-
Notifications
You must be signed in to change notification settings - Fork 14.3k
[AMDGPU][NFC] Use templates to decode AV operands. #79313
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Merged
Merged
Conversation
This file contains hidden or 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
Eliminates the need to define them manually. Part of <llvm#62629>.
@llvm/pr-subscribers-backend-amdgpu Author: Ivan Kosarev (kosarev) ChangesEliminates the need to define them manually. Part of <#62629>. Full diff: https://github.com/llvm/llvm-project/pull/79313.diff 2 Files Affected:
diff --git a/llvm/lib/Target/AMDGPU/Disassembler/AMDGPUDisassembler.cpp b/llvm/lib/Target/AMDGPU/Disassembler/AMDGPUDisassembler.cpp
index 4e2e57e943196d..93c0f86c4e41be 100644
--- a/llvm/lib/Target/AMDGPU/Disassembler/AMDGPUDisassembler.cpp
+++ b/llvm/lib/Target/AMDGPU/Disassembler/AMDGPUDisassembler.cpp
@@ -145,6 +145,17 @@ static DecodeStatus decodeSplitBarrier(MCInst &Inst, unsigned Val,
MandatoryLiteral, ImmWidth)); \
}
+static DecodeStatus decodeSrcOp(MCInst &Inst, unsigned EncSize,
+ AMDGPUDisassembler::OpWidthTy OpWidth,
+ unsigned Imm, unsigned EncImm,
+ bool MandatoryLiteral, unsigned ImmWidth,
+ const MCDisassembler *Decoder) {
+ assert(Imm < (1 << EncSize) && "Operand doesn't fit encoding!");
+ auto DAsm = static_cast<const AMDGPUDisassembler *>(Decoder);
+ return addOperand(
+ Inst, DAsm->decodeSrcOp(OpWidth, EncImm, MandatoryLiteral, ImmWidth));
+}
+
// Decoder for registers. Imm(7-bit) is number of register, uses decodeSrcOp to
// get register class. Used by SGPR only operands.
#define DECODE_OPERAND_REG_7(RegClass, OpWidth) \
@@ -154,9 +165,12 @@ static DecodeStatus decodeSplitBarrier(MCInst &Inst, unsigned Val,
// Imm{9} is acc(agpr or vgpr) Imm{8} should be 0 (see VOP3Pe_SMFMAC).
// Set Imm{8} to 1 (IS_VGPR) to decode using 'enum10' from decodeSrcOp.
// Used by AV_ register classes (AGPR or VGPR only register operands).
-#define DECODE_OPERAND_REG_AV10(RegClass, OpWidth) \
- DECODE_SrcOp(Decode##RegClass##RegisterClass, 10, OpWidth, \
- Imm | AMDGPU::EncValues::IS_VGPR, false, 0)
+template <AMDGPUDisassembler::OpWidthTy OpWidth>
+static DecodeStatus decodeAV10(MCInst &Inst, unsigned Imm, uint64_t /* Addr */,
+ const MCDisassembler *Decoder) {
+ return decodeSrcOp(Inst, 10, OpWidth, Imm, Imm | AMDGPU::EncValues::IS_VGPR,
+ false, 0, Decoder);
+}
// Decoder for Src(9-bit encoding) registers only.
#define DECODE_OPERAND_SRC_REG_9(RegClass, OpWidth) \
@@ -165,13 +179,20 @@ static DecodeStatus decodeSplitBarrier(MCInst &Inst, unsigned Val,
// Decoder for Src(9-bit encoding) AGPR, register number encoded in 9bits, set
// Imm{9} to 1 (set acc) and decode using 'enum10' from decodeSrcOp, registers
// only.
-#define DECODE_OPERAND_SRC_REG_A9(RegClass, OpWidth) \
- DECODE_SrcOp(decodeOperand_##RegClass, 9, OpWidth, Imm | 512, false, 0)
+template <AMDGPUDisassembler::OpWidthTy OpWidth>
+static DecodeStatus decodeSrcA9(MCInst &Inst, unsigned Imm, uint64_t /* Addr */,
+ const MCDisassembler *Decoder) {
+ return decodeSrcOp(Inst, 9, OpWidth, Imm, Imm | 512, false, 0, Decoder);
+}
// Decoder for 'enum10' from decodeSrcOp, Imm{0-8} is 9-bit Src encoding
// Imm{9} is acc, registers only.
-#define DECODE_SRC_OPERAND_REG_AV10(RegClass, OpWidth) \
- DECODE_SrcOp(decodeOperand_##RegClass, 10, OpWidth, Imm, false, 0)
+template <AMDGPUDisassembler::OpWidthTy OpWidth>
+static DecodeStatus decodeSrcAV10(MCInst &Inst, unsigned Imm,
+ uint64_t /* Addr */,
+ const MCDisassembler *Decoder) {
+ return decodeSrcOp(Inst, 10, OpWidth, Imm, Imm, false, 0, Decoder);
+}
// Decoder for RegisterOperands using 9-bit Src encoding. Operand can be
// register from RegClass or immediate. Registers that don't belong to RegClass
@@ -229,9 +250,6 @@ DECODE_OPERAND_REG_8(AReg_256)
DECODE_OPERAND_REG_8(AReg_512)
DECODE_OPERAND_REG_8(AReg_1024)
-DECODE_OPERAND_REG_AV10(AVDst_128, OPW128)
-DECODE_OPERAND_REG_AV10(AVDst_512, OPW512)
-
// Decoders for register only source RegisterOperands that use use 9-bit Src
// encoding: 'decodeOperand_<RegClass>'.
@@ -241,12 +259,6 @@ DECODE_OPERAND_SRC_REG_9(VReg_128, OPW128)
DECODE_OPERAND_SRC_REG_9(VReg_256, OPW256)
DECODE_OPERAND_SRC_REG_9(VRegOrLds_32, OPW32)
-DECODE_OPERAND_SRC_REG_A9(AGPR_32, OPW32)
-
-DECODE_SRC_OPERAND_REG_AV10(AV_32, OPW32)
-DECODE_SRC_OPERAND_REG_AV10(AV_64, OPW64)
-DECODE_SRC_OPERAND_REG_AV10(AV_128, OPW128)
-
// Decoders for register or immediate RegisterOperands that use 9-bit Src
// encoding: 'decodeOperand_<RegClass>_Imm<ImmWidth>'.
diff --git a/llvm/lib/Target/AMDGPU/SIRegisterInfo.td b/llvm/lib/Target/AMDGPU/SIRegisterInfo.td
index 404601d981138b..b50ea2a1db4a73 100644
--- a/llvm/lib/Target/AMDGPU/SIRegisterInfo.td
+++ b/llvm/lib/Target/AMDGPU/SIRegisterInfo.td
@@ -1321,11 +1321,14 @@ def VGPRSrc_16_Lo128 : RegisterOperand<VGPR_16_Lo128> {
// ASrc_* Operands with an AccVGPR
//===----------------------------------------------------------------------===//
-def ARegSrc_32 : RegisterOperand<AGPR_32> {
- let DecoderMethod = "decodeOperand_AGPR_32";
+class AVOperand<RegisterClass regClass, string decoder, string width>
+ : RegisterOperand<regClass> {
+ let DecoderMethod = decoder # "<AMDGPUDisassembler::" # width # ">";
let EncoderMethod = "getAVOperandEncoding";
}
+def ARegSrc_32 : AVOperand<AGPR_32, "decodeSrcA9", "OPW32">;
+
//===----------------------------------------------------------------------===//
// VCSrc_* Operands with an SGPR, VGPR or an inline constant
//===----------------------------------------------------------------------===//
@@ -1359,36 +1362,21 @@ def VISrc_1024_f32 : RegOrF32 <"VReg_1024", "OPERAND_REG_INLINE_C">;
// AVSrc_*, AVDst_*, AVLdSt_* Operands with an AGPR or VGPR
//===----------------------------------------------------------------------===//
-def AVSrc_32 : RegisterOperand<AV_32> {
- let DecoderMethod = "decodeOperand_AV_32";
- let EncoderMethod = "getAVOperandEncoding";
-}
-
-def AVSrc_64 : RegisterOperand<AV_64> {
- let DecoderMethod = "decodeOperand_AV_64";
- let EncoderMethod = "getAVOperandEncoding";
-}
+class AVSrcOperand<RegisterClass regClass, string width>
+ : AVOperand<regClass, "decodeSrcAV10", width>;
-def AVSrc_128 : RegisterOperand<AV_128> {
- let DecoderMethod = "decodeOperand_AV_128";
- let EncoderMethod = "getAVOperandEncoding";
-}
+def AVSrc_32 : AVSrcOperand<AV_32, "OPW32">;
+def AVSrc_64 : AVSrcOperand<AV_64, "OPW64">;
+def AVSrc_128 : AVSrcOperand<AV_128, "OPW128">;
-def AVDst_128 : RegisterOperand<AV_128> {
- let DecoderMethod = "DecodeAVDst_128RegisterClass";
- let EncoderMethod = "getAVOperandEncoding";
-}
+class AVDstOperand<RegisterClass regClass, string width>
+ : AVOperand<regClass, "decodeAV10", width>;
-def AVDst_512 : RegisterOperand<AV_512> {
- let DecoderMethod = "DecodeAVDst_512RegisterClass";
- let EncoderMethod = "getAVOperandEncoding";
-}
+def AVDst_128 : AVDstOperand<AV_128, "OPW128">;
+def AVDst_512 : AVDstOperand<AV_512, "OPW512">;
class AVLdStOperand<RegisterClass regClass, string width>
- : RegisterOperand<regClass> {
- let DecoderMethod = "decodeAVLdSt<AMDGPUDisassembler::" # width # ">";
- let EncoderMethod = "getAVOperandEncoding";
-}
+ : AVOperand<regClass, "decodeAVLdSt", width>;
def AVLdSt_32 : AVLdStOperand<AV_32, "OPW32">;
def AVLdSt_64 : AVLdStOperand<AV_64, "OPW64">;
|
jayfoad
approved these changes
Jan 24, 2024
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
LGTM
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.
Eliminates the need to define them manually.
Part of #62629.