on X86 Atom. Some of our tests failed because the tail merging part of
the BranchFolding pass was creating new basic blocks which did not
contain live-in information. When the anti-dependency code in the Post-RA
scheduler ran, it would sometimes rename the register containing
the function return value because the fact that the return value was
live-in to the subsequent block had been lost. To fix this, it is necessary
to run the RegisterScavenging code in the BranchFolding pass.
This patch makes sure that the register scavenging code is invoked
in the X86 subtarget only when post-RA scheduling is being done.
Post RA scheduling in the X86 subtarget is only done for Atom.
This patch adds a new function to the TargetRegisterClass to control
whether or not live-ins should be preserved during branch folding.
This is necessary in order for the anti-dependency optimizations done
during the PostRASchedulerList pass to work properly when doing
Post-RA scheduling for the X86 in general and for the Intel Atom in particular.
The patch adds and invokes the new function trackLivenessAfterRegAlloc()
instead of using the existing requiresRegisterScavenging().
It changes BranchFolding.cpp to call trackLivenessAfterRegAlloc() instead of
requiresRegisterScavenging(). It changes the all the targets that
implemented requiresRegisterScavenging() to also implement
trackLivenessAfterRegAlloc().
It adds an assertion in the Post RA scheduler to make sure that post RA
liveness information is available when it is needed.
It changes the X86 break-anti-dependencies test to use –mcpu=atom, in order
to avoid running into the added assertion.
Finally, this patch restores the use of anti-dependency checking
(which was turned off temporarily for the 3.1 release) for
Intel Atom in the Post RA scheduler.
Patch by Andy Zhang!
Thanks to Jakob and Anton for their reviews.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@155395
91177308-0d34-0410-b5e6-
96231b3b80d8
return false;
}
+ /// trackLivenessAfterRegAlloc - returns true if the live-ins should be tracked
+ /// after register allocation.
+ virtual bool trackLivenessAfterRegAlloc(const MachineFunction &MF) const {
+ return false;
+ }
+
/// needsStackRealignment - true if storage within the function requires the
/// stack pointer to be aligned more than the normal calling convention calls
/// for.
// Use a RegScavenger to help update liveness when required.
MachineRegisterInfo &MRI = MF.getRegInfo();
- if (MRI.tracksLiveness() && TRI->requiresRegisterScavenging(MF))
+ if (MRI.tracksLiveness() && TRI->trackLivenessAfterRegAlloc(MF))
RS = new RegScavenger();
else
MRI.invalidateLiveness();
const InstrItineraryData *InstrItins = TM.getInstrItineraryData();
HazardRec =
TM.getInstrInfo()->CreateTargetPostRAHazardRecognizer(InstrItins, this);
+
+ assert((AntiDepMode == TargetSubtargetInfo::ANTIDEP_NONE ||
+ MRI.tracksLiveness()) &&
+ "Live-ins must be accurate for anti-dependency breaking");
AntiDepBreak =
((AntiDepMode == TargetSubtargetInfo::ANTIDEP_ALL) ?
(AntiDepBreaker *)new AggressiveAntiDepBreaker(MF, RCI, CriticalPathRCs) :
return true;
}
+bool ARMBaseRegisterInfo::
+trackLivenessAfterRegAlloc(const MachineFunction &MF) const {
+ return true;
+}
+
bool ARMBaseRegisterInfo::
requiresFrameIndexScavenging(const MachineFunction &MF) const {
return true;
virtual bool requiresRegisterScavenging(const MachineFunction &MF) const;
+ virtual bool trackLivenessAfterRegAlloc(const MachineFunction &MF) const;
+
virtual bool requiresFrameIndexScavenging(const MachineFunction &MF) const;
virtual bool requiresVirtualBaseRegisters(const MachineFunction &MF) const;
virtual bool requiresRegisterScavenging(const MachineFunction &MF) const
{ return true; }
+ //! Enable tracking of liveness after register allocation, since register
+ // scavenging is enabled.
+ virtual bool trackLivenessAfterRegAlloc(const MachineFunction &MF) const
+ { return true; }
+
//! Return the reserved registers
BitVector getReservedRegs(const MachineFunction &MF) const;
return true;
}
+ bool trackLivenessAfterRegAlloc(const MachineFunction &MF) const {
+ return true;
+ }
+
// Debug information queries.
unsigned getRARegister() const;
unsigned getFrameRegister(const MachineFunction &MF) const;
return true;
}
+bool
+MipsRegisterInfo::trackLivenessAfterRegAlloc(const MachineFunction &MF) const {
+ return true;
+}
+
// This function eliminate ADJCALLSTACKDOWN,
// ADJCALLSTACKUP pseudo instructions
void MipsRegisterInfo::
virtual bool requiresRegisterScavenging(const MachineFunction &MF) const;
+ virtual bool trackLivenessAfterRegAlloc(const MachineFunction &MF) const;
+
void eliminateCallFramePseudoInstr(MachineFunction &MF,
MachineBasicBlock &MBB,
MachineBasicBlock::iterator I) const;
ImmToIdxMap[PPC::ADDI8] = PPC::ADD8; ImmToIdxMap[PPC::STD_32] = PPC::STDX_32;
}
+bool
+PPCRegisterInfo::trackLivenessAfterRegAlloc(const MachineFunction &MF) const {
+ return requiresRegisterScavenging(MF);
+}
+
+
/// getPointerRegClass - Return the register class to use to hold pointers.
/// This is used for addressing modes.
const TargetRegisterClass *
/// FIXME (64-bit): Should be inlined.
bool requiresRegisterScavenging(const MachineFunction &MF) const;
+ bool trackLivenessAfterRegAlloc(const MachineFunction &MF) const;
+
void eliminateCallFramePseudoInstr(MachineFunction &MF,
MachineBasicBlock &MBB,
MachineBasicBlock::iterator I) const;
return -1;
}
+bool
+X86RegisterInfo::trackLivenessAfterRegAlloc(const MachineFunction &MF) const {
+ // Only enable when post-RA scheduling is enabled and this is needed.
+ return TM.getSubtargetImpl()->postRAScheduler();
+}
+
int
X86RegisterInfo::getSEHRegNum(unsigned i) const {
int reg = X86_MC::getX86RegNum(i);
int getCompactUnwindRegNum(unsigned RegNum, bool isEH) const;
/// Code Generation virtual methods...
- ///
+ ///
+ virtual bool trackLivenessAfterRegAlloc(const MachineFunction &MF) const;
/// getMatchingSuperRegClass - Return a subclass of the specified register
/// class A so that each register in it has a sub-register of the
CodeGenOpt::Level OptLevel,
TargetSubtargetInfo::AntiDepBreakMode& Mode,
RegClassVector& CriticalPathRCs) const {
- //TODO: change back to ANTIDEP_CRITICAL when the
- // X86 subtarget properly sets up post RA liveness.
- Mode = TargetSubtargetInfo::ANTIDEP_NONE;
+ Mode = TargetSubtargetInfo::ANTIDEP_CRITICAL;
CriticalPathRCs.clear();
return PostRAScheduler && OptLevel >= CodeGenOpt::Default;
}
TargetSubtargetInfo::AntiDepBreakMode& Mode,
RegClassVector& CriticalPathRCs) const;
+ bool postRAScheduler() const { return PostRAScheduler; }
+
/// getInstrItins = Return the instruction itineraries based on the
/// subtarget selection.
const InstrItineraryData &getInstrItineraryData() const { return InstrItins; }
return TFI->hasFP(MF);
}
+bool
+XCoreRegisterInfo::trackLivenessAfterRegAlloc(const MachineFunction &MF) const {
+ return requiresRegisterScavenging(MF);
+}
+
bool
XCoreRegisterInfo::useFPForScavengingIndex(const MachineFunction &MF) const {
return false;
bool requiresRegisterScavenging(const MachineFunction &MF) const;
+ bool trackLivenessAfterRegAlloc(const MachineFunction &MF) const;
+
bool useFPForScavengingIndex(const MachineFunction &MF) const;
void eliminateCallFramePseudoInstr(MachineFunction &MF,
-; XFAIL: *
; RUN: llc <%s -O2 -mcpu=atom -march=x86 -relocation-model=static | FileCheck -check-prefix=atom %s
; RUN: llc <%s -O2 -mcpu=core2 -march=x86 -relocation-model=static | FileCheck %s
;
-; FIXME: Atom's scheduler is temporarily disabled.
-; XFAIL: *
@a = common global i32 0, align 4
@b = common global i32 0, align 4
; Without list-burr scheduling we may not see the difference in codegen here.
-; RUN: llc < %s -march=x86-64 -post-RA-scheduler -pre-RA-sched=list-burr -break-anti-dependencies=none > %t
+; Use a subtarget that has post-RA scheduling enabled because the anti-dependency
+; breaker requires liveness information to be kept.
+; RUN: llc < %s -march=x86-64 -mcpu=atom -post-RA-scheduler -pre-RA-sched=list-burr -break-anti-dependencies=none > %t
; RUN: grep {%xmm0} %t | count 14
; RUN: not grep {%xmm1} %t
-; RUN: llc < %s -march=x86-64 -post-RA-scheduler -break-anti-dependencies=critical > %t
+; RUN: llc < %s -march=x86-64 -mcpu=atom -post-RA-scheduler -break-anti-dependencies=critical > %t
; RUN: grep {%xmm0} %t | count 7
; RUN: grep {%xmm1} %t | count 7