[Mesa-dev] Mesa 17.0.6 release candidate

Wed May 10 23:20:14 UTC 2017

Hello list,

The candidate for the Mesa 17.0.6 is now available. Currently we have:
 - 57 queued
 - 0 nominated (outstanding)
 - and 4 rejected patch(es)

The current queue contains fixes in several places.

In Mesa Core we include a fix for validating that SSA references use
the same number of channels as there are in the SSA value and a minor
fix for the YUV color conversion matrix.

Gallivm has received a fix when setting bool values in Big Endian
architectures.

The state tracker has gotten some fixes regarding incomplete
framebuffers.

Several fixes, including some leak fixes and have been added to
renderonly.

radv has now support enabled for POLARIS12 and reports presentation
error per image request.

radeonsi has seen adjusted the ESGS ring buffer size computation on VI
while, on SI, it has gotten a fix for the gl_PrimitiveID in
tessellation with instanced draws. Additionally, a hang workaround has
been applied to POLARIS12 too.

L8A8_UNORM texture format has been added for etnaviv and freedreno got
a fix for a hang with large render targets on devices with small gmem.

Some more fixes have made it for Intel's Vulkan driver including some
corrections when returning error codes, while i965 has gotten several
fixes dealing with DF.

EGL has seen fixed a segfault and, now emits an error when EGLSurface
is lost. Also, in Android, there has been included a fix to prevent a
deadlock and yet another to ensure the emission of an error when
needed.

Finally, in the building front, DRI modules now always link against
shared glapi. The travis continuous integration has received many
updates to be able to follow the changes in the building of stable
branch closer and Scons has been updated for LLVM 4.0.

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 17.0.6 this Friday (12th of May), around or
shortly after 23:00 GMT.

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

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

commit db815dc91b84ea3c907d05da9588e9bae87fa0e3
Author: Marek Olšák <mar
ek.olsak at amd.com>

    radeonsi: apply the tess+GS hang workaround to Polaris12 as well

    (cherry picked from commit ee5908396e4b0d4be64d8bc486d25c2e95b7fd71)

commit d0467f63bf7279df29d159183ee4619efc848c63
Author: Dave Airlie <airlied at redhat.com>

    radv: enable POLARIS12 support.

    (cherry picked from commit a096d8d3f7a8c6dffc91b2ac1794fdd215c054d3)

commit ff4d6de5c772ef9de6be62373ca74e36ac4506d2
Author: Dave Airlie <airlied at redhat.com>

    radv/wsi: report presentation error per image request

    (cherry picked from commit 09034aab64c7a6022a2c508658fead1442f08576)

commit afa2b097c047c86ff12a49e5ffdf54dde95c714a
Author: Jason Ekstrand <jason.ekstrand at intel.com>

    anv/cmd_buffer: Use the device allocator for QueueSubmit

    (cherry picked from commit bd3a9813b92bd2e116b58f0932bc7f1f722a9f63)

commit e38b645c48cc726e2e99e07ff600250e6cfaf6db
Author: Samuel Iglesias Gonsálvez <siglesias at igalia.com>

    i965/vec4: don't modify regioning parameters to the sources of DF align1 instructions

    (cherry picked from commit f57e234fdd52331d0aa6656a36efdebea9d11e9d)

commit 660868b6291ebd01855ad29269db51677494e4f1
Author: Samuel Iglesias Gonsálvez <siglesias at igalia.com>

    i965/vec4: fix vertical stride to avoid breaking region parameter rule

    (cherry picked from commit 7f728bce811fc283e672e3a07b008bb7b52de35e)

commit a4043bb5c60c402ba694ee481a17b72ed9459dc8
Author: Nicolai Hähnle <nicolai.haehnle at amd.com>

    st/mesa: remove redundant stfb->iface checks

    (cherry picked from commit 9d346af322e3f9d6d52e00759e8c3cd37c192004)

commit 997d81e5375a6447f93b6e76183e9846e3775cb6
Author: Emil Velikov <emil.velikov at collabora.com>

    travis: enable wayland support

    (cherry picked from commit 27a0b383b9edbacb177553685272408945ec630c)

commit 629d623685908db2f96acf5a10c05d20bf15bcff
Author: Emil Velikov <emil.velikov at collabora.com>

    travis: add Gallium state-tracker targets

    (cherry picked from commit 0e6a36cd3f2db7208e85864d4d552f8575200fbe)

commit 79cca4bc71135fefc246d7dab41f14af49381c88
Author: Emil Velikov <emil.velikov at collabora.com>

    travis: split the make target to three separate ones

    (cherry picked from commit 7e2af374742f6f387cada8c3e1e8585be3831405)

commit 9e8629d52005fb6ea8da68b6374961fb500235a0
Author: Emil Velikov <emil.velikov at collabora.com>

    travis: add "make swr" to the build matrix

    (cherry picked from commit 8479fd8a1099c4bf37fb8cded7a3040b4d366fbc)

Cheers,
Andres

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

Nominated (0)
=============

Queued (38)
===========

Adam Jackson (1):
      egl/platform/drm: Don't take display ownership
until gbm is initialized

Andres Gomez (6):
      docs: add sha256 checksums for 17.0.5
      travis: replace Trusty-based LLVM toolchain apt-get with apt addon
      travis: add the possibility of using the txc-dxtn library
      cherry-ignore: 17.1 nominations only
      cherry-ignore: fix regression in descriptor set freeing.
      cherry-ignore: rejected commits

Ben Boeckel (1):
      scons: update for LLVM 4.0

Brian Paul (1):
      st/mesa: move duplicated st_ws_framebuffer() function into header file

Chad Versace (3):
      egl: Emit error when EGLSurface is lost
      egl/android: Cancel any outstanding ANativeBuffer in surface destructor
      egl/android: Mark surface as lost when dequeueBuffer fails

Christian Gmeiner (1):
      etnaviv: add L8A8_UNORM texture format

Dave Airlie (2):
      radv/wsi: report presentation error per image request
      radv: enable POLARIS12 support.

Emil Velikov (21):
      travis: correct libdrm required regex to also track libdrm itself
      travis: add nearly all gallium drivers to the list
      travis: use both cores for make/make check
      travis: bring the scons build on par with AppVeyor
      travis: explicitly LD_LIBRARY_PATH the local libraries
      travis: enable apt cache
      travis: automatically manage ccache caching
      travis: remove unused -dev packages
      travis: rework "if test" blocks in the script section
      travis: split out matrix from env
      travis: add separate "scons" and "scons llvm" targets
      travis: add "scons swr" to the build matrix
      travis: add "make swr" to the build matrix
      travis: split the make target to three separate ones
      travis: model scons check target like the make one
      travis: add Gallium state-tracker targets
      travis: enable wayland support
      travis: bump MAKEFLAGS to -j4
      gallium/dri: always link against shared glapi
      mesa/dri: always link against shared glapi
      glx: glX_proto_send.py: use correct compile guard GLX_INDIRECT_RENDERING

Eric Anholt (1):
      nir: Pick just the channels we want for bitmap and drawpixels lowering.

Ilia Mirkin (1):
      gallium/targets: fix bool setting on BE architectures

Jason Ekstrand (1):
      anv/cmd_buffer: Use the device allocator for QueueSubmit

Johnson Lin (1):
      nir/lower_tex: Fix minor error in YUV color conversion matrix

Marek Olšák (2):
      radeonsi: adjust ESGS ring buffer size computation on VI
      radeonsi: apply the tess+GS hang workaround to Polaris12 as well

Nicolai Hähnle (3):
      st/mesa: don't cast the incomplete framebufer to st_framebuffer
      st/mesa: remove redundant stfb->iface checks
      radeonsi: fix gl_PrimitiveID in tessellation with instanced draws on SI

Philipp Zabel (3):
      renderonly: close transfer prime_fd
      renderonly: drop resources on destroy
      renderonly: use drmIoctl

Rhys Kidd (3):
      travis: Support LLVM 3.8+ on Trusty-based Travis-CI via apt-get not apt addon
      travis: Add radv vulkan driver to continuous integration
      travis: Add radeonsi to continuous integration

Rob Clark (1):
      freedreno/a3xx: fix hang w/ large render targets and small gmem

Samuel Iglesias Gonsálvez (5):
      i965/vec4: fix vertical stride to avoid breaking region parameter rule
      i965/vec4: fix register width for DF VGRF and UNIFORM
      i965/vec4: don't modify regioning parameters to the sources of DF align1 instructions
      anv: anv_gem_mmap() returns MAP_FAILED as mapping error
      anv: vkBindImageMemory() should return VK_ERROR_OUT_OF_{HOST,DEVICE}_MEMORY on failure

Rejected (4)
============

Eric Anholt (1):
      vc4: Only build the NEON code on arm32.

Fixes earlier commit 4d30024238e which did not land in branch.

Randy Xu (1):
      i965: Solve Android native fence fd double close

Fixes earlier commit 6403e376511 which did not land in branch.

Nicolai Hähnle (2):
      radeonsi: fix primitive ID in fragment shader when using tessellation
      radeonsi: load patch_id for TES-as-ES when exporting for PS

Commits depend on 5f99c490089 and 9fd9a7d0ba3, respectively, that did
not land in branch.