freedreno: allow ctx->draw_vbo to fail

Pretty much only happens if shader variant compile fails.  But in this
case, if we haven't emitted cmdstream, we don't want to set needs_flush.

Signed-off-by: Rob Clark <robclark@freedesktop.org>

freedreno: move shader-stage dirty bits to global dirty flag

This was always a bit overly complicated, and had some issues (like
ctx->prog.dirty not getting reset at the end of the batch).  It also
required some special hacks to avoid resetting dirty state on binning
pass.  So just move it all into ctx->dirty (leaving some free bits
for future shader stages), and make FD_DIRTY_PROG just be the union
of all FD_SHADER_DIRTY_*.

Signed-off-by: Rob Clark <robclark@freedesktop.org>

freedreno/a4xx: fix bogus offset for f32x24s8 stencil restore

fixes: $piglit/bin/fbo-clear-formats GL_ARB_depth_buffer_float

Signed-off-by: Rob Clark <robclark@freedesktop.org>

freedreno: add some debug_asserts() to catch insane offsets

Ofc won't catch *all* faults, but at least helpful for catching offsets
which are completely bogus.

Signed-off-by: Rob Clark <robclark@freedesktop.org>

freedreno/a4xx: deal with VS which do not write position

Fixes $piglit/bin/glsl-1.40-tf-no-position

a3xx may need similar?

Signed-off-by: Rob Clark <robclark@freedesktop.org>

freedreno/ir3: remove a couple redundant is_flow()s

Now that the opc's encode the instruction category (making them unique)
we no longer need to check the category in addition to the opc.

Signed-off-by: Rob Clark <robclark@freedesktop.org>

freedreno/ir3: cp small negative integers too

Signed-off-by: Rob Clark <robclark@freedesktop.org>

freedreno/ir3: fix # of registers

The instruction encoding allows for more registers, but at least on
a3xx/a4xx they don't actually exist.

Signed-off-by: Rob Clark <robclark@freedesktop.org>

freedreno/ir3: lower immeds to const

Helps reduce register pressure and instruction counts for immediates
that would otherwise require a mov into gpr.

total instructions in shared programs:          4455332 -> 4369297 (-1.93%)
total dwords in shared programs:                8807872 -> 8614432 (-2.20%)
total full registers used in shared programs:   263062 -> 250846 (-4.64%)
total half registers used in shader programs:   9845 -> 9845 (0.00%)
total const registers used in shared programs:  1029735 -> 1466993 (42.46%)

                 half       full      const      instr     dwords
    helped           0       10415           0       17861        5912
      hurt           0        1157       21458         947          33

Signed-off-by: Rob Clark <robclark@freedesktop.org>

freedreno/ir3: add ir3_cp_ctx

Needed in next commit.. just split out to reduce noise.

Signed-off-by: Rob Clark <robclark@freedesktop.org>

add REVIEWERS and get_reviewer.pl script

Copied from linux kernel (where it is called MAINTAINERS and
get_maintainer.pl), with minimal changes to script (to recognize
mesa src tree rather than linux kernel src tree, and to avoid
accidentaly CC'ing Linus Torvalds on mesa patches), and slimmed
down MAINTAINER file syntax to recognize that we don't really have
subsystem "maintainers" in the same sense as the linux kernel (ie. no
different mailing lists and git trees per subsystem).

The main point is to automate slapping on the correct CC's for patches
via git's --cc-cmd feature, more than anything else.

I didn't attempt to fully populate the REVIEWERS file, by a long shot.
This is an opt-in system and anyone else can add their own entries.

To utilize:

  git send-email --cc-cmd ./scripts/get_reviewer.pl ...

or to configure it to be the default:

  git config sendemail.cccmd ./scripts/get_reviewer.pl

Signed-off-by: Rob Clark <robclark@freedesktop.org>

nouveau/video: properly detect the decoder class for availability checks

The kernel is now more strict with the class ids it exposes, so we need
to check the G98 and MCP89 classes as well as the GT215 class. This
effectively caused us to decide there were no decoding capabilities on
newer kernel for VP3 chips.

Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=95251
Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
Cc: "11.2" <mesa-stable@lists.freedesktop.org>

i965: Delete stale perf_debug().

MOCS for 3DSTATE_SO_BUFFER has existed for ages.

i965: Silence unused variable warning

I added this when deleting some unnecessary code in a rebase.

mesa/main: handle double uniform matrices properly

When computing the offset in the uniform storage table, take into account
the size multiplier so double precision matrices are handled correctly.

Signed-off-by: Juan A. Suarez Romero <jasuarez@igalia.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>

nir: Separate 32 and 64-bit fmod lowering

Split 32-bit and 64-bit fmod lowering as the drivers might need to
lower them separately inside NIR depending on the HW support.

Signed-off-by: Samuel Iglesias Gonsálvez <siglesias@igalia.com>
Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>

nir/lower_double_ops: lower mod()

There are rounding errors with the division in i965 that affect
the mod(x,y) result when x = N * y. Instead of returning '0' it
was returning 'y'.

This lowering pass fixes those cases.

Signed-off-by: Samuel Iglesias Gonsálvez <siglesias@igalia.com>
Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>

i965: Define GEN_GE/GEN_LE macros in terms of GEN_LT.

GEN_LT has a straightforward implementation on which we can build the
GEN_GE and GEN_LE macros.

Suggested-by: Ilia Mirkin <imirkin@alum.mit.edu>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>

i965: Add disassembler support for remaining opcodes.

For opcodes that changed meaning on different generations, we store a
pointer to a secondary table and the table's size in a tagged union in
place of the mnemonic and number of sources.

Acked-by: Francisco Jerez <currojerez@riseup.net>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>

i965: Make opcode_descs and gen_from_devinfo() static.

The previous commit replaced direct uses of opcode_descs with calls to
the wrapper function, which should be the only method of accessing
opcode_descs's data. As a result gen_from_devinfo() can also be made
static.

Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>

i965: Actually check whether the opcode is supported.

Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>

i965: Merge inst_info and opcode_desc tables.

I merged opcode_desc into inst_info (instead of the other way around)
because inst_info was sorted by opcode number.

Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>

i965: Move inst_info from brw_eu_validate.c to brw_eu.c.

Drop the uses of 'enum gen' to a plain int, so that we don't have to
expose the bitfield definitions and GEN_GE/GEN_LE macros to other users
of brw_eu.h. As a result, s/.gen/.gens/ to avoid confusion with
devinfo->gen.

Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>

i965/disasm: Wrap opcode_desc look-up in a function.

The function takes a device info struct as argument in addition to the
opcode number in order to disambiguate between multiple opcode_desc
entries for different instructions with the same opcode number.

Reviewed-by: Iago Toral Quiroga <itoral@igalia.com> [v1]
[v2] mattst88: Put brw_opcode_desc() in brw_eu.c instead of moving it
               there in a later patch.
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org> [v2]
[v3] mattst88: Return NULL if opcode >= ARRAY_SIZE(opcode_descs)
Reviewed-by: Matt Turner <mattst88@gmail.com>

i965: Pass devinfo pointer to is_3src() helpers.

This is not strictly required for the following changes because none
of the three-source opcodes we support at the moment in the compiler
back-end has been removed or redefined, but that's likely to change in
the future.  In any case having hardware instructions specified as a
pair of hardware device and opcode number explicitly in all cases will
simplify the opcode look-up interface introduced in a subsequent
commit, since the opcode number alone is in general ambiguous.

Reviewed-by: Iago Toral Quiroga <itoral@igalia.com>
Reviewed-by: Matt Turner <mattst88@gmail.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>

i965: Pass devinfo pointer to brw_instruction_name().

A future series will implement support for an instruction that happens
to have the same opcode number as another instruction we support
already on a disjoint set of hardware generations.  In order to
disambiguate which instruction it is brw_instruction_name() will need
some way to find out which device we are generating code for.

Reviewed-by: Iago Toral Quiroga <itoral@igalia.com>
Reviewed-by: Matt Turner <mattst88@gmail.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>

i965: Write a scalar TCS backend that runs in SINGLE_PATCH mode.

Unlike most shader stages, the Hull Shader hardware makes us explicitly
tell it how many threads to dispatch and manually configure the channel
mask.  One perk of this is that we have a lot of flexibility - we can
run it in either SIMD4x2 or SIMD8 mode.

Treating it as SIMD8 means that shaders with 8 or fewer output vertices
(which is overwhemingly the common case) can be handled by a single
thread.  This has several intriguing properties:

- Accessing input arrays with gl_InvocationID as the index is a simple
  SIMD8 URB read with g1 as the header.  No indirect addressing required.
- Barriers are no-ops.
- We could potentially do output shadowing to combine writes, as the
  concurrency concerns are gone.  (We don't do this yet, though.)

v2: Drop first_non_payload_grf change, as it was always adding 0
    (caught by Jordan Justen).

Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>

i965: Rework the TCS passthrough shader to use NIR.

I'm about to implement a scalar TCS backend, and I'd rather not
duplicate all of this code there.

One change is that we now write the tessellation levels from all
TCS threads, rather than just the first.  This is pretty harmless,
and was easier.  The IF/ENDIF needed for that are gone; otherwise
the generated code is basically identical.

I chose to emit load/store intrinsics directly because it was easier.

Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>

gallium/util: change assertion to conditional in util_bitmask_destroy()

If we fail to create a context in the VMware driver we call this function
unconditionally to free a bunch of bit vectors.  Instead of asserting on
a null pointer, just no-op.

Reviewed-by: Jose Fonseca <jfonseca@vmware.com>

cso: null-out previously bound sampler states

If, for example, we previously had 2 sampler states bound and now we
are binding one, we'd leave the second sampler state unchanged.
This change nulls-out the second sampler state in this situation.
We're already doing the same thing for sampler views.

This silences an occasional warning issued by the VMware driver when
the number of sampler views and sampler states disagreed.

Reviewed-by: Charmaine Lee <charmainel@vmware.com>
Reviewed-by: Ilia Mirkin <imirkin@alum.mit.edu>

svga: try to flag surfaces for sampling, in addition to rendering

This silences some warnings when we try to sample from surfaces that were
created for drawing, such as when blitting from one of the framebuffer
surfaces.  We were already doing the opposite situation (adding a bind
flag for rendering to surfaces declared as texture sources).

Reviewed-by: Charmaine Lee <charmainel@vmware.com>

svga: fix copying non-zero layers of 1D array textures

Like cube maps, we need to convert the z information to a layer index.
Also rename the *_face vars to *_face_layer to make things a little more
understandable.

Reviewed-by: Charmaine Lee <charmainel@vmware.com>

svga: clean up svga_pipe_blit.c

Remove dead code.  Fix formatting.

Reviewed-by: Charmaine Lee <charmainel@vmware.com>

rbug: s/Elements/ARRAY_SIZE/

Signed-off-by: Brian Paul <brianp@vmware.com>

freedreno: s/Elements/ARRAY_SIZE/

Signed-off-by: Brian Paul <brianp@vmware.com>
Reviewed-by: Rob Clark <robdclark@gmail.com>

trace: s/Elements/ARRAY_SIZE/

Signed-off-by: Brian Paul <brianp@vmware.com>

ilo: s/Elements/ARRAY_SIZE/

Signed-off-by: Brian Paul <brianp@vmware.com>

i915g: s/Elements/ARRAY_SIZE/

Signed-off-by: Brian Paul <brianp@vmware.com>

nvc0: compute a percentage for metric-achieved_occupancy

metric-issue_slot_utilization and metric-branch_efficiency are already
computed as percentages.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>

nvc0: display some performance metrics with a percentage

This makes more sense for them.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>

nvc0: store the driver query type for performance metrics

This will allow to use percentages for some metrics because the Gallium
HUD doesn't allow to display floating point numbers and 0 is printed
instead.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>

nvc0: fix exposing of metric-issue_slots for SM21/SM30

This is most likely a copy-paste error when I reworked this area few
weeks ago. For SM20, metric-issue_slots is equal to inst_issued because
there is only one pipeline, so the metric is not exposed there.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reported-by: Karol Herbst <nouveau@karolherbst.de>

mesa/objectlabel: handle NULL src string

This prevents a crash when a NULL src is passed with a non-NULL length.

fixes: dEQP-GLES31.functional.debug.object_labels.query_length_only
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=95252

Signed-off-by: Mark Janes <mark.a.janes@intel.com>
Reviewed-by: Dave Airlie <airlied@redhat.com>

glsl: subroutine types cannot be used in constructors.

This fixes two of the cases in
GL43-CTS.shader_subroutine.subroutines_not_allowed_as_variables_constructors_and_argument_or_return_types

Reviewed-by: Iago Toral Quiroga <itoral@igalia.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>

glsl: resource is a reserved keyword in GLSL 4.20 as well

resource just appears in GLSL 4.20 without any fanfare.

Fixes GL43-CTX.CommonBugs.CommonBug_ReservedNames

Reviewed-by: Iago Toral Quiroga <itoral@igalia.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>

gallium,utils: Fix trivial sign compare warnings

Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
Reviewed-by: Jakob Sinclair <sinclair.jakob@openmailbox.org>
Reviewed-by: Jose Fonseca <jfonseca@vmware.com>

anv: fix hang during generation of dev_icd.json.

Fixes: b370ec7c76d ("anv: tweak the %.json rule")
Reviewed-by: Emil Velikov <emil.velikov@collabora.com>

swrast: Add texfetch_funcs entries for astc 3d formats

Signed-off-by: Anuj Phogat <anuj.phogat@gmail.com>
Reviewed-by: Brian Paul <brianp@vmware.com>

mesa: Enable translation between astc 3d gl formats and mesa formats

Signed-off-by: Anuj Phogat <anuj.phogat@gmail.com>
Reviewed-by: Brian Paul <brianp@vmware.com>

mesa: Handle astc 3d formats in _mesa_get_compressed_formats()

Signed-off-by: Anuj Phogat <anuj.phogat@gmail.com>
Reviewed-by: Brian Paul <brianp@vmware.com>

mesa: Handle astc 3d formats in _mesa_base_tex_format()

Signed-off-by: Anuj Phogat <anuj.phogat@gmail.com>
Reviewed-by: Brian Paul <brianp@vmware.com>

mesa: Account for astc 3d formats in _mesa_is_astc_format()

Signed-off-by: Anuj Phogat <anuj.phogat@gmail.com>
Reviewed-by: Brian Paul <brianp@vmware.com>

mesa: Add a helper function is_astc_3d_format()

Signed-off-by: Anuj Phogat <anuj.phogat@gmail.com>
Reviewed-by: Brian Paul <brianp@vmware.com>

mesa: Add the missing defines for GL_OES_texture_compression_astc

Signed-off-by: Anuj Phogat <anuj.phogat@gmail.com>
Reviewed-by: Brian Paul <brianp@vmware.com>

mesa: Align the values of #define's in glheader.h

Signed-off-by: Anuj Phogat <anuj.phogat@gmail.com>
Reviewed-by: Brian Paul <brianp@vmware.com>

mesa: Add OES_texture_compression_astc to extension table and gl_extensions

Signed-off-by: Anuj Phogat <anuj.phogat@gmail.com>
Reviewed-by: Brian Paul <brianp@vmware.com>

mesa: Add entries for astc 3d formats initializing struct gl_format_info

Signed-off-by: Anuj Phogat <anuj.phogat@gmail.com>
Reviewed-by: Brian Paul <brianp@vmware.com>

mesa: Add mesa formats for astc 3d formats

Signed-off-by: Anuj Phogat <anuj.phogat@gmail.com>
Reviewed-by: Brian Paul <brianp@vmware.com>

glapi: Update dispatch XML files for OES_texture_compression_astc.xml

Signed-off-by: Anuj Phogat <anuj.phogat@gmail.com>
Reviewed-by: Brian Paul <brianp@vmware.com>

mesa: Account for block depth in _mesa_format_image_size()

Signed-off-by: Anuj Phogat <anuj.phogat@gmail.com>
Reviewed-by: Brian Paul <brianp@vmware.com>

mesa: Handle 3d block sizes in _mesa_compute_compressed_pixelstore

Signed-off-by: Anuj Phogat <anuj.phogat@gmail.com>
Reviewed-by: Brian Paul <brianp@vmware.com>

mesa: Handle 3d block sizes in teximage error checks

Signed-off-by: Anuj Phogat <anuj.phogat@gmail.com>
Reviewed-by: Brian Paul <brianp@vmware.com>

mesa: Handle 3d block sizes in getteximage error checks

Signed-off-by: Anuj Phogat <anuj.phogat@gmail.com>
Reviewed-by: Brian Paul <brianp@vmware.com>

mesa: Add an assert for BlockDepth in _mesa_get_format_block_size()

Signed-off-by: Anuj Phogat <anuj.phogat@gmail.com>
Reviewed-by: Brian Paul <brianp@vmware.com>

mesa: Add a helper function to query 3D block sizes

Signed-off-by: Anuj Phogat <anuj.phogat@gmail.com>
Reviewed-by: Brian Paul <brianp@vmware.com>

mesa: Add block depth field in struct gl_format_info

This will be later required for 3D ASTC formats.

Signed-off-by: Anuj Phogat <anuj.phogat@gmail.com>
Reviewed-by: Brian Paul <brianp@vmware.com>

mesa/copyimage: make sure number of samples match.

This fixes
GL43-CTS.copy_image.samples_missmatch
which otherwise asserts in the radeonsi driver.

Reviewed-by: Anuj Phogat <anuj.phogat@gmail.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>

mesa/objectlabel: don't do memcpy if bufSize is 0 (v2)

This prevents GL43-CTS.khr_debug.labels_non_debug from
memcpying all over the stack and crashing.

v2: actually fix the test.

Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>

mesa/textureview: move error checks up higher

GL43-CTS.texture_view.errors checks for GL_INVALID_VALUE
here but we catch these problems in the dimensionsOK check
and return the wrong error value.

This fixes:
GL43-CTS.texture_view.errors.

Reviewed-by: Anuj Phogat <anuj.phogat@gmail.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>

gallium/radeon: remove stencil_tile_split from metadata

this is a leftover from the days when depth-stencil buffers were
allocated by the DDX

Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
Reviewed-by: Michel Dänzer <michel.daenzer@amd.com>

gallium/radeon: remove tile_mode_array_valid flags

Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
Reviewed-by: Michel Dänzer <michel.daenzer@amd.com>

winsys/amdgpu: pass PIPE_CONFIG to addrlib on texture import

This hasn't been needed, but I think we should set it.

Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
Reviewed-by: Michel Dänzer <michel.daenzer@amd.com>

winsys/amdgpu: read NUM_BANKS from buffer metadata

Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
Reviewed-by: Michel Dänzer <michel.daenzer@amd.com>

radeonsi: remove unused tile mode getters

Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
Reviewed-by: Michel Dänzer <michel.daenzer@amd.com>

radeonsi: just read tile mode arrays in SDMA setup

Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
Reviewed-by: Michel Dänzer <michel.daenzer@amd.com>

radeonsi: just read tile mode arrays in SI DMA setup

Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
Reviewed-by: Michel Dänzer <michel.daenzer@amd.com>

radeonsi: just read tile mode arrays in DB setup

Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
Reviewed-by: Michel Dänzer <michel.daenzer@amd.com>

gallium/radeon: add radeon_surf::macro_tile_index

for indexing cik_macrotile_mode_array

Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
Reviewed-by: Michel Dänzer <michel.daenzer@amd.com>

winsys/radeon: drop support for kernels lacking tile mode array queries

This will allow us to simplify a lot of code around tiling.

Kernel 3.10 is required for SI support.
Kernel 3.13 is required for CIK support.

Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
Reviewed-by: Michel Dänzer <michel.daenzer@amd.com>

st/mesa: fix blit-based GetTexImage for non-finalized textures

This fixes getteximage-depth piglit failures on radeonsi.

Cc: 11.1 11.2 <mesa-stable@lists.freedesktop.org>
Reviewed-by: Ilia Mirkin <imirkin@alum.mit.edu>
Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>

winsys/radeon: count buffer size only once

Reviewed-by: Michel Dänzer <michel.daenzer@amd.com>
Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>

winsys/amdgpu: count buffer size only once

Reviewed-by: Michel Dänzer <michel.daenzer@amd.com>
Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>

winsys/amdgpu: loosen up requirements for how much memory IBs can use

ported from winsys/radeon.

Reviewed-by: Michel Dänzer <michel.daenzer@amd.com>
Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>

radeonsi: when parsing dmesg, skip empty lines

Reviewed-by: Michel Dänzer <michel.daenzer@amd.com>
Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>

radeonsi: use the hw MSAA resolving if formats are compatible

This allows resolving RGBA into RGBX.
This should improve HL2 Lost Coast performance.

Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>

nv50,nvc0: re-bind old compute state after reading MP perf counters

This might be useful to avoid breaking the current compute state when
monitoring MP perf counters because we use a compute kernel to read out
those counters. This has been initially suggested by Ilia Mirkin.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>

nir: make lower_clamp_color pass work after lower i/o

Kinda important to work with tgsi_to_nir, which generates nir which
already has i/o lowered.

Signed-off-by: Rob Clark <robclark@freedesktop.org>

vc4: Use NIR lowering for sRGB decode.

This should get us the same decode code generated, but with a lot less
custom code in the driver.

vc4: Just use NIR lowering for texture projection.

This means doing Newton-Raphson on the RCP, but it's probably actually a
good thing to be accurate on.

vc4: Scalarize phi nodes as well.

This makes fewer programs with loops assertion fail, replacing them with
the rendering failure warning.

vc4: Add whitespace after each program stage dump.

In particular it's been hard to find the point where we switch from
dumping pre-optimization QIR and post-optimization QIR.

vc4: Remove the CSE pass.

It's not doing anything according to shader-db now that we're using NIR.
It would have had to be reworked significantly anyway, to handle control
flow.

vc4: Emit only one FRAG_Z or FRAG_W QIR opcode.

We were generating piles of FRAG_W for interpolation, only to CSE them
away immediately.  Since this is the only thing that CSE is doing for us
any more, just avoid making the CSE work necessary.

vc4: Use the NIR cubemap normalization instead of our own.

This is one of two uses of the current QIR CSE pass according to
shader-db.  The NIR pass means that we'll end up doing Newton-Raphson on
our RCP, which we weren't doing before, but that's probably actually a
good thing.

vc4: Drop the support for DCE of texture instructions.

Now that we're using NIR for our optimization, there's no need for this
tricky code.

radeonsi: fix PIPE_FORMAT_R11G11B10_FLOAT handling

That format has first_non_void < 0. This fixes a regression in piglit
arb_shader_image_load_store-semantics that was introduced by commit 76b8c5cc602,
while hopefully still shutting Coverity up (and failing in a more obvious way
if a similar error should re-appear).

Reviewed-by: Jakob Sinclair <sinclair.jakob@openmailbox.org>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>

radeonsi: correct NULL-pointer check in si_upload_const_buffer

Cc: "11.1 11.2" <mesa-stable@lists.freedesktop.org>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>

softpipe: bump 3D texture limit to 2048

The GL4.1 spec bumps this to 2048, so we should do so.

Reviewed-by: Roland Scheidegger <sroland@vmware.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>

softpipe: allow r32 xchg on shader images.

This is part of OES_shader_image_atomic.txt.

Reviewed-by: Roland Scheidegger <sroland@vmware.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>

softpipe: avoid leaking local_mem on machines alloc failure

Spotted by Coverity

Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
Reviewed-by: Vinson Lee <vlee@freedesktop.org>