anv: emit 3DSTATE_HS for each primitive on gfx12

This is Wa_16011107343, same workaround as commit 880a3efe6c2
but for gfx12.

Signed-off-by: Tapani Pälli <tapani.palli@intel.com>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21551>

iris: emit 3DSTATE_HS for each primitive on gfx12

This is Wa_16011107343, same workaround as commit a043ae8e249
but for gfx12.

Signed-off-by: Tapani Pälli <tapani.palli@intel.com>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21551>

aco: only ls and ps use store output now

Reviewed-by: Timur Kristóf <timur.kristof@gmail.com>
Signed-off-by: Qiang Yu <yuq825@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21435>

ac/llvm: only init outputs when fragment shader for radv

LS pass output to TCS by reg is not enabled when LLVM.

Reviewed-by: Timur Kristóf <timur.kristof@gmail.com>
Signed-off-by: Qiang Yu <yuq825@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21435>

radeonsi: only init llvm output when needed.

These are the cases which still uses nir_store_output.

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Signed-off-by: Qiang Yu <yuq825@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21435>

glsl: Move lower_vector_insert to GLSL-to-NIR.

We already have a nir_builder equivalent for generating this code, just
use that instead of doing it in GLSL.

No change on r300 shader-db.

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21476>

panfrost/midgard: Drop redundant arg to emit_explicit_constant.

Every caller passed the same value twice.  Just reuse it?

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21476>

panfrost/midgard: Fix handling of csel with a vector constant condition.

If it's not all true or all false, then you'll have a csel with a vector
constant, and the backend failed to translate appropriately.  Expand the
constant to fix it.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21476>

venus: fix device memory export alloc info

Signed-off-by: Yiwei Zhang <zzyiwei@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21379>

venus: fix external image creation

Signed-off-by: Yiwei Zhang <zzyiwei@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21379>

venus: simplify support for non-AHB external images

Signed-off-by: Yiwei Zhang <zzyiwei@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21379>

venus: simplify ahb image creation

Those excessive asserts are only useful when bringing up Android.

Signed-off-by: Yiwei Zhang <zzyiwei@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21379>

venus: refactor image create info pnext tracking

Signed-off-by: Yiwei Zhang <zzyiwei@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21379>

venus: remove redundant abstractions for wsi struct search

Doing that won't avoid linking wsi headers, and in fact we have already
included both android and common wsi headers. For swapchain info, it's
currently disabled by the swapchain spec version advertised on Android.

Signed-off-by: Yiwei Zhang <zzyiwei@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21379>

venus: fix external buffer creation

Move buffer create info scrub into core venus.

Signed-off-by: Yiwei Zhang <zzyiwei@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21379>

venus: distinguish external memory from mappable memory support

Mappable memory support is a must for Venus core, but the support of
such can be transparent to the driver. Thus the renderer external memory
type won't expose opaque fd type.

External memory over vtest can be exposed and the wsi support on top can
be made explicit as long as masking out the importable bit.

Signed-off-by: Yiwei Zhang <zzyiwei@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21379>

venus: tighten up the sync fd requirements for Android wsi

Signed-off-by: Yiwei Zhang <zzyiwei@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21379>

venus: refactor sync fd fence and sempahore features

Signed-off-by: Yiwei Zhang <zzyiwei@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21379>

venus: clean up globalFencing

Signed-off-by: Yiwei Zhang <zzyiwei@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21379>

venus: clean up memoryResourceAllocationSize

Signed-off-by: Yiwei Zhang <zzyiwei@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21379>

venus: start requiring all experimental features

This is to prepare for promoting exp features to core.

Signed-off-by: Yiwei Zhang <zzyiwei@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21379>

venus: replace binary search with hardcode for max buffer size

Avoid the couple test buffer creations without maintenance4.

Signed-off-by: Yiwei Zhang <zzyiwei@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21379>

venus: further disallow sparse resource

Signed-off-by: Yiwei Zhang <zzyiwei@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21379>

aco/optimizer_postRA: assume all registers are untrackable in loop headers

Register writes from the pre-header might not be correct for any but
the first loop iteration because they can be clobbered inside the loop.

Foz-DB Navi21:
Totals from 18 (0.01% of 134913) affected shaders:
CodeSize: 251384 -> 251508 (+0.05%)
Instrs: 47644 -> 47664 (+0.04%)
Latency: 801801 -> 801852 (+0.01%)
InvThroughput: 177579 -> 177593 (+0.01%)
Copies: 4752 -> 4771 (+0.40%)

Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/8376
Fixes: d3b0f781103 ("aco/optimizer_postRA: Initialize loop header with preheader information")

Reviewed-by: Timur Kristóf <timur.kristof@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21540>

zink: skip buffer barriers for ACCESS_NONE -> ACCESS_READ / ACCESS_WRITE

these do nothing and may cause unnecessary driver flushing

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21427>

ci: Update traces expectations for gutting glsl opt_algebraic.

All look like harmless changes.

Acked-by: Timothy Arceri <tarceri@itsqueeze.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21475>

glsl: Remove unused as_rvalue_to_saturate().

This is not where saturate recognition happens.  Dead code since
5598458e6923 ("i965/vec4: Remove try_emit_saturate") in 2014!

Acked-by: Timothy Arceri <tarceri@itsqueeze.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21475>

glsl/opt_algebraic: Drop the flrp recognizer.

No change to r300.

freedreno looks mixed but slightly positive in instructions:

total instructions in shared programs: 11012472 -> 11012453 (<.01%)
instructions in affected programs: 8250 -> 8231 (-0.23%)
helped: 16
HURT: 50

Acked-by: Timothy Arceri <tarceri@itsqueeze.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21475>

glsl/opt_algebraic: Drop the ftrunc pattern recognizer.

Now that it's in NIR, there's no change to r300 or freedreno shader-db
when we do.

Acked-by: Timothy Arceri <tarceri@itsqueeze.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21475>

nir: Port a floor->truncate algebraic opt pattern from GLSL.

Prevents regression when dropping code from the GLSL optimizer.

Acked-by: Timothy Arceri <tarceri@itsqueeze.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21475>

glsl/opt_algebraic: Drop some fmul simplifications.

Looks like mostly noise, trending slightly positively.

freedreno:

total instructions in shared programs: 11012781 -> 11012472 (<.01%)
instructions in affected programs: 114072 -> 113763 (-0.27%)
helped: 123
HURT: 153

r300:

total instructions in shared programs: 1338236 -> 1337897 (-0.03%)
instructions in affected programs: 3460 -> 3121 (-9.80%)
helped: 61
HURT: 11

Acked-by: Timothy Arceri <tarceri@itsqueeze.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21475>

glsl/opt_algebraic: Drop the flrp/ffma simplifiers.

NIR seems to do a better job.  Freedreno:

total instructions in shared programs: 11013096 -> 11012781 (<.01%)
instructions in affected programs: 258358 -> 258043 (-0.12%)
helped: 470
HURT: 269

r300:

total instructions in shared programs: 1338237 -> 1338236 (<.01%)
instructions in affected programs: 161 -> 160 (-0.62%)
helped: 1
HURT: 0
total presub in shared programs: 45127 -> 44881 (-0.55%)
presub in affected programs: 1719 -> 1473 (-14.31%)
helped: 246
HURT: 0

Acked-by: Timothy Arceri <tarceri@itsqueeze.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21475>

glsl/opt_algebraic: Drop no-op pack/unpack optimization.

No change on freedreno shader-db.

Acked-by: Timothy Arceri <tarceri@itsqueeze.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21475>

glsl/opt_algebraic: Drop the eq/neq add-removal optimization.

No change on freedreno or r300 shader-db.

Acked-by: Timothy Arceri <tarceri@itsqueeze.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21475>

glsl/opt_algebraic: Drop scalar all_eq/any_neq -> eq/neq opt.

No change in r300 or freedreno shader-db.

Acked-by: Timothy Arceri <tarceri@itsqueeze.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21475>

glsl/opt_algebraic: Drop fdot 0-channel optimizations.

No change on i915g shader-db.

Acked-by: Timothy Arceri <tarceri@itsqueeze.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21475>

nir: Add optimization for fdot(x, 0) -> 0.

We had all these nice fdot opts to drop individual channels that were 0,
but nothing handling it being entirely 0!  Avoids r300g regression when
dropping them from GLSL.

Acked-by: Timothy Arceri <tarceri@itsqueeze.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21475>

glsl/opt_algebraic: Drop csel(true/false, x, y) optimization.

No change on freedreno shader-db.

Acked-by: Timothy Arceri <tarceri@itsqueeze.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21475>

glsl/opt_algebraic: Drop x + -x -> 0 optimization.

No change on freedreno shader-db.

Acked-by: Timothy Arceri <tarceri@itsqueeze.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21475>

glsl/opt_algebraic: Drop add/sub with 0 optimizations.

Looks like minor instruction selection noise in freedreno shader-db:

total instructions in shared programs: 11013100 -> 11013096 (<.01%)
instructions in affected programs: 2714 -> 2710 (-0.15%)
helped: 8
HURT: 6

Acked-by: Timothy Arceri <tarceri@itsqueeze.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21475>

glsl/opt_algebraic: Drop fdiv(1,x) -> frcp(x) and fdiv(x,1) -> x optimizations.

No change on freedreno shader-db.

Acked-by: Timothy Arceri <tarceri@itsqueeze.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21475>

glsl/opt_algebraic: Drop and/or/xor optimizations.

NIR has them, and if anything freedreno shader-db prefers that NIR sees them:

total instructions in shared programs: 11013112 -> 11013100 (<.01%)
instructions in affected programs: 26266 -> 26254 (-0.05%)
helped: 4
HURT: 0

Acked-by: Timothy Arceri <tarceri@itsqueeze.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21475>

glsl/opt_algebraic: Drop rcp optimizations.

No change on freedreno shader-db.

Acked-by: Timothy Arceri <tarceri@itsqueeze.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21475>

glsl/opt_algebraic: Drop pow optimizations.

These should all be covered by NIR.  Minor shader-db changes on freedreno,
which appear to be scheduling noise.

total instructions in shared programs: 11013132 -> 11013112 (<.01%)
instructions in affected programs: 3408 -> 3388 (-0.59%)

Acked-by: Timothy Arceri <tarceri@itsqueeze.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21475>

glsl/opt_algebraic: Drop shifts of 0 optimizations.

No change on freedreno shader-db.

Acked-by: Timothy Arceri <tarceri@itsqueeze.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21475>

glsl/opt_algebraic: drop fsat(fadd(b2f(x), b2f(y))) -> b2f(ior(x, y)) opt.

No change on freedreno shader-db.

Acked-by: Timothy Arceri <tarceri@itsqueeze.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21475>

glsl/opt_algebraic: Drop f2i(trunc(x)) -> f2i(x) optimization.

No change on freedreno shader-db.

Acked-by: Timothy Arceri <tarceri@itsqueeze.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21475>

glsl/opt_algebraic: Drop -(-x) -> x optimization.

No change on freedreno shader-db.

Acked-by: Timothy Arceri <tarceri@itsqueeze.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21475>

glsl/opt_algebraic: Drop abs(-x) -> abs(x) and abs(abs(x)) -> abs(x).

NIR does this.  No change on freedreno shader-db.

Acked-by: Timothy Arceri <tarceri@itsqueeze.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21475>

glsl/opt_algebraic: Drop pow-recognizer.

NIR handles pow recognizing, too.  No change on freedreno shader-db.

Acked-by: Timothy Arceri <tarceri@itsqueeze.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21475>

glsl/opt_algebraic: Drop log(exp(x)) -> x and exp(log(x)) -> x optimisations.

No change on freedreno shader-db.

Acked-by: Timothy Arceri <tarceri@itsqueeze.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21475>

glsl/opt_algebraic: Drop ~~x == x transformation.

No change on freedreno shader-db.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21475>

zink: use tc renderpass optimizing to more optimally start queries

if tc has determined that no queries are ended inside a renderpass,
it becomes possible to start queries outside of a renderpass, which
is more performant on some hw

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21534>

zink: always start/stop/resume queries inside renderpasses

this avoids potentially splitting renderpasses by ensuring that
all (non-cs) query operations always occur inside renderpasses

zink_query_update_gs_states() now has to be called inside renderpass
to catch the active queries

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21534>

zink: pull 'was_line_loop' into ctx for query updating

make this a bit more flexible

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21534>

zink: use more consistent check for deleting zink_query::stats_list links

the linkage of this may vary depending on the query's active state,
so avoid issues by just checking the linkage directly

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21534>

zink: break out query suspend functionality for reuse

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21534>

zink: track whether queries were started in a renderpass

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21534>

zink: rework query pool overflow

now when a query pool is full, a new query pool can be created and the
previous one can be dropped from reuse to be freed at a later time

this has the added benefit of avoiding yet another place where a renderpass
might get split

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21534>

zink: create/use query pools dynamically

this fits the concept of "pools are shared between queries" a bit
better and leaves the pool ownership/destruction to their users

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21534>

zink: rework find_or_allocate_qp()

this and its caller are now a bit more streamlined with regards to
functionality of the surrounding code

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21534>

zink: reorder some query code

no functional changes

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21534>

zink: try updating qbos on query resume if !in_rp

this is another logical place to do the update since it won't split
a renderpass

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21534>

zink: only update qbo for TIME_ELAPSED on start if !in_rp

this otherwise splits the current renderpass for no reason

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21534>

zink: handle null query results for conditional render

this just needs to have a zero blasted into the buffer

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21534>

zink: handle null query results

it's possible that a query may have no results at all, so shortcut
everything and return zero as fast as possible for those cases

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21534>

zink: only try doing qbo updates on query suspend if !in_rp

this otherwise creates infinite recursion trying to end a currently
ending renderpass

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21534>

zink: refuse to start cs invocation queries in renderpass

these can't produce data in renderpasses, so ensure they aren't accidentally
started inside a renderpass where they'll have to later be restarted outside
of one

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21534>

zink: don't double suspend queries

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21534>

zink: un-suspend queries if they end while suspended

ensure they don't accidentally resume themselves after suspend

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21534>

zink: add zink_query::suspended to indicate suspended state of queries

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21534>

zink: don't auto-sync qbos on query end

these can now be synced later all at once

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21534>

zink: handle multiple query starts in qbo update

this still isn't used due to how queries are started/stopped, but it should
be a bit more robust

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21534>

zink: delete zink_query::last_start_idx

this was always zero

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21534>

zink: make zink_vk_query unref consistent

the no-check free seems suspicious

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21534>

zink: fix possible query destroy leak

if the starts array has been reset, then the counter will be inaccurate,
and some of the members will leak, so this needs to iterate over the capacity
of the array instead of the contents

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21534>

zink: fix zink_query_start initialization

not sure if this was a bug, but it would have fully zeroed the struct
and prevented previous zink_vk_query members from being unrefed

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21534>

zink: rewrite zink_query_start struct

making the non-pool members easier to zero

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21534>

zink: reset queries on the promoted cmdbuf when possible

this avoids splitting renderpasses just to reset queries

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21534>

zink: remove suspended queries from list before resuming

this avoids double-starting primgen queries from recursion when
needs_rast_discard_workaround is true

cc: mesa-stable

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21534>

zink: move zink_batch_no_rp call for query reset

this could otherwise pointlessly terminate a renderpass

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21534>

mesa/shaderapi: fix path memory leak

For instance, with "piglit/arb_shading_language_include-api -auto -fbo" or
"piglit/shader_runner tests/spec/arb_shading_language_include/execution/replacement.shader_test -auto -fbo":
Direct leak of 66 byte(s) in 6 object(s) allocated from:
    #0 0x7fa4b59050 in __interceptor_strdup (/usr/lib64/libasan.so.6+0x59050)
    #1 0x7f9a098fe0 in validate_and_tokenise_sh_incl ../src/mesa/main/shaderapi.c:3383
    #2 0x7f9a0a43e8 in _mesa_NamedStringARB ../src/mesa/main/shaderapi.c:3547

Signed-off-by: Patrick Lerda <patrick9876@free.fr>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21553>

d3d12: Don't loop in update_draw_indirect_with_sysvals

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21567>

d3d12: Create varying structures as necessary, reference them

This changes instances of d3d12_varying_info to d3d12_varying_info*,
significantly reducing the size of the d3d12_shader_key,
d3d12_gs_variant_key, and d3d12_tcs_variant_key.

Associated changes to key fill, compare, hashing, and gs and tcs variant
maps significantly reduce the amount of time spent clearing and
comparing memory.

The biggest win here is not having to re-zero _or_ re-fill varyings in
d3d12_fill_shader_key, validate_geometry_shader_variant, and
validate_tess_ctrl_shader_variant.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21527>

d3d12: Don't memcmp gs/tcs keys

This is unnecessary and incomplete. Not only is it covered by the hash
map lookup, but also does not consider the other key members.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21527>

d3d12: Don't unnecessarily zero out gs/tcs keys

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21527>

d3d12: Track max varying slot, set and compare less bytes

Often, the full range of available slots don't need to participate in
the comparison or be zeroed out.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21527>

d3d12: Don't recompute has_flat_varyings or missing_dual_src_outputs

Per shader selection, these two flags can be expensive to compute.
Instead, recompute their values as neccessary, and save them in the
context.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21527>

anv: Skip BTI RT flush if we're doing an op that doesn't use render targets.

rt_flushes emitted on zink sauer.trace --loop=500 -2.02118% +/- 1.15992% (n=8).

Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21508>

anv: Skip the RT flush when doing depth-only rendering.

The spec citation says it's just for when the RT write message BTI might
point to a different RT, and if we don't have any color attachments then
we won't have one of those at all.

Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21508>

aux/tc: only call tc_parse_draw() when parsing renderpass info

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21533>

aux/tc: track whether queries have been terminated in a renderpass

on tilers it's important to know whether a query is ended mid-renderpass
so that the query begin can occur inside/outside of the renderpass

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21533>

panfrost: Use NIR scoped barriers instead of memory barriers

Now both GLSL and SPIR-V will produce the scoped barriers, so no
need to handle the old ones.

Control barriers are still present in some cases, so keep that
for now.

Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3339>

pan/midgard: Handle nir_intrinsic_scoped_barrier in Midgard compiler

Behave the same as the existing more specific barrier intrinsics.

Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3339>

pan/compiler: Fix handling of nir_intrinsic_scoped_barrier

Only emit anything if this is a control barrier, i.e. contains an
execution scope.  Also change the assertion to look at that scope
instead of the memory scope.

Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3339>

intel/compiler: Mark various memory barriers intrinsics unreachable

Now that both SPIR-V and GLSL are using scoped barriers, we can stop
handling the specialized ones.

Reviewed-by: Faith Ekstrand <faith.ekstrand@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3339>

glsl: Implement use_scoped_barrier option for lowering memory barriers

When the option is enabled, lower memory barriers to the
unified nir_intrinsic_scoped_barrier.

The translation of the following is based on
https://www.khronos.org/registry/OpenGL/extensions/ARB/ARB_gl_spirv.txt

- memoryBarrier()
- memoryBarrierBuffer()
- memoryBarrierImage()
- memoryBarrierShared()
- groupMemoryBarrier()

Also use scoped barrier for the memory counterparts of the GLSL
(control) barrier() when the option is enabled.  The execution
part of a (control) barrier() remains using the old intrinsic.

For memoryBarrierAtomicCounter() there's no corresponding
nir_var_atomic_counter mode. Since atomic counters are lowered
to SSBOs, use the nir_var_mem_ssbo mode in the scoped barrier
instead.

Reviewed-by: Faith Ekstrand <faith.ekstrand@collabora.com>
Acked-by: Rob Clark <robclark@freedesktop.org>
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3339>

gallivm: Fix handling of nir_intrinsic_scoped_barrier

A scoped barrier only has a control barrier behavior if there's
an execution scope.

Fixes: 13d900de0dcf ("llvmpipe: set nir_shader_compiler_options::use_scoped_barrier")
Reviewed-by: Faith Ekstrand <faith.ekstrand@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3339>

microsoft/compiler: Handle scoped barrier in Tess splitting

Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3339>

nir: Support use_scoped_barrier in nir_lower_atomics_to_ssbo

Reviewed-by: Faith Ekstrand <faith.ekstrand@collabora.com>
Acked-by: Rob Clark <robclark@freedesktop.org>
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3339>