[Mesa-dev] Mesa 13.0.3 release candidate

Thu Dec 22 18:20:31 UTC 2016

Hello list,

The candidate for the Mesa 13.0.3 is now available. Currently we have:
 - 52 queued
 - 18 nominated (outstanding)
 - and 6 rejected patch(es)

With this series we have - a handful of fixes for i965 and radeonsi. The ANV
driver has gained a few improved smoke testing, less memory leaks and others.

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 13.0.3 this Friday (23th of December), around or shortly
after 20:00 GMT.

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

Trivial merge conflicts
-----------------------
commit 7f2ee55aacaf1aae80d276ef9b7a0b12cc1c71f1
Author: Timothy Arceri <timothy.arceri at collabora.com>

    mesa: fix active subroutine uniforms properly

    (cherry picked from commit 0303201dfb73c16751d5519cca7480fa678d429a)

commit 590366320d3f14ff6b677a5f9449a476715a6172
Author: Marek Olšák <marek.olsak at amd.com>

    radeonsi: apply a tessellation bug workaround for SI

    (cherry picked from commit 78c4528ae7709fbe94d917d034cfd60535b5dcf3)

Cheers,
Emil

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

Nominated (18)
==============

Chad Versace (2):
      42011be i965/mt: Disable HiZ when sharing depth buffer externally (v2)
Note: backport was needed, Chad sent one to the list.

      fbb4af9 egl: Check config's surface types in eglCreate*Surface()

Dave Airlie (1):
      9d23b8a radv: flush smem for uniform buffer bit.

Francisco Jerez (1):
      79d08ed anv: Fix uniform and storage buffer offset alignment limits.

Fredrik Höglund (1):
      27a8aab radv: fix dual source blending

Ian Romanick (9):
      8862fef glsl: Split process_block_array into two functions
      300de78 glsl: Fix wonkey indentation left from previous commit
      d56bd07 glsl: Track the linearized array index for each UBO
instance array element
      5085b64 glsl: Use simpler visitor to determine which UBO and
SSBO blocks are used
      b7053b8 glsl: Add tracking for elements of an array-of-arrays
that have been accessed
      8d499f6 glsl: Add structures to track accessed elements of a single array
      e929350 glsl: Mark a set of array elements as accessed using a
list of array_deref_range
      d329569 glsl: Walk a list of ir_dereference_array to mark array
elements as accessed
      ceea514 linker: Accurately mark a uniform block instance array
element as used in a stage

Kenneth Graunke (2):
      8fc5443 i965: Don't bail on vertex element processing if we need
draw params.
      62b8bcd glsl: Use ir_var_temporary when generating inline functions.
Note: late nomination

Michel Dänzer (1):
      3d661a1 cso: Don't restore nr_samplers in cso_restore_fragment_samplers

Timothy Arceri (1):
      3421b3f nir: Turn imov/fmov of undef into undef

Queued (52)
===========

Chad Versace (2):
      i965/mt: Disable aux surfaces after making miptree shareable
      egl: Fix crashes in eglCreate*Surface()

Dave Airlie (4):
      anv: set maxFragmentDualSrcAttachments to 1
      radv: set maxFragmentDualSrcAttachments to 1
      radv: fix another regression since shadow fixes.
      radv: add missing license file to radv_meta_bufimage.

Emil Velikov (3):
      anv: don't double-close the same fd
      anv: don't leak memory if anv_init_wsi() fails
      radv: don't leak the fd if radv_physical_device_init() succeeds

Eric Anholt (1):
      vc4: In a loop break/continue, jump if everyone has taken the path.

Gwan-gyeong Mun (3):
      anv: Add missing error-checking to anv_block_pool_init (v2)
      anv: Update the teardown in reverse order of the anv_CreateDevice
      vulkan/wsi: Fix resource leak in success path of wsi_queue_init()

Haixia Shi (1):
      compiler/glsl: fix precision problem of tanh

Ilia Mirkin (1):
      mesa: only verify that enabled arrays have backing buffers

Jason Ekstrand (8):
      anv/cmd_buffer: Re-emit MEDIA_CURBE_LOAD when CS push constants are dirty
      anv/image: Rename hiz_surface to aux_surface
      anv/cmd_buffer: Remove the 1-D case from the HiZ QPitch calculation
      genxml/gen9: Change the default of MI_SEMAPHORE_WAIT::RegisterPoleMode
      anv/device: Return the right error for failed maps
      anv/device: Implicitly unmap memory objects in FreeMemory
      anv/descriptor_set: Write the state offset in the surface state free list.
      spirv: Use a simpler and more correct implementaiton of tanh()

Kenneth Graunke (1):
      i965: Allocate at least some URB space even when max_vertices = 0.

Marek Olšák (17):
      radeonsi: always set all blend registers
      radeonsi: set CB_BLEND1_CONTROL.ENABLE for dual source blending
      radeonsi: disable RB+ blend optimizations for dual source blending
      radeonsi: consolidate max-work-group-size computation
      radeonsi: apply a multi-wave workgroup SPI bug workaround to
affected CIK chips
      radeonsi: apply a TC L1 write corruption workaround for SI
      radeonsi: apply a tessellation bug workaround for SI
      radeonsi: add a tess+GS hang workaround for VI dGPUs
      radeonsi: apply the double EVENT_WRITE_EOP workaround to VI as well
      cso: don't release sampler states that are bound
      radeonsi: always restore sampler states when unbinding sampler views
      radeonsi: fix incorrect FMASK checking in bind_sampler_states
      radeonsi: allow specifying simm16 of emit_waitcnt at call sites
      radeonsi: wait for outstanding memory instructions in TCS barriers
      tgsi: fix the src type of TGSI_OPCODE_MEMBAR
      radeonsi: wait for outstanding LDS instructions in memory
barriers if needed
      radeonsi: disable the constant engine (CE) on Carrizo and Stoney

Matt Turner (3):
      i965/fs: Rename opt_copy_propagate -> opt_copy_propagation.
      i965/fs: Add unit tests for copy propagation pass.
      i965/fs: Reject copy propagation into SEL if not min/max.

Nanley Chery (1):
      mesa/fbobject: Update CubeMapFace when reusing textures

Nicolai Hähnle (4):
      radeonsi: fix isolines tess factor writes to control ring
      radeonsi: update all GSVS ring descriptors for new buffer allocations
      radeonsi: do not kill GS with memory writes
      radeonsi: fix an off-by-one error in the bounds check for max_vertices

Rhys Kidd (1):
      glsl: Add pthread libs to cache_test

Timothy Arceri (2):
      mesa: fix active subroutine uniforms properly
      Revert "nir: Turn imov/fmov of undef into undef."

Rejected (6)
============

Jason Ekstrand (6):
      174f490 i965/blit: Remove a bogus assertion
      80d3af8 i965/blit: Break blits into chunks in intel_miptree_blit
Note: not nominated, yet required by the below.

      6c74e7f i965/blit: Break the guts of intel_miptree_blit into a helper
      06d8649 i965/copy_image: Re-implement the blitter path with
emit_miptree_blit
      157971e i965/blit: Fix the src dimension sanity check in miptree_copy
      b18cd8c i965/miptree: Use intel_miptree_copy for maps
Reason: The patches are very invasive which is indicated by the bug introduced
with 06d8649 and fixed with 157971e.

Note: Might reconsider bringing these on, if we have noticeable usecases where
these are required.