From 5ef7c5482975537ca838b51ca356f654ecf7a9ad Mon Sep 17 00:00:00 2001
From: Qiang Yu <yuq825@gmail.com>
Date: Sat, 19 Aug 2023 15:36:00 +0800
Subject: [PATCH] aco: wait memory ops done before go to next shader part
MIME-Version: 1.0
Content-Type: text/plain; charset=utf8
Content-Transfer-Encoding: 8bit
Next part don't know whether p_end_with_regs args are loaded from
memory ops or not, need to wait it's done here.
Other memory load needs to be waited too like:
a = load_mem()
b = ...
if (...) {
wait_mem(a)
store_mem(a)
}
p_end_with_regs(b)
"a" still needs to be waited, otherwise next shader part regs may
be overwritten by unfinished memory loads.
Memory stores are waited too. When >=gfx10 and last VGT has no
parameter export, we need to wait all memeory stores done before
pos export (see ac_nir_export_position). So when merged shader
(ES+GS or VS+GS) is partially built, first stage needs to wait
all memory stores done, otherwise second stage don't know if
any memory stores pending before.
Reviewed-by: Daniel Schürmann <daniel@schuermann.dev>
Signe-off-by: Qiang Yu <yuq825@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24973>
---
src/amd/compiler/aco_insert_waitcnt.cpp | 11 ++++++-----
src/amd/compiler/aco_lower_to_hw_instr.cpp | 6 ++++++
2 files changed, 12 insertions(+), 5 deletions(-)
diff --git a/src/amd/compiler/aco_insert_waitcnt.cpp b/src/amd/compiler/aco_insert_waitcnt.cpp
index 71d073965f8..6e292a9bb72 100644
--- a/src/amd/compiler/aco_insert_waitcnt.cpp
+++ b/src/amd/compiler/aco_insert_waitcnt.cpp
@@ -1082,6 +1082,12 @@ handle_block(Program* program, Block& block, wait_ctx& ctx)
}
}
+ /* For last block of a program which has succeed shader part, wait all memory ops done
+ * before go to next shader part.
+ */
+ if (block.kind & block_kind_end_with_regs)
+ force_waitcnt(ctx, queued_imm);
+
if (!queued_imm.empty())
emit_waitcnt(ctx, new_instructions, queued_imm);
if (!queued_delay.empty())
@@ -1153,11 +1159,6 @@ insert_wait_states(Program* program)
in_ctx[current.index] = ctx;
}
- if (current.instructions.empty()) {
- out_ctx[current.index] = std::move(ctx);
- continue;
- }
-
loop_progress = std::max<unsigned>(loop_progress, current.loop_nest_depth);
done[current.index] = true;
diff --git a/src/amd/compiler/aco_lower_to_hw_instr.cpp b/src/amd/compiler/aco_lower_to_hw_instr.cpp
index 480ea19dd58..edadca72bb9 100644
--- a/src/amd/compiler/aco_lower_to_hw_instr.cpp
+++ b/src/amd/compiler/aco_lower_to_hw_instr.cpp
@@ -3065,6 +3065,12 @@ lower_to_hw_instr(Program* program)
Builder bld(program, end_with_regs_block);
bld.sopp(aco_opcode::s_branch, exit_block->index);
+
+ /* For insert waitcnt pass to add waitcnt in exit block, otherwise waitcnt will be added
+ * after the s_branch which won't be executed.
+ */
+ end_with_regs_block->kind &= ~block_kind_end_with_regs;
+ exit_block->kind |= block_kind_end_with_regs;
}
}
--
2.34.1