[PPC64] Use mfocrf in prologue when we only need to save 1 nonvolatile CR field
In the ELFv2 ABI, we are not required to save all CR fields. If only one
nonvolatile CR field is clobbered, use mfocrf instead of mfcr to
selectively save the field, because mfocrf has short latency compares to
mfcr.
Thanks Nemanja's invaluable hint!
Reviewers: nemanjai tjablin hfinkel kbarton
http://reviews.llvm.org/D17749
llvm-svn: 266038
diff --git a/llvm/lib/Target/PowerPC/PPCFrameLowering.cpp b/llvm/lib/Target/PowerPC/PPCFrameLowering.cpp
index aadf64d..40a2e8b 100644
--- a/llvm/lib/Target/PowerPC/PPCFrameLowering.cpp
+++ b/llvm/lib/Target/PowerPC/PPCFrameLowering.cpp
@@ -838,11 +838,15 @@
// If we need to spill the CR and the LR but we don't have two separate
// registers available, we must spill them one at a time
if (MustSaveCR && SingleScratchReg && MustSaveLR) {
- // FIXME: In the ELFv2 ABI, we are not required to save all CR fields.
- // If only one or two CR fields are clobbered, it could be more
- // efficient to use mfocrf to selectively save just those fields.
+ // In the ELFv2 ABI, we are not required to save all CR fields.
+ // If only one or two CR fields are clobbered, it is more efficient to use
+ // mfocrf to selectively save just those fields, because mfocrf has short
+ // latency compares to mfcr.
+ unsigned MfcrOpcode = PPC::MFCR8;
+ if (isELFv2ABI && MustSaveCRs.size() == 1)
+ MfcrOpcode = PPC::MFOCRF8;
MachineInstrBuilder MIB =
- BuildMI(MBB, MBBI, dl, TII.get(PPC::MFCR8), TempReg);
+ BuildMI(MBB, MBBI, dl, TII.get(MfcrOpcode), TempReg);
for (unsigned i = 0, e = MustSaveCRs.size(); i != e; ++i)
MIB.addReg(MustSaveCRs[i], RegState::ImplicitKill);
BuildMI(MBB, MBBI, dl, TII.get(PPC::STW8))
@@ -856,11 +860,15 @@
if (MustSaveCR &&
!(SingleScratchReg && MustSaveLR)) { // will only occur for PPC64
- // FIXME: In the ELFv2 ABI, we are not required to save all CR fields.
- // If only one or two CR fields are clobbered, it could be more
- // efficient to use mfocrf to selectively save just those fields.
+ // In the ELFv2 ABI, we are not required to save all CR fields.
+ // If only one or two CR fields are clobbered, it is more efficient to use
+ // mfocrf to selectively save just those fields, because mfocrf has short
+ // latency compares to mfcr.
+ unsigned MfcrOpcode = PPC::MFCR8;
+ if (isELFv2ABI && MustSaveCRs.size() == 1)
+ MfcrOpcode = PPC::MFOCRF8;
MachineInstrBuilder MIB =
- BuildMI(MBB, MBBI, dl, TII.get(PPC::MFCR8), TempReg);
+ BuildMI(MBB, MBBI, dl, TII.get(MfcrOpcode), TempReg);
for (unsigned i = 0, e = MustSaveCRs.size(); i != e; ++i)
MIB.addReg(MustSaveCRs[i], RegState::ImplicitKill);
}
diff --git a/llvm/test/CodeGen/PowerPC/crsave.ll b/llvm/test/CodeGen/PowerPC/crsave.ll
index 8121e1b..4975afc 100644
--- a/llvm/test/CodeGen/PowerPC/crsave.ll
+++ b/llvm/test/CodeGen/PowerPC/crsave.ll
@@ -1,5 +1,6 @@
; RUN: llc -O0 -disable-fp-elim -mtriple=powerpc-unknown-linux-gnu -mcpu=g5 < %s | FileCheck %s -check-prefix=PPC32
; RUN: llc -O0 -mtriple=powerpc64-unknown-linux-gnu -mcpu=g5 < %s | FileCheck %s -check-prefix=PPC64
+; RUN: llc -O0 -mtriple=powerpc64le-unknown-linux-gnu < %s | FileCheck %s -check-prefix=PPC64-ELFv2
declare void @foo()
@@ -60,3 +61,22 @@
; PPC64: mtocrf 16, 12
; PPC64: mtocrf 8, 12
+; Generate mfocrf in prologue when we need to save 1 nonvolatile CR field
+define void @cloberOneNvCrField() {
+entry:
+ tail call void asm sideeffect "# clobbers", "~{cr2}"()
+ ret void
+
+; PPC64-ELFv2-LABEL: @cloberOneNvCrField
+; PPC64-ELFv2: mfocrf [[REG1:[0-9]+]], 32
+}
+
+; Generate mfcr in prologue when we need to save all nonvolatile CR field
+define void @cloberAllNvCrField() {
+entry:
+ tail call void asm sideeffect "# clobbers", "~{cr2},~{cr3},~{cr4}"()
+ ret void
+
+; PPC64-ELFv2-LABEL: @cloberAllNvCrField
+; PPC64-ELFv2: mfcr [[REG1:[0-9]+]]
+}
diff --git a/llvm/test/CodeGen/PowerPC/pr26690.ll b/llvm/test/CodeGen/PowerPC/pr26690.ll
index 524e8b5..e1c3c49 100644
--- a/llvm/test/CodeGen/PowerPC/pr26690.ll
+++ b/llvm/test/CodeGen/PowerPC/pr26690.ll
@@ -101,7 +101,7 @@
ret i32 2
}
-; CHECK: mfcr {{[0-9]+}}
+; CHECK: mfocrf {{[0-9]+}}
!llvm.ident = !{!0}