[llvm] [X86][GlobalISel] Enable POW/EXP*/LOG* functions with libcall mapping (PR #130328)
via llvm-commits
llvm-commits at lists.llvm.org
Fri Mar 7 10:57:19 PST 2025
https://github.com/JaydeepChauhan14 created https://github.com/llvm/llvm-project/pull/130328
None
>From 4ef7ec624c3be2da883156c4b6ac69519791d7b0 Mon Sep 17 00:00:00 2001
From: Chauhan Jaydeep Ashwinbhai <chauhan.jaydeep.ashwinbhai at intel.com>
Date: Fri, 7 Mar 2025 10:56:28 -0800
Subject: [PATCH] [X86][GlobalISel] Enable POW/EXP*/LOG* functions with libcall
mapping
---
.../lib/Target/X86/GISel/X86LegalizerInfo.cpp | 3 +-
llvm/test/CodeGen/X86/exp10-libcall-names.ll | 29 ++-
llvm/test/CodeGen/X86/finite-libcalls.ll | 165 +++++++++++++-----
3 files changed, 139 insertions(+), 58 deletions(-)
diff --git a/llvm/lib/Target/X86/GISel/X86LegalizerInfo.cpp b/llvm/lib/Target/X86/GISel/X86LegalizerInfo.cpp
index 684580d6fca06..24bf0dd378641 100644
--- a/llvm/lib/Target/X86/GISel/X86LegalizerInfo.cpp
+++ b/llvm/lib/Target/X86/GISel/X86LegalizerInfo.cpp
@@ -101,7 +101,8 @@ X86LegalizerInfo::X86LegalizerInfo(const X86Subtarget &STI,
getActionDefinitionsBuilder({G_LROUND, G_LLROUND, G_FCOS, G_FCOSH, G_FACOS,
G_FSIN, G_FSINH, G_FASIN, G_FTAN, G_FTANH,
- G_FATAN, G_FATAN2})
+ G_FATAN, G_FATAN2, G_FPOW, G_FEXP, G_FEXP2,
+ G_FEXP10, G_FLOG, G_FLOG2, G_FLOG10})
.libcall();
// merge/unmerge
diff --git a/llvm/test/CodeGen/X86/exp10-libcall-names.ll b/llvm/test/CodeGen/X86/exp10-libcall-names.ll
index 94265813adc26..96e3aae408e94 100644
--- a/llvm/test/CodeGen/X86/exp10-libcall-names.ll
+++ b/llvm/test/CodeGen/X86/exp10-libcall-names.ll
@@ -10,8 +10,8 @@
; RUN: llc -mtriple=x86_64-apple-xros8.0 < %s | FileCheck -check-prefix=APPLE %s
; RUN: llc -mtriple=x86_64-apple-driverkit < %s | FileCheck -check-prefix=APPLE %s
; RUN: llc -mtriple=x86_64-apple-driverkit24.0 < %s | FileCheck -check-prefix=APPLE %s
-; RUN: llc < %s -mtriple=i686-linux-gnu -global-isel -global-isel-abort=2 | FileCheck %s --check-prefix=GISEL-X86
-; RUN: llc < %s -mtriple=x86_64-linux-gnu -global-isel -global-isel-abort=2 | FileCheck %s --check-prefix=GISEL-X64
+; RUN: llc < %s -mtriple=i686-linux-gnu -global-isel -global-isel-abort=1 | FileCheck %s --check-prefix=GISEL-X86
+; RUN: llc < %s -mtriple=x86_64-linux-gnu -global-isel -global-isel-abort=1 | FileCheck %s --check-prefix=GISEL-X64
; RUN: not llc -mtriple=x86_64-apple-macos10.8 -filetype=null %s 2>&1 | FileCheck -check-prefix=ERR %s
; Check exp10/exp10f is emitted as __exp10/__exp10f on assorted systems.
@@ -30,15 +30,18 @@ define float @test_exp10_f32(float %x) nounwind {
; GISEL-X86-LABEL: test_exp10_f32:
; GISEL-X86: # %bb.0:
; GISEL-X86-NEXT: subl $12, %esp
-; GISEL-X86-NEXT: flds {{[0-9]+}}(%esp)
-; GISEL-X86-NEXT: fstps (%esp)
+; GISEL-X86-NEXT: movl {{[0-9]+}}(%esp), %eax
+; GISEL-X86-NEXT: movl %eax, (%esp)
; GISEL-X86-NEXT: calll exp10f
; GISEL-X86-NEXT: addl $12, %esp
; GISEL-X86-NEXT: retl
;
; GISEL-X64-LABEL: test_exp10_f32:
; GISEL-X64: # %bb.0:
-; GISEL-X64-NEXT: jmp exp10f at PLT # TAILCALL
+; GISEL-X64-NEXT: pushq %rax
+; GISEL-X64-NEXT: callq exp10f
+; GISEL-X64-NEXT: popq %rax
+; GISEL-X64-NEXT: retq
%ret = call float @llvm.exp10.f32(float %x)
ret float %ret
}
@@ -55,15 +58,23 @@ define double @test_exp10_f64(double %x) nounwind {
; GISEL-X86-LABEL: test_exp10_f64:
; GISEL-X86: # %bb.0:
; GISEL-X86-NEXT: subl $12, %esp
-; GISEL-X86-NEXT: fldl {{[0-9]+}}(%esp)
-; GISEL-X86-NEXT: fstpl (%esp)
+; GISEL-X86-NEXT: leal {{[0-9]+}}(%esp), %eax
+; GISEL-X86-NEXT: movl {{[0-9]+}}(%esp), %ecx
+; GISEL-X86-NEXT: movl 4(%eax), %eax
+; GISEL-X86-NEXT: xorl %edx, %edx
+; GISEL-X86-NEXT: addl %esp, %edx
+; GISEL-X86-NEXT: movl %ecx, (%esp)
+; GISEL-X86-NEXT: movl %eax, 4(%edx)
; GISEL-X86-NEXT: calll exp10
; GISEL-X86-NEXT: addl $12, %esp
; GISEL-X86-NEXT: retl
;
; GISEL-X64-LABEL: test_exp10_f64:
; GISEL-X64: # %bb.0:
-; GISEL-X64-NEXT: jmp exp10 at PLT # TAILCALL
+; GISEL-X64-NEXT: pushq %rax
+; GISEL-X64-NEXT: callq exp10
+; GISEL-X64-NEXT: popq %rax
+; GISEL-X64-NEXT: retq
%ret = call double @llvm.exp10.f64(double %x)
ret double %ret
}
@@ -101,7 +112,7 @@ define x86_fp80 @test_exp10_f80(x86_fp80 %x) nounwind {
; GISEL-X64-NEXT: subq $24, %rsp
; GISEL-X64-NEXT: fldt {{[0-9]+}}(%rsp)
; GISEL-X64-NEXT: fstpt (%rsp)
-; GISEL-X64-NEXT: callq exp10l at PLT
+; GISEL-X64-NEXT: callq exp10l
; GISEL-X64-NEXT: addq $24, %rsp
; GISEL-X64-NEXT: retq
%ret = call x86_fp80 @llvm.exp10.f80(x86_fp80 %x)
diff --git a/llvm/test/CodeGen/X86/finite-libcalls.ll b/llvm/test/CodeGen/X86/finite-libcalls.ll
index 9d0e188259c17..ea2695dc96997 100644
--- a/llvm/test/CodeGen/X86/finite-libcalls.ll
+++ b/llvm/test/CodeGen/X86/finite-libcalls.ll
@@ -2,8 +2,8 @@
; RUN: llc < %s -mtriple=x86_64-pc-linux-gnu | FileCheck %s --check-prefix=GNU
; RUN: llc < %s -mtriple=x86_64-pc-windows-msvc | FileCheck %s --check-prefix=WIN
; RUN: llc < %s -mtriple=x86_64-apple-darwin | FileCheck %s --check-prefix=MAC
-; RUN: llc < %s -mtriple=i686-linux-gnu -global-isel -global-isel-abort=2 | FileCheck %s --check-prefixes=GISEL-X86
-; RUN: llc < %s -mtriple=x86_64-linux-gnu -global-isel -global-isel-abort=2 | FileCheck %s --check-prefixes=GISEL-X64
+; RUN: llc < %s -mtriple=i686-linux-gnu -global-isel -global-isel-abort=1 | FileCheck %s --check-prefixes=GISEL-X86
+; RUN: llc < %s -mtriple=x86_64-linux-gnu -global-isel -global-isel-abort=1 | FileCheck %s --check-prefixes=GISEL-X64
; PR35672 - https://bugs.llvm.org/show_bug.cgi?id=35672
; FIXME: We would not need the function-level attributes if FMF were propagated to DAG nodes for this case.
@@ -24,15 +24,18 @@ define float @exp_f32(float %x) #0 {
; GISEL-X86-LABEL: exp_f32:
; GISEL-X86: # %bb.0:
; GISEL-X86-NEXT: subl $12, %esp
-; GISEL-X86-NEXT: flds {{[0-9]+}}(%esp)
-; GISEL-X86-NEXT: fstps (%esp)
+; GISEL-X86-NEXT: movl {{[0-9]+}}(%esp), %eax
+; GISEL-X86-NEXT: movl %eax, (%esp)
; GISEL-X86-NEXT: calll expf
; GISEL-X86-NEXT: addl $12, %esp
; GISEL-X86-NEXT: retl
;
; GISEL-X64-LABEL: exp_f32:
; GISEL-X64: # %bb.0:
-; GISEL-X64-NEXT: jmp expf at PLT # TAILCALL
+; GISEL-X64-NEXT: pushq %rax
+; GISEL-X64-NEXT: callq expf
+; GISEL-X64-NEXT: popq %rax
+; GISEL-X64-NEXT: retq
%r = tail call nnan ninf float @llvm.exp.f32(float %x)
ret float %r
}
@@ -53,15 +56,23 @@ define double @exp_f64(double %x) #0 {
; GISEL-X86-LABEL: exp_f64:
; GISEL-X86: # %bb.0:
; GISEL-X86-NEXT: subl $12, %esp
-; GISEL-X86-NEXT: fldl {{[0-9]+}}(%esp)
-; GISEL-X86-NEXT: fstpl (%esp)
+; GISEL-X86-NEXT: leal {{[0-9]+}}(%esp), %eax
+; GISEL-X86-NEXT: movl {{[0-9]+}}(%esp), %ecx
+; GISEL-X86-NEXT: movl 4(%eax), %eax
+; GISEL-X86-NEXT: xorl %edx, %edx
+; GISEL-X86-NEXT: addl %esp, %edx
+; GISEL-X86-NEXT: movl %ecx, (%esp)
+; GISEL-X86-NEXT: movl %eax, 4(%edx)
; GISEL-X86-NEXT: calll exp
; GISEL-X86-NEXT: addl $12, %esp
; GISEL-X86-NEXT: retl
;
; GISEL-X64-LABEL: exp_f64:
; GISEL-X64: # %bb.0:
-; GISEL-X64-NEXT: jmp exp at PLT # TAILCALL
+; GISEL-X64-NEXT: pushq %rax
+; GISEL-X64-NEXT: callq exp
+; GISEL-X64-NEXT: popq %rax
+; GISEL-X64-NEXT: retq
%r = tail call nnan ninf double @llvm.exp.f64(double %x)
ret double %r
}
@@ -116,7 +127,7 @@ define x86_fp80 @exp_f80(x86_fp80 %x) #0 {
; GISEL-X64-NEXT: subq $24, %rsp
; GISEL-X64-NEXT: fldt {{[0-9]+}}(%rsp)
; GISEL-X64-NEXT: fstpt (%rsp)
-; GISEL-X64-NEXT: callq expl at PLT
+; GISEL-X64-NEXT: callq expl
; GISEL-X64-NEXT: addq $24, %rsp
; GISEL-X64-NEXT: retq
%r = tail call nnan ninf x86_fp80 @llvm.exp.f80(x86_fp80 %x)
@@ -139,15 +150,18 @@ define float @exp2_f32(float %x) #0 {
; GISEL-X86-LABEL: exp2_f32:
; GISEL-X86: # %bb.0:
; GISEL-X86-NEXT: subl $12, %esp
-; GISEL-X86-NEXT: flds {{[0-9]+}}(%esp)
-; GISEL-X86-NEXT: fstps (%esp)
+; GISEL-X86-NEXT: movl {{[0-9]+}}(%esp), %eax
+; GISEL-X86-NEXT: movl %eax, (%esp)
; GISEL-X86-NEXT: calll exp2f
; GISEL-X86-NEXT: addl $12, %esp
; GISEL-X86-NEXT: retl
;
; GISEL-X64-LABEL: exp2_f32:
; GISEL-X64: # %bb.0:
-; GISEL-X64-NEXT: jmp exp2f at PLT # TAILCALL
+; GISEL-X64-NEXT: pushq %rax
+; GISEL-X64-NEXT: callq exp2f
+; GISEL-X64-NEXT: popq %rax
+; GISEL-X64-NEXT: retq
%r = tail call nnan ninf float @llvm.exp2.f32(float %x)
ret float %r
}
@@ -168,15 +182,23 @@ define double @exp2_f64(double %x) #0 {
; GISEL-X86-LABEL: exp2_f64:
; GISEL-X86: # %bb.0:
; GISEL-X86-NEXT: subl $12, %esp
-; GISEL-X86-NEXT: fldl {{[0-9]+}}(%esp)
-; GISEL-X86-NEXT: fstpl (%esp)
+; GISEL-X86-NEXT: leal {{[0-9]+}}(%esp), %eax
+; GISEL-X86-NEXT: movl {{[0-9]+}}(%esp), %ecx
+; GISEL-X86-NEXT: movl 4(%eax), %eax
+; GISEL-X86-NEXT: xorl %edx, %edx
+; GISEL-X86-NEXT: addl %esp, %edx
+; GISEL-X86-NEXT: movl %ecx, (%esp)
+; GISEL-X86-NEXT: movl %eax, 4(%edx)
; GISEL-X86-NEXT: calll exp2
; GISEL-X86-NEXT: addl $12, %esp
; GISEL-X86-NEXT: retl
;
; GISEL-X64-LABEL: exp2_f64:
; GISEL-X64: # %bb.0:
-; GISEL-X64-NEXT: jmp exp2 at PLT # TAILCALL
+; GISEL-X64-NEXT: pushq %rax
+; GISEL-X64-NEXT: callq exp2
+; GISEL-X64-NEXT: popq %rax
+; GISEL-X64-NEXT: retq
%r = tail call nnan ninf double @llvm.exp2.f64(double %x)
ret double %r
}
@@ -231,7 +253,7 @@ define x86_fp80 @exp2_f80(x86_fp80 %x) #0 {
; GISEL-X64-NEXT: subq $24, %rsp
; GISEL-X64-NEXT: fldt {{[0-9]+}}(%rsp)
; GISEL-X64-NEXT: fstpt (%rsp)
-; GISEL-X64-NEXT: callq exp2l at PLT
+; GISEL-X64-NEXT: callq exp2l
; GISEL-X64-NEXT: addq $24, %rsp
; GISEL-X64-NEXT: retq
%r = tail call nnan ninf x86_fp80 @llvm.exp2.f80(x86_fp80 %x)
@@ -254,15 +276,18 @@ define float @log_f32(float %x) #0 {
; GISEL-X86-LABEL: log_f32:
; GISEL-X86: # %bb.0:
; GISEL-X86-NEXT: subl $12, %esp
-; GISEL-X86-NEXT: flds {{[0-9]+}}(%esp)
-; GISEL-X86-NEXT: fstps (%esp)
+; GISEL-X86-NEXT: movl {{[0-9]+}}(%esp), %eax
+; GISEL-X86-NEXT: movl %eax, (%esp)
; GISEL-X86-NEXT: calll logf
; GISEL-X86-NEXT: addl $12, %esp
; GISEL-X86-NEXT: retl
;
; GISEL-X64-LABEL: log_f32:
; GISEL-X64: # %bb.0:
-; GISEL-X64-NEXT: jmp logf at PLT # TAILCALL
+; GISEL-X64-NEXT: pushq %rax
+; GISEL-X64-NEXT: callq logf
+; GISEL-X64-NEXT: popq %rax
+; GISEL-X64-NEXT: retq
%r = tail call nnan ninf float @llvm.log.f32(float %x)
ret float %r
}
@@ -283,15 +308,23 @@ define double @log_f64(double %x) #0 {
; GISEL-X86-LABEL: log_f64:
; GISEL-X86: # %bb.0:
; GISEL-X86-NEXT: subl $12, %esp
-; GISEL-X86-NEXT: fldl {{[0-9]+}}(%esp)
-; GISEL-X86-NEXT: fstpl (%esp)
+; GISEL-X86-NEXT: leal {{[0-9]+}}(%esp), %eax
+; GISEL-X86-NEXT: movl {{[0-9]+}}(%esp), %ecx
+; GISEL-X86-NEXT: movl 4(%eax), %eax
+; GISEL-X86-NEXT: xorl %edx, %edx
+; GISEL-X86-NEXT: addl %esp, %edx
+; GISEL-X86-NEXT: movl %ecx, (%esp)
+; GISEL-X86-NEXT: movl %eax, 4(%edx)
; GISEL-X86-NEXT: calll log
; GISEL-X86-NEXT: addl $12, %esp
; GISEL-X86-NEXT: retl
;
; GISEL-X64-LABEL: log_f64:
; GISEL-X64: # %bb.0:
-; GISEL-X64-NEXT: jmp log at PLT # TAILCALL
+; GISEL-X64-NEXT: pushq %rax
+; GISEL-X64-NEXT: callq log
+; GISEL-X64-NEXT: popq %rax
+; GISEL-X64-NEXT: retq
%r = tail call nnan ninf double @llvm.log.f64(double %x)
ret double %r
}
@@ -346,7 +379,7 @@ define x86_fp80 @log_f80(x86_fp80 %x) #0 {
; GISEL-X64-NEXT: subq $24, %rsp
; GISEL-X64-NEXT: fldt {{[0-9]+}}(%rsp)
; GISEL-X64-NEXT: fstpt (%rsp)
-; GISEL-X64-NEXT: callq logl at PLT
+; GISEL-X64-NEXT: callq logl
; GISEL-X64-NEXT: addq $24, %rsp
; GISEL-X64-NEXT: retq
%r = tail call nnan ninf x86_fp80 @llvm.log.f80(x86_fp80 %x)
@@ -369,15 +402,18 @@ define float @log2_f32(float %x) #0 {
; GISEL-X86-LABEL: log2_f32:
; GISEL-X86: # %bb.0:
; GISEL-X86-NEXT: subl $12, %esp
-; GISEL-X86-NEXT: flds {{[0-9]+}}(%esp)
-; GISEL-X86-NEXT: fstps (%esp)
+; GISEL-X86-NEXT: movl {{[0-9]+}}(%esp), %eax
+; GISEL-X86-NEXT: movl %eax, (%esp)
; GISEL-X86-NEXT: calll log2f
; GISEL-X86-NEXT: addl $12, %esp
; GISEL-X86-NEXT: retl
;
; GISEL-X64-LABEL: log2_f32:
; GISEL-X64: # %bb.0:
-; GISEL-X64-NEXT: jmp log2f at PLT # TAILCALL
+; GISEL-X64-NEXT: pushq %rax
+; GISEL-X64-NEXT: callq log2f
+; GISEL-X64-NEXT: popq %rax
+; GISEL-X64-NEXT: retq
%r = tail call nnan ninf float @llvm.log2.f32(float %x)
ret float %r
}
@@ -398,15 +434,23 @@ define double @log2_f64(double %x) #0 {
; GISEL-X86-LABEL: log2_f64:
; GISEL-X86: # %bb.0:
; GISEL-X86-NEXT: subl $12, %esp
-; GISEL-X86-NEXT: fldl {{[0-9]+}}(%esp)
-; GISEL-X86-NEXT: fstpl (%esp)
+; GISEL-X86-NEXT: leal {{[0-9]+}}(%esp), %eax
+; GISEL-X86-NEXT: movl {{[0-9]+}}(%esp), %ecx
+; GISEL-X86-NEXT: movl 4(%eax), %eax
+; GISEL-X86-NEXT: xorl %edx, %edx
+; GISEL-X86-NEXT: addl %esp, %edx
+; GISEL-X86-NEXT: movl %ecx, (%esp)
+; GISEL-X86-NEXT: movl %eax, 4(%edx)
; GISEL-X86-NEXT: calll log2
; GISEL-X86-NEXT: addl $12, %esp
; GISEL-X86-NEXT: retl
;
; GISEL-X64-LABEL: log2_f64:
; GISEL-X64: # %bb.0:
-; GISEL-X64-NEXT: jmp log2 at PLT # TAILCALL
+; GISEL-X64-NEXT: pushq %rax
+; GISEL-X64-NEXT: callq log2
+; GISEL-X64-NEXT: popq %rax
+; GISEL-X64-NEXT: retq
%r = tail call nnan ninf double @llvm.log2.f64(double %x)
ret double %r
}
@@ -461,7 +505,7 @@ define x86_fp80 @log2_f80(x86_fp80 %x) #0 {
; GISEL-X64-NEXT: subq $24, %rsp
; GISEL-X64-NEXT: fldt {{[0-9]+}}(%rsp)
; GISEL-X64-NEXT: fstpt (%rsp)
-; GISEL-X64-NEXT: callq log2l at PLT
+; GISEL-X64-NEXT: callq log2l
; GISEL-X64-NEXT: addq $24, %rsp
; GISEL-X64-NEXT: retq
%r = tail call nnan ninf x86_fp80 @llvm.log2.f80(x86_fp80 %x)
@@ -484,15 +528,18 @@ define float @log10_f32(float %x) #0 {
; GISEL-X86-LABEL: log10_f32:
; GISEL-X86: # %bb.0:
; GISEL-X86-NEXT: subl $12, %esp
-; GISEL-X86-NEXT: flds {{[0-9]+}}(%esp)
-; GISEL-X86-NEXT: fstps (%esp)
+; GISEL-X86-NEXT: movl {{[0-9]+}}(%esp), %eax
+; GISEL-X86-NEXT: movl %eax, (%esp)
; GISEL-X86-NEXT: calll log10f
; GISEL-X86-NEXT: addl $12, %esp
; GISEL-X86-NEXT: retl
;
; GISEL-X64-LABEL: log10_f32:
; GISEL-X64: # %bb.0:
-; GISEL-X64-NEXT: jmp log10f at PLT # TAILCALL
+; GISEL-X64-NEXT: pushq %rax
+; GISEL-X64-NEXT: callq log10f
+; GISEL-X64-NEXT: popq %rax
+; GISEL-X64-NEXT: retq
%r = tail call nnan ninf float @llvm.log10.f32(float %x)
ret float %r
}
@@ -513,15 +560,23 @@ define double @log10_f64(double %x) #0 {
; GISEL-X86-LABEL: log10_f64:
; GISEL-X86: # %bb.0:
; GISEL-X86-NEXT: subl $12, %esp
-; GISEL-X86-NEXT: fldl {{[0-9]+}}(%esp)
-; GISEL-X86-NEXT: fstpl (%esp)
+; GISEL-X86-NEXT: leal {{[0-9]+}}(%esp), %eax
+; GISEL-X86-NEXT: movl {{[0-9]+}}(%esp), %ecx
+; GISEL-X86-NEXT: movl 4(%eax), %eax
+; GISEL-X86-NEXT: xorl %edx, %edx
+; GISEL-X86-NEXT: addl %esp, %edx
+; GISEL-X86-NEXT: movl %ecx, (%esp)
+; GISEL-X86-NEXT: movl %eax, 4(%edx)
; GISEL-X86-NEXT: calll log10
; GISEL-X86-NEXT: addl $12, %esp
; GISEL-X86-NEXT: retl
;
; GISEL-X64-LABEL: log10_f64:
; GISEL-X64: # %bb.0:
-; GISEL-X64-NEXT: jmp log10 at PLT # TAILCALL
+; GISEL-X64-NEXT: pushq %rax
+; GISEL-X64-NEXT: callq log10
+; GISEL-X64-NEXT: popq %rax
+; GISEL-X64-NEXT: retq
%r = tail call nnan ninf double @llvm.log10.f64(double %x)
ret double %r
}
@@ -576,7 +631,7 @@ define x86_fp80 @log10_f80(x86_fp80 %x) #0 {
; GISEL-X64-NEXT: subq $24, %rsp
; GISEL-X64-NEXT: fldt {{[0-9]+}}(%rsp)
; GISEL-X64-NEXT: fstpt (%rsp)
-; GISEL-X64-NEXT: callq log10l at PLT
+; GISEL-X64-NEXT: callq log10l
; GISEL-X64-NEXT: addq $24, %rsp
; GISEL-X64-NEXT: retq
%r = tail call nnan ninf x86_fp80 @llvm.log10.f80(x86_fp80 %x)
@@ -602,17 +657,20 @@ define float @pow_f32(float %x) #0 {
; GISEL-X86-LABEL: pow_f32:
; GISEL-X86: # %bb.0:
; GISEL-X86-NEXT: subl $12, %esp
-; GISEL-X86-NEXT: flds {{[0-9]+}}(%esp)
-; GISEL-X86-NEXT: fsts {{[0-9]+}}(%esp)
-; GISEL-X86-NEXT: fstps (%esp)
+; GISEL-X86-NEXT: movl {{[0-9]+}}(%esp), %eax
+; GISEL-X86-NEXT: movl %eax, (%esp)
+; GISEL-X86-NEXT: movl %eax, {{[0-9]+}}(%esp)
; GISEL-X86-NEXT: calll powf
; GISEL-X86-NEXT: addl $12, %esp
; GISEL-X86-NEXT: retl
;
; GISEL-X64-LABEL: pow_f32:
; GISEL-X64: # %bb.0:
+; GISEL-X64-NEXT: pushq %rax
; GISEL-X64-NEXT: movaps %xmm0, %xmm1
-; GISEL-X64-NEXT: jmp powf at PLT # TAILCALL
+; GISEL-X64-NEXT: callq powf
+; GISEL-X64-NEXT: popq %rax
+; GISEL-X64-NEXT: retq
%r = tail call nnan ninf float @llvm.pow.f32(float %x, float %x)
ret float %r
}
@@ -636,17 +694,28 @@ define double @pow_f64(double %x) #0 {
; GISEL-X86-LABEL: pow_f64:
; GISEL-X86: # %bb.0:
; GISEL-X86-NEXT: subl $28, %esp
-; GISEL-X86-NEXT: fldl {{[0-9]+}}(%esp)
-; GISEL-X86-NEXT: fstl {{[0-9]+}}(%esp)
-; GISEL-X86-NEXT: fstpl (%esp)
+; GISEL-X86-NEXT: leal {{[0-9]+}}(%esp), %eax
+; GISEL-X86-NEXT: movl {{[0-9]+}}(%esp), %ecx
+; GISEL-X86-NEXT: movl 4(%eax), %eax
+; GISEL-X86-NEXT: xorl %edx, %edx
+; GISEL-X86-NEXT: addl %esp, %edx
+; GISEL-X86-NEXT: movl %ecx, (%esp)
+; GISEL-X86-NEXT: movl %eax, 4(%edx)
+; GISEL-X86-NEXT: movl $8, %edx
+; GISEL-X86-NEXT: addl %esp, %edx
+; GISEL-X86-NEXT: movl %ecx, {{[0-9]+}}(%esp)
+; GISEL-X86-NEXT: movl %eax, 4(%edx)
; GISEL-X86-NEXT: calll pow
; GISEL-X86-NEXT: addl $28, %esp
; GISEL-X86-NEXT: retl
;
; GISEL-X64-LABEL: pow_f64:
; GISEL-X64: # %bb.0:
+; GISEL-X64-NEXT: pushq %rax
; GISEL-X64-NEXT: movaps %xmm0, %xmm1
-; GISEL-X64-NEXT: jmp pow at PLT # TAILCALL
+; GISEL-X64-NEXT: callq pow
+; GISEL-X64-NEXT: popq %rax
+; GISEL-X64-NEXT: retq
%r = tail call nnan ninf double @llvm.pow.f64(double %x, double %x)
ret double %r
}
@@ -699,8 +768,8 @@ define x86_fp80 @pow_f80(x86_fp80 %x) #0 {
; GISEL-X86-NEXT: subl $28, %esp
; GISEL-X86-NEXT: fldt {{[0-9]+}}(%esp)
; GISEL-X86-NEXT: fld %st(0)
-; GISEL-X86-NEXT: fstpt {{[0-9]+}}(%esp)
; GISEL-X86-NEXT: fstpt (%esp)
+; GISEL-X86-NEXT: fstpt {{[0-9]+}}(%esp)
; GISEL-X86-NEXT: calll powl
; GISEL-X86-NEXT: addl $28, %esp
; GISEL-X86-NEXT: retl
@@ -710,9 +779,9 @@ define x86_fp80 @pow_f80(x86_fp80 %x) #0 {
; GISEL-X64-NEXT: subq $40, %rsp
; GISEL-X64-NEXT: fldt {{[0-9]+}}(%rsp)
; GISEL-X64-NEXT: fld %st(0)
-; GISEL-X64-NEXT: fstpt {{[0-9]+}}(%rsp)
; GISEL-X64-NEXT: fstpt (%rsp)
-; GISEL-X64-NEXT: callq powl at PLT
+; GISEL-X64-NEXT: fstpt {{[0-9]+}}(%rsp)
+; GISEL-X64-NEXT: callq powl
; GISEL-X64-NEXT: addq $40, %rsp
; GISEL-X64-NEXT: retq
%r = tail call nnan ninf x86_fp80 @llvm.pow.f80(x86_fp80 %x, x86_fp80 %x)
More information about the llvm-commits
mailing list