Icecream95 [Tue, 22 Dec 2020 12:54:01 +0000 (01:54 +1300)]
pan/mdg: Implement load_global_invocation_id
NIR generates global_invocation_id, but Clover lowers it to
global_invocation_id_zero_base. It isn't lowered for OpenGL,
so we need to support both intrinsics.
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8264>
Icecream95 [Tue, 22 Dec 2020 10:08:21 +0000 (23:08 +1300)]
pan/mdg: Add load_kernel_input support
Similar to uniform loads, except that the offset is in bytes instead
of vec4 units.
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8264>
Icecream95 [Mon, 7 Sep 2020 09:29:28 +0000 (21:29 +1200)]
pan/mdg: Fix promoted uniform moves with 64-bit types
The move source is the second src, not the first one.
Fixes:
5e5ea25a0da ("pan/mdg: Explicitly type 64-bit uniform moves")
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8264>
Icecream95 [Tue, 22 Dec 2020 10:03:19 +0000 (23:03 +1300)]
nir: Handle load_kernel_input in nir_get_io_offset_src
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8264>
Vinson Lee [Sat, 12 Dec 2020 03:52:30 +0000 (19:52 -0800)]
virgl: Fix typos.
Signed-off-by: Vinson Lee <vlee@freedesktop.org>
Reviewed-by: Gert Wollny <gert.wollny@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8072>
Eric Anholt [Thu, 31 Dec 2020 15:57:11 +0000 (07:57 -0800)]
ci/softpipe: Update the comment about the rasterpos flake.
The noted fix landed, but a test fix was needed.
Reviewed-by: Adam Jackson <ajax@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8275>
Eric Anholt [Wed, 30 Dec 2020 21:16:10 +0000 (13:16 -0800)]
ci/softpipe: Ban glx-multithread-texture, too.
This one also flakes on us.
Reviewed-by: Adam Jackson <ajax@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8275>
Eric Anholt [Tue, 29 Dec 2020 19:21:26 +0000 (11:21 -0800)]
ci/softpipe: Skip flaky triangle-rasterization-overdraw.
It runs a different test each run based on current time, and fails
occasionally. Until the test can get fixed, remove it.
Reviewed-by: Adam Jackson <ajax@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8275>
Daniel Schürmann [Fri, 11 Sep 2020 10:05:17 +0000 (11:05 +0100)]
nir/opt_vectorize: hash whether a swizzle accesses elements beyond the maximum vectorization factor
Swizzles that access components outside of the maximum
vector size cannot be vectorized with each other.
This patch creates different hash bins for this case.
For example accesses to .x and .y are considered different variables
compared to accesses to .z and .w for 16-bit vec2.
This prevents the vectorization of things like
vec2 16 ssa_3 = iadd ssa_1.xz, ssa_2.xz
Reviewed-by: Connor Abbott <cwabbott0@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6666>
Daniel Schürmann [Tue, 22 Sep 2020 17:16:26 +0000 (18:16 +0100)]
nir/opt_vectorize: rehash users of vectorized instructions
This ensures that chains of ALU instructions are vectorized
in a single iteration.
Reviewed-by: Connor Abbott <cwabbott0@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6666>
Daniel Schürmann [Fri, 9 Oct 2020 09:54:46 +0000 (11:54 +0200)]
nir/opt_vectorize: don't hash filtered instructions
This patch also changes nir_opt_vectorize_cb to
use only one instruction as parameter.
Reviewed-by: Connor Abbott <cwabbott0@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6666>
Daniel Schürmann [Wed, 9 Sep 2020 16:44:48 +0000 (17:44 +0100)]
nir/opt_vectorize: don't hash instructions which are already vectorized
This guarantees that the hashset contains exactly the instructions
which can be vectorized.
Reviewed-by: Connor Abbott <cwabbott0@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6666>
Daniel Schürmann [Wed, 23 Sep 2020 14:08:13 +0000 (15:08 +0100)]
nir/opt_vectorize: use a single instruction per hash entry instead of a vector
This drastically simplifies vectorization but may potentially
lead to slightly worse vectorizations.
Reviewed-by: Connor Abbott <cwabbott0@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6666>
Samuel Pitoiset [Wed, 30 Dec 2020 09:28:54 +0000 (10:28 +0100)]
nir: fix determining if an addition might overflow for phi sources
nir_addition_might_overflow() expects the parent instruction to be
an alu instr but it might be a phi instr. Fix it by assuming that
the addition might overflow.
This fixes compiler crashes with Horizon Zero Dawn.
No fossils-db changes.
Cc: mesa-stable
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Daniel Schürmann <daniel@schuermann.dev>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8268>
Alyssa Rosenzweig [Thu, 31 Dec 2020 14:57:37 +0000 (09:57 -0500)]
pan/mdg: Fix bound setting in RA for sources
The bound parameter allows us to prevent allocations from crossing
particular boundaries (typically 128-bit boundaries). For 16-bit, we
don't want to cross 64-bit boundaries, in order to keep swizzles
possible to encode. We already handle this for 16-bit destinations, but
it _also_ needs to be (redundantly) handled for 16-bit sources, in case
types don't match (for example, with a vectorized size conversion
instruction).
Fixes a few newer dEQP fails.
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8282>
Boris Brezillon [Mon, 30 Nov 2020 12:25:22 +0000 (13:25 +0100)]
pan/bi: Fix ATEST emission
The ATEST FAU index must point to the ATEST parameter datum slot
otherwise the cmdstream config is ignored.
(Rebased by Alyssa)
Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8135>
Alyssa Rosenzweig [Wed, 16 Dec 2020 19:37:17 +0000 (14:37 -0500)]
pan/bi: Remove old IR
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8135>
Alyssa Rosenzweig [Mon, 28 Dec 2020 17:12:53 +0000 (12:12 -0500)]
pan/bi: Use new instruction types
And remove now redundant casts
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8135>
Alyssa Rosenzweig [Wed, 16 Dec 2020 19:29:43 +0000 (14:29 -0500)]
pan/bi: Remove old IR instruction emit
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8135>
Alyssa Rosenzweig [Wed, 16 Dec 2020 19:28:18 +0000 (14:28 -0500)]
pan/bi: Remove old IR opcode table
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8135>
Alyssa Rosenzweig [Wed, 16 Dec 2020 19:27:38 +0000 (14:27 -0500)]
pan/bi: Remove old IR helpers
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8135>
Alyssa Rosenzweig [Wed, 16 Dec 2020 19:24:18 +0000 (14:24 -0500)]
pan/bi: Remove NIR->old IR
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8135>
Alyssa Rosenzweig [Wed, 16 Dec 2020 19:23:53 +0000 (14:23 -0500)]
pan/bi: Remove old IR scheduling
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8135>
Alyssa Rosenzweig [Wed, 16 Dec 2020 19:22:43 +0000 (14:22 -0500)]
pan/bi: Remove old IR spill code
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8135>
Alyssa Rosenzweig [Wed, 16 Dec 2020 19:15:11 +0000 (14:15 -0500)]
pan/bi: Remove old IR prints
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8135>
Alyssa Rosenzweig [Wed, 16 Dec 2020 19:11:59 +0000 (14:11 -0500)]
pan/bi: Remove packing helpers
Support code for the old IR packing.
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8135>
Alyssa Rosenzweig [Wed, 16 Dec 2020 19:07:32 +0000 (14:07 -0500)]
pan/bi: Remove old IR packs
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8135>
Alyssa Rosenzweig [Wed, 16 Dec 2020 19:06:33 +0000 (14:06 -0500)]
pan/bi: Remove combine lowering
Unused.
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8135>
Alyssa Rosenzweig [Wed, 30 Dec 2020 20:50:50 +0000 (15:50 -0500)]
pan/bi: Switch to new IR
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8135>
Alyssa Rosenzweig [Wed, 30 Dec 2020 20:50:36 +0000 (15:50 -0500)]
pan/bi: Stub FAU lowering pass
A given tuple can only load a single entry from FAU-RAM, so we need a
lowering pass before scheduling to assign FAU and lower the other
sources to moves.
Right now this only handles constants. Uniforms and general FAU are
todo.
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8135>
Alyssa Rosenzweig [Wed, 16 Dec 2020 17:58:35 +0000 (12:58 -0500)]
pan/bi: Add bi_branch, bi_jump helpers
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8135>
Alyssa Rosenzweig [Thu, 10 Dec 2020 21:05:11 +0000 (16:05 -0500)]
pan/bi: Schedule new instructions singletons
We'll let the routines coexist for a moment, just to keep the commits
digestible, but this is modified from bi_make_singleton.
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8135>
Alyssa Rosenzweig [Thu, 10 Dec 2020 21:04:49 +0000 (16:04 -0500)]
pan/bi: Add bi_message_type_for_instr helper
Greatly simplified by the use of the table.
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8135>
Alyssa Rosenzweig [Wed, 16 Dec 2020 18:35:51 +0000 (13:35 -0500)]
pan/bi: Add instruction emit for builder
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8135>
Alyssa Rosenzweig [Wed, 16 Dec 2020 18:03:04 +0000 (13:03 -0500)]
pan/bi: Add emit tex for builder
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8135>
Alyssa Rosenzweig [Wed, 16 Dec 2020 18:02:22 +0000 (13:02 -0500)]
pan/bi: Fix TEXS/TEXC check prototype
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8135>
Alyssa Rosenzweig [Wed, 16 Dec 2020 18:02:03 +0000 (13:02 -0500)]
pan/bi: Emit TEXC with builder
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8135>
Alyssa Rosenzweig [Wed, 16 Dec 2020 18:01:45 +0000 (13:01 -0500)]
pan/bi: Add builder-using helpers for TEXC structs
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8135>
Alyssa Rosenzweig [Wed, 16 Dec 2020 18:01:12 +0000 (13:01 -0500)]
pan/bi: Add TEXS emit with builder
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8135>
Alyssa Rosenzweig [Sat, 12 Dec 2020 03:33:09 +0000 (22:33 -0500)]
pan/bi: Implement jumps with the builder
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8135>
Alyssa Rosenzweig [Wed, 16 Dec 2020 18:58:12 +0000 (13:58 -0500)]
pan/bi: Implement ALU with the builder
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8135>
Alyssa Rosenzweig [Sat, 12 Dec 2020 03:00:01 +0000 (22:00 -0500)]
pan/bi: Add bi_cmpf_nir helper
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8135>
Alyssa Rosenzweig [Sat, 12 Dec 2020 02:24:59 +0000 (21:24 -0500)]
pan/bi: Add bi_nir_round helper
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8135>
Alyssa Rosenzweig [Sat, 12 Dec 2020 01:30:19 +0000 (20:30 -0500)]
pan/bi: Add bi_alu_src_index helper
Generates bi_index from nir_alu_src, taking into account the applied
swizzle, and using (swizzle / 32-bit) portion as an offset, to be
applied later during RA. The sub 32-bit portion only applies for 8-bit
and 16-bit instructions, which need to either handle them explicitly as
a swizzle specifier, or lower to a swizzle explicitly.
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8135>
Alyssa Rosenzweig [Wed, 16 Dec 2020 18:35:42 +0000 (13:35 -0500)]
pan/bi: Add intrinsic emits for builder
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8135>
Alyssa Rosenzweig [Mon, 21 Dec 2020 22:35:28 +0000 (17:35 -0500)]
pan/bi: Implement load attribute with the builder
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8135>
Alyssa Rosenzweig [Sat, 12 Dec 2020 03:49:54 +0000 (22:49 -0500)]
pan/bi: Implement frag coord with the builder
Much slicker :-)
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8135>
Alyssa Rosenzweig [Sat, 12 Dec 2020 02:00:26 +0000 (21:00 -0500)]
pan/bi: Implement load_ubo with the builder
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8135>
Alyssa Rosenzweig [Wed, 9 Dec 2020 14:00:04 +0000 (09:00 -0500)]
pan/bi: Implement store_vary with the builder
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8135>
Alyssa Rosenzweig [Wed, 9 Dec 2020 13:42:41 +0000 (08:42 -0500)]
pan/bi: Implement fragment_out by builder
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8135>
Alyssa Rosenzweig [Wed, 9 Dec 2020 13:42:26 +0000 (08:42 -0500)]
pan/bi: Implement BLEND by builder
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8135>
Alyssa Rosenzweig [Tue, 8 Dec 2020 13:39:20 +0000 (08:39 -0500)]
pan/bi: Implement load_vary via the builder
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8135>
Alyssa Rosenzweig [Tue, 8 Dec 2020 13:39:31 +0000 (08:39 -0500)]
pan/bi: Implement bi_reg_fmt_for_nir helper
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8135>
Alyssa Rosenzweig [Tue, 8 Dec 2020 13:39:07 +0000 (08:39 -0500)]
pan/bi: Implement load_blend_input via the builder
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8135>
Alyssa Rosenzweig [Tue, 8 Dec 2020 00:15:24 +0000 (19:15 -0500)]
pan/bi: Implement bi_emit_load_const via the builder
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8135>
Alyssa Rosenzweig [Tue, 8 Dec 2020 00:15:10 +0000 (19:15 -0500)]
pan/bi: Implement bi_load_sysval via the builder
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8135>
Alyssa Rosenzweig [Tue, 8 Dec 2020 00:14:49 +0000 (19:14 -0500)]
pan/bi: Implement bi_emit_ld_tile via the builder
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8135>
Alyssa Rosenzweig [Wed, 9 Dec 2020 19:06:15 +0000 (14:06 -0500)]
pan/bi: Add bi_make_vec_to helper
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8135>
Alyssa Rosenzweig [Wed, 9 Dec 2020 13:58:57 +0000 (08:58 -0500)]
pan/bi: Add bi_is_intr_immediate helper
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8135>
Alyssa Rosenzweig [Wed, 16 Dec 2020 17:50:07 +0000 (12:50 -0500)]
pan/bi: Add builder initialization helper
When going out of NIR, it's useful to start inserting instructions at
the end of the current block.
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8135>
Alyssa Rosenzweig [Fri, 11 Dec 2020 15:52:11 +0000 (10:52 -0500)]
pan/bi: Add new style read/writemask helpers
In the medium term we'll want to overhaul these masks entirely since
they don't make much sense anymore, but as a stop gap, this will let us
reuse the existing liveness, RA, and DCE infrastructure.
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8135>
Alyssa Rosenzweig [Fri, 11 Dec 2020 15:30:06 +0000 (10:30 -0500)]
pan/bi: Add bi_count_staging_registers helper
This will shortly be used for liveness analysis on the new IR.
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8135>
Alyssa Rosenzweig [Fri, 4 Dec 2020 19:54:00 +0000 (14:54 -0500)]
pan/bi: Generate instruction packer for new IR
Based on existing packing infrastructure, though modifier handling is
greatly simplified by using canonical modifier encodings in the IR.
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8135>
Alyssa Rosenzweig [Sat, 28 Nov 2020 00:01:00 +0000 (19:01 -0500)]
pan/bi: Generate builder routines
To simplify construction of Bifrost IR. Ideas from NIR's builder, as
well as IBC.
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8135>
Alyssa Rosenzweig [Fri, 27 Nov 2020 19:39:40 +0000 (14:39 -0500)]
pan/bi: Generate instruction printer
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8135>
Alyssa Rosenzweig [Mon, 28 Dec 2020 14:26:07 +0000 (09:26 -0500)]
pan/bi: Add bi_not alias of bi_neg
For bitwise ops, we define .neg to mean .not to keep the IR manageable
("complement" more generally), but it's convenient to have specialized
names.
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8135>
Alyssa Rosenzweig [Mon, 28 Dec 2020 13:47:48 +0000 (08:47 -0500)]
pan/bi: Rename bi_load
Need to free up the symbol so it doesn't clash with the builder.
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8135>
Alyssa Rosenzweig [Fri, 4 Dec 2020 20:06:48 +0000 (15:06 -0500)]
pan/bi: Rename bi_pack_{fma, add} to free up symbols
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8135>
Alyssa Rosenzweig [Tue, 22 Dec 2020 23:33:56 +0000 (18:33 -0500)]
pan/bi: Ensure fneg of a constant isn't reached
Fix the optimization loop like ir3 does.
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8135>
Alyssa Rosenzweig [Thu, 31 Dec 2020 14:20:17 +0000 (09:20 -0500)]
ci/panfrost: Skip test with 4096 byte shader
This has flake potential. There is clearly a hardware bug here but I
can't figure out what the correct solution is (padding out with nops
causes other tests to flake in turn).
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8135>
Mike Blumenkrantz [Mon, 28 Dec 2020 20:47:37 +0000 (15:47 -0500)]
features: mark off multidraw for zink
Reviewed-by: Hoe Hao Cheng <haochengho12907@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8233>
Mike Blumenkrantz [Thu, 20 Aug 2020 13:22:37 +0000 (09:22 -0400)]
zink: enable PIPE_CAP_MULTI_DRAW_INDIRECT(_PARAMS) caps
Reviewed-by: Erik Faye-Lund <erik.faye-lund@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8233>
Mike Blumenkrantz [Thu, 20 Aug 2020 13:42:52 +0000 (09:42 -0400)]
zink: hook up IndirectCount draw commands
these take a draw count buffer
Reviewed-by: Erik Faye-Lund <erik.faye-lund@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8233>
Mike Blumenkrantz [Mon, 28 Dec 2020 20:39:18 +0000 (15:39 -0500)]
zink: add KHR_draw_indirect_count detection
this is a vk 1.2 feature
Reviewed-by: Hoe Hao Cheng <haochengho12907@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8233>
Samuel Pitoiset [Mon, 28 Dec 2020 07:44:03 +0000 (08:44 +0100)]
radv: disable stippledBresenhamLines on GFX9
Some CTS fail on Vega10 but work on Raven.
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8242>
Eric Anholt [Wed, 30 Dec 2020 20:51:04 +0000 (12:51 -0800)]
softpipe: Sanity check that the SSBO view offset is within the BO.
If we're going to check that the size matches, we should consider the
buffer offset too. I haven't tracked down any testcases doing this, but
it seems obviously correct.
Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8276>
Eric Anholt [Wed, 30 Dec 2020 20:44:42 +0000 (12:44 -0800)]
gallium/tgsi_exec: Replace the SSBO RESQ-specific interface with lookup.
We've got the new lookup with size+ptr, just use that one for querying
buffer size.
This means we now return 0 instead of undefined for unbound buffers, but
it also means we return 0 for a buffer view with a size larger than that
of the underlying buffer.
Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8276>
Eric Anholt [Wed, 30 Dec 2020 20:35:54 +0000 (12:35 -0800)]
gallium/tgsi_exec: Move the SSBO store path to tgsi_exec, too.
Now that we have lookups, we can just fold this in.
Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8276>
Eric Anholt [Wed, 30 Dec 2020 20:17:21 +0000 (12:17 -0800)]
gallium/tgsi_exec: Use the new SSBO lookup interface for SSBO loads.
We were ignoring the requested size of the load in the overflow handling
and would read past the end of buffers, rather than just returning 0 as
robustness would like us to do.
Fixes valgrind complaint on softpipe in:
EQP-GLES31.functional.shaders.builtin_functions.common.sign.float_mediump_compute
Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8276>
Eric Anholt [Wed, 30 Dec 2020 19:59:15 +0000 (11:59 -0800)]
gallium/tgsi_exec: Reuse the atomic helper for SSBO atomics.
I needed to refactor SSBOs to deal with a buffer overflow, and it's easier
to just delete the SSBO atomic code from the interface.
Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8276>
Eric Anholt [Wed, 30 Dec 2020 19:21:49 +0000 (11:21 -0800)]
gallium/tgsi_exec: Fix shared memory atomic ops.
It was only executing the first channel, ignoring the rest. I also
cleaned things up to not loop over rgba, since atomics are only ever to a
single 32-bit value per invocation.
This worked on softpipe previously because it only dispatches 1 CS
invocation per TGSI exec machine anyway, wasting the other 3 slots.
Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8276>
Eric Anholt [Thu, 31 Dec 2020 00:09:13 +0000 (16:09 -0800)]
ci/softpipe: Re-enable GS tests that had been banned for being flaky.
Now that the flakes should be fixed, turn the coverage on.
Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8277>
Eric Anholt [Thu, 31 Dec 2020 00:09:13 +0000 (16:09 -0800)]
gallium/draw: Fix intermittent failure to bind new geometry shaders.
If you deleted your old GS and created a new one, then it would
occasionally skip binding the new GS because the token pointers were
equal. Clear the current token pointer in the machine when we're deleting
its token.
Cc: mesa-stable
Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8277>
Dylan Baker [Thu, 31 Dec 2020 04:19:51 +0000 (20:19 -0800)]
docs: update calendar and link releases notes for 20.3.2
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8280>
Dylan Baker [Thu, 31 Dec 2020 04:11:04 +0000 (20:11 -0800)]
docs: Add sha256 sum for 20.3.2
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8280>
Dylan Baker [Thu, 31 Dec 2020 03:57:40 +0000 (19:57 -0800)]
docs: add release notes for 20.3.2
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8280>
Mike Blumenkrantz [Thu, 24 Dec 2020 22:14:14 +0000 (17:14 -0500)]
zink: fix tess shader i/o variables
this was a mismerge during refactoring for review
Fixes:
d09f9da4c4d ("zink: add ntv handling for tess shader i/o variables")
Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8232>
Mike Blumenkrantz [Wed, 30 Dec 2020 21:14:54 +0000 (16:14 -0500)]
zink: require KHR_maintenance2 for tessellation and set bottom-left origin
this makes tessellation work as expected in apps but has no impact on unit tests
Reviewed-by: Dave Airlie <airlied@redhat.com>
Reviewed-by: Witold Baryluk <witold.baryluk@gmail.com>
Tested-by: Witold Baryluk <witold.baryluk@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8274>
Adam Jackson [Wed, 9 Dec 2020 22:22:04 +0000 (17:22 -0500)]
swrast: Remove the classic swrast DRI driver
Following up from the classic swrast OSMesa removal in favor of
llvmpipe/softpipe, remove the classic swrast DRI driver. It's unused by
any distribution -- debian was the last holdout until this week, which had
it enabled instead of softpipe on obscure platforms (non-LLVM and hurd).
Now that debian has switched, remove the driver so nobody can accidentally
enable it again.
Fixes: #325, #324, #322, #321, #319, #318, #317, #315, #313, #311, #310,
#306, #305, #304, #302, #301, #300, #299, #298, #297, #293, #289
Reviewed-by: Eric Anholt <eric@anholt.net> (commit message rewritten)
Reviewed-by: Matt Turner <mattst88@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8069>
Nanley Chery [Mon, 28 Dec 2020 17:31:33 +0000 (09:31 -0800)]
iris: Support clears in more GPU-based copies
Commit
7779b1d71bf053f0c73a1b717e6d2ed91f948378, disabled clear support
when copying to/from color buffers. According to the performance CI, it
falls within a range of commits that introduced a performance regression
on Bioshock Infinite with Tigerlake. Icelake isn't noticeably affected.
By analyzing a trace of the game, I found a couple cases where that
commit added new partial resolves. Update get_copy_region_aux_settings
to avoid them:
- The trace uploads to R8_UNORM textures. On TGL, these enter the
COMPRESSED_CLEAR state on the upload and are partially resolved before
every subsequent upload. Thankfully, they keep their initial clear
color of all zeroes. Since zeros can survive format reinterpretation,
allow clear support for it.
- The trace copies between RGBA16_FLOAT textures. The ones with zero
clear color are helped by the optimization above. The ones with
non-zero clear color are used as source textures. Thankfully on ICL+,
the clear color used for sampling is in pixel form and can thus be
sampled from with format reinterpretation. Allow clear support for
this case.
I haven't tested the actual performance impact of this change, but it
should be beneficial regardless.
Reported-by: Clayton Craft <clayton.a.craft@intel.com>
Reported-by: Jordan Justen <jordan.l.justen@intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8262>
Nanley Chery [Mon, 28 Dec 2020 18:27:13 +0000 (10:27 -0800)]
iris: Move STC case in get_copy_region_aux_settings
Combine the STC_CCS case with the HiZ cases.
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8262>
Eric Anholt [Tue, 29 Dec 2020 19:44:26 +0000 (11:44 -0800)]
ci/panfrost: Disable the flaky gimark trace.
Fixes:
dcb8f625543e ("ci: add piglit jobs to LAVA and remove tracie ones")
Acked-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8263>
Vinson Lee [Thu, 24 Dec 2020 04:14:42 +0000 (20:14 -0800)]
swr: Fix typos.
Signed-off-by: Vinson Lee <vlee@freedesktop.org>
Reviewed-by: Adam Jackson <ajax@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8223>
Tony Wasserka [Tue, 29 Dec 2020 16:29:41 +0000 (17:29 +0100)]
aco/ra: Avoid redundant RegisterFile copies in get_reg_impl
Now that this function does not block RegisterFile entries anymore,
the temporary copy is only needed upon reaching the collect_vars call.
Reviewed-by: Daniel Schürmann <daniel@schuermann.dev>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8261>
Mike Blumenkrantz [Wed, 30 Dec 2020 14:55:58 +0000 (09:55 -0500)]
zink: avoid replacing valid tcs with injected one
need to check both arrays here since we only want to use an injected shader
if we don't have a real one, not just if the real one isn't being updated
Fixes:
334759d8509 ("zink: implement passthrough tcs shader injection")
Reviewed-by: Erik Faye-Lund <erik.faye-lund@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8271>
Danylo Piliaiev [Fri, 4 Dec 2020 14:44:41 +0000 (16:44 +0200)]
freedreno/a6xx: support layered framebuffers in blitter_clear
Clear is done with one instanced draw call, where the layer
to clear is controlled by gl_Layer.
Same as how util_blitter_clear does this.
Fixes test:
gl-3.2-layered-rendering-clear-color-all-types 2d_multisample_array single_level
Signed-off-by: Danylo Piliaiev <dpiliaiev@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7919>
Danylo Piliaiev [Fri, 4 Dec 2020 14:39:04 +0000 (16:39 +0200)]
freedreno/a6xx: add support for gl_Layer in vertex shader
Passes amd_vertex_shader_layer-layered-2d-texture-render
Don't enable GL_AMD_vertex_shader_layer because we do not pass
amd_vertex_shader_layer-layered-depth-texture-render due to
the assert:
emit_blit: Assertion `psurf->u.tex.first_layer == psurf->u.tex.last_layer'
However, in current state it is still useful for clearing
of arrayed framebuffers.
Signed-off-by: Danylo Piliaiev <dpiliaiev@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7919>
Danylo Piliaiev [Wed, 16 Dec 2020 16:08:04 +0000 (18:08 +0200)]
freedreno/a6xx: fix array pitch for layer-first layouts
Fixes:
gl-3.2-layered-rendering-gl-layer-render
gl-3.2-layered-rendering-gl-layer-render-clipped
gl-3.2-layered-rendering-gl-layer-render-storage
gl-3.2-layered-rendering-gl-layer-cube-map
Would fix:
amd_vertex_shader_layer-layered-2d-texture-render
when GL_AMD_vertex_shader_layer is enabled.
Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/3929
Signed-off-by: Danylo Piliaiev <dpiliaiev@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7919>
Gert Wollny [Mon, 28 Dec 2020 16:55:58 +0000 (17:55 +0100)]
doc: virgl supports ARB_texture_filter_anisotropic already
Signed-off-by: Gert Wollny <gert.wollny@collabora.com>
Reviewed-by: Adam Jackson <ajax@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8257>
Samuel Pitoiset [Fri, 11 Dec 2020 14:44:29 +0000 (15:44 +0100)]
radv: add a Python script to check if a VA was ever valid
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7891>