[Mesa-dev] Mesa 17.0.3 release candidate

Andres Gomez agomez at igalia.com
Thu Mar 30 15:09:43 UTC 2017


Hello list,

The candidate for the Mesa 17.0.3 is now available. Currently we have:
 - 32 queued
 - 1 nominated (outstanding)
 - and 2 rejected patch(es)


The current queue consists of fixes in several fronts.

There are several patches addressing improvements in the API validation
and the GLSL compiler

We have several patches for the Intel drivers (both GL and Vulkan),
including a fix for a hang with GfxBench 4.0 CarChase, several
improvements in the Vulkan driver and other crash fixes.

Galleon drivers have seen fixed an existing deadlock too.

For radeonsi, we now include the polaris12 pci id and a fix for a hang
after a shader compile failure.

nouveau is benefiting now from performance improvements, specially for
the Feral-ported games and some other fixes.

radv, freedreno and clover include some few fixes too.

On integration side - we had a swr build fix with llvm >= 5.0 and also
a fix for building with MSVC.

For those of you wondering - the delay was caused mainly by a
miscommunication on my size while checking the proposed fix involving
the use of the v4 for DRI2's flush extension.


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 17.0.3 this Saturday (1st of April), around or
shortly after 15:00 GMT.

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


Trivial merge conflicts
-----------------------

commit d46506ac2289c7d13493ffc9fbb43513a66b3139
Author: Jason Ekstrand <jason.ekstrand at intel.com>

    anv/query: Invalidate the correct range

    (cherry picked from commit 81840130c0f147ed6ae4c26872c2f04a2167bc54)

commit a6616a84d87699669961cb256b8c2feec3be86f5
Author: Jason Ekstrand <jason.ekstrand at intel.com>

    anv/GetQueryPoolResults: Actually implement the spec

    (cherry picked from commit 08df015b9de8ccb16ce6db93890910f8a02be4c6)

commit 3218aed24d3adcd7e9dd015c8c5751c492ff8a46
Author: Bas Nieuwenhuizen <bas at basnieuwenhuizen.nl>

    radv: flush DB cache before and after HTILE decompress.

    (cherry picked from commit a8c51b1cd9168b621e27cf5308d0dd8fc08f8a4a)

commit eb6372c0b93d575eecbc3f7f8c6f0ec700c15041
Author: Jason Ekstrand <jason.ekstrand at intel.com>

    anv/query: Fix the location of timestamp availability

    (cherry picked from commit 4bbb4b95b8ba02693f5e6990b983ebb66dc6241a)

commit 6cabd40211413bbf780fa61949da7dde47608792
Author: Jason Ekstrand <jason.ekstrand at intel.com>

    anv: Make anv_get_layerCount a macro

    (cherry picked from commit 1b8fa8dd794c22aba43b16470e75ecaebf902b11)

commit 9973db02a04621598bf504fcc7797522ee5046de
Author: Jason Ekstrand <jason.ekstrand at intel.com>

    anv/cmd_buffer: Apply flush operations prior to executing secondaries

    (cherry picked from commit 01a65dc43be3a4bf6b8a901586f7222218f4b6b3)


Cheers,
Andres


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

Nominated (1)
=============

Craig Stout (1):
      anv/cmd_buffer: fix host memory leak


Queued (32)
===========

Alex Deucher (1):
      radeonsi: add new polaris12 pci id

Andres Gomez (1):
      glsl: on UBO/SSBOs link error reset the number of active blocks to 0

Axel Davy (2):
      st/nine: Resolve deadlock in surface/volume dtors when using csmt
      st/nine: Use atomics for available_texture_mem

Bas Nieuwenhuizen (1):
      radv: flush DB cache before and after HTILE decompress.

Dave Airlie (1):
      radv: fix primitive reset index emission

Ilia Mirkin (1):
      st/mesa: set result writemask based on ir type

Jan Vesely (1):
      clover: use pipe_resource references

Jason Ekstrand (9):
      anv/query: Invalidate the correct range
      anv/GetQueryPoolResults: Actually implement the spec
      anv/image: Return early when unbinding an image
      anv/query: Fix the location of timestamp availability
      anv: Make anv_get_layerCount a macro
      anv/blorp: Use anv_get_layerCount everywhere
      anv/cmd_buffer: Apply flush operations prior to executing secondaries
      anv/cmd_buffer: Fix bad indentation
      anv: Flush caches prior to PIPELINE_SELECT on all gens

José Fonseca (1):
      c11/threads: Include thr/xtimec.h for xtime definition when building with MSVC.

Juan A. Suarez Romero (1):
      tests/cache_test: allow crossing mount points

Karol Herbst (1):
      nvc0/ir: treat FMA like MAD for operand propagation

Kenneth Graunke (1):
      i965: Fall back to GL 4.2/4.3 on Haswell if the kernel isn't new enough.

Marek Olšák (1):
      radeonsi: don't hang on shader compile failure

Matt Turner (1):
      i965/fs: Don't emit SEL instructions for type-converting MOVs.

Nanley Chery (1):
      intel: Correct the BDW surface state size

Nicolai Hähnle (1):
      mesa/main: fix MultiDrawElements[BaseVertex] validation of primcount

Rob Clark (1):
      freedreno: fix memory leak

Tim Rowley (1):
      swr: [rasterizer jitter] fix llvm >= 5.0 build break

Timothy Arceri (2):
      glsl: fix lower jumps for returns when loop is inside an if
      mesa: update lower_jumps tests after bug fix

Topi Pohjolainen (1):
      i965/gen8+: Do full stall when switching pipeline

Xu Randy (2):
      anv/blorp: Fix a crash in CmdClearColorImage
      anv/genX: Solve the vkCreateGraphicsPipelines crash


Rejected (2)
============

Alex Smith (1):
      radv: Invalidate L2 for TRANSFER_WRITE barriers

The commit addressed an earlier one [0567ab0407e] which did not land in
branch. A stable specific backported patch will follow later ...

Thomas Hellstrom (1):
      gbm/dri: Flush after unmap

The commit caused a regression in i965 (and possibly others) since it
didn't implement v4 of DRI2's flush extension.



More information about the mesa-dev mailing list