diff options
author | Timur Kristóf <timur.kristof@gmail.com> | 2021-03-18 10:15:14 +0100 |
---|---|---|
committer | Dylan Baker <dylan.c.baker@intel.com> | 2021-03-19 13:51:02 -0700 |
commit | fed5454f5020e83efc02697a25d9bb35e34b4d53 (patch) | |
tree | b109adf3ca8cff417060b5d245f84d37761477f5 | |
parent | 6041e3b07a891fe89c3b2081ce212022f0d40d33 (diff) | |
download | mesa-fed5454f5020e83efc02697a25d9bb35e34b4d53.tar.gz |
aco: Fix constant address offset calculation for ds_read2 instructions.
Cc: mesa-stable
Signed-off-by: Timur Kristóf <timur.kristof@gmail.com>
Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9678>
(cherry picked from commit 89c8e22cc65bae26160134145b6c3ef313174713)
-rw-r--r-- | .pick_status.json | 2 | ||||
-rw-r--r-- | src/amd/compiler/aco_instruction_selection.cpp | 14 |
2 files changed, 9 insertions, 7 deletions
diff --git a/.pick_status.json b/.pick_status.json index 88623d93c96..88ab81f0200 100644 --- a/.pick_status.json +++ b/.pick_status.json @@ -670,7 +670,7 @@ "description": "aco: Fix constant address offset calculation for ds_read2 instructions.", "nominated": true, "nomination_type": 0, - "resolution": 0, + "resolution": 1, "master_sha": null, "because_sha": null }, diff --git a/src/amd/compiler/aco_instruction_selection.cpp b/src/amd/compiler/aco_instruction_selection.cpp index 9c6976a4ff7..71507db3f1f 100644 --- a/src/amd/compiler/aco_instruction_selection.cpp +++ b/src/amd/compiler/aco_instruction_selection.cpp @@ -3361,14 +3361,16 @@ Temp lds_load_callback(Builder& bld, const LoadEmitInfo &info, op = aco_opcode::ds_read_u8; } - unsigned max_offset_plus_one = read2 ? 254 * (size / 2u) + 1 : 65536; - if (const_offset >= max_offset_plus_one) { - offset = bld.vadd32(bld.def(v1), offset, Operand(const_offset / max_offset_plus_one)); - const_offset %= max_offset_plus_one; + unsigned const_offset_unit = read2 ? size / 2u : 1u; + unsigned const_offset_range = read2 ? 255 * const_offset_unit : 65536; + + if (const_offset > (const_offset_range - const_offset_unit)) { + unsigned excess = const_offset - (const_offset % const_offset_range); + offset = bld.vadd32(bld.def(v1), offset, Operand(excess)); + const_offset -= excess; } - if (read2) - const_offset /= (size / 2u); + const_offset /= const_offset_unit; RegClass rc = RegClass(RegType::vgpr, DIV_ROUND_UP(size, 4)); Temp val = rc == info.dst.regClass() && dst_hint.id() ? dst_hint : bld.tmp(rc); |