mirror of
https://github.com/RPCSX/llvm.git
synced 2024-12-23 04:28:40 +00:00
CodeGen: ensure that libcalls are always AAPCS CC
All of the builtins are designed to be invoked with ARM AAPCS CC even on ARM AAPCS VFP CC hosts. Tweak the default initialisation to ARM AAPCS CC rather than C CC for ARM/thumb targets. The changes to the tests are necessary to ensure that the calling convention for the lowered library calls are honoured. Furthermore, these adjustments cause certain branch invocations to change to branch-and-link since the returned value needs to be moved across registers (d0 -> r0, r1). git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@280683 91177308-0d34-0410-b5e6-96231b3b80d8
This commit is contained in:
Saleem Abdulrasool
parent
610e45c3d2
commit
7471df8d42
@ -488,12 +488,11 @@ static void InitLibcallNames(const char **Names, const Triple &TT) {
|
||||
Names[RTLIB::DEOPTIMIZE] = "__llvm_deoptimize";
|
||||
}
|
||||
|
||||
/// InitLibcallCallingConvs - Set default libcall CallingConvs.
|
||||
///
|
||||
static void InitLibcallCallingConvs(CallingConv::ID *CCs) {
|
||||
for (int i = 0; i < RTLIB::UNKNOWN_LIBCALL; ++i) {
|
||||
CCs[i] = CallingConv::C;
|
||||
}
|
||||
/// Set default libcall CallingConvs.
|
||||
static void InitLibcallCallingConvs(CallingConv::ID *CCs, const Triple &T) {
|
||||
bool IsARM = T.getArch() == Triple::arm || T.getArch() == Triple::thumb;
|
||||
for (int LC = 0; LC < RTLIB::UNKNOWN_LIBCALL; ++LC)
|
||||
CCs[LC] = IsARM ? CallingConv::ARM_AAPCS : CallingConv::C;
|
||||
}
|
||||
|
||||
/// getFPEXT - Return the FPEXT_*_* value for the given types, or
|
||||
@ -835,7 +834,7 @@ TargetLoweringBase::TargetLoweringBase(const TargetMachine &tm) : TM(tm) {
|
||||
|
||||
InitLibcallNames(LibcallRoutineNames, TM.getTargetTriple());
|
||||
InitCmpLibcallCCs(CmpLibcallCCs);
|
||||
InitLibcallCallingConvs(LibcallCallingConvs);
|
||||
InitLibcallCallingConvs(LibcallCallingConvs, TM.getTargetTriple());
|
||||
}
|
||||
|
||||
void TargetLoweringBase::initActions() {
|
||||
|
||||
@ -4,115 +4,116 @@
|
||||
|
||||
; CHECK-LABEL: test1
|
||||
; CHECK: vrintm.f32
|
||||
define float @test1(float %a) {
|
||||
define arm_aapcs_vfpcc float @test1(float %a) {
|
||||
entry:
|
||||
%call = call float @floorf(float %a) nounwind readnone
|
||||
%call = call arm_aapcs_vfpcc float @floorf(float %a) nounwind readnone
|
||||
ret float %call
|
||||
}
|
||||
|
||||
; CHECK-LABEL: test2
|
||||
; SP: b floor
|
||||
; SP: bl floor
|
||||
; DP: vrintm.f64
|
||||
define double @test2(double %a) {
|
||||
define arm_aapcs_vfpcc double @test2(double %a) {
|
||||
entry:
|
||||
%call = call double @floor(double %a) nounwind readnone
|
||||
%call = call arm_aapcscc double @floor(double %a) nounwind readnone
|
||||
ret double %call
|
||||
}
|
||||
|
||||
; CHECK-LABEL: test3
|
||||
; CHECK: vrintp.f32
|
||||
define float @test3(float %a) {
|
||||
define arm_aapcs_vfpcc float @test3(float %a) {
|
||||
entry:
|
||||
%call = call float @ceilf(float %a) nounwind readnone
|
||||
%call = call arm_aapcs_vfpcc float @ceilf(float %a) nounwind readnone
|
||||
ret float %call
|
||||
}
|
||||
|
||||
; CHECK-LABEL: test4
|
||||
; SP: b ceil
|
||||
; SP: bl ceil
|
||||
; DP: vrintp.f64
|
||||
define double @test4(double %a) {
|
||||
define arm_aapcs_vfpcc double @test4(double %a) {
|
||||
entry:
|
||||
%call = call double @ceil(double %a) nounwind readnone
|
||||
%call = call arm_aapcscc double @ceil(double %a) nounwind readnone
|
||||
ret double %call
|
||||
}
|
||||
|
||||
; CHECK-LABEL: test5
|
||||
; CHECK: vrinta.f32
|
||||
define float @test5(float %a) {
|
||||
define arm_aapcs_vfpcc float @test5(float %a) {
|
||||
entry:
|
||||
%call = call float @roundf(float %a) nounwind readnone
|
||||
%call = call arm_aapcs_vfpcc float @roundf(float %a) nounwind readnone
|
||||
ret float %call
|
||||
}
|
||||
|
||||
; CHECK-LABEL: test6
|
||||
; SP: b round
|
||||
; SP: bl round
|
||||
; DP: vrinta.f64
|
||||
define double @test6(double %a) {
|
||||
define arm_aapcs_vfpcc double @test6(double %a) {
|
||||
entry:
|
||||
%call = call double @round(double %a) nounwind readnone
|
||||
%call = call arm_aapcscc double @round(double %a) nounwind readnone
|
||||
ret double %call
|
||||
}
|
||||
|
||||
; CHECK-LABEL: test7
|
||||
; CHECK: vrintz.f32
|
||||
define float @test7(float %a) {
|
||||
define arm_aapcs_vfpcc float @test7(float %a) {
|
||||
entry:
|
||||
%call = call float @truncf(float %a) nounwind readnone
|
||||
%call = call arm_aapcs_vfpcc float @truncf(float %a) nounwind readnone
|
||||
ret float %call
|
||||
}
|
||||
|
||||
; CHECK-LABEL: test8
|
||||
; SP: b trunc
|
||||
; SP: bl trunc
|
||||
; DP: vrintz.f64
|
||||
define double @test8(double %a) {
|
||||
define arm_aapcs_vfpcc double @test8(double %a) {
|
||||
entry:
|
||||
%call = call double @trunc(double %a) nounwind readnone
|
||||
%call = call arm_aapcscc double @trunc(double %a) nounwind readnone
|
||||
ret double %call
|
||||
}
|
||||
|
||||
; CHECK-LABEL: test9
|
||||
; CHECK: vrintr.f32
|
||||
define float @test9(float %a) {
|
||||
define arm_aapcs_vfpcc float @test9(float %a) {
|
||||
entry:
|
||||
%call = call float @nearbyintf(float %a) nounwind readnone
|
||||
%call = call arm_aapcs_vfpcc float @nearbyintf(float %a) nounwind readnone
|
||||
ret float %call
|
||||
}
|
||||
|
||||
; CHECK-LABEL: test10
|
||||
; SP: b nearbyint
|
||||
; SP: bl nearbyint
|
||||
; DP: vrintr.f64
|
||||
define double @test10(double %a) {
|
||||
define arm_aapcs_vfpcc double @test10(double %a) {
|
||||
entry:
|
||||
%call = call double @nearbyint(double %a) nounwind readnone
|
||||
%call = call arm_aapcscc double @nearbyint(double %a) nounwind readnone
|
||||
ret double %call
|
||||
}
|
||||
|
||||
; CHECK-LABEL: test11
|
||||
; CHECK: vrintx.f32
|
||||
define float @test11(float %a) {
|
||||
define arm_aapcs_vfpcc float @test11(float %a) {
|
||||
entry:
|
||||
%call = call float @rintf(float %a) nounwind readnone
|
||||
%call = call arm_aapcs_vfpcc float @rintf(float %a) nounwind readnone
|
||||
ret float %call
|
||||
}
|
||||
|
||||
; CHECK-LABEL: test12
|
||||
; SP: b rint
|
||||
; SP: bl rint
|
||||
; DP: vrintx.f64
|
||||
define double @test12(double %a) {
|
||||
define arm_aapcs_vfpcc double @test12(double %a) {
|
||||
entry:
|
||||
%call = call double @rint(double %a) nounwind readnone
|
||||
%call = call arm_aapcscc double @rint(double %a) nounwind readnone
|
||||
ret double %call
|
||||
}
|
||||
|
||||
declare float @floorf(float) nounwind readnone
|
||||
declare double @floor(double) nounwind readnone
|
||||
declare float @ceilf(float) nounwind readnone
|
||||
declare double @ceil(double) nounwind readnone
|
||||
declare float @roundf(float) nounwind readnone
|
||||
declare double @round(double) nounwind readnone
|
||||
declare float @truncf(float) nounwind readnone
|
||||
declare double @trunc(double) nounwind readnone
|
||||
declare float @nearbyintf(float) nounwind readnone
|
||||
declare double @nearbyint(double) nounwind readnone
|
||||
declare float @rintf(float) nounwind readnone
|
||||
declare double @rint(double) nounwind readnone
|
||||
declare arm_aapcs_vfpcc float @floorf(float) nounwind readnone
|
||||
declare arm_aapcscc double @floor(double) nounwind readnone
|
||||
declare arm_aapcs_vfpcc float @ceilf(float) nounwind readnone
|
||||
declare arm_aapcscc double @ceil(double) nounwind readnone
|
||||
declare arm_aapcs_vfpcc float @roundf(float) nounwind readnone
|
||||
declare arm_aapcscc double @round(double) nounwind readnone
|
||||
declare arm_aapcs_vfpcc float @truncf(float) nounwind readnone
|
||||
declare arm_aapcscc double @trunc(double) nounwind readnone
|
||||
declare arm_aapcs_vfpcc float @nearbyintf(float) nounwind readnone
|
||||
declare arm_aapcscc double @nearbyint(double) nounwind readnone
|
||||
declare arm_aapcs_vfpcc float @rintf(float) nounwind readnone
|
||||
declare arm_aapcscc double @rint(double) nounwind readnone
|
||||
|
||||
|
||||
@ -1,37 +1,38 @@
|
||||
; RUN: llc -mtriple=thumbv7-linux-gnueabihf %s -o - | FileCheck %s
|
||||
|
||||
define i64 @test_mismatched_call(double %in) {
|
||||
define arm_aapcs_vfpcc i64 @test_mismatched_call(double %in) {
|
||||
; CHECK-LABEL: test_mismatched_call:
|
||||
; CHECK: bl floor
|
||||
; CHECK: vmov r0, r1, d0
|
||||
; CHECK: bl floor
|
||||
|
||||
%val = tail call double @floor(double %in)
|
||||
%val = tail call arm_aapcscc double @floor(double %in)
|
||||
%res = bitcast double %val to i64
|
||||
ret i64 %res
|
||||
}
|
||||
|
||||
define double @test_matched_call(double %in) {
|
||||
define arm_aapcs_vfpcc double @test_matched_call(double %in) {
|
||||
; CHECK-LABEL: test_matched_call:
|
||||
; CHECK: b floor
|
||||
; CHECK: b _floor
|
||||
|
||||
%val = tail call double @floor(double %in)
|
||||
%val = tail call arm_aapcs_vfpcc double @_floor(double %in)
|
||||
ret double %val
|
||||
}
|
||||
|
||||
define void @test_irrelevant_call(double %in) {
|
||||
define arm_aapcs_vfpcc void @test_irrelevant_call(double %in) {
|
||||
; CHECK-LABEL: test_irrelevant_call:
|
||||
; CHECK-NOT: bl floor
|
||||
|
||||
%val = tail call double @floor(double %in)
|
||||
%val = tail call arm_aapcscc double @floor(double %in)
|
||||
ret void
|
||||
}
|
||||
|
||||
define arm_aapcscc double @test_callingconv(double %in) {
|
||||
define arm_aapcs_vfpcc double @test_callingconv(double %in) {
|
||||
; CHECK: test_callingconv:
|
||||
; CHECK: bl floor
|
||||
|
||||
%val = tail call double @floor(double %in)
|
||||
%val = tail call arm_aapcscc double @floor(double %in)
|
||||
ret double %val
|
||||
}
|
||||
|
||||
declare double @floor(double) nounwind readonly
|
||||
declare arm_aapcs_vfpcc double @_floor(double) nounwind readonly
|
||||
declare arm_aapcscc double @floor(double) nounwind readonly
|
||||
|
||||
@ -5,107 +5,107 @@
|
||||
; RUN: llc < %s -mtriple=thumbv7-none-eabihf -mcpu=cortex-a7 | FileCheck %s -check-prefix=CHECK -check-prefix=HARD -check-prefix=DP -check-prefix=NEON -check-prefix=VFP4
|
||||
; RUN: llc < %s -mtriple=thumbv7-none-eabihf -mcpu=cortex-a57 | FileCheck %s -check-prefix=CHECK -check-prefix=HARD -check-prefix=DP -check-prefix=NEON -check-prefix=FP-ARMv8
|
||||
|
||||
declare double @llvm.sqrt.f64(double %Val)
|
||||
declare arm_aapcscc double @llvm.sqrt.f64(double %Val)
|
||||
define double @sqrt_d(double %a) {
|
||||
; CHECK-LABEL: sqrt_d:
|
||||
; SOFT: {{(bl|b)}} sqrt
|
||||
; HARD: vsqrt.f64 d0, d0
|
||||
%1 = call double @llvm.sqrt.f64(double %a)
|
||||
%1 = call arm_aapcscc double @llvm.sqrt.f64(double %a)
|
||||
ret double %1
|
||||
}
|
||||
|
||||
declare double @llvm.powi.f64(double %Val, i32 %power)
|
||||
declare arm_aapcscc double @llvm.powi.f64(double %Val, i32 %power)
|
||||
define double @powi_d(double %a, i32 %b) {
|
||||
; CHECK-LABEL: powi_d:
|
||||
; SOFT: {{(bl|b)}} __powidf2
|
||||
; HARD: b __powidf2
|
||||
%1 = call double @llvm.powi.f64(double %a, i32 %b)
|
||||
; HARD: bl __powidf2
|
||||
%1 = call arm_aapcscc double @llvm.powi.f64(double %a, i32 %b)
|
||||
ret double %1
|
||||
}
|
||||
|
||||
declare double @llvm.sin.f64(double %Val)
|
||||
declare arm_aapcscc double @llvm.sin.f64(double %Val)
|
||||
define double @sin_d(double %a) {
|
||||
; CHECK-LABEL: sin_d:
|
||||
; SOFT: {{(bl|b)}} sin
|
||||
; HARD: b sin
|
||||
%1 = call double @llvm.sin.f64(double %a)
|
||||
; HARD: bl sin
|
||||
%1 = call arm_aapcscc double @llvm.sin.f64(double %a)
|
||||
ret double %1
|
||||
}
|
||||
|
||||
declare double @llvm.cos.f64(double %Val)
|
||||
declare arm_aapcscc double @llvm.cos.f64(double %Val)
|
||||
define double @cos_d(double %a) {
|
||||
; CHECK-LABEL: cos_d:
|
||||
; SOFT: {{(bl|b)}} cos
|
||||
; HARD: b cos
|
||||
%1 = call double @llvm.cos.f64(double %a)
|
||||
; HARD: bl cos
|
||||
%1 = call arm_aapcscc double @llvm.cos.f64(double %a)
|
||||
ret double %1
|
||||
}
|
||||
|
||||
declare double @llvm.pow.f64(double %Val, double %power)
|
||||
declare arm_aapcscc double @llvm.pow.f64(double %Val, double %power)
|
||||
define double @pow_d(double %a, double %b) {
|
||||
; CHECK-LABEL: pow_d:
|
||||
; SOFT: {{(bl|b)}} pow
|
||||
; HARD: b pow
|
||||
%1 = call double @llvm.pow.f64(double %a, double %b)
|
||||
; HARD: bl pow
|
||||
%1 = call arm_aapcscc double @llvm.pow.f64(double %a, double %b)
|
||||
ret double %1
|
||||
}
|
||||
|
||||
declare double @llvm.exp.f64(double %Val)
|
||||
declare arm_aapcscc double @llvm.exp.f64(double %Val)
|
||||
define double @exp_d(double %a) {
|
||||
; CHECK-LABEL: exp_d:
|
||||
; SOFT: {{(bl|b)}} exp
|
||||
; HARD: b exp
|
||||
%1 = call double @llvm.exp.f64(double %a)
|
||||
; HARD: bl exp
|
||||
%1 = call arm_aapcscc double @llvm.exp.f64(double %a)
|
||||
ret double %1
|
||||
}
|
||||
|
||||
declare double @llvm.exp2.f64(double %Val)
|
||||
declare arm_aapcscc double @llvm.exp2.f64(double %Val)
|
||||
define double @exp2_d(double %a) {
|
||||
; CHECK-LABEL: exp2_d:
|
||||
; SOFT: {{(bl|b)}} exp2
|
||||
; HARD: b exp2
|
||||
%1 = call double @llvm.exp2.f64(double %a)
|
||||
; HARD: bl exp2
|
||||
%1 = call arm_aapcscc double @llvm.exp2.f64(double %a)
|
||||
ret double %1
|
||||
}
|
||||
|
||||
declare double @llvm.log.f64(double %Val)
|
||||
declare arm_aapcscc double @llvm.log.f64(double %Val)
|
||||
define double @log_d(double %a) {
|
||||
; CHECK-LABEL: log_d:
|
||||
; SOFT: {{(bl|b)}} log
|
||||
; HARD: b log
|
||||
%1 = call double @llvm.log.f64(double %a)
|
||||
; HARD: bl log
|
||||
%1 = call arm_aapcscc double @llvm.log.f64(double %a)
|
||||
ret double %1
|
||||
}
|
||||
|
||||
declare double @llvm.log10.f64(double %Val)
|
||||
declare arm_aapcscc double @llvm.log10.f64(double %Val)
|
||||
define double @log10_d(double %a) {
|
||||
; CHECK-LABEL: log10_d:
|
||||
; SOFT: {{(bl|b)}} log10
|
||||
; HARD: b log10
|
||||
%1 = call double @llvm.log10.f64(double %a)
|
||||
; HARD: bl log10
|
||||
%1 = call arm_aapcscc double @llvm.log10.f64(double %a)
|
||||
ret double %1
|
||||
}
|
||||
|
||||
declare double @llvm.log2.f64(double %Val)
|
||||
declare arm_aapcscc double @llvm.log2.f64(double %Val)
|
||||
define double @log2_d(double %a) {
|
||||
; CHECK-LABEL: log2_d:
|
||||
; SOFT: {{(bl|b)}} log2
|
||||
; HARD: b log2
|
||||
%1 = call double @llvm.log2.f64(double %a)
|
||||
; HARD: bl log2
|
||||
%1 = call arm_aapcscc double @llvm.log2.f64(double %a)
|
||||
ret double %1
|
||||
}
|
||||
|
||||
declare double @llvm.fma.f64(double %a, double %b, double %c)
|
||||
declare arm_aapcscc double @llvm.fma.f64(double %a, double %b, double %c)
|
||||
define double @fma_d(double %a, double %b, double %c) {
|
||||
; CHECK-LABEL: fma_d:
|
||||
; SOFT: {{(bl|b)}} fma
|
||||
; HARD: vfma.f64
|
||||
%1 = call double @llvm.fma.f64(double %a, double %b, double %c)
|
||||
%1 = call arm_aapcscc double @llvm.fma.f64(double %a, double %b, double %c)
|
||||
ret double %1
|
||||
}
|
||||
|
||||
; FIXME: the FPv4-SP version is less efficient than the no-FPU version
|
||||
declare double @llvm.fabs.f64(double %Val)
|
||||
declare arm_aapcscc double @llvm.fabs.f64(double %Val)
|
||||
define double @abs_d(double %a) {
|
||||
; CHECK-LABEL: abs_d:
|
||||
; NONE: bic r1, r1, #-2147483648
|
||||
@ -116,11 +116,11 @@ define double @abs_d(double %a) {
|
||||
; SP: bfi r1, r2, #31, #1
|
||||
; SP: vmov d0, r0, r1
|
||||
; DP: vabs.f64 d0, d0
|
||||
%1 = call double @llvm.fabs.f64(double %a)
|
||||
%1 = call arm_aapcscc double @llvm.fabs.f64(double %a)
|
||||
ret double %1
|
||||
}
|
||||
|
||||
declare double @llvm.copysign.f64(double %Mag, double %Sgn)
|
||||
declare arm_aapcscc double @llvm.copysign.f64(double %Mag, double %Sgn)
|
||||
define double @copysign_d(double %a, double %b) {
|
||||
; CHECK-LABEL: copysign_d:
|
||||
; SOFT: lsrs [[REG:r[0-9]+]], r3, #31
|
||||
@ -130,71 +130,71 @@ define double @copysign_d(double %a, double %b) {
|
||||
; NEON: vmov.i32 [[REG:d[0-9]+]], #0x80000000
|
||||
; NEON: vshl.i64 [[REG]], [[REG]], #32
|
||||
; NEON: vbsl [[REG]], d
|
||||
%1 = call double @llvm.copysign.f64(double %a, double %b)
|
||||
%1 = call arm_aapcscc double @llvm.copysign.f64(double %a, double %b)
|
||||
ret double %1
|
||||
}
|
||||
|
||||
declare double @llvm.floor.f64(double %Val)
|
||||
declare arm_aapcscc double @llvm.floor.f64(double %Val)
|
||||
define double @floor_d(double %a) {
|
||||
; CHECK-LABEL: floor_d:
|
||||
; SOFT: {{(bl|b)}} floor
|
||||
; VFP4: b floor
|
||||
; VFP4: bl floor
|
||||
; FP-ARMv8: vrintm.f64
|
||||
%1 = call double @llvm.floor.f64(double %a)
|
||||
%1 = call arm_aapcscc double @llvm.floor.f64(double %a)
|
||||
ret double %1
|
||||
}
|
||||
|
||||
declare double @llvm.ceil.f64(double %Val)
|
||||
declare arm_aapcscc double @llvm.ceil.f64(double %Val)
|
||||
define double @ceil_d(double %a) {
|
||||
; CHECK-LABEL: ceil_d:
|
||||
; SOFT: {{(bl|b)}} ceil
|
||||
; VFP4: b ceil
|
||||
; VFP4: bl ceil
|
||||
; FP-ARMv8: vrintp.f64
|
||||
%1 = call double @llvm.ceil.f64(double %a)
|
||||
%1 = call arm_aapcscc double @llvm.ceil.f64(double %a)
|
||||
ret double %1
|
||||
}
|
||||
|
||||
declare double @llvm.trunc.f64(double %Val)
|
||||
declare arm_aapcscc double @llvm.trunc.f64(double %Val)
|
||||
define double @trunc_d(double %a) {
|
||||
; CHECK-LABEL: trunc_d:
|
||||
; SOFT: {{(bl|b)}} trunc
|
||||
; FFP4: b trunc
|
||||
; FFP4: bl trunc
|
||||
; FP-ARMv8: vrintz.f64
|
||||
%1 = call double @llvm.trunc.f64(double %a)
|
||||
%1 = call arm_aapcscc double @llvm.trunc.f64(double %a)
|
||||
ret double %1
|
||||
}
|
||||
|
||||
declare double @llvm.rint.f64(double %Val)
|
||||
declare arm_aapcscc double @llvm.rint.f64(double %Val)
|
||||
define double @rint_d(double %a) {
|
||||
; CHECK-LABEL: rint_d:
|
||||
; SOFT: {{(bl|b)}} rint
|
||||
; VFP4: b rint
|
||||
; VFP4: bl rint
|
||||
; FP-ARMv8: vrintx.f64
|
||||
%1 = call double @llvm.rint.f64(double %a)
|
||||
%1 = call arm_aapcscc double @llvm.rint.f64(double %a)
|
||||
ret double %1
|
||||
}
|
||||
|
||||
declare double @llvm.nearbyint.f64(double %Val)
|
||||
declare arm_aapcscc double @llvm.nearbyint.f64(double %Val)
|
||||
define double @nearbyint_d(double %a) {
|
||||
; CHECK-LABEL: nearbyint_d:
|
||||
; SOFT: {{(bl|b)}} nearbyint
|
||||
; VFP4: b nearbyint
|
||||
; VFP4: bl nearbyint
|
||||
; FP-ARMv8: vrintr.f64
|
||||
%1 = call double @llvm.nearbyint.f64(double %a)
|
||||
%1 = call arm_aapcscc double @llvm.nearbyint.f64(double %a)
|
||||
ret double %1
|
||||
}
|
||||
|
||||
declare double @llvm.round.f64(double %Val)
|
||||
declare arm_aapcscc double @llvm.round.f64(double %Val)
|
||||
define double @round_d(double %a) {
|
||||
; CHECK-LABEL: round_d:
|
||||
; SOFT: {{(bl|b)}} round
|
||||
; VFP4: b round
|
||||
; VFP4: bl round
|
||||
; FP-ARMv8: vrinta.f64
|
||||
%1 = call double @llvm.round.f64(double %a)
|
||||
%1 = call arm_aapcscc double @llvm.round.f64(double %a)
|
||||
ret double %1
|
||||
}
|
||||
|
||||
declare double @llvm.fmuladd.f64(double %a, double %b, double %c)
|
||||
declare arm_aapcscc double @llvm.fmuladd.f64(double %a, double %b, double %c)
|
||||
define double @fmuladd_d(double %a, double %b, double %c) {
|
||||
; CHECK-LABEL: fmuladd_d:
|
||||
; SOFT: bl __aeabi_dmul
|
||||
@ -202,21 +202,21 @@ define double @fmuladd_d(double %a, double %b, double %c) {
|
||||
; VFP4: vmul.f64
|
||||
; VFP4: vadd.f64
|
||||
; FP-ARMv8: vmla.f64
|
||||
%1 = call double @llvm.fmuladd.f64(double %a, double %b, double %c)
|
||||
%1 = call arm_aapcscc double @llvm.fmuladd.f64(double %a, double %b, double %c)
|
||||
ret double %1
|
||||
}
|
||||
|
||||
declare i16 @llvm.convert.to.fp16.f64(double %a)
|
||||
declare arm_aapcscc i16 @llvm.convert.to.fp16.f64(double %a)
|
||||
define i16 @d_to_h(double %a) {
|
||||
; CHECK-LABEL: d_to_h:
|
||||
; SOFT: bl __aeabi_d2h
|
||||
; VFP4: bl __aeabi_d2h
|
||||
; FP-ARMv8: vcvt{{[bt]}}.f16.f64
|
||||
%1 = call i16 @llvm.convert.to.fp16.f64(double %a)
|
||||
%1 = call arm_aapcscc i16 @llvm.convert.to.fp16.f64(double %a)
|
||||
ret i16 %1
|
||||
}
|
||||
|
||||
declare double @llvm.convert.from.fp16.f64(i16 %a)
|
||||
declare arm_aapcscc double @llvm.convert.from.fp16.f64(i16 %a)
|
||||
define double @h_to_d(i16 %a) {
|
||||
; CHECK-LABEL: h_to_d:
|
||||
; NONE: bl __aeabi_h2f
|
||||
@ -226,6 +226,7 @@ define double @h_to_d(i16 %a) {
|
||||
; VFPv4: vcvt{{[bt]}}.f32.f16
|
||||
; VFPv4: vcvt.f64.f32
|
||||
; FP-ARMv8: vcvt{{[bt]}}.f64.f16
|
||||
%1 = call double @llvm.convert.from.fp16.f64(i16 %a)
|
||||
%1 = call arm_aapcscc double @llvm.convert.from.fp16.f64(i16 %a)
|
||||
ret double %1
|
||||
}
|
||||
|
||||
|
||||
@ -14,106 +14,106 @@ define float @sqrt_f(float %a) {
|
||||
ret float %1
|
||||
}
|
||||
|
||||
declare float @llvm.powi.f32(float %Val, i32 %power)
|
||||
declare arm_aapcscc float @llvm.powi.f32(float %Val, i32 %power)
|
||||
define float @powi_f(float %a, i32 %b) {
|
||||
; CHECK-LABEL: powi_f:
|
||||
; SOFT: bl __powisf2
|
||||
; HARD: b __powisf2
|
||||
%1 = call float @llvm.powi.f32(float %a, i32 %b)
|
||||
; HARD: bl __powisf2
|
||||
%1 = call arm_aapcscc float @llvm.powi.f32(float %a, i32 %b)
|
||||
ret float %1
|
||||
}
|
||||
|
||||
declare float @llvm.sin.f32(float %Val)
|
||||
declare arm_aapcscc float @llvm.sin.f32(float %Val)
|
||||
define float @sin_f(float %a) {
|
||||
; CHECK-LABEL: sin_f:
|
||||
; SOFT: bl sinf
|
||||
; HARD: b sinf
|
||||
%1 = call float @llvm.sin.f32(float %a)
|
||||
; HARD: bl sinf
|
||||
%1 = call arm_aapcscc float @llvm.sin.f32(float %a)
|
||||
ret float %1
|
||||
}
|
||||
|
||||
declare float @llvm.cos.f32(float %Val)
|
||||
declare arm_aapcscc float @llvm.cos.f32(float %Val)
|
||||
define float @cos_f(float %a) {
|
||||
; CHECK-LABEL: cos_f:
|
||||
; SOFT: bl cosf
|
||||
; HARD: b cosf
|
||||
%1 = call float @llvm.cos.f32(float %a)
|
||||
; HARD: bl cosf
|
||||
%1 = call arm_aapcscc float @llvm.cos.f32(float %a)
|
||||
ret float %1
|
||||
}
|
||||
|
||||
declare float @llvm.pow.f32(float %Val, float %power)
|
||||
declare arm_aapcscc float @llvm.pow.f32(float %Val, float %power)
|
||||
define float @pow_f(float %a, float %b) {
|
||||
; CHECK-LABEL: pow_f:
|
||||
; SOFT: bl powf
|
||||
; HARD: b powf
|
||||
%1 = call float @llvm.pow.f32(float %a, float %b)
|
||||
; HARD: bl powf
|
||||
%1 = call arm_aapcscc float @llvm.pow.f32(float %a, float %b)
|
||||
ret float %1
|
||||
}
|
||||
|
||||
declare float @llvm.exp.f32(float %Val)
|
||||
declare arm_aapcscc float @llvm.exp.f32(float %Val)
|
||||
define float @exp_f(float %a) {
|
||||
; CHECK-LABEL: exp_f:
|
||||
; SOFT: bl expf
|
||||
; HARD: b expf
|
||||
%1 = call float @llvm.exp.f32(float %a)
|
||||
; HARD: bl expf
|
||||
%1 = call arm_aapcscc float @llvm.exp.f32(float %a)
|
||||
ret float %1
|
||||
}
|
||||
|
||||
declare float @llvm.exp2.f32(float %Val)
|
||||
declare arm_aapcscc float @llvm.exp2.f32(float %Val)
|
||||
define float @exp2_f(float %a) {
|
||||
; CHECK-LABEL: exp2_f:
|
||||
; SOFT: bl exp2f
|
||||
; HARD: b exp2f
|
||||
%1 = call float @llvm.exp2.f32(float %a)
|
||||
; HARD: bl exp2f
|
||||
%1 = call arm_aapcscc float @llvm.exp2.f32(float %a)
|
||||
ret float %1
|
||||
}
|
||||
|
||||
declare float @llvm.log.f32(float %Val)
|
||||
declare arm_aapcscc float @llvm.log.f32(float %Val)
|
||||
define float @log_f(float %a) {
|
||||
; CHECK-LABEL: log_f:
|
||||
; SOFT: bl logf
|
||||
; HARD: b logf
|
||||
%1 = call float @llvm.log.f32(float %a)
|
||||
; HARD: bl logf
|
||||
%1 = call arm_aapcscc float @llvm.log.f32(float %a)
|
||||
ret float %1
|
||||
}
|
||||
|
||||
declare float @llvm.log10.f32(float %Val)
|
||||
declare arm_aapcscc float @llvm.log10.f32(float %Val)
|
||||
define float @log10_f(float %a) {
|
||||
; CHECK-LABEL: log10_f:
|
||||
; SOFT: bl log10f
|
||||
; HARD: b log10f
|
||||
%1 = call float @llvm.log10.f32(float %a)
|
||||
; HARD: bl log10f
|
||||
%1 = call arm_aapcscc float @llvm.log10.f32(float %a)
|
||||
ret float %1
|
||||
}
|
||||
|
||||
declare float @llvm.log2.f32(float %Val)
|
||||
declare arm_aapcscc float @llvm.log2.f32(float %Val)
|
||||
define float @log2_f(float %a) {
|
||||
; CHECK-LABEL: log2_f:
|
||||
; SOFT: bl log2f
|
||||
; HARD: b log2f
|
||||
%1 = call float @llvm.log2.f32(float %a)
|
||||
; HARD: bl log2f
|
||||
%1 = call arm_aapcscc float @llvm.log2.f32(float %a)
|
||||
ret float %1
|
||||
}
|
||||
|
||||
declare float @llvm.fma.f32(float %a, float %b, float %c)
|
||||
declare arm_aapcscc float @llvm.fma.f32(float %a, float %b, float %c)
|
||||
define float @fma_f(float %a, float %b, float %c) {
|
||||
; CHECK-LABEL: fma_f:
|
||||
; SOFT: bl fmaf
|
||||
; HARD: vfma.f32
|
||||
%1 = call float @llvm.fma.f32(float %a, float %b, float %c)
|
||||
%1 = call arm_aapcscc float @llvm.fma.f32(float %a, float %b, float %c)
|
||||
ret float %1
|
||||
}
|
||||
|
||||
declare float @llvm.fabs.f32(float %Val)
|
||||
declare arm_aapcscc float @llvm.fabs.f32(float %Val)
|
||||
define float @abs_f(float %a) {
|
||||
; CHECK-LABEL: abs_f:
|
||||
; SOFT: bic r0, r0, #-2147483648
|
||||
; HARD: vabs.f32
|
||||
%1 = call float @llvm.fabs.f32(float %a)
|
||||
%1 = call arm_aapcscc float @llvm.fabs.f32(float %a)
|
||||
ret float %1
|
||||
}
|
||||
|
||||
declare float @llvm.copysign.f32(float %Mag, float %Sgn)
|
||||
declare arm_aapcscc float @llvm.copysign.f32(float %Mag, float %Sgn)
|
||||
define float @copysign_f(float %a, float %b) {
|
||||
; CHECK-LABEL: copysign_f:
|
||||
; NONE: lsrs [[REG:r[0-9]+]], r{{[0-9]+}}, #31
|
||||
@ -124,73 +124,73 @@ define float @copysign_f(float %a, float %b) {
|
||||
; VFP: bfi r{{[0-9]+}}, [[REG]], #31, #1
|
||||
; NEON: vmov.i32 [[REG:d[0-9]+]], #0x80000000
|
||||
; NEON: vbsl [[REG]], d
|
||||
%1 = call float @llvm.copysign.f32(float %a, float %b)
|
||||
%1 = call arm_aapcscc float @llvm.copysign.f32(float %a, float %b)
|
||||
ret float %1
|
||||
}
|
||||
|
||||
declare float @llvm.floor.f32(float %Val)
|
||||
declare arm_aapcscc float @llvm.floor.f32(float %Val)
|
||||
define float @floor_f(float %a) {
|
||||
; CHECK-LABEL: floor_f:
|
||||
; SOFT: bl floorf
|
||||
; VFP4: b floorf
|
||||
; VFP4: bl floorf
|
||||
; FP-ARMv8: vrintm.f32
|
||||
%1 = call float @llvm.floor.f32(float %a)
|
||||
%1 = call arm_aapcscc float @llvm.floor.f32(float %a)
|
||||
ret float %1
|
||||
}
|
||||
|
||||
declare float @llvm.ceil.f32(float %Val)
|
||||
declare arm_aapcscc float @llvm.ceil.f32(float %Val)
|
||||
define float @ceil_f(float %a) {
|
||||
; CHECK-LABEL: ceil_f:
|
||||
; SOFT: bl ceilf
|
||||
; VFP4: b ceilf
|
||||
; VFP4: bl ceilf
|
||||
; FP-ARMv8: vrintp.f32
|
||||
%1 = call float @llvm.ceil.f32(float %a)
|
||||
%1 = call arm_aapcscc float @llvm.ceil.f32(float %a)
|
||||
ret float %1
|
||||
}
|
||||
|
||||
declare float @llvm.trunc.f32(float %Val)
|
||||
declare arm_aapcscc float @llvm.trunc.f32(float %Val)
|
||||
define float @trunc_f(float %a) {
|
||||
; CHECK-LABEL: trunc_f:
|
||||
; SOFT: bl truncf
|
||||
; VFP4: b truncf
|
||||
; VFP4: bl truncf
|
||||
; FP-ARMv8: vrintz.f32
|
||||
%1 = call float @llvm.trunc.f32(float %a)
|
||||
%1 = call arm_aapcscc float @llvm.trunc.f32(float %a)
|
||||
ret float %1
|
||||
}
|
||||
|
||||
declare float @llvm.rint.f32(float %Val)
|
||||
declare arm_aapcscc float @llvm.rint.f32(float %Val)
|
||||
define float @rint_f(float %a) {
|
||||
; CHECK-LABEL: rint_f:
|
||||
; SOFT: bl rintf
|
||||
; VFP4: b rintf
|
||||
; VFP4: bl rintf
|
||||
; FP-ARMv8: vrintx.f32
|
||||
%1 = call float @llvm.rint.f32(float %a)
|
||||
%1 = call arm_aapcscc float @llvm.rint.f32(float %a)
|
||||
ret float %1
|
||||
}
|
||||
|
||||
declare float @llvm.nearbyint.f32(float %Val)
|
||||
declare arm_aapcscc float @llvm.nearbyint.f32(float %Val)
|
||||
define float @nearbyint_f(float %a) {
|
||||
; CHECK-LABEL: nearbyint_f:
|
||||
; SOFT: bl nearbyintf
|
||||
; VFP4: b nearbyintf
|
||||
; VFP4: bl nearbyintf
|
||||
; FP-ARMv8: vrintr.f32
|
||||
%1 = call float @llvm.nearbyint.f32(float %a)
|
||||
%1 = call arm_aapcscc float @llvm.nearbyint.f32(float %a)
|
||||
ret float %1
|
||||
}
|
||||
|
||||
declare float @llvm.round.f32(float %Val)
|
||||
declare arm_aapcscc float @llvm.round.f32(float %Val)
|
||||
define float @round_f(float %a) {
|
||||
; CHECK-LABEL: round_f:
|
||||
; SOFT: bl roundf
|
||||
; VFP4: b roundf
|
||||
; VFP4: bl roundf
|
||||
; FP-ARMv8: vrinta.f32
|
||||
%1 = call float @llvm.round.f32(float %a)
|
||||
%1 = call arm_aapcscc float @llvm.round.f32(float %a)
|
||||
ret float %1
|
||||
}
|
||||
|
||||
; FIXME: why does cortex-m4 use vmla, while cortex-a7 uses vmul+vadd?
|
||||
; (these should be equivalent, even the rounding is the same)
|
||||
declare float @llvm.fmuladd.f32(float %a, float %b, float %c)
|
||||
declare arm_aapcscc float @llvm.fmuladd.f32(float %a, float %b, float %c)
|
||||
define float @fmuladd_f(float %a, float %b, float %c) {
|
||||
; CHECK-LABEL: fmuladd_f:
|
||||
; SOFT: bl __aeabi_fmul
|
||||
@ -198,24 +198,25 @@ define float @fmuladd_f(float %a, float %b, float %c) {
|
||||
; VMLA: vmla.f32
|
||||
; NO-VMLA: vmul.f32
|
||||
; NO-VMLA: vadd.f32
|
||||
%1 = call float @llvm.fmuladd.f32(float %a, float %b, float %c)
|
||||
%1 = call arm_aapcscc float @llvm.fmuladd.f32(float %a, float %b, float %c)
|
||||
ret float %1
|
||||
}
|
||||
|
||||
declare i16 @llvm.convert.to.fp16.f32(float %a)
|
||||
declare arm_aapcscc i16 @llvm.convert.to.fp16.f32(float %a)
|
||||
define i16 @f_to_h(float %a) {
|
||||
; CHECK-LABEL: f_to_h:
|
||||
; SOFT: bl __aeabi_f2h
|
||||
; HARD: vcvt{{[bt]}}.f16.f32
|
||||
%1 = call i16 @llvm.convert.to.fp16.f32(float %a)
|
||||
%1 = call arm_aapcscc i16 @llvm.convert.to.fp16.f32(float %a)
|
||||
ret i16 %1
|
||||
}
|
||||
|
||||
declare float @llvm.convert.from.fp16.f32(i16 %a)
|
||||
declare arm_aapcscc float @llvm.convert.from.fp16.f32(i16 %a)
|
||||
define float @h_to_f(i16 %a) {
|
||||
; CHECK-LABEL: h_to_f:
|
||||
; SOFT: bl __aeabi_h2f
|
||||
; HARD: vcvt{{[bt]}}.f32.f16
|
||||
%1 = call float @llvm.convert.from.fp16.f32(i16 %a)
|
||||
%1 = call arm_aapcscc float @llvm.convert.from.fp16.f32(i16 %a)
|
||||
ret float %1
|
||||
}
|
||||
|
||||
|
||||
@ -4,6 +4,7 @@
|
||||
; RUN: llc < %s -mtriple=thumbv7-none-eabihf -mcpu=cortex-a8 | FileCheck %s -check-prefix=CHECK -check-prefix=HARD -check-prefix=DP -check-prefix=VFP4-ALL -check-prefix=VFP4-DP
|
||||
|
||||
define float @add_f(float %a, float %b) {
|
||||
|
||||
entry:
|
||||
; CHECK-LABEL: add_f:
|
||||
; NONE: bl __aeabi_fadd
|
||||
@ -83,7 +84,7 @@ define float @rem_f(float %a, float %b) {
|
||||
entry:
|
||||
; CHECK-LABEL: rem_f:
|
||||
; NONE: bl fmodf
|
||||
; HARD: b fmodf
|
||||
; HARD: bl fmodf
|
||||
%0 = frem float %a, %b
|
||||
ret float %0
|
||||
}
|
||||
@ -92,7 +93,7 @@ define double @rem_d(double %a, double %b) {
|
||||
entry:
|
||||
; CHECK-LABEL: rem_d:
|
||||
; NONE: bl fmod
|
||||
; HARD: b fmod
|
||||
; HARD: bl fmod
|
||||
%0 = frem double %a, %b
|
||||
ret double %0
|
||||
}
|
||||
|
||||
Loading…
Reference in New Issue
Block a user