[AMDGPU] Emit backend_stack_size PAL metadata #72509
Merged
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
For chain functions, PAL uses a `backend_stack_size` metadata item, which at the moment has the same meaning as `stack_frame_size_in_bytes`. We emit both for now in order to simplify coordination with PAL. The new item must be emitted in the `shader_functions` section, just as the metadata for other module entry functions. For simplicity, we mark chain functions as module entry functions and emit the same metadata for all of them.
rovka
requested review from
jayfoad,
arsenm,
nhaehnle,
ruiling and
jasilvanus
|
@llvm/pr-subscribers-backend-amdgpu Author: Diana (rovka) ChangesFor chain functions, PAL uses a The new item must be emitted in the Full diff: https://github.com/llvm/llvm-project/pull/72509.diff 4 Files Affected:
diff --git a/llvm/lib/Target/AMDGPU/Utils/AMDGPUBaseInfo.cpp b/llvm/lib/Target/AMDGPU/Utils/AMDGPUBaseInfo.cpp
index a09abc639d7590f..fdc59281c50d0b3 100644
--- a/llvm/lib/Target/AMDGPU/Utils/AMDGPUBaseInfo.cpp
+++ b/llvm/lib/Target/AMDGPU/Utils/AMDGPUBaseInfo.cpp
@@ -1910,7 +1910,7 @@ bool isModuleEntryFunctionCC(CallingConv::ID CC) {
case CallingConv::AMDGPU_Gfx:
return true;
default:
- return isEntryFunctionCC(CC);
+ return isEntryFunctionCC(CC) || isChainCC(CC);
}
}
diff --git a/llvm/lib/Target/AMDGPU/Utils/AMDGPUPALMetadata.cpp b/llvm/lib/Target/AMDGPU/Utils/AMDGPUPALMetadata.cpp
index 278341944c52305..48d0bde139d7728 100644
--- a/llvm/lib/Target/AMDGPU/Utils/AMDGPUPALMetadata.cpp
+++ b/llvm/lib/Target/AMDGPU/Utils/AMDGPUPALMetadata.cpp
@@ -244,6 +244,7 @@ void AMDGPUPALMetadata::setScratchSize(CallingConv::ID CC, unsigned Val) {
void AMDGPUPALMetadata::setFunctionScratchSize(StringRef FnName, unsigned Val) {
auto Node = getShaderFunction(FnName);
Node[".stack_frame_size_in_bytes"] = MsgPackDoc.getNode(Val);
+ Node[".backend_stack_size"] = MsgPackDoc.getNode(Val);
}
// Set the amount of LDS used in bytes in the metadata.
diff --git a/llvm/test/CodeGen/AMDGPU/amdpal-callable.ll b/llvm/test/CodeGen/AMDGPU/amdpal-callable.ll
index 0061c0a3118ad94..b7b2cb22c1b6269 100644
--- a/llvm/test/CodeGen/AMDGPU/amdpal-callable.ll
+++ b/llvm/test/CodeGen/AMDGPU/amdpal-callable.ll
@@ -146,12 +146,14 @@ attributes #0 = { nounwind }
; GCN-NEXT: 0x2e13 (COMPUTE_PGM_RSRC2): 0x8001{{$}}
; GCN-NEXT: .shader_functions:
; GCN-NEXT: dynamic_stack:
+; GCN-NEXT: .backend_stack_size: 0x10{{$}}
; GCN-NEXT: .lds_size: 0{{$}}
; GCN-NEXT: .sgpr_count: 0x28{{$}}
; GCN-NEXT: .stack_frame_size_in_bytes: 0x10{{$}}
; SDAG-NEXT: .vgpr_count: 0x2{{$}}
; GISEL-NEXT: .vgpr_count: 0x3{{$}}
; GCN-NEXT: dynamic_stack_loop:
+; GCN-NEXT: .backend_stack_size: 0x10{{$}}
; GCN-NEXT: .lds_size: 0{{$}}
; SDAG-NEXT: .sgpr_count: 0x25{{$}}
; GISEL-NEXT: .sgpr_count: 0x26{{$}}
@@ -159,71 +161,84 @@ attributes #0 = { nounwind }
; SDAG-NEXT: .vgpr_count: 0x3{{$}}
; GISEL-NEXT: .vgpr_count: 0x4{{$}}
; GCN-NEXT: multiple_stack:
+; GCN-NEXT: .backend_stack_size: 0x24{{$}}
; GCN-NEXT: .lds_size: 0{{$}}
; GCN-NEXT: .sgpr_count: 0x21{{$}}
; GCN-NEXT: .stack_frame_size_in_bytes: 0x24{{$}}
; GCN-NEXT: .vgpr_count: 0x3{{$}}
; GCN-NEXT: no_stack:
+; GCN-NEXT: .backend_stack_size: 0{{$}}
; GCN-NEXT: .lds_size: 0{{$}}
; GCN-NEXT: .sgpr_count: 0x20{{$}}
; GCN-NEXT: .stack_frame_size_in_bytes: 0{{$}}
; GCN-NEXT: .vgpr_count: 0x1{{$}}
; GCN-NEXT: no_stack_call:
+; GCN-NEXT: .backend_stack_size: 0x10{{$}}
; GCN-NEXT: .lds_size: 0{{$}}
; GCN-NEXT: .sgpr_count: 0x25{{$}}
; GCN-NEXT: .stack_frame_size_in_bytes: 0x10{{$}}
; GCN-NEXT: .vgpr_count: 0x3{{$}}
; GCN-NEXT: no_stack_extern_call:
+; GCN-NEXT: .backend_stack_size: 0x10{{$}}
; GCN-NEXT: .lds_size: 0{{$}}
; GFX8-NEXT: .sgpr_count: 0x28{{$}}
; GFX9-NEXT: .sgpr_count: 0x2c{{$}}
; GCN-NEXT: .stack_frame_size_in_bytes: 0x10{{$}}
; GCN-NEXT: .vgpr_count: 0x2b{{$}}
; GCN-NEXT: no_stack_extern_call_many_args:
+; GCN-NEXT: .backend_stack_size: 0x90{{$}}
; GCN-NEXT: .lds_size: 0{{$}}
; GFX8-NEXT: .sgpr_count: 0x28{{$}}
; GFX9-NEXT: .sgpr_count: 0x2c{{$}}
; GCN-NEXT: .stack_frame_size_in_bytes: 0x90{{$}}
; GCN-NEXT: .vgpr_count: 0x2b{{$}}
; GCN-NEXT: no_stack_indirect_call:
+; GCN-NEXT: .backend_stack_size: 0x10{{$}}
; GCN-NEXT: .lds_size: 0{{$}}
; GFX8-NEXT: .sgpr_count: 0x28{{$}}
; GFX9-NEXT: .sgpr_count: 0x2c{{$}}
; GCN-NEXT: .stack_frame_size_in_bytes: 0x10{{$}}
; GCN-NEXT: .vgpr_count: 0x2b{{$}}
; GCN-NEXT: simple_lds:
+; GCN-NEXT: .backend_stack_size: 0{{$}}
; GCN-NEXT: .lds_size: 0x100{{$}}
; GCN-NEXT: .sgpr_count: 0x20{{$}}
; GCN-NEXT: .stack_frame_size_in_bytes: 0{{$}}
; GCN-NEXT: .vgpr_count: 0x1{{$}}
; GCN-NEXT: simple_lds_recurse:
+; GCN-NEXT: .backend_stack_size: 0x10{{$}}
; GCN-NEXT: .lds_size: 0x100{{$}}
; GCN-NEXT: .sgpr_count: 0x28{{$}}
; GCN-NEXT: .stack_frame_size_in_bytes: 0x10{{$}}
; GCN-NEXT: .vgpr_count: 0x29{{$}}
; GCN-NEXT: simple_stack:
+; GCN-NEXT: .backend_stack_size: 0x14{{$}}
; GCN-NEXT: .lds_size: 0{{$}}
; GCN-NEXT: .sgpr_count: 0x21{{$}}
; GCN-NEXT: .stack_frame_size_in_bytes: 0x14{{$}}
; GCN-NEXT: .vgpr_count: 0x2{{$}}
; GCN-NEXT: simple_stack_call:
+; GCN-NEXT: .backend_stack_size: 0x20{{$}}
; GCN-NEXT: .lds_size: 0{{$}}
; GCN-NEXT: .sgpr_count: 0x25{{$}}
; GCN-NEXT: .stack_frame_size_in_bytes: 0x20{{$}}
; GCN-NEXT: .vgpr_count: 0x4{{$}}
; GCN-NEXT: simple_stack_extern_call:
+; GCN-NEXT: .backend_stack_size: 0x20{{$}}
; GCN-NEXT: .lds_size: 0{{$}}
; GFX8-NEXT: .sgpr_count: 0x28{{$}}
; GFX9-NEXT: .sgpr_count: 0x2c{{$}}
; GCN-NEXT: .stack_frame_size_in_bytes: 0x20{{$}}
; GCN-NEXT: .vgpr_count: 0x2b{{$}}
; GCN-NEXT: simple_stack_indirect_call:
+; GCN-NEXT: .backend_stack_size: 0x20{{$}}
; GCN-NEXT: .lds_size: 0{{$}}
; GFX8-NEXT: .sgpr_count: 0x28{{$}}
; GFX9-NEXT: .sgpr_count: 0x2c{{$}}
; GCN-NEXT: .stack_frame_size_in_bytes: 0x20{{$}}
; GCN-NEXT: .vgpr_count: 0x2b{{$}}
; GCN-NEXT: simple_stack_recurse:
+; GCN-NEXT: .backend_stack_size: 0x20{{$}}
; GCN-NEXT: .lds_size: 0{{$}}
; GCN-NEXT: .sgpr_count: 0x28{{$}}
; GCN-NEXT: .stack_frame_size_in_bytes: 0x20{{$}}
diff --git a/llvm/test/CodeGen/AMDGPU/amdpal-chain-metadata.ll b/llvm/test/CodeGen/AMDGPU/amdpal-chain-metadata.ll
new file mode 100644
index 000000000000000..044abe15c5d8cdd
--- /dev/null
+++ b/llvm/test/CodeGen/AMDGPU/amdpal-chain-metadata.ll
@@ -0,0 +1,31 @@
+; RUN: llc -mtriple=amdgcn--amdpal -mcpu=gfx1100 <%s | FileCheck %s
+
+; CHECK-LABEL: {{^}}amdgpu_cs_chain_func:
+; CHECK: .amdgpu_pal_metadata
+; CHECK-NEXT: ---
+; CHECK-NEXT: amdpal.pipelines:
+; CHECK-NEXT: - .api: Vulkan
+; CHECK: .shader_functions:
+; CHECK-NEXT: amdgpu_cs_chain_func:
+; CHECK: .backend_stack_size: 0x10{{$}}
+; CHECK: .stack_frame_size_in_bytes: 0x10{{$}}
+; CHECK:amdpal.version:
+; CHECK-NEXT: - 0x3
+; CHECK-NEXT: - 0
+; CHECK-NEXT:...
+; CHECK-NEXT: .end_amdgpu_pal_metadata
+
+define amdgpu_cs_chain void @amdgpu_cs_chain_func(<40 x i32> %should_spill) {
+.entry:
+ %v = alloca [3 x i32], addrspace(5)
+ store i32 42, ptr addrspace(5) %v
+ call amdgpu_gfx void @use(<40 x i32> %should_spill, ptr addrspace(5) %v)
+ ret void
+}
+
+declare amdgpu_gfx void @use(...)
+
+!amdgpu.pal.metadata.msgpack = !{!0}
+
+!0 = !{!"\82\B0amdpal.pipelines\91\8A\A4.api\A6Vulkan\B2.compute_registers\85\AB.tg_size_en\C3\AA.tgid_x_en\C2\AA.tgid_y_en\C2\AA.tgid_z_en\C2\AF.tidig_comp_cnt\01\B0.hardware_stages\81\A3.cs\8C\AF.checksum_value\CE\94D\D7\D0\AB.debug_mode\00\AB.float_mode\CC\C0\A9.image_op\C2\AC.mem_ordered\C3\AB.sgpr_limitj\B7.threadgroup_dimensions\93\01\CD\04\00\01\AD.trap_present\00\B2.user_data_reg_map\DC\00 \CE\10\00\00\00\CE\FF\FF\FF\FF\00\CE\FF\FF\FF\FF\CE\FF\FF\FF\FF\CE\FF\FF\FF\FF\CE\FF\FF\FF\FF\CE\FF\FF\FF\FF\CE\FF\FF\FF\FF\CE\FF\FF\FF\FF\CE\FF\FF\FF\FF\CE\FF\FF\FF\FF\CE\FF\FF\FF\FF\CE\FF\FF\FF\FF\CE\FF\FF\FF\FF\CE\FF\FF\FF\FF\CE\FF\FF\FF\FF\CE\FF\FF\FF\FF\CE\FF\FF\FF\FF\CE\FF\FF\FF\FF\CE\FF\FF\FF\FF\CE\FF\FF\FF\FF\CE\FF\FF\FF\FF\CE\FF\FF\FF\FF\CE\FF\FF\FF\FF\CE\FF\FF\FF\FF\CE\FF\FF\FF\FF\CE\FF\FF\FF\FF\CE\FF\FF\FF\FF\CE\FF\FF\FF\FF\CE\FF\FF\FF\FF\CE\FF\FF\FF\FF\AB.user_sgprs\03\AB.vgpr_limit\CD\01\00\AF.wavefront_size@\B7.internal_pipeline_hash\92\CF\E7\10k\A6:\A6%\F7\CF\B2\1F\1A\D4{\DA\E1T\AA.registers\80\A8.shaders\81\A8.compute\82\B0.api_shader_hash\92\CF\E9Zn7}\1E\B9\E7\00\B1.hardware_mapping\91\A3.cs\B0.spill_threshold\CE\FF\FF\FF\FF\A5.type\A2Cs\B0.user_data_limit\01\AF.xgl_cache_info\82\B3.128_bit_cache_hash\92\CF\B4X\B8\11[\A4\88P\CF\A0;\B0\AF\FF\B4\BE\C0\AD.llpc_version\A461.1\AEamdpal.version\92\03\00"}
+!1 = !{i32 7}
|
nhaehnle
approved these changes
Nov 17, 2023
sr-tream
pushed a commit
to sr-tream/llvm-project
that referenced
this pull request
Nov 20, 2023
For chain functions, PAL uses a `backend_stack_size` metadata item, which at the moment has the same meaning as `stack_frame_size_in_bytes`. We emit both for now in order to simplify coordination with PAL. The new item must be emitted in the `shader_functions` section, just as the metadata for other module entry functions. For simplicity, we mark chain functions as module entry functions and emit the same metadata for all of them.
zahiraam
pushed a commit
to zahiraam/llvm-project
that referenced
this pull request
Nov 20, 2023
For chain functions, PAL uses a `backend_stack_size` metadata item, which at the moment has the same meaning as `stack_frame_size_in_bytes`. We emit both for now in order to simplify coordination with PAL. The new item must be emitted in the `shader_functions` section, just as the metadata for other module entry functions. For simplicity, we mark chain functions as module entry functions and emit the same metadata for all of them.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
For chain functions, PAL uses a
backend_stack_sizemetadata item, which at the moment has the same meaning asstack_frame_size_in_bytes. We emit both for now in order to simplify coordination with PAL.The new item must be emitted in the
shader_functionssection, just as the metadata for other module entry functions. For simplicity, we mark chain functions as module entry functions and emit the same metadata for all of them.