aco/gfx10: Fix VcmpxExecWARHazard mitigation.
The SOPP instruction shouldn't have a definition, and its block
should be set to -1 in order to prevent it from being recognized
as a branch.
Also fix a typo in the readme.
Fixes: d6dfce02d074d615a3b88a3fccd8ee8c7e13c010
Signed-off-by: Timur Kristóf <timur.kristof@gmail.com>
Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Tested-by: Marge Bot <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/3552>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/3552>
(cherry picked from commit c787b8d2a16d5e2950f209b1fcbec6e6c0388845)
diff --git a/.pick_status.json b/.pick_status.json
index b4c3d70..931c9f1 100644
--- a/.pick_status.json
+++ b/.pick_status.json
@@ -4,7 +4,7 @@
"description": "aco/gfx10: Fix VcmpxExecWARHazard mitigation.",
"nominated": true,
"nomination_type": 1,
- "resolution": 0,
+ "resolution": 1,
"master_sha": null,
"because_sha": "d6dfce02d074d615a3b88a3fccd8ee8c7e13c010"
},
diff --git a/src/amd/compiler/README.md b/src/amd/compiler/README.md
index 822ecd7..7c7e68f 100644
--- a/src/amd/compiler/README.md
+++ b/src/amd/compiler/README.md
@@ -154,7 +154,6 @@
Triggered by:
An SMEM instruction reads an SGPR. Then, a VALU instruction writes that same SGPR.
-Despite LLVM
Mitigated by:
Any non-SOPP SALU instruction (except `s_setvskip`, `s_version`, and any non-lgkmcnt `s_waitcnt`).
diff --git a/src/amd/compiler/aco_insert_NOPs.cpp b/src/amd/compiler/aco_insert_NOPs.cpp
index 2422db8..76b7ddd 100644
--- a/src/amd/compiler/aco_insert_NOPs.cpp
+++ b/src/amd/compiler/aco_insert_NOPs.cpp
@@ -455,9 +455,9 @@
ctx.has_nonVALU_exec_read = false;
/* Insert s_waitcnt_depctr instruction with magic imm to mitigate the problem */
- aco_ptr<SOPP_instruction> depctr{create_instruction<SOPP_instruction>(aco_opcode::s_waitcnt_depctr, Format::SOPP, 0, 1)};
+ aco_ptr<SOPP_instruction> depctr{create_instruction<SOPP_instruction>(aco_opcode::s_waitcnt_depctr, Format::SOPP, 0, 0)};
depctr->imm = 0xfffe;
- depctr->definitions[0] = Definition(sgpr_null, s1);
+ depctr->block = -1;
new_instructions.emplace_back(std::move(depctr));
} else if (instr_writes_sgpr(instr)) {
/* Any VALU instruction that writes an SGPR mitigates the problem */
