[llvm] 9b3c0f9 - [RISCV] Add Zbb+Zbt command lines to the signed saturing add/sub tests.
Craig Topper via llvm-commits
llvm-commits at lists.llvm.org
Thu Mar 25 17:26:08 PDT 2021
Author: Craig Topper
Date: 2021-03-25T17:25:36-07:00
New Revision: 9b3c0f9a544c26ddbc266a7eeb0a8d3af6f71e87
URL: https://github.com/llvm/llvm-project/commit/9b3c0f9a544c26ddbc266a7eeb0a8d3af6f71e87
DIFF: https://github.com/llvm/llvm-project/commit/9b3c0f9a544c26ddbc266a7eeb0a8d3af6f71e87.diff
LOG: [RISCV] Add Zbb+Zbt command lines to the signed saturing add/sub tests.
This will enable cmov to be used for select. I improve the codegen
of select_cc in D99021, but that patch doesn't work for cmov.
Added:
Modified:
llvm/test/CodeGen/RISCV/sadd_sat.ll
llvm/test/CodeGen/RISCV/sadd_sat_plus.ll
llvm/test/CodeGen/RISCV/ssub_sat.ll
llvm/test/CodeGen/RISCV/ssub_sat_plus.ll
Removed:
################################################################################
diff --git a/llvm/test/CodeGen/RISCV/sadd_sat.ll b/llvm/test/CodeGen/RISCV/sadd_sat.ll
index d35e23c0a1cc..98b366939dab 100644
--- a/llvm/test/CodeGen/RISCV/sadd_sat.ll
+++ b/llvm/test/CodeGen/RISCV/sadd_sat.ll
@@ -1,8 +1,10 @@
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
; RUN: llc < %s -mtriple=riscv32 -mattr=+m | FileCheck %s --check-prefix=RV32I
; RUN: llc < %s -mtriple=riscv64 -mattr=+m | FileCheck %s --check-prefix=RV64I
-; RUN: llc < %s -mtriple=riscv32 -mattr=+m,+experimental-zbb | FileCheck %s --check-prefix=RV32IZbb
-; RUN: llc < %s -mtriple=riscv64 -mattr=+m,+experimental-zbb | FileCheck %s --check-prefix=RV64IZbb
+; RUN: llc < %s -mtriple=riscv32 -mattr=+m,+experimental-zbb | FileCheck %s --check-prefixes=RV32IZbb,RV32IZbbNOZbt
+; RUN: llc < %s -mtriple=riscv64 -mattr=+m,+experimental-zbb | FileCheck %s --check-prefixes=RV64IZbb,RV64IZbbNOZbt
+; RUN: llc < %s -mtriple=riscv32 -mattr=+m,+experimental-zbb,+experimental-zbt | FileCheck %s --check-prefixes=RV32IZbb,RV32IZbbZbt
+; RUN: llc < %s -mtriple=riscv64 -mattr=+m,+experimental-zbb,+experimental-zbt | FileCheck %s --check-prefixes=RV64IZbb,RV64IZbbZbt
declare i4 @llvm.sadd.sat.i4(i4, i4)
declare i8 @llvm.sadd.sat.i8(i8, i8)
@@ -45,22 +47,22 @@ define signext i32 @func(i32 signext %x, i32 signext %y) nounwind {
; RV64I-NEXT: lui a0, 524288
; RV64I-NEXT: ret
;
-; RV32IZbb-LABEL: func:
-; RV32IZbb: # %bb.0:
-; RV32IZbb-NEXT: mv a2, a0
-; RV32IZbb-NEXT: add a3, a0, a1
-; RV32IZbb-NEXT: lui a0, 524288
-; RV32IZbb-NEXT: bgez a3, .LBB0_2
-; RV32IZbb-NEXT: # %bb.1:
-; RV32IZbb-NEXT: addi a0, a0, -1
-; RV32IZbb-NEXT: .LBB0_2:
-; RV32IZbb-NEXT: slt a2, a3, a2
-; RV32IZbb-NEXT: slti a1, a1, 0
-; RV32IZbb-NEXT: bne a1, a2, .LBB0_4
-; RV32IZbb-NEXT: # %bb.3:
-; RV32IZbb-NEXT: mv a0, a3
-; RV32IZbb-NEXT: .LBB0_4:
-; RV32IZbb-NEXT: ret
+; RV32IZbbNOZbt-LABEL: func:
+; RV32IZbbNOZbt: # %bb.0:
+; RV32IZbbNOZbt-NEXT: mv a2, a0
+; RV32IZbbNOZbt-NEXT: add a3, a0, a1
+; RV32IZbbNOZbt-NEXT: lui a0, 524288
+; RV32IZbbNOZbt-NEXT: bgez a3, .LBB0_2
+; RV32IZbbNOZbt-NEXT: # %bb.1:
+; RV32IZbbNOZbt-NEXT: addi a0, a0, -1
+; RV32IZbbNOZbt-NEXT: .LBB0_2:
+; RV32IZbbNOZbt-NEXT: slt a2, a3, a2
+; RV32IZbbNOZbt-NEXT: slti a1, a1, 0
+; RV32IZbbNOZbt-NEXT: bne a1, a2, .LBB0_4
+; RV32IZbbNOZbt-NEXT: # %bb.3:
+; RV32IZbbNOZbt-NEXT: mv a0, a3
+; RV32IZbbNOZbt-NEXT: .LBB0_4:
+; RV32IZbbNOZbt-NEXT: ret
;
; RV64IZbb-LABEL: func:
; RV64IZbb: # %bb.0:
@@ -70,6 +72,19 @@ define signext i32 @func(i32 signext %x, i32 signext %y) nounwind {
; RV64IZbb-NEXT: min a0, a0, a2
; RV64IZbb-NEXT: max a0, a0, a1
; RV64IZbb-NEXT: ret
+;
+; RV32IZbbZbt-LABEL: func:
+; RV32IZbbZbt: # %bb.0:
+; RV32IZbbZbt-NEXT: add a2, a0, a1
+; RV32IZbbZbt-NEXT: slti a3, a2, 0
+; RV32IZbbZbt-NEXT: lui a4, 524288
+; RV32IZbbZbt-NEXT: addi a5, a4, -1
+; RV32IZbbZbt-NEXT: cmov a3, a3, a5, a4
+; RV32IZbbZbt-NEXT: slt a0, a2, a0
+; RV32IZbbZbt-NEXT: slti a1, a1, 0
+; RV32IZbbZbt-NEXT: xor a0, a1, a0
+; RV32IZbbZbt-NEXT: cmov a0, a0, a3, a2
+; RV32IZbbZbt-NEXT: ret
%tmp = call i32 @llvm.sadd.sat.i32(i32 %x, i32 %y);
ret i32 %tmp;
}
@@ -118,47 +133,80 @@ define i64 @func2(i64 %x, i64 %y) nounwind {
; RV64I-NEXT: .LBB1_4:
; RV64I-NEXT: ret
;
-; RV32IZbb-LABEL: func2:
-; RV32IZbb: # %bb.0:
-; RV32IZbb-NEXT: mv a4, a1
-; RV32IZbb-NEXT: mv a1, a0
-; RV32IZbb-NEXT: add a5, a4, a3
-; RV32IZbb-NEXT: add a0, a0, a2
-; RV32IZbb-NEXT: sltu a1, a0, a1
-; RV32IZbb-NEXT: add a2, a5, a1
-; RV32IZbb-NEXT: lui a1, 524288
-; RV32IZbb-NEXT: bgez a2, .LBB1_2
-; RV32IZbb-NEXT: # %bb.1:
-; RV32IZbb-NEXT: addi a1, a1, -1
-; RV32IZbb-NEXT: .LBB1_2:
-; RV32IZbb-NEXT: xor a5, a4, a2
-; RV32IZbb-NEXT: xor a3, a4, a3
-; RV32IZbb-NEXT: andn a3, a5, a3
-; RV32IZbb-NEXT: bltz a3, .LBB1_4
-; RV32IZbb-NEXT: # %bb.3:
-; RV32IZbb-NEXT: mv a1, a2
-; RV32IZbb-NEXT: ret
-; RV32IZbb-NEXT: .LBB1_4:
-; RV32IZbb-NEXT: srai a0, a2, 31
-; RV32IZbb-NEXT: ret
+; RV32IZbbNOZbt-LABEL: func2:
+; RV32IZbbNOZbt: # %bb.0:
+; RV32IZbbNOZbt-NEXT: mv a4, a1
+; RV32IZbbNOZbt-NEXT: mv a1, a0
+; RV32IZbbNOZbt-NEXT: add a5, a4, a3
+; RV32IZbbNOZbt-NEXT: add a0, a0, a2
+; RV32IZbbNOZbt-NEXT: sltu a1, a0, a1
+; RV32IZbbNOZbt-NEXT: add a2, a5, a1
+; RV32IZbbNOZbt-NEXT: lui a1, 524288
+; RV32IZbbNOZbt-NEXT: bgez a2, .LBB1_2
+; RV32IZbbNOZbt-NEXT: # %bb.1:
+; RV32IZbbNOZbt-NEXT: addi a1, a1, -1
+; RV32IZbbNOZbt-NEXT: .LBB1_2:
+; RV32IZbbNOZbt-NEXT: xor a5, a4, a2
+; RV32IZbbNOZbt-NEXT: xor a3, a4, a3
+; RV32IZbbNOZbt-NEXT: andn a3, a5, a3
+; RV32IZbbNOZbt-NEXT: bltz a3, .LBB1_4
+; RV32IZbbNOZbt-NEXT: # %bb.3:
+; RV32IZbbNOZbt-NEXT: mv a1, a2
+; RV32IZbbNOZbt-NEXT: ret
+; RV32IZbbNOZbt-NEXT: .LBB1_4:
+; RV32IZbbNOZbt-NEXT: srai a0, a2, 31
+; RV32IZbbNOZbt-NEXT: ret
;
-; RV64IZbb-LABEL: func2:
-; RV64IZbb: # %bb.0:
-; RV64IZbb-NEXT: mv a2, a0
-; RV64IZbb-NEXT: add a3, a0, a1
-; RV64IZbb-NEXT: addi a0, zero, -1
-; RV64IZbb-NEXT: slli a0, a0, 63
-; RV64IZbb-NEXT: bgez a3, .LBB1_2
-; RV64IZbb-NEXT: # %bb.1:
-; RV64IZbb-NEXT: addi a0, a0, -1
-; RV64IZbb-NEXT: .LBB1_2:
-; RV64IZbb-NEXT: slt a2, a3, a2
-; RV64IZbb-NEXT: slti a1, a1, 0
-; RV64IZbb-NEXT: bne a1, a2, .LBB1_4
-; RV64IZbb-NEXT: # %bb.3:
-; RV64IZbb-NEXT: mv a0, a3
-; RV64IZbb-NEXT: .LBB1_4:
-; RV64IZbb-NEXT: ret
+; RV64IZbbNOZbt-LABEL: func2:
+; RV64IZbbNOZbt: # %bb.0:
+; RV64IZbbNOZbt-NEXT: mv a2, a0
+; RV64IZbbNOZbt-NEXT: add a3, a0, a1
+; RV64IZbbNOZbt-NEXT: addi a0, zero, -1
+; RV64IZbbNOZbt-NEXT: slli a0, a0, 63
+; RV64IZbbNOZbt-NEXT: bgez a3, .LBB1_2
+; RV64IZbbNOZbt-NEXT: # %bb.1:
+; RV64IZbbNOZbt-NEXT: addi a0, a0, -1
+; RV64IZbbNOZbt-NEXT: .LBB1_2:
+; RV64IZbbNOZbt-NEXT: slt a2, a3, a2
+; RV64IZbbNOZbt-NEXT: slti a1, a1, 0
+; RV64IZbbNOZbt-NEXT: bne a1, a2, .LBB1_4
+; RV64IZbbNOZbt-NEXT: # %bb.3:
+; RV64IZbbNOZbt-NEXT: mv a0, a3
+; RV64IZbbNOZbt-NEXT: .LBB1_4:
+; RV64IZbbNOZbt-NEXT: ret
+;
+; RV32IZbbZbt-LABEL: func2:
+; RV32IZbbZbt: # %bb.0:
+; RV32IZbbZbt-NEXT: add a4, a1, a3
+; RV32IZbbZbt-NEXT: add a2, a0, a2
+; RV32IZbbZbt-NEXT: sltu a0, a2, a0
+; RV32IZbbZbt-NEXT: add a0, a4, a0
+; RV32IZbbZbt-NEXT: slti a4, a0, 0
+; RV32IZbbZbt-NEXT: lui a6, 524288
+; RV32IZbbZbt-NEXT: addi a5, a6, -1
+; RV32IZbbZbt-NEXT: cmov a4, a4, a5, a6
+; RV32IZbbZbt-NEXT: xor a5, a1, a0
+; RV32IZbbZbt-NEXT: xor a1, a1, a3
+; RV32IZbbZbt-NEXT: andn a1, a5, a1
+; RV32IZbbZbt-NEXT: slti a3, a1, 0
+; RV32IZbbZbt-NEXT: cmov a1, a3, a4, a0
+; RV32IZbbZbt-NEXT: srai a0, a0, 31
+; RV32IZbbZbt-NEXT: cmov a0, a3, a0, a2
+; RV32IZbbZbt-NEXT: ret
+;
+; RV64IZbbZbt-LABEL: func2:
+; RV64IZbbZbt: # %bb.0:
+; RV64IZbbZbt-NEXT: add a2, a0, a1
+; RV64IZbbZbt-NEXT: slti a3, a2, 0
+; RV64IZbbZbt-NEXT: addi a4, zero, -1
+; RV64IZbbZbt-NEXT: slli a4, a4, 63
+; RV64IZbbZbt-NEXT: addi a5, a4, -1
+; RV64IZbbZbt-NEXT: cmov a3, a3, a5, a4
+; RV64IZbbZbt-NEXT: slt a0, a2, a0
+; RV64IZbbZbt-NEXT: slti a1, a1, 0
+; RV64IZbbZbt-NEXT: xor a0, a1, a0
+; RV64IZbbZbt-NEXT: cmov a0, a0, a3, a2
+; RV64IZbbZbt-NEXT: ret
%tmp = call i64 @llvm.sadd.sat.i64(i64 %x, i64 %y);
ret i64 %tmp;
}
diff --git a/llvm/test/CodeGen/RISCV/sadd_sat_plus.ll b/llvm/test/CodeGen/RISCV/sadd_sat_plus.ll
index 1767701e6eaf..69240024025b 100644
--- a/llvm/test/CodeGen/RISCV/sadd_sat_plus.ll
+++ b/llvm/test/CodeGen/RISCV/sadd_sat_plus.ll
@@ -1,8 +1,10 @@
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
; RUN: llc < %s -mtriple=riscv32 -mattr=+m | FileCheck %s --check-prefix=RV32I
; RUN: llc < %s -mtriple=riscv64 -mattr=+m | FileCheck %s --check-prefix=RV64I
-; RUN: llc < %s -mtriple=riscv32 -mattr=+m,+experimental-zbb | FileCheck %s --check-prefix=RV32IZbb
-; RUN: llc < %s -mtriple=riscv64 -mattr=+m,+experimental-zbb | FileCheck %s --check-prefix=RV64IZbb
+; RUN: llc < %s -mtriple=riscv32 -mattr=+m,+experimental-zbb | FileCheck %s --check-prefixes=RV32IZbb,RV32IZbbNOZbt
+; RUN: llc < %s -mtriple=riscv64 -mattr=+m,+experimental-zbb | FileCheck %s --check-prefixes=RV64IZbb,RV64IZbbNOZbt
+; RUN: llc < %s -mtriple=riscv32 -mattr=+m,+experimental-zbb,+experimental-zbt | FileCheck %s --check-prefixes=RV32IZbb,RV32IZbbZbt
+; RUN: llc < %s -mtriple=riscv64 -mattr=+m,+experimental-zbb,+experimental-zbt | FileCheck %s --check-prefixes=RV64IZbb,RV64IZbbZbt
declare i4 @llvm.sadd.sat.i4(i4, i4)
declare i8 @llvm.sadd.sat.i8(i8, i8)
@@ -48,23 +50,23 @@ define i32 @func32(i32 %x, i32 %y, i32 %z) nounwind {
; RV64I-NEXT: lui a0, 524288
; RV64I-NEXT: ret
;
-; RV32IZbb-LABEL: func32:
-; RV32IZbb: # %bb.0:
-; RV32IZbb-NEXT: mv a3, a0
-; RV32IZbb-NEXT: mul a2, a1, a2
-; RV32IZbb-NEXT: add a1, a0, a2
-; RV32IZbb-NEXT: lui a0, 524288
-; RV32IZbb-NEXT: bgez a1, .LBB0_2
-; RV32IZbb-NEXT: # %bb.1:
-; RV32IZbb-NEXT: addi a0, a0, -1
-; RV32IZbb-NEXT: .LBB0_2:
-; RV32IZbb-NEXT: slt a3, a1, a3
-; RV32IZbb-NEXT: slti a2, a2, 0
-; RV32IZbb-NEXT: bne a2, a3, .LBB0_4
-; RV32IZbb-NEXT: # %bb.3:
-; RV32IZbb-NEXT: mv a0, a1
-; RV32IZbb-NEXT: .LBB0_4:
-; RV32IZbb-NEXT: ret
+; RV32IZbbNOZbt-LABEL: func32:
+; RV32IZbbNOZbt: # %bb.0:
+; RV32IZbbNOZbt-NEXT: mv a3, a0
+; RV32IZbbNOZbt-NEXT: mul a2, a1, a2
+; RV32IZbbNOZbt-NEXT: add a1, a0, a2
+; RV32IZbbNOZbt-NEXT: lui a0, 524288
+; RV32IZbbNOZbt-NEXT: bgez a1, .LBB0_2
+; RV32IZbbNOZbt-NEXT: # %bb.1:
+; RV32IZbbNOZbt-NEXT: addi a0, a0, -1
+; RV32IZbbNOZbt-NEXT: .LBB0_2:
+; RV32IZbbNOZbt-NEXT: slt a3, a1, a3
+; RV32IZbbNOZbt-NEXT: slti a2, a2, 0
+; RV32IZbbNOZbt-NEXT: bne a2, a3, .LBB0_4
+; RV32IZbbNOZbt-NEXT: # %bb.3:
+; RV32IZbbNOZbt-NEXT: mv a0, a1
+; RV32IZbbNOZbt-NEXT: .LBB0_4:
+; RV32IZbbNOZbt-NEXT: ret
;
; RV64IZbb-LABEL: func32:
; RV64IZbb: # %bb.0:
@@ -76,6 +78,20 @@ define i32 @func32(i32 %x, i32 %y, i32 %z) nounwind {
; RV64IZbb-NEXT: min a0, a0, a2
; RV64IZbb-NEXT: max a0, a0, a1
; RV64IZbb-NEXT: ret
+;
+; RV32IZbbZbt-LABEL: func32:
+; RV32IZbbZbt: # %bb.0:
+; RV32IZbbZbt-NEXT: mul a1, a1, a2
+; RV32IZbbZbt-NEXT: add a2, a0, a1
+; RV32IZbbZbt-NEXT: slt a0, a2, a0
+; RV32IZbbZbt-NEXT: slti a1, a1, 0
+; RV32IZbbZbt-NEXT: xor a0, a1, a0
+; RV32IZbbZbt-NEXT: slti a1, a2, 0
+; RV32IZbbZbt-NEXT: lui a3, 524288
+; RV32IZbbZbt-NEXT: addi a4, a3, -1
+; RV32IZbbZbt-NEXT: cmov a1, a1, a4, a3
+; RV32IZbbZbt-NEXT: cmov a0, a0, a1, a2
+; RV32IZbbZbt-NEXT: ret
%a = mul i32 %y, %z
%tmp = call i32 @llvm.sadd.sat.i32(i32 %x, i32 %a)
ret i32 %tmp
@@ -125,47 +141,80 @@ define i64 @func64(i64 %x, i64 %y, i64 %z) nounwind {
; RV64I-NEXT: .LBB1_4:
; RV64I-NEXT: ret
;
-; RV32IZbb-LABEL: func64:
-; RV32IZbb: # %bb.0:
-; RV32IZbb-NEXT: mv a2, a1
-; RV32IZbb-NEXT: mv a1, a0
-; RV32IZbb-NEXT: add a3, a2, a5
-; RV32IZbb-NEXT: add a0, a0, a4
-; RV32IZbb-NEXT: sltu a1, a0, a1
-; RV32IZbb-NEXT: add a3, a3, a1
-; RV32IZbb-NEXT: lui a1, 524288
-; RV32IZbb-NEXT: bgez a3, .LBB1_2
-; RV32IZbb-NEXT: # %bb.1:
-; RV32IZbb-NEXT: addi a1, a1, -1
-; RV32IZbb-NEXT: .LBB1_2:
-; RV32IZbb-NEXT: xor a4, a2, a3
-; RV32IZbb-NEXT: xor a2, a2, a5
-; RV32IZbb-NEXT: andn a2, a4, a2
-; RV32IZbb-NEXT: bltz a2, .LBB1_4
-; RV32IZbb-NEXT: # %bb.3:
-; RV32IZbb-NEXT: mv a1, a3
-; RV32IZbb-NEXT: ret
-; RV32IZbb-NEXT: .LBB1_4:
-; RV32IZbb-NEXT: srai a0, a3, 31
-; RV32IZbb-NEXT: ret
+; RV32IZbbNOZbt-LABEL: func64:
+; RV32IZbbNOZbt: # %bb.0:
+; RV32IZbbNOZbt-NEXT: mv a2, a1
+; RV32IZbbNOZbt-NEXT: mv a1, a0
+; RV32IZbbNOZbt-NEXT: add a3, a2, a5
+; RV32IZbbNOZbt-NEXT: add a0, a0, a4
+; RV32IZbbNOZbt-NEXT: sltu a1, a0, a1
+; RV32IZbbNOZbt-NEXT: add a3, a3, a1
+; RV32IZbbNOZbt-NEXT: lui a1, 524288
+; RV32IZbbNOZbt-NEXT: bgez a3, .LBB1_2
+; RV32IZbbNOZbt-NEXT: # %bb.1:
+; RV32IZbbNOZbt-NEXT: addi a1, a1, -1
+; RV32IZbbNOZbt-NEXT: .LBB1_2:
+; RV32IZbbNOZbt-NEXT: xor a4, a2, a3
+; RV32IZbbNOZbt-NEXT: xor a2, a2, a5
+; RV32IZbbNOZbt-NEXT: andn a2, a4, a2
+; RV32IZbbNOZbt-NEXT: bltz a2, .LBB1_4
+; RV32IZbbNOZbt-NEXT: # %bb.3:
+; RV32IZbbNOZbt-NEXT: mv a1, a3
+; RV32IZbbNOZbt-NEXT: ret
+; RV32IZbbNOZbt-NEXT: .LBB1_4:
+; RV32IZbbNOZbt-NEXT: srai a0, a3, 31
+; RV32IZbbNOZbt-NEXT: ret
;
-; RV64IZbb-LABEL: func64:
-; RV64IZbb: # %bb.0:
-; RV64IZbb-NEXT: mv a1, a0
-; RV64IZbb-NEXT: add a3, a0, a2
-; RV64IZbb-NEXT: addi a0, zero, -1
-; RV64IZbb-NEXT: slli a0, a0, 63
-; RV64IZbb-NEXT: bgez a3, .LBB1_2
-; RV64IZbb-NEXT: # %bb.1:
-; RV64IZbb-NEXT: addi a0, a0, -1
-; RV64IZbb-NEXT: .LBB1_2:
-; RV64IZbb-NEXT: slt a1, a3, a1
-; RV64IZbb-NEXT: slti a2, a2, 0
-; RV64IZbb-NEXT: bne a2, a1, .LBB1_4
-; RV64IZbb-NEXT: # %bb.3:
-; RV64IZbb-NEXT: mv a0, a3
-; RV64IZbb-NEXT: .LBB1_4:
-; RV64IZbb-NEXT: ret
+; RV64IZbbNOZbt-LABEL: func64:
+; RV64IZbbNOZbt: # %bb.0:
+; RV64IZbbNOZbt-NEXT: mv a1, a0
+; RV64IZbbNOZbt-NEXT: add a3, a0, a2
+; RV64IZbbNOZbt-NEXT: addi a0, zero, -1
+; RV64IZbbNOZbt-NEXT: slli a0, a0, 63
+; RV64IZbbNOZbt-NEXT: bgez a3, .LBB1_2
+; RV64IZbbNOZbt-NEXT: # %bb.1:
+; RV64IZbbNOZbt-NEXT: addi a0, a0, -1
+; RV64IZbbNOZbt-NEXT: .LBB1_2:
+; RV64IZbbNOZbt-NEXT: slt a1, a3, a1
+; RV64IZbbNOZbt-NEXT: slti a2, a2, 0
+; RV64IZbbNOZbt-NEXT: bne a2, a1, .LBB1_4
+; RV64IZbbNOZbt-NEXT: # %bb.3:
+; RV64IZbbNOZbt-NEXT: mv a0, a3
+; RV64IZbbNOZbt-NEXT: .LBB1_4:
+; RV64IZbbNOZbt-NEXT: ret
+;
+; RV32IZbbZbt-LABEL: func64:
+; RV32IZbbZbt: # %bb.0:
+; RV32IZbbZbt-NEXT: add a2, a1, a5
+; RV32IZbbZbt-NEXT: add a3, a0, a4
+; RV32IZbbZbt-NEXT: sltu a0, a3, a0
+; RV32IZbbZbt-NEXT: add a0, a2, a0
+; RV32IZbbZbt-NEXT: slti a2, a0, 0
+; RV32IZbbZbt-NEXT: lui a6, 524288
+; RV32IZbbZbt-NEXT: addi a4, a6, -1
+; RV32IZbbZbt-NEXT: cmov a2, a2, a4, a6
+; RV32IZbbZbt-NEXT: xor a4, a1, a0
+; RV32IZbbZbt-NEXT: xor a1, a1, a5
+; RV32IZbbZbt-NEXT: andn a1, a4, a1
+; RV32IZbbZbt-NEXT: slti a4, a1, 0
+; RV32IZbbZbt-NEXT: cmov a1, a4, a2, a0
+; RV32IZbbZbt-NEXT: srai a0, a0, 31
+; RV32IZbbZbt-NEXT: cmov a0, a4, a0, a3
+; RV32IZbbZbt-NEXT: ret
+;
+; RV64IZbbZbt-LABEL: func64:
+; RV64IZbbZbt: # %bb.0:
+; RV64IZbbZbt-NEXT: add a1, a0, a2
+; RV64IZbbZbt-NEXT: slti a3, a1, 0
+; RV64IZbbZbt-NEXT: addi a4, zero, -1
+; RV64IZbbZbt-NEXT: slli a4, a4, 63
+; RV64IZbbZbt-NEXT: addi a5, a4, -1
+; RV64IZbbZbt-NEXT: cmov a3, a3, a5, a4
+; RV64IZbbZbt-NEXT: slt a0, a1, a0
+; RV64IZbbZbt-NEXT: slti a2, a2, 0
+; RV64IZbbZbt-NEXT: xor a0, a2, a0
+; RV64IZbbZbt-NEXT: cmov a0, a0, a3, a1
+; RV64IZbbZbt-NEXT: ret
%a = mul i64 %y, %z
%tmp = call i64 @llvm.sadd.sat.i64(i64 %x, i64 %z)
ret i64 %tmp
diff --git a/llvm/test/CodeGen/RISCV/ssub_sat.ll b/llvm/test/CodeGen/RISCV/ssub_sat.ll
index 61d5a4833ba9..008a3d0cf8ad 100644
--- a/llvm/test/CodeGen/RISCV/ssub_sat.ll
+++ b/llvm/test/CodeGen/RISCV/ssub_sat.ll
@@ -1,8 +1,10 @@
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
; RUN: llc < %s -mtriple=riscv32 -mattr=+m | FileCheck %s --check-prefix=RV32I
; RUN: llc < %s -mtriple=riscv64 -mattr=+m | FileCheck %s --check-prefix=RV64I
-; RUN: llc < %s -mtriple=riscv32 -mattr=+m,+experimental-zbb | FileCheck %s --check-prefix=RV32IZbb
-; RUN: llc < %s -mtriple=riscv64 -mattr=+m,+experimental-zbb | FileCheck %s --check-prefix=RV64IZbb
+; RUN: llc < %s -mtriple=riscv32 -mattr=+m,+experimental-zbb | FileCheck %s --check-prefixes=RV32IZbb,RV32IZbbNOZbt
+; RUN: llc < %s -mtriple=riscv64 -mattr=+m,+experimental-zbb | FileCheck %s --check-prefixes=RV64IZbb,RV64IZbbNOZbt
+; RUN: llc < %s -mtriple=riscv32 -mattr=+m,+experimental-zbb,+experimental-zbt | FileCheck %s --check-prefixes=RV32IZbb,RV32IZbbZbt
+; RUN: llc < %s -mtriple=riscv64 -mattr=+m,+experimental-zbb,+experimental-zbt | FileCheck %s --check-prefixes=RV64IZbb,RV64IZbbZbt
declare i4 @llvm.ssub.sat.i4(i4, i4)
declare i8 @llvm.ssub.sat.i8(i8, i8)
@@ -45,22 +47,22 @@ define signext i32 @func(i32 signext %x, i32 signext %y) nounwind {
; RV64I-NEXT: lui a0, 524288
; RV64I-NEXT: ret
;
-; RV32IZbb-LABEL: func:
-; RV32IZbb: # %bb.0:
-; RV32IZbb-NEXT: mv a2, a0
-; RV32IZbb-NEXT: sub a3, a0, a1
-; RV32IZbb-NEXT: lui a0, 524288
-; RV32IZbb-NEXT: bgez a3, .LBB0_2
-; RV32IZbb-NEXT: # %bb.1:
-; RV32IZbb-NEXT: addi a0, a0, -1
-; RV32IZbb-NEXT: .LBB0_2:
-; RV32IZbb-NEXT: sgtz a1, a1
-; RV32IZbb-NEXT: slt a2, a3, a2
-; RV32IZbb-NEXT: bne a1, a2, .LBB0_4
-; RV32IZbb-NEXT: # %bb.3:
-; RV32IZbb-NEXT: mv a0, a3
-; RV32IZbb-NEXT: .LBB0_4:
-; RV32IZbb-NEXT: ret
+; RV32IZbbNOZbt-LABEL: func:
+; RV32IZbbNOZbt: # %bb.0:
+; RV32IZbbNOZbt-NEXT: mv a2, a0
+; RV32IZbbNOZbt-NEXT: sub a3, a0, a1
+; RV32IZbbNOZbt-NEXT: lui a0, 524288
+; RV32IZbbNOZbt-NEXT: bgez a3, .LBB0_2
+; RV32IZbbNOZbt-NEXT: # %bb.1:
+; RV32IZbbNOZbt-NEXT: addi a0, a0, -1
+; RV32IZbbNOZbt-NEXT: .LBB0_2:
+; RV32IZbbNOZbt-NEXT: sgtz a1, a1
+; RV32IZbbNOZbt-NEXT: slt a2, a3, a2
+; RV32IZbbNOZbt-NEXT: bne a1, a2, .LBB0_4
+; RV32IZbbNOZbt-NEXT: # %bb.3:
+; RV32IZbbNOZbt-NEXT: mv a0, a3
+; RV32IZbbNOZbt-NEXT: .LBB0_4:
+; RV32IZbbNOZbt-NEXT: ret
;
; RV64IZbb-LABEL: func:
; RV64IZbb: # %bb.0:
@@ -70,6 +72,19 @@ define signext i32 @func(i32 signext %x, i32 signext %y) nounwind {
; RV64IZbb-NEXT: min a0, a0, a2
; RV64IZbb-NEXT: max a0, a0, a1
; RV64IZbb-NEXT: ret
+;
+; RV32IZbbZbt-LABEL: func:
+; RV32IZbbZbt: # %bb.0:
+; RV32IZbbZbt-NEXT: sgtz a2, a1
+; RV32IZbbZbt-NEXT: sub a1, a0, a1
+; RV32IZbbZbt-NEXT: slt a0, a1, a0
+; RV32IZbbZbt-NEXT: xor a0, a2, a0
+; RV32IZbbZbt-NEXT: slti a2, a1, 0
+; RV32IZbbZbt-NEXT: lui a3, 524288
+; RV32IZbbZbt-NEXT: addi a4, a3, -1
+; RV32IZbbZbt-NEXT: cmov a2, a2, a4, a3
+; RV32IZbbZbt-NEXT: cmov a0, a0, a2, a1
+; RV32IZbbZbt-NEXT: ret
%tmp = call i32 @llvm.ssub.sat.i32(i32 %x, i32 %y);
ret i32 %tmp;
}
@@ -116,46 +131,79 @@ define i64 @func2(i64 %x, i64 %y) nounwind {
; RV64I-NEXT: .LBB1_4:
; RV64I-NEXT: ret
;
-; RV32IZbb-LABEL: func2:
-; RV32IZbb: # %bb.0:
-; RV32IZbb-NEXT: mv a4, a1
-; RV32IZbb-NEXT: sltu a1, a0, a2
-; RV32IZbb-NEXT: sub a5, a4, a3
-; RV32IZbb-NEXT: sub a5, a5, a1
-; RV32IZbb-NEXT: lui a1, 524288
-; RV32IZbb-NEXT: bgez a5, .LBB1_2
-; RV32IZbb-NEXT: # %bb.1:
-; RV32IZbb-NEXT: addi a1, a1, -1
-; RV32IZbb-NEXT: .LBB1_2:
-; RV32IZbb-NEXT: xor a6, a4, a5
-; RV32IZbb-NEXT: xor a3, a4, a3
-; RV32IZbb-NEXT: and a3, a3, a6
-; RV32IZbb-NEXT: bltz a3, .LBB1_4
-; RV32IZbb-NEXT: # %bb.3:
-; RV32IZbb-NEXT: sub a0, a0, a2
-; RV32IZbb-NEXT: mv a1, a5
-; RV32IZbb-NEXT: ret
-; RV32IZbb-NEXT: .LBB1_4:
-; RV32IZbb-NEXT: srai a0, a5, 31
-; RV32IZbb-NEXT: ret
+; RV32IZbbNOZbt-LABEL: func2:
+; RV32IZbbNOZbt: # %bb.0:
+; RV32IZbbNOZbt-NEXT: mv a4, a1
+; RV32IZbbNOZbt-NEXT: sltu a1, a0, a2
+; RV32IZbbNOZbt-NEXT: sub a5, a4, a3
+; RV32IZbbNOZbt-NEXT: sub a5, a5, a1
+; RV32IZbbNOZbt-NEXT: lui a1, 524288
+; RV32IZbbNOZbt-NEXT: bgez a5, .LBB1_2
+; RV32IZbbNOZbt-NEXT: # %bb.1:
+; RV32IZbbNOZbt-NEXT: addi a1, a1, -1
+; RV32IZbbNOZbt-NEXT: .LBB1_2:
+; RV32IZbbNOZbt-NEXT: xor a6, a4, a5
+; RV32IZbbNOZbt-NEXT: xor a3, a4, a3
+; RV32IZbbNOZbt-NEXT: and a3, a3, a6
+; RV32IZbbNOZbt-NEXT: bltz a3, .LBB1_4
+; RV32IZbbNOZbt-NEXT: # %bb.3:
+; RV32IZbbNOZbt-NEXT: sub a0, a0, a2
+; RV32IZbbNOZbt-NEXT: mv a1, a5
+; RV32IZbbNOZbt-NEXT: ret
+; RV32IZbbNOZbt-NEXT: .LBB1_4:
+; RV32IZbbNOZbt-NEXT: srai a0, a5, 31
+; RV32IZbbNOZbt-NEXT: ret
;
-; RV64IZbb-LABEL: func2:
-; RV64IZbb: # %bb.0:
-; RV64IZbb-NEXT: mv a2, a0
-; RV64IZbb-NEXT: sub a3, a0, a1
-; RV64IZbb-NEXT: addi a0, zero, -1
-; RV64IZbb-NEXT: slli a0, a0, 63
-; RV64IZbb-NEXT: bgez a3, .LBB1_2
-; RV64IZbb-NEXT: # %bb.1:
-; RV64IZbb-NEXT: addi a0, a0, -1
-; RV64IZbb-NEXT: .LBB1_2:
-; RV64IZbb-NEXT: sgtz a1, a1
-; RV64IZbb-NEXT: slt a2, a3, a2
-; RV64IZbb-NEXT: bne a1, a2, .LBB1_4
-; RV64IZbb-NEXT: # %bb.3:
-; RV64IZbb-NEXT: mv a0, a3
-; RV64IZbb-NEXT: .LBB1_4:
-; RV64IZbb-NEXT: ret
+; RV64IZbbNOZbt-LABEL: func2:
+; RV64IZbbNOZbt: # %bb.0:
+; RV64IZbbNOZbt-NEXT: mv a2, a0
+; RV64IZbbNOZbt-NEXT: sub a3, a0, a1
+; RV64IZbbNOZbt-NEXT: addi a0, zero, -1
+; RV64IZbbNOZbt-NEXT: slli a0, a0, 63
+; RV64IZbbNOZbt-NEXT: bgez a3, .LBB1_2
+; RV64IZbbNOZbt-NEXT: # %bb.1:
+; RV64IZbbNOZbt-NEXT: addi a0, a0, -1
+; RV64IZbbNOZbt-NEXT: .LBB1_2:
+; RV64IZbbNOZbt-NEXT: sgtz a1, a1
+; RV64IZbbNOZbt-NEXT: slt a2, a3, a2
+; RV64IZbbNOZbt-NEXT: bne a1, a2, .LBB1_4
+; RV64IZbbNOZbt-NEXT: # %bb.3:
+; RV64IZbbNOZbt-NEXT: mv a0, a3
+; RV64IZbbNOZbt-NEXT: .LBB1_4:
+; RV64IZbbNOZbt-NEXT: ret
+;
+; RV32IZbbZbt-LABEL: func2:
+; RV32IZbbZbt: # %bb.0:
+; RV32IZbbZbt-NEXT: sltu a4, a0, a2
+; RV32IZbbZbt-NEXT: sub a5, a1, a3
+; RV32IZbbZbt-NEXT: sub a4, a5, a4
+; RV32IZbbZbt-NEXT: slti a7, a4, 0
+; RV32IZbbZbt-NEXT: lui a6, 524288
+; RV32IZbbZbt-NEXT: addi a5, a6, -1
+; RV32IZbbZbt-NEXT: cmov a6, a7, a5, a6
+; RV32IZbbZbt-NEXT: xor a5, a1, a4
+; RV32IZbbZbt-NEXT: xor a1, a1, a3
+; RV32IZbbZbt-NEXT: and a1, a1, a5
+; RV32IZbbZbt-NEXT: slti a3, a1, 0
+; RV32IZbbZbt-NEXT: cmov a1, a3, a6, a4
+; RV32IZbbZbt-NEXT: srai a4, a4, 31
+; RV32IZbbZbt-NEXT: sub a0, a0, a2
+; RV32IZbbZbt-NEXT: cmov a0, a3, a4, a0
+; RV32IZbbZbt-NEXT: ret
+;
+; RV64IZbbZbt-LABEL: func2:
+; RV64IZbbZbt: # %bb.0:
+; RV64IZbbZbt-NEXT: sgtz a2, a1
+; RV64IZbbZbt-NEXT: sub a1, a0, a1
+; RV64IZbbZbt-NEXT: slt a0, a1, a0
+; RV64IZbbZbt-NEXT: xor a0, a2, a0
+; RV64IZbbZbt-NEXT: slti a2, a1, 0
+; RV64IZbbZbt-NEXT: addi a3, zero, -1
+; RV64IZbbZbt-NEXT: slli a3, a3, 63
+; RV64IZbbZbt-NEXT: addi a4, a3, -1
+; RV64IZbbZbt-NEXT: cmov a2, a2, a4, a3
+; RV64IZbbZbt-NEXT: cmov a0, a0, a2, a1
+; RV64IZbbZbt-NEXT: ret
%tmp = call i64 @llvm.ssub.sat.i64(i64 %x, i64 %y);
ret i64 %tmp;
}
diff --git a/llvm/test/CodeGen/RISCV/ssub_sat_plus.ll b/llvm/test/CodeGen/RISCV/ssub_sat_plus.ll
index 7f2e1595a98d..a30ad68631c4 100644
--- a/llvm/test/CodeGen/RISCV/ssub_sat_plus.ll
+++ b/llvm/test/CodeGen/RISCV/ssub_sat_plus.ll
@@ -1,8 +1,10 @@
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
; RUN: llc < %s -mtriple=riscv32 -mattr=+m | FileCheck %s --check-prefix=RV32I
; RUN: llc < %s -mtriple=riscv64 -mattr=+m | FileCheck %s --check-prefix=RV64I
-; RUN: llc < %s -mtriple=riscv32 -mattr=+m,+experimental-zbb | FileCheck %s --check-prefix=RV32IZbb
-; RUN: llc < %s -mtriple=riscv64 -mattr=+m,+experimental-zbb | FileCheck %s --check-prefix=RV64IZbb
+; RUN: llc < %s -mtriple=riscv32 -mattr=+m,+experimental-zbb | FileCheck %s --check-prefixes=RV32IZbb,RV32IZbbNOZbt
+; RUN: llc < %s -mtriple=riscv64 -mattr=+m,+experimental-zbb | FileCheck %s --check-prefixes=RV64IZbb,RV64IZbbNOZbt
+; RUN: llc < %s -mtriple=riscv32 -mattr=+m,+experimental-zbb,+experimental-zbt | FileCheck %s --check-prefixes=RV32IZbb,RV32IZbbZbt
+; RUN: llc < %s -mtriple=riscv64 -mattr=+m,+experimental-zbb,+experimental-zbt | FileCheck %s --check-prefixes=RV64IZbb,RV64IZbbZbt
declare i4 @llvm.ssub.sat.i4(i4, i4)
declare i8 @llvm.ssub.sat.i8(i8, i8)
@@ -48,23 +50,23 @@ define i32 @func32(i32 %x, i32 %y, i32 %z) nounwind {
; RV64I-NEXT: lui a0, 524288
; RV64I-NEXT: ret
;
-; RV32IZbb-LABEL: func32:
-; RV32IZbb: # %bb.0:
-; RV32IZbb-NEXT: mv a3, a0
-; RV32IZbb-NEXT: mul a2, a1, a2
-; RV32IZbb-NEXT: sub a1, a0, a2
-; RV32IZbb-NEXT: lui a0, 524288
-; RV32IZbb-NEXT: bgez a1, .LBB0_2
-; RV32IZbb-NEXT: # %bb.1:
-; RV32IZbb-NEXT: addi a0, a0, -1
-; RV32IZbb-NEXT: .LBB0_2:
-; RV32IZbb-NEXT: sgtz a2, a2
-; RV32IZbb-NEXT: slt a3, a1, a3
-; RV32IZbb-NEXT: bne a2, a3, .LBB0_4
-; RV32IZbb-NEXT: # %bb.3:
-; RV32IZbb-NEXT: mv a0, a1
-; RV32IZbb-NEXT: .LBB0_4:
-; RV32IZbb-NEXT: ret
+; RV32IZbbNOZbt-LABEL: func32:
+; RV32IZbbNOZbt: # %bb.0:
+; RV32IZbbNOZbt-NEXT: mv a3, a0
+; RV32IZbbNOZbt-NEXT: mul a2, a1, a2
+; RV32IZbbNOZbt-NEXT: sub a1, a0, a2
+; RV32IZbbNOZbt-NEXT: lui a0, 524288
+; RV32IZbbNOZbt-NEXT: bgez a1, .LBB0_2
+; RV32IZbbNOZbt-NEXT: # %bb.1:
+; RV32IZbbNOZbt-NEXT: addi a0, a0, -1
+; RV32IZbbNOZbt-NEXT: .LBB0_2:
+; RV32IZbbNOZbt-NEXT: sgtz a2, a2
+; RV32IZbbNOZbt-NEXT: slt a3, a1, a3
+; RV32IZbbNOZbt-NEXT: bne a2, a3, .LBB0_4
+; RV32IZbbNOZbt-NEXT: # %bb.3:
+; RV32IZbbNOZbt-NEXT: mv a0, a1
+; RV32IZbbNOZbt-NEXT: .LBB0_4:
+; RV32IZbbNOZbt-NEXT: ret
;
; RV64IZbb-LABEL: func32:
; RV64IZbb: # %bb.0:
@@ -76,6 +78,20 @@ define i32 @func32(i32 %x, i32 %y, i32 %z) nounwind {
; RV64IZbb-NEXT: min a0, a0, a2
; RV64IZbb-NEXT: max a0, a0, a1
; RV64IZbb-NEXT: ret
+;
+; RV32IZbbZbt-LABEL: func32:
+; RV32IZbbZbt: # %bb.0:
+; RV32IZbbZbt-NEXT: mul a1, a1, a2
+; RV32IZbbZbt-NEXT: sgtz a2, a1
+; RV32IZbbZbt-NEXT: sub a1, a0, a1
+; RV32IZbbZbt-NEXT: slt a0, a1, a0
+; RV32IZbbZbt-NEXT: xor a0, a2, a0
+; RV32IZbbZbt-NEXT: slti a2, a1, 0
+; RV32IZbbZbt-NEXT: lui a3, 524288
+; RV32IZbbZbt-NEXT: addi a4, a3, -1
+; RV32IZbbZbt-NEXT: cmov a2, a2, a4, a3
+; RV32IZbbZbt-NEXT: cmov a0, a0, a2, a1
+; RV32IZbbZbt-NEXT: ret
%a = mul i32 %y, %z
%tmp = call i32 @llvm.ssub.sat.i32(i32 %x, i32 %a)
ret i32 %tmp
@@ -123,46 +139,79 @@ define i64 @func64(i64 %x, i64 %y, i64 %z) nounwind {
; RV64I-NEXT: .LBB1_4:
; RV64I-NEXT: ret
;
-; RV32IZbb-LABEL: func64:
-; RV32IZbb: # %bb.0:
-; RV32IZbb-NEXT: mv a2, a1
-; RV32IZbb-NEXT: sltu a1, a0, a4
-; RV32IZbb-NEXT: sub a3, a2, a5
-; RV32IZbb-NEXT: sub a3, a3, a1
-; RV32IZbb-NEXT: lui a1, 524288
-; RV32IZbb-NEXT: bgez a3, .LBB1_2
-; RV32IZbb-NEXT: # %bb.1:
-; RV32IZbb-NEXT: addi a1, a1, -1
-; RV32IZbb-NEXT: .LBB1_2:
-; RV32IZbb-NEXT: xor a6, a2, a3
-; RV32IZbb-NEXT: xor a2, a2, a5
-; RV32IZbb-NEXT: and a2, a2, a6
-; RV32IZbb-NEXT: bltz a2, .LBB1_4
-; RV32IZbb-NEXT: # %bb.3:
-; RV32IZbb-NEXT: sub a0, a0, a4
-; RV32IZbb-NEXT: mv a1, a3
-; RV32IZbb-NEXT: ret
-; RV32IZbb-NEXT: .LBB1_4:
-; RV32IZbb-NEXT: srai a0, a3, 31
-; RV32IZbb-NEXT: ret
+; RV32IZbbNOZbt-LABEL: func64:
+; RV32IZbbNOZbt: # %bb.0:
+; RV32IZbbNOZbt-NEXT: mv a2, a1
+; RV32IZbbNOZbt-NEXT: sltu a1, a0, a4
+; RV32IZbbNOZbt-NEXT: sub a3, a2, a5
+; RV32IZbbNOZbt-NEXT: sub a3, a3, a1
+; RV32IZbbNOZbt-NEXT: lui a1, 524288
+; RV32IZbbNOZbt-NEXT: bgez a3, .LBB1_2
+; RV32IZbbNOZbt-NEXT: # %bb.1:
+; RV32IZbbNOZbt-NEXT: addi a1, a1, -1
+; RV32IZbbNOZbt-NEXT: .LBB1_2:
+; RV32IZbbNOZbt-NEXT: xor a6, a2, a3
+; RV32IZbbNOZbt-NEXT: xor a2, a2, a5
+; RV32IZbbNOZbt-NEXT: and a2, a2, a6
+; RV32IZbbNOZbt-NEXT: bltz a2, .LBB1_4
+; RV32IZbbNOZbt-NEXT: # %bb.3:
+; RV32IZbbNOZbt-NEXT: sub a0, a0, a4
+; RV32IZbbNOZbt-NEXT: mv a1, a3
+; RV32IZbbNOZbt-NEXT: ret
+; RV32IZbbNOZbt-NEXT: .LBB1_4:
+; RV32IZbbNOZbt-NEXT: srai a0, a3, 31
+; RV32IZbbNOZbt-NEXT: ret
;
-; RV64IZbb-LABEL: func64:
-; RV64IZbb: # %bb.0:
-; RV64IZbb-NEXT: mv a1, a0
-; RV64IZbb-NEXT: sub a3, a0, a2
-; RV64IZbb-NEXT: addi a0, zero, -1
-; RV64IZbb-NEXT: slli a0, a0, 63
-; RV64IZbb-NEXT: bgez a3, .LBB1_2
-; RV64IZbb-NEXT: # %bb.1:
-; RV64IZbb-NEXT: addi a0, a0, -1
-; RV64IZbb-NEXT: .LBB1_2:
-; RV64IZbb-NEXT: sgtz a2, a2
-; RV64IZbb-NEXT: slt a1, a3, a1
-; RV64IZbb-NEXT: bne a2, a1, .LBB1_4
-; RV64IZbb-NEXT: # %bb.3:
-; RV64IZbb-NEXT: mv a0, a3
-; RV64IZbb-NEXT: .LBB1_4:
-; RV64IZbb-NEXT: ret
+; RV64IZbbNOZbt-LABEL: func64:
+; RV64IZbbNOZbt: # %bb.0:
+; RV64IZbbNOZbt-NEXT: mv a1, a0
+; RV64IZbbNOZbt-NEXT: sub a3, a0, a2
+; RV64IZbbNOZbt-NEXT: addi a0, zero, -1
+; RV64IZbbNOZbt-NEXT: slli a0, a0, 63
+; RV64IZbbNOZbt-NEXT: bgez a3, .LBB1_2
+; RV64IZbbNOZbt-NEXT: # %bb.1:
+; RV64IZbbNOZbt-NEXT: addi a0, a0, -1
+; RV64IZbbNOZbt-NEXT: .LBB1_2:
+; RV64IZbbNOZbt-NEXT: sgtz a2, a2
+; RV64IZbbNOZbt-NEXT: slt a1, a3, a1
+; RV64IZbbNOZbt-NEXT: bne a2, a1, .LBB1_4
+; RV64IZbbNOZbt-NEXT: # %bb.3:
+; RV64IZbbNOZbt-NEXT: mv a0, a3
+; RV64IZbbNOZbt-NEXT: .LBB1_4:
+; RV64IZbbNOZbt-NEXT: ret
+;
+; RV32IZbbZbt-LABEL: func64:
+; RV32IZbbZbt: # %bb.0:
+; RV32IZbbZbt-NEXT: sltu a2, a0, a4
+; RV32IZbbZbt-NEXT: sub a3, a1, a5
+; RV32IZbbZbt-NEXT: sub a2, a3, a2
+; RV32IZbbZbt-NEXT: slti a7, a2, 0
+; RV32IZbbZbt-NEXT: lui a6, 524288
+; RV32IZbbZbt-NEXT: addi a3, a6, -1
+; RV32IZbbZbt-NEXT: cmov a6, a7, a3, a6
+; RV32IZbbZbt-NEXT: xor a3, a1, a2
+; RV32IZbbZbt-NEXT: xor a1, a1, a5
+; RV32IZbbZbt-NEXT: and a1, a1, a3
+; RV32IZbbZbt-NEXT: slti a3, a1, 0
+; RV32IZbbZbt-NEXT: cmov a1, a3, a6, a2
+; RV32IZbbZbt-NEXT: srai a2, a2, 31
+; RV32IZbbZbt-NEXT: sub a0, a0, a4
+; RV32IZbbZbt-NEXT: cmov a0, a3, a2, a0
+; RV32IZbbZbt-NEXT: ret
+;
+; RV64IZbbZbt-LABEL: func64:
+; RV64IZbbZbt: # %bb.0:
+; RV64IZbbZbt-NEXT: sgtz a1, a2
+; RV64IZbbZbt-NEXT: sub a2, a0, a2
+; RV64IZbbZbt-NEXT: slt a0, a2, a0
+; RV64IZbbZbt-NEXT: xor a0, a1, a0
+; RV64IZbbZbt-NEXT: slti a1, a2, 0
+; RV64IZbbZbt-NEXT: addi a3, zero, -1
+; RV64IZbbZbt-NEXT: slli a3, a3, 63
+; RV64IZbbZbt-NEXT: addi a4, a3, -1
+; RV64IZbbZbt-NEXT: cmov a1, a1, a4, a3
+; RV64IZbbZbt-NEXT: cmov a0, a0, a1, a2
+; RV64IZbbZbt-NEXT: ret
%a = mul i64 %y, %z
%tmp = call i64 @llvm.ssub.sat.i64(i64 %x, i64 %z)
ret i64 %tmp
More information about the llvm-commits
mailing list