mesa.git
6 years agoradeonsi: if there's just const buffer 0, set it in place of CONST/SSBO pointer
Marek Olšák [Sun, 8 Oct 2017 01:44:07 +0000 (03:44 +0200)]
radeonsi: if there's just const buffer 0, set it in place of CONST/SSBO pointer

SI_SGPR_CONST_AND_SHADER_BUFFERS now contains the pointer to const buffer 0
if there is no other buffer there.

Benefits:
- there is no constbuf descriptor upload and shader load

It's assumed that all constant addresses are within bounds. Non-constant
addresses are clamped against the last declared CONST variable.
This only works if the state tracker ensures the bound constant buffer
matches what the shader needs.

Once we get 32-bit pointers, we can only do this for user constant buffers
where the driver is in charge of the upload so that it can guarantee a 32-bit
address.

The real performance benefit might not be measurable.

These apps get 100% theoretical benefit in all shaders (except where noted):
- antichamber
- barman arkham origins
- borderlands 2
- borderlands pre-sequel
- brutal legend
- civilization BE
- CS:GO
- deadcore
- dota 2 -- most shaders
- europa universalis
- grid autosport -- most shaders
- left 4 dead 2
- legend of grimrock
- life is strange
- payday 2
- portal
- rocket league
- serious sam 3 bfe
- talos principle
- team fortress 2
- thea
- unigine heaven
- unigine valley -- also sanctuary and tropics
- wasteland 2
- xcom: enemy unknown & enemy within
- tesseract
- unity (engine)

Changed stats only:
    SGPRS: 2059998 -> 2086238 (1.27 %)
    VGPRS: 1626888 -> 1626904 (0.00 %)
    Spilled SGPRs: 7902 -> 7865 (-0.47 %)
    Code Size: 60924520 -> 60982660 (0.10 %) bytes
    Max Waves: 374539 -> 374526 (-0.00 %)

Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
6 years agoac: clean up ac_build_indexed_load function interfaces
Marek Olšák [Sun, 8 Oct 2017 18:05:44 +0000 (20:05 +0200)]
ac: clean up ac_build_indexed_load function interfaces

Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
6 years agoradeonsi: handle 64-bit loads earlier in fetch_constant
Marek Olšák [Sun, 8 Oct 2017 16:20:38 +0000 (18:20 +0200)]
radeonsi: handle 64-bit loads earlier in fetch_constant

Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
6 years agoradeonsi: add si_descriptors::gpu_address and remove buffer_offset
Marek Olšák [Sun, 8 Oct 2017 15:52:57 +0000 (17:52 +0200)]
radeonsi: add si_descriptors::gpu_address and remove buffer_offset

This allows us to change the pointer arbitrarily.

Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
6 years agoradeonsi: unify code for extracting a buffer address from a descriptor
Marek Olšák [Sun, 8 Oct 2017 15:50:31 +0000 (17:50 +0200)]
radeonsi: unify code for extracting a buffer address from a descriptor

Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
6 years agoradeonsi: remove atom parameter from si_upload_descriptors
Marek Olšák [Sun, 8 Oct 2017 15:22:20 +0000 (17:22 +0200)]
radeonsi: remove atom parameter from si_upload_descriptors

Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
6 years agoradeonsi: pack si_descriptors better again
Marek Olšák [Sun, 8 Oct 2017 15:10:56 +0000 (17:10 +0200)]
radeonsi: pack si_descriptors better again

Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
6 years agoradeonsi: emit dirty consecutive pointers in one SET_SH_REG packet
Marek Olšák [Sat, 7 Oct 2017 20:23:34 +0000 (22:23 +0200)]
radeonsi: emit dirty consecutive pointers in one SET_SH_REG packet

IB size: -1.6%

Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
6 years agoradeonsi: split si_emit_shader_pointer
Marek Olšák [Sat, 7 Oct 2017 20:16:46 +0000 (22:16 +0200)]
radeonsi: split si_emit_shader_pointer

Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
6 years agoradeonsi: generalize the SI_VS_SHADER_POINTER_MASK macro
Marek Olšák [Sat, 7 Oct 2017 19:53:10 +0000 (21:53 +0200)]
radeonsi: generalize the SI_VS_SHADER_POINTER_MASK macro

Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
6 years agoradeonsi/gfx9: use SPI_SHADER_USER_DATA_COMMON
Marek Olšák [Sat, 7 Oct 2017 19:34:03 +0000 (21:34 +0200)]
radeonsi/gfx9: use SPI_SHADER_USER_DATA_COMMON

IB size: -0.4%

Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
6 years agoradeonsi/gfx9: move RW_BUFFERS from s[0:1] to s[8:9] for HS and GS
Marek Olšák [Sat, 7 Oct 2017 19:26:14 +0000 (21:26 +0200)]
radeonsi/gfx9: move RW_BUFFERS from s[0:1] to s[8:9] for HS and GS

Let's use the same user data SGPRs in all stages.
(for SPI_SHADER_USER_DATA_COMMON_0)

Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
6 years agoradeonsi: add GFX-IB-size query to the HUD
Marek Olšák [Sat, 7 Oct 2017 22:35:59 +0000 (00:35 +0200)]
radeonsi: add GFX-IB-size query to the HUD

It shows the sum of all IBs per frame.

Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
6 years agowinsys/amdgpu: disable CPU caching for GFX & SDMA IBs
Marek Olšák [Sun, 8 Oct 2017 22:06:13 +0000 (00:06 +0200)]
winsys/amdgpu: disable CPU caching for GFX & SDMA IBs

This should decrease IB fetch latency.

Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
6 years agowinsys/amdgpu: don't do read-modify-write on command buffers
Marek Olšák [Sun, 8 Oct 2017 21:47:30 +0000 (23:47 +0200)]
winsys/amdgpu: don't do read-modify-write on command buffers

i.e. don't use |=

Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
6 years agobroadcom/vc4: Fix false-positive for the tiling ioctls on simulator mode.
Eric Anholt [Fri, 13 Oct 2017 19:22:33 +0000 (12:22 -0700)]
broadcom/vc4: Fix false-positive for the tiling ioctls on simulator mode.

If there happened to be an ENOENT laying around, we would try using the
ioctls later and fail out resource allocation.

6 years agobroadcom/vc4: Skip BO labeling when in simulator mode.
Eric Anholt [Fri, 13 Oct 2017 19:16:04 +0000 (12:16 -0700)]
broadcom/vc4: Skip BO labeling when in simulator mode.

It was calling down into i915 trying to label the BO, which is definitely
not the right thing.

6 years agobroadcom/vc5: Don't forget to set the RT format for 1555 textures.
Eric Anholt [Fri, 13 Oct 2017 20:13:36 +0000 (13:13 -0700)]
broadcom/vc5: Don't forget to set the RT format for 1555 textures.

Fixes dEQP-GLES3.functional.fbo.completeness.renderable.texture.color0.rgb5_a1

6 years agoanv: Add func anv_gem_get_tiling()
Chad Versace [Tue, 29 Aug 2017 08:34:44 +0000 (01:34 -0700)]
anv: Add func anv_gem_get_tiling()

Will use in VK_ANDROID_native_buffer.

Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
6 years agoanv: Move close(fd) from anv_bo_cache_import to its callers (v2)
Chad Versace [Mon, 11 Sep 2017 23:41:21 +0000 (16:41 -0700)]
anv: Move close(fd) from anv_bo_cache_import to its callers (v2)

This will allow us to implement VK_ANDROID_native_buffer without dup'ing
the fd. We must close the fd in VK_KHR_external_memory_fd, but we should
not in VK_ANDROID_native_buffer.

v2:
  - Add missing close(fd) for case
    VK_EXTERNAL_SEMAPHORE_HANDLE_TYPE_OPAQUE_FD_BIT_KHR, subcase
    ANV_SEMAPHORE_TYPE_BO.

Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
6 years agoanv: Add field anv_image::planes[]::bo_is_owned (v2)
Chad Versace [Tue, 29 Aug 2017 08:36:42 +0000 (01:36 -0700)]
anv: Add field anv_image::planes[]::bo_is_owned (v2)

If this flag is set, then the image and the bo have the same lifetime.
vkDestroyImage will release the bo.

We need this for VK_ANDROID_native_buffer, because that extension
creates the VkImage *and* imports its memory during the same
call, vkCreateImage.

v2: Rebase onto VK_KHR_bind_memory2.

Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
6 years agoanv: Better support for Android logging (v2)
Chad Versace [Mon, 10 Jul 2017 18:11:40 +0000 (11:11 -0700)]
anv: Better support for Android logging (v2)

In src/intel/vulkan/*, redirect all instances of printf, vk_error,
anv_loge, anv_debug, anv_finishme, anv_perf_warn, anv_assert, and their
many variants to the new intel_log functions. I believe I caught them
all.

The other subdirs of src/intel are left for a future exercise.

v2:
  - Rebase onto Tapani's VK_EXT_debug_report changes.
  - Drop unused #include <cutils/log.h>.

Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
6 years agointel: Add simple logging façade for Android (v2)
Chad Versace [Fri, 25 Aug 2017 21:14:34 +0000 (14:14 -0700)]
intel: Add simple logging façade for Android (v2)

I'm bringing up Vulkan in the Android container of Chrome OS (ARC++).

On Android, stdio goes to /dev/null. On Android, remote gdb is even more
painful than the usual remote gdb. On Android, nothing works like you
expect and debugging is hell. I need logging.

This patch introduces a small, simple logging API that can easily wrap
Android's API. On non-Android platforms, this logger does nothing fancy.
It follows the time-honored Unix tradition of spewing everything to
stderr with minimal fuss.

My goal here is not perfection. My goal is to make a minimal, clean API,
that people hate merely a little instead of a lot, and that's good
enough to let me bring up Android Vulkan.  And it needs to be fast,
which means it must be small. No one wants to their game to miss frames
while aiming a flaming bow into the jaws of an angry robot t-rex, and
thus become t-rex breakfast, because some fool had too much fun desiging
a bloated, ideal logging API.

If people like it, perhaps we should quickly promote it to src/util.

The API looks like this:

    #define INTEL_LOG_TAG "intel-vulkan"
    #define DEBUG

    intel_logd("try hard thing with foo=%d", foo);

    n = try_foo(...);
    if (n < 0) {
        intel_loge("%s:%d: foo failed bigtime", __FILE__, __LINE__);
        return VK_ERROR_DEVICE_LOST;
    }

And produces this on non-Android:

    intel-vulkan: debug: try hard thing with foo=93
    intel-vulkan: error: anv_device.c:182: foo failed bigtime

v2: Fix meson build. [for dcbaker]

Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
6 years agoanv/android: Link to libsync, liblog in Android.mk
Tapani Pälli [Thu, 14 Sep 2017 06:57:40 +0000 (09:57 +0300)]
anv/android: Link to libsync, liblog in Android.mk

chadv: I made this patch by extracting the hunk from Tapani's patch in
https://lists.freedesktop.org/archives/mesa-dev/2017-September/169602.html.

Signed-off-by: Chad Versace <chadversary@chromium.org>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
6 years agoanv/android: Link to Android libraries in the autotools build
Chad Versace [Mon, 10 Jul 2017 18:15:35 +0000 (11:15 -0700)]
anv/android: Link to Android libraries in the autotools build

A first step to supporting Vulkan on ARC++. Mesa on ARC++ uses
Autotools, not Android.mk.

Doing this now, even before VK_ANDROID_native_buffer is implemented,
allows us to incrementally add Android support to the Autotools build.

Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
6 years agomeson: s/radv_extensions/radv_extensions_c/ to respect var convention
Eric Engestrom [Tue, 17 Oct 2017 17:58:28 +0000 (18:58 +0100)]
meson: s/radv_extensions/radv_extensions_c/ to respect var convention

Suggested-by: Dylan Baker <dylan@pnwbakers.com>
Signed-off-by: Eric Engestrom <eric.engestrom@imgtec.com>
6 years agomeson: track python script dependency
Eric Engestrom [Tue, 17 Oct 2017 15:21:54 +0000 (16:21 +0100)]
meson: track python script dependency

Suggested-by: Andres Gomez <agomez@igalia.com>
Signed-off-by: Eric Engestrom <eric.engestrom@imgtec.com>
Reviewed-by: Dylan Baker <dylan@pnwbakers.com>
6 years agovulkan/wsi: Free the event in x11_manage_fifo_queues().
Henri Verbeet [Fri, 13 Oct 2017 15:10:26 +0000 (17:10 +0200)]
vulkan/wsi: Free the event in x11_manage_fifo_queues().

Cc: mesa-stable@lists.freedesktop.org
Signed-off-by: Henri Verbeet <hverbeet@gmail.com>
Fixes: e73d136a023 ("vulkan/wsi/x11: Implement FIFO mode.")
Reviewed-by: Emil Velikov <emil.velikov@collabora.com>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com
6 years agomeson: add missing radv_extensions.c generation for libvulkan_radeon
Eric Engestrom [Tue, 17 Oct 2017 10:59:48 +0000 (11:59 +0100)]
meson: add missing radv_extensions.c generation for libvulkan_radeon

Fixes: 17201a2eb0b1b8538713 "radv: port to using updated anv entrypoint/extension generator."
Signed-off-by: Eric Engestrom <eric.engestrom@imgtec.com>
Reviewed-by: Andres Gomez <agomez@igalia.com>
6 years agoanv/apply_pipeline_layout: Use nir_tex_instr_remove_src
Jason Ekstrand [Mon, 16 Oct 2017 15:50:44 +0000 (08:50 -0700)]
anv/apply_pipeline_layout: Use nir_tex_instr_remove_src

Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
6 years agonir: Add a helper for adding texture instruction sources
Jason Ekstrand [Mon, 16 Oct 2017 15:50:23 +0000 (08:50 -0700)]
nir: Add a helper for adding texture instruction sources

Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
6 years agost/va: Return correct width and height for encode/decode support
Mark Thompson [Mon, 9 Oct 2017 20:45:03 +0000 (21:45 +0100)]
st/va: Return correct width and height for encode/decode support

Previously this would return the largest possible buffer size, which is
much larger than the codecs themselves support.  This caused confusion
when client applications attempted to decode 8K video thinking it was
supported when it isn't.

Signed-off-by: Mark Thompson <sw@jkqxz.net>
Reviewed-by: Christian König <christian.koenig@amd.com>
6 years agost/va: Fix config entrypoint handling
Mark Thompson [Tue, 10 Oct 2017 21:21:04 +0000 (22:21 +0100)]
st/va: Fix config entrypoint handling

Consistently use it as a PIPE_VIDEO_ENTRYPOINT.

v2: Return an error if the entrypoint is not set (Christian).

Signed-off-by: Mark Thompson <sw@jkqxz.net>
Reviewed-by: Christian König <christian.koenig@amd.com>
6 years agost/va: Disable vaExportSurfaceHandle()
Mark Thompson [Sun, 15 Oct 2017 19:57:21 +0000 (20:57 +0100)]
st/va: Disable vaExportSurfaceHandle()

This is not in libva 2.0, so it shouldn't be enabled yet.

Signed-off-by: Mark Thompson <sw@jkqxz.net>
Acked-by: Christian König <christian.koenig@amd.com>
6 years agoradv/image: bump all the offset to uint64_t.
Dave Airlie [Tue, 17 Oct 2017 07:17:12 +0000 (08:17 +0100)]
radv/image: bump all the offset to uint64_t.

So one of the CTS tests tries to allocate a 16384x1 2048 array
texture. This overflows a bunch of calculations when we want it
tiled as the heights goes to 128.

addrlib returns us the correct size (16GB or so), but we mangle
it in the htile calcs due to the 32-bit offset fields, then
userspace gives us the reduced number and we try to allocate
it on a heap and things blow up.

We really need to give the app back the correct size for the
image so we can blow up properly in memory allocation later.

This should fix hangs in
dEQP-VK.pipeline.render_to_image.core.1d_array.huge.width_layers.r8g8b8a8_unorm_d32_sfloat_s8_uint
since
Fixes: ad3d98da9f (radv: enable tc compatible htile for d32s8 also.)
Now there's an open question if we should be enabling tc-compat
htile at all for shallow textures like the above.

This might cause some other wierd side effects in CTS even
without the tc compat so:
Cc: "17.2" <mesa-stable@lists.freedesktop.org>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Signed-off-by: Dave Airlie <airlied@redhat.com>
6 years agoradv: port to using updated anv entrypoint/extension generator.
Dave Airlie [Tue, 17 Oct 2017 05:18:36 +0000 (15:18 +1000)]
radv: port to using updated anv entrypoint/extension generator.

This ports radv to using the anv entrypoint/extension generator
code.

No differences on enabled extensions list in vulkaninfo.

Acked-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Signed-off-by: Dave Airlie <airlied@redhat.com>
6 years agoradv: enable VK_KHX_multiview always.
Dave Airlie [Tue, 17 Oct 2017 06:47:16 +0000 (16:47 +1000)]
radv: enable VK_KHX_multiview always.

This was in the wrong place.

Fixes: ba51ad2f2 (radv: Expose VK_KHX_multiview.)
Acked-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Signed-off-by: Dave Airlie <airlied@redhat.com>
6 years agoRevert "mesa: fix texture updates for ATI_fragment_shader"
Marek Olšák [Tue, 17 Oct 2017 01:08:47 +0000 (03:08 +0200)]
Revert "mesa: fix texture updates for ATI_fragment_shader"

This reverts commit 9d54025cd1eee3f42b80c062d1f644904bf2ba41.

It breaks KOTOR.

Cc: 17.1 17.2 <mesa-stable@lists.freedesktop.org>
6 years agomesa: remove redundant NULL check in update_single_program_texture_state
Miklós Máté [Sun, 15 Oct 2017 17:46:05 +0000 (19:46 +0200)]
mesa: remove redundant NULL check in update_single_program_texture_state

update_single_program_texture() never returns NULL.

Signed-off-by: Miklós Máté <mtmkls@gmail.com>
Signed-off-by: Marek Olšák <marek.olsak@amd.com>
Reviewed-by: Timothy Arceri <tarceri@itsqueeze.com>
6 years agomeson: build mesa test.
Dylan Baker [Fri, 13 Oct 2017 19:07:04 +0000 (12:07 -0700)]
meson: build mesa test.

v2: - add dependency on dispatch.h generator (which this test needs)

Signed-off-by: Dylan Baker <dylanx.c.baker@intel.com>
Reviewed-by: Eric Anholt <eric@anholt.net> (v1)
6 years ago.travis: Don't build gallium drivers in non-gallium test targets
Dylan Baker [Wed, 11 Oct 2017 17:50:33 +0000 (10:50 -0700)]
.travis: Don't build gallium drivers in non-gallium test targets

Simply disable gallium in non-gallium builds. For some reason the
gallium driver wont link on ubuntu 14.04 (it will on 16.04, debian
testing, and arch)

Signed-off-by: Dylan Baker <dylanx.c.baker@intel.com>
6 years agomeson: refactor meson_options
Dylan Baker [Tue, 10 Oct 2017 22:44:14 +0000 (15:44 -0700)]
meson: refactor meson_options

To put one argument on each line. This results in the file being much
longer, but I think much more readable.

Suggested-by: Eero Tamminen <eero.t.tamminen@intel.com>
Signed-off-by: Dylan Baker <dylanx.c.baker@intel.com>
Reviewed-by: Eric Anholt <eric@anholt.net>
6 years agomeson: build llvmpipe
Dylan Baker [Tue, 10 Oct 2017 21:56:39 +0000 (14:56 -0700)]
meson: build llvmpipe

Signed-off-by: Dylan Baker <dylanx.c.baker@intel.com>
Reviewed-by: Eric Anholt <eric@anholt.net>
6 years agomeson: build softpipe
Dylan Baker [Tue, 10 Oct 2017 21:27:19 +0000 (14:27 -0700)]
meson: build softpipe

This doesn't include llvmpipe.

v2: - Fix inconsistent use of with_gallium_swrast and
      with_gallium_softpipe.

Signed-off-by: Dylan Baker <dylanx.c.baker@intel.com>
Reviewed-by: Eric Anholt <eric@anholt.net>
6 years agomeson: build nouveau (gallium) driver
Dylan Baker [Mon, 9 Oct 2017 21:59:35 +0000 (14:59 -0700)]
meson: build nouveau (gallium) driver

Tested with a GK107.

v2: - Add target for nouveau standalone compiler. This target is not
      built by default.
v3: - Add nouveau to list of drivers built by default

Signed-off-by: Dylan Baker <dylanx.c.baker@intel.com>
Reviewed-by: Eric Anholt <eric at anholt.net>
6 years agomeson: build radeonsi gallium driver
Dylan Baker [Sat, 30 Sep 2017 21:04:28 +0000 (14:04 -0700)]
meson: build radeonsi gallium driver

This hooks up the bits necessary to build gallium dri drivers, with
radeonSI as the first example driver. This isn't tested yet.

v4: - drop radeonsi generated header from sources.

Signed-off-by: Dylan Baker <dylanx.c.baker@intel.com>
Reviewed-by: Eric Anholt <eric at anholt.net>
6 years agomeson: build gallium winsys for dri, null, and wrapper
Dylan Baker [Sat, 30 Sep 2017 20:48:34 +0000 (13:48 -0700)]
meson: build gallium winsys for dri, null, and wrapper

Signed-off-by: Dylan Baker <dylanx.c.baker@intel.com>
Reviewed-by: Eric Anholt <eric at anholt.net>
6 years agomeson: build radeonsi
Dylan Baker [Sat, 30 Sep 2017 16:03:51 +0000 (09:03 -0700)]
meson: build radeonsi

This builds the radeonsi (and radeon) window system bits and gallium
driver bits.

Signed-off-by: Dylan Baker <dylanx.c.baker@intel.com>
Reviewed-by: Eric Anholt <eric at anholt.net>
6 years agomeson: Build gallium dri state tracker
Dylan Baker [Tue, 10 Oct 2017 18:57:50 +0000 (11:57 -0700)]
meson: Build gallium dri state tracker

Signed-off-by: Dylan Baker <dylanx.c.baker@intel.com>
Reviewed-by: Eric Anholt <eric at anholt.net>
6 years agomeson: build gallium helper drivers
Dylan Baker [Sat, 30 Sep 2017 03:21:23 +0000 (20:21 -0700)]
meson: build gallium helper drivers

This builds ddebug, noop, rbug, and trace drivers.

Signed-off-by: Dylan Baker <dylanx.c.baker@intel.com>
Reviewed-by: Eric Anholt <eric at anholt.net>
6 years agomeson: Build gallium pipe-loader
Dylan Baker [Sat, 30 Sep 2017 03:03:20 +0000 (20:03 -0700)]
meson: Build gallium pipe-loader

Signed-off-by: Dylan Baker <dylanx.c.baker@intel.com>
6 years agomeson: split and simplify dependencies
Dylan Baker [Sat, 30 Sep 2017 00:53:01 +0000 (17:53 -0700)]
meson: split and simplify dependencies

Rather than group dependencies in complex groups, use a flatter
structure with split dependencies to avoid checking for the same
dependencies twice.

v2: - Fix building vulkan drivers without gallium or dri drivers
v3: - Drop TODO comment that is done
    - Fix typo in commit message

Signed-off-by: Dylan Baker <dylanx.c.baker@intel.com>
Reviewed-by: Eric Anholt <eric@anholt.net>
6 years agomeson: Build gallium auxiliary
Dylan Baker [Fri, 29 Sep 2017 05:25:02 +0000 (22:25 -0700)]
meson: Build gallium auxiliary

v2: - guard gallivm files with "with_llvm" instead of "dep_llvm.found()"

Signed-off-by: Dylan Baker <dylanx.c.baker@intel.com>
Reviewed-by: Eric Anholt <eric@anholt.net> (v1)
6 years agomeson: build libmesa_gallium
Dylan Baker [Fri, 29 Sep 2017 04:03:07 +0000 (21:03 -0700)]
meson: build libmesa_gallium

Signed-off-by: Dylan Baker <dylanx.c.baker@intel.com>
Reviewed-by: Eric Anholt <eric@anholt.net>
6 years agomeson: Add option to toggle LLVM
Dylan Baker [Thu, 12 Oct 2017 16:47:30 +0000 (09:47 -0700)]
meson: Add option to toggle LLVM

Signed-off-by: Dylan Baker <dylanx.c.baker@intel.com>
Reviewed-by: Eric Anholt <eric@anholt.net>
6 years agomeson: always set GLX_USE_TLS
Dylan Baker [Wed, 11 Oct 2017 19:49:31 +0000 (12:49 -0700)]
meson: always set GLX_USE_TLS

This can be applied to all GLX implementations, and in autotools this is
guarded only by the --enable-glx-tls flag. Since this is on by default
in autotools, and is strictly better than being off, the meson build
doesn't even have a toggle for it.

Signed-off-by: Dylan Baker <dylanx.c.baker@intel.com>
Reviewed-by: Eric Anholt <eric@anholt.net>
6 years agomeson: Don't try to install dri drivers unless one is built
Dylan Baker [Mon, 9 Oct 2017 23:34:03 +0000 (16:34 -0700)]
meson: Don't try to install dri drivers unless one is built

This confused the with_dri flag which is meant to control Direct
Rendering Infrastructure, not classic drivers

Signed-off-by: Dylan Baker <dylanx.c.baker@intel.com>
Reviewed-by: Eric Anholt <eric@anholt.net>
6 years agomeson: Set _GNU_SOURCE
Dylan Baker [Mon, 9 Oct 2017 21:22:07 +0000 (14:22 -0700)]
meson: Set _GNU_SOURCE

When we start adding non-free software platforms support we'll need to
guard this, but for now it should be fine as is.

Signed-off-by: Dylan Baker <dylanx.c.baker@intel.com>
Reviewed-by: Eric Anholt <eric@anholt.net>
6 years agomeson: add checks for version script and dynamic list
Dylan Baker [Sat, 30 Sep 2017 20:15:52 +0000 (13:15 -0700)]
meson: add checks for version script and dynamic list

These are used by gallium drivers.

Signed-off-by: Dylan Baker <dylanx.c.baker@intel.com>
Reviewed-by: Eric Anholt <eric@anholt.net>
6 years agoconfigure: commit test files
Dylan Baker [Sat, 30 Sep 2017 19:14:02 +0000 (12:14 -0700)]
configure: commit test files

These are currently auto-generated, but meson needs the same files, so
lets commit them to reduce duplication.

v3: - Rename .build to build-support

Reviewed-by: Emil Velikov <emil.l.velikov@gmail.com>
Reviewed-by: Eric Anholt <eric@anholt.net>
Signed-off-by: Dylan Baker <dylanx.c.baker@intel.com>
6 years agomeson: Add switch for texture float
Dylan Baker [Tue, 10 Oct 2017 22:25:07 +0000 (15:25 -0700)]
meson: Add switch for texture float

Signed-off-by: Dylan Baker <dylanx.c.baker@intel.com>
Reviewed-by: Eric Anholt <eric@anholt.net>
6 years agoRevert "i965/tex_image: Reference the renderbuffer miptree in setTexBuffer2"
Kenneth Graunke [Mon, 16 Oct 2017 22:30:47 +0000 (15:30 -0700)]
Revert "i965/tex_image: Reference the renderbuffer miptree in setTexBuffer2"

This reverts commit d80cbbeaff9329fdc78ae3d97097c1e65dfcdd61.

It turns out that formats do matter - the framebuffer's miptree has an
sRGB format, and the one we created did not.  This broke rendering when
using KWin compositing, GNOME Terminal Fedora (with a transparent
background), and Qt menu rendering in general, to name a few.

It's been a month and this hasn't been fixed, and I'm sick of reverting
this patch or applying NAK'd hacks and restarting various programs at
random times every day, multiple times a day, to keep my desktop
environment functional.

The only benefit of this patch was to prepare the way for modifiers,
which AFAIK aren't finished yet anyway, so there's really no downside
to reverting it.

Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=102924

6 years agoAndroid: add libmesa_nir dependency to libmesa_dricore
Rob Herring [Mon, 16 Oct 2017 12:43:36 +0000 (07:43 -0500)]
Android: add libmesa_nir dependency to libmesa_dricore

Commit 32fcced7b479 ("meta: Unset the textures_used_by_txf bitfield.")
added a dependency in libmesa_dricore to NIR headers, but failed to add
libmesa_nir as a dependency resulting in a build error:

In file included from external/mesa3d/src/mesa/drivers/common/meta.c:90:
external/mesa3d/src/compiler/nir/nir.h:48:10: fatal error: 'nir_opcodes.h' file not found

Add libmesa_nir as a static library dependency to libmesa_dricore.

Fixes: 32fcced7b479 ("meta: Unset the textures_used_by_txf bitfield.")
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Signed-off-by: Rob Herring <robh@kernel.org>
6 years agoi965: Only put external handles into the handle ht
Chris Wilson [Fri, 13 Oct 2017 09:34:51 +0000 (10:34 +0100)]
i965: Only put external handles into the handle ht

We know that we will only ever need to lookup an external handle and so
can defer adding a bo to the external ht until it is ever exported or
imported, keeping that hashtable compact.

Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
6 years agosvga: format the version string like the rest of mesa
Eric Engestrom [Mon, 16 Oct 2017 16:14:28 +0000 (17:14 +0100)]
svga: format the version string like the rest of mesa

All 4 other version strings do it like this.
((Also, double parentheses just look confusing))

Signed-off-by: Eric Engestrom <eric.engestrom@imgtec.com>
Reviewed-by: Brian Paul <brianp@vmware.com>
6 years agogit_sha1_gen: use git_sha1.h.in on all build systems
Eric Engestrom [Mon, 16 Oct 2017 16:02:03 +0000 (17:02 +0100)]
git_sha1_gen: use git_sha1.h.in on all build systems

Meson already uses this, let's get the other build sys to use it too.

Note: rstrip() was dropped, as truncating to the first 10 chars already
gets rid of the terminating newline (not an issue with the env var
either, unless maliciously crafted to break the build... not sure this
is a real-world issue).

Verified to work and give the same output as before on both python 2
and 3 :)

Signed-off-by: Eric Engestrom <eric.engestrom@imgtec.com>
Reviewed-by: Brian Paul <brianp@vmware.com>
6 years agosvga: fix format_conversion_table breakage
Brian Paul [Sat, 14 Oct 2017 16:50:21 +0000 (10:50 -0600)]
svga: fix format_conversion_table breakage

The new A1B5G5R5_UNORM, X1B5G5R5_UNORM formats were added in the
wrong place in commit ef874ee450b18e.

Fixes: ef874ee450b18e "gallium: Add support for 5551 with the 1-bit field in the low bit."
Reviewed-by: Emil Velikov <emil.l.velikov@gmail.com>
Reviewed-by: Eric Engestrom <eric.engestrom@imgtec.com>
6 years agoi965/miptree: Drop the invalidate parameter form copy_teximage
Jason Ekstrand [Mon, 9 Oct 2017 22:08:33 +0000 (15:08 -0700)]
i965/miptree: Drop the invalidate parameter form copy_teximage

This was a leftover from i915.  The one caller in i965 always passes in
false so there's no point in having the parameter.

Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
6 years agoanv: Get rid of gen fall-through
Jason Ekstrand [Mon, 11 Sep 2017 04:27:13 +0000 (21:27 -0700)]
anv: Get rid of gen fall-through

In the early days of the Vulkan driver, we thought it would be a good
idea to just make genN just fall back to the genN-1 code if it didn't
need to be any different for genN.  While this seemed like a good idea,
it ultimately ended up being far simpler to just recompile everything.
We haven't been using the fall-through functionality for some time so
we're better off just deleting it so it doesn't accidentally start
causing problems.

Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
6 years agointel/common: Improve the comments for sample positions
Jason Ekstrand [Fri, 18 Aug 2017 02:05:30 +0000 (19:05 -0700)]
intel/common: Improve the comments for sample positions

These are pulled directly from brw_multisample_state.h

Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
6 years agoradv: update ia_multi_vgt when executing secondary buffers
Samuel Pitoiset [Mon, 16 Oct 2017 12:15:55 +0000 (14:15 +0200)]
radv: update ia_multi_vgt when executing secondary buffers

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
6 years agoradv: be smarter with the draw packets when executing secondary buffers
Samuel Pitoiset [Mon, 16 Oct 2017 12:15:54 +0000 (14:15 +0200)]
radv: be smarter with the draw packets when executing secondary buffers

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
6 years agoradv: always dirty some states after executing secondary buffers
Samuel Pitoiset [Wed, 11 Oct 2017 14:10:43 +0000 (16:10 +0200)]
radv: always dirty some states after executing secondary buffers

The spec requires the number of buffer to be greater than 0.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
6 years agoradv: be smarter with pipelines when emitting secondary buffers
Samuel Pitoiset [Wed, 11 Oct 2017 13:54:40 +0000 (15:54 +0200)]
radv: be smarter with pipelines when emitting secondary buffers

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
6 years agodocs: Add EXT_memory_objects extensions to features.txt
Jakob Bornecrantz [Sun, 15 Oct 2017 16:42:48 +0000 (17:42 +0100)]
docs: Add EXT_memory_objects extensions to features.txt

These extensions are good for Vulkan interop, so track them.

Signed-off-by: Jakob Bornecrantz <jakob.bornecrantz@collabora.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
6 years agonir: add component level support to remove_unused_io_vars()
Timothy Arceri [Tue, 26 Sep 2017 03:18:58 +0000 (13:18 +1000)]
nir: add component level support to remove_unused_io_vars()

Reviewed-by: Eric Anholt <eric@anholt.net>
6 years agoglsl: mark xfb inputs as always_active_io
Timothy Arceri [Fri, 22 Sep 2017 06:04:17 +0000 (16:04 +1000)]
glsl: mark xfb inputs as always_active_io

We won't split varyings marked as always active because there
is no point in doing so. This means we need to mark both
sides of the interface as always active otherwise we will have
a mismatch and start removing things we shouldn't.

Reviewed-by: Eric Anholt <eric@anholt.net>
6 years agonir: add variant of lower_io_to_scalar to be called earlier
Timothy Arceri [Thu, 5 Oct 2017 22:19:35 +0000 (09:19 +1100)]
nir: add variant of lower_io_to_scalar to be called earlier

This is intended to be called before nir_lower_io() so that we
can do some linking optimisations with the results. It can also
be used with drivers that don't use nir_lower_io() at all such
as RADV.

v2: pass mode mask rather than first and last stage integer.

Reviewed-by: Eric Anholt <eric@anholt.net>
6 years agonir: add glsl_channel_type() helper
Timothy Arceri [Tue, 19 Sep 2017 23:49:57 +0000 (09:49 +1000)]
nir: add glsl_channel_type() helper

Reviewed-by: Eric Anholt <eric@anholt.net>
6 years agonir: add glsl_type_is_64bit() to nir_types
Timothy Arceri [Sun, 4 Dec 2016 06:21:16 +0000 (17:21 +1100)]
nir: add glsl_type_is_64bit() to nir_types

Reviewed-by: Eric Anholt <eric@anholt.net>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
6 years agoa2xx: add support for a few 16-bit color rendering formats
Ilia Mirkin [Fri, 25 Aug 2017 04:47:05 +0000 (00:47 -0400)]
a2xx: add support for a few 16-bit color rendering formats

The rest should be possible too, just needs some additional
investigation. Passes fbo-*-formats piglit tests.

Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
Reviewed-by: Wladimir J. van der Laan <laanwj@gmail.com>
6 years agofreedreno/a20x: Enable rendering to RGBA/RGBX
Wladimir J. van der Laan [Fri, 25 Aug 2017 14:34:08 +0000 (14:34 +0000)]
freedreno/a20x: Enable rendering to RGBA/RGBX

Signed-off-by: Wladimir J. van der Laan <laanwj@gmail.com>
Reviewed-by: Ilia Mirkin <imirkin@alum.mit.edu>
6 years agofreedreno/a20x: Fix rendering to BGRX
Wladimir J. van der Laan [Fri, 25 Aug 2017 14:29:53 +0000 (14:29 +0000)]
freedreno/a20x: Fix rendering to BGRX

Make sure that BGRX rendering is swapped the correct way around.

Signed-off-by: Wladimir J. van der Laan <laanwj@gmail.com>
Reviewed-by: Ilia Mirkin <imirkin@alum.mit.edu>
6 years agomesa: minor simplification in test_attachment_completeness()
Brian Paul [Fri, 13 Oct 2017 22:16:48 +0000 (16:16 -0600)]
mesa: minor simplification in test_attachment_completeness()

We already have a pointer to the texture object.  Use it here.

Reviewed-by: Timothy Arceri <tarceri@itsqueeze.com>
6 years agoetnaviv: rework TS enable to be a derived state
Lucas Stach [Thu, 12 Oct 2017 14:07:48 +0000 (16:07 +0200)]
etnaviv: rework TS enable to be a derived state

Draw operations should not use the TS if the TS buffer content is invalid,
as this leads to wrong rendering or even GPU hangs. As the TS valid status
can change between draws (clear operations changing it to valid, blits using
the RS to the color or ZS buffer changing it to invalid), the TS_MEM_CONFIG
must be updated before each draw if the status has changed.

This fixes the remaining TS related piglit failures (regressions of a
standard run against a piglit run with TS completely disabled).

Signed-off-by: Lucas Stach <l.stach@pengutronix.de>
Reviewed-by: Wladimir J. van der Laan <laanwj@gmail.com>
Reviewed-by: Christian Gmeiner <christian.gmeiner@gmail.com>
6 years agoetnaviv: skip unused vertex attributes when assigning VS inputs
Lucas Stach [Thu, 12 Oct 2017 14:07:47 +0000 (16:07 +0200)]
etnaviv: skip unused vertex attributes when assigning VS inputs

When not all of the vertex attributes are actually used in the shader,
we end up with some inputs without an assigned reg. Those are marked
as invalid and must be skipped when assigning the inputs, as those would
overwrite other valid inputs otherwise.

Fixes piglit drawpixels and a bunch of other tests using the st_draw path.

Signed-off-by: Lucas Stach <l.stach@pengutronix.de>
Reviewed-by: Wladimir J. van der Laan <laanwj@gmail.com>
Reviewed-by: Christian Gmeiner <christian.gmeiner@gmail.com>
6 years agoradv: do not allocate CMASK for non-MSSA images with 128 bit formats
Samuel Pitoiset [Thu, 12 Oct 2017 20:55:32 +0000 (22:55 +0200)]
radv: do not allocate CMASK for non-MSSA images with 128 bit formats

This saves some useless CMASK initializations/eliminations in
the Vulkan SSAO demo.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
6 years agoradv: set correct INDEX_TYPE for indexed indirect draws on GFX9
Samuel Pitoiset [Fri, 13 Oct 2017 16:01:56 +0000 (18:01 +0200)]
radv: set correct INDEX_TYPE for indexed indirect draws on GFX9

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
6 years agoradv: add the draw count buffer to the list of buffers
Samuel Pitoiset [Fri, 13 Oct 2017 16:20:21 +0000 (18:20 +0200)]
radv: add the draw count buffer to the list of buffers

My guess is that the GPU is going to report VM faults if
vkCmdDrawIndirectCountAMD() (and friends) are used.

Cc: mesa-stable@lists.freedesktop.org
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
6 years agoblob: Use intptr_t instead of ssize_t
Jason Ekstrand [Fri, 13 Oct 2017 17:40:45 +0000 (10:40 -0700)]
blob: Use intptr_t instead of ssize_t

ssize_t is a GNU extension and is not available on Windows or MacOS.
Instead, we use intptr_t which should be effectively equivalent and is
part of the C standard.  This should fix the Windows and Mac OS builds.

Fixes: 3af1c829891a4530682bce113fdd512d4f2de3c6
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=103253
Reviewed-by: Jose Fonseca <jfonseca@vmware.com>
Tested-by: Vinson Lee <vlee@freedesktop.org>
6 years agoi965: Rename brw->no_batch_wrap to intel_batchbuffer::no_wrap
Kenneth Graunke [Fri, 13 Oct 2017 07:35:30 +0000 (00:35 -0700)]
i965: Rename brw->no_batch_wrap to intel_batchbuffer::no_wrap

This really makes more sense in the intel_batchbuffer struct.

Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
6 years agoi965: Delete dead brw_context fields.
Kenneth Graunke [Fri, 13 Oct 2017 07:31:27 +0000 (00:31 -0700)]
i965: Delete dead brw_context fields.

fast_clear_op is leftover from the meta-fast-clear days.
No idea what the other thing was for, but it isn't used now.

Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
6 years agomapi/shared-glapi/test: rework glapitable.h handling
Emil Velikov [Fri, 13 Oct 2017 15:35:47 +0000 (16:35 +0100)]
mapi/shared-glapi/test: rework glapitable.h handling

Currently all the build systems but Meson generate the header in
src/mapi/glapi. Meson cannot do that since:
 - it does not allow user control over the location of output files
 - moving the generation rule(s) causes explosion due to the unusual
structure of glapi and friends
 - copying the file into the correct location is a non-trivial task

To workaround the above deficiency in the least invasive way, let's
adjust the #include directive and add a few -I flags to the autotools
build.

Note: both builddir and srcdir, should be used. Otherwise building from
a release tarball fails badly.

Cc: Dylan Baker <dylanx.c.baker@intel.com>
Signed-off-by: Emil Velikov <emil.velikov@collabora.com>
Tested-by: Mark Janes <mark.a.janes@intel.com>
Reviewed-by: Dylan Baker <dylan@pnwbakers.com>
6 years agomeson: fix blob test includes
Dylan Baker [Fri, 13 Oct 2017 17:34:07 +0000 (10:34 -0700)]
meson: fix blob test includes

Since blob.h moved up to src/compiler the test should include that
instead of src/compiler/glsl

fixes: 0e3bd56c6ea783dbc ("compiler: Move blob up a level")
Signed-off-by: Dylan Baker <dylanx.c.baker@intel.com>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
6 years agoRevert "make: Fix test to be meson compatible"
Emil Velikov [Fri, 13 Oct 2017 13:54:53 +0000 (14:54 +0100)]
Revert "make: Fix test to be meson compatible"

This reverts commit fc48ad24272799b154de37f6a944eb391bdbbb82.

There commit reference the previous commit as it justification of
changing behaviour. Although unlike the said commit, there's nothing
obviously wrong there.

I'll take a look close why Meson fails to pick the file, but in the
interim reverting this commit fixes the normal distcheck target.

6 years agost/dri: Add definitions to allow importing 16-bit surfaces
Mark Thompson [Sun, 1 Oct 2017 17:39:05 +0000 (18:39 +0100)]
st/dri: Add definitions to allow importing 16-bit surfaces

Necessary to support P010/P016 surfaces for video.

Signed-off-by: Mark Thompson <sw@jkqxz.net>
Acked-by: Leo Liu <leo.liu@amd.com>
6 years agoi965: Complete 'expose RGBA visuals only on Android'
Mario Kleiner [Fri, 6 Oct 2017 16:11:31 +0000 (18:11 +0200)]
i965: Complete 'expose RGBA visuals only on Android'

Commit 731ba6924a2ed6cdf47a78fd637a91a19ddcf9ed
"expose RGBA visuals only on Android" replaced
ARRAY_SIZE(formats) by num_formats, but there are
3 loops which add configs, and only one was updated
to num_formats.

Also update loops for configs with accumulation buffer
and multisample configs.

Fixes: 731ba6924a2 "i965: expose RGBA visuals only on Android"
Signed-off-by: Mario Kleiner <mario.kleiner.de@gmail.com>
Reviewed-by: Tapani Pälli <tapani.palli@intel.com>
Reviewed-by: Emil Velikov <emil.velikov@collabora.com>
6 years agoconfigure.ac: add missing LLVM components for OpenCL
Emil Velikov [Thu, 5 Oct 2017 10:19:08 +0000 (11:19 +0100)]
configure.ac: add missing LLVM components for OpenCL

Coverage and LTO seems to be hard requirements for Clang, while
coroutines is needed as of LLVM/Clang 4.0.

Mark the last one as "optional" so we handle every case.

Cc: mesa-stable@lists.freedesktop.org
Signed-off-by: Emil Velikov <emil.velikov@collabora.com>
Reviewed-by: Tobias Droste <tdroste@gmx.de>
6 years agoconfigure.ac: add llvm_add_optional_component helper
Emil Velikov [Thu, 5 Oct 2017 10:19:07 +0000 (11:19 +0100)]
configure.ac: add llvm_add_optional_component helper

We want to add "optional" components, which have been added with later
LLVM versions.

One such in-tree example is inteljitevents. Others are to follow
shortly.

v2: Use the correct function, add blank line between functions (Tobias)

Cc: mesa-stable@lists.freedesktop.org
Signed-off-by: Emil Velikov <emil.velikov@collabora.com>
Reviewed-by: Tobias Droste <tdroste@gmx.de>
6 years agoTravis: add binutils 2.26 for a few more LLVM 3.9 builds
Emil Velikov [Thu, 5 Oct 2017 10:19:06 +0000 (11:19 +0100)]
Travis: add binutils 2.26 for a few more LLVM 3.9 builds

Otherwise we error out at link stage as follows:

/usr/lib/llvm-3.9/lib/libLLVMAMDGPUCodeGen.a(R600OptimizeVectorRegisters.cpp.o):
unrecognized relocation (0x2a) in section
`.text._ZNK12_GLOBAL__N_119R600VectorRegMerger16getAnalysisUsageERN4llvm13AnalysisUsageE'
/usr/bin/ld: final link failed: Bad value

Cc: mesa-stable@lists.freedesktop.org
Cc: Jan Vesely <jan.vesely@rutgers.edu
Signed-off-by: Emil Velikov <emil.velikov@collabora.com>