From 0d9f557b6c36da3aa92daff4c0d37ea821d7ae1e Mon Sep 17 00:00:00 2001
From: hev <wangrui@loongson.cn>
Date: Thu, 23 Nov 2023 19:34:50 +0800
Subject: [PATCH] [LoongArch] Disable mulodi4 and muloti4 libcalls (#73199)

This library function only exists in compiler-rt not libgcc. So this
would fail to link unless we were linking with compiler-rt.

Fixes https://github.com/ClangBuiltLinux/linux/issues/1958
---
 .../LoongArch/LoongArchISelLowering.cpp       |   5 +
 .../CodeGen/LoongArch/smul-with-overflow.ll   | 463 +++++++++++++++---
 2 files changed, 397 insertions(+), 71 deletions(-)

diff --git a/llvm/lib/Target/LoongArch/LoongArchISelLowering.cpp b/llvm/lib/Target/LoongArch/LoongArchISelLowering.cpp
index f3f72e74ef085..ac78789c2c331 100644
--- a/llvm/lib/Target/LoongArch/LoongArchISelLowering.cpp
+++ b/llvm/lib/Target/LoongArch/LoongArchISelLowering.cpp
@@ -152,8 +152,13 @@ LoongArchTargetLowering::LoongArchTargetLowering(const TargetMachine &TM,
 
     // Set libcalls.
     setLibcallName(RTLIB::MUL_I128, nullptr);
+    // The MULO libcall is not part of libgcc, only compiler-rt.
+    setLibcallName(RTLIB::MULO_I64, nullptr);
   }
 
+  // The MULO libcall is not part of libgcc, only compiler-rt.
+  setLibcallName(RTLIB::MULO_I128, nullptr);
+
   setOperationAction(ISD::ATOMIC_FENCE, MVT::Other, Custom);
 
   static const ISD::CondCode FPCCToExpand[] = {
diff --git a/llvm/test/CodeGen/LoongArch/smul-with-overflow.ll b/llvm/test/CodeGen/LoongArch/smul-with-overflow.ll
index a53e77e5aa4b3..6cba4108d63c6 100644
--- a/llvm/test/CodeGen/LoongArch/smul-with-overflow.ll
+++ b/llvm/test/CodeGen/LoongArch/smul-with-overflow.ll
@@ -5,23 +5,53 @@
 define zeroext i1 @smuloi64(i64 %v1, i64 %v2, ptr %res) {
 ; LA32-LABEL: smuloi64:
 ; LA32:       # %bb.0:
-; LA32-NEXT:    addi.w $sp, $sp, -16
-; LA32-NEXT:    .cfi_def_cfa_offset 16
-; LA32-NEXT:    st.w $ra, $sp, 12 # 4-byte Folded Spill
-; LA32-NEXT:    st.w $fp, $sp, 8 # 4-byte Folded Spill
-; LA32-NEXT:    .cfi_offset 1, -4
-; LA32-NEXT:    .cfi_offset 22, -8
-; LA32-NEXT:    move $fp, $a4
-; LA32-NEXT:    st.w $zero, $sp, 4
-; LA32-NEXT:    addi.w $a4, $sp, 4
-; LA32-NEXT:    bl %plt(__mulodi4)
-; LA32-NEXT:    st.w $a1, $fp, 4
-; LA32-NEXT:    st.w $a0, $fp, 0
-; LA32-NEXT:    ld.w $a0, $sp, 4
+; LA32-NEXT:    srai.w $a5, $a1, 31
+; LA32-NEXT:    mul.w $a6, $a2, $a5
+; LA32-NEXT:    mulh.wu $a7, $a2, $a5
+; LA32-NEXT:    add.w $a7, $a7, $a6
+; LA32-NEXT:    mul.w $a5, $a3, $a5
+; LA32-NEXT:    add.w $a5, $a7, $a5
+; LA32-NEXT:    srai.w $a7, $a3, 31
+; LA32-NEXT:    mul.w $t0, $a7, $a1
+; LA32-NEXT:    mulh.wu $t1, $a7, $a0
+; LA32-NEXT:    add.w $t0, $t1, $t0
+; LA32-NEXT:    mul.w $a7, $a7, $a0
+; LA32-NEXT:    add.w $t0, $t0, $a7
+; LA32-NEXT:    add.w $a5, $t0, $a5
+; LA32-NEXT:    mulh.wu $t0, $a0, $a2
+; LA32-NEXT:    mul.w $t1, $a1, $a2
+; LA32-NEXT:    add.w $t0, $t1, $t0
+; LA32-NEXT:    sltu $t1, $t0, $t1
+; LA32-NEXT:    mulh.wu $t2, $a1, $a2
+; LA32-NEXT:    add.w $t1, $t2, $t1
+; LA32-NEXT:    mul.w $t2, $a0, $a3
+; LA32-NEXT:    add.w $t0, $t2, $t0
+; LA32-NEXT:    sltu $t2, $t0, $t2
+; LA32-NEXT:    mulh.wu $t3, $a0, $a3
+; LA32-NEXT:    add.w $t2, $t3, $t2
+; LA32-NEXT:    add.w $a6, $a7, $a6
+; LA32-NEXT:    sltu $a7, $a6, $a7
+; LA32-NEXT:    add.w $a5, $a5, $a7
+; LA32-NEXT:    mul.w $a0, $a0, $a2
+; LA32-NEXT:    mul.w $a2, $a1, $a3
+; LA32-NEXT:    mulh.wu $a1, $a1, $a3
+; LA32-NEXT:    add.w $a3, $t1, $t2
+; LA32-NEXT:    sltu $a7, $a3, $t1
+; LA32-NEXT:    add.w $a1, $a1, $a7
+; LA32-NEXT:    st.w $a0, $a4, 0
+; LA32-NEXT:    add.w $a0, $a2, $a3
+; LA32-NEXT:    sltu $a2, $a0, $a2
+; LA32-NEXT:    add.w $a1, $a1, $a2
+; LA32-NEXT:    st.w $t0, $a4, 4
+; LA32-NEXT:    add.w $a1, $a1, $a5
+; LA32-NEXT:    add.w $a2, $a0, $a6
+; LA32-NEXT:    sltu $a0, $a2, $a0
+; LA32-NEXT:    add.w $a0, $a1, $a0
+; LA32-NEXT:    srai.w $a1, $t0, 31
+; LA32-NEXT:    xor $a0, $a0, $a1
+; LA32-NEXT:    xor $a1, $a2, $a1
+; LA32-NEXT:    or $a0, $a1, $a0
 ; LA32-NEXT:    sltu $a0, $zero, $a0
-; LA32-NEXT:    ld.w $fp, $sp, 8 # 4-byte Folded Reload
-; LA32-NEXT:    ld.w $ra, $sp, 12 # 4-byte Folded Reload
-; LA32-NEXT:    addi.w $sp, $sp, 16
 ; LA32-NEXT:    ret
 ;
 ; LA64-LABEL: smuloi64:
@@ -43,69 +73,360 @@ define zeroext i1 @smuloi64(i64 %v1, i64 %v2, ptr %res) {
 define zeroext i1 @smuloi128(i128 %v1, i128 %v2, ptr %res) {
 ; LA32-LABEL: smuloi128:
 ; LA32:       # %bb.0:
-; LA32-NEXT:    addi.w $sp, $sp, -64
-; LA32-NEXT:    .cfi_def_cfa_offset 64
-; LA32-NEXT:    st.w $ra, $sp, 60 # 4-byte Folded Spill
-; LA32-NEXT:    st.w $fp, $sp, 56 # 4-byte Folded Spill
+; LA32-NEXT:    addi.w $sp, $sp, -96
+; LA32-NEXT:    .cfi_def_cfa_offset 96
+; LA32-NEXT:    st.w $ra, $sp, 92 # 4-byte Folded Spill
+; LA32-NEXT:    st.w $fp, $sp, 88 # 4-byte Folded Spill
+; LA32-NEXT:    st.w $s0, $sp, 84 # 4-byte Folded Spill
+; LA32-NEXT:    st.w $s1, $sp, 80 # 4-byte Folded Spill
+; LA32-NEXT:    st.w $s2, $sp, 76 # 4-byte Folded Spill
+; LA32-NEXT:    st.w $s3, $sp, 72 # 4-byte Folded Spill
+; LA32-NEXT:    st.w $s4, $sp, 68 # 4-byte Folded Spill
+; LA32-NEXT:    st.w $s5, $sp, 64 # 4-byte Folded Spill
+; LA32-NEXT:    st.w $s6, $sp, 60 # 4-byte Folded Spill
+; LA32-NEXT:    st.w $s7, $sp, 56 # 4-byte Folded Spill
+; LA32-NEXT:    st.w $s8, $sp, 52 # 4-byte Folded Spill
 ; LA32-NEXT:    .cfi_offset 1, -4
 ; LA32-NEXT:    .cfi_offset 22, -8
-; LA32-NEXT:    move $fp, $a2
-; LA32-NEXT:    st.w $zero, $sp, 52
-; LA32-NEXT:    ld.w $a2, $a1, 12
-; LA32-NEXT:    st.w $a2, $sp, 12
-; LA32-NEXT:    ld.w $a2, $a1, 8
-; LA32-NEXT:    st.w $a2, $sp, 8
-; LA32-NEXT:    ld.w $a2, $a1, 4
-; LA32-NEXT:    st.w $a2, $sp, 4
-; LA32-NEXT:    ld.w $a1, $a1, 0
-; LA32-NEXT:    st.w $a1, $sp, 0
-; LA32-NEXT:    ld.w $a1, $a0, 12
-; LA32-NEXT:    st.w $a1, $sp, 28
-; LA32-NEXT:    ld.w $a1, $a0, 8
-; LA32-NEXT:    st.w $a1, $sp, 24
-; LA32-NEXT:    ld.w $a1, $a0, 4
-; LA32-NEXT:    st.w $a1, $sp, 20
-; LA32-NEXT:    ld.w $a0, $a0, 0
-; LA32-NEXT:    st.w $a0, $sp, 16
-; LA32-NEXT:    addi.w $a0, $sp, 32
-; LA32-NEXT:    addi.w $a1, $sp, 16
-; LA32-NEXT:    addi.w $a2, $sp, 0
-; LA32-NEXT:    addi.w $a3, $sp, 52
-; LA32-NEXT:    bl %plt(__muloti4)
-; LA32-NEXT:    ld.w $a0, $sp, 44
-; LA32-NEXT:    st.w $a0, $fp, 12
-; LA32-NEXT:    ld.w $a0, $sp, 40
-; LA32-NEXT:    st.w $a0, $fp, 8
-; LA32-NEXT:    ld.w $a0, $sp, 36
-; LA32-NEXT:    st.w $a0, $fp, 4
-; LA32-NEXT:    ld.w $a0, $sp, 32
-; LA32-NEXT:    st.w $a0, $fp, 0
-; LA32-NEXT:    ld.w $a0, $sp, 52
+; LA32-NEXT:    .cfi_offset 23, -12
+; LA32-NEXT:    .cfi_offset 24, -16
+; LA32-NEXT:    .cfi_offset 25, -20
+; LA32-NEXT:    .cfi_offset 26, -24
+; LA32-NEXT:    .cfi_offset 27, -28
+; LA32-NEXT:    .cfi_offset 28, -32
+; LA32-NEXT:    .cfi_offset 29, -36
+; LA32-NEXT:    .cfi_offset 30, -40
+; LA32-NEXT:    .cfi_offset 31, -44
+; LA32-NEXT:    st.w $a2, $sp, 12 # 4-byte Folded Spill
+; LA32-NEXT:    ld.w $a6, $a1, 0
+; LA32-NEXT:    ld.w $a7, $a0, 0
+; LA32-NEXT:    mulh.wu $a3, $a7, $a6
+; LA32-NEXT:    ld.w $a5, $a0, 4
+; LA32-NEXT:    mul.w $a4, $a5, $a6
+; LA32-NEXT:    add.w $a3, $a4, $a3
+; LA32-NEXT:    sltu $a4, $a3, $a4
+; LA32-NEXT:    mulh.wu $t0, $a5, $a6
+; LA32-NEXT:    add.w $a4, $t0, $a4
+; LA32-NEXT:    ld.w $t0, $a1, 4
+; LA32-NEXT:    mul.w $t1, $a7, $t0
+; LA32-NEXT:    add.w $a3, $t1, $a3
+; LA32-NEXT:    st.w $a3, $sp, 44 # 4-byte Folded Spill
+; LA32-NEXT:    sltu $t1, $a3, $t1
+; LA32-NEXT:    mulh.wu $t2, $a7, $t0
+; LA32-NEXT:    add.w $t1, $t2, $t1
+; LA32-NEXT:    ld.w $t4, $a0, 12
+; LA32-NEXT:    ld.w $t2, $a0, 8
+; LA32-NEXT:    ld.w $t3, $a1, 8
+; LA32-NEXT:    mulh.wu $a0, $t2, $t3
+; LA32-NEXT:    mul.w $t5, $t4, $t3
+; LA32-NEXT:    add.w $a0, $t5, $a0
+; LA32-NEXT:    sltu $t5, $a0, $t5
+; LA32-NEXT:    mulh.wu $t6, $t4, $t3
+; LA32-NEXT:    add.w $t5, $t6, $t5
+; LA32-NEXT:    ld.w $t7, $a1, 12
+; LA32-NEXT:    mul.w $a1, $t2, $t7
+; LA32-NEXT:    add.w $a0, $a1, $a0
+; LA32-NEXT:    st.w $a0, $sp, 48 # 4-byte Folded Spill
+; LA32-NEXT:    sltu $a1, $a0, $a1
+; LA32-NEXT:    mulh.wu $t6, $t2, $t7
+; LA32-NEXT:    add.w $t6, $t6, $a1
+; LA32-NEXT:    srai.w $s7, $t4, 31
+; LA32-NEXT:    mul.w $a1, $s7, $t7
+; LA32-NEXT:    mulh.wu $t8, $s7, $t3
+; LA32-NEXT:    add.w $t8, $t8, $a1
+; LA32-NEXT:    mulh.wu $fp, $a6, $s7
+; LA32-NEXT:    mul.w $s6, $t0, $s7
+; LA32-NEXT:    add.w $s8, $s6, $fp
+; LA32-NEXT:    mul.w $a1, $a6, $s7
+; LA32-NEXT:    add.w $ra, $a1, $s8
+; LA32-NEXT:    sltu $s0, $ra, $a1
+; LA32-NEXT:    add.w $a0, $fp, $s0
+; LA32-NEXT:    add.w $a3, $a4, $t1
+; LA32-NEXT:    st.w $a3, $sp, 20 # 4-byte Folded Spill
+; LA32-NEXT:    sltu $a4, $a3, $a4
+; LA32-NEXT:    mulh.wu $t1, $a5, $t0
+; LA32-NEXT:    add.w $a3, $t1, $a4
+; LA32-NEXT:    st.w $a3, $sp, 28 # 4-byte Folded Spill
+; LA32-NEXT:    srai.w $s4, $t7, 31
+; LA32-NEXT:    mul.w $fp, $a7, $s4
+; LA32-NEXT:    mulh.wu $a4, $a7, $s4
+; LA32-NEXT:    add.w $s1, $a4, $fp
+; LA32-NEXT:    sltu $s0, $s1, $fp
+; LA32-NEXT:    add.w $s5, $a4, $s0
+; LA32-NEXT:    mul.w $a4, $s7, $t3
+; LA32-NEXT:    add.w $t8, $t8, $a4
+; LA32-NEXT:    add.w $s0, $ra, $t8
+; LA32-NEXT:    add.w $a3, $a1, $a4
+; LA32-NEXT:    st.w $a3, $sp, 32 # 4-byte Folded Spill
+; LA32-NEXT:    sltu $a4, $a3, $a1
+; LA32-NEXT:    add.w $a3, $s0, $a4
+; LA32-NEXT:    st.w $a3, $sp, 24 # 4-byte Folded Spill
+; LA32-NEXT:    add.w $s3, $t5, $t6
+; LA32-NEXT:    sltu $a4, $s3, $t5
+; LA32-NEXT:    mulh.wu $t5, $t4, $t7
+; LA32-NEXT:    add.w $a3, $t5, $a4
+; LA32-NEXT:    st.w $a3, $sp, 16 # 4-byte Folded Spill
+; LA32-NEXT:    mul.w $a4, $a7, $a6
+; LA32-NEXT:    st.w $a4, $a2, 0
+; LA32-NEXT:    sltu $a4, $s8, $s6
+; LA32-NEXT:    mulh.wu $t5, $t0, $s7
+; LA32-NEXT:    add.w $a4, $t5, $a4
+; LA32-NEXT:    add.w $t1, $a4, $a0
+; LA32-NEXT:    sltu $a4, $t1, $a4
+; LA32-NEXT:    add.w $s2, $t5, $a4
+; LA32-NEXT:    mulh.wu $a4, $a7, $t3
+; LA32-NEXT:    mul.w $t5, $a5, $t3
+; LA32-NEXT:    add.w $a4, $t5, $a4
+; LA32-NEXT:    sltu $t5, $a4, $t5
+; LA32-NEXT:    mulh.wu $t6, $a5, $t3
+; LA32-NEXT:    add.w $a3, $t6, $t5
+; LA32-NEXT:    mul.w $t6, $a7, $t7
+; LA32-NEXT:    add.w $t5, $t6, $a4
+; LA32-NEXT:    sltu $a4, $t5, $t6
+; LA32-NEXT:    mulh.wu $t6, $a7, $t7
+; LA32-NEXT:    add.w $a4, $t6, $a4
+; LA32-NEXT:    mulh.wu $t6, $t2, $a6
+; LA32-NEXT:    mul.w $s7, $t4, $a6
+; LA32-NEXT:    add.w $t6, $s7, $t6
+; LA32-NEXT:    sltu $s7, $t6, $s7
+; LA32-NEXT:    mulh.wu $s8, $t4, $a6
+; LA32-NEXT:    add.w $a0, $s8, $s7
+; LA32-NEXT:    mul.w $s7, $t2, $t0
+; LA32-NEXT:    add.w $t6, $s7, $t6
+; LA32-NEXT:    sltu $s7, $t6, $s7
+; LA32-NEXT:    mulh.wu $s8, $t2, $t0
+; LA32-NEXT:    add.w $a2, $s8, $s7
+; LA32-NEXT:    mul.w $s8, $a5, $s4
+; LA32-NEXT:    add.w $s7, $s1, $s8
+; LA32-NEXT:    add.w $s1, $s7, $ra
+; LA32-NEXT:    add.w $a1, $fp, $a1
+; LA32-NEXT:    st.w $a1, $sp, 40 # 4-byte Folded Spill
+; LA32-NEXT:    sltu $ra, $a1, $fp
+; LA32-NEXT:    add.w $a1, $s1, $ra
+; LA32-NEXT:    st.w $a1, $sp, 36 # 4-byte Folded Spill
+; LA32-NEXT:    xor $s0, $a1, $s7
+; LA32-NEXT:    sltui $s0, $s0, 1
+; LA32-NEXT:    sltu $a1, $a1, $s7
+; LA32-NEXT:    masknez $s1, $a1, $s0
+; LA32-NEXT:    maskeqz $s0, $ra, $s0
+; LA32-NEXT:    add.w $t1, $s6, $t1
+; LA32-NEXT:    sltu $s6, $t1, $s6
+; LA32-NEXT:    add.w $s2, $s2, $s6
+; LA32-NEXT:    add.w $a2, $a0, $a2
+; LA32-NEXT:    sltu $a0, $a2, $a0
+; LA32-NEXT:    mulh.wu $s6, $t4, $t0
+; LA32-NEXT:    add.w $t8, $s6, $a0
+; LA32-NEXT:    add.w $a4, $a3, $a4
+; LA32-NEXT:    sltu $a3, $a4, $a3
+; LA32-NEXT:    mulh.wu $s6, $a5, $t7
+; LA32-NEXT:    add.w $a3, $s6, $a3
+; LA32-NEXT:    mul.w $s6, $t4, $t7
+; LA32-NEXT:    mul.w $t7, $a5, $t7
+; LA32-NEXT:    mul.w $ra, $t4, $t0
+; LA32-NEXT:    mul.w $t0, $a5, $t0
+; LA32-NEXT:    mul.w $t4, $t4, $s4
+; LA32-NEXT:    mul.w $a7, $a7, $t3
+; LA32-NEXT:    mul.w $a6, $t2, $a6
+; LA32-NEXT:    mul.w $t3, $t2, $t3
+; LA32-NEXT:    mul.w $a0, $t2, $s4
+; LA32-NEXT:    mulh.wu $t2, $t2, $s4
+; LA32-NEXT:    mulh.wu $a5, $s4, $a5
+; LA32-NEXT:    sltu $s4, $s7, $s8
+; LA32-NEXT:    add.w $s4, $a5, $s4
+; LA32-NEXT:    add.w $s4, $s5, $s4
+; LA32-NEXT:    sltu $s5, $s4, $s5
+; LA32-NEXT:    add.w $s5, $a5, $s5
+; LA32-NEXT:    ld.w $a1, $sp, 20 # 4-byte Folded Reload
+; LA32-NEXT:    add.w $a1, $t0, $a1
+; LA32-NEXT:    sltu $a5, $a1, $t0
+; LA32-NEXT:    ld.w $t0, $sp, 28 # 4-byte Folded Reload
+; LA32-NEXT:    add.w $t0, $t0, $a5
+; LA32-NEXT:    or $s0, $s0, $s1
+; LA32-NEXT:    add.w $a4, $t7, $a4
+; LA32-NEXT:    sltu $a5, $a4, $t7
+; LA32-NEXT:    add.w $t7, $a3, $a5
+; LA32-NEXT:    add.w $s1, $ra, $a2
+; LA32-NEXT:    sltu $a2, $s1, $ra
+; LA32-NEXT:    add.w $t8, $t8, $a2
+; LA32-NEXT:    add.w $a5, $s6, $s3
+; LA32-NEXT:    sltu $a2, $a5, $s6
+; LA32-NEXT:    ld.w $a3, $sp, 16 # 4-byte Folded Reload
+; LA32-NEXT:    add.w $a2, $a3, $a2
+; LA32-NEXT:    ld.w $s6, $sp, 12 # 4-byte Folded Reload
+; LA32-NEXT:    ld.w $a3, $sp, 44 # 4-byte Folded Reload
+; LA32-NEXT:    st.w $a3, $s6, 4
+; LA32-NEXT:    ld.w $a3, $sp, 24 # 4-byte Folded Reload
+; LA32-NEXT:    add.w $a3, $s2, $a3
+; LA32-NEXT:    ld.w $s2, $sp, 32 # 4-byte Folded Reload
+; LA32-NEXT:    add.w $s2, $t1, $s2
+; LA32-NEXT:    sltu $t1, $s2, $t1
+; LA32-NEXT:    add.w $a3, $a3, $t1
+; LA32-NEXT:    add.w $t1, $s8, $s4
+; LA32-NEXT:    sltu $s3, $t1, $s8
+; LA32-NEXT:    add.w $s3, $s5, $s3
+; LA32-NEXT:    add.w $t2, $t2, $a0
+; LA32-NEXT:    add.w $t2, $t2, $t4
+; LA32-NEXT:    add.w $t2, $t2, $s7
+; LA32-NEXT:    add.w $t4, $a0, $fp
+; LA32-NEXT:    sltu $a0, $t4, $a0
+; LA32-NEXT:    add.w $a0, $t2, $a0
+; LA32-NEXT:    add.w $a0, $s3, $a0
+; LA32-NEXT:    add.w $t2, $t1, $t4
+; LA32-NEXT:    sltu $t1, $t2, $t1
+; LA32-NEXT:    add.w $a0, $a0, $t1
+; LA32-NEXT:    add.w $a0, $a0, $a3
+; LA32-NEXT:    add.w $t1, $t2, $s2
+; LA32-NEXT:    sltu $a3, $t1, $t2
+; LA32-NEXT:    add.w $a0, $a0, $a3
+; LA32-NEXT:    add.w $a3, $t6, $t0
+; LA32-NEXT:    add.w $a1, $a6, $a1
+; LA32-NEXT:    sltu $a6, $a1, $a6
+; LA32-NEXT:    add.w $t0, $a3, $a6
+; LA32-NEXT:    add.w $a1, $a7, $a1
+; LA32-NEXT:    sltu $a7, $a1, $a7
+; LA32-NEXT:    add.w $a3, $t5, $t0
+; LA32-NEXT:    add.w $a3, $a3, $a7
+; LA32-NEXT:    sltu $t2, $a3, $t5
+; LA32-NEXT:    xor $t4, $a3, $t5
+; LA32-NEXT:    sltui $t4, $t4, 1
+; LA32-NEXT:    masknez $t2, $t2, $t4
+; LA32-NEXT:    maskeqz $a7, $a7, $t4
+; LA32-NEXT:    st.w $a1, $s6, 8
+; LA32-NEXT:    or $a1, $a7, $t2
+; LA32-NEXT:    sltu $a7, $t0, $t6
+; LA32-NEXT:    xor $t0, $t0, $t6
+; LA32-NEXT:    sltui $t0, $t0, 1
+; LA32-NEXT:    masknez $a7, $a7, $t0
+; LA32-NEXT:    maskeqz $a6, $a6, $t0
+; LA32-NEXT:    or $a6, $a6, $a7
+; LA32-NEXT:    add.w $a6, $s1, $a6
+; LA32-NEXT:    sltu $a7, $a6, $s1
+; LA32-NEXT:    add.w $a7, $t8, $a7
+; LA32-NEXT:    add.w $a1, $a4, $a1
+; LA32-NEXT:    sltu $a4, $a1, $a4
+; LA32-NEXT:    add.w $a4, $t7, $a4
+; LA32-NEXT:    add.w $t0, $t1, $s0
+; LA32-NEXT:    sltu $t1, $t0, $t1
+; LA32-NEXT:    add.w $a0, $a0, $t1
+; LA32-NEXT:    st.w $a3, $s6, 12
+; LA32-NEXT:    add.w $a1, $a6, $a1
+; LA32-NEXT:    sltu $a6, $a1, $a6
+; LA32-NEXT:    add.w $a4, $a7, $a4
+; LA32-NEXT:    add.w $a4, $a4, $a6
+; LA32-NEXT:    sltu $t1, $a4, $a7
+; LA32-NEXT:    xor $a7, $a4, $a7
+; LA32-NEXT:    sltui $a7, $a7, 1
+; LA32-NEXT:    masknez $t1, $t1, $a7
+; LA32-NEXT:    maskeqz $a6, $a6, $a7
+; LA32-NEXT:    or $a6, $a6, $t1
+; LA32-NEXT:    add.w $a6, $a5, $a6
+; LA32-NEXT:    sltu $a5, $a6, $a5
+; LA32-NEXT:    add.w $a2, $a2, $a5
+; LA32-NEXT:    ld.w $t1, $sp, 48 # 4-byte Folded Reload
+; LA32-NEXT:    add.w $a4, $t1, $a4
+; LA32-NEXT:    add.w $a1, $t3, $a1
+; LA32-NEXT:    sltu $a5, $a1, $t3
+; LA32-NEXT:    add.w $a4, $a4, $a5
+; LA32-NEXT:    sltu $a7, $a4, $t1
+; LA32-NEXT:    xor $t1, $a4, $t1
+; LA32-NEXT:    sltui $t1, $t1, 1
+; LA32-NEXT:    masknez $a7, $a7, $t1
+; LA32-NEXT:    maskeqz $a5, $a5, $t1
+; LA32-NEXT:    or $a5, $a5, $a7
+; LA32-NEXT:    add.w $a5, $a6, $a5
+; LA32-NEXT:    sltu $a6, $a5, $a6
+; LA32-NEXT:    add.w $a2, $a2, $a6
+; LA32-NEXT:    add.w $a0, $a2, $a0
+; LA32-NEXT:    add.w $a2, $a5, $t0
+; LA32-NEXT:    sltu $a5, $a2, $a5
+; LA32-NEXT:    add.w $a0, $a0, $a5
+; LA32-NEXT:    ld.w $a5, $sp, 40 # 4-byte Folded Reload
+; LA32-NEXT:    add.w $a5, $a1, $a5
+; LA32-NEXT:    sltu $a1, $a5, $a1
+; LA32-NEXT:    ld.w $a6, $sp, 36 # 4-byte Folded Reload
+; LA32-NEXT:    add.w $a6, $a4, $a6
+; LA32-NEXT:    add.w $a6, $a6, $a1
+; LA32-NEXT:    sltu $a7, $a6, $a4
+; LA32-NEXT:    xor $a4, $a6, $a4
+; LA32-NEXT:    sltui $a4, $a4, 1
+; LA32-NEXT:    masknez $a7, $a7, $a4
+; LA32-NEXT:    maskeqz $a1, $a1, $a4
+; LA32-NEXT:    or $a1, $a1, $a7
+; LA32-NEXT:    add.w $a1, $a2, $a1
+; LA32-NEXT:    sltu $a2, $a1, $a2
+; LA32-NEXT:    add.w $a0, $a0, $a2
+; LA32-NEXT:    srai.w $a2, $a3, 31
+; LA32-NEXT:    xor $a3, $a6, $a2
+; LA32-NEXT:    xor $a0, $a0, $a2
+; LA32-NEXT:    or $a0, $a3, $a0
+; LA32-NEXT:    xor $a3, $a5, $a2
+; LA32-NEXT:    xor $a1, $a1, $a2
+; LA32-NEXT:    or $a1, $a3, $a1
+; LA32-NEXT:    or $a0, $a1, $a0
 ; LA32-NEXT:    sltu $a0, $zero, $a0
-; LA32-NEXT:    ld.w $fp, $sp, 56 # 4-byte Folded Reload
-; LA32-NEXT:    ld.w $ra, $sp, 60 # 4-byte Folded Reload
-; LA32-NEXT:    addi.w $sp, $sp, 64
+; LA32-NEXT:    ld.w $s8, $sp, 52 # 4-byte Folded Reload
+; LA32-NEXT:    ld.w $s7, $sp, 56 # 4-byte Folded Reload
+; LA32-NEXT:    ld.w $s6, $sp, 60 # 4-byte Folded Reload
+; LA32-NEXT:    ld.w $s5, $sp, 64 # 4-byte Folded Reload
+; LA32-NEXT:    ld.w $s4, $sp, 68 # 4-byte Folded Reload
+; LA32-NEXT:    ld.w $s3, $sp, 72 # 4-byte Folded Reload
+; LA32-NEXT:    ld.w $s2, $sp, 76 # 4-byte Folded Reload
+; LA32-NEXT:    ld.w $s1, $sp, 80 # 4-byte Folded Reload
+; LA32-NEXT:    ld.w $s0, $sp, 84 # 4-byte Folded Reload
+; LA32-NEXT:    ld.w $fp, $sp, 88 # 4-byte Folded Reload
+; LA32-NEXT:    ld.w $ra, $sp, 92 # 4-byte Folded Reload
+; LA32-NEXT:    addi.w $sp, $sp, 96
 ; LA32-NEXT:    ret
 ;
 ; LA64-LABEL: smuloi128:
 ; LA64:       # %bb.0:
-; LA64-NEXT:    addi.d $sp, $sp, -32
-; LA64-NEXT:    .cfi_def_cfa_offset 32
-; LA64-NEXT:    st.d $ra, $sp, 24 # 8-byte Folded Spill
-; LA64-NEXT:    st.d $fp, $sp, 16 # 8-byte Folded Spill
-; LA64-NEXT:    .cfi_offset 1, -8
-; LA64-NEXT:    .cfi_offset 22, -16
-; LA64-NEXT:    move $fp, $a4
-; LA64-NEXT:    st.d $zero, $sp, 8
-; LA64-NEXT:    addi.d $a4, $sp, 8
-; LA64-NEXT:    bl %plt(__muloti4)
-; LA64-NEXT:    st.d $a1, $fp, 8
-; LA64-NEXT:    st.d $a0, $fp, 0
-; LA64-NEXT:    ld.d $a0, $sp, 8
+; LA64-NEXT:    srai.d $a5, $a1, 63
+; LA64-NEXT:    mul.d $a6, $a2, $a5
+; LA64-NEXT:    mulh.du $a7, $a2, $a5
+; LA64-NEXT:    add.d $a7, $a7, $a6
+; LA64-NEXT:    mul.d $a5, $a3, $a5
+; LA64-NEXT:    add.d $a5, $a7, $a5
+; LA64-NEXT:    srai.d $a7, $a3, 63
+; LA64-NEXT:    mul.d $t0, $a7, $a1
+; LA64-NEXT:    mulh.du $t1, $a7, $a0
+; LA64-NEXT:    add.d $t0, $t1, $t0
+; LA64-NEXT:    mul.d $a7, $a7, $a0
+; LA64-NEXT:    add.d $t0, $t0, $a7
+; LA64-NEXT:    add.d $a5, $t0, $a5
+; LA64-NEXT:    mulh.du $t0, $a0, $a2
+; LA64-NEXT:    mul.d $t1, $a1, $a2
+; LA64-NEXT:    add.d $t0, $t1, $t0
+; LA64-NEXT:    sltu $t1, $t0, $t1
+; LA64-NEXT:    mulh.du $t2, $a1, $a2
+; LA64-NEXT:    add.d $t1, $t2, $t1
+; LA64-NEXT:    mul.d $t2, $a0, $a3
+; LA64-NEXT:    add.d $t0, $t2, $t0
+; LA64-NEXT:    sltu $t2, $t0, $t2
+; LA64-NEXT:    mulh.du $t3, $a0, $a3
+; LA64-NEXT:    add.d $t2, $t3, $t2
+; LA64-NEXT:    add.d $a6, $a7, $a6
+; LA64-NEXT:    sltu $a7, $a6, $a7
+; LA64-NEXT:    add.d $a5, $a5, $a7
+; LA64-NEXT:    mul.d $a0, $a0, $a2
+; LA64-NEXT:    mul.d $a2, $a1, $a3
+; LA64-NEXT:    mulh.du $a1, $a1, $a3
+; LA64-NEXT:    add.d $a3, $t1, $t2
+; LA64-NEXT:    sltu $a7, $a3, $t1
+; LA64-NEXT:    add.d $a1, $a1, $a7
+; LA64-NEXT:    st.d $a0, $a4, 0
+; LA64-NEXT:    add.d $a0, $a2, $a3
+; LA64-NEXT:    sltu $a2, $a0, $a2
+; LA64-NEXT:    add.d $a1, $a1, $a2
+; LA64-NEXT:    st.d $t0, $a4, 8
+; LA64-NEXT:    add.d $a1, $a1, $a5
+; LA64-NEXT:    add.d $a2, $a0, $a6
+; LA64-NEXT:    sltu $a0, $a2, $a0
+; LA64-NEXT:    add.d $a0, $a1, $a0
+; LA64-NEXT:    srai.d $a1, $t0, 63
+; LA64-NEXT:    xor $a0, $a0, $a1
+; LA64-NEXT:    xor $a1, $a2, $a1
+; LA64-NEXT:    or $a0, $a1, $a0
 ; LA64-NEXT:    sltu $a0, $zero, $a0
-; LA64-NEXT:    ld.d $fp, $sp, 16 # 8-byte Folded Reload
-; LA64-NEXT:    ld.d $ra, $sp, 24 # 8-byte Folded Reload
-; LA64-NEXT:    addi.d $sp, $sp, 32
 ; LA64-NEXT:    ret
   %t = call {i128, i1} @llvm.smul.with.overflow.i128(i128 %v1, i128 %v2)
   %val = extractvalue {i128, i1} %t, 0