINING_MIP_LEVELS

Lucas Stach (1):
      gbm: request correct version of the DRI2_FENCE extension

Nicolai Hähnle (2):
      radeonsi: store group_size_variable in struct si_compute
      glsl/lower_output_reads: fix geometry shader output handling
with conditional emit

Steinar H. Gunderson (1):
      Fix races during _mesa_HashWalk().

Tapani Pälli (1):
      mesa: fix empty program log length

Rejected (1)
============

Emil Velikov (1):
      a39ad18 configure.ac: honour LLVM_LIBDIR when linking against LLVM

Reason: Patch is wrong/incomplete and was reverted in master.