Skip to content

[AArch64] Fix incorrect big-endian spill in foldMemoryOperandImpl #65601

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 1 commit into from
Oct 12, 2023
+515 −246
Merged

[AArch64] Fix incorrect big-endian spill in foldMemoryOperandImpl #65601

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
40 changes: 6 additions & 34 deletions llvm/lib/Target/AArch64/AArch64InstrInfo.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -5492,42 +5492,14 @@ MachineInstr *AArch64InstrInfo::foldMemoryOperandImpl(
//
// STRXui %xzr, %stack.0
//
if (IsSpill && DstMO.isUndef() && SrcReg.isPhysical()) {
if (IsSpill && DstMO.isUndef() && SrcReg == AArch64::WZR &&
TRI.getRegSizeInBits(*getRegClass(DstReg)) == 64) {
assert(SrcMO.getSubReg() == 0 &&
"Unexpected subreg on physical register");
const TargetRegisterClass *SpillRC;
unsigned SpillSubreg;
switch (DstMO.getSubReg()) {
default:
SpillRC = nullptr;
break;
case AArch64::sub_32:
case AArch64::ssub:
if (AArch64::GPR32RegClass.contains(SrcReg)) {
SpillRC = &AArch64::GPR64RegClass;
SpillSubreg = AArch64::sub_32;
} else if (AArch64::FPR32RegClass.contains(SrcReg)) {
SpillRC = &AArch64::FPR64RegClass;
SpillSubreg = AArch64::ssub;
} else
SpillRC = nullptr;
break;
case AArch64::dsub:
if (AArch64::FPR64RegClass.contains(SrcReg)) {
SpillRC = &AArch64::FPR128RegClass;
SpillSubreg = AArch64::dsub;
} else
SpillRC = nullptr;
break;
}

if (SpillRC)
if (unsigned WidenedSrcReg =
TRI.getMatchingSuperReg(SrcReg, SpillSubreg, SpillRC)) {
storeRegToStackSlot(MBB, InsertPt, WidenedSrcReg, SrcMO.isKill(),
FrameIndex, SpillRC, &TRI, Register());
return &*--InsertPt;
}
storeRegToStackSlot(MBB, InsertPt, AArch64::XZR, SrcMO.isKill(),
FrameIndex, &AArch64::GPR64RegClass, &TRI,
Register());
return &*--InsertPt;
}

// Handle cases like filling use of:
Expand Down
7 changes: 5 additions & 2 deletions llvm/test/CodeGen/AArch64/arm64-neon-mul-div.ll
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1486,7 +1486,8 @@ define <2 x float> @frem2f32(<2 x float> %A, <2 x float> %B) {
; CHECK-NEXT: mov s0, v0.s[1]
; CHECK-NEXT: mov s1, v1.s[1]
; CHECK-NEXT: bl fmodf
; CHECK-NEXT: str d0, [sp, #32] // 16-byte Folded Spill
; CHECK-NEXT: // kill: def $s0 killed $s0 def $q0
; CHECK-NEXT: str q0, [sp, #32] // 16-byte Folded Spill
; CHECK-NEXT: ldp q0, q1, [sp] // 32-byte Folded Reload
; CHECK-NEXT: // kill: def $s0 killed $s0 killed $q0
; CHECK-NEXT: // kill: def $s1 killed $s1 killed $q1
Expand All @@ -1513,7 +1514,8 @@ define <4 x float> @frem4f32(<4 x float> %A, <4 x float> %B) {
; CHECK-NEXT: mov s0, v0.s[1]
; CHECK-NEXT: mov s1, v1.s[1]
; CHECK-NEXT: bl fmodf
; CHECK-NEXT: str d0, [sp] // 16-byte Folded Spill
; CHECK-NEXT: // kill: def $s0 killed $s0 def $q0
; CHECK-NEXT: str q0, [sp] // 16-byte Folded Spill
; CHECK-NEXT: ldp q0, q1, [sp, #16] // 32-byte Folded Reload
; CHECK-NEXT: // kill: def $s0 killed $s0 killed $q0
; CHECK-NEXT: // kill: def $s1 killed $s1 killed $q1
Expand Down Expand Up @@ -1569,6 +1571,7 @@ define <2 x double> @frem2d64(<2 x double> %A, <2 x double> %B) {
; CHECK-NEXT: mov d0, v0.d[1]
; CHECK-NEXT: mov d1, v1.d[1]
; CHECK-NEXT: bl fmod
; CHECK-NEXT: // kill: def $d0 killed $d0 def $q0
; CHECK-NEXT: str q0, [sp, #32] // 16-byte Folded Spill
; CHECK-NEXT: ldp q0, q1, [sp] // 32-byte Folded Reload
; CHECK-NEXT: // kill: def $d0 killed $d0 killed $q0
Expand Down
305 changes: 215 additions & 90 deletions llvm/test/CodeGen/AArch64/fexplog.ll
View file
Open in desktop

Large diffs are not rendered by default.

3 changes: 2 additions & 1 deletion llvm/test/CodeGen/AArch64/fold-int-pow2-with-fmul-or-fdiv.ll
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -52,7 +52,8 @@ define <4 x float> @fmul_pow2_ldexp_4xfloat(<4 x i32> %i) {
; CHECK-NEON-NEXT: fmov s0, #9.00000000
; CHECK-NEON-NEXT: bl ldexpf
; CHECK-NEON-NEXT: ldr q1, [sp, #16] // 16-byte Folded Reload
; CHECK-NEON-NEXT: str d0, [sp] // 16-byte Folded Spill
; CHECK-NEON-NEXT: // kill: def $s0 killed $s0 def $q0
; CHECK-NEON-NEXT: str q0, [sp] // 16-byte Folded Spill
; CHECK-NEON-NEXT: fmov s0, #9.00000000
; CHECK-NEON-NEXT: fmov w0, s1
; CHECK-NEON-NEXT: bl ldexpf
Expand Down
89 changes: 57 additions & 32 deletions llvm/test/CodeGen/AArch64/fpow.ll
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -66,6 +66,7 @@ define <2 x double> @pow_v2f64(<2 x double> %a, <2 x double> %b) {
; CHECK-SD-NEXT: mov d0, v0.d[1]
; CHECK-SD-NEXT: mov d1, v1.d[1]
; CHECK-SD-NEXT: bl pow
; CHECK-SD-NEXT: // kill: def $d0 killed $d0 def $q0
; CHECK-SD-NEXT: str q0, [sp, #32] // 16-byte Folded Spill
; CHECK-SD-NEXT: ldp q0, q1, [sp] // 32-byte Folded Reload
; CHECK-SD-NEXT: // kill: def $d0 killed $d0 killed $q0
Expand All @@ -92,9 +93,10 @@ define <2 x double> @pow_v2f64(<2 x double> %a, <2 x double> %b) {
; CHECK-GI-NEXT: // kill: def $d0 killed $d0 killed $q0
; CHECK-GI-NEXT: // kill: def $d1 killed $d1 killed $q1
; CHECK-GI-NEXT: bl pow
; CHECK-GI-NEXT: // kill: def $d0 killed $d0 def $q0
; CHECK-GI-NEXT: str q0, [sp] // 16-byte Folded Spill
; CHECK-GI-NEXT: fmov d0, d8
; CHECK-GI-NEXT: fmov d1, d9
; CHECK-GI-NEXT: fmov d0, d8
; CHECK-GI-NEXT: bl pow
; CHECK-GI-NEXT: ldr q1, [sp] // 16-byte Folded Reload
; CHECK-GI-NEXT: // kill: def $d0 killed $d0 def $q0
Expand Down Expand Up @@ -198,6 +200,7 @@ define <4 x double> @pow_v4f64(<4 x double> %a, <4 x double> %b) {
; CHECK-SD-NEXT: stp q1, q3, [sp, #48] // 32-byte Folded Spill
; CHECK-SD-NEXT: mov d1, v2.d[1]
; CHECK-SD-NEXT: bl pow
; CHECK-SD-NEXT: // kill: def $d0 killed $d0 def $q0
; CHECK-SD-NEXT: str q0, [sp, #32] // 16-byte Folded Spill
; CHECK-SD-NEXT: ldp q0, q1, [sp] // 32-byte Folded Reload
; CHECK-SD-NEXT: // kill: def $d0 killed $d0 killed $q0
Expand All @@ -211,6 +214,7 @@ define <4 x double> @pow_v4f64(<4 x double> %a, <4 x double> %b) {
; CHECK-SD-NEXT: mov d0, v0.d[1]
; CHECK-SD-NEXT: mov d1, v1.d[1]
; CHECK-SD-NEXT: bl pow
; CHECK-SD-NEXT: // kill: def $d0 killed $d0 def $q0
; CHECK-SD-NEXT: str q0, [sp, #16] // 16-byte Folded Spill
; CHECK-SD-NEXT: ldp q0, q1, [sp, #48] // 32-byte Folded Reload
; CHECK-SD-NEXT: // kill: def $d0 killed $d0 killed $q0
Expand Down Expand Up @@ -246,19 +250,22 @@ define <4 x double> @pow_v4f64(<4 x double> %a, <4 x double> %b) {
; CHECK-GI-NEXT: // kill: def $d1 killed $d1 killed $q1
; CHECK-GI-NEXT: mov d9, v4.d[1]
; CHECK-GI-NEXT: bl pow
; CHECK-GI-NEXT: // kill: def $d0 killed $d0 def $q0
; CHECK-GI-NEXT: str q0, [sp, #48] // 16-byte Folded Spill
; CHECK-GI-NEXT: fmov d0, d8
; CHECK-GI-NEXT: fmov d1, d10
; CHECK-GI-NEXT: fmov d0, d8
; CHECK-GI-NEXT: bl pow
; CHECK-GI-NEXT: // kill: def $d0 killed $d0 def $q0
; CHECK-GI-NEXT: str q0, [sp, #16] // 16-byte Folded Spill
; CHECK-GI-NEXT: ldr q0, [sp, #32] // 16-byte Folded Reload
; CHECK-GI-NEXT: ldr q1, [sp] // 16-byte Folded Reload
; CHECK-GI-NEXT: // kill: def $d0 killed $d0 killed $q0
; CHECK-GI-NEXT: ldr q0, [sp, #32] // 16-byte Folded Reload
; CHECK-GI-NEXT: // kill: def $d1 killed $d1 killed $q1
; CHECK-GI-NEXT: // kill: def $d0 killed $d0 killed $q0
; CHECK-GI-NEXT: bl pow
; CHECK-GI-NEXT: // kill: def $d0 killed $d0 def $q0
; CHECK-GI-NEXT: str q0, [sp, #32] // 16-byte Folded Spill
; CHECK-GI-NEXT: fmov d0, d9
; CHECK-GI-NEXT: fmov d1, d11
; CHECK-GI-NEXT: fmov d0, d9
; CHECK-GI-NEXT: bl pow
; CHECK-GI-NEXT: ldr q2, [sp, #48] // 16-byte Folded Reload
; CHECK-GI-NEXT: ldr q1, [sp, #16] // 16-byte Folded Reload
Expand Down Expand Up @@ -290,7 +297,8 @@ define <2 x float> @pow_v2f32(<2 x float> %a, <2 x float> %b) {
; CHECK-SD-NEXT: mov s0, v0.s[1]
; CHECK-SD-NEXT: mov s1, v1.s[1]
; CHECK-SD-NEXT: bl powf
; CHECK-SD-NEXT: str d0, [sp, #32] // 16-byte Folded Spill
; CHECK-SD-NEXT: // kill: def $s0 killed $s0 def $q0
; CHECK-SD-NEXT: str q0, [sp, #32] // 16-byte Folded Spill
; CHECK-SD-NEXT: ldp q0, q1, [sp] // 32-byte Folded Reload
; CHECK-SD-NEXT: // kill: def $s0 killed $s0 killed $q0
; CHECK-SD-NEXT: // kill: def $s1 killed $s1 killed $q1
Expand Down Expand Up @@ -319,9 +327,10 @@ define <2 x float> @pow_v2f32(<2 x float> %a, <2 x float> %b) {
; CHECK-GI-NEXT: // kill: def $s0 killed $s0 killed $q0
; CHECK-GI-NEXT: // kill: def $s1 killed $s1 killed $q1
; CHECK-GI-NEXT: bl powf
; CHECK-GI-NEXT: str d0, [sp] // 16-byte Folded Spill
; CHECK-GI-NEXT: fmov s0, s8
; CHECK-GI-NEXT: // kill: def $s0 killed $s0 def $q0
; CHECK-GI-NEXT: str q0, [sp] // 16-byte Folded Spill
; CHECK-GI-NEXT: fmov s1, s9
; CHECK-GI-NEXT: fmov s0, s8
; CHECK-GI-NEXT: bl powf
; CHECK-GI-NEXT: ldr q1, [sp] // 16-byte Folded Reload
; CHECK-GI-NEXT: // kill: def $s0 killed $s0 def $q0
Expand All @@ -347,7 +356,8 @@ define <3 x float> @pow_v3f32(<3 x float> %a, <3 x float> %b) {
; CHECK-SD-NEXT: mov s0, v0.s[1]
; CHECK-SD-NEXT: mov s1, v1.s[1]
; CHECK-SD-NEXT: bl powf
; CHECK-SD-NEXT: str d0, [sp] // 16-byte Folded Spill
; CHECK-SD-NEXT: // kill: def $s0 killed $s0 def $q0
; CHECK-SD-NEXT: str q0, [sp] // 16-byte Folded Spill
; CHECK-SD-NEXT: ldp q0, q1, [sp, #16] // 32-byte Folded Reload
; CHECK-SD-NEXT: // kill: def $s0 killed $s0 killed $q0
; CHECK-SD-NEXT: // kill: def $s1 killed $s1 killed $q1
Expand Down Expand Up @@ -387,13 +397,15 @@ define <3 x float> @pow_v3f32(<3 x float> %a, <3 x float> %b) {
; CHECK-GI-NEXT: mov s11, v1.s[2]
; CHECK-GI-NEXT: // kill: def $s1 killed $s1 killed $q1
; CHECK-GI-NEXT: bl powf
; CHECK-GI-NEXT: str d0, [sp, #16] // 16-byte Folded Spill
; CHECK-GI-NEXT: fmov s0, s8
; CHECK-GI-NEXT: // kill: def $s0 killed $s0 def $q0
; CHECK-GI-NEXT: str q0, [sp, #16] // 16-byte Folded Spill
; CHECK-GI-NEXT: fmov s1, s10
; CHECK-GI-NEXT: fmov s0, s8
; CHECK-GI-NEXT: bl powf
; CHECK-GI-NEXT: str d0, [sp] // 16-byte Folded Spill
; CHECK-GI-NEXT: fmov s0, s9
; CHECK-GI-NEXT: // kill: def $s0 killed $s0 def $q0
; CHECK-GI-NEXT: str q0, [sp] // 16-byte Folded Spill
; CHECK-GI-NEXT: fmov s1, s11
; CHECK-GI-NEXT: fmov s0, s9
; CHECK-GI-NEXT: bl powf
; CHECK-GI-NEXT: ldp q2, q1, [sp] // 32-byte Folded Reload
; CHECK-GI-NEXT: // kill: def $s0 killed $s0 def $q0
Expand Down Expand Up @@ -422,7 +434,8 @@ define <4 x float> @pow_v4f32(<4 x float> %a, <4 x float> %b) {
; CHECK-SD-NEXT: mov s0, v0.s[1]
; CHECK-SD-NEXT: mov s1, v1.s[1]
; CHECK-SD-NEXT: bl powf
; CHECK-SD-NEXT: str d0, [sp] // 16-byte Folded Spill
; CHECK-SD-NEXT: // kill: def $s0 killed $s0 def $q0
; CHECK-SD-NEXT: str q0, [sp] // 16-byte Folded Spill
; CHECK-SD-NEXT: ldp q0, q1, [sp, #16] // 32-byte Folded Reload
; CHECK-SD-NEXT: // kill: def $s0 killed $s0 killed $q0
; CHECK-SD-NEXT: // kill: def $s1 killed $s1 killed $q1
Expand Down Expand Up @@ -475,17 +488,20 @@ define <4 x float> @pow_v4f32(<4 x float> %a, <4 x float> %b) {
; CHECK-GI-NEXT: mov s13, v1.s[3]
; CHECK-GI-NEXT: // kill: def $s1 killed $s1 killed $q1
; CHECK-GI-NEXT: bl powf
; CHECK-GI-NEXT: str d0, [sp, #32] // 16-byte Folded Spill
; CHECK-GI-NEXT: fmov s0, s8
; CHECK-GI-NEXT: // kill: def $s0 killed $s0 def $q0
; CHECK-GI-NEXT: str q0, [sp, #32] // 16-byte Folded Spill
; CHECK-GI-NEXT: fmov s1, s11
; CHECK-GI-NEXT: fmov s0, s8
; CHECK-GI-NEXT: bl powf
; CHECK-GI-NEXT: str d0, [sp, #16] // 16-byte Folded Spill
; CHECK-GI-NEXT: fmov s0, s9
; CHECK-GI-NEXT: // kill: def $s0 killed $s0 def $q0
; CHECK-GI-NEXT: str q0, [sp, #16] // 16-byte Folded Spill
; CHECK-GI-NEXT: fmov s1, s12
; CHECK-GI-NEXT: fmov s0, s9
; CHECK-GI-NEXT: bl powf
; CHECK-GI-NEXT: str d0, [sp] // 16-byte Folded Spill
; CHECK-GI-NEXT: fmov s0, s10
; CHECK-GI-NEXT: // kill: def $s0 killed $s0 def $q0
; CHECK-GI-NEXT: str q0, [sp] // 16-byte Folded Spill
; CHECK-GI-NEXT: fmov s1, s13
; CHECK-GI-NEXT: fmov s0, s10
; CHECK-GI-NEXT: bl powf
; CHECK-GI-NEXT: ldp q2, q1, [sp, #16] // 32-byte Folded Reload
; CHECK-GI-NEXT: // kill: def $s0 killed $s0 def $q0
Expand Down Expand Up @@ -517,7 +533,8 @@ define <8 x float> @pow_v8f32(<8 x float> %a, <8 x float> %b) {
; CHECK-SD-NEXT: stp q1, q3, [sp, #32] // 32-byte Folded Spill
; CHECK-SD-NEXT: mov s1, v2.s[1]
; CHECK-SD-NEXT: bl powf
; CHECK-SD-NEXT: str d0, [sp, #64] // 16-byte Folded Spill
; CHECK-SD-NEXT: // kill: def $s0 killed $s0 def $q0
; CHECK-SD-NEXT: str q0, [sp, #64] // 16-byte Folded Spill
; CHECK-SD-NEXT: ldp q0, q1, [sp] // 32-byte Folded Reload
; CHECK-SD-NEXT: // kill: def $s0 killed $s0 killed $q0
; CHECK-SD-NEXT: // kill: def $s1 killed $s1 killed $q1
Expand Down Expand Up @@ -546,7 +563,8 @@ define <8 x float> @pow_v8f32(<8 x float> %a, <8 x float> %b) {
; CHECK-SD-NEXT: mov s0, v0.s[1]
; CHECK-SD-NEXT: mov s1, v1.s[1]
; CHECK-SD-NEXT: bl powf
; CHECK-SD-NEXT: str d0, [sp, #16] // 16-byte Folded Spill
; CHECK-SD-NEXT: // kill: def $s0 killed $s0 def $q0
; CHECK-SD-NEXT: str q0, [sp, #16] // 16-byte Folded Spill
; CHECK-SD-NEXT: ldp q0, q1, [sp, #32] // 32-byte Folded Reload
; CHECK-SD-NEXT: // kill: def $s0 killed $s0 killed $q0
; CHECK-SD-NEXT: // kill: def $s1 killed $s1 killed $q1
Expand Down Expand Up @@ -614,32 +632,39 @@ define <8 x float> @pow_v8f32(<8 x float> %a, <8 x float> %b) {
; CHECK-GI-NEXT: mov s2, v3.s[3]
; CHECK-GI-NEXT: stp s2, s5, [sp, #200] // 8-byte Folded Spill
; CHECK-GI-NEXT: bl powf
; CHECK-GI-NEXT: str d0, [sp, #96] // 16-byte Folded Spill
; CHECK-GI-NEXT: fmov s0, s8
; CHECK-GI-NEXT: // kill: def $s0 killed $s0 def $q0
; CHECK-GI-NEXT: str q0, [sp, #96] // 16-byte Folded Spill
; CHECK-GI-NEXT: fmov s1, s14
; CHECK-GI-NEXT: fmov s0, s8
; CHECK-GI-NEXT: bl powf
; CHECK-GI-NEXT: str d0, [sp, #32] // 16-byte Folded Spill
; CHECK-GI-NEXT: fmov s0, s9
; CHECK-GI-NEXT: // kill: def $s0 killed $s0 def $q0
; CHECK-GI-NEXT: str q0, [sp, #32] // 16-byte Folded Spill
; CHECK-GI-NEXT: fmov s1, s15
; CHECK-GI-NEXT: fmov s0, s9
; CHECK-GI-NEXT: bl powf
; CHECK-GI-NEXT: str d0, [sp, #80] // 16-byte Folded Spill
; CHECK-GI-NEXT: fmov s0, s10
; CHECK-GI-NEXT: // kill: def $s0 killed $s0 def $q0
; CHECK-GI-NEXT: str q0, [sp, #80] // 16-byte Folded Spill
; CHECK-GI-NEXT: fmov s1, s13
; CHECK-GI-NEXT: fmov s0, s10
; CHECK-GI-NEXT: bl powf
; CHECK-GI-NEXT: str d0, [sp, #64] // 16-byte Folded Spill
; CHECK-GI-NEXT: // kill: def $s0 killed $s0 def $q0
; CHECK-GI-NEXT: str q0, [sp, #64] // 16-byte Folded Spill
; CHECK-GI-NEXT: ldp q1, q0, [sp] // 32-byte Folded Reload
; CHECK-GI-NEXT: // kill: def $s0 killed $s0 killed $q0
; CHECK-GI-NEXT: // kill: def $s1 killed $s1 killed $q1
; CHECK-GI-NEXT: bl powf
; CHECK-GI-NEXT: fmov s1, s12
; CHECK-GI-NEXT: str d0, [sp, #16] // 16-byte Folded Spill
; CHECK-GI-NEXT: // kill: def $s0 killed $s0 def $q0
; CHECK-GI-NEXT: str q0, [sp, #16] // 16-byte Folded Spill
; CHECK-GI-NEXT: ldr s0, [sp, #48] // 4-byte Folded Reload
; CHECK-GI-NEXT: bl powf
; CHECK-GI-NEXT: fmov s1, s11
; CHECK-GI-NEXT: str d0, [sp, #48] // 16-byte Folded Spill
; CHECK-GI-NEXT: // kill: def $s0 killed $s0 def $q0
; CHECK-GI-NEXT: str q0, [sp, #48] // 16-byte Folded Spill
; CHECK-GI-NEXT: ldr s0, [sp, #112] // 4-byte Folded Reload
; CHECK-GI-NEXT: bl powf
; CHECK-GI-NEXT: str d0, [sp, #112] // 16-byte Folded Spill
; CHECK-GI-NEXT: // kill: def $s0 killed $s0 def $q0
; CHECK-GI-NEXT: str q0, [sp, #112] // 16-byte Folded Spill
; CHECK-GI-NEXT: ldp s1, s0, [sp, #200] // 8-byte Folded Reload
; CHECK-GI-NEXT: bl powf
; CHECK-GI-NEXT: ldp q3, q2, [sp, #16] // 32-byte Folded Reload
Expand Down
Loading