[RISCV] Add Zba command lines to xaluo.ll. NFC

Some of the patterns end up with 32 to 64 bit zero extends on RV64
which can be handled by zext.w.

diff --git a/llvm/test/CodeGen/RISCV/xaluo.ll b/llvm/test/CodeGen/RISCV/xaluo.ll
index facc0f2..758cf4c 100644 (file)
--- a/llvm/test/CodeGen/RISCV/xaluo.ll
+++ b/llvm/test/CodeGen/RISCV/xaluo.ll
@@ -1,6 +1,8 @@
 ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
 ; RUN: llc < %s -mtriple=riscv32 -mattr=+m -verify-machineinstrs | FileCheck %s --check-prefix=RV32
 ; RUN: llc < %s -mtriple=riscv64 -mattr=+m -verify-machineinstrs | FileCheck %s --check-prefix=RV64
+; RUN: llc < %s -mtriple=riscv32 -mattr=+m,+experimental-zba -verify-machineinstrs | FileCheck %s --check-prefix=RV32ZBA
+; RUN: llc < %s -mtriple=riscv64 -mattr=+m,+experimental-zba -verify-machineinstrs | FileCheck %s --check-prefix=RV64ZBA
 
 ;
 ; Get the actual value of the overflow bit.
@@ -25,6 +27,26 @@ define zeroext i1 @saddo1.i32(i32 %v1, i32 %v2, i32* %res) {
 ; RV64-NEXT:    snez a0, a0
 ; RV64-NEXT:    sw a3, 0(a2)
 ; RV64-NEXT:    ret
+;
+; RV32ZBA-LABEL: saddo1.i32:
+; RV32ZBA:       # %bb.0: # %entry
+; RV32ZBA-NEXT:    add a3, a0, a1
+; RV32ZBA-NEXT:    slt a0, a3, a0
+; RV32ZBA-NEXT:    slti a1, a1, 0
+; RV32ZBA-NEXT:    xor a0, a1, a0
+; RV32ZBA-NEXT:    sw a3, 0(a2)
+; RV32ZBA-NEXT:    ret
+;
+; RV64ZBA-LABEL: saddo1.i32:
+; RV64ZBA:       # %bb.0: # %entry
+; RV64ZBA-NEXT:    sext.w a1, a1
+; RV64ZBA-NEXT:    sext.w a0, a0
+; RV64ZBA-NEXT:    add a3, a0, a1
+; RV64ZBA-NEXT:    addw a0, a0, a1
+; RV64ZBA-NEXT:    xor a0, a0, a3
+; RV64ZBA-NEXT:    snez a0, a0
+; RV64ZBA-NEXT:    sw a3, 0(a2)
+; RV64ZBA-NEXT:    ret
 entry:
   %t = call {i32, i1} @llvm.sadd.with.overflow.i32(i32 %v1, i32 %v2)
   %val = extractvalue {i32, i1} %t, 0
@@ -51,6 +73,23 @@ define zeroext i1 @saddo2.i32(i32 %v1, i32* %res) {
 ; RV64-NEXT:    snez a0, a0
 ; RV64-NEXT:    sw a2, 0(a1)
 ; RV64-NEXT:    ret
+;
+; RV32ZBA-LABEL: saddo2.i32:
+; RV32ZBA:       # %bb.0: # %entry
+; RV32ZBA-NEXT:    addi a2, a0, 4
+; RV32ZBA-NEXT:    slt a0, a2, a0
+; RV32ZBA-NEXT:    sw a2, 0(a1)
+; RV32ZBA-NEXT:    ret
+;
+; RV64ZBA-LABEL: saddo2.i32:
+; RV64ZBA:       # %bb.0: # %entry
+; RV64ZBA-NEXT:    sext.w a0, a0
+; RV64ZBA-NEXT:    addi a2, a0, 4
+; RV64ZBA-NEXT:    addiw a0, a0, 4
+; RV64ZBA-NEXT:    xor a0, a0, a2
+; RV64ZBA-NEXT:    snez a0, a0
+; RV64ZBA-NEXT:    sw a2, 0(a1)
+; RV64ZBA-NEXT:    ret
 entry:
   %t = call {i32, i1} @llvm.sadd.with.overflow.i32(i32 %v1, i32 4)
   %val = extractvalue {i32, i1} %t, 0
@@ -78,6 +117,24 @@ define zeroext i1 @saddo3.i32(i32 %v1, i32* %res) {
 ; RV64-NEXT:    snez a0, a0
 ; RV64-NEXT:    sw a2, 0(a1)
 ; RV64-NEXT:    ret
+;
+; RV32ZBA-LABEL: saddo3.i32:
+; RV32ZBA:       # %bb.0: # %entry
+; RV32ZBA-NEXT:    addi a2, a0, -4
+; RV32ZBA-NEXT:    slt a0, a2, a0
+; RV32ZBA-NEXT:    xori a0, a0, 1
+; RV32ZBA-NEXT:    sw a2, 0(a1)
+; RV32ZBA-NEXT:    ret
+;
+; RV64ZBA-LABEL: saddo3.i32:
+; RV64ZBA:       # %bb.0: # %entry
+; RV64ZBA-NEXT:    sext.w a0, a0
+; RV64ZBA-NEXT:    addi a2, a0, -4
+; RV64ZBA-NEXT:    addiw a0, a0, -4
+; RV64ZBA-NEXT:    xor a0, a0, a2
+; RV64ZBA-NEXT:    snez a0, a0
+; RV64ZBA-NEXT:    sw a2, 0(a1)
+; RV64ZBA-NEXT:    ret
 entry:
   %t = call {i32, i1} @llvm.sadd.with.overflow.i32(i32 %v1, i32 -4)
   %val = extractvalue {i32, i1} %t, 0
@@ -108,6 +165,27 @@ define zeroext i1 @saddo4.i32(i32 %v1, i32* %res) {
 ; RV64-NEXT:    snez a0, a0
 ; RV64-NEXT:    sw a3, 0(a1)
 ; RV64-NEXT:    ret
+;
+; RV32ZBA-LABEL: saddo4.i32:
+; RV32ZBA:       # %bb.0: # %entry
+; RV32ZBA-NEXT:    lui a2, 4096
+; RV32ZBA-NEXT:    addi a2, a2, -1
+; RV32ZBA-NEXT:    add a2, a0, a2
+; RV32ZBA-NEXT:    slt a0, a2, a0
+; RV32ZBA-NEXT:    sw a2, 0(a1)
+; RV32ZBA-NEXT:    ret
+;
+; RV64ZBA-LABEL: saddo4.i32:
+; RV64ZBA:       # %bb.0: # %entry
+; RV64ZBA-NEXT:    sext.w a0, a0
+; RV64ZBA-NEXT:    lui a2, 4096
+; RV64ZBA-NEXT:    addiw a2, a2, -1
+; RV64ZBA-NEXT:    add a3, a0, a2
+; RV64ZBA-NEXT:    addw a0, a0, a2
+; RV64ZBA-NEXT:    xor a0, a0, a3
+; RV64ZBA-NEXT:    snez a0, a0
+; RV64ZBA-NEXT:    sw a3, 0(a1)
+; RV64ZBA-NEXT:    ret
 entry:
   %t = call {i32, i1} @llvm.sadd.with.overflow.i32(i32 %v1, i32 16777215)
   %val = extractvalue {i32, i1} %t, 0
@@ -140,6 +218,30 @@ define zeroext i1 @saddo1.i64(i64 %v1, i64 %v2, i64* %res) {
 ; RV64-NEXT:    xor a0, a1, a0
 ; RV64-NEXT:    sd a3, 0(a2)
 ; RV64-NEXT:    ret
+;
+; RV32ZBA-LABEL: saddo1.i64:
+; RV32ZBA:       # %bb.0: # %entry
+; RV32ZBA-NEXT:    add a5, a1, a3
+; RV32ZBA-NEXT:    add a2, a0, a2
+; RV32ZBA-NEXT:    sltu a0, a2, a0
+; RV32ZBA-NEXT:    add a5, a5, a0
+; RV32ZBA-NEXT:    xor a0, a1, a5
+; RV32ZBA-NEXT:    xor a1, a1, a3
+; RV32ZBA-NEXT:    not a1, a1
+; RV32ZBA-NEXT:    and a0, a1, a0
+; RV32ZBA-NEXT:    slti a0, a0, 0
+; RV32ZBA-NEXT:    sw a2, 0(a4)
+; RV32ZBA-NEXT:    sw a5, 4(a4)
+; RV32ZBA-NEXT:    ret
+;
+; RV64ZBA-LABEL: saddo1.i64:
+; RV64ZBA:       # %bb.0: # %entry
+; RV64ZBA-NEXT:    add a3, a0, a1
+; RV64ZBA-NEXT:    slt a0, a3, a0
+; RV64ZBA-NEXT:    slti a1, a1, 0
+; RV64ZBA-NEXT:    xor a0, a1, a0
+; RV64ZBA-NEXT:    sd a3, 0(a2)
+; RV64ZBA-NEXT:    ret
 entry:
   %t = call {i64, i1} @llvm.sadd.with.overflow.i64(i64 %v1, i64 %v2)
   %val = extractvalue {i64, i1} %t, 0
@@ -168,6 +270,26 @@ define zeroext i1 @saddo2.i64(i64 %v1, i64* %res) {
 ; RV64-NEXT:    slt a0, a2, a0
 ; RV64-NEXT:    sd a2, 0(a1)
 ; RV64-NEXT:    ret
+;
+; RV32ZBA-LABEL: saddo2.i64:
+; RV32ZBA:       # %bb.0: # %entry
+; RV32ZBA-NEXT:    addi a3, a0, 4
+; RV32ZBA-NEXT:    sltu a0, a3, a0
+; RV32ZBA-NEXT:    add a4, a1, a0
+; RV32ZBA-NEXT:    xor a0, a1, a4
+; RV32ZBA-NEXT:    not a1, a1
+; RV32ZBA-NEXT:    and a0, a1, a0
+; RV32ZBA-NEXT:    slti a0, a0, 0
+; RV32ZBA-NEXT:    sw a3, 0(a2)
+; RV32ZBA-NEXT:    sw a4, 4(a2)
+; RV32ZBA-NEXT:    ret
+;
+; RV64ZBA-LABEL: saddo2.i64:
+; RV64ZBA:       # %bb.0: # %entry
+; RV64ZBA-NEXT:    addi a2, a0, 4
+; RV64ZBA-NEXT:    slt a0, a2, a0
+; RV64ZBA-NEXT:    sd a2, 0(a1)
+; RV64ZBA-NEXT:    ret
 entry:
   %t = call {i64, i1} @llvm.sadd.with.overflow.i64(i64 %v1, i64 4)
   %val = extractvalue {i64, i1} %t, 0
@@ -197,6 +319,27 @@ define zeroext i1 @saddo3.i64(i64 %v1, i64* %res) {
 ; RV64-NEXT:    xori a0, a0, 1
 ; RV64-NEXT:    sd a2, 0(a1)
 ; RV64-NEXT:    ret
+;
+; RV32ZBA-LABEL: saddo3.i64:
+; RV32ZBA:       # %bb.0: # %entry
+; RV32ZBA-NEXT:    addi a3, a0, -4
+; RV32ZBA-NEXT:    sltu a0, a3, a0
+; RV32ZBA-NEXT:    add a0, a1, a0
+; RV32ZBA-NEXT:    addi a4, a0, -1
+; RV32ZBA-NEXT:    xor a0, a1, a4
+; RV32ZBA-NEXT:    and a0, a1, a0
+; RV32ZBA-NEXT:    slti a0, a0, 0
+; RV32ZBA-NEXT:    sw a3, 0(a2)
+; RV32ZBA-NEXT:    sw a4, 4(a2)
+; RV32ZBA-NEXT:    ret
+;
+; RV64ZBA-LABEL: saddo3.i64:
+; RV64ZBA:       # %bb.0: # %entry
+; RV64ZBA-NEXT:    addi a2, a0, -4
+; RV64ZBA-NEXT:    slt a0, a2, a0
+; RV64ZBA-NEXT:    xori a0, a0, 1
+; RV64ZBA-NEXT:    sd a2, 0(a1)
+; RV64ZBA-NEXT:    ret
 entry:
   %t = call {i64, i1} @llvm.sadd.with.overflow.i64(i64 %v1, i64 -4)
   %val = extractvalue {i64, i1} %t, 0
@@ -222,6 +365,23 @@ define zeroext i1 @uaddo.i32(i32 %v1, i32 %v2, i32* %res) {
 ; RV64-NEXT:    sw a0, 0(a2)
 ; RV64-NEXT:    mv a0, a3
 ; RV64-NEXT:    ret
+;
+; RV32ZBA-LABEL: uaddo.i32:
+; RV32ZBA:       # %bb.0: # %entry
+; RV32ZBA-NEXT:    add a1, a0, a1
+; RV32ZBA-NEXT:    sltu a0, a1, a0
+; RV32ZBA-NEXT:    sw a1, 0(a2)
+; RV32ZBA-NEXT:    ret
+;
+; RV64ZBA-LABEL: uaddo.i32:
+; RV64ZBA:       # %bb.0: # %entry
+; RV64ZBA-NEXT:    addw a3, a0, a1
+; RV64ZBA-NEXT:    sext.w a4, a0
+; RV64ZBA-NEXT:    sltu a3, a3, a4
+; RV64ZBA-NEXT:    add a0, a0, a1
+; RV64ZBA-NEXT:    sw a0, 0(a2)
+; RV64ZBA-NEXT:    mv a0, a3
+; RV64ZBA-NEXT:    ret
 entry:
   %t = call {i32, i1} @llvm.uadd.with.overflow.i32(i32 %v1, i32 %v2)
   %val = extractvalue {i32, i1} %t, 0
@@ -251,6 +411,27 @@ define zeroext i1 @uaddo.i64(i64 %v1, i64 %v2, i64* %res) {
 ; RV64-NEXT:    sltu a0, a1, a0
 ; RV64-NEXT:    sd a1, 0(a2)
 ; RV64-NEXT:    ret
+;
+; RV32ZBA-LABEL: uaddo.i64:
+; RV32ZBA:       # %bb.0: # %entry
+; RV32ZBA-NEXT:    add a3, a1, a3
+; RV32ZBA-NEXT:    add a2, a0, a2
+; RV32ZBA-NEXT:    sltu a0, a2, a0
+; RV32ZBA-NEXT:    add a3, a3, a0
+; RV32ZBA-NEXT:    beq a3, a1, .LBB8_2
+; RV32ZBA-NEXT:  # %bb.1: # %entry
+; RV32ZBA-NEXT:    sltu a0, a3, a1
+; RV32ZBA-NEXT:  .LBB8_2: # %entry
+; RV32ZBA-NEXT:    sw a2, 0(a4)
+; RV32ZBA-NEXT:    sw a3, 4(a4)
+; RV32ZBA-NEXT:    ret
+;
+; RV64ZBA-LABEL: uaddo.i64:
+; RV64ZBA:       # %bb.0: # %entry
+; RV64ZBA-NEXT:    add a1, a0, a1
+; RV64ZBA-NEXT:    sltu a0, a1, a0
+; RV64ZBA-NEXT:    sd a1, 0(a2)
+; RV64ZBA-NEXT:    ret
 entry:
   %t = call {i64, i1} @llvm.uadd.with.overflow.i64(i64 %v1, i64 %v2)
   %val = extractvalue {i64, i1} %t, 0
@@ -279,6 +460,26 @@ define zeroext i1 @ssubo1.i32(i32 %v1, i32 %v2, i32* %res) {
 ; RV64-NEXT:    snez a0, a0
 ; RV64-NEXT:    sw a3, 0(a2)
 ; RV64-NEXT:    ret
+;
+; RV32ZBA-LABEL: ssubo1.i32:
+; RV32ZBA:       # %bb.0: # %entry
+; RV32ZBA-NEXT:    sgtz a3, a1
+; RV32ZBA-NEXT:    sub a1, a0, a1
+; RV32ZBA-NEXT:    slt a0, a1, a0
+; RV32ZBA-NEXT:    xor a0, a3, a0
+; RV32ZBA-NEXT:    sw a1, 0(a2)
+; RV32ZBA-NEXT:    ret
+;
+; RV64ZBA-LABEL: ssubo1.i32:
+; RV64ZBA:       # %bb.0: # %entry
+; RV64ZBA-NEXT:    sext.w a1, a1
+; RV64ZBA-NEXT:    sext.w a0, a0
+; RV64ZBA-NEXT:    sub a3, a0, a1
+; RV64ZBA-NEXT:    subw a0, a0, a1
+; RV64ZBA-NEXT:    xor a0, a0, a3
+; RV64ZBA-NEXT:    snez a0, a0
+; RV64ZBA-NEXT:    sw a3, 0(a2)
+; RV64ZBA-NEXT:    ret
 entry:
   %t = call {i32, i1} @llvm.ssub.with.overflow.i32(i32 %v1, i32 %v2)
   %val = extractvalue {i32, i1} %t, 0
@@ -304,6 +505,23 @@ define zeroext i1 @ssubo2.i32(i32 %v1, i32* %res) {
 ; RV64-NEXT:    snez a0, a0
 ; RV64-NEXT:    sw a2, 0(a1)
 ; RV64-NEXT:    ret
+;
+; RV32ZBA-LABEL: ssubo2.i32:
+; RV32ZBA:       # %bb.0: # %entry
+; RV32ZBA-NEXT:    addi a2, a0, 4
+; RV32ZBA-NEXT:    slt a0, a2, a0
+; RV32ZBA-NEXT:    sw a2, 0(a1)
+; RV32ZBA-NEXT:    ret
+;
+; RV64ZBA-LABEL: ssubo2.i32:
+; RV64ZBA:       # %bb.0: # %entry
+; RV64ZBA-NEXT:    sext.w a0, a0
+; RV64ZBA-NEXT:    addi a2, a0, 4
+; RV64ZBA-NEXT:    addiw a0, a0, 4
+; RV64ZBA-NEXT:    xor a0, a0, a2
+; RV64ZBA-NEXT:    snez a0, a0
+; RV64ZBA-NEXT:    sw a2, 0(a1)
+; RV64ZBA-NEXT:    ret
 entry:
   %t = call {i32, i1} @llvm.ssub.with.overflow.i32(i32 %v1, i32 -4)
   %val = extractvalue {i32, i1} %t, 0
@@ -336,6 +554,30 @@ define zeroext i1 @ssubo.i64(i64 %v1, i64 %v2, i64* %res) {
 ; RV64-NEXT:    xor a0, a3, a0
 ; RV64-NEXT:    sd a1, 0(a2)
 ; RV64-NEXT:    ret
+;
+; RV32ZBA-LABEL: ssubo.i64:
+; RV32ZBA:       # %bb.0: # %entry
+; RV32ZBA-NEXT:    sltu a6, a0, a2
+; RV32ZBA-NEXT:    sub a5, a1, a3
+; RV32ZBA-NEXT:    sub a5, a5, a6
+; RV32ZBA-NEXT:    xor a6, a1, a5
+; RV32ZBA-NEXT:    xor a1, a1, a3
+; RV32ZBA-NEXT:    and a1, a1, a6
+; RV32ZBA-NEXT:    slti a1, a1, 0
+; RV32ZBA-NEXT:    sub a0, a0, a2
+; RV32ZBA-NEXT:    sw a0, 0(a4)
+; RV32ZBA-NEXT:    sw a5, 4(a4)
+; RV32ZBA-NEXT:    mv a0, a1
+; RV32ZBA-NEXT:    ret
+;
+; RV64ZBA-LABEL: ssubo.i64:
+; RV64ZBA:       # %bb.0: # %entry
+; RV64ZBA-NEXT:    sgtz a3, a1
+; RV64ZBA-NEXT:    sub a1, a0, a1
+; RV64ZBA-NEXT:    slt a0, a1, a0
+; RV64ZBA-NEXT:    xor a0, a3, a0
+; RV64ZBA-NEXT:    sd a1, 0(a2)
+; RV64ZBA-NEXT:    ret
 entry:
   %t = call {i64, i1} @llvm.ssub.with.overflow.i64(i64 %v1, i64 %v2)
   %val = extractvalue {i64, i1} %t, 0
@@ -361,6 +603,23 @@ define zeroext i1 @usubo.i32(i32 %v1, i32 %v2, i32* %res) {
 ; RV64-NEXT:    sw a0, 0(a2)
 ; RV64-NEXT:    mv a0, a3
 ; RV64-NEXT:    ret
+;
+; RV32ZBA-LABEL: usubo.i32:
+; RV32ZBA:       # %bb.0: # %entry
+; RV32ZBA-NEXT:    sub a1, a0, a1
+; RV32ZBA-NEXT:    sltu a0, a0, a1
+; RV32ZBA-NEXT:    sw a1, 0(a2)
+; RV32ZBA-NEXT:    ret
+;
+; RV64ZBA-LABEL: usubo.i32:
+; RV64ZBA:       # %bb.0: # %entry
+; RV64ZBA-NEXT:    subw a3, a0, a1
+; RV64ZBA-NEXT:    sext.w a4, a0
+; RV64ZBA-NEXT:    sltu a3, a4, a3
+; RV64ZBA-NEXT:    sub a0, a0, a1
+; RV64ZBA-NEXT:    sw a0, 0(a2)
+; RV64ZBA-NEXT:    mv a0, a3
+; RV64ZBA-NEXT:    ret
 entry:
   %t = call {i32, i1} @llvm.usub.with.overflow.i32(i32 %v1, i32 %v2)
   %val = extractvalue {i32, i1} %t, 0
@@ -393,6 +652,30 @@ define zeroext i1 @usubo.i64(i64 %v1, i64 %v2, i64* %res) {
 ; RV64-NEXT:    sltu a0, a0, a1
 ; RV64-NEXT:    sd a1, 0(a2)
 ; RV64-NEXT:    ret
+;
+; RV32ZBA-LABEL: usubo.i64:
+; RV32ZBA:       # %bb.0: # %entry
+; RV32ZBA-NEXT:    sltu a5, a0, a2
+; RV32ZBA-NEXT:    sub a3, a1, a3
+; RV32ZBA-NEXT:    sub a3, a3, a5
+; RV32ZBA-NEXT:    sub a2, a0, a2
+; RV32ZBA-NEXT:    beq a3, a1, .LBB13_2
+; RV32ZBA-NEXT:  # %bb.1: # %entry
+; RV32ZBA-NEXT:    sltu a0, a1, a3
+; RV32ZBA-NEXT:    j .LBB13_3
+; RV32ZBA-NEXT:  .LBB13_2:
+; RV32ZBA-NEXT:    sltu a0, a0, a2
+; RV32ZBA-NEXT:  .LBB13_3: # %entry
+; RV32ZBA-NEXT:    sw a2, 0(a4)
+; RV32ZBA-NEXT:    sw a3, 4(a4)
+; RV32ZBA-NEXT:    ret
+;
+; RV64ZBA-LABEL: usubo.i64:
+; RV64ZBA:       # %bb.0: # %entry
+; RV64ZBA-NEXT:    sub a1, a0, a1
+; RV64ZBA-NEXT:    sltu a0, a0, a1
+; RV64ZBA-NEXT:    sd a1, 0(a2)
+; RV64ZBA-NEXT:    ret
 entry:
   %t = call {i64, i1} @llvm.usub.with.overflow.i64(i64 %v1, i64 %v2)
   %val = extractvalue {i64, i1} %t, 0
@@ -422,6 +705,27 @@ define zeroext i1 @smulo.i32(i32 %v1, i32 %v2, i32* %res) {
 ; RV64-NEXT:    snez a0, a0
 ; RV64-NEXT:    sw a3, 0(a2)
 ; RV64-NEXT:    ret
+;
+; RV32ZBA-LABEL: smulo.i32:
+; RV32ZBA:       # %bb.0: # %entry
+; RV32ZBA-NEXT:    mulh a3, a0, a1
+; RV32ZBA-NEXT:    mul a1, a0, a1
+; RV32ZBA-NEXT:    srai a0, a1, 31
+; RV32ZBA-NEXT:    xor a0, a3, a0
+; RV32ZBA-NEXT:    snez a0, a0
+; RV32ZBA-NEXT:    sw a1, 0(a2)
+; RV32ZBA-NEXT:    ret
+;
+; RV64ZBA-LABEL: smulo.i32:
+; RV64ZBA:       # %bb.0: # %entry
+; RV64ZBA-NEXT:    sext.w a1, a1
+; RV64ZBA-NEXT:    sext.w a0, a0
+; RV64ZBA-NEXT:    mul a3, a0, a1
+; RV64ZBA-NEXT:    mulw a0, a0, a1
+; RV64ZBA-NEXT:    xor a0, a0, a3
+; RV64ZBA-NEXT:    snez a0, a0
+; RV64ZBA-NEXT:    sw a3, 0(a2)
+; RV64ZBA-NEXT:    ret
 entry:
   %t = call {i32, i1} @llvm.smul.with.overflow.i32(i32 %v1, i32 %v2)
   %val = extractvalue {i32, i1} %t, 0
@@ -452,6 +756,28 @@ define zeroext i1 @smulo2.i32(i32 %v1, i32* %res) {
 ; RV64-NEXT:    snez a0, a0
 ; RV64-NEXT:    sw a3, 0(a1)
 ; RV64-NEXT:    ret
+;
+; RV32ZBA-LABEL: smulo2.i32:
+; RV32ZBA:       # %bb.0: # %entry
+; RV32ZBA-NEXT:    addi a2, zero, 13
+; RV32ZBA-NEXT:    mulh a3, a0, a2
+; RV32ZBA-NEXT:    mul a2, a0, a2
+; RV32ZBA-NEXT:    srai a0, a2, 31
+; RV32ZBA-NEXT:    xor a0, a3, a0
+; RV32ZBA-NEXT:    snez a0, a0
+; RV32ZBA-NEXT:    sw a2, 0(a1)
+; RV32ZBA-NEXT:    ret
+;
+; RV64ZBA-LABEL: smulo2.i32:
+; RV64ZBA:       # %bb.0: # %entry
+; RV64ZBA-NEXT:    sext.w a0, a0
+; RV64ZBA-NEXT:    addi a2, zero, 13
+; RV64ZBA-NEXT:    mul a3, a0, a2
+; RV64ZBA-NEXT:    mulw a0, a0, a2
+; RV64ZBA-NEXT:    xor a0, a0, a3
+; RV64ZBA-NEXT:    snez a0, a0
+; RV64ZBA-NEXT:    sw a3, 0(a1)
+; RV64ZBA-NEXT:    ret
 entry:
   %t = call {i32, i1} @llvm.smul.with.overflow.i32(i32 %v1, i32 13)
   %val = extractvalue {i32, i1} %t, 0
@@ -492,6 +818,38 @@ define zeroext i1 @smulo.i64(i64 %v1, i64 %v2, i64* %res) {
 ; RV64-NEXT:    snez a0, a0
 ; RV64-NEXT:    sd a1, 0(a2)
 ; RV64-NEXT:    ret
+;
+; RV32ZBA-LABEL: smulo.i64:
+; RV32ZBA:       # %bb.0: # %entry
+; RV32ZBA-NEXT:    addi sp, sp, -16
+; RV32ZBA-NEXT:    .cfi_def_cfa_offset 16
+; RV32ZBA-NEXT:    sw ra, 12(sp) # 4-byte Folded Spill
+; RV32ZBA-NEXT:    sw s0, 8(sp) # 4-byte Folded Spill
+; RV32ZBA-NEXT:    .cfi_offset ra, -4
+; RV32ZBA-NEXT:    .cfi_offset s0, -8
+; RV32ZBA-NEXT:    mv s0, a4
+; RV32ZBA-NEXT:    sw zero, 4(sp)
+; RV32ZBA-NEXT:    addi a4, sp, 4
+; RV32ZBA-NEXT:    call __mulodi4@plt
+; RV32ZBA-NEXT:    lw a2, 4(sp)
+; RV32ZBA-NEXT:    snez a2, a2
+; RV32ZBA-NEXT:    sw a1, 4(s0)
+; RV32ZBA-NEXT:    sw a0, 0(s0)
+; RV32ZBA-NEXT:    mv a0, a2
+; RV32ZBA-NEXT:    lw s0, 8(sp) # 4-byte Folded Reload
+; RV32ZBA-NEXT:    lw ra, 12(sp) # 4-byte Folded Reload
+; RV32ZBA-NEXT:    addi sp, sp, 16
+; RV32ZBA-NEXT:    ret
+;
+; RV64ZBA-LABEL: smulo.i64:
+; RV64ZBA:       # %bb.0: # %entry
+; RV64ZBA-NEXT:    mulh a3, a0, a1
+; RV64ZBA-NEXT:    mul a1, a0, a1
+; RV64ZBA-NEXT:    srai a0, a1, 63
+; RV64ZBA-NEXT:    xor a0, a3, a0
+; RV64ZBA-NEXT:    snez a0, a0
+; RV64ZBA-NEXT:    sd a1, 0(a2)
+; RV64ZBA-NEXT:    ret
 entry:
   %t = call {i64, i1} @llvm.smul.with.overflow.i64(i64 %v1, i64 %v2)
   %val = extractvalue {i64, i1} %t, 0
@@ -535,6 +893,41 @@ define zeroext i1 @smulo2.i64(i64 %v1, i64* %res) {
 ; RV64-NEXT:    snez a0, a0
 ; RV64-NEXT:    sd a2, 0(a1)
 ; RV64-NEXT:    ret
+;
+; RV32ZBA-LABEL: smulo2.i64:
+; RV32ZBA:       # %bb.0: # %entry
+; RV32ZBA-NEXT:    addi sp, sp, -16
+; RV32ZBA-NEXT:    .cfi_def_cfa_offset 16
+; RV32ZBA-NEXT:    sw ra, 12(sp) # 4-byte Folded Spill
+; RV32ZBA-NEXT:    sw s0, 8(sp) # 4-byte Folded Spill
+; RV32ZBA-NEXT:    .cfi_offset ra, -4
+; RV32ZBA-NEXT:    .cfi_offset s0, -8
+; RV32ZBA-NEXT:    mv s0, a2
+; RV32ZBA-NEXT:    sw zero, 4(sp)
+; RV32ZBA-NEXT:    addi a2, zero, 13
+; RV32ZBA-NEXT:    addi a4, sp, 4
+; RV32ZBA-NEXT:    mv a3, zero
+; RV32ZBA-NEXT:    call __mulodi4@plt
+; RV32ZBA-NEXT:    lw a2, 4(sp)
+; RV32ZBA-NEXT:    snez a2, a2
+; RV32ZBA-NEXT:    sw a1, 4(s0)
+; RV32ZBA-NEXT:    sw a0, 0(s0)
+; RV32ZBA-NEXT:    mv a0, a2
+; RV32ZBA-NEXT:    lw s0, 8(sp) # 4-byte Folded Reload
+; RV32ZBA-NEXT:    lw ra, 12(sp) # 4-byte Folded Reload
+; RV32ZBA-NEXT:    addi sp, sp, 16
+; RV32ZBA-NEXT:    ret
+;
+; RV64ZBA-LABEL: smulo2.i64:
+; RV64ZBA:       # %bb.0: # %entry
+; RV64ZBA-NEXT:    addi a2, zero, 13
+; RV64ZBA-NEXT:    mulh a3, a0, a2
+; RV64ZBA-NEXT:    mul a2, a0, a2
+; RV64ZBA-NEXT:    srai a0, a2, 63
+; RV64ZBA-NEXT:    xor a0, a3, a0
+; RV64ZBA-NEXT:    snez a0, a0
+; RV64ZBA-NEXT:    sd a2, 0(a1)
+; RV64ZBA-NEXT:    ret
 entry:
   %t = call {i64, i1} @llvm.smul.with.overflow.i64(i64 %v1, i64 13)
   %val = extractvalue {i64, i1} %t, 0
@@ -562,6 +955,25 @@ define zeroext i1 @umulo.i32(i32 %v1, i32 %v2, i32* %res) {
 ; RV64-NEXT:    snez a0, a0
 ; RV64-NEXT:    sw a1, 0(a2)
 ; RV64-NEXT:    ret
+;
+; RV32ZBA-LABEL: umulo.i32:
+; RV32ZBA:       # %bb.0: # %entry
+; RV32ZBA-NEXT:    mulhu a3, a0, a1
+; RV32ZBA-NEXT:    snez a3, a3
+; RV32ZBA-NEXT:    mul a0, a0, a1
+; RV32ZBA-NEXT:    sw a0, 0(a2)
+; RV32ZBA-NEXT:    mv a0, a3
+; RV32ZBA-NEXT:    ret
+;
+; RV64ZBA-LABEL: umulo.i32:
+; RV64ZBA:       # %bb.0: # %entry
+; RV64ZBA-NEXT:    slli a1, a1, 32
+; RV64ZBA-NEXT:    slli a0, a0, 32
+; RV64ZBA-NEXT:    mulhu a1, a0, a1
+; RV64ZBA-NEXT:    srli a0, a1, 32
+; RV64ZBA-NEXT:    snez a0, a0
+; RV64ZBA-NEXT:    sw a1, 0(a2)
+; RV64ZBA-NEXT:    ret
 entry:
   %t = call {i32, i1} @llvm.umul.with.overflow.i32(i32 %v1, i32 %v2)
   %val = extractvalue {i32, i1} %t, 0
@@ -591,6 +1003,26 @@ define zeroext i1 @umulo2.i32(i32 %v1, i32* %res) {
 ; RV64-NEXT:    snez a0, a0
 ; RV64-NEXT:    sw a2, 0(a1)
 ; RV64-NEXT:    ret
+;
+; RV32ZBA-LABEL: umulo2.i32:
+; RV32ZBA:       # %bb.0: # %entry
+; RV32ZBA-NEXT:    addi a3, zero, 13
+; RV32ZBA-NEXT:    mulhu a2, a0, a3
+; RV32ZBA-NEXT:    snez a2, a2
+; RV32ZBA-NEXT:    mul a0, a0, a3
+; RV32ZBA-NEXT:    sw a0, 0(a1)
+; RV32ZBA-NEXT:    mv a0, a2
+; RV32ZBA-NEXT:    ret
+;
+; RV64ZBA-LABEL: umulo2.i32:
+; RV64ZBA:       # %bb.0: # %entry
+; RV64ZBA-NEXT:    zext.w a0, a0
+; RV64ZBA-NEXT:    addi a2, zero, 13
+; RV64ZBA-NEXT:    mul a2, a0, a2
+; RV64ZBA-NEXT:    srli a0, a2, 32
+; RV64ZBA-NEXT:    snez a0, a0
+; RV64ZBA-NEXT:    sw a2, 0(a1)
+; RV64ZBA-NEXT:    ret
 entry:
   %t = call {i32, i1} @llvm.umul.with.overflow.i32(i32 %v1, i32 13)
   %val = extractvalue {i32, i1} %t, 0
@@ -632,6 +1064,39 @@ define zeroext i1 @umulo.i64(i64 %v1, i64 %v2, i64* %res) {
 ; RV64-NEXT:    sd a0, 0(a2)
 ; RV64-NEXT:    mv a0, a3
 ; RV64-NEXT:    ret
+;
+; RV32ZBA-LABEL: umulo.i64:
+; RV32ZBA:       # %bb.0: # %entry
+; RV32ZBA-NEXT:    mul a6, a3, a0
+; RV32ZBA-NEXT:    mul a5, a1, a2
+; RV32ZBA-NEXT:    add a6, a5, a6
+; RV32ZBA-NEXT:    mulhu a5, a0, a2
+; RV32ZBA-NEXT:    add a6, a5, a6
+; RV32ZBA-NEXT:    sltu a7, a6, a5
+; RV32ZBA-NEXT:    snez t0, a3
+; RV32ZBA-NEXT:    snez a5, a1
+; RV32ZBA-NEXT:    and a5, a5, t0
+; RV32ZBA-NEXT:    mulhu a1, a1, a2
+; RV32ZBA-NEXT:    snez a1, a1
+; RV32ZBA-NEXT:    or a1, a5, a1
+; RV32ZBA-NEXT:    mulhu a3, a3, a0
+; RV32ZBA-NEXT:    snez a3, a3
+; RV32ZBA-NEXT:    or a1, a1, a3
+; RV32ZBA-NEXT:    or a1, a1, a7
+; RV32ZBA-NEXT:    mul a0, a0, a2
+; RV32ZBA-NEXT:    sw a0, 0(a4)
+; RV32ZBA-NEXT:    sw a6, 4(a4)
+; RV32ZBA-NEXT:    mv a0, a1
+; RV32ZBA-NEXT:    ret
+;
+; RV64ZBA-LABEL: umulo.i64:
+; RV64ZBA:       # %bb.0: # %entry
+; RV64ZBA-NEXT:    mulhu a3, a0, a1
+; RV64ZBA-NEXT:    snez a3, a3
+; RV64ZBA-NEXT:    mul a0, a0, a1
+; RV64ZBA-NEXT:    sd a0, 0(a2)
+; RV64ZBA-NEXT:    mv a0, a3
+; RV64ZBA-NEXT:    ret
 entry:
   %t = call {i64, i1} @llvm.umul.with.overflow.i64(i64 %v1, i64 %v2)
   %val = extractvalue {i64, i1} %t, 0
@@ -666,6 +1131,32 @@ define zeroext i1 @umulo2.i64(i64 %v1, i64* %res) {
 ; RV64-NEXT:    sd a0, 0(a1)
 ; RV64-NEXT:    mv a0, a2
 ; RV64-NEXT:    ret
+;
+; RV32ZBA-LABEL: umulo2.i64:
+; RV32ZBA:       # %bb.0: # %entry
+; RV32ZBA-NEXT:    addi a3, zero, 13
+; RV32ZBA-NEXT:    mul a4, a1, a3
+; RV32ZBA-NEXT:    mulhu a5, a0, a3
+; RV32ZBA-NEXT:    add a4, a5, a4
+; RV32ZBA-NEXT:    sltu a5, a4, a5
+; RV32ZBA-NEXT:    mulhu a1, a1, a3
+; RV32ZBA-NEXT:    snez a1, a1
+; RV32ZBA-NEXT:    or a1, a1, a5
+; RV32ZBA-NEXT:    mul a0, a0, a3
+; RV32ZBA-NEXT:    sw a0, 0(a2)
+; RV32ZBA-NEXT:    sw a4, 4(a2)
+; RV32ZBA-NEXT:    mv a0, a1
+; RV32ZBA-NEXT:    ret
+;
+; RV64ZBA-LABEL: umulo2.i64:
+; RV64ZBA:       # %bb.0: # %entry
+; RV64ZBA-NEXT:    addi a3, zero, 13
+; RV64ZBA-NEXT:    mulhu a2, a0, a3
+; RV64ZBA-NEXT:    snez a2, a2
+; RV64ZBA-NEXT:    mul a0, a0, a3
+; RV64ZBA-NEXT:    sd a0, 0(a1)
+; RV64ZBA-NEXT:    mv a0, a2
+; RV64ZBA-NEXT:    ret
 entry:
   %t = call {i64, i1} @llvm.umul.with.overflow.i64(i64 %v1, i64 13)
   %val = extractvalue {i64, i1} %t, 0
@@ -701,6 +1192,29 @@ define i32 @saddo.select.i32(i32 %v1, i32 %v2) {
 ; RV64-NEXT:    mv a0, a1
 ; RV64-NEXT:  .LBB22_2: # %entry
 ; RV64-NEXT:    ret
+;
+; RV32ZBA-LABEL: saddo.select.i32:
+; RV32ZBA:       # %bb.0: # %entry
+; RV32ZBA-NEXT:    add a2, a0, a1
+; RV32ZBA-NEXT:    slt a2, a2, a0
+; RV32ZBA-NEXT:    slti a3, a1, 0
+; RV32ZBA-NEXT:    bne a3, a2, .LBB22_2
+; RV32ZBA-NEXT:  # %bb.1: # %entry
+; RV32ZBA-NEXT:    mv a0, a1
+; RV32ZBA-NEXT:  .LBB22_2: # %entry
+; RV32ZBA-NEXT:    ret
+;
+; RV64ZBA-LABEL: saddo.select.i32:
+; RV64ZBA:       # %bb.0: # %entry
+; RV64ZBA-NEXT:    sext.w a2, a1
+; RV64ZBA-NEXT:    sext.w a3, a0
+; RV64ZBA-NEXT:    add a4, a3, a2
+; RV64ZBA-NEXT:    addw a2, a3, a2
+; RV64ZBA-NEXT:    bne a2, a4, .LBB22_2
+; RV64ZBA-NEXT:  # %bb.1: # %entry
+; RV64ZBA-NEXT:    mv a0, a1
+; RV64ZBA-NEXT:  .LBB22_2: # %entry
+; RV64ZBA-NEXT:    ret
 entry:
   %t = call {i32, i1} @llvm.sadd.with.overflow.i32(i32 %v1, i32 %v2)
   %obit = extractvalue {i32, i1} %t, 1
@@ -727,6 +1241,25 @@ define i1 @saddo.not.i32(i32 %v1, i32 %v2) {
 ; RV64-NEXT:    xor a0, a0, a2
 ; RV64-NEXT:    seqz a0, a0
 ; RV64-NEXT:    ret
+;
+; RV32ZBA-LABEL: saddo.not.i32:
+; RV32ZBA:       # %bb.0: # %entry
+; RV32ZBA-NEXT:    add a2, a0, a1
+; RV32ZBA-NEXT:    slt a0, a2, a0
+; RV32ZBA-NEXT:    slti a1, a1, 0
+; RV32ZBA-NEXT:    xor a0, a1, a0
+; RV32ZBA-NEXT:    xori a0, a0, 1
+; RV32ZBA-NEXT:    ret
+;
+; RV64ZBA-LABEL: saddo.not.i32:
+; RV64ZBA:       # %bb.0: # %entry
+; RV64ZBA-NEXT:    sext.w a1, a1
+; RV64ZBA-NEXT:    sext.w a0, a0
+; RV64ZBA-NEXT:    add a2, a0, a1
+; RV64ZBA-NEXT:    addw a0, a0, a1
+; RV64ZBA-NEXT:    xor a0, a0, a2
+; RV64ZBA-NEXT:    seqz a0, a0
+; RV64ZBA-NEXT:    ret
 entry:
   %t = call {i32, i1} @llvm.sadd.with.overflow.i32(i32 %v1, i32 %v2)
   %obit = extractvalue {i32, i1} %t, 1
@@ -762,6 +1295,34 @@ define i64 @saddo.select.i64(i64 %v1, i64 %v2) {
 ; RV64-NEXT:    mv a0, a1
 ; RV64-NEXT:  .LBB24_2: # %entry
 ; RV64-NEXT:    ret
+;
+; RV32ZBA-LABEL: saddo.select.i64:
+; RV32ZBA:       # %bb.0: # %entry
+; RV32ZBA-NEXT:    add a4, a1, a3
+; RV32ZBA-NEXT:    add a5, a0, a2
+; RV32ZBA-NEXT:    sltu a5, a5, a0
+; RV32ZBA-NEXT:    add a4, a4, a5
+; RV32ZBA-NEXT:    xor a4, a1, a4
+; RV32ZBA-NEXT:    xor a5, a1, a3
+; RV32ZBA-NEXT:    not a5, a5
+; RV32ZBA-NEXT:    and a4, a5, a4
+; RV32ZBA-NEXT:    bltz a4, .LBB24_2
+; RV32ZBA-NEXT:  # %bb.1: # %entry
+; RV32ZBA-NEXT:    mv a0, a2
+; RV32ZBA-NEXT:    mv a1, a3
+; RV32ZBA-NEXT:  .LBB24_2: # %entry
+; RV32ZBA-NEXT:    ret
+;
+; RV64ZBA-LABEL: saddo.select.i64:
+; RV64ZBA:       # %bb.0: # %entry
+; RV64ZBA-NEXT:    add a2, a0, a1
+; RV64ZBA-NEXT:    slt a2, a2, a0
+; RV64ZBA-NEXT:    slti a3, a1, 0
+; RV64ZBA-NEXT:    bne a3, a2, .LBB24_2
+; RV64ZBA-NEXT:  # %bb.1: # %entry
+; RV64ZBA-NEXT:    mv a0, a1
+; RV64ZBA-NEXT:  .LBB24_2: # %entry
+; RV64ZBA-NEXT:    ret
 entry:
   %t = call {i64, i1} @llvm.sadd.with.overflow.i64(i64 %v1, i64 %v2)
   %obit = extractvalue {i64, i1} %t, 1
@@ -792,6 +1353,29 @@ define i1 @saddo.not.i64(i64 %v1, i64 %v2) {
 ; RV64-NEXT:    xor a0, a1, a0
 ; RV64-NEXT:    xori a0, a0, 1
 ; RV64-NEXT:    ret
+;
+; RV32ZBA-LABEL: saddo.not.i64:
+; RV32ZBA:       # %bb.0: # %entry
+; RV32ZBA-NEXT:    add a4, a1, a3
+; RV32ZBA-NEXT:    add a2, a0, a2
+; RV32ZBA-NEXT:    sltu a0, a2, a0
+; RV32ZBA-NEXT:    add a0, a4, a0
+; RV32ZBA-NEXT:    xor a0, a1, a0
+; RV32ZBA-NEXT:    xor a1, a1, a3
+; RV32ZBA-NEXT:    not a1, a1
+; RV32ZBA-NEXT:    and a0, a1, a0
+; RV32ZBA-NEXT:    addi a1, zero, -1
+; RV32ZBA-NEXT:    slt a0, a1, a0
+; RV32ZBA-NEXT:    ret
+;
+; RV64ZBA-LABEL: saddo.not.i64:
+; RV64ZBA:       # %bb.0: # %entry
+; RV64ZBA-NEXT:    add a2, a0, a1
+; RV64ZBA-NEXT:    slt a0, a2, a0
+; RV64ZBA-NEXT:    slti a1, a1, 0
+; RV64ZBA-NEXT:    xor a0, a1, a0
+; RV64ZBA-NEXT:    xori a0, a0, 1
+; RV64ZBA-NEXT:    ret
 entry:
   %t = call {i64, i1} @llvm.sadd.with.overflow.i64(i64 %v1, i64 %v2)
   %obit = extractvalue {i64, i1} %t, 1
@@ -818,6 +1402,25 @@ define i32 @uaddo.select.i32(i32 %v1, i32 %v2) {
 ; RV64-NEXT:    mv a0, a1
 ; RV64-NEXT:  .LBB26_2: # %entry
 ; RV64-NEXT:    ret
+;
+; RV32ZBA-LABEL: uaddo.select.i32:
+; RV32ZBA:       # %bb.0: # %entry
+; RV32ZBA-NEXT:    add a2, a0, a1
+; RV32ZBA-NEXT:    bltu a2, a0, .LBB26_2
+; RV32ZBA-NEXT:  # %bb.1: # %entry
+; RV32ZBA-NEXT:    mv a0, a1
+; RV32ZBA-NEXT:  .LBB26_2: # %entry
+; RV32ZBA-NEXT:    ret
+;
+; RV64ZBA-LABEL: uaddo.select.i32:
+; RV64ZBA:       # %bb.0: # %entry
+; RV64ZBA-NEXT:    addw a2, a0, a1
+; RV64ZBA-NEXT:    sext.w a3, a0
+; RV64ZBA-NEXT:    bltu a2, a3, .LBB26_2
+; RV64ZBA-NEXT:  # %bb.1: # %entry
+; RV64ZBA-NEXT:    mv a0, a1
+; RV64ZBA-NEXT:  .LBB26_2: # %entry
+; RV64ZBA-NEXT:    ret
 entry:
   %t = call {i32, i1} @llvm.uadd.with.overflow.i32(i32 %v1, i32 %v2)
   %obit = extractvalue {i32, i1} %t, 1
@@ -840,6 +1443,21 @@ define i1 @uaddo.not.i32(i32 %v1, i32 %v2) {
 ; RV64-NEXT:    sltu a0, a1, a0
 ; RV64-NEXT:    xori a0, a0, 1
 ; RV64-NEXT:    ret
+;
+; RV32ZBA-LABEL: uaddo.not.i32:
+; RV32ZBA:       # %bb.0: # %entry
+; RV32ZBA-NEXT:    add a1, a0, a1
+; RV32ZBA-NEXT:    sltu a0, a1, a0
+; RV32ZBA-NEXT:    xori a0, a0, 1
+; RV32ZBA-NEXT:    ret
+;
+; RV64ZBA-LABEL: uaddo.not.i32:
+; RV64ZBA:       # %bb.0: # %entry
+; RV64ZBA-NEXT:    addw a1, a0, a1
+; RV64ZBA-NEXT:    sext.w a0, a0
+; RV64ZBA-NEXT:    sltu a0, a1, a0
+; RV64ZBA-NEXT:    xori a0, a0, 1
+; RV64ZBA-NEXT:    ret
 entry:
   %t = call {i32, i1} @llvm.uadd.with.overflow.i32(i32 %v1, i32 %v2)
   %obit = extractvalue {i32, i1} %t, 1
@@ -875,6 +1493,34 @@ define i64 @uaddo.select.i64(i64 %v1, i64 %v2) {
 ; RV64-NEXT:    mv a0, a1
 ; RV64-NEXT:  .LBB28_2: # %entry
 ; RV64-NEXT:    ret
+;
+; RV32ZBA-LABEL: uaddo.select.i64:
+; RV32ZBA:       # %bb.0: # %entry
+; RV32ZBA-NEXT:    add a5, a1, a3
+; RV32ZBA-NEXT:    add a4, a0, a2
+; RV32ZBA-NEXT:    sltu a4, a4, a0
+; RV32ZBA-NEXT:    add a5, a5, a4
+; RV32ZBA-NEXT:    bne a5, a1, .LBB28_3
+; RV32ZBA-NEXT:  # %bb.1: # %entry
+; RV32ZBA-NEXT:    beqz a4, .LBB28_4
+; RV32ZBA-NEXT:  .LBB28_2: # %entry
+; RV32ZBA-NEXT:    ret
+; RV32ZBA-NEXT:  .LBB28_3: # %entry
+; RV32ZBA-NEXT:    sltu a4, a5, a1
+; RV32ZBA-NEXT:    bnez a4, .LBB28_2
+; RV32ZBA-NEXT:  .LBB28_4: # %entry
+; RV32ZBA-NEXT:    mv a0, a2
+; RV32ZBA-NEXT:    mv a1, a3
+; RV32ZBA-NEXT:    ret
+;
+; RV64ZBA-LABEL: uaddo.select.i64:
+; RV64ZBA:       # %bb.0: # %entry
+; RV64ZBA-NEXT:    add a2, a0, a1
+; RV64ZBA-NEXT:    bltu a2, a0, .LBB28_2
+; RV64ZBA-NEXT:  # %bb.1: # %entry
+; RV64ZBA-NEXT:    mv a0, a1
+; RV64ZBA-NEXT:  .LBB28_2: # %entry
+; RV64ZBA-NEXT:    ret
 entry:
   %t = call {i64, i1} @llvm.uadd.with.overflow.i64(i64 %v1, i64 %v2)
   %obit = extractvalue {i64, i1} %t, 1
@@ -902,6 +1548,26 @@ define i1 @uaddo.not.i64(i64 %v1, i64 %v2) {
 ; RV64-NEXT:    sltu a0, a1, a0
 ; RV64-NEXT:    xori a0, a0, 1
 ; RV64-NEXT:    ret
+;
+; RV32ZBA-LABEL: uaddo.not.i64:
+; RV32ZBA:       # %bb.0: # %entry
+; RV32ZBA-NEXT:    add a3, a1, a3
+; RV32ZBA-NEXT:    add a2, a0, a2
+; RV32ZBA-NEXT:    sltu a0, a2, a0
+; RV32ZBA-NEXT:    add a2, a3, a0
+; RV32ZBA-NEXT:    beq a2, a1, .LBB29_2
+; RV32ZBA-NEXT:  # %bb.1: # %entry
+; RV32ZBA-NEXT:    sltu a0, a2, a1
+; RV32ZBA-NEXT:  .LBB29_2: # %entry
+; RV32ZBA-NEXT:    xori a0, a0, 1
+; RV32ZBA-NEXT:    ret
+;
+; RV64ZBA-LABEL: uaddo.not.i64:
+; RV64ZBA:       # %bb.0: # %entry
+; RV64ZBA-NEXT:    add a1, a0, a1
+; RV64ZBA-NEXT:    sltu a0, a1, a0
+; RV64ZBA-NEXT:    xori a0, a0, 1
+; RV64ZBA-NEXT:    ret
 entry:
   %t = call {i64, i1} @llvm.uadd.with.overflow.i64(i64 %v1, i64 %v2)
   %obit = extractvalue {i64, i1} %t, 1
@@ -932,6 +1598,29 @@ define i32 @ssubo.select.i32(i32 %v1, i32 %v2) {
 ; RV64-NEXT:    mv a0, a1
 ; RV64-NEXT:  .LBB30_2: # %entry
 ; RV64-NEXT:    ret
+;
+; RV32ZBA-LABEL: ssubo.select.i32:
+; RV32ZBA:       # %bb.0: # %entry
+; RV32ZBA-NEXT:    sgtz a2, a1
+; RV32ZBA-NEXT:    sub a3, a0, a1
+; RV32ZBA-NEXT:    slt a3, a3, a0
+; RV32ZBA-NEXT:    bne a2, a3, .LBB30_2
+; RV32ZBA-NEXT:  # %bb.1: # %entry
+; RV32ZBA-NEXT:    mv a0, a1
+; RV32ZBA-NEXT:  .LBB30_2: # %entry
+; RV32ZBA-NEXT:    ret
+;
+; RV64ZBA-LABEL: ssubo.select.i32:
+; RV64ZBA:       # %bb.0: # %entry
+; RV64ZBA-NEXT:    sext.w a2, a1
+; RV64ZBA-NEXT:    sext.w a3, a0
+; RV64ZBA-NEXT:    sub a4, a3, a2
+; RV64ZBA-NEXT:    subw a2, a3, a2
+; RV64ZBA-NEXT:    bne a2, a4, .LBB30_2
+; RV64ZBA-NEXT:  # %bb.1: # %entry
+; RV64ZBA-NEXT:    mv a0, a1
+; RV64ZBA-NEXT:  .LBB30_2: # %entry
+; RV64ZBA-NEXT:    ret
 entry:
   %t = call {i32, i1} @llvm.ssub.with.overflow.i32(i32 %v1, i32 %v2)
   %obit = extractvalue {i32, i1} %t, 1
@@ -958,6 +1647,25 @@ define i1 @ssubo.not.i32(i32 %v1, i32 %v2) {
 ; RV64-NEXT:    xor a0, a0, a2
 ; RV64-NEXT:    seqz a0, a0
 ; RV64-NEXT:    ret
+;
+; RV32ZBA-LABEL: ssubo.not.i32:
+; RV32ZBA:       # %bb.0: # %entry
+; RV32ZBA-NEXT:    sgtz a2, a1
+; RV32ZBA-NEXT:    sub a1, a0, a1
+; RV32ZBA-NEXT:    slt a0, a1, a0
+; RV32ZBA-NEXT:    xor a0, a2, a0
+; RV32ZBA-NEXT:    xori a0, a0, 1
+; RV32ZBA-NEXT:    ret
+;
+; RV64ZBA-LABEL: ssubo.not.i32:
+; RV64ZBA:       # %bb.0: # %entry
+; RV64ZBA-NEXT:    sext.w a1, a1
+; RV64ZBA-NEXT:    sext.w a0, a0
+; RV64ZBA-NEXT:    sub a2, a0, a1
+; RV64ZBA-NEXT:    subw a0, a0, a1
+; RV64ZBA-NEXT:    xor a0, a0, a2
+; RV64ZBA-NEXT:    seqz a0, a0
+; RV64ZBA-NEXT:    ret
 entry:
   %t = call {i32, i1} @llvm.ssub.with.overflow.i32(i32 %v1, i32 %v2)
   %obit = extractvalue {i32, i1} %t, 1
@@ -991,6 +1699,32 @@ define i64 @ssubo.select.i64(i64 %v1, i64 %v2) {
 ; RV64-NEXT:    mv a0, a1
 ; RV64-NEXT:  .LBB32_2: # %entry
 ; RV64-NEXT:    ret
+;
+; RV32ZBA-LABEL: ssubo.select.i64:
+; RV32ZBA:       # %bb.0: # %entry
+; RV32ZBA-NEXT:    sltu a4, a0, a2
+; RV32ZBA-NEXT:    sub a5, a1, a3
+; RV32ZBA-NEXT:    sub a4, a5, a4
+; RV32ZBA-NEXT:    xor a4, a1, a4
+; RV32ZBA-NEXT:    xor a5, a1, a3
+; RV32ZBA-NEXT:    and a4, a5, a4
+; RV32ZBA-NEXT:    bltz a4, .LBB32_2
+; RV32ZBA-NEXT:  # %bb.1: # %entry
+; RV32ZBA-NEXT:    mv a0, a2
+; RV32ZBA-NEXT:    mv a1, a3
+; RV32ZBA-NEXT:  .LBB32_2: # %entry
+; RV32ZBA-NEXT:    ret
+;
+; RV64ZBA-LABEL: ssubo.select.i64:
+; RV64ZBA:       # %bb.0: # %entry
+; RV64ZBA-NEXT:    sgtz a2, a1
+; RV64ZBA-NEXT:    sub a3, a0, a1
+; RV64ZBA-NEXT:    slt a3, a3, a0
+; RV64ZBA-NEXT:    bne a2, a3, .LBB32_2
+; RV64ZBA-NEXT:  # %bb.1: # %entry
+; RV64ZBA-NEXT:    mv a0, a1
+; RV64ZBA-NEXT:  .LBB32_2: # %entry
+; RV64ZBA-NEXT:    ret
 entry:
   %t = call {i64, i1} @llvm.ssub.with.overflow.i64(i64 %v1, i64 %v2)
   %obit = extractvalue {i64, i1} %t, 1
@@ -1019,6 +1753,27 @@ define i1 @ssub.not.i64(i64 %v1, i64 %v2) {
 ; RV64-NEXT:    xor a0, a2, a0
 ; RV64-NEXT:    xori a0, a0, 1
 ; RV64-NEXT:    ret
+;
+; RV32ZBA-LABEL: ssub.not.i64:
+; RV32ZBA:       # %bb.0: # %entry
+; RV32ZBA-NEXT:    sltu a0, a0, a2
+; RV32ZBA-NEXT:    sub a2, a1, a3
+; RV32ZBA-NEXT:    sub a0, a2, a0
+; RV32ZBA-NEXT:    xor a0, a1, a0
+; RV32ZBA-NEXT:    xor a1, a1, a3
+; RV32ZBA-NEXT:    and a0, a1, a0
+; RV32ZBA-NEXT:    addi a1, zero, -1
+; RV32ZBA-NEXT:    slt a0, a1, a0
+; RV32ZBA-NEXT:    ret
+;
+; RV64ZBA-LABEL: ssub.not.i64:
+; RV64ZBA:       # %bb.0: # %entry
+; RV64ZBA-NEXT:    sgtz a2, a1
+; RV64ZBA-NEXT:    sub a1, a0, a1
+; RV64ZBA-NEXT:    slt a0, a1, a0
+; RV64ZBA-NEXT:    xor a0, a2, a0
+; RV64ZBA-NEXT:    xori a0, a0, 1
+; RV64ZBA-NEXT:    ret
 entry:
   %t = call {i64, i1} @llvm.ssub.with.overflow.i64(i64 %v1, i64 %v2)
   %obit = extractvalue {i64, i1} %t, 1
@@ -1045,6 +1800,25 @@ define i32 @usubo.select.i32(i32 %v1, i32 %v2) {
 ; RV64-NEXT:    mv a0, a1
 ; RV64-NEXT:  .LBB34_2: # %entry
 ; RV64-NEXT:    ret
+;
+; RV32ZBA-LABEL: usubo.select.i32:
+; RV32ZBA:       # %bb.0: # %entry
+; RV32ZBA-NEXT:    sub a2, a0, a1
+; RV32ZBA-NEXT:    bltu a0, a2, .LBB34_2
+; RV32ZBA-NEXT:  # %bb.1: # %entry
+; RV32ZBA-NEXT:    mv a0, a1
+; RV32ZBA-NEXT:  .LBB34_2: # %entry
+; RV32ZBA-NEXT:    ret
+;
+; RV64ZBA-LABEL: usubo.select.i32:
+; RV64ZBA:       # %bb.0: # %entry
+; RV64ZBA-NEXT:    subw a2, a0, a1
+; RV64ZBA-NEXT:    sext.w a3, a0
+; RV64ZBA-NEXT:    bltu a3, a2, .LBB34_2
+; RV64ZBA-NEXT:  # %bb.1: # %entry
+; RV64ZBA-NEXT:    mv a0, a1
+; RV64ZBA-NEXT:  .LBB34_2: # %entry
+; RV64ZBA-NEXT:    ret
 entry:
   %t = call {i32, i1} @llvm.usub.with.overflow.i32(i32 %v1, i32 %v2)
   %obit = extractvalue {i32, i1} %t, 1
@@ -1067,6 +1841,21 @@ define i1 @usubo.not.i32(i32 %v1, i32 %v2) {
 ; RV64-NEXT:    sltu a0, a0, a1
 ; RV64-NEXT:    xori a0, a0, 1
 ; RV64-NEXT:    ret
+;
+; RV32ZBA-LABEL: usubo.not.i32:
+; RV32ZBA:       # %bb.0: # %entry
+; RV32ZBA-NEXT:    sub a1, a0, a1
+; RV32ZBA-NEXT:    sltu a0, a0, a1
+; RV32ZBA-NEXT:    xori a0, a0, 1
+; RV32ZBA-NEXT:    ret
+;
+; RV64ZBA-LABEL: usubo.not.i32:
+; RV64ZBA:       # %bb.0: # %entry
+; RV64ZBA-NEXT:    subw a1, a0, a1
+; RV64ZBA-NEXT:    sext.w a0, a0
+; RV64ZBA-NEXT:    sltu a0, a0, a1
+; RV64ZBA-NEXT:    xori a0, a0, 1
+; RV64ZBA-NEXT:    ret
 entry:
   %t = call {i32, i1} @llvm.usub.with.overflow.i32(i32 %v1, i32 %v2)
   %obit = extractvalue {i32, i1} %t, 1
@@ -1103,6 +1892,35 @@ define i64 @usubo.select.i64(i64 %v1, i64 %v2) {
 ; RV64-NEXT:    mv a0, a1
 ; RV64-NEXT:  .LBB36_2: # %entry
 ; RV64-NEXT:    ret
+;
+; RV32ZBA-LABEL: usubo.select.i64:
+; RV32ZBA:       # %bb.0: # %entry
+; RV32ZBA-NEXT:    sltu a4, a0, a2
+; RV32ZBA-NEXT:    sub a5, a1, a3
+; RV32ZBA-NEXT:    sub a4, a5, a4
+; RV32ZBA-NEXT:    beq a4, a1, .LBB36_2
+; RV32ZBA-NEXT:  # %bb.1: # %entry
+; RV32ZBA-NEXT:    sltu a4, a1, a4
+; RV32ZBA-NEXT:    beqz a4, .LBB36_3
+; RV32ZBA-NEXT:    j .LBB36_4
+; RV32ZBA-NEXT:  .LBB36_2:
+; RV32ZBA-NEXT:    sub a4, a0, a2
+; RV32ZBA-NEXT:    sltu a4, a0, a4
+; RV32ZBA-NEXT:    bnez a4, .LBB36_4
+; RV32ZBA-NEXT:  .LBB36_3: # %entry
+; RV32ZBA-NEXT:    mv a0, a2
+; RV32ZBA-NEXT:    mv a1, a3
+; RV32ZBA-NEXT:  .LBB36_4: # %entry
+; RV32ZBA-NEXT:    ret
+;
+; RV64ZBA-LABEL: usubo.select.i64:
+; RV64ZBA:       # %bb.0: # %entry
+; RV64ZBA-NEXT:    sub a2, a0, a1
+; RV64ZBA-NEXT:    bltu a0, a2, .LBB36_2
+; RV64ZBA-NEXT:  # %bb.1: # %entry
+; RV64ZBA-NEXT:    mv a0, a1
+; RV64ZBA-NEXT:  .LBB36_2: # %entry
+; RV64ZBA-NEXT:    ret
 entry:
   %t = call {i64, i1} @llvm.usub.with.overflow.i64(i64 %v1, i64 %v2)
   %obit = extractvalue {i64, i1} %t, 1
@@ -1133,6 +1951,29 @@ define i1 @usubo.not.i64(i64 %v1, i64 %v2) {
 ; RV64-NEXT:    sltu a0, a0, a1
 ; RV64-NEXT:    xori a0, a0, 1
 ; RV64-NEXT:    ret
+;
+; RV32ZBA-LABEL: usubo.not.i64:
+; RV32ZBA:       # %bb.0: # %entry
+; RV32ZBA-NEXT:    sltu a4, a0, a2
+; RV32ZBA-NEXT:    sub a3, a1, a3
+; RV32ZBA-NEXT:    sub a3, a3, a4
+; RV32ZBA-NEXT:    beq a3, a1, .LBB37_2
+; RV32ZBA-NEXT:  # %bb.1: # %entry
+; RV32ZBA-NEXT:    sltu a0, a1, a3
+; RV32ZBA-NEXT:    xori a0, a0, 1
+; RV32ZBA-NEXT:    ret
+; RV32ZBA-NEXT:  .LBB37_2:
+; RV32ZBA-NEXT:    sub a1, a0, a2
+; RV32ZBA-NEXT:    sltu a0, a0, a1
+; RV32ZBA-NEXT:    xori a0, a0, 1
+; RV32ZBA-NEXT:    ret
+;
+; RV64ZBA-LABEL: usubo.not.i64:
+; RV64ZBA:       # %bb.0: # %entry
+; RV64ZBA-NEXT:    sub a1, a0, a1
+; RV64ZBA-NEXT:    sltu a0, a0, a1
+; RV64ZBA-NEXT:    xori a0, a0, 1
+; RV64ZBA-NEXT:    ret
 entry:
   %t = call {i64, i1} @llvm.usub.with.overflow.i64(i64 %v1, i64 %v2)
   %obit = extractvalue {i64, i1} %t, 1
@@ -1163,6 +2004,29 @@ define i32 @smulo.select.i32(i32 %v1, i32 %v2) {
 ; RV64-NEXT:    mv a0, a1
 ; RV64-NEXT:  .LBB38_2: # %entry
 ; RV64-NEXT:    ret
+;
+; RV32ZBA-LABEL: smulo.select.i32:
+; RV32ZBA:       # %bb.0: # %entry
+; RV32ZBA-NEXT:    mulh a2, a0, a1
+; RV32ZBA-NEXT:    mul a3, a0, a1
+; RV32ZBA-NEXT:    srai a3, a3, 31
+; RV32ZBA-NEXT:    bne a2, a3, .LBB38_2
+; RV32ZBA-NEXT:  # %bb.1: # %entry
+; RV32ZBA-NEXT:    mv a0, a1
+; RV32ZBA-NEXT:  .LBB38_2: # %entry
+; RV32ZBA-NEXT:    ret
+;
+; RV64ZBA-LABEL: smulo.select.i32:
+; RV64ZBA:       # %bb.0: # %entry
+; RV64ZBA-NEXT:    sext.w a2, a1
+; RV64ZBA-NEXT:    sext.w a3, a0
+; RV64ZBA-NEXT:    mul a4, a3, a2
+; RV64ZBA-NEXT:    mulw a2, a3, a2
+; RV64ZBA-NEXT:    bne a2, a4, .LBB38_2
+; RV64ZBA-NEXT:  # %bb.1: # %entry
+; RV64ZBA-NEXT:    mv a0, a1
+; RV64ZBA-NEXT:  .LBB38_2: # %entry
+; RV64ZBA-NEXT:    ret
 entry:
   %t = call {i32, i1} @llvm.smul.with.overflow.i32(i32 %v1, i32 %v2)
   %obit = extractvalue {i32, i1} %t, 1
@@ -1189,6 +2053,25 @@ define i1 @smulo.not.i32(i32 %v1, i32 %v2) {
 ; RV64-NEXT:    xor a0, a0, a2
 ; RV64-NEXT:    seqz a0, a0
 ; RV64-NEXT:    ret
+;
+; RV32ZBA-LABEL: smulo.not.i32:
+; RV32ZBA:       # %bb.0: # %entry
+; RV32ZBA-NEXT:    mulh a2, a0, a1
+; RV32ZBA-NEXT:    mul a0, a0, a1
+; RV32ZBA-NEXT:    srai a0, a0, 31
+; RV32ZBA-NEXT:    xor a0, a2, a0
+; RV32ZBA-NEXT:    seqz a0, a0
+; RV32ZBA-NEXT:    ret
+;
+; RV64ZBA-LABEL: smulo.not.i32:
+; RV64ZBA:       # %bb.0: # %entry
+; RV64ZBA-NEXT:    sext.w a1, a1
+; RV64ZBA-NEXT:    sext.w a0, a0
+; RV64ZBA-NEXT:    mul a2, a0, a1
+; RV64ZBA-NEXT:    mulw a0, a0, a1
+; RV64ZBA-NEXT:    xor a0, a0, a2
+; RV64ZBA-NEXT:    seqz a0, a0
+; RV64ZBA-NEXT:    ret
 entry:
   %t = call {i32, i1} @llvm.smul.with.overflow.i32(i32 %v1, i32 %v2)
   %obit = extractvalue {i32, i1} %t, 1
@@ -1244,6 +2127,54 @@ define i64 @smulo.select.i64(i64 %v1, i64 %v2) {
 ; RV64-NEXT:    mv a0, a1
 ; RV64-NEXT:  .LBB40_2: # %entry
 ; RV64-NEXT:    ret
+;
+; RV32ZBA-LABEL: smulo.select.i64:
+; RV32ZBA:       # %bb.0: # %entry
+; RV32ZBA-NEXT:    addi sp, sp, -32
+; RV32ZBA-NEXT:    .cfi_def_cfa_offset 32
+; RV32ZBA-NEXT:    sw ra, 28(sp) # 4-byte Folded Spill
+; RV32ZBA-NEXT:    sw s0, 24(sp) # 4-byte Folded Spill
+; RV32ZBA-NEXT:    sw s1, 20(sp) # 4-byte Folded Spill
+; RV32ZBA-NEXT:    sw s2, 16(sp) # 4-byte Folded Spill
+; RV32ZBA-NEXT:    sw s3, 12(sp) # 4-byte Folded Spill
+; RV32ZBA-NEXT:    .cfi_offset ra, -4
+; RV32ZBA-NEXT:    .cfi_offset s0, -8
+; RV32ZBA-NEXT:    .cfi_offset s1, -12
+; RV32ZBA-NEXT:    .cfi_offset s2, -16
+; RV32ZBA-NEXT:    .cfi_offset s3, -20
+; RV32ZBA-NEXT:    mv s2, a3
+; RV32ZBA-NEXT:    mv s3, a2
+; RV32ZBA-NEXT:    mv s0, a1
+; RV32ZBA-NEXT:    mv s1, a0
+; RV32ZBA-NEXT:    sw zero, 8(sp)
+; RV32ZBA-NEXT:    addi a4, sp, 8
+; RV32ZBA-NEXT:    call __mulodi4@plt
+; RV32ZBA-NEXT:    lw a0, 8(sp)
+; RV32ZBA-NEXT:    bnez a0, .LBB40_2
+; RV32ZBA-NEXT:  # %bb.1: # %entry
+; RV32ZBA-NEXT:    mv s1, s3
+; RV32ZBA-NEXT:    mv s0, s2
+; RV32ZBA-NEXT:  .LBB40_2: # %entry
+; RV32ZBA-NEXT:    mv a0, s1
+; RV32ZBA-NEXT:    mv a1, s0
+; RV32ZBA-NEXT:    lw s3, 12(sp) # 4-byte Folded Reload
+; RV32ZBA-NEXT:    lw s2, 16(sp) # 4-byte Folded Reload
+; RV32ZBA-NEXT:    lw s1, 20(sp) # 4-byte Folded Reload
+; RV32ZBA-NEXT:    lw s0, 24(sp) # 4-byte Folded Reload
+; RV32ZBA-NEXT:    lw ra, 28(sp) # 4-byte Folded Reload
+; RV32ZBA-NEXT:    addi sp, sp, 32
+; RV32ZBA-NEXT:    ret
+;
+; RV64ZBA-LABEL: smulo.select.i64:
+; RV64ZBA:       # %bb.0: # %entry
+; RV64ZBA-NEXT:    mulh a2, a0, a1
+; RV64ZBA-NEXT:    mul a3, a0, a1
+; RV64ZBA-NEXT:    srai a3, a3, 63
+; RV64ZBA-NEXT:    bne a2, a3, .LBB40_2
+; RV64ZBA-NEXT:  # %bb.1: # %entry
+; RV64ZBA-NEXT:    mv a0, a1
+; RV64ZBA-NEXT:  .LBB40_2: # %entry
+; RV64ZBA-NEXT:    ret
 entry:
   %t = call {i64, i1} @llvm.smul.with.overflow.i64(i64 %v1, i64 %v2)
   %obit = extractvalue {i64, i1} %t, 1
@@ -1275,6 +2206,30 @@ define i1 @smulo.not.i64(i64 %v1, i64 %v2) {
 ; RV64-NEXT:    xor a0, a2, a0
 ; RV64-NEXT:    seqz a0, a0
 ; RV64-NEXT:    ret
+;
+; RV32ZBA-LABEL: smulo.not.i64:
+; RV32ZBA:       # %bb.0: # %entry
+; RV32ZBA-NEXT:    addi sp, sp, -16
+; RV32ZBA-NEXT:    .cfi_def_cfa_offset 16
+; RV32ZBA-NEXT:    sw ra, 12(sp) # 4-byte Folded Spill
+; RV32ZBA-NEXT:    .cfi_offset ra, -4
+; RV32ZBA-NEXT:    sw zero, 8(sp)
+; RV32ZBA-NEXT:    addi a4, sp, 8
+; RV32ZBA-NEXT:    call __mulodi4@plt
+; RV32ZBA-NEXT:    lw a0, 8(sp)
+; RV32ZBA-NEXT:    seqz a0, a0
+; RV32ZBA-NEXT:    lw ra, 12(sp) # 4-byte Folded Reload
+; RV32ZBA-NEXT:    addi sp, sp, 16
+; RV32ZBA-NEXT:    ret
+;
+; RV64ZBA-LABEL: smulo.not.i64:
+; RV64ZBA:       # %bb.0: # %entry
+; RV64ZBA-NEXT:    mulh a2, a0, a1
+; RV64ZBA-NEXT:    mul a0, a0, a1
+; RV64ZBA-NEXT:    srai a0, a0, 63
+; RV64ZBA-NEXT:    xor a0, a2, a0
+; RV64ZBA-NEXT:    seqz a0, a0
+; RV64ZBA-NEXT:    ret
 entry:
   %t = call {i64, i1} @llvm.smul.with.overflow.i64(i64 %v1, i64 %v2)
   %obit = extractvalue {i64, i1} %t, 1
@@ -1303,6 +2258,27 @@ define i32 @umulo.select.i32(i32 %v1, i32 %v2) {
 ; RV64-NEXT:    mv a0, a1
 ; RV64-NEXT:  .LBB42_2: # %entry
 ; RV64-NEXT:    ret
+;
+; RV32ZBA-LABEL: umulo.select.i32:
+; RV32ZBA:       # %bb.0: # %entry
+; RV32ZBA-NEXT:    mulhu a2, a0, a1
+; RV32ZBA-NEXT:    bnez a2, .LBB42_2
+; RV32ZBA-NEXT:  # %bb.1: # %entry
+; RV32ZBA-NEXT:    mv a0, a1
+; RV32ZBA-NEXT:  .LBB42_2: # %entry
+; RV32ZBA-NEXT:    ret
+;
+; RV64ZBA-LABEL: umulo.select.i32:
+; RV64ZBA:       # %bb.0: # %entry
+; RV64ZBA-NEXT:    slli a2, a1, 32
+; RV64ZBA-NEXT:    slli a3, a0, 32
+; RV64ZBA-NEXT:    mulhu a2, a3, a2
+; RV64ZBA-NEXT:    srli a2, a2, 32
+; RV64ZBA-NEXT:    bnez a2, .LBB42_2
+; RV64ZBA-NEXT:  # %bb.1: # %entry
+; RV64ZBA-NEXT:    mv a0, a1
+; RV64ZBA-NEXT:  .LBB42_2: # %entry
+; RV64ZBA-NEXT:    ret
 entry:
   %t = call {i32, i1} @llvm.umul.with.overflow.i32(i32 %v1, i32 %v2)
   %obit = extractvalue {i32, i1} %t, 1
@@ -1325,6 +2301,21 @@ define i1 @umulo.not.i32(i32 %v1, i32 %v2) {
 ; RV64-NEXT:    srli a0, a0, 32
 ; RV64-NEXT:    seqz a0, a0
 ; RV64-NEXT:    ret
+;
+; RV32ZBA-LABEL: umulo.not.i32:
+; RV32ZBA:       # %bb.0: # %entry
+; RV32ZBA-NEXT:    mulhu a0, a0, a1
+; RV32ZBA-NEXT:    seqz a0, a0
+; RV32ZBA-NEXT:    ret
+;
+; RV64ZBA-LABEL: umulo.not.i32:
+; RV64ZBA:       # %bb.0: # %entry
+; RV64ZBA-NEXT:    slli a1, a1, 32
+; RV64ZBA-NEXT:    slli a0, a0, 32
+; RV64ZBA-NEXT:    mulhu a0, a0, a1
+; RV64ZBA-NEXT:    srli a0, a0, 32
+; RV64ZBA-NEXT:    seqz a0, a0
+; RV64ZBA-NEXT:    ret
 entry:
   %t = call {i32, i1} @llvm.umul.with.overflow.i32(i32 %v1, i32 %v2)
   %obit = extractvalue {i32, i1} %t, 1
@@ -1366,6 +2357,40 @@ define i64 @umulo.select.i64(i64 %v1, i64 %v2) {
 ; RV64-NEXT:    mv a0, a1
 ; RV64-NEXT:  .LBB44_2: # %entry
 ; RV64-NEXT:    ret
+;
+; RV32ZBA-LABEL: umulo.select.i64:
+; RV32ZBA:       # %bb.0: # %entry
+; RV32ZBA-NEXT:    mul a4, a3, a0
+; RV32ZBA-NEXT:    mul a5, a1, a2
+; RV32ZBA-NEXT:    add a4, a5, a4
+; RV32ZBA-NEXT:    mulhu a5, a0, a2
+; RV32ZBA-NEXT:    add a4, a5, a4
+; RV32ZBA-NEXT:    sltu a6, a4, a5
+; RV32ZBA-NEXT:    snez a5, a3
+; RV32ZBA-NEXT:    snez a4, a1
+; RV32ZBA-NEXT:    and a4, a4, a5
+; RV32ZBA-NEXT:    mulhu a5, a1, a2
+; RV32ZBA-NEXT:    snez a5, a5
+; RV32ZBA-NEXT:    or a4, a4, a5
+; RV32ZBA-NEXT:    mulhu a5, a3, a0
+; RV32ZBA-NEXT:    snez a5, a5
+; RV32ZBA-NEXT:    or a4, a4, a5
+; RV32ZBA-NEXT:    or a4, a4, a6
+; RV32ZBA-NEXT:    bnez a4, .LBB44_2
+; RV32ZBA-NEXT:  # %bb.1: # %entry
+; RV32ZBA-NEXT:    mv a0, a2
+; RV32ZBA-NEXT:    mv a1, a3
+; RV32ZBA-NEXT:  .LBB44_2: # %entry
+; RV32ZBA-NEXT:    ret
+;
+; RV64ZBA-LABEL: umulo.select.i64:
+; RV64ZBA:       # %bb.0: # %entry
+; RV64ZBA-NEXT:    mulhu a2, a0, a1
+; RV64ZBA-NEXT:    bnez a2, .LBB44_2
+; RV64ZBA-NEXT:  # %bb.1: # %entry
+; RV64ZBA-NEXT:    mv a0, a1
+; RV64ZBA-NEXT:  .LBB44_2: # %entry
+; RV64ZBA-NEXT:    ret
 entry:
   %t = call {i64, i1} @llvm.umul.with.overflow.i64(i64 %v1, i64 %v2)
   %obit = extractvalue {i64, i1} %t, 1
@@ -1400,6 +2425,33 @@ define i1 @umulo.not.i64(i64 %v1, i64 %v2) {
 ; RV64-NEXT:    mulhu a0, a0, a1
 ; RV64-NEXT:    seqz a0, a0
 ; RV64-NEXT:    ret
+;
+; RV32ZBA-LABEL: umulo.not.i64:
+; RV32ZBA:       # %bb.0: # %entry
+; RV32ZBA-NEXT:    mul a4, a3, a0
+; RV32ZBA-NEXT:    mul a5, a1, a2
+; RV32ZBA-NEXT:    add a4, a5, a4
+; RV32ZBA-NEXT:    mulhu a5, a0, a2
+; RV32ZBA-NEXT:    add a4, a5, a4
+; RV32ZBA-NEXT:    sltu a6, a4, a5
+; RV32ZBA-NEXT:    snez a5, a3
+; RV32ZBA-NEXT:    snez a4, a1
+; RV32ZBA-NEXT:    and a4, a4, a5
+; RV32ZBA-NEXT:    mulhu a1, a1, a2
+; RV32ZBA-NEXT:    snez a1, a1
+; RV32ZBA-NEXT:    or a1, a4, a1
+; RV32ZBA-NEXT:    mulhu a0, a3, a0
+; RV32ZBA-NEXT:    snez a0, a0
+; RV32ZBA-NEXT:    or a0, a1, a0
+; RV32ZBA-NEXT:    or a0, a0, a6
+; RV32ZBA-NEXT:    xori a0, a0, 1
+; RV32ZBA-NEXT:    ret
+;
+; RV64ZBA-LABEL: umulo.not.i64:
+; RV64ZBA:       # %bb.0: # %entry
+; RV64ZBA-NEXT:    mulhu a0, a0, a1
+; RV64ZBA-NEXT:    seqz a0, a0
+; RV64ZBA-NEXT:    ret
 entry:
   %t = call {i64, i1} @llvm.umul.with.overflow.i64(i64 %v1, i64 %v2)
   %obit = extractvalue {i64, i1} %t, 1
@@ -1438,6 +2490,33 @@ define zeroext i1 @saddo.br.i32(i32 %v1, i32 %v2) {
 ; RV64-NEXT:  .LBB46_2: # %continue
 ; RV64-NEXT:    addi a0, zero, 1
 ; RV64-NEXT:    ret
+;
+; RV32ZBA-LABEL: saddo.br.i32:
+; RV32ZBA:       # %bb.0: # %entry
+; RV32ZBA-NEXT:    add a2, a0, a1
+; RV32ZBA-NEXT:    slt a0, a2, a0
+; RV32ZBA-NEXT:    slti a1, a1, 0
+; RV32ZBA-NEXT:    beq a1, a0, .LBB46_2
+; RV32ZBA-NEXT:  # %bb.1: # %overflow
+; RV32ZBA-NEXT:    mv a0, zero
+; RV32ZBA-NEXT:    ret
+; RV32ZBA-NEXT:  .LBB46_2: # %continue
+; RV32ZBA-NEXT:    addi a0, zero, 1
+; RV32ZBA-NEXT:    ret
+;
+; RV64ZBA-LABEL: saddo.br.i32:
+; RV64ZBA:       # %bb.0: # %entry
+; RV64ZBA-NEXT:    sext.w a1, a1
+; RV64ZBA-NEXT:    sext.w a0, a0
+; RV64ZBA-NEXT:    add a2, a0, a1
+; RV64ZBA-NEXT:    addw a0, a0, a1
+; RV64ZBA-NEXT:    beq a0, a2, .LBB46_2
+; RV64ZBA-NEXT:  # %bb.1: # %overflow
+; RV64ZBA-NEXT:    mv a0, zero
+; RV64ZBA-NEXT:    ret
+; RV64ZBA-NEXT:  .LBB46_2: # %continue
+; RV64ZBA-NEXT:    addi a0, zero, 1
+; RV64ZBA-NEXT:    ret
 entry:
   %t = call {i32, i1} @llvm.sadd.with.overflow.i32(i32 %v1, i32 %v2)
   %val = extractvalue {i32, i1} %t, 0
@@ -1482,6 +2561,37 @@ define zeroext i1 @saddo.br.i64(i64 %v1, i64 %v2) {
 ; RV64-NEXT:  .LBB47_2: # %continue
 ; RV64-NEXT:    addi a0, zero, 1
 ; RV64-NEXT:    ret
+;
+; RV32ZBA-LABEL: saddo.br.i64:
+; RV32ZBA:       # %bb.0: # %entry
+; RV32ZBA-NEXT:    add a4, a1, a3
+; RV32ZBA-NEXT:    add a2, a0, a2
+; RV32ZBA-NEXT:    sltu a0, a2, a0
+; RV32ZBA-NEXT:    add a0, a4, a0
+; RV32ZBA-NEXT:    xor a0, a1, a0
+; RV32ZBA-NEXT:    xor a1, a1, a3
+; RV32ZBA-NEXT:    not a1, a1
+; RV32ZBA-NEXT:    and a0, a1, a0
+; RV32ZBA-NEXT:    bgez a0, .LBB47_2
+; RV32ZBA-NEXT:  # %bb.1: # %overflow
+; RV32ZBA-NEXT:    mv a0, zero
+; RV32ZBA-NEXT:    ret
+; RV32ZBA-NEXT:  .LBB47_2: # %continue
+; RV32ZBA-NEXT:    addi a0, zero, 1
+; RV32ZBA-NEXT:    ret
+;
+; RV64ZBA-LABEL: saddo.br.i64:
+; RV64ZBA:       # %bb.0: # %entry
+; RV64ZBA-NEXT:    add a2, a0, a1
+; RV64ZBA-NEXT:    slt a0, a2, a0
+; RV64ZBA-NEXT:    slti a1, a1, 0
+; RV64ZBA-NEXT:    beq a1, a0, .LBB47_2
+; RV64ZBA-NEXT:  # %bb.1: # %overflow
+; RV64ZBA-NEXT:    mv a0, zero
+; RV64ZBA-NEXT:    ret
+; RV64ZBA-NEXT:  .LBB47_2: # %continue
+; RV64ZBA-NEXT:    addi a0, zero, 1
+; RV64ZBA-NEXT:    ret
 entry:
   %t = call {i64, i1} @llvm.sadd.with.overflow.i64(i64 %v1, i64 %v2)
   %val = extractvalue {i64, i1} %t, 0
@@ -1518,6 +2628,29 @@ define zeroext i1 @uaddo.br.i32(i32 %v1, i32 %v2) {
 ; RV64-NEXT:  .LBB48_2: # %continue
 ; RV64-NEXT:    addi a0, zero, 1
 ; RV64-NEXT:    ret
+;
+; RV32ZBA-LABEL: uaddo.br.i32:
+; RV32ZBA:       # %bb.0: # %entry
+; RV32ZBA-NEXT:    add a1, a0, a1
+; RV32ZBA-NEXT:    bgeu a1, a0, .LBB48_2
+; RV32ZBA-NEXT:  # %bb.1: # %overflow
+; RV32ZBA-NEXT:    mv a0, zero
+; RV32ZBA-NEXT:    ret
+; RV32ZBA-NEXT:  .LBB48_2: # %continue
+; RV32ZBA-NEXT:    addi a0, zero, 1
+; RV32ZBA-NEXT:    ret
+;
+; RV64ZBA-LABEL: uaddo.br.i32:
+; RV64ZBA:       # %bb.0: # %entry
+; RV64ZBA-NEXT:    addw a1, a0, a1
+; RV64ZBA-NEXT:    sext.w a0, a0
+; RV64ZBA-NEXT:    bgeu a1, a0, .LBB48_2
+; RV64ZBA-NEXT:  # %bb.1: # %overflow
+; RV64ZBA-NEXT:    mv a0, zero
+; RV64ZBA-NEXT:    ret
+; RV64ZBA-NEXT:  .LBB48_2: # %continue
+; RV64ZBA-NEXT:    addi a0, zero, 1
+; RV64ZBA-NEXT:    ret
 entry:
   %t = call {i32, i1} @llvm.uadd.with.overflow.i32(i32 %v1, i32 %v2)
   %val = extractvalue {i32, i1} %t, 0
@@ -1560,6 +2693,35 @@ define zeroext i1 @uaddo.br.i64(i64 %v1, i64 %v2) {
 ; RV64-NEXT:  .LBB49_2: # %continue
 ; RV64-NEXT:    addi a0, zero, 1
 ; RV64-NEXT:    ret
+;
+; RV32ZBA-LABEL: uaddo.br.i64:
+; RV32ZBA:       # %bb.0: # %entry
+; RV32ZBA-NEXT:    add a3, a1, a3
+; RV32ZBA-NEXT:    add a2, a0, a2
+; RV32ZBA-NEXT:    sltu a0, a2, a0
+; RV32ZBA-NEXT:    add a2, a3, a0
+; RV32ZBA-NEXT:    beq a2, a1, .LBB49_2
+; RV32ZBA-NEXT:  # %bb.1: # %entry
+; RV32ZBA-NEXT:    sltu a0, a2, a1
+; RV32ZBA-NEXT:  .LBB49_2: # %entry
+; RV32ZBA-NEXT:    beqz a0, .LBB49_4
+; RV32ZBA-NEXT:  # %bb.3: # %overflow
+; RV32ZBA-NEXT:    mv a0, zero
+; RV32ZBA-NEXT:    ret
+; RV32ZBA-NEXT:  .LBB49_4: # %continue
+; RV32ZBA-NEXT:    addi a0, zero, 1
+; RV32ZBA-NEXT:    ret
+;
+; RV64ZBA-LABEL: uaddo.br.i64:
+; RV64ZBA:       # %bb.0: # %entry
+; RV64ZBA-NEXT:    add a1, a0, a1
+; RV64ZBA-NEXT:    bgeu a1, a0, .LBB49_2
+; RV64ZBA-NEXT:  # %bb.1: # %overflow
+; RV64ZBA-NEXT:    mv a0, zero
+; RV64ZBA-NEXT:    ret
+; RV64ZBA-NEXT:  .LBB49_2: # %continue
+; RV64ZBA-NEXT:    addi a0, zero, 1
+; RV64ZBA-NEXT:    ret
 entry:
   %t = call {i64, i1} @llvm.uadd.with.overflow.i64(i64 %v1, i64 %v2)
   %val = extractvalue {i64, i1} %t, 0
@@ -1600,6 +2762,33 @@ define zeroext i1 @ssubo.br.i32(i32 %v1, i32 %v2) {
 ; RV64-NEXT:  .LBB50_2: # %continue
 ; RV64-NEXT:    addi a0, zero, 1
 ; RV64-NEXT:    ret
+;
+; RV32ZBA-LABEL: ssubo.br.i32:
+; RV32ZBA:       # %bb.0: # %entry
+; RV32ZBA-NEXT:    sgtz a2, a1
+; RV32ZBA-NEXT:    sub a1, a0, a1
+; RV32ZBA-NEXT:    slt a0, a1, a0
+; RV32ZBA-NEXT:    beq a2, a0, .LBB50_2
+; RV32ZBA-NEXT:  # %bb.1: # %overflow
+; RV32ZBA-NEXT:    mv a0, zero
+; RV32ZBA-NEXT:    ret
+; RV32ZBA-NEXT:  .LBB50_2: # %continue
+; RV32ZBA-NEXT:    addi a0, zero, 1
+; RV32ZBA-NEXT:    ret
+;
+; RV64ZBA-LABEL: ssubo.br.i32:
+; RV64ZBA:       # %bb.0: # %entry
+; RV64ZBA-NEXT:    sext.w a1, a1
+; RV64ZBA-NEXT:    sext.w a0, a0
+; RV64ZBA-NEXT:    sub a2, a0, a1
+; RV64ZBA-NEXT:    subw a0, a0, a1
+; RV64ZBA-NEXT:    beq a0, a2, .LBB50_2
+; RV64ZBA-NEXT:  # %bb.1: # %overflow
+; RV64ZBA-NEXT:    mv a0, zero
+; RV64ZBA-NEXT:    ret
+; RV64ZBA-NEXT:  .LBB50_2: # %continue
+; RV64ZBA-NEXT:    addi a0, zero, 1
+; RV64ZBA-NEXT:    ret
 entry:
   %t = call {i32, i1} @llvm.ssub.with.overflow.i32(i32 %v1, i32 %v2)
   %val = extractvalue {i32, i1} %t, 0
@@ -1642,6 +2831,35 @@ define zeroext i1 @ssubo.br.i64(i64 %v1, i64 %v2) {
 ; RV64-NEXT:  .LBB51_2: # %continue
 ; RV64-NEXT:    addi a0, zero, 1
 ; RV64-NEXT:    ret
+;
+; RV32ZBA-LABEL: ssubo.br.i64:
+; RV32ZBA:       # %bb.0: # %entry
+; RV32ZBA-NEXT:    sltu a0, a0, a2
+; RV32ZBA-NEXT:    sub a2, a1, a3
+; RV32ZBA-NEXT:    sub a0, a2, a0
+; RV32ZBA-NEXT:    xor a0, a1, a0
+; RV32ZBA-NEXT:    xor a1, a1, a3
+; RV32ZBA-NEXT:    and a0, a1, a0
+; RV32ZBA-NEXT:    bgez a0, .LBB51_2
+; RV32ZBA-NEXT:  # %bb.1: # %overflow
+; RV32ZBA-NEXT:    mv a0, zero
+; RV32ZBA-NEXT:    ret
+; RV32ZBA-NEXT:  .LBB51_2: # %continue
+; RV32ZBA-NEXT:    addi a0, zero, 1
+; RV32ZBA-NEXT:    ret
+;
+; RV64ZBA-LABEL: ssubo.br.i64:
+; RV64ZBA:       # %bb.0: # %entry
+; RV64ZBA-NEXT:    sgtz a2, a1
+; RV64ZBA-NEXT:    sub a1, a0, a1
+; RV64ZBA-NEXT:    slt a0, a1, a0
+; RV64ZBA-NEXT:    beq a2, a0, .LBB51_2
+; RV64ZBA-NEXT:  # %bb.1: # %overflow
+; RV64ZBA-NEXT:    mv a0, zero
+; RV64ZBA-NEXT:    ret
+; RV64ZBA-NEXT:  .LBB51_2: # %continue
+; RV64ZBA-NEXT:    addi a0, zero, 1
+; RV64ZBA-NEXT:    ret
 entry:
   %t = call {i64, i1} @llvm.ssub.with.overflow.i64(i64 %v1, i64 %v2)
   %val = extractvalue {i64, i1} %t, 0
@@ -1678,6 +2896,29 @@ define zeroext i1 @usubo.br.i32(i32 %v1, i32 %v2) {
 ; RV64-NEXT:  .LBB52_2: # %continue
 ; RV64-NEXT:    addi a0, zero, 1
 ; RV64-NEXT:    ret
+;
+; RV32ZBA-LABEL: usubo.br.i32:
+; RV32ZBA:       # %bb.0: # %entry
+; RV32ZBA-NEXT:    sub a1, a0, a1
+; RV32ZBA-NEXT:    bgeu a0, a1, .LBB52_2
+; RV32ZBA-NEXT:  # %bb.1: # %overflow
+; RV32ZBA-NEXT:    mv a0, zero
+; RV32ZBA-NEXT:    ret
+; RV32ZBA-NEXT:  .LBB52_2: # %continue
+; RV32ZBA-NEXT:    addi a0, zero, 1
+; RV32ZBA-NEXT:    ret
+;
+; RV64ZBA-LABEL: usubo.br.i32:
+; RV64ZBA:       # %bb.0: # %entry
+; RV64ZBA-NEXT:    subw a1, a0, a1
+; RV64ZBA-NEXT:    sext.w a0, a0
+; RV64ZBA-NEXT:    bgeu a0, a1, .LBB52_2
+; RV64ZBA-NEXT:  # %bb.1: # %overflow
+; RV64ZBA-NEXT:    mv a0, zero
+; RV64ZBA-NEXT:    ret
+; RV64ZBA-NEXT:  .LBB52_2: # %continue
+; RV64ZBA-NEXT:    addi a0, zero, 1
+; RV64ZBA-NEXT:    ret
 entry:
   %t = call {i32, i1} @llvm.usub.with.overflow.i32(i32 %v1, i32 %v2)
   %val = extractvalue {i32, i1} %t, 0
@@ -1722,6 +2963,37 @@ define zeroext i1 @usubo.br.i64(i64 %v1, i64 %v2) {
 ; RV64-NEXT:  .LBB53_2: # %continue
 ; RV64-NEXT:    addi a0, zero, 1
 ; RV64-NEXT:    ret
+;
+; RV32ZBA-LABEL: usubo.br.i64:
+; RV32ZBA:       # %bb.0: # %entry
+; RV32ZBA-NEXT:    sltu a4, a0, a2
+; RV32ZBA-NEXT:    sub a3, a1, a3
+; RV32ZBA-NEXT:    sub a3, a3, a4
+; RV32ZBA-NEXT:    beq a3, a1, .LBB53_3
+; RV32ZBA-NEXT:  # %bb.1: # %entry
+; RV32ZBA-NEXT:    sltu a0, a1, a3
+; RV32ZBA-NEXT:    bnez a0, .LBB53_4
+; RV32ZBA-NEXT:  .LBB53_2: # %continue
+; RV32ZBA-NEXT:    addi a0, zero, 1
+; RV32ZBA-NEXT:    ret
+; RV32ZBA-NEXT:  .LBB53_3:
+; RV32ZBA-NEXT:    sub a1, a0, a2
+; RV32ZBA-NEXT:    sltu a0, a0, a1
+; RV32ZBA-NEXT:    beqz a0, .LBB53_2
+; RV32ZBA-NEXT:  .LBB53_4: # %overflow
+; RV32ZBA-NEXT:    mv a0, zero
+; RV32ZBA-NEXT:    ret
+;
+; RV64ZBA-LABEL: usubo.br.i64:
+; RV64ZBA:       # %bb.0: # %entry
+; RV64ZBA-NEXT:    sub a1, a0, a1
+; RV64ZBA-NEXT:    bgeu a0, a1, .LBB53_2
+; RV64ZBA-NEXT:  # %bb.1: # %overflow
+; RV64ZBA-NEXT:    mv a0, zero
+; RV64ZBA-NEXT:    ret
+; RV64ZBA-NEXT:  .LBB53_2: # %continue
+; RV64ZBA-NEXT:    addi a0, zero, 1
+; RV64ZBA-NEXT:    ret
 entry:
   %t = call {i64, i1} @llvm.usub.with.overflow.i64(i64 %v1, i64 %v2)
   %val = extractvalue {i64, i1} %t, 0
@@ -1762,6 +3034,33 @@ define zeroext i1 @smulo.br.i32(i32 %v1, i32 %v2) {
 ; RV64-NEXT:  .LBB54_2: # %continue
 ; RV64-NEXT:    addi a0, zero, 1
 ; RV64-NEXT:    ret
+;
+; RV32ZBA-LABEL: smulo.br.i32:
+; RV32ZBA:       # %bb.0: # %entry
+; RV32ZBA-NEXT:    mulh a2, a0, a1
+; RV32ZBA-NEXT:    mul a0, a0, a1
+; RV32ZBA-NEXT:    srai a0, a0, 31
+; RV32ZBA-NEXT:    beq a2, a0, .LBB54_2
+; RV32ZBA-NEXT:  # %bb.1: # %overflow
+; RV32ZBA-NEXT:    mv a0, zero
+; RV32ZBA-NEXT:    ret
+; RV32ZBA-NEXT:  .LBB54_2: # %continue
+; RV32ZBA-NEXT:    addi a0, zero, 1
+; RV32ZBA-NEXT:    ret
+;
+; RV64ZBA-LABEL: smulo.br.i32:
+; RV64ZBA:       # %bb.0: # %entry
+; RV64ZBA-NEXT:    sext.w a1, a1
+; RV64ZBA-NEXT:    sext.w a0, a0
+; RV64ZBA-NEXT:    mul a2, a0, a1
+; RV64ZBA-NEXT:    mulw a0, a0, a1
+; RV64ZBA-NEXT:    beq a0, a2, .LBB54_2
+; RV64ZBA-NEXT:  # %bb.1: # %overflow
+; RV64ZBA-NEXT:    mv a0, zero
+; RV64ZBA-NEXT:    ret
+; RV64ZBA-NEXT:  .LBB54_2: # %continue
+; RV64ZBA-NEXT:    addi a0, zero, 1
+; RV64ZBA-NEXT:    ret
 entry:
   %t = call {i32, i1} @llvm.smul.with.overflow.i32(i32 %v1, i32 %v2)
   %val = extractvalue {i32, i1} %t, 0
@@ -1809,6 +3108,40 @@ define zeroext i1 @smulo.br.i64(i64 %v1, i64 %v2) {
 ; RV64-NEXT:  .LBB55_2: # %continue
 ; RV64-NEXT:    addi a0, zero, 1
 ; RV64-NEXT:    ret
+;
+; RV32ZBA-LABEL: smulo.br.i64:
+; RV32ZBA:       # %bb.0: # %entry
+; RV32ZBA-NEXT:    addi sp, sp, -16
+; RV32ZBA-NEXT:    .cfi_def_cfa_offset 16
+; RV32ZBA-NEXT:    sw ra, 12(sp) # 4-byte Folded Spill
+; RV32ZBA-NEXT:    .cfi_offset ra, -4
+; RV32ZBA-NEXT:    sw zero, 8(sp)
+; RV32ZBA-NEXT:    addi a4, sp, 8
+; RV32ZBA-NEXT:    call __mulodi4@plt
+; RV32ZBA-NEXT:    lw a0, 8(sp)
+; RV32ZBA-NEXT:    beqz a0, .LBB55_2
+; RV32ZBA-NEXT:  # %bb.1: # %overflow
+; RV32ZBA-NEXT:    mv a0, zero
+; RV32ZBA-NEXT:    j .LBB55_3
+; RV32ZBA-NEXT:  .LBB55_2: # %continue
+; RV32ZBA-NEXT:    addi a0, zero, 1
+; RV32ZBA-NEXT:  .LBB55_3: # %overflow
+; RV32ZBA-NEXT:    lw ra, 12(sp) # 4-byte Folded Reload
+; RV32ZBA-NEXT:    addi sp, sp, 16
+; RV32ZBA-NEXT:    ret
+;
+; RV64ZBA-LABEL: smulo.br.i64:
+; RV64ZBA:       # %bb.0: # %entry
+; RV64ZBA-NEXT:    mulh a2, a0, a1
+; RV64ZBA-NEXT:    mul a0, a0, a1
+; RV64ZBA-NEXT:    srai a0, a0, 63
+; RV64ZBA-NEXT:    beq a2, a0, .LBB55_2
+; RV64ZBA-NEXT:  # %bb.1: # %overflow
+; RV64ZBA-NEXT:    mv a0, zero
+; RV64ZBA-NEXT:    ret
+; RV64ZBA-NEXT:  .LBB55_2: # %continue
+; RV64ZBA-NEXT:    addi a0, zero, 1
+; RV64ZBA-NEXT:    ret
 entry:
   %t = call {i64, i1} @llvm.smul.with.overflow.i64(i64 %v1, i64 %v2)
   %val = extractvalue {i64, i1} %t, 0
@@ -1859,6 +3192,43 @@ define zeroext i1 @smulo2.br.i64(i64 %v1) {
 ; RV64-NEXT:  .LBB56_2: # %continue
 ; RV64-NEXT:    addi a0, zero, 1
 ; RV64-NEXT:    ret
+;
+; RV32ZBA-LABEL: smulo2.br.i64:
+; RV32ZBA:       # %bb.0: # %entry
+; RV32ZBA-NEXT:    addi sp, sp, -16
+; RV32ZBA-NEXT:    .cfi_def_cfa_offset 16
+; RV32ZBA-NEXT:    sw ra, 12(sp) # 4-byte Folded Spill
+; RV32ZBA-NEXT:    .cfi_offset ra, -4
+; RV32ZBA-NEXT:    sw zero, 8(sp)
+; RV32ZBA-NEXT:    addi a2, zero, -13
+; RV32ZBA-NEXT:    addi a3, zero, -1
+; RV32ZBA-NEXT:    addi a4, sp, 8
+; RV32ZBA-NEXT:    call __mulodi4@plt
+; RV32ZBA-NEXT:    lw a0, 8(sp)
+; RV32ZBA-NEXT:    beqz a0, .LBB56_2
+; RV32ZBA-NEXT:  # %bb.1: # %overflow
+; RV32ZBA-NEXT:    mv a0, zero
+; RV32ZBA-NEXT:    j .LBB56_3
+; RV32ZBA-NEXT:  .LBB56_2: # %continue
+; RV32ZBA-NEXT:    addi a0, zero, 1
+; RV32ZBA-NEXT:  .LBB56_3: # %overflow
+; RV32ZBA-NEXT:    lw ra, 12(sp) # 4-byte Folded Reload
+; RV32ZBA-NEXT:    addi sp, sp, 16
+; RV32ZBA-NEXT:    ret
+;
+; RV64ZBA-LABEL: smulo2.br.i64:
+; RV64ZBA:       # %bb.0: # %entry
+; RV64ZBA-NEXT:    addi a1, zero, -13
+; RV64ZBA-NEXT:    mulh a2, a0, a1
+; RV64ZBA-NEXT:    mul a0, a0, a1
+; RV64ZBA-NEXT:    srai a0, a0, 63
+; RV64ZBA-NEXT:    beq a2, a0, .LBB56_2
+; RV64ZBA-NEXT:  # %bb.1: # %overflow
+; RV64ZBA-NEXT:    mv a0, zero
+; RV64ZBA-NEXT:    ret
+; RV64ZBA-NEXT:  .LBB56_2: # %continue
+; RV64ZBA-NEXT:    addi a0, zero, 1
+; RV64ZBA-NEXT:    ret
 entry:
   %t = call {i64, i1} @llvm.smul.with.overflow.i64(i64 %v1, i64 -13)
   %val = extractvalue {i64, i1} %t, 0
@@ -1897,6 +3267,31 @@ define zeroext i1 @umulo.br.i32(i32 %v1, i32 %v2) {
 ; RV64-NEXT:  .LBB57_2: # %continue
 ; RV64-NEXT:    addi a0, zero, 1
 ; RV64-NEXT:    ret
+;
+; RV32ZBA-LABEL: umulo.br.i32:
+; RV32ZBA:       # %bb.0: # %entry
+; RV32ZBA-NEXT:    mulhu a0, a0, a1
+; RV32ZBA-NEXT:    beqz a0, .LBB57_2
+; RV32ZBA-NEXT:  # %bb.1: # %overflow
+; RV32ZBA-NEXT:    mv a0, zero
+; RV32ZBA-NEXT:    ret
+; RV32ZBA-NEXT:  .LBB57_2: # %continue
+; RV32ZBA-NEXT:    addi a0, zero, 1
+; RV32ZBA-NEXT:    ret
+;
+; RV64ZBA-LABEL: umulo.br.i32:
+; RV64ZBA:       # %bb.0: # %entry
+; RV64ZBA-NEXT:    slli a1, a1, 32
+; RV64ZBA-NEXT:    slli a0, a0, 32
+; RV64ZBA-NEXT:    mulhu a0, a0, a1
+; RV64ZBA-NEXT:    srli a0, a0, 32
+; RV64ZBA-NEXT:    beqz a0, .LBB57_2
+; RV64ZBA-NEXT:  # %bb.1: # %overflow
+; RV64ZBA-NEXT:    mv a0, zero
+; RV64ZBA-NEXT:    ret
+; RV64ZBA-NEXT:  .LBB57_2: # %continue
+; RV64ZBA-NEXT:    addi a0, zero, 1
+; RV64ZBA-NEXT:    ret
 entry:
   %t = call {i32, i1} @llvm.umul.with.overflow.i32(i32 %v1, i32 %v2)
   %val = extractvalue {i32, i1} %t, 0
@@ -1947,6 +3342,43 @@ define zeroext i1 @umulo.br.i64(i64 %v1, i64 %v2) {
 ; RV64-NEXT:  .LBB58_2: # %continue
 ; RV64-NEXT:    addi a0, zero, 1
 ; RV64-NEXT:    ret
+;
+; RV32ZBA-LABEL: umulo.br.i64:
+; RV32ZBA:       # %bb.0: # %entry
+; RV32ZBA-NEXT:    mul a4, a3, a0
+; RV32ZBA-NEXT:    mul a5, a1, a2
+; RV32ZBA-NEXT:    add a4, a5, a4
+; RV32ZBA-NEXT:    mulhu a5, a0, a2
+; RV32ZBA-NEXT:    add a4, a5, a4
+; RV32ZBA-NEXT:    sltu a6, a4, a5
+; RV32ZBA-NEXT:    snez a5, a3
+; RV32ZBA-NEXT:    snez a4, a1
+; RV32ZBA-NEXT:    and a4, a4, a5
+; RV32ZBA-NEXT:    mulhu a1, a1, a2
+; RV32ZBA-NEXT:    snez a1, a1
+; RV32ZBA-NEXT:    or a1, a4, a1
+; RV32ZBA-NEXT:    mulhu a0, a3, a0
+; RV32ZBA-NEXT:    snez a0, a0
+; RV32ZBA-NEXT:    or a0, a1, a0
+; RV32ZBA-NEXT:    or a0, a0, a6
+; RV32ZBA-NEXT:    beqz a0, .LBB58_2
+; RV32ZBA-NEXT:  # %bb.1: # %overflow
+; RV32ZBA-NEXT:    mv a0, zero
+; RV32ZBA-NEXT:    ret
+; RV32ZBA-NEXT:  .LBB58_2: # %continue
+; RV32ZBA-NEXT:    addi a0, zero, 1
+; RV32ZBA-NEXT:    ret
+;
+; RV64ZBA-LABEL: umulo.br.i64:
+; RV64ZBA:       # %bb.0: # %entry
+; RV64ZBA-NEXT:    mulhu a0, a0, a1
+; RV64ZBA-NEXT:    beqz a0, .LBB58_2
+; RV64ZBA-NEXT:  # %bb.1: # %overflow
+; RV64ZBA-NEXT:    mv a0, zero
+; RV64ZBA-NEXT:    ret
+; RV64ZBA-NEXT:  .LBB58_2: # %continue
+; RV64ZBA-NEXT:    addi a0, zero, 1
+; RV64ZBA-NEXT:    ret
 entry:
   %t = call {i64, i1} @llvm.umul.with.overflow.i64(i64 %v1, i64 %v2)
   %val = extractvalue {i64, i1} %t, 0
@@ -1989,6 +3421,35 @@ define zeroext i1 @umulo2.br.i64(i64 %v1) {
 ; RV64-NEXT:  .LBB59_2: # %continue
 ; RV64-NEXT:    addi a0, zero, 1
 ; RV64-NEXT:    ret
+;
+; RV32ZBA-LABEL: umulo2.br.i64:
+; RV32ZBA:       # %bb.0: # %entry
+; RV32ZBA-NEXT:    add a2, a0, a0
+; RV32ZBA-NEXT:    sltu a0, a2, a0
+; RV32ZBA-NEXT:    add a2, a1, a1
+; RV32ZBA-NEXT:    add a2, a2, a0
+; RV32ZBA-NEXT:    beq a2, a1, .LBB59_2
+; RV32ZBA-NEXT:  # %bb.1: # %entry
+; RV32ZBA-NEXT:    sltu a0, a2, a1
+; RV32ZBA-NEXT:  .LBB59_2: # %entry
+; RV32ZBA-NEXT:    beqz a0, .LBB59_4
+; RV32ZBA-NEXT:  # %bb.3: # %overflow
+; RV32ZBA-NEXT:    mv a0, zero
+; RV32ZBA-NEXT:    ret
+; RV32ZBA-NEXT:  .LBB59_4: # %continue
+; RV32ZBA-NEXT:    addi a0, zero, 1
+; RV32ZBA-NEXT:    ret
+;
+; RV64ZBA-LABEL: umulo2.br.i64:
+; RV64ZBA:       # %bb.0: # %entry
+; RV64ZBA-NEXT:    add a1, a0, a0
+; RV64ZBA-NEXT:    bgeu a1, a0, .LBB59_2
+; RV64ZBA-NEXT:  # %bb.1: # %overflow
+; RV64ZBA-NEXT:    mv a0, zero
+; RV64ZBA-NEXT:    ret
+; RV64ZBA-NEXT:  .LBB59_2: # %continue
+; RV64ZBA-NEXT:    addi a0, zero, 1
+; RV64ZBA-NEXT:    ret
 entry:
   %t = call {i64, i1} @llvm.umul.with.overflow.i64(i64 %v1, i64 2)
   %val = extractvalue {i64, i1} %t, 0