[llvm] [MIPS] Updated MIPS N calling conventions so that fp16 arguments no longer cause a crash (PR #116569)
via llvm-commits
llvm-commits at lists.llvm.org
Sun Nov 17 18:01:30 PST 2024
https://github.com/Tazdevil971 created https://github.com/llvm/llvm-project/pull/116569
This PR fixes a bug introduced by #110199, which causes any half float argument to crash the compiler on MIPS64.
Currently compiling this bit of code with `llc -mtriple=mips64`:
```
define void @half_args(half %a) nounwind {
entry:
ret void
}
```
Crashes with the following log:
```
LLVM ERROR: unable to allocate function argument #0
PLEASE submit a bug report to https://github.com/llvm/llvm-project/issues/ and include the crash backtrace.
Stack dump:
0. Program arguments: llc -mtriple=mips64
1. Running pass 'Function Pass Manager' on module '<stdin>'.
2. Running pass 'MIPS DAG->DAG Pattern Instruction Selection' on function '@half_args'
#0 0x000055a3a4013df8 llvm::sys::PrintStackTrace(llvm::raw_ostream&, int) (/home/davide/Ps2/rps2-tools/prefix/bin/llc+0x32d0df8)
#1 0x000055a3a401199e llvm::sys::RunSignalHandlers() (/home/davide/Ps2/rps2-tools/prefix/bin/llc+0x32ce99e)
#2 0x000055a3a40144a8 SignalHandler(int) Signals.cpp:0:0
#3 0x00007f00bde558c0 __restore_rt libc_sigaction.c:0:0
#4 0x00007f00bdea462c __pthread_kill_implementation ./nptl/pthread_kill.c:44:76
#5 0x00007f00bde55822 gsignal ./signal/../sysdeps/posix/raise.c:27:6
#6 0x00007f00bde3e4af abort ./stdlib/abort.c:81:7
#7 0x000055a3a3f80e3c llvm::report_fatal_error(llvm::Twine const&, bool) (/home/davide/Ps2/rps2-tools/prefix/bin/llc+0x323de3c)
#8 0x000055a3a2e20dfa (/home/davide/Ps2/rps2-tools/prefix/bin/llc+0x20dddfa)
#9 0x000055a3a2a34e20 llvm::MipsTargetLowering::LowerFormalArguments(llvm::SDValue, unsigned int, bool, llvm::SmallVectorImpl<llvm::ISD::InputArg> const&, llvm::SDLoc const&, llvm::SelectionDAG&, llvm::SmallVectorImpl<llvm::SDValue>&) const MipsISelLowering.cpp:0:0
#10 0x000055a3a3d896a9 llvm::SelectionDAGISel::LowerArguments(llvm::Function const&) (/home/davide/Ps2/rps2-tools/prefix/bin/llc+0x30466a9)
#11 0x000055a3a3e0b3ec llvm::SelectionDAGISel::SelectAllBasicBlocks(llvm::Function const&) (/home/davide/Ps2/rps2-tools/prefix/bin/llc+0x30c83ec)
#12 0x000055a3a3e09e21 llvm::SelectionDAGISel::runOnMachineFunction(llvm::MachineFunction&) (/home/davide/Ps2/rps2-tools/prefix/bin/llc+0x30c6e21)
#13 0x000055a3a2aae1ca llvm::MipsDAGToDAGISel::runOnMachineFunction(llvm::MachineFunction&) MipsISelDAGToDAG.cpp:0:0
#14 0x000055a3a3e07706 llvm::SelectionDAGISelLegacy::runOnMachineFunction(llvm::MachineFunction&) (/home/davide/Ps2/rps2-tools/prefix/bin/llc+0x30c4706)
#15 0x000055a3a3051ed6 llvm::MachineFunctionPass::runOnFunction(llvm::Function&) (/home/davide/Ps2/rps2-tools/prefix/bin/llc+0x230eed6)
#16 0x000055a3a35a3ec9 llvm::FPPassManager::runOnFunction(llvm::Function&) (/home/davide/Ps2/rps2-tools/prefix/bin/llc+0x2860ec9)
#17 0x000055a3a35ac3b2 llvm::FPPassManager::runOnModule(llvm::Module&) (/home/davide/Ps2/rps2-tools/prefix/bin/llc+0x28693b2)
#18 0x000055a3a35a499c llvm::legacy::PassManagerImpl::run(llvm::Module&) (/home/davide/Ps2/rps2-tools/prefix/bin/llc+0x286199c)
#19 0x000055a3a262abbb main (/home/davide/Ps2/rps2-tools/prefix/bin/llc+0x18e7bbb)
#20 0x00007f00bde3fc4c __libc_start_call_main ./csu/../sysdeps/nptl/libc_start_call_main.h:74:3
#21 0x00007f00bde3fd05 call_init ./csu/../csu/libc-start.c:128:20
#22 0x00007f00bde3fd05 __libc_start_main at GLIBC_2.2.5 ./csu/../csu/libc-start.c:347:5
#23 0x000055a3a2624921 _start /builddir/glibc-2.39/csu/../sysdeps/x86_64/start.S:117:0
```
This is caused by the fact that after the change, `f16`s are no longer lowered as `f32`s in calls.
Two possible fixes are available:
- Update calling conventions to properly support passing `f16` as integers.
- Update `useFPRegsForHalfType()` to return `true` so that `f16` are still kept in `f32` registers, as before #110199.
This PR implements the first solution to not introduce any more ABI changes as #110199 already did.
As of what is the correct ABI for halfs, I don't think there is a correct answer. GCC doesn't support halfs on MIPS, and I couldn't find any information on old MIPS ABI manuals either.
>From 29762d0264290bb1c71192aff87d88664b416fb4 Mon Sep 17 00:00:00 2001
From: Davide Mor <tazdevil971 at gmail.com>
Date: Mon, 18 Nov 2024 02:20:00 +0100
Subject: [PATCH] [MIPS] Updated MIPS N calling conventions so that fp16
arguments no longer cause a crash
---
llvm/lib/Target/Mips/MipsCallingConv.td | 15 +-
llvm/test/CodeGen/Mips/fp16-promote.ll | 734 ++++++++++++++++--------
2 files changed, 503 insertions(+), 246 deletions(-)
diff --git a/llvm/lib/Target/Mips/MipsCallingConv.td b/llvm/lib/Target/Mips/MipsCallingConv.td
index 866161bf50638d..25384a3fe8de37 100644
--- a/llvm/lib/Target/Mips/MipsCallingConv.td
+++ b/llvm/lib/Target/Mips/MipsCallingConv.td
@@ -20,8 +20,8 @@ class CCIfSubtargetNot<string F, CCAction A> : CCIfSubtarget<F, A, "!">;
/// Match if the original argument (before lowering) was a float.
/// For example, this is true for i32's that were lowered from soft-float.
-class CCIfOrigArgWasNotFloat<CCAction A>
- : CCIf<"!static_cast<MipsCCState *>(&State)->WasOriginalArgFloat(ValNo)",
+class CCIfOrigArgWasFloat<CCAction A>
+ : CCIf<"static_cast<MipsCCState *>(&State)->WasOriginalArgFloat(ValNo)",
A>;
/// Match if the original argument (before lowering) was a 128-bit float (i.e.
@@ -134,12 +134,13 @@ def CC_MipsN : CallingConv<[
CCIfSubtargetNot<"isLittle()",
CCIfInReg<CCPromoteToUpperBitsInType<i64>>>>,
- // All integers (except soft-float integers) are promoted to 64-bit.
- CCIfType<[i8, i16, i32], CCIfOrigArgWasNotFloat<CCPromoteToType<i64>>>,
+ // First deal with f32s that got turned into i32s due to soft-float
+ CCIfType<[i32],
+ CCIfSubtarget<"useSoftFloat()",
+ CCIfOrigArgWasFloat<CCDelegateTo<CC_MipsN_SoftFloat>>>>,
- // The only i32's we have left are soft-float arguments.
- CCIfSubtarget<"useSoftFloat()", CCIfType<[i32],
- CCDelegateTo<CC_MipsN_SoftFloat>>>,
+ // All remaining integers are promoted to 64-bit.
+ CCIfType<[i8, i16, i32], CCPromoteToType<i64>>,
// Integer arguments are passed in integer registers.
CCIfType<[i64], CCAssignToRegWithShadow<[A0_64, A1_64, A2_64, A3_64,
diff --git a/llvm/test/CodeGen/Mips/fp16-promote.ll b/llvm/test/CodeGen/Mips/fp16-promote.ll
index c104ffb3c72eb9..47bace9f5c03f6 100644
--- a/llvm/test/CodeGen/Mips/fp16-promote.ll
+++ b/llvm/test/CodeGen/Mips/fp16-promote.ll
@@ -1,28 +1,51 @@
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 5
-; RUN: llc -mtriple=mipsel-linux-gnu < %s | FileCheck %s -check-prefix=CHECK-LIBCALL
+; RUN: llc -mtriple=mipsel-linux-gnu < %s | FileCheck %s -check-prefix=MIPS32
+; RUN: llc -mtriple=mips64el-linux-gnu < %s | FileCheck %s -check-prefix=MIPS64
define void @test_fadd(ptr %p, ptr %q) nounwind {
-; CHECK-LIBCALL-LABEL: test_fadd:
-; CHECK-LIBCALL: # %bb.0:
-; CHECK-LIBCALL-NEXT: addiu $sp, $sp, -32
-; CHECK-LIBCALL-NEXT: sdc1 $f20, 24($sp) # 8-byte Folded Spill
-; CHECK-LIBCALL-NEXT: sw $ra, 20($sp) # 4-byte Folded Spill
-; CHECK-LIBCALL-NEXT: sw $16, 16($sp) # 4-byte Folded Spill
-; CHECK-LIBCALL-NEXT: move $16, $4
-; CHECK-LIBCALL-NEXT: lhu $4, 0($5)
-; CHECK-LIBCALL-NEXT: jal __gnu_h2f_ieee
-; CHECK-LIBCALL-NEXT: nop
-; CHECK-LIBCALL-NEXT: lhu $4, 0($16)
-; CHECK-LIBCALL-NEXT: jal __gnu_h2f_ieee
-; CHECK-LIBCALL-NEXT: mov.s $f20, $f0
-; CHECK-LIBCALL-NEXT: jal __gnu_f2h_ieee
-; CHECK-LIBCALL-NEXT: add.s $f12, $f0, $f20
-; CHECK-LIBCALL-NEXT: sh $2, 0($16)
-; CHECK-LIBCALL-NEXT: lw $16, 16($sp) # 4-byte Folded Reload
-; CHECK-LIBCALL-NEXT: lw $ra, 20($sp) # 4-byte Folded Reload
-; CHECK-LIBCALL-NEXT: ldc1 $f20, 24($sp) # 8-byte Folded Reload
-; CHECK-LIBCALL-NEXT: jr $ra
-; CHECK-LIBCALL-NEXT: addiu $sp, $sp, 32
+; MIPS32-LABEL: test_fadd:
+; MIPS32: # %bb.0:
+; MIPS32-NEXT: addiu $sp, $sp, -32
+; MIPS32-NEXT: sdc1 $f20, 24($sp) # 8-byte Folded Spill
+; MIPS32-NEXT: sw $ra, 20($sp) # 4-byte Folded Spill
+; MIPS32-NEXT: sw $16, 16($sp) # 4-byte Folded Spill
+; MIPS32-NEXT: move $16, $4
+; MIPS32-NEXT: lhu $4, 0($5)
+; MIPS32-NEXT: jal __gnu_h2f_ieee
+; MIPS32-NEXT: nop
+; MIPS32-NEXT: lhu $4, 0($16)
+; MIPS32-NEXT: jal __gnu_h2f_ieee
+; MIPS32-NEXT: mov.s $f20, $f0
+; MIPS32-NEXT: jal __gnu_f2h_ieee
+; MIPS32-NEXT: add.s $f12, $f0, $f20
+; MIPS32-NEXT: sh $2, 0($16)
+; MIPS32-NEXT: lw $16, 16($sp) # 4-byte Folded Reload
+; MIPS32-NEXT: lw $ra, 20($sp) # 4-byte Folded Reload
+; MIPS32-NEXT: ldc1 $f20, 24($sp) # 8-byte Folded Reload
+; MIPS32-NEXT: jr $ra
+; MIPS32-NEXT: addiu $sp, $sp, 32
+;
+; MIPS64-LABEL: test_fadd:
+; MIPS64: # %bb.0:
+; MIPS64-NEXT: daddiu $sp, $sp, -32
+; MIPS64-NEXT: sdc1 $f24, 24($sp) # 8-byte Folded Spill
+; MIPS64-NEXT: sd $ra, 16($sp) # 8-byte Folded Spill
+; MIPS64-NEXT: sd $16, 8($sp) # 8-byte Folded Spill
+; MIPS64-NEXT: move $16, $4
+; MIPS64-NEXT: lhu $4, 0($5)
+; MIPS64-NEXT: jal __gnu_h2f_ieee
+; MIPS64-NEXT: nop
+; MIPS64-NEXT: lhu $4, 0($16)
+; MIPS64-NEXT: jal __gnu_h2f_ieee
+; MIPS64-NEXT: mov.s $f24, $f0
+; MIPS64-NEXT: jal __gnu_f2h_ieee
+; MIPS64-NEXT: add.s $f12, $f0, $f24
+; MIPS64-NEXT: sh $2, 0($16)
+; MIPS64-NEXT: ld $16, 8($sp) # 8-byte Folded Reload
+; MIPS64-NEXT: ld $ra, 16($sp) # 8-byte Folded Reload
+; MIPS64-NEXT: ldc1 $f24, 24($sp) # 8-byte Folded Reload
+; MIPS64-NEXT: jr $ra
+; MIPS64-NEXT: daddiu $sp, $sp, 32
%a = load half, ptr %p, align 2
%b = load half, ptr %q, align 2
%r = fadd half %a, %b
@@ -31,282 +54,515 @@ define void @test_fadd(ptr %p, ptr %q) nounwind {
}
define float @test_fpext_float(ptr %p) nounwind {
-; CHECK-LIBCALL-LABEL: test_fpext_float:
-; CHECK-LIBCALL: # %bb.0:
-; CHECK-LIBCALL-NEXT: addiu $sp, $sp, -24
-; CHECK-LIBCALL-NEXT: sw $ra, 20($sp) # 4-byte Folded Spill
-; CHECK-LIBCALL-NEXT: lhu $4, 0($4)
-; CHECK-LIBCALL-NEXT: jal __gnu_h2f_ieee
-; CHECK-LIBCALL-NEXT: nop
-; CHECK-LIBCALL-NEXT: lw $ra, 20($sp) # 4-byte Folded Reload
-; CHECK-LIBCALL-NEXT: jr $ra
-; CHECK-LIBCALL-NEXT: addiu $sp, $sp, 24
+; MIPS32-LABEL: test_fpext_float:
+; MIPS32: # %bb.0:
+; MIPS32-NEXT: addiu $sp, $sp, -24
+; MIPS32-NEXT: sw $ra, 20($sp) # 4-byte Folded Spill
+; MIPS32-NEXT: lhu $4, 0($4)
+; MIPS32-NEXT: jal __gnu_h2f_ieee
+; MIPS32-NEXT: nop
+; MIPS32-NEXT: lw $ra, 20($sp) # 4-byte Folded Reload
+; MIPS32-NEXT: jr $ra
+; MIPS32-NEXT: addiu $sp, $sp, 24
+;
+; MIPS64-LABEL: test_fpext_float:
+; MIPS64: # %bb.0:
+; MIPS64-NEXT: daddiu $sp, $sp, -16
+; MIPS64-NEXT: sd $ra, 8($sp) # 8-byte Folded Spill
+; MIPS64-NEXT: lhu $4, 0($4)
+; MIPS64-NEXT: jal __gnu_h2f_ieee
+; MIPS64-NEXT: nop
+; MIPS64-NEXT: ld $ra, 8($sp) # 8-byte Folded Reload
+; MIPS64-NEXT: jr $ra
+; MIPS64-NEXT: daddiu $sp, $sp, 16
%a = load half, ptr %p, align 2
%r = fpext half %a to float
ret float %r
}
define double @test_fpext_double(ptr %p) nounwind {
-; CHECK-LIBCALL-LABEL: test_fpext_double:
-; CHECK-LIBCALL: # %bb.0:
-; CHECK-LIBCALL-NEXT: addiu $sp, $sp, -24
-; CHECK-LIBCALL-NEXT: sw $ra, 20($sp) # 4-byte Folded Spill
-; CHECK-LIBCALL-NEXT: lhu $4, 0($4)
-; CHECK-LIBCALL-NEXT: jal __gnu_h2f_ieee
-; CHECK-LIBCALL-NEXT: nop
-; CHECK-LIBCALL-NEXT: cvt.d.s $f0, $f0
-; CHECK-LIBCALL-NEXT: lw $ra, 20($sp) # 4-byte Folded Reload
-; CHECK-LIBCALL-NEXT: jr $ra
-; CHECK-LIBCALL-NEXT: addiu $sp, $sp, 24
+; MIPS32-LABEL: test_fpext_double:
+; MIPS32: # %bb.0:
+; MIPS32-NEXT: addiu $sp, $sp, -24
+; MIPS32-NEXT: sw $ra, 20($sp) # 4-byte Folded Spill
+; MIPS32-NEXT: lhu $4, 0($4)
+; MIPS32-NEXT: jal __gnu_h2f_ieee
+; MIPS32-NEXT: nop
+; MIPS32-NEXT: cvt.d.s $f0, $f0
+; MIPS32-NEXT: lw $ra, 20($sp) # 4-byte Folded Reload
+; MIPS32-NEXT: jr $ra
+; MIPS32-NEXT: addiu $sp, $sp, 24
+;
+; MIPS64-LABEL: test_fpext_double:
+; MIPS64: # %bb.0:
+; MIPS64-NEXT: daddiu $sp, $sp, -16
+; MIPS64-NEXT: sd $ra, 8($sp) # 8-byte Folded Spill
+; MIPS64-NEXT: lhu $4, 0($4)
+; MIPS64-NEXT: jal __gnu_h2f_ieee
+; MIPS64-NEXT: nop
+; MIPS64-NEXT: cvt.d.s $f0, $f0
+; MIPS64-NEXT: ld $ra, 8($sp) # 8-byte Folded Reload
+; MIPS64-NEXT: jr $ra
+; MIPS64-NEXT: daddiu $sp, $sp, 16
%a = load half, ptr %p, align 2
%r = fpext half %a to double
ret double %r
}
define void @test_fptrunc_float(float %f, ptr %p) nounwind {
-; CHECK-LIBCALL-LABEL: test_fptrunc_float:
-; CHECK-LIBCALL: # %bb.0:
-; CHECK-LIBCALL-NEXT: addiu $sp, $sp, -24
-; CHECK-LIBCALL-NEXT: sw $ra, 20($sp) # 4-byte Folded Spill
-; CHECK-LIBCALL-NEXT: sw $16, 16($sp) # 4-byte Folded Spill
-; CHECK-LIBCALL-NEXT: jal __gnu_f2h_ieee
-; CHECK-LIBCALL-NEXT: move $16, $5
-; CHECK-LIBCALL-NEXT: sh $2, 0($16)
-; CHECK-LIBCALL-NEXT: lw $16, 16($sp) # 4-byte Folded Reload
-; CHECK-LIBCALL-NEXT: lw $ra, 20($sp) # 4-byte Folded Reload
-; CHECK-LIBCALL-NEXT: jr $ra
-; CHECK-LIBCALL-NEXT: addiu $sp, $sp, 24
+; MIPS32-LABEL: test_fptrunc_float:
+; MIPS32: # %bb.0:
+; MIPS32-NEXT: addiu $sp, $sp, -24
+; MIPS32-NEXT: sw $ra, 20($sp) # 4-byte Folded Spill
+; MIPS32-NEXT: sw $16, 16($sp) # 4-byte Folded Spill
+; MIPS32-NEXT: jal __gnu_f2h_ieee
+; MIPS32-NEXT: move $16, $5
+; MIPS32-NEXT: sh $2, 0($16)
+; MIPS32-NEXT: lw $16, 16($sp) # 4-byte Folded Reload
+; MIPS32-NEXT: lw $ra, 20($sp) # 4-byte Folded Reload
+; MIPS32-NEXT: jr $ra
+; MIPS32-NEXT: addiu $sp, $sp, 24
+;
+; MIPS64-LABEL: test_fptrunc_float:
+; MIPS64: # %bb.0:
+; MIPS64-NEXT: daddiu $sp, $sp, -16
+; MIPS64-NEXT: sd $ra, 8($sp) # 8-byte Folded Spill
+; MIPS64-NEXT: sd $16, 0($sp) # 8-byte Folded Spill
+; MIPS64-NEXT: jal __gnu_f2h_ieee
+; MIPS64-NEXT: move $16, $5
+; MIPS64-NEXT: sh $2, 0($16)
+; MIPS64-NEXT: ld $16, 0($sp) # 8-byte Folded Reload
+; MIPS64-NEXT: ld $ra, 8($sp) # 8-byte Folded Reload
+; MIPS64-NEXT: jr $ra
+; MIPS64-NEXT: daddiu $sp, $sp, 16
%a = fptrunc float %f to half
store half %a, ptr %p
ret void
}
define void @test_fptrunc_double(double %d, ptr %p) nounwind {
-; CHECK-LIBCALL-LABEL: test_fptrunc_double:
-; CHECK-LIBCALL: # %bb.0:
-; CHECK-LIBCALL-NEXT: addiu $sp, $sp, -24
-; CHECK-LIBCALL-NEXT: sw $ra, 20($sp) # 4-byte Folded Spill
-; CHECK-LIBCALL-NEXT: sw $16, 16($sp) # 4-byte Folded Spill
-; CHECK-LIBCALL-NEXT: jal __truncdfhf2
-; CHECK-LIBCALL-NEXT: move $16, $6
-; CHECK-LIBCALL-NEXT: sh $2, 0($16)
-; CHECK-LIBCALL-NEXT: lw $16, 16($sp) # 4-byte Folded Reload
-; CHECK-LIBCALL-NEXT: lw $ra, 20($sp) # 4-byte Folded Reload
-; CHECK-LIBCALL-NEXT: jr $ra
-; CHECK-LIBCALL-NEXT: addiu $sp, $sp, 24
+; MIPS32-LABEL: test_fptrunc_double:
+; MIPS32: # %bb.0:
+; MIPS32-NEXT: addiu $sp, $sp, -24
+; MIPS32-NEXT: sw $ra, 20($sp) # 4-byte Folded Spill
+; MIPS32-NEXT: sw $16, 16($sp) # 4-byte Folded Spill
+; MIPS32-NEXT: jal __truncdfhf2
+; MIPS32-NEXT: move $16, $6
+; MIPS32-NEXT: sh $2, 0($16)
+; MIPS32-NEXT: lw $16, 16($sp) # 4-byte Folded Reload
+; MIPS32-NEXT: lw $ra, 20($sp) # 4-byte Folded Reload
+; MIPS32-NEXT: jr $ra
+; MIPS32-NEXT: addiu $sp, $sp, 24
+;
+; MIPS64-LABEL: test_fptrunc_double:
+; MIPS64: # %bb.0:
+; MIPS64-NEXT: daddiu $sp, $sp, -16
+; MIPS64-NEXT: sd $ra, 8($sp) # 8-byte Folded Spill
+; MIPS64-NEXT: sd $16, 0($sp) # 8-byte Folded Spill
+; MIPS64-NEXT: jal __truncdfhf2
+; MIPS64-NEXT: move $16, $5
+; MIPS64-NEXT: sh $2, 0($16)
+; MIPS64-NEXT: ld $16, 0($sp) # 8-byte Folded Reload
+; MIPS64-NEXT: ld $ra, 8($sp) # 8-byte Folded Reload
+; MIPS64-NEXT: jr $ra
+; MIPS64-NEXT: daddiu $sp, $sp, 16
%a = fptrunc double %d to half
store half %a, ptr %p
ret void
}
define <4 x float> @test_vec_fpext_float(ptr %p) nounwind {
-; CHECK-LIBCALL-LABEL: test_vec_fpext_float:
-; CHECK-LIBCALL: # %bb.0:
-; CHECK-LIBCALL-NEXT: addiu $sp, $sp, -32
-; CHECK-LIBCALL-NEXT: sw $ra, 28($sp) # 4-byte Folded Spill
-; CHECK-LIBCALL-NEXT: sw $17, 24($sp) # 4-byte Folded Spill
-; CHECK-LIBCALL-NEXT: sw $16, 20($sp) # 4-byte Folded Spill
-; CHECK-LIBCALL-NEXT: move $17, $4
-; CHECK-LIBCALL-NEXT: lhu $4, 6($5)
-; CHECK-LIBCALL-NEXT: jal __gnu_h2f_ieee
-; CHECK-LIBCALL-NEXT: move $16, $5
-; CHECK-LIBCALL-NEXT: lhu $4, 4($16)
-; CHECK-LIBCALL-NEXT: jal __gnu_h2f_ieee
-; CHECK-LIBCALL-NEXT: swc1 $f0, 12($17)
-; CHECK-LIBCALL-NEXT: swc1 $f0, 8($17)
-; CHECK-LIBCALL-NEXT: lhu $4, 2($16)
-; CHECK-LIBCALL-NEXT: jal __gnu_h2f_ieee
-; CHECK-LIBCALL-NEXT: nop
-; CHECK-LIBCALL-NEXT: swc1 $f0, 4($17)
-; CHECK-LIBCALL-NEXT: lhu $4, 0($16)
-; CHECK-LIBCALL-NEXT: jal __gnu_h2f_ieee
-; CHECK-LIBCALL-NEXT: nop
-; CHECK-LIBCALL-NEXT: swc1 $f0, 0($17)
-; CHECK-LIBCALL-NEXT: lw $16, 20($sp) # 4-byte Folded Reload
-; CHECK-LIBCALL-NEXT: lw $17, 24($sp) # 4-byte Folded Reload
-; CHECK-LIBCALL-NEXT: lw $ra, 28($sp) # 4-byte Folded Reload
-; CHECK-LIBCALL-NEXT: jr $ra
-; CHECK-LIBCALL-NEXT: addiu $sp, $sp, 32
+; MIPS32-LABEL: test_vec_fpext_float:
+; MIPS32: # %bb.0:
+; MIPS32-NEXT: addiu $sp, $sp, -32
+; MIPS32-NEXT: sw $ra, 28($sp) # 4-byte Folded Spill
+; MIPS32-NEXT: sw $17, 24($sp) # 4-byte Folded Spill
+; MIPS32-NEXT: sw $16, 20($sp) # 4-byte Folded Spill
+; MIPS32-NEXT: move $17, $4
+; MIPS32-NEXT: lhu $4, 6($5)
+; MIPS32-NEXT: jal __gnu_h2f_ieee
+; MIPS32-NEXT: move $16, $5
+; MIPS32-NEXT: lhu $4, 4($16)
+; MIPS32-NEXT: jal __gnu_h2f_ieee
+; MIPS32-NEXT: swc1 $f0, 12($17)
+; MIPS32-NEXT: swc1 $f0, 8($17)
+; MIPS32-NEXT: lhu $4, 2($16)
+; MIPS32-NEXT: jal __gnu_h2f_ieee
+; MIPS32-NEXT: nop
+; MIPS32-NEXT: swc1 $f0, 4($17)
+; MIPS32-NEXT: lhu $4, 0($16)
+; MIPS32-NEXT: jal __gnu_h2f_ieee
+; MIPS32-NEXT: nop
+; MIPS32-NEXT: swc1 $f0, 0($17)
+; MIPS32-NEXT: lw $16, 20($sp) # 4-byte Folded Reload
+; MIPS32-NEXT: lw $17, 24($sp) # 4-byte Folded Reload
+; MIPS32-NEXT: lw $ra, 28($sp) # 4-byte Folded Reload
+; MIPS32-NEXT: jr $ra
+; MIPS32-NEXT: addiu $sp, $sp, 32
+;
+; MIPS64-LABEL: test_vec_fpext_float:
+; MIPS64: # %bb.0:
+; MIPS64-NEXT: daddiu $sp, $sp, -32
+; MIPS64-NEXT: sd $ra, 24($sp) # 8-byte Folded Spill
+; MIPS64-NEXT: sd $18, 16($sp) # 8-byte Folded Spill
+; MIPS64-NEXT: sd $17, 8($sp) # 8-byte Folded Spill
+; MIPS64-NEXT: sd $16, 0($sp) # 8-byte Folded Spill
+; MIPS64-NEXT: move $16, $4
+; MIPS64-NEXT: lhu $4, 2($4)
+; MIPS64-NEXT: jal __gnu_h2f_ieee
+; MIPS64-NEXT: nop
+; MIPS64-NEXT: lhu $4, 6($16)
+; MIPS64-NEXT: jal __gnu_h2f_ieee
+; MIPS64-NEXT: mfc1 $17, $f0
+; MIPS64-NEXT: mfc1 $18, $f0
+; MIPS64-NEXT: lhu $4, 0($16)
+; MIPS64-NEXT: jal __gnu_h2f_ieee
+; MIPS64-NEXT: dsll $17, $17, 32
+; MIPS64-NEXT: mfc1 $1, $f0
+; MIPS64-NEXT: dsll $1, $1, 32
+; MIPS64-NEXT: dsrl $1, $1, 32
+; MIPS64-NEXT: or $17, $1, $17
+; MIPS64-NEXT: lhu $4, 4($16)
+; MIPS64-NEXT: jal __gnu_h2f_ieee
+; MIPS64-NEXT: dsll $18, $18, 32
+; MIPS64-NEXT: mfc1 $1, $f0
+; MIPS64-NEXT: dsll $1, $1, 32
+; MIPS64-NEXT: dsrl $1, $1, 32
+; MIPS64-NEXT: or $3, $1, $18
+; MIPS64-NEXT: move $2, $17
+; MIPS64-NEXT: ld $16, 0($sp) # 8-byte Folded Reload
+; MIPS64-NEXT: ld $17, 8($sp) # 8-byte Folded Reload
+; MIPS64-NEXT: ld $18, 16($sp) # 8-byte Folded Reload
+; MIPS64-NEXT: ld $ra, 24($sp) # 8-byte Folded Reload
+; MIPS64-NEXT: jr $ra
+; MIPS64-NEXT: daddiu $sp, $sp, 32
%a = load <4 x half>, ptr %p, align 8
%b = fpext <4 x half> %a to <4 x float>
ret <4 x float> %b
}
define <4 x double> @test_vec_fpext_double(ptr %p) nounwind {
-; CHECK-LIBCALL-LABEL: test_vec_fpext_double:
-; CHECK-LIBCALL: # %bb.0:
-; CHECK-LIBCALL-NEXT: addiu $sp, $sp, -40
-; CHECK-LIBCALL-NEXT: sdc1 $f20, 32($sp) # 8-byte Folded Spill
-; CHECK-LIBCALL-NEXT: sw $ra, 28($sp) # 4-byte Folded Spill
-; CHECK-LIBCALL-NEXT: sw $17, 24($sp) # 4-byte Folded Spill
-; CHECK-LIBCALL-NEXT: sw $16, 20($sp) # 4-byte Folded Spill
-; CHECK-LIBCALL-NEXT: move $17, $4
-; CHECK-LIBCALL-NEXT: lhu $4, 6($5)
-; CHECK-LIBCALL-NEXT: jal __gnu_h2f_ieee
-; CHECK-LIBCALL-NEXT: move $16, $5
-; CHECK-LIBCALL-NEXT: lhu $4, 4($16)
-; CHECK-LIBCALL-NEXT: jal __gnu_h2f_ieee
-; CHECK-LIBCALL-NEXT: mov.s $f20, $f0
-; CHECK-LIBCALL-NEXT: lhu $4, 2($16)
-; CHECK-LIBCALL-NEXT: cvt.d.s $f0, $f0
-; CHECK-LIBCALL-NEXT: cvt.d.s $f2, $f20
-; CHECK-LIBCALL-NEXT: sdc1 $f2, 24($17)
-; CHECK-LIBCALL-NEXT: jal __gnu_h2f_ieee
-; CHECK-LIBCALL-NEXT: sdc1 $f0, 16($17)
-; CHECK-LIBCALL-NEXT: cvt.d.s $f0, $f0
-; CHECK-LIBCALL-NEXT: sdc1 $f0, 8($17)
-; CHECK-LIBCALL-NEXT: lhu $4, 0($16)
-; CHECK-LIBCALL-NEXT: jal __gnu_h2f_ieee
-; CHECK-LIBCALL-NEXT: nop
-; CHECK-LIBCALL-NEXT: cvt.d.s $f0, $f0
-; CHECK-LIBCALL-NEXT: sdc1 $f0, 0($17)
-; CHECK-LIBCALL-NEXT: lw $16, 20($sp) # 4-byte Folded Reload
-; CHECK-LIBCALL-NEXT: lw $17, 24($sp) # 4-byte Folded Reload
-; CHECK-LIBCALL-NEXT: lw $ra, 28($sp) # 4-byte Folded Reload
-; CHECK-LIBCALL-NEXT: ldc1 $f20, 32($sp) # 8-byte Folded Reload
-; CHECK-LIBCALL-NEXT: jr $ra
-; CHECK-LIBCALL-NEXT: addiu $sp, $sp, 40
+; MIPS32-LABEL: test_vec_fpext_double:
+; MIPS32: # %bb.0:
+; MIPS32-NEXT: addiu $sp, $sp, -40
+; MIPS32-NEXT: sdc1 $f20, 32($sp) # 8-byte Folded Spill
+; MIPS32-NEXT: sw $ra, 28($sp) # 4-byte Folded Spill
+; MIPS32-NEXT: sw $17, 24($sp) # 4-byte Folded Spill
+; MIPS32-NEXT: sw $16, 20($sp) # 4-byte Folded Spill
+; MIPS32-NEXT: move $17, $4
+; MIPS32-NEXT: lhu $4, 6($5)
+; MIPS32-NEXT: jal __gnu_h2f_ieee
+; MIPS32-NEXT: move $16, $5
+; MIPS32-NEXT: lhu $4, 4($16)
+; MIPS32-NEXT: jal __gnu_h2f_ieee
+; MIPS32-NEXT: mov.s $f20, $f0
+; MIPS32-NEXT: lhu $4, 2($16)
+; MIPS32-NEXT: cvt.d.s $f0, $f0
+; MIPS32-NEXT: cvt.d.s $f2, $f20
+; MIPS32-NEXT: sdc1 $f2, 24($17)
+; MIPS32-NEXT: jal __gnu_h2f_ieee
+; MIPS32-NEXT: sdc1 $f0, 16($17)
+; MIPS32-NEXT: cvt.d.s $f0, $f0
+; MIPS32-NEXT: sdc1 $f0, 8($17)
+; MIPS32-NEXT: lhu $4, 0($16)
+; MIPS32-NEXT: jal __gnu_h2f_ieee
+; MIPS32-NEXT: nop
+; MIPS32-NEXT: cvt.d.s $f0, $f0
+; MIPS32-NEXT: sdc1 $f0, 0($17)
+; MIPS32-NEXT: lw $16, 20($sp) # 4-byte Folded Reload
+; MIPS32-NEXT: lw $17, 24($sp) # 4-byte Folded Reload
+; MIPS32-NEXT: lw $ra, 28($sp) # 4-byte Folded Reload
+; MIPS32-NEXT: ldc1 $f20, 32($sp) # 8-byte Folded Reload
+; MIPS32-NEXT: jr $ra
+; MIPS32-NEXT: addiu $sp, $sp, 40
+;
+; MIPS64-LABEL: test_vec_fpext_double:
+; MIPS64: # %bb.0:
+; MIPS64-NEXT: daddiu $sp, $sp, -32
+; MIPS64-NEXT: sdc1 $f24, 24($sp) # 8-byte Folded Spill
+; MIPS64-NEXT: sd $ra, 16($sp) # 8-byte Folded Spill
+; MIPS64-NEXT: sd $17, 8($sp) # 8-byte Folded Spill
+; MIPS64-NEXT: sd $16, 0($sp) # 8-byte Folded Spill
+; MIPS64-NEXT: move $17, $4
+; MIPS64-NEXT: lhu $4, 6($5)
+; MIPS64-NEXT: jal __gnu_h2f_ieee
+; MIPS64-NEXT: move $16, $5
+; MIPS64-NEXT: lhu $4, 4($16)
+; MIPS64-NEXT: jal __gnu_h2f_ieee
+; MIPS64-NEXT: mov.s $f24, $f0
+; MIPS64-NEXT: lhu $4, 2($16)
+; MIPS64-NEXT: cvt.d.s $f0, $f0
+; MIPS64-NEXT: cvt.d.s $f1, $f24
+; MIPS64-NEXT: sdc1 $f1, 24($17)
+; MIPS64-NEXT: jal __gnu_h2f_ieee
+; MIPS64-NEXT: sdc1 $f0, 16($17)
+; MIPS64-NEXT: cvt.d.s $f0, $f0
+; MIPS64-NEXT: sdc1 $f0, 8($17)
+; MIPS64-NEXT: lhu $4, 0($16)
+; MIPS64-NEXT: jal __gnu_h2f_ieee
+; MIPS64-NEXT: nop
+; MIPS64-NEXT: cvt.d.s $f0, $f0
+; MIPS64-NEXT: sdc1 $f0, 0($17)
+; MIPS64-NEXT: ld $16, 0($sp) # 8-byte Folded Reload
+; MIPS64-NEXT: ld $17, 8($sp) # 8-byte Folded Reload
+; MIPS64-NEXT: ld $ra, 16($sp) # 8-byte Folded Reload
+; MIPS64-NEXT: ldc1 $f24, 24($sp) # 8-byte Folded Reload
+; MIPS64-NEXT: jr $ra
+; MIPS64-NEXT: daddiu $sp, $sp, 32
%a = load <4 x half>, ptr %p, align 8
%b = fpext <4 x half> %a to <4 x double>
ret <4 x double> %b
}
define void @test_vec_fptrunc_float(<4 x float> %a, ptr %p) nounwind {
-; CHECK-LIBCALL-LABEL: test_vec_fptrunc_float:
-; CHECK-LIBCALL: # %bb.0:
-; CHECK-LIBCALL-NEXT: addiu $sp, $sp, -40
-; CHECK-LIBCALL-NEXT: sw $ra, 36($sp) # 4-byte Folded Spill
-; CHECK-LIBCALL-NEXT: sw $19, 32($sp) # 4-byte Folded Spill
-; CHECK-LIBCALL-NEXT: sw $18, 28($sp) # 4-byte Folded Spill
-; CHECK-LIBCALL-NEXT: sw $17, 24($sp) # 4-byte Folded Spill
-; CHECK-LIBCALL-NEXT: sw $16, 20($sp) # 4-byte Folded Spill
-; CHECK-LIBCALL-NEXT: move $16, $7
-; CHECK-LIBCALL-NEXT: move $17, $5
-; CHECK-LIBCALL-NEXT: move $18, $4
-; CHECK-LIBCALL-NEXT: jal __gnu_f2h_ieee
-; CHECK-LIBCALL-NEXT: mtc1 $6, $f12
-; CHECK-LIBCALL-NEXT: move $19, $2
-; CHECK-LIBCALL-NEXT: jal __gnu_f2h_ieee
-; CHECK-LIBCALL-NEXT: mtc1 $16, $f12
-; CHECK-LIBCALL-NEXT: mtc1 $17, $f12
-; CHECK-LIBCALL-NEXT: lw $16, 56($sp)
-; CHECK-LIBCALL-NEXT: sh $2, 6($16)
-; CHECK-LIBCALL-NEXT: jal __gnu_f2h_ieee
-; CHECK-LIBCALL-NEXT: sh $19, 4($16)
-; CHECK-LIBCALL-NEXT: sh $2, 2($16)
-; CHECK-LIBCALL-NEXT: jal __gnu_f2h_ieee
-; CHECK-LIBCALL-NEXT: mtc1 $18, $f12
-; CHECK-LIBCALL-NEXT: sh $2, 0($16)
-; CHECK-LIBCALL-NEXT: lw $16, 20($sp) # 4-byte Folded Reload
-; CHECK-LIBCALL-NEXT: lw $17, 24($sp) # 4-byte Folded Reload
-; CHECK-LIBCALL-NEXT: lw $18, 28($sp) # 4-byte Folded Reload
-; CHECK-LIBCALL-NEXT: lw $19, 32($sp) # 4-byte Folded Reload
-; CHECK-LIBCALL-NEXT: lw $ra, 36($sp) # 4-byte Folded Reload
-; CHECK-LIBCALL-NEXT: jr $ra
-; CHECK-LIBCALL-NEXT: addiu $sp, $sp, 40
+; MIPS32-LABEL: test_vec_fptrunc_float:
+; MIPS32: # %bb.0:
+; MIPS32-NEXT: addiu $sp, $sp, -40
+; MIPS32-NEXT: sw $ra, 36($sp) # 4-byte Folded Spill
+; MIPS32-NEXT: sw $19, 32($sp) # 4-byte Folded Spill
+; MIPS32-NEXT: sw $18, 28($sp) # 4-byte Folded Spill
+; MIPS32-NEXT: sw $17, 24($sp) # 4-byte Folded Spill
+; MIPS32-NEXT: sw $16, 20($sp) # 4-byte Folded Spill
+; MIPS32-NEXT: move $16, $7
+; MIPS32-NEXT: move $17, $5
+; MIPS32-NEXT: move $18, $4
+; MIPS32-NEXT: jal __gnu_f2h_ieee
+; MIPS32-NEXT: mtc1 $6, $f12
+; MIPS32-NEXT: move $19, $2
+; MIPS32-NEXT: jal __gnu_f2h_ieee
+; MIPS32-NEXT: mtc1 $16, $f12
+; MIPS32-NEXT: mtc1 $17, $f12
+; MIPS32-NEXT: lw $16, 56($sp)
+; MIPS32-NEXT: sh $2, 6($16)
+; MIPS32-NEXT: jal __gnu_f2h_ieee
+; MIPS32-NEXT: sh $19, 4($16)
+; MIPS32-NEXT: sh $2, 2($16)
+; MIPS32-NEXT: jal __gnu_f2h_ieee
+; MIPS32-NEXT: mtc1 $18, $f12
+; MIPS32-NEXT: sh $2, 0($16)
+; MIPS32-NEXT: lw $16, 20($sp) # 4-byte Folded Reload
+; MIPS32-NEXT: lw $17, 24($sp) # 4-byte Folded Reload
+; MIPS32-NEXT: lw $18, 28($sp) # 4-byte Folded Reload
+; MIPS32-NEXT: lw $19, 32($sp) # 4-byte Folded Reload
+; MIPS32-NEXT: lw $ra, 36($sp) # 4-byte Folded Reload
+; MIPS32-NEXT: jr $ra
+; MIPS32-NEXT: addiu $sp, $sp, 40
+;
+; MIPS64-LABEL: test_vec_fptrunc_float:
+; MIPS64: # %bb.0:
+; MIPS64-NEXT: daddiu $sp, $sp, -48
+; MIPS64-NEXT: sd $ra, 40($sp) # 8-byte Folded Spill
+; MIPS64-NEXT: sd $19, 32($sp) # 8-byte Folded Spill
+; MIPS64-NEXT: sd $18, 24($sp) # 8-byte Folded Spill
+; MIPS64-NEXT: sd $17, 16($sp) # 8-byte Folded Spill
+; MIPS64-NEXT: sd $16, 8($sp) # 8-byte Folded Spill
+; MIPS64-NEXT: move $16, $6
+; MIPS64-NEXT: move $17, $5
+; MIPS64-NEXT: move $18, $4
+; MIPS64-NEXT: sll $1, $18, 0
+; MIPS64-NEXT: jal __gnu_f2h_ieee
+; MIPS64-NEXT: mtc1 $1, $f12
+; MIPS64-NEXT: move $19, $2
+; MIPS64-NEXT: sll $1, $17, 0
+; MIPS64-NEXT: jal __gnu_f2h_ieee
+; MIPS64-NEXT: mtc1 $1, $f12
+; MIPS64-NEXT: dsrl $1, $17, 32
+; MIPS64-NEXT: sll $1, $1, 0
+; MIPS64-NEXT: mtc1 $1, $f12
+; MIPS64-NEXT: sh $2, 4($16)
+; MIPS64-NEXT: jal __gnu_f2h_ieee
+; MIPS64-NEXT: sh $19, 0($16)
+; MIPS64-NEXT: sh $2, 6($16)
+; MIPS64-NEXT: dsrl $1, $18, 32
+; MIPS64-NEXT: sll $1, $1, 0
+; MIPS64-NEXT: jal __gnu_f2h_ieee
+; MIPS64-NEXT: mtc1 $1, $f12
+; MIPS64-NEXT: sh $2, 2($16)
+; MIPS64-NEXT: ld $16, 8($sp) # 8-byte Folded Reload
+; MIPS64-NEXT: ld $17, 16($sp) # 8-byte Folded Reload
+; MIPS64-NEXT: ld $18, 24($sp) # 8-byte Folded Reload
+; MIPS64-NEXT: ld $19, 32($sp) # 8-byte Folded Reload
+; MIPS64-NEXT: ld $ra, 40($sp) # 8-byte Folded Reload
+; MIPS64-NEXT: jr $ra
+; MIPS64-NEXT: daddiu $sp, $sp, 48
%b = fptrunc <4 x float> %a to <4 x half>
store <4 x half> %b, ptr %p, align 8
ret void
}
define void @test_vec_fptrunc_double(<4 x double> %a, ptr %p) nounwind {
-; CHECK-LIBCALL-LABEL: test_vec_fptrunc_double:
-; CHECK-LIBCALL: # %bb.0:
-; CHECK-LIBCALL-NEXT: addiu $sp, $sp, -72
-; CHECK-LIBCALL-NEXT: sw $ra, 68($sp) # 4-byte Folded Spill
-; CHECK-LIBCALL-NEXT: sw $19, 64($sp) # 4-byte Folded Spill
-; CHECK-LIBCALL-NEXT: sw $18, 60($sp) # 4-byte Folded Spill
-; CHECK-LIBCALL-NEXT: sw $17, 56($sp) # 4-byte Folded Spill
-; CHECK-LIBCALL-NEXT: sw $16, 52($sp) # 4-byte Folded Spill
-; CHECK-LIBCALL-NEXT: move $16, $5
-; CHECK-LIBCALL-NEXT: move $17, $4
-; CHECK-LIBCALL-NEXT: lw $1, 92($sp)
-; CHECK-LIBCALL-NEXT: sw $1, 36($sp)
-; CHECK-LIBCALL-NEXT: lw $1, 88($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: ldc1 $f12, 32($sp)
-; CHECK-LIBCALL-NEXT: sw $7, 28($sp)
-; CHECK-LIBCALL-NEXT: jal __truncdfhf2
-; CHECK-LIBCALL-NEXT: sw $6, 24($sp)
-; CHECK-LIBCALL-NEXT: move $18, $2
-; CHECK-LIBCALL-NEXT: jal __truncdfhf2
-; CHECK-LIBCALL-NEXT: ldc1 $f12, 40($sp)
-; CHECK-LIBCALL-NEXT: ldc1 $f12, 24($sp)
-; CHECK-LIBCALL-NEXT: lw $19, 104($sp)
-; CHECK-LIBCALL-NEXT: sh $2, 6($19)
-; CHECK-LIBCALL-NEXT: jal __truncdfhf2
-; CHECK-LIBCALL-NEXT: sh $18, 4($19)
-; CHECK-LIBCALL-NEXT: sh $2, 2($19)
-; CHECK-LIBCALL-NEXT: sw $16, 20($sp)
-; CHECK-LIBCALL-NEXT: sw $17, 16($sp)
-; CHECK-LIBCALL-NEXT: jal __truncdfhf2
-; CHECK-LIBCALL-NEXT: ldc1 $f12, 16($sp)
-; CHECK-LIBCALL-NEXT: sh $2, 0($19)
-; CHECK-LIBCALL-NEXT: lw $16, 52($sp) # 4-byte Folded Reload
-; CHECK-LIBCALL-NEXT: lw $17, 56($sp) # 4-byte Folded Reload
-; CHECK-LIBCALL-NEXT: lw $18, 60($sp) # 4-byte Folded Reload
-; CHECK-LIBCALL-NEXT: lw $19, 64($sp) # 4-byte Folded Reload
-; CHECK-LIBCALL-NEXT: lw $ra, 68($sp) # 4-byte Folded Reload
-; CHECK-LIBCALL-NEXT: jr $ra
-; CHECK-LIBCALL-NEXT: addiu $sp, $sp, 72
+; MIPS32-LABEL: test_vec_fptrunc_double:
+; MIPS32: # %bb.0:
+; MIPS32-NEXT: addiu $sp, $sp, -72
+; MIPS32-NEXT: sw $ra, 68($sp) # 4-byte Folded Spill
+; MIPS32-NEXT: sw $19, 64($sp) # 4-byte Folded Spill
+; MIPS32-NEXT: sw $18, 60($sp) # 4-byte Folded Spill
+; MIPS32-NEXT: sw $17, 56($sp) # 4-byte Folded Spill
+; MIPS32-NEXT: sw $16, 52($sp) # 4-byte Folded Spill
+; MIPS32-NEXT: move $16, $5
+; MIPS32-NEXT: move $17, $4
+; MIPS32-NEXT: lw $1, 92($sp)
+; MIPS32-NEXT: sw $1, 36($sp)
+; MIPS32-NEXT: lw $1, 88($sp)
+; MIPS32-NEXT: sw $1, 32($sp)
+; MIPS32-NEXT: lw $1, 96($sp)
+; MIPS32-NEXT: lw $2, 100($sp)
+; MIPS32-NEXT: sw $2, 44($sp)
+; MIPS32-NEXT: sw $1, 40($sp)
+; MIPS32-NEXT: ldc1 $f12, 32($sp)
+; MIPS32-NEXT: sw $7, 28($sp)
+; MIPS32-NEXT: jal __truncdfhf2
+; MIPS32-NEXT: sw $6, 24($sp)
+; MIPS32-NEXT: move $18, $2
+; MIPS32-NEXT: jal __truncdfhf2
+; MIPS32-NEXT: ldc1 $f12, 40($sp)
+; MIPS32-NEXT: ldc1 $f12, 24($sp)
+; MIPS32-NEXT: lw $19, 104($sp)
+; MIPS32-NEXT: sh $2, 6($19)
+; MIPS32-NEXT: jal __truncdfhf2
+; MIPS32-NEXT: sh $18, 4($19)
+; MIPS32-NEXT: sh $2, 2($19)
+; MIPS32-NEXT: sw $16, 20($sp)
+; MIPS32-NEXT: sw $17, 16($sp)
+; MIPS32-NEXT: jal __truncdfhf2
+; MIPS32-NEXT: ldc1 $f12, 16($sp)
+; MIPS32-NEXT: sh $2, 0($19)
+; MIPS32-NEXT: lw $16, 52($sp) # 4-byte Folded Reload
+; MIPS32-NEXT: lw $17, 56($sp) # 4-byte Folded Reload
+; MIPS32-NEXT: lw $18, 60($sp) # 4-byte Folded Reload
+; MIPS32-NEXT: lw $19, 64($sp) # 4-byte Folded Reload
+; MIPS32-NEXT: lw $ra, 68($sp) # 4-byte Folded Reload
+; MIPS32-NEXT: jr $ra
+; MIPS32-NEXT: addiu $sp, $sp, 72
+;
+; MIPS64-LABEL: test_vec_fptrunc_double:
+; MIPS64: # %bb.0:
+; MIPS64-NEXT: daddiu $sp, $sp, -48
+; MIPS64-NEXT: sd $ra, 40($sp) # 8-byte Folded Spill
+; MIPS64-NEXT: sd $20, 32($sp) # 8-byte Folded Spill
+; MIPS64-NEXT: sd $19, 24($sp) # 8-byte Folded Spill
+; MIPS64-NEXT: sd $18, 16($sp) # 8-byte Folded Spill
+; MIPS64-NEXT: sd $17, 8($sp) # 8-byte Folded Spill
+; MIPS64-NEXT: sd $16, 0($sp) # 8-byte Folded Spill
+; MIPS64-NEXT: move $16, $8
+; MIPS64-NEXT: move $17, $7
+; MIPS64-NEXT: move $18, $5
+; MIPS64-NEXT: move $19, $4
+; MIPS64-NEXT: jal __truncdfhf2
+; MIPS64-NEXT: dmtc1 $6, $f12
+; MIPS64-NEXT: move $20, $2
+; MIPS64-NEXT: jal __truncdfhf2
+; MIPS64-NEXT: dmtc1 $17, $f12
+; MIPS64-NEXT: dmtc1 $18, $f12
+; MIPS64-NEXT: sh $2, 6($16)
+; MIPS64-NEXT: jal __truncdfhf2
+; MIPS64-NEXT: sh $20, 4($16)
+; MIPS64-NEXT: sh $2, 2($16)
+; MIPS64-NEXT: jal __truncdfhf2
+; MIPS64-NEXT: dmtc1 $19, $f12
+; MIPS64-NEXT: sh $2, 0($16)
+; MIPS64-NEXT: ld $16, 0($sp) # 8-byte Folded Reload
+; MIPS64-NEXT: ld $17, 8($sp) # 8-byte Folded Reload
+; MIPS64-NEXT: ld $18, 16($sp) # 8-byte Folded Reload
+; MIPS64-NEXT: ld $19, 24($sp) # 8-byte Folded Reload
+; MIPS64-NEXT: ld $20, 32($sp) # 8-byte Folded Reload
+; MIPS64-NEXT: ld $ra, 40($sp) # 8-byte Folded Reload
+; MIPS64-NEXT: jr $ra
+; MIPS64-NEXT: daddiu $sp, $sp, 48
%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) nounwind {
-; CHECK-LIBCALL-LABEL: test_fadd_fadd:
-; CHECK-LIBCALL: # %bb.0:
-; CHECK-LIBCALL-NEXT: addiu $sp, $sp, -40
-; CHECK-LIBCALL-NEXT: sdc1 $f20, 32($sp) # 8-byte Folded Spill
-; CHECK-LIBCALL-NEXT: sw $ra, 28($sp) # 4-byte Folded Spill
-; CHECK-LIBCALL-NEXT: sw $17, 24($sp) # 4-byte Folded Spill
-; CHECK-LIBCALL-NEXT: sw $16, 20($sp) # 4-byte Folded Spill
-; CHECK-LIBCALL-NEXT: move $16, $6
-; CHECK-LIBCALL-NEXT: move $17, $4
-; CHECK-LIBCALL-NEXT: jal __gnu_h2f_ieee
-; CHECK-LIBCALL-NEXT: move $4, $5
-; CHECK-LIBCALL-NEXT: mov.s $f20, $f0
-; CHECK-LIBCALL-NEXT: jal __gnu_h2f_ieee
-; CHECK-LIBCALL-NEXT: move $4, $17
-; CHECK-LIBCALL-NEXT: jal __gnu_f2h_ieee
-; CHECK-LIBCALL-NEXT: add.s $f12, $f0, $f20
-; CHECK-LIBCALL-NEXT: jal __gnu_h2f_ieee
-; CHECK-LIBCALL-NEXT: move $4, $2
-; CHECK-LIBCALL-NEXT: mov.s $f20, $f0
-; CHECK-LIBCALL-NEXT: jal __gnu_h2f_ieee
-; CHECK-LIBCALL-NEXT: move $4, $16
-; CHECK-LIBCALL-NEXT: jal __gnu_f2h_ieee
-; CHECK-LIBCALL-NEXT: add.s $f12, $f20, $f0
-; CHECK-LIBCALL-NEXT: lw $16, 20($sp) # 4-byte Folded Reload
-; CHECK-LIBCALL-NEXT: lw $17, 24($sp) # 4-byte Folded Reload
-; CHECK-LIBCALL-NEXT: lw $ra, 28($sp) # 4-byte Folded Reload
-; CHECK-LIBCALL-NEXT: ldc1 $f20, 32($sp) # 8-byte Folded Reload
-; CHECK-LIBCALL-NEXT: jr $ra
-; CHECK-LIBCALL-NEXT: addiu $sp, $sp, 40
+; MIPS32-LABEL: test_fadd_fadd:
+; MIPS32: # %bb.0:
+; MIPS32-NEXT: addiu $sp, $sp, -40
+; MIPS32-NEXT: sdc1 $f20, 32($sp) # 8-byte Folded Spill
+; MIPS32-NEXT: sw $ra, 28($sp) # 4-byte Folded Spill
+; MIPS32-NEXT: sw $17, 24($sp) # 4-byte Folded Spill
+; MIPS32-NEXT: sw $16, 20($sp) # 4-byte Folded Spill
+; MIPS32-NEXT: move $16, $6
+; MIPS32-NEXT: move $17, $4
+; MIPS32-NEXT: jal __gnu_h2f_ieee
+; MIPS32-NEXT: move $4, $5
+; MIPS32-NEXT: mov.s $f20, $f0
+; MIPS32-NEXT: jal __gnu_h2f_ieee
+; MIPS32-NEXT: move $4, $17
+; MIPS32-NEXT: jal __gnu_f2h_ieee
+; MIPS32-NEXT: add.s $f12, $f0, $f20
+; MIPS32-NEXT: jal __gnu_h2f_ieee
+; MIPS32-NEXT: move $4, $2
+; MIPS32-NEXT: mov.s $f20, $f0
+; MIPS32-NEXT: jal __gnu_h2f_ieee
+; MIPS32-NEXT: move $4, $16
+; MIPS32-NEXT: jal __gnu_f2h_ieee
+; MIPS32-NEXT: add.s $f12, $f20, $f0
+; MIPS32-NEXT: lw $16, 20($sp) # 4-byte Folded Reload
+; MIPS32-NEXT: lw $17, 24($sp) # 4-byte Folded Reload
+; MIPS32-NEXT: lw $ra, 28($sp) # 4-byte Folded Reload
+; MIPS32-NEXT: ldc1 $f20, 32($sp) # 8-byte Folded Reload
+; MIPS32-NEXT: jr $ra
+; MIPS32-NEXT: addiu $sp, $sp, 40
+;
+; MIPS64-LABEL: test_fadd_fadd:
+; MIPS64: # %bb.0:
+; MIPS64-NEXT: daddiu $sp, $sp, -32
+; MIPS64-NEXT: sdc1 $f24, 24($sp) # 8-byte Folded Spill
+; MIPS64-NEXT: sd $ra, 16($sp) # 8-byte Folded Spill
+; MIPS64-NEXT: sd $17, 8($sp) # 8-byte Folded Spill
+; MIPS64-NEXT: sd $16, 0($sp) # 8-byte Folded Spill
+; MIPS64-NEXT: move $16, $6
+; MIPS64-NEXT: move $17, $4
+; MIPS64-NEXT: jal __gnu_h2f_ieee
+; MIPS64-NEXT: sll $4, $5, 0
+; MIPS64-NEXT: mov.s $f24, $f0
+; MIPS64-NEXT: jal __gnu_h2f_ieee
+; MIPS64-NEXT: sll $4, $17, 0
+; MIPS64-NEXT: jal __gnu_f2h_ieee
+; MIPS64-NEXT: add.s $f12, $f0, $f24
+; MIPS64-NEXT: jal __gnu_h2f_ieee
+; MIPS64-NEXT: sll $4, $2, 0
+; MIPS64-NEXT: mov.s $f24, $f0
+; MIPS64-NEXT: jal __gnu_h2f_ieee
+; MIPS64-NEXT: sll $4, $16, 0
+; MIPS64-NEXT: jal __gnu_f2h_ieee
+; MIPS64-NEXT: add.s $f12, $f24, $f0
+; MIPS64-NEXT: ld $16, 0($sp) # 8-byte Folded Reload
+; MIPS64-NEXT: ld $17, 8($sp) # 8-byte Folded Reload
+; MIPS64-NEXT: ld $ra, 16($sp) # 8-byte Folded Reload
+; MIPS64-NEXT: ldc1 $f24, 24($sp) # 8-byte Folded Reload
+; MIPS64-NEXT: jr $ra
+; MIPS64-NEXT: daddiu $sp, $sp, 32
%d = fadd half %a, %b
%e = fadd half %d, %c
ret half %e
}
define half @to_half(i16 %bits) nounwind {
-; CHECK-LIBCALL-LABEL: to_half:
-; CHECK-LIBCALL: # %bb.0:
-; CHECK-LIBCALL-NEXT: jr $ra
-; CHECK-LIBCALL-NEXT: move $2, $4
+; MIPS32-LABEL: to_half:
+; MIPS32: # %bb.0:
+; MIPS32-NEXT: jr $ra
+; MIPS32-NEXT: move $2, $4
+;
+; MIPS64-LABEL: to_half:
+; MIPS64: # %bb.0:
+; MIPS64-NEXT: jr $ra
+; MIPS64-NEXT: sll $2, $4, 0
%f = bitcast i16 %bits to half
ret half %f
}
define i16 @from_half(half %f) nounwind {
-; CHECK-LIBCALL-LABEL: from_half:
-; CHECK-LIBCALL: # %bb.0:
-; CHECK-LIBCALL-NEXT: jr $ra
-; CHECK-LIBCALL-NEXT: move $2, $4
+; MIPS32-LABEL: from_half:
+; MIPS32: # %bb.0:
+; MIPS32-NEXT: jr $ra
+; MIPS32-NEXT: move $2, $4
+;
+; MIPS64-LABEL: from_half:
+; MIPS64: # %bb.0:
+; MIPS64-NEXT: jr $ra
+; MIPS64-NEXT: sll $2, $4, 0
%bits = bitcast half %f to i16
ret i16 %bits
}
More information about the llvm-commits
mailing list