From 38fcdfa0da5fc3a405a79880c7b3322f8764ce23 Mon Sep 17 00:00:00 2001
From: Ying Huang <ying.huang@oss.cipunited.com>
Date: Thu, 26 Sep 2024 22:39:13 -0400
Subject: [PATCH] [MIPS] Use softPromoteHalf legalization for fp16 rather than
 PromoteFloat

Fix part of #97975.
---
 llvm/lib/Target/Mips/MipsISelLowering.h |   2 +
 llvm/test/CodeGen/Mips/fp16-promote.ll  | 438 ++++++++++++++++++++++--
 2 files changed, 411 insertions(+), 29 deletions(-)

diff --git a/llvm/lib/Target/Mips/MipsISelLowering.h b/llvm/lib/Target/Mips/MipsISelLowering.h
index 2b18b299180926..8033898091c756 100644
--- a/llvm/lib/Target/Mips/MipsISelLowering.h
+++ b/llvm/lib/Target/Mips/MipsISelLowering.h
@@ -365,6 +365,8 @@ class TargetRegisterClass;
       return ABI.IsN64() ? Mips::A1_64 : Mips::A1;
     }
 
+    bool softPromoteHalfType() const override { return true; }
+
     bool isJumpTableRelative() const override {
       return getTargetMachine().isPositionIndependent();
     }
diff --git a/llvm/test/CodeGen/Mips/fp16-promote.ll b/llvm/test/CodeGen/Mips/fp16-promote.ll
index 6cbfab5f776698..15bb32e201cdb1 100644
--- a/llvm/test/CodeGen/Mips/fp16-promote.ll
+++ b/llvm/test/CodeGen/Mips/fp16-promote.ll
@@ -1,10 +1,46 @@
 ; RUN: llc -asm-verbose=false -mtriple=mipsel-linux-gnu -relocation-model=pic < %s | FileCheck %s -check-prefix=CHECK-LIBCALL
 
 ; CHECK-LIBCALL-LABEL: test_fadd:
-; CHECK-LIBCALL: %call16(__gnu_h2f_ieee)
-; CHECK-LIBCALL: %call16(__gnu_h2f_ieee)
-; CHECK-LIBCALL-DAG: add.s
-; CHECK-LIBCALL-DAG: %call16(__gnu_f2h_ieee)
+; CHECK-LIBCALL: lui $2, %hi(_gp_disp)
+; CHECK-LIBCALL-NEXT: addiu $2, $2, %lo(_gp_disp)
+; CHECK-LIBCALL-NEXT: addiu $sp, $sp, -40
+; CHECK-LIBCALL-NEXT: .cfi_def_cfa_offset 40
+; CHECK-LIBCALL-NEXT: sdc1 $f20, 32($sp)
+; CHECK-LIBCALL-NEXT: sw $ra, 28($sp)
+; CHECK-LIBCALL-NEXT: sw $17, 24($sp)
+; CHECK-LIBCALL-NEXT: sw $16, 20($sp)
+; CHECK-LIBCALL-NEXT: .cfi_offset 52, -8
+; CHECK-LIBCALL-NEXT: .cfi_offset 53, -4
+; CHECK-LIBCALL-NEXT: .cfi_offset 31, -12
+; CHECK-LIBCALL-NEXT: .cfi_offset 17, -16
+; CHECK-LIBCALL-NEXT: .cfi_offset 16, -20
+; CHECK-LIBCALL-NEXT: addu $16, $2, $25
+; CHECK-LIBCALL-NEXT: move $17, $4
+; CHECK-LIBCALL-NEXT: lhu $4, 0($5)
+; CHECK-LIBCALL-NEXT: lw $25, %call16(__gnu_h2f_ieee)($16)
+; CHECK-LIBCALL-NEXT: .reloc ($tmp0), R_MIPS_JALR, __gnu_h2f_ieee
+; CHECK-LIBCALL-NEXT: $tmp0:
+; CHECK-LIBCALL-NEXT: jalr $25
+; CHECK-LIBCALL-NEXT: move $gp, $16
+; CHECK-LIBCALL-NEXT: lhu $4, 0($17)
+; CHECK-LIBCALL-NEXT: lw $25, %call16(__gnu_h2f_ieee)($16)
+; CHECK-LIBCALL-NEXT: .reloc ($tmp1), R_MIPS_JALR, __gnu_h2f_ieee
+; CHECK-LIBCALL-NEXT: $tmp1:
+; CHECK-LIBCALL-NEXT: jalr $25
+; CHECK-LIBCALL-NEXT: mov.s $f20, $f0
+; CHECK-LIBCALL-NEXT: add.s $f12, $f0, $f20
+; CHECK-LIBCALL-NEXT: lw $25, %call16(__gnu_f2h_ieee)($16)
+; CHECK-LIBCALL-NEXT: .reloc ($tmp2), R_MIPS_JALR, __gnu_f2h_ieee
+; CHECK-LIBCALL-NEXT: $tmp2:
+; CHECK-LIBCALL-NEXT: jalr $25
+; CHECK-LIBCALL-NEXT: move $gp, $16
+; CHECK-LIBCALL-NEXT: sh $2, 0($17)
+; CHECK-LIBCALL-NEXT: lw $16, 20($sp)
+; CHECK-LIBCALL-NEXT: lw $17, 24($sp)
+; CHECK-LIBCALL-NEXT: lw $ra, 28($sp)
+; CHECK-LIBCALL-NEXT: ldc1 $f20, 32($sp)
+; CHECK-LIBCALL-NEXT: jr $ra
+; CHECK-LIBCALL-NEXT: addiu $sp, $sp, 40
 define void @test_fadd(ptr %p, ptr %q) #0 {
   %a = load half, ptr %p, align 2
   %b = load half, ptr %q, align 2
@@ -14,7 +50,22 @@ define void @test_fadd(ptr %p, ptr %q) #0 {
 }
 
 ; CHECK-LIBCALL-LABEL: test_fpext_float:
-; CHECK-LIBCALL: %call16(__gnu_h2f_ieee)
+; CHECK-LIBCALL: lui $2, %hi(_gp_disp)
+; CHECK-LIBCALL-NEXT: addiu $2, $2, %lo(_gp_disp)
+; CHECK-LIBCALL-NEXT: addiu $sp, $sp, -24
+; CHECK-LIBCALL-NEXT: .cfi_def_cfa_offset 24
+; CHECK-LIBCALL-NEXT: sw $ra, 20($sp)
+; CHECK-LIBCALL-NEXT: .cfi_offset 31, -4
+; CHECK-LIBCALL-NEXT: addu $gp, $2, $25
+; CHECK-LIBCALL-NEXT: lhu $4, 0($4)
+; CHECK-LIBCALL-NEXT: lw $25, %call16(__gnu_h2f_ieee)($gp)
+; CHECK-LIBCALL-NEXT: .reloc ($tmp3), R_MIPS_JALR, __gnu_h2f_ieee
+; CHECK-LIBCALL-NEXT: $tmp3:
+; CHECK-LIBCALL-NEXT: jalr $25
+; CHECK-LIBCALL-NEXT: nop
+; CHECK-LIBCALL-NEXT: lw $ra, 20($sp)
+; CHECK-LIBCALL-NEXT: jr $ra
+; CHECK-LIBCALL-NEXT: addiu $sp, $sp, 24
 define float @test_fpext_float(ptr %p) {
   %a = load half, ptr %p, align 2
   %r = fpext half %a to float
@@ -22,8 +73,23 @@ define float @test_fpext_float(ptr %p) {
 }
 
 ; CHECK-LIBCALL-LABEL: test_fpext_double:
-; CHECK-LIBCALL: %call16(__gnu_h2f_ieee)
-; CHECK-LIBCALL: cvt.d.s
+; CHECK-LIBCALL: lui $2, %hi(_gp_disp)
+; CHECK-LIBCALL-NEXT: addiu $2, $2, %lo(_gp_disp)
+; CHECK-LIBCALL-NEXT: addiu $sp, $sp, -24
+; CHECK-LIBCALL-NEXT: .cfi_def_cfa_offset 24
+; CHECK-LIBCALL-NEXT: sw $ra, 20($sp)
+; CHECK-LIBCALL-NEXT: .cfi_offset 31, -4
+; CHECK-LIBCALL-NEXT: addu $gp, $2, $25
+; CHECK-LIBCALL-NEXT: lhu $4, 0($4)
+; CHECK-LIBCALL-NEXT: lw $25, %call16(__gnu_h2f_ieee)($gp)
+; CHECK-LIBCALL-NEXT: .reloc ($tmp4), R_MIPS_JALR, __gnu_h2f_ieee
+; CHECK-LIBCALL-NEXT: $tmp4:
+; CHECK-LIBCALL-NEXT: jalr $25
+; CHECK-LIBCALL-NEXT: nop
+; CHECK-LIBCALL-NEXT: cvt.d.s $f0, $f0
+; CHECK-LIBCALL-NEXT: lw $ra, 20($sp)
+; CHECK-LIBCALL-NEXT: jr $ra
+; CHECK-LIBCALL-NEXT: addiu $sp, $sp, 24
 define double @test_fpext_double(ptr %p) {
   %a = load half, ptr %p, align 2
   %r = fpext half %a to double
@@ -31,7 +97,25 @@ define double @test_fpext_double(ptr %p) {
 }
 
 ; CHECK-LIBCALL-LABEL: test_fptrunc_float:
-; CHECK-LIBCALL: %call16(__gnu_f2h_ieee)
+; CHECK-LIBCALL: lui $2, %hi(_gp_disp)
+; CHECK-LIBCALL-NEXT: addiu $2, $2, %lo(_gp_disp)
+; CHECK-LIBCALL-NEXT: addiu $sp, $sp, -24
+; CHECK-LIBCALL-NEXT: .cfi_def_cfa_offset 24
+; CHECK-LIBCALL-NEXT: sw $ra, 20($sp)
+; CHECK-LIBCALL-NEXT: sw $16, 16($sp)
+; CHECK-LIBCALL-NEXT: .cfi_offset 31, -4
+; CHECK-LIBCALL-NEXT: .cfi_offset 16, -8
+; CHECK-LIBCALL-NEXT: addu $gp, $2, $25
+; CHECK-LIBCALL-NEXT: lw $25, %call16(__gnu_f2h_ieee)($gp)
+; CHECK-LIBCALL-NEXT: .reloc ($tmp5), R_MIPS_JALR, __gnu_f2h_ieee
+; CHECK-LIBCALL-NEXT: $tmp5:
+; CHECK-LIBCALL-NEXT: jalr $25
+; CHECK-LIBCALL-NEXT: move $16, $5
+; CHECK-LIBCALL-NEXT: sh $2, 0($16)
+; CHECK-LIBCALL-NEXT: lw $16, 16($sp)
+; CHECK-LIBCALL-NEXT: lw $ra, 20($sp)
+; CHECK-LIBCALL-NEXT: jr $ra
+; CHECK-LIBCALL-NEXT: addiu $sp, $sp, 24
 define void @test_fptrunc_float(float %f, ptr %p) #0 {
   %a = fptrunc float %f to half
   store half %a, ptr %p
@@ -39,7 +123,25 @@ define void @test_fptrunc_float(float %f, ptr %p) #0 {
 }
 
 ; CHECK-LIBCALL-LABEL: test_fptrunc_double:
-; CHECK-LIBCALL: %call16(__truncdfhf2)
+; CHECK-LIBCALL: lui $2, %hi(_gp_disp)
+; CHECK-LIBCALL-NEXT: addiu $2, $2, %lo(_gp_disp)
+; CHECK-LIBCALL-NEXT: addiu $sp, $sp, -24
+; CHECK-LIBCALL-NEXT: .cfi_def_cfa_offset 24
+; CHECK-LIBCALL-NEXT: sw $ra, 20($sp)
+; CHECK-LIBCALL-NEXT: sw $16, 16($sp)
+; CHECK-LIBCALL-NEXT: .cfi_offset 31, -4
+; CHECK-LIBCALL-NEXT: .cfi_offset 16, -8
+; CHECK-LIBCALL-NEXT: addu $gp, $2, $25
+; CHECK-LIBCALL-NEXT: lw $25, %call16(__truncdfhf2)($gp)
+; CHECK-LIBCALL-NEXT: .reloc ($tmp6), R_MIPS_JALR, __truncdfhf2
+; CHECK-LIBCALL-NEXT: $tmp6:
+; CHECK-LIBCALL-NEXT: jalr $25
+; CHECK-LIBCALL-NEXT: move $16, $6
+; CHECK-LIBCALL-NEXT: sh $2, 0($16)
+; CHECK-LIBCALL-NEXT: lw $16, 16($sp)
+; CHECK-LIBCALL-NEXT: lw $ra, 20($sp)
+; CHECK-LIBCALL-NEXT: jr $ra
+; CHECK-LIBCALL-NEXT: addiu $sp, $sp, 24
 define void @test_fptrunc_double(double %d, ptr %p) #0 {
   %a = fptrunc double %d to half
   store half %a, ptr %p
@@ -47,10 +149,58 @@ define void @test_fptrunc_double(double %d, ptr %p) #0 {
 }
 
 ; CHECK-LIBCALL-LABEL: test_vec_fpext_float:
-; CHECK-LIBCALL: %call16(__gnu_h2f_ieee)
-; CHECK-LIBCALL: %call16(__gnu_h2f_ieee)
-; CHECK-LIBCALL: %call16(__gnu_h2f_ieee)
-; CHECK-LIBCALL: %call16(__gnu_h2f_ieee)
+; CHECK-LIBCALL: lui $2, %hi(_gp_disp)
+; CHECK-LIBCALL-NEXT: addiu $2, $2, %lo(_gp_disp)
+; CHECK-LIBCALL-NEXT: addiu $sp, $sp, -40
+; CHECK-LIBCALL-NEXT: .cfi_def_cfa_offset 40
+; CHECK-LIBCALL-NEXT: sdc1 $f20, 32($sp)
+; CHECK-LIBCALL-NEXT: sw $ra, 28($sp)
+; CHECK-LIBCALL-NEXT: sw $18, 24($sp)
+; CHECK-LIBCALL-NEXT: sw $17, 20($sp)
+; CHECK-LIBCALL-NEXT: sw $16, 16($sp)
+; CHECK-LIBCALL-NEXT: .cfi_offset 52, -8
+; CHECK-LIBCALL-NEXT: .cfi_offset 53, -4
+; CHECK-LIBCALL-NEXT: .cfi_offset 31, -12
+; CHECK-LIBCALL-NEXT: .cfi_offset 18, -16
+; CHECK-LIBCALL-NEXT: .cfi_offset 17, -20
+; CHECK-LIBCALL-NEXT: .cfi_offset 16, -24
+; CHECK-LIBCALL-NEXT: addu $16, $2, $25
+; CHECK-LIBCALL-NEXT: move $17, $5
+; CHECK-LIBCALL-NEXT: move $18, $4
+; CHECK-LIBCALL-NEXT: lhu $4, 4($5)
+; CHECK-LIBCALL-NEXT: lw $25, %call16(__gnu_h2f_ieee)($16)
+; CHECK-LIBCALL-NEXT: .reloc ($tmp7), R_MIPS_JALR, __gnu_h2f_ieee
+; CHECK-LIBCALL-NEXT: $tmp7:
+; CHECK-LIBCALL-NEXT: jalr $25
+; CHECK-LIBCALL-NEXT: move $gp, $16
+; CHECK-LIBCALL-NEXT: lhu $4, 6($17)
+; CHECK-LIBCALL-NEXT: lw $25, %call16(__gnu_h2f_ieee)($16)
+; CHECK-LIBCALL-NEXT: .reloc ($tmp8), R_MIPS_JALR, __gnu_h2f_ieee
+; CHECK-LIBCALL-NEXT: $tmp8:
+; CHECK-LIBCALL-NEXT: jalr $25
+; CHECK-LIBCALL-NEXT: mov.s $f20, $f0
+; CHECK-LIBCALL-NEXT: lhu $4, 2($17)
+; CHECK-LIBCALL-NEXT: lw $25, %call16(__gnu_h2f_ieee)($16)
+; CHECK-LIBCALL-NEXT: swc1 $f0, 12($18)
+; CHECK-LIBCALL-NEXT: .reloc ($tmp9), R_MIPS_JALR, __gnu_h2f_ieee
+; CHECK-LIBCALL-NEXT: $tmp9:
+; CHECK-LIBCALL-NEXT: jalr $25
+; CHECK-LIBCALL-NEXT: swc1 $f20, 8($18)
+; CHECK-LIBCALL-NEXT: swc1 $f0, 4($18)
+; CHECK-LIBCALL-NEXT: lhu $4, 0($17)
+; CHECK-LIBCALL-NEXT: lw $25, %call16(__gnu_h2f_ieee)($16)
+; CHECK-LIBCALL-NEXT: .reloc ($tmp10), R_MIPS_JALR, __gnu_h2f_ieee
+; CHECK-LIBCALL-NEXT: $tmp10:
+; CHECK-LIBCALL-NEXT: jalr $25
+; CHECK-LIBCALL-NEXT: nop
+; CHECK-LIBCALL-NEXT: swc1 $f0, 0($18)
+; CHECK-LIBCALL-NEXT: lw $16, 16($sp)
+; CHECK-LIBCALL-NEXT: lw $17, 20($sp)
+; CHECK-LIBCALL-NEXT: lw $18, 24($sp)
+; CHECK-LIBCALL-NEXT: lw $ra, 28($sp)
+; CHECK-LIBCALL-NEXT: ldc1 $f20, 32($sp)
+; CHECK-LIBCALL-NEXT: jr $ra
+; CHECK-LIBCALL-NEXT: addiu $sp, $sp, 40
 define <4 x float> @test_vec_fpext_float(ptr %p) #0 {
   %a = load <4 x half>, ptr %p, align 8
   %b = fpext <4 x half> %a to <4 x float>
@@ -60,14 +210,65 @@ define <4 x float> @test_vec_fpext_float(ptr %p) #0 {
 ; This test is not robust against variations in instruction scheduling.
 ; See the discussion in http://reviews.llvm.org/D8804
 ; CHECK-LIBCALL-LABEL: test_vec_fpext_double:
-; CHECK-LIBCALL: %call16(__gnu_h2f_ieee)
-; CHECK-LIBCALL: %call16(__gnu_h2f_ieee)
-; CHECK-LIBCALL: %call16(__gnu_h2f_ieee)
-; CHECK-LIBCALL: cvt.d.s
-; CHECK-LIBCALL: cvt.d.s
-; CHECK-LIBCALL: cvt.d.s
-; CHECK-LIBCALL: %call16(__gnu_h2f_ieee)
-; CHECK-LIBCALL: cvt.d.s
+; CHECK-LIBCALL: lui $2, %hi(_gp_disp)
+; CHECK-LIBCALL-NEXT: addiu $2, $2, %lo(_gp_disp)
+; CHECK-LIBCALL-NEXT: addiu $sp, $sp, -48
+; CHECK-LIBCALL-NEXT: .cfi_def_cfa_offset 48
+; CHECK-LIBCALL-NEXT: sdc1 $f22, 40($sp)
+; CHECK-LIBCALL-NEXT: sdc1 $f20, 32($sp)
+; CHECK-LIBCALL-NEXT: sw $ra, 28($sp)
+; CHECK-LIBCALL-NEXT: sw $18, 24($sp)
+; CHECK-LIBCALL-NEXT: sw $17, 20($sp)
+; CHECK-LIBCALL-NEXT: sw $16, 16($sp)
+; CHECK-LIBCALL-NEXT: .cfi_offset 54, -8
+; CHECK-LIBCALL-NEXT: .cfi_offset 55, -4
+; CHECK-LIBCALL-NEXT: .cfi_offset 52, -16
+; CHECK-LIBCALL-NEXT: .cfi_offset 53, -12
+; CHECK-LIBCALL-NEXT: .cfi_offset 31, -20
+; CHECK-LIBCALL-NEXT: .cfi_offset 18, -24
+; CHECK-LIBCALL-NEXT: .cfi_offset 17, -28
+; CHECK-LIBCALL-NEXT: .cfi_offset 16, -32
+; CHECK-LIBCALL-NEXT: addu $16, $2, $25
+; CHECK-LIBCALL-NEXT: move $17, $5
+; CHECK-LIBCALL-NEXT: move $18, $4
+; CHECK-LIBCALL-NEXT: lhu $4, 2($5)
+; CHECK-LIBCALL-NEXT: lw $25, %call16(__gnu_h2f_ieee)($16)
+; CHECK-LIBCALL-NEXT: .reloc ($tmp11), R_MIPS_JALR, __gnu_h2f_ieee
+; CHECK-LIBCALL-NEXT: $tmp11:
+; CHECK-LIBCALL-NEXT: jalr $25
+; CHECK-LIBCALL-NEXT: move $gp, $16
+; CHECK-LIBCALL-NEXT: lhu $4, 4($17)
+; CHECK-LIBCALL-NEXT: lw $25, %call16(__gnu_h2f_ieee)($16)
+; CHECK-LIBCALL-NEXT: .reloc ($tmp12), R_MIPS_JALR, __gnu_h2f_ieee
+; CHECK-LIBCALL-NEXT: $tmp12:
+; CHECK-LIBCALL-NEXT: jalr $25
+; CHECK-LIBCALL-NEXT: cvt.d.s $f20, $f0
+; CHECK-LIBCALL-NEXT: lhu $4, 6($17)
+; CHECK-LIBCALL-NEXT: lw $25, %call16(__gnu_h2f_ieee)($16)
+; CHECK-LIBCALL-NEXT: .reloc ($tmp13), R_MIPS_JALR, __gnu_h2f_ieee
+; CHECK-LIBCALL-NEXT: $tmp13:
+; CHECK-LIBCALL-NEXT: jalr $25
+; CHECK-LIBCALL-NEXT: cvt.d.s $f22, $f0
+; CHECK-LIBCALL-NEXT: cvt.d.s $f0, $f0
+; CHECK-LIBCALL-NEXT: sdc1 $f0, 24($18)
+; CHECK-LIBCALL-NEXT: sdc1 $f22, 16($18)
+; CHECK-LIBCALL-NEXT: sdc1 $f20, 8($18)
+; CHECK-LIBCALL-NEXT: lhu $4, 0($17)
+; CHECK-LIBCALL-NEXT: lw $25, %call16(__gnu_h2f_ieee)($16)
+; CHECK-LIBCALL-NEXT: .reloc ($tmp14), R_MIPS_JALR, __gnu_h2f_ieee
+; CHECK-LIBCALL-NEXT: $tmp14:
+; CHECK-LIBCALL-NEXT: jalr $25
+; CHECK-LIBCALL-NEXT: nop
+; CHECK-LIBCALL-NEXT: cvt.d.s $f0, $f0
+; CHECK-LIBCALL-NEXT: sdc1 $f0, 0($18)
+; CHECK-LIBCALL-NEXT: lw $16, 16($sp)
+; CHECK-LIBCALL-NEXT: lw $17, 20($sp)
+; CHECK-LIBCALL-NEXT: lw $18, 24($sp)
+; CHECK-LIBCALL-NEXT: lw $ra, 28($sp)
+; CHECK-LIBCALL-NEXT: ldc1 $f20, 32($sp)
+; CHECK-LIBCALL-NEXT: ldc1 $f22, 40($sp)
+; CHECK-LIBCALL-NEXT: jr $ra
+; CHECK-LIBCALL-NEXT: addiu $sp, $sp, 48
 define <4 x double> @test_vec_fpext_double(ptr %p) #0 {
   %a = load <4 x half>, ptr %p, align 8
   %b = fpext <4 x half> %a to <4 x double>
@@ -75,10 +276,61 @@ define <4 x double> @test_vec_fpext_double(ptr %p) #0 {
 }
 
 ; CHECK-LIBCALL-LABEL: test_vec_fptrunc_float:
-; CHECK-LIBCALL: %call16(__gnu_f2h_ieee)
-; CHECK-LIBCALL: %call16(__gnu_f2h_ieee)
-; CHECK-LIBCALL: %call16(__gnu_f2h_ieee)
-; CHECK-LIBCALL: %call16(__gnu_f2h_ieee)
+; CHECK-LIBCALL: lui $2, %hi(_gp_disp)
+; CHECK-LIBCALL-NEXT: addiu $2, $2, %lo(_gp_disp)
+; CHECK-LIBCALL-NEXT: addiu $sp, $sp, -40
+; CHECK-LIBCALL-NEXT: .cfi_def_cfa_offset 40
+; CHECK-LIBCALL-NEXT: sw $ra, 36($sp)
+; CHECK-LIBCALL-NEXT: sw $20, 32($sp)
+; CHECK-LIBCALL-NEXT: sw $19, 28($sp)
+; CHECK-LIBCALL-NEXT: sw $18, 24($sp)
+; CHECK-LIBCALL-NEXT: sw $17, 20($sp)
+; CHECK-LIBCALL-NEXT: sw $16, 16($sp)
+; CHECK-LIBCALL-NEXT: .cfi_offset 31, -4
+; CHECK-LIBCALL-NEXT: .cfi_offset 20, -8
+; CHECK-LIBCALL-NEXT: .cfi_offset 19, -12
+; CHECK-LIBCALL-NEXT: .cfi_offset 18, -16
+; CHECK-LIBCALL-NEXT: .cfi_offset 17, -20
+; CHECK-LIBCALL-NEXT: .cfi_offset 16, -24
+; CHECK-LIBCALL-NEXT: addu $16, $2, $25
+; CHECK-LIBCALL-NEXT: move $17, $7
+; CHECK-LIBCALL-NEXT: move $18, $5
+; CHECK-LIBCALL-NEXT: move $19, $4
+; CHECK-LIBCALL-NEXT: mtc1 $6, $f12
+; CHECK-LIBCALL-NEXT: lw $25, %call16(__gnu_f2h_ieee)($16)
+; CHECK-LIBCALL-NEXT: .reloc ($tmp15), R_MIPS_JALR, __gnu_f2h_ieee
+; CHECK-LIBCALL-NEXT: $tmp15:
+; CHECK-LIBCALL-NEXT: jalr $25
+; CHECK-LIBCALL-NEXT: move $gp, $16
+; CHECK-LIBCALL-NEXT: move $20, $2
+; CHECK-LIBCALL-NEXT: lw $25, %call16(__gnu_f2h_ieee)($16)
+; CHECK-LIBCALL-NEXT: .reloc ($tmp16), R_MIPS_JALR, __gnu_f2h_ieee
+; CHECK-LIBCALL-NEXT: $tmp16:
+; CHECK-LIBCALL-NEXT: jalr $25
+; CHECK-LIBCALL-NEXT: mtc1 $17, $f12
+; CHECK-LIBCALL-NEXT: mtc1 $18, $f12
+; CHECK-LIBCALL-NEXT: lw $25, %call16(__gnu_f2h_ieee)($16)
+; CHECK-LIBCALL-NEXT: lw $17, 56($sp)
+; CHECK-LIBCALL-NEXT: sh $2, 6($17)
+; CHECK-LIBCALL-NEXT: .reloc ($tmp17), R_MIPS_JALR, __gnu_f2h_ieee
+; CHECK-LIBCALL-NEXT: $tmp17:
+; CHECK-LIBCALL-NEXT: jalr $25
+; CHECK-LIBCALL-NEXT: sh $20, 4($17)
+; CHECK-LIBCALL-NEXT: sh $2, 2($17)
+; CHECK-LIBCALL-NEXT: lw $25, %call16(__gnu_f2h_ieee)($16)
+; CHECK-LIBCALL-NEXT: .reloc ($tmp18), R_MIPS_JALR, __gnu_f2h_ieee
+; CHECK-LIBCALL-NEXT: $tmp18:
+; CHECK-LIBCALL-NEXT: jalr $25
+; CHECK-LIBCALL-NEXT: mtc1 $19, $f12
+; CHECK-LIBCALL-NEXT: sh $2, 0($17)
+; CHECK-LIBCALL-NEXT: lw $16, 16($sp)
+; CHECK-LIBCALL-NEXT: lw $17, 20($sp)
+; CHECK-LIBCALL-NEXT: lw $18, 24($sp)
+; CHECK-LIBCALL-NEXT: lw $19, 28($sp)
+; CHECK-LIBCALL-NEXT: lw $20, 32($sp)
+; CHECK-LIBCALL-NEXT: lw $ra, 36($sp)
+; CHECK-LIBCALL-NEXT: jr $ra
+; CHECK-LIBCALL-NEXT: addiu $sp, $sp, 40
 define void @test_vec_fptrunc_float(<4 x float> %a, ptr %p) #0 {
   %b = fptrunc <4 x float> %a to <4 x half>
   store <4 x half> %b, ptr %p, align 8
@@ -86,13 +338,141 @@ define void @test_vec_fptrunc_float(<4 x float> %a, ptr %p) #0 {
 }
 
 ; CHECK-LIBCALL-LABEL: test_vec_fptrunc_double:
-; CHECK-LIBCALL: %call16(__truncdfhf2)
-; CHECK-LIBCALL: %call16(__truncdfhf2)
-; CHECK-LIBCALL: %call16(__truncdfhf2)
-; CHECK-LIBCALL: %call16(__truncdfhf2)
+; CHECK-LIBCALL: lui $2, %hi(_gp_disp)
+; CHECK-LIBCALL-NEXT: addiu $2, $2, %lo(_gp_disp)
+; CHECK-LIBCALL-NEXT: addiu $sp, $sp, -72
+; CHECK-LIBCALL-NEXT: .cfi_def_cfa_offset 72
+; CHECK-LIBCALL-NEXT: sw $ra, 68($sp)
+; CHECK-LIBCALL-NEXT: sw $20, 64($sp)
+; CHECK-LIBCALL-NEXT: sw $19, 60($sp)
+; CHECK-LIBCALL-NEXT: sw $18, 56($sp)
+; CHECK-LIBCALL-NEXT: sw $17, 52($sp)
+; CHECK-LIBCALL-NEXT: sw $16, 48($sp)
+; CHECK-LIBCALL-NEXT: .cfi_offset 31, -4
+; CHECK-LIBCALL-NEXT: .cfi_offset 20, -8
+; CHECK-LIBCALL-NEXT: .cfi_offset 19, -12
+; CHECK-LIBCALL-NEXT: .cfi_offset 18, -16
+; CHECK-LIBCALL-NEXT: .cfi_offset 17, -20
+; CHECK-LIBCALL-NEXT: .cfi_offset 16, -24
+; CHECK-LIBCALL-NEXT: addu $16, $2, $25
+; CHECK-LIBCALL-NEXT: move $17, $5
+; CHECK-LIBCALL-NEXT: move $18, $4
+; CHECK-LIBCALL-NEXT: lw $1, 88($sp)
+; CHECK-LIBCALL-NEXT: lw $2, 92($sp)
+; CHECK-LIBCALL-NEXT: sw $2, 36($sp)
+; CHECK-LIBCALL-NEXT: sw $1, 32($sp)
+; CHECK-LIBCALL-NEXT: lw $1, 96($sp)
+; CHECK-LIBCALL-NEXT: lw $2, 100($sp)
+; CHECK-LIBCALL-NEXT: sw $2, 44($sp)
+; CHECK-LIBCALL-NEXT: sw $1, 40($sp)
+; CHECK-LIBCALL-NEXT: lw $25, %call16(__truncdfhf2)($16)
+; CHECK-LIBCALL-NEXT: ldc1 $f12, 32($sp)
+; CHECK-LIBCALL-NEXT: sw $7, 28($sp)
+; CHECK-LIBCALL-NEXT: sw $6, 24($sp)
+; CHECK-LIBCALL-NEXT: .reloc ($tmp19), R_MIPS_JALR, __truncdfhf2
+; CHECK-LIBCALL-NEXT: $tmp19:
+; CHECK-LIBCALL-NEXT: jalr $25
+; CHECK-LIBCALL-NEXT: move $gp, $16
+; CHECK-LIBCALL-NEXT: move $19, $2
+; CHECK-LIBCALL-NEXT: lw $25, %call16(__truncdfhf2)($16)
+; CHECK-LIBCALL-NEXT: .reloc ($tmp20), R_MIPS_JALR, __truncdfhf2
+; CHECK-LIBCALL-NEXT: $tmp20:
+; CHECK-LIBCALL-NEXT: jalr $25
+; CHECK-LIBCALL-NEXT: ldc1 $f12, 40($sp)
+; CHECK-LIBCALL-NEXT: lw $25, %call16(__truncdfhf2)($16)
+; CHECK-LIBCALL-NEXT: ldc1 $f12, 24($sp)
+; CHECK-LIBCALL-NEXT: lw $20, 104($sp)
+; CHECK-LIBCALL-NEXT: sh $2, 6($20)
+; CHECK-LIBCALL-NEXT: .reloc ($tmp21), R_MIPS_JALR, __truncdfhf2
+; CHECK-LIBCALL-NEXT: $tmp21:
+; CHECK-LIBCALL-NEXT: jalr $25
+; CHECK-LIBCALL-NEXT: sh $19, 4($20)
+; CHECK-LIBCALL-NEXT: sh $2, 2($20)
+; CHECK-LIBCALL-NEXT: sw $17, 20($sp)
+; CHECK-LIBCALL-NEXT: sw $18, 16($sp)
+; CHECK-LIBCALL-NEXT: lw $25, %call16(__truncdfhf2)($16)
+; CHECK-LIBCALL-NEXT: .reloc ($tmp22), R_MIPS_JALR, __truncdfhf2
+; CHECK-LIBCALL-NEXT: $tmp22:
+; CHECK-LIBCALL-NEXT: jalr $25
+; CHECK-LIBCALL-NEXT: ldc1 $f12, 16($sp)
+; CHECK-LIBCALL-NEXT: sh $2, 0($20)
+; CHECK-LIBCALL-NEXT: lw $16, 48($sp)
+; CHECK-LIBCALL-NEXT: lw $17, 52($sp)
+; CHECK-LIBCALL-NEXT: lw $18, 56($sp)
+; CHECK-LIBCALL-NEXT: lw $19, 60($sp)
+; CHECK-LIBCALL-NEXT: lw $20, 64($sp)
+; CHECK-LIBCALL-NEXT: lw $ra, 68($sp)
+; CHECK-LIBCALL-NEXT: jr $ra
+; CHECK-LIBCALL-NEXT: addiu $sp, $sp, 72
 define void @test_vec_fptrunc_double(<4 x double> %a, ptr %p) #0 {
   %b = fptrunc <4 x double> %a to <4 x half>
   store <4 x half> %b, ptr %p, align 8
   ret void
 }
 
+define half @test_fadd_fadd(half %a, half %b, half %c) {
+; CHECK-LIBCALL-LABEL: test_fadd_fadd:
+; CHECK-LIBCALL: lui $2, %hi(_gp_disp)
+; CHECK-LIBCALL-NEXT: addiu $2, $2, %lo(_gp_disp)
+; CHECK-LIBCALL-NEXT: addiu $sp, $sp, -40
+; CHECK-LIBCALL-NEXT: .cfi_def_cfa_offset 40
+; CHECK-LIBCALL-NEXT: sdc1 $f20, 32($sp)
+; CHECK-LIBCALL-NEXT: sw $ra, 28($sp)
+; CHECK-LIBCALL-NEXT: sw $18, 24($sp)
+; CHECK-LIBCALL-NEXT: sw $17, 20($sp)
+; CHECK-LIBCALL-NEXT: sw $16, 16($sp)
+; CHECK-LIBCALL-NEXT: .cfi_offset 52, -8
+; CHECK-LIBCALL-NEXT: .cfi_offset 53, -4
+; CHECK-LIBCALL-NEXT: .cfi_offset 31, -12
+; CHECK-LIBCALL-NEXT: .cfi_offset 18, -16
+; CHECK-LIBCALL-NEXT: .cfi_offset 17, -20
+; CHECK-LIBCALL-NEXT: .cfi_offset 16, -24
+; CHECK-LIBCALL-NEXT: addu $16, $2, $25
+; CHECK-LIBCALL-NEXT: move $17, $6
+; CHECK-LIBCALL-NEXT: move $18, $4
+; CHECK-LIBCALL-NEXT: lw $25, %call16(__gnu_h2f_ieee)($16)
+; CHECK-LIBCALL-NEXT: move $4, $5
+; CHECK-LIBCALL-NEXT: .reloc ($tmp23), R_MIPS_JALR, __gnu_h2f_ieee
+; CHECK-LIBCALL-NEXT: $tmp23:
+; CHECK-LIBCALL-NEXT: jalr $25
+; CHECK-LIBCALL-NEXT: move $gp, $16
+; CHECK-LIBCALL-NEXT: mov.s $f20, $f0
+; CHECK-LIBCALL-NEXT: lw $25, %call16(__gnu_h2f_ieee)($16)
+; CHECK-LIBCALL-NEXT: .reloc ($tmp24), R_MIPS_JALR, __gnu_h2f_ieee
+; CHECK-LIBCALL-NEXT: $tmp24:
+; CHECK-LIBCALL-NEXT: jalr $25
+; CHECK-LIBCALL-NEXT: move $4, $18
+; CHECK-LIBCALL-NEXT: add.s $f12, $f0, $f20
+; CHECK-LIBCALL-NEXT: lw $25, %call16(__gnu_f2h_ieee)($16)
+; CHECK-LIBCALL-NEXT: .reloc ($tmp25), R_MIPS_JALR, __gnu_f2h_ieee
+; CHECK-LIBCALL-NEXT: $tmp25:
+; CHECK-LIBCALL-NEXT: jalr $25
+; CHECK-LIBCALL-NEXT: move $gp, $16
+; CHECK-LIBCALL-NEXT: lw $25, %call16(__gnu_h2f_ieee)($16)
+; CHECK-LIBCALL-NEXT: .reloc ($tmp26), R_MIPS_JALR, __gnu_h2f_ieee
+; CHECK-LIBCALL-NEXT: $tmp26:
+; CHECK-LIBCALL-NEXT: jalr $25
+; CHECK-LIBCALL-NEXT: move $4, $2
+; CHECK-LIBCALL-NEXT: mov.s $f20, $f0
+; CHECK-LIBCALL-NEXT: lw $25, %call16(__gnu_h2f_ieee)($16)
+; CHECK-LIBCALL-NEXT: .reloc ($tmp27), R_MIPS_JALR, __gnu_h2f_ieee
+; CHECK-LIBCALL-NEXT: $tmp27:
+; CHECK-LIBCALL-NEXT: jalr $25
+; CHECK-LIBCALL-NEXT: move $4, $17
+; CHECK-LIBCALL-NEXT: lw $25, %call16(__gnu_f2h_ieee)($16)
+; CHECK-LIBCALL-NEXT: .reloc ($tmp28), R_MIPS_JALR, __gnu_f2h_ieee
+; CHECK-LIBCALL-NEXT: $tmp28:
+; CHECK-LIBCALL-NEXT: jalr $25
+; CHECK-LIBCALL-NEXT: add.s $f12, $f20, $f0
+; CHECK-LIBCALL-NEXT: lw $16, 16($sp)
+; CHECK-LIBCALL-NEXT: lw $17, 20($sp)
+; CHECK-LIBCALL-NEXT: lw $18, 24($sp)
+; CHECK-LIBCALL-NEXT: lw $ra, 28($sp)
+; CHECK-LIBCALL-NEXT: ldc1 $f20, 32($sp)
+; CHECK-LIBCALL-NEXT: jr $ra
+; CHECK-LIBCALL-NEXT: addiu $sp, $sp, 40
+    %d = fadd half %a, %b
+    %e = fadd half %d, %c
+    ret half %e
+}
+