Skip to content

[RISCV] Support zext.h mnemonic with Zbkb. #96821

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 2 commits into from
Jun 27, 2024
Merged

Conversation

topperc
Copy link
Collaborator

@topperc topperc commented Jun 26, 2024

Zbkb adds generic pack insructions. The zext.h encodings from Zbb are subsets of the generic encodings with rs2=x0. zext.h is pack on RV32 and packw on RV64.

Previously we only supported zext.h as a single instruction mnemonic in the assembler when Zbb was enabled. Otherwise we would emit it as 2 shifts. This patches recognizes it when either Zbkb or Zbb is enabled.

This patch also enables the zext.h isel patterns when Zbkb is enabled without Zbb.

Zbkb adds generic pack insructions. The zext.h encodings from Zbb
are subsets of the generic encodings with rs2=x0. zext.h is pack
on RV32 and packw on RV64.

Previously we only supported zext.h as a single instruction mnemonic
in the assembler when Zbb was enabled. Otherwise we would emit it
as 2 shifts. This patches recognizes it when either Zbkb or Zbb is
enabled.

This patch also enables the zext.h isel patterns when Zbkb is enabled
without Zbb.
@llvmbot llvmbot added backend:RISC-V mc Machine (object) code labels Jun 26, 2024
@llvmbot
Copy link
Member

llvmbot commented Jun 26, 2024

@llvm/pr-subscribers-backend-risc-v

@llvm/pr-subscribers-mc

Author: Craig Topper (topperc)

Changes

Zbkb adds generic pack insructions. The zext.h encodings from Zbb are subsets of the generic encodings with rs2=x0. zext.h is pack on RV32 and packw on RV64.

Previously we only supported zext.h as a single instruction mnemonic in the assembler when Zbb was enabled. Otherwise we would emit it as 2 shifts. This patches recognizes it when either Zbkb or Zbb is enabled.

This patch also enables the zext.h isel patterns when Zbkb is enabled without Zbb.


Full diff: https://github.com/llvm/llvm-project/pull/96821.diff

7 Files Affected:

  • (modified) llvm/lib/Target/RISCV/RISCVInstrInfoZb.td (+10-12)
  • (modified) llvm/test/CodeGen/RISCV/rv32zbkb.ll (+64)
  • (modified) llvm/test/CodeGen/RISCV/rv64-legal-i32/rv64zbkb.ll (+60)
  • (modified) llvm/test/CodeGen/RISCV/rv64zbkb.ll (+60)
  • (modified) llvm/test/MC/RISCV/rv32zbkb-only-valid.s (+11-1)
  • (modified) llvm/test/MC/RISCV/rv32zbkb-valid.s (-5)
  • (modified) llvm/test/MC/RISCV/rv64zbkb-valid.s (+13-3)
diff --git a/llvm/lib/Target/RISCV/RISCVInstrInfoZb.td b/llvm/lib/Target/RISCV/RISCVInstrInfoZb.td
index 8a0bbf6abd334..a5747eef4454d 100644
--- a/llvm/lib/Target/RISCV/RISCVInstrInfoZb.td
+++ b/llvm/lib/Target/RISCV/RISCVInstrInfoZb.td
@@ -415,22 +415,18 @@ let Predicates = [HasStdExtZbkb, IsRV64], IsSignExtendingOpW = 1 in
 def PACKW  : ALUW_rr<0b0000100, 0b100, "packw">,
              Sched<[WritePACK32, ReadPACK32, ReadPACK32]>;
 
-let Predicates = [HasStdExtZbb, IsRV32] in {
+let Predicates = [HasStdExtZbbOrZbkb, IsRV32] in {
+let IsSignExtendingOpW = 1 in
 def ZEXT_H_RV32 : RVBUnary<0b000010000000, 0b100, OPC_OP, "zext.h">,
                   Sched<[WriteIALU, ReadIALU]>;
-} // Predicates = [HasStdExtZbb, IsRV32]
-
-let Predicates = [HasStdExtZbb, IsRV64], IsSignExtendingOpW = 1 in {
-def ZEXT_H_RV64 : RVBUnary<0b000010000000, 0b100, OPC_OP_32, "zext.h">,
-                  Sched<[WriteIALU, ReadIALU]>;
-} // Predicates = [HasStdExtZbb, IsRV64]
-
-let Predicates = [HasStdExtZbbOrZbkb, IsRV32] in {
 def REV8_RV32 : RVBUnary<0b011010011000, 0b101, OPC_OP_IMM, "rev8">,
                 Sched<[WriteREV8, ReadREV8]>;
 } // Predicates = [HasStdExtZbbOrZbkb, IsRV32]
 
 let Predicates = [HasStdExtZbbOrZbkb, IsRV64] in {
+let IsSignExtendingOpW = 1 in
+def ZEXT_H_RV64 : RVBUnary<0b000010000000, 0b100, OPC_OP_32, "zext.h">,
+                  Sched<[WriteIALU, ReadIALU]>;
 def REV8_RV64 : RVBUnary<0b011010111000, 0b101, OPC_OP_IMM, "rev8">,
                 Sched<[WriteREV8, ReadREV8]>;
 } // Predicates = [HasStdExtZbbOrZbkb, IsRV64]
@@ -637,9 +633,9 @@ def : Pat<(i64 (or (sext_inreg (shl GPR:$rs2, (i64 16)), i32),
           (PACKW GPR:$rs1, GPR:$rs2)>;
 } // Predicates = [HasStdExtZbkb, IsRV64]
 
-let Predicates = [HasStdExtZbb, IsRV32] in
+let Predicates = [HasStdExtZbbOrZbkb, IsRV32] in
 def : Pat<(i32 (and GPR:$rs, 0xFFFF)), (ZEXT_H_RV32 GPR:$rs)>;
-let Predicates = [HasStdExtZbb, IsRV64] in
+let Predicates = [HasStdExtZbbOrZbkb, IsRV64] in
 def : Pat<(i64 (and GPR:$rs, 0xFFFF)), (ZEXT_H_RV64 GPR:$rs)>;
 
 let Predicates = [HasStdExtZba] in {
@@ -748,9 +744,11 @@ def : PatGpr<ctpop, CPOPW, i32>;
 
 def : Pat<(i32 (sext_inreg GPR:$rs1, i8)), (SEXT_B GPR:$rs1)>;
 def : Pat<(i32 (sext_inreg GPR:$rs1, i16)), (SEXT_H GPR:$rs1)>;
+} // Predicates = [HasStdExtZbb, IsRV64]
 
+let Predicates = [HasStdExtZbbOrZbkb, IsRV64] in {
 def : Pat<(i32 (and GPR:$rs, 0xFFFF)), (ZEXT_H_RV64 GPR:$rs)>;
-} // Predicates = [HasStdExtZbb, IsRV64]
+} // Predicates = [HasStdExtZbbOrZbkb, IsRV64]
 
 let Predicates = [HasStdExtZbbOrZbkb, IsRV64] in {
 def : Pat<(i32 (and GPR:$rs1, (not GPR:$rs2))), (ANDN GPR:$rs1, GPR:$rs2)>;
diff --git a/llvm/test/CodeGen/RISCV/rv32zbkb.ll b/llvm/test/CodeGen/RISCV/rv32zbkb.ll
index 33116d16595a4..b194c0ed52dfb 100644
--- a/llvm/test/CodeGen/RISCV/rv32zbkb.ll
+++ b/llvm/test/CodeGen/RISCV/rv32zbkb.ll
@@ -255,3 +255,67 @@ define void @packh_i16_3(i8 zeroext %0, i8 zeroext %1, i8 zeroext %2, ptr %p) {
   store i16 %8, ptr %p
   ret void
 }
+
+define i32 @zexth_i32(i32 %a) nounwind {
+; RV32I-LABEL: zexth_i32:
+; RV32I:       # %bb.0:
+; RV32I-NEXT:    slli a0, a0, 16
+; RV32I-NEXT:    srli a0, a0, 16
+; RV32I-NEXT:    ret
+;
+; RV32ZBKB-LABEL: zexth_i32:
+; RV32ZBKB:       # %bb.0:
+; RV32ZBKB-NEXT:    zext.h a0, a0
+; RV32ZBKB-NEXT:    ret
+  %and = and i32 %a, 65535
+  ret i32 %and
+}
+
+define i64 @zexth_i64(i64 %a) nounwind {
+; RV32I-LABEL: zexth_i64:
+; RV32I:       # %bb.0:
+; RV32I-NEXT:    slli a0, a0, 16
+; RV32I-NEXT:    srli a0, a0, 16
+; RV32I-NEXT:    li a1, 0
+; RV32I-NEXT:    ret
+;
+; RV32ZBKB-LABEL: zexth_i64:
+; RV32ZBKB:       # %bb.0:
+; RV32ZBKB-NEXT:    zext.h a0, a0
+; RV32ZBKB-NEXT:    li a1, 0
+; RV32ZBKB-NEXT:    ret
+  %and = and i64 %a, 65535
+  ret i64 %and
+}
+
+define i32 @zext_i16_to_i32(i16 %a) nounwind {
+; RV32I-LABEL: zext_i16_to_i32:
+; RV32I:       # %bb.0:
+; RV32I-NEXT:    slli a0, a0, 16
+; RV32I-NEXT:    srli a0, a0, 16
+; RV32I-NEXT:    ret
+;
+; RV32ZBKB-LABEL: zext_i16_to_i32:
+; RV32ZBKB:       # %bb.0:
+; RV32ZBKB-NEXT:    zext.h a0, a0
+; RV32ZBKB-NEXT:    ret
+  %1 = zext i16 %a to i32
+  ret i32 %1
+}
+
+define i64 @zext_i16_to_i64(i16 %a) nounwind {
+; RV32I-LABEL: zext_i16_to_i64:
+; RV32I:       # %bb.0:
+; RV32I-NEXT:    slli a0, a0, 16
+; RV32I-NEXT:    srli a0, a0, 16
+; RV32I-NEXT:    li a1, 0
+; RV32I-NEXT:    ret
+;
+; RV32ZBKB-LABEL: zext_i16_to_i64:
+; RV32ZBKB:       # %bb.0:
+; RV32ZBKB-NEXT:    zext.h a0, a0
+; RV32ZBKB-NEXT:    li a1, 0
+; RV32ZBKB-NEXT:    ret
+  %1 = zext i16 %a to i64
+  ret i64 %1
+}
diff --git a/llvm/test/CodeGen/RISCV/rv64-legal-i32/rv64zbkb.ll b/llvm/test/CodeGen/RISCV/rv64-legal-i32/rv64zbkb.ll
index 806f2e0838e29..8e26598380d4a 100644
--- a/llvm/test/CodeGen/RISCV/rv64-legal-i32/rv64zbkb.ll
+++ b/llvm/test/CodeGen/RISCV/rv64-legal-i32/rv64zbkb.ll
@@ -308,3 +308,63 @@ define i64 @pack_i64_imm() {
 ; RV64ZBKB-NEXT:    ret
   ret i64 1157442765409226768 ; 0x0101010101010101
 }
+
+define i32 @zexth_i32(i32 %a) nounwind {
+; RV64I-LABEL: zexth_i32:
+; RV64I:       # %bb.0:
+; RV64I-NEXT:    slli a0, a0, 48
+; RV64I-NEXT:    srli a0, a0, 48
+; RV64I-NEXT:    ret
+;
+; RV64ZBKB-LABEL: zexth_i32:
+; RV64ZBKB:       # %bb.0:
+; RV64ZBKB-NEXT:    zext.h a0, a0
+; RV64ZBKB-NEXT:    ret
+  %and = and i32 %a, 65535
+  ret i32 %and
+}
+
+define i64 @zexth_i64(i64 %a) nounwind {
+; RV64I-LABEL: zexth_i64:
+; RV64I:       # %bb.0:
+; RV64I-NEXT:    slli a0, a0, 48
+; RV64I-NEXT:    srli a0, a0, 48
+; RV64I-NEXT:    ret
+;
+; RV64ZBKB-LABEL: zexth_i64:
+; RV64ZBKB:       # %bb.0:
+; RV64ZBKB-NEXT:    zext.h a0, a0
+; RV64ZBKB-NEXT:    ret
+  %and = and i64 %a, 65535
+  ret i64 %and
+}
+
+define i32 @zext_i16_to_i32(i16 %a) nounwind {
+; RV64I-LABEL: zext_i16_to_i32:
+; RV64I:       # %bb.0:
+; RV64I-NEXT:    slli a0, a0, 48
+; RV64I-NEXT:    srli a0, a0, 48
+; RV64I-NEXT:    ret
+;
+; RV64ZBKB-LABEL: zext_i16_to_i32:
+; RV64ZBKB:       # %bb.0:
+; RV64ZBKB-NEXT:    zext.h a0, a0
+; RV64ZBKB-NEXT:    ret
+  %1 = zext i16 %a to i32
+  ret i32 %1
+}
+
+define i64 @zext_i16_to_i64(i16 %a) nounwind {
+; RV64I-LABEL: zext_i16_to_i64:
+; RV64I:       # %bb.0:
+; RV64I-NEXT:    slli a0, a0, 48
+; RV64I-NEXT:    srli a0, a0, 48
+; RV64I-NEXT:    ret
+;
+; RV64ZBKB-LABEL: zext_i16_to_i64:
+; RV64ZBKB:       # %bb.0:
+; RV64ZBKB-NEXT:    zext.h a0, a0
+; RV64ZBKB-NEXT:    ret
+  %1 = zext i16 %a to i64
+  ret i64 %1
+}
diff --git a/llvm/test/CodeGen/RISCV/rv64zbkb.ll b/llvm/test/CodeGen/RISCV/rv64zbkb.ll
index db1330b0a63bd..4aa6cd42ab099 100644
--- a/llvm/test/CodeGen/RISCV/rv64zbkb.ll
+++ b/llvm/test/CodeGen/RISCV/rv64zbkb.ll
@@ -332,3 +332,63 @@ define i64 @pack_i64_imm() {
 ; RV64ZBKB-NEXT:    ret
   ret i64 1157442765409226768 ; 0x0101010101010101
 }
+
+define i32 @zexth_i32(i32 %a) nounwind {
+; RV64I-LABEL: zexth_i32:
+; RV64I:       # %bb.0:
+; RV64I-NEXT:    slli a0, a0, 48
+; RV64I-NEXT:    srli a0, a0, 48
+; RV64I-NEXT:    ret
+;
+; RV64ZBKB-LABEL: zexth_i32:
+; RV64ZBKB:       # %bb.0:
+; RV64ZBKB-NEXT:    zext.h a0, a0
+; RV64ZBKB-NEXT:    ret
+  %and = and i32 %a, 65535
+  ret i32 %and
+}
+
+define i64 @zexth_i64(i64 %a) nounwind {
+; RV64I-LABEL: zexth_i64:
+; RV64I:       # %bb.0:
+; RV64I-NEXT:    slli a0, a0, 48
+; RV64I-NEXT:    srli a0, a0, 48
+; RV64I-NEXT:    ret
+;
+; RV64ZBKB-LABEL: zexth_i64:
+; RV64ZBKB:       # %bb.0:
+; RV64ZBKB-NEXT:    zext.h a0, a0
+; RV64ZBKB-NEXT:    ret
+  %and = and i64 %a, 65535
+  ret i64 %and
+}
+
+define i32 @zext_i16_to_i32(i16 %a) nounwind {
+; RV64I-LABEL: zext_i16_to_i32:
+; RV64I:       # %bb.0:
+; RV64I-NEXT:    slli a0, a0, 48
+; RV64I-NEXT:    srli a0, a0, 48
+; RV64I-NEXT:    ret
+;
+; RV64ZBKB-LABEL: zext_i16_to_i32:
+; RV64ZBKB:       # %bb.0:
+; RV64ZBKB-NEXT:    zext.h a0, a0
+; RV64ZBKB-NEXT:    ret
+  %1 = zext i16 %a to i32
+  ret i32 %1
+}
+
+define i64 @zext_i16_to_i64(i16 %a) nounwind {
+; RV64I-LABEL: zext_i16_to_i64:
+; RV64I:       # %bb.0:
+; RV64I-NEXT:    slli a0, a0, 48
+; RV64I-NEXT:    srli a0, a0, 48
+; RV64I-NEXT:    ret
+;
+; RV64ZBKB-LABEL: zext_i16_to_i64:
+; RV64ZBKB:       # %bb.0:
+; RV64ZBKB-NEXT:    zext.h a0, a0
+; RV64ZBKB-NEXT:    ret
+  %1 = zext i16 %a to i64
+  ret i64 %1
+}
diff --git a/llvm/test/MC/RISCV/rv32zbkb-only-valid.s b/llvm/test/MC/RISCV/rv32zbkb-only-valid.s
index a04ff04ba3f10..3120365fc3545 100644
--- a/llvm/test/MC/RISCV/rv32zbkb-only-valid.s
+++ b/llvm/test/MC/RISCV/rv32zbkb-only-valid.s
@@ -2,7 +2,7 @@
 # RUN:     | FileCheck -check-prefixes=CHECK-ASM,CHECK-ASM-AND-OBJ %s
 # RUN: llvm-mc -filetype=obj -triple=riscv32 -mattr=+zbkb < %s \
 # RUN:     | llvm-objdump --mattr=+zbkb -d -r - \
-# RUN:     | FileCheck --check-prefix=CHECK-ASM-AND-OBJ %s
+# RUN:     | FileCheck --check-prefixes=CHECK-OBJ,CHECK-ASM-AND-OBJ %s
 
 # CHECK-ASM-AND-OBJ: rev8 t0, t1
 # CHECK-ASM: encoding: [0x93,0x52,0x83,0x69]
@@ -14,3 +14,13 @@ zip t0, t1
 # CHECK-S-OBJ-NOALIAS: unzip t0, t1
 # CHECK-ASM: encoding: [0x93,0x52,0xf3,0x08]
 unzip t0, t1
+
+# Test the encoding used for zext.h for RV32.
+# CHECK-ASM: pack t0, t1, zero
+# CHECK-OBJ: zext.h t0, t1
+# CHECK-ASM: encoding: [0xb3,0x42,0x03,0x08]
+pack t0, t1, x0
+
+# CHECK-ASM-AND-OBJ: zext.h t0, t1
+# CHECK-ASM: encoding: [0xb3,0x42,0x03,0x08]
+zext.h t0, t1
diff --git a/llvm/test/MC/RISCV/rv32zbkb-valid.s b/llvm/test/MC/RISCV/rv32zbkb-valid.s
index 9849430a46f82..0e5ee4cb5fd6e 100644
--- a/llvm/test/MC/RISCV/rv32zbkb-valid.s
+++ b/llvm/test/MC/RISCV/rv32zbkb-valid.s
@@ -36,11 +36,6 @@ xnor t0, t1, t2
 # CHECK-ASM: encoding: [0xb3,0x42,0x73,0x08]
 pack t0, t1, t2
 
-# Test the encoding used for zext.h for RV32.
-# CHECK-ASM-AND-OBJ: pack t0, t1, zero
-# CHECK-ASM: encoding: [0xb3,0x42,0x03,0x08]
-pack t0, t1, x0
-
 # CHECK-ASM-AND-OBJ: packh t0, t1, t2
 # CHECK-ASM: encoding: [0xb3,0x72,0x73,0x08]
 packh t0, t1, t2
diff --git a/llvm/test/MC/RISCV/rv64zbkb-valid.s b/llvm/test/MC/RISCV/rv64zbkb-valid.s
index fc0541c86ebed..f885c4a56172a 100644
--- a/llvm/test/MC/RISCV/rv64zbkb-valid.s
+++ b/llvm/test/MC/RISCV/rv64zbkb-valid.s
@@ -2,7 +2,7 @@
 # RUN:     | FileCheck -check-prefixes=CHECK-ASM,CHECK-ASM-AND-OBJ %s
 # RUN: llvm-mc -filetype=obj -triple=riscv64 -mattr=+zbkb < %s \
 # RUN:     | llvm-objdump --mattr=+zbkb --no-print-imm-hex -d -r - \
-# RUN:     | FileCheck --check-prefix=CHECK-ASM-AND-OBJ %s
+# RUN:     | FileCheck --check-prefixes=CHECK-OBJ,CHECK-ASM-AND-OBJ %s
 
 # CHECK-ASM-AND-OBJ: rev8 t0, t1
 # CHECK-ASM: encoding: [0x93,0x52,0x83,0x6b]
@@ -28,7 +28,17 @@ roriw t0, t1, 0
 # CHECK-ASM: encoding: [0xbb,0x42,0x73,0x08]
 packw t0, t1, t2
 
-# Test the encoding used for zext.h
-# CHECK-ASM-AND-OBJ: packw t0, t1, zero
+# Test the encoding used for zext.h on RV64
+# CHECK-ASM: packw t0, t1, zero
+# CHECK-OBJ: zext.h t0, t1
 # CHECK-ASM: encoding: [0xbb,0x42,0x03,0x08]
 packw t0, t1, zero
+
+# Test the encoding used for zext.h on RV32
+# CHECK-ASM-AND-OBJ: pack t0, t1, zero
+# CHECK-ASM: encoding: [0xb3,0x42,0x03,0x08]
+pack t0, t1, zero
+
+# CHECK-ASM-AND-OBJ: zext.h t0, t1
+# CHECK-ASM: encoding: [0xbb,0x42,0x03,0x08]
+zext.h t0, t1

@@ -14,3 +14,13 @@ zip t0, t1
# CHECK-S-OBJ-NOALIAS: unzip t0, t1
# CHECK-ASM: encoding: [0x93,0x52,0xf3,0x08]
unzip t0, t1

# Test the encoding used for zext.h for RV32.
# CHECK-ASM: pack t0, t1, zero
Copy link
Collaborator Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Note we get different printing depending on whether its printed from the assembler or disassembler. Assembler just echoes what we parsed. Disassembler will treat it as zext.h.

Copy link
Contributor

@wangpc-pp wangpc-pp left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

LGTM.

@dtcxzyw
Copy link
Member

dtcxzyw commented Jun 27, 2024

FYI, GNU binutils as doesn't emit pack/packw for zext.h when zbkb is available:

dtcxzyw@dtcxzyw:~/WorkSpace/Projects/compilers/LLVM/llvm-build$ cat xxx.S
zext.h t0, t1
dtcxzyw@dtcxzyw:~/WorkSpace/Projects/compilers/LLVM/llvm-build$ riscv64-linux-gnu-gcc -march=rv64gc_zbkb xxx.S -c
dtcxzyw@dtcxzyw:~/WorkSpace/Projects/compilers/LLVM/llvm-build$ riscv64-linux-gnu-objdump xxx.o -S

xxx.o:     file format elf64-littleriscv


Disassembly of section .text:

0000000000000000 <.text>:
   0:   03031293                slli    t0,t1,0x30
   4:   0302d293                srli    t0,t0,0x30

@topperc
Copy link
Collaborator Author

topperc commented Jun 27, 2024

FYI, GNU binutils as doesn't emit pack/packw for zext.h when zbkb is available:

dtcxzyw@dtcxzyw:~/WorkSpace/Projects/compilers/LLVM/llvm-build$ cat xxx.S
zext.h t0, t1
dtcxzyw@dtcxzyw:~/WorkSpace/Projects/compilers/LLVM/llvm-build$ riscv64-linux-gnu-gcc -march=rv64gc_zbkb xxx.S -c
dtcxzyw@dtcxzyw:~/WorkSpace/Projects/compilers/LLVM/llvm-build$ riscv64-linux-gnu-objdump xxx.o -S

xxx.o:     file format elf64-littleriscv


Disassembly of section .text:

0000000000000000 <.text>:
   0:   03031293                slli    t0,t1,0x30
   4:   0302d293                srli    t0,t0,0x30

I know. I'm going to try to get that changed too.

@topperc topperc merged commit cad7263 into llvm:main Jun 27, 2024
7 checks passed
@topperc topperc deleted the pr/zext-h branch June 27, 2024 16:31
@jrtc27
Copy link
Collaborator

jrtc27 commented Jun 27, 2024

Hm, I’m not sold on the assembly vs disassembly difference. Do we really need that weirdness, or can we limit that kind of thing to only be for real pseudoinstructions?

@topperc
Copy link
Collaborator Author

topperc commented Jun 27, 2024

Hm, I’m not sold on the assembly vs disassembly difference. Do we really need that weirdness, or can we limit that kind of thing to only be for real pseudoinstructions?

#96957

lravenclaw pushed a commit to lravenclaw/llvm-project that referenced this pull request Jul 3, 2024
Zbkb adds generic pack insructions. The zext.h encodings from Zbb are
subsets of the generic encodings with rs2=x0. zext.h is pack on RV32 and
packw on RV64.

Previously we only supported zext.h as a single instruction mnemonic in
the assembler when Zbb was enabled. Otherwise we would emit it as 2
shifts. This patches recognizes it when either Zbkb or Zbb is enabled.

This patch also enables the zext.h isel patterns when Zbkb is enabled
without Zbb.
AlexisPerry pushed a commit to llvm-project-tlp/llvm-project that referenced this pull request Jul 9, 2024
Zbkb adds generic pack insructions. The zext.h encodings from Zbb are
subsets of the generic encodings with rs2=x0. zext.h is pack on RV32 and
packw on RV64.

Previously we only supported zext.h as a single instruction mnemonic in
the assembler when Zbb was enabled. Otherwise we would emit it as 2
shifts. This patches recognizes it when either Zbkb or Zbb is enabled.

This patch also enables the zext.h isel patterns when Zbkb is enabled
without Zbb.
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
backend:RISC-V mc Machine (object) code
Projects
None yet
Development

Successfully merging this pull request may close these issues.

5 participants