[llvm] r309651 - [StackColoring] Update AliasAnalysis information in stack coloring pass
Hiroshi 7 Inoue via llvm-commits
llvm-commits at lists.llvm.org
Tue Oct 3 08:47:53 PDT 2017
Hi Mikael, Mattias,
It looks like a bug in my patch. I do not intend to change that behavior.
Maybe we need `return false` depending on the results of
getUnderlyingObjectsForCodeGen.
I am sorry for bothering you.
-----
Hiroshi Inoue <inouehrs at jp.ibm.com>
IBM Research - Tokyo
Mikael Holmén <mikael.holmen at ericsson.com> wrote on 2017/10/03 14:22:19:
> From: Mikael Holmén <mikael.holmen at ericsson.com>
> To: Hiroshi 7 Inoue <INOUEHRS at jp.ibm.com>
> Cc: Mattias Eriksson V <mattias.v.eriksson at ericsson.com>, <llvm-
> commits at lists.llvm.org>, Hal Finkel <hfinkel at anl.gov>, Matthias
> Braun <matze at braunis.de>
> Date: 2017/10/03 14:22
> Subject: Re: [llvm] r309651 - [StackColoring] Update AliasAnalysis
> information in stack coloring pass
>
> Hi Hiroshi,
>
> On 10/02/2017 06:37 PM, Mattias Eriksson V wrote:
> > On 10/02/2017 06:20 PM, Hiroshi 7 Inoue wrote:
> >> Hi Mikael,
> >>
> >> Thank you for sharing your problem. I will investigate it.
> >> Do you have this problem with the latest code base? I have updated my
> >> first patch (r309651) in https://urldefense.proofpoint.com/v2/
> url?u=https-3A__reviews.llvm.org_rL309849&d=DwIF-g&c=jf_iaSHvJObTbx-
> siA1ZOg&r=m0X4DAZw5Qh-99BsgZP4_s5n1wQ8Sz4aTCPvRS-
>
YkVI&m=LpTapfEiXCfaCXpeIggFCkGTbgEzp0_1t3xivK5Dkqg&s=FNJP45aHFjcZvjKgUenRJIUj7zz4zJUeAG5kNqCAsTE&e=
> .
> >
> > Hi, Hiroshi!
> >
> > Yes, we see the issue also with part 2 applied.
> >
>
> Yes as Mattias said part 2 suffers from the same problem since it also
> happily continues with memop #2 even after memop #1 was found to be
> non-identified, and fails to inform buildSchedGraph about it.
>
> Regards,
> Mikael
>
> > --
> > Mattias
> >
> >>
> >> Regards,
> >> Hiroshi
> >>
> >> -----
> >> Hiroshi Inoue <inouehrs at jp.ibm.com>
> >> IBM Research - Tokyo
> >>
> >>
> >> Mikael Holmén <mikael.holmen at ericsson.com> wrote on 2017/10/02
20:34:01:
> >>
> >> > From: Mikael Holmén <mikael.holmen at ericsson.com>
> >> > To: Hiroshi Inoue <inouehrs at jp.ibm.com>, <llvm-
> >> > commits at lists.llvm.org>, "Hal Finkel" <hfinkel at anl.gov>, Matthias
> >> > Braun <matze at braunis.de>
> >> > Cc: Mattias Eriksson V <mattias.v.eriksson at ericsson.com>
> >> > Date: 2017/10/02 20:35
> >> > Subject: Re: [llvm] r309651 - [StackColoring] Update AliasAnalysis
> >> > information in stack coloring pass
> >> >
> >> > Hi Hiroshi, Matthias and Hal,
> >> >
> >> > (CC my colleague Mattias)
> >> >
> >> > I think I've found a problem with this patch.
> >> >
> >> > Found it on my out-of-tree target, but I'll try to describe what's
> >> > happening.
> >> >
> >> > The problem is during scheduling where we fail to identify a
> >> dependency
> >> > between a store and a following load. The store _might_ store at
the
> >> > same address as the load loads from, but it doesn't always do it.
It
> >> > depends on earlier code flow.
> >> >
> >> > The store looks like:
> >> >
> >> > mv_a32_r16_rmod1 %a0_32, %r1; mem:ST2[%_tmp9] ST2[@g_290]
> >> >
> >> > I.e. it has two memops describing the two different memory
> >> locations it
> >> > might store to (is this weird?). They memory locations are
completely
> >> > unrelated and describe the two locations the pointer %r1 can point
to.
> >> >
> >> > Before this patch, when ScheduleDAGInstrs::buildSchedGraph called
> >> >
> >> > UnderlyingObjectsVector Objs;
> >> > getUnderlyingObjectsForInstr(&MI, MFI, Objs, MF.getDataLayout
());
> >> >
> >> > it got an empty Objs vector back since getUnderlyingObjectsForInstr
> >> > found one non-identified object (%_tmp9) and then cleared the whole
> >> vector.
> >> >
> >> > With this patch, I instead get a non-empty Objs, containing only
> >> @g_290
> >> > and because of that we don't add a chain dependency towards the
load
> >> > following the store, and we get wrong code as a result.
> >> >
> >> > I know there is code in the method getUnderlyingObjectsForCodeGen
that
> >> > looks for non-identified objects and if found it clears the Objects
> >> > arrays, but the way it's implemented it doesn't help!
> >> >
> >> > In my current case where there are two memops and the first one is
> >> > non-identified, we still continue with the second memop afterwards
and
> >> > add that underlying object to the vector, and leave that to the
> >> > schedgraph builder where we don't see any traces of any
non-identified
> >> > object, and we build a graph with missing dependencies
> >> >
> >> > I suppose this difference is not intended?
> >> >
> >> > Regards,
> >> > Mikael
> >> >
> >> > On 08/01/2017 05:32 AM, Hiroshi Inoue via llvm-commits wrote:
> >> > > Author: inouehrs
> >> > > Date: Mon Jul 31 20:32:15 2017
> >> > > New Revision: 309651
> >> > >
> >> > > URL: https://urldefense.proofpoint.com/v2/url?
> >> >
> >>
>
u=http-3A__llvm.org_viewvc_llvm-2Dproject-3Frev-3D309651-26view-3Drev&d=DwICaQ&c=jf_iaSHvJObTbx-
> >>
> >> > siA1ZOg&r=m0X4DAZw5Qh-99BsgZP4_s5n1wQ8Sz4aTCPvRS-
> >> >
> >>
>
YkVI&m=mIxa9fsPE91JBDyRBM9jO3Gnmf1VjsdJcurWlZTiDmA&s=e3Q5lQrWHghbIG_b8_wJ0PPVh5IF218IDKqkTOkGGV0&e=
> >>
> >> > > Log:
> >> > > [StackColoring] Update AliasAnalysis information in stack
> >> coloring pass
> >> > >
> >> > > Stack coloring pass need to maintain AliasAnalysis information
> >> > when merging stack slots of different types.
> >> > > Actually, there is a FIXME comment in StackColoring.cpp
> >> > >
> >> > > // FIXME: In order to enable the use of TBAA when using AA in
> >> CodeGen,
> >> > > // we'll also need to update the TBAA nodes in MMOs with values
> >> > > // derived from the merged allocas.
> >> > >
> >> > > But, TBAA has been already enabled in CodeGen without fixing this
> >> pass.
> >> > > The incorrect TBAA metadata results in recent failures in
> >> > bootstrap test on ppc64le (PR33928) by allowing unsafe instruction
> >> scheduling.
> >> > > Although we observed the problem on ppc64le, this is a platform
> >> > neutral issue.
> >> > >
> >> > > This patch makes the stack coloring pass maintains AliasAnalysis
> >> > information when merging multiple stack slots.
> >> > >
> >> > >
> >> > > Modified:
> >> > > llvm/trunk/include/llvm/Analysis/ValueTracking.h
> >> > > llvm/trunk/include/llvm/CodeGen/MachineFunction.h
> >> > > llvm/trunk/include/llvm/CodeGen/MachineInstr.h
> >> > > llvm/trunk/lib/Analysis/ValueTracking.cpp
> >> > > llvm/trunk/lib/CodeGen/MachineFunction.cpp
> >> > > llvm/trunk/lib/CodeGen/ScheduleDAGInstrs.cpp
> >> > > llvm/trunk/lib/CodeGen/StackColoring.cpp
> >> > >
> >> > > Modified: llvm/trunk/include/llvm/Analysis/ValueTracking.h
> >> > > URL: https://urldefense.proofpoint.com/v2/url?
> >> >
> >>
>
u=http-3A__llvm.org_viewvc_llvm-2Dproject_llvm_trunk_include_llvm_Analysis_ValueTracking.h-3Frev-3D309651-26r1-3D309650-26r2-3D309651-26view-3Ddiff&d=DwICaQ&c=jf_iaSHvJObTbx-
> >>
> >> > siA1ZOg&r=m0X4DAZw5Qh-99BsgZP4_s5n1wQ8Sz4aTCPvRS-
> >> >
> >>
>
YkVI&m=mIxa9fsPE91JBDyRBM9jO3Gnmf1VjsdJcurWlZTiDmA&s=KsDAR0RUJBPSTneO3i2jMCIKtjKOSAi1260oq3nZfkA&e=
> >>
> >> > >
> >> >
> >>
>
==============================================================================
> >>
> >> > > --- llvm/trunk/include/llvm/Analysis/ValueTracking.h (original)
> >> > > +++ llvm/trunk/include/llvm/Analysis/ValueTracking.h Mon Jul 31
> >> > 20:32:15 2017
> >> > > @@ -312,6 +312,12 @@ template <typename T> class ArrayRef;
> >> > > const DataLayout &DL, LoopInfo *LI
> >> = nullptr,
> >> > > unsigned MaxLookup = 6);
> >> > >
> >> > > + /// This is a wrapper around GetUnderlyingObjects and adds
> >> > support for basic
> >> > > + /// ptrtoint+arithmetic+inttoptr sequences.
> >> > > + void getUnderlyingObjectsForCodeGen(const Value *V,
> >> > > + SmallVectorImpl<Value *> &Objects,
> >> > > + const DataLayout &DL);
> >> > > +
> >> > > /// Return true if the only users of this pointer are
> >> lifetime markers.
> >> > > bool onlyUsedByLifetimeMarkers(const Value *V);
> >> > >
> >> > >
> >> > > Modified: llvm/trunk/include/llvm/CodeGen/MachineFunction.h
> >> > > URL: https://urldefense.proofpoint.com/v2/url?
> >> >
> >>
>
u=http-3A__llvm.org_viewvc_llvm-2Dproject_llvm_trunk_include_llvm_CodeGen_MachineFunction.h-3Frev-3D309651-26r1-3D309650-26r2-3D309651-26view-3Ddiff&d=DwICaQ&c=jf_iaSHvJObTbx-
> >>
> >> > siA1ZOg&r=m0X4DAZw5Qh-99BsgZP4_s5n1wQ8Sz4aTCPvRS-
> >> >
> >>
>
YkVI&m=mIxa9fsPE91JBDyRBM9jO3Gnmf1VjsdJcurWlZTiDmA&s=WJpDpSXhK9W0O2Y7ry9udwfGJezS0aC1GhsuQrPJ2a8&e=
> >>
> >> > >
> >> >
> >>
>
==============================================================================
> >>
> >> > > --- llvm/trunk/include/llvm/CodeGen/MachineFunction.h (original)
> >> > > +++ llvm/trunk/include/llvm/CodeGen/MachineFunction.h Mon Jul 31
> >> > 20:32:15 2017
> >> > > @@ -661,6 +661,12 @@ public:
> >> > > MachineMemOperand *getMachineMemOperand(const
> >> MachineMemOperand *MMO,
> >> > > int64_t Offset,
> >> uint64_t Size);
> >> > >
> >> > > + /// Allocate a new MachineMemOperand by copying an existing
one,
> >> > > + /// replacing only AliasAnalysis information.
> >> > MachineMemOperands are owned
> >> > > + /// by the MachineFunction and need not be explicitly
> >> deallocated.
> >> > > + MachineMemOperand *getMachineMemOperand(const
> >> MachineMemOperand *MMO,
> >> > > + const AAMDNodes
&AAInfo);
> >> > > +
> >> > > using OperandCapacity =
ArrayRecycler<MachineOperand>::Capacity;
> >> > >
> >> > > /// Allocate an array of MachineOperands. This is only
> >> > intended for use by
> >> > >
> >> > > Modified: llvm/trunk/include/llvm/CodeGen/MachineInstr.h
> >> > > URL: https://urldefense.proofpoint.com/v2/url?
> >> >
> >>
>
u=http-3A__llvm.org_viewvc_llvm-2Dproject_llvm_trunk_include_llvm_CodeGen_MachineInstr.h-3Frev-3D309651-26r1-3D309650-26r2-3D309651-26view-3Ddiff&d=DwICaQ&c=jf_iaSHvJObTbx-
> >>
> >> > siA1ZOg&r=m0X4DAZw5Qh-99BsgZP4_s5n1wQ8Sz4aTCPvRS-
> >> >
> >>
>
YkVI&m=mIxa9fsPE91JBDyRBM9jO3Gnmf1VjsdJcurWlZTiDmA&s=P1817BTam8xe_gWSipjZpR88qqNAdwIFFUXlfMpsB8I&e=
> >>
> >> > >
> >> >
> >>
>
==============================================================================
> >>
> >> > > --- llvm/trunk/include/llvm/CodeGen/MachineInstr.h (original)
> >> > > +++ llvm/trunk/include/llvm/CodeGen/MachineInstr.h Mon Jul 31
> >> 20:32:15 2017
> >> > > @@ -379,6 +379,9 @@ public:
> >> > > return NumMemRefs == 1;
> >> > > }
> >> > >
> >> > > + /// Return the number of memory operands.
> >> > > + unsigned getNumMemOperands() const { return NumMemRefs; }
> >> > > +
> >> > > /// API for querying MachineInstr properties. They are the
> >> > same as MCInstrDesc
> >> > > /// queries but they are bundle aware.
> >> > >
> >> > >
> >> > > Modified: llvm/trunk/lib/Analysis/ValueTracking.cpp
> >> > > URL: https://urldefense.proofpoint.com/v2/url?
> >> >
> >>
>
u=http-3A__llvm.org_viewvc_llvm-2Dproject_llvm_trunk_lib_Analysis_ValueTracking.cpp-3Frev-3D309651-26r1-3D309650-26r2-3D309651-26view-3Ddiff&d=DwICaQ&c=jf_iaSHvJObTbx-
> >>
> >> > siA1ZOg&r=m0X4DAZw5Qh-99BsgZP4_s5n1wQ8Sz4aTCPvRS-
> >> >
> >>
>
YkVI&m=mIxa9fsPE91JBDyRBM9jO3Gnmf1VjsdJcurWlZTiDmA&s=kZRZpGyrwG9zEe89JDudBCkYnFPMzYQafnJFPQaCzCg&e=
> >>
> >> > >
> >> >
> >>
>
==============================================================================
> >>
> >> > > --- llvm/trunk/lib/Analysis/ValueTracking.cpp (original)
> >> > > +++ llvm/trunk/lib/Analysis/ValueTracking.cpp Mon Jul 31 20:32:15
> >> 2017
> >> > > @@ -3277,6 +3277,70 @@ void llvm::GetUnderlyingObjects(Value *V
> >> > > } while (!Worklist.empty());
> >> > > }
> >> > >
> >> > > +/// This is the function that does the work of looking through
> >> basic
> >> > > +/// ptrtoint+arithmetic+inttoptr sequences.
> >> > > +static const Value *getUnderlyingObjectFromInt(const Value *V) {
> >> > > + do {
> >> > > + if (const Operator *U = dyn_cast<Operator>(V)) {
> >> > > + // If we find a ptrtoint, we can transfer control back to
the
> >> > > + // regular getUnderlyingObjectFromInt.
> >> > > + if (U->getOpcode() == Instruction::PtrToInt)
> >> > > + return U->getOperand(0);
> >> > > + // If we find an add of a constant, a multiplied value, or
> >> > a phi, it's
> >> > > + // likely that the other operand will lead us to the base
> >> > > + // object. We don't have to worry about the case where the
> >> > > + // object address is somehow being computed by the
multiply,
> >> > > + // because our callers only care when the result is an
> >> > > + // identifiable object.
> >> > > + if (U->getOpcode() != Instruction::Add ||
> >> > > + (!isa<ConstantInt>(U->getOperand(1)) &&
> >> > > + Operator::getOpcode(U->getOperand(1)) !=
> >> Instruction::Mul &&
> >> > > + !isa<PHINode>(U->getOperand(1))))
> >> > > + return V;
> >> > > + V = U->getOperand(0);
> >> > > + } else {
> >> > > + return V;
> >> > > + }
> >> > > + assert(V->getType()->isIntegerTy() && "Unexpected operand
> >> type!");
> >> > > + } while (true);
> >> > > +}
> >> > > +
> >> > > +/// This is a wrapper around GetUnderlyingObjects and adds
> >> > support for basic
> >> > > +/// ptrtoint+arithmetic+inttoptr sequences.
> >> > > +void llvm::getUnderlyingObjectsForCodeGen(const Value *V,
> >> > > + SmallVectorImpl<Value *> &Objects,
> >> > > + const DataLayout &DL) {
> >> > > + SmallPtrSet<const Value *, 16> Visited;
> >> > > + SmallVector<const Value *, 4> Working(1, V);
> >> > > + do {
> >> > > + V = Working.pop_back_val();
> >> > > +
> >> > > + SmallVector<Value *, 4> Objs;
> >> > > + GetUnderlyingObjects(const_cast<Value *>(V), Objs, DL);
> >> > > +
> >> > > + for (Value *V : Objs) {
> >> > > + // If GetUnderlyingObjects fails to find an identifiable
> >> object,
> >> > > + // getUnderlyingObjectsForCodeGen also fails for safety.
> >> > > + if (!isIdentifiedObject(V)) {
> >> > > + Objects.clear();
> >> > > + return;
> >> > > + }
> >> > > +
> >> > > + if (!Visited.insert(V).second)
> >> > > + continue;
> >> > > + if (Operator::getOpcode(V) == Instruction::IntToPtr) {
> >> > > + const Value *O =
> >> > > + getUnderlyingObjectFromInt(cast<User>(V)->getOperand
(0));
> >> > > + if (O->getType()->isPointerTy()) {
> >> > > + Working.push_back(O);
> >> > > + continue;
> >> > > + }
> >> > > + }
> >> > > + Objects.push_back(const_cast<Value *>(V));
> >> > > + }
> >> > > + } while (!Working.empty());
> >> > > +}
> >> > > +
> >> > > /// Return true if the only users of this pointer are lifetime
> >> markers.
> >> > > bool llvm::onlyUsedByLifetimeMarkers(const Value *V) {
> >> > > for (const User *U : V->users()) {
> >> > >
> >> > > Modified: llvm/trunk/lib/CodeGen/MachineFunction.cpp
> >> > > URL: https://urldefense.proofpoint.com/v2/url?
> >> >
> >>
>
u=http-3A__llvm.org_viewvc_llvm-2Dproject_llvm_trunk_lib_CodeGen_MachineFunction.cpp-3Frev-3D309651-26r1-3D309650-26r2-3D309651-26view-3Ddiff&d=DwICaQ&c=jf_iaSHvJObTbx-
> >>
> >> > siA1ZOg&r=m0X4DAZw5Qh-99BsgZP4_s5n1wQ8Sz4aTCPvRS-
> >> >
> >>
>
YkVI&m=mIxa9fsPE91JBDyRBM9jO3Gnmf1VjsdJcurWlZTiDmA&s=allS8M0h249MGa_Gx6zUREWD57UANqNdgqwsYI7vnBc&e=
> >>
> >> > >
> >> >
> >>
>
==============================================================================
> >>
> >> > > --- llvm/trunk/lib/CodeGen/MachineFunction.cpp (original)
> >> > > +++ llvm/trunk/lib/CodeGen/MachineFunction.cpp Mon Jul 31
> >> 20:32:15 2017
> >> > > @@ -330,6 +330,20 @@ MachineFunction::getMachineMemOperand(co
> >> > > MMO->getOrdering(),
> >> > MMO->getFailureOrdering());
> >> > > }
> >> > >
> >> > > +MachineMemOperand *
> >> > > +MachineFunction::getMachineMemOperand(const MachineMemOperand
*MMO,
> >> > > + const AAMDNodes &AAInfo) {
> >> > > + MachinePointerInfo MPI = MMO->getValue() ?
> >> > > + MachinePointerInfo(MMO->getValue(),
> >> MMO->getOffset()) :
> >> > > + MachinePointerInfo(MMO->getPseudoValue(),
> >> MMO->getOffset());
> >> > > +
> >> > > + return new (Allocator)
> >> > > + MachineMemOperand(MPI, MMO->getFlags(),
> >> MMO->getSize(),
> >> > > + MMO->getBaseAlignment(), AAInfo,
> >> > > + MMO->getRanges(),
> >> MMO->getSyncScopeID(),
> >> > > + MMO->getOrdering(),
> >> > MMO->getFailureOrdering());
> >> > > +}
> >> > > +
> >> > > MachineInstr::mmo_iterator
> >> > > MachineFunction::allocateMemRefsArray(unsigned long Num) {
> >> > > return Allocator.Allocate<MachineMemOperand *>(Num);
> >> > >
> >> > > Modified: llvm/trunk/lib/CodeGen/ScheduleDAGInstrs.cpp
> >> > > URL: https://urldefense.proofpoint.com/v2/url?
> >> >
> >>
>
u=http-3A__llvm.org_viewvc_llvm-2Dproject_llvm_trunk_lib_CodeGen_ScheduleDAGInstrs.cpp-3Frev-3D309651-26r1-3D309650-26r2-3D309651-26view-3Ddiff&d=DwICaQ&c=jf_iaSHvJObTbx-
> >>
> >> > siA1ZOg&r=m0X4DAZw5Qh-99BsgZP4_s5n1wQ8Sz4aTCPvRS-
> >> >
> >>
YkVI&m=mIxa9fsPE91JBDyRBM9jO3Gnmf1VjsdJcurWlZTiDmA&s=Ivq088QzLr4wNpGbN-
> >> > k6QDpkRPbrJE6Z8WYbQgbAfXE&e=
> >> > >
> >> >
> >>
>
==============================================================================
> >>
> >> > > --- llvm/trunk/lib/CodeGen/ScheduleDAGInstrs.cpp (original)
> >> > > +++ llvm/trunk/lib/CodeGen/ScheduleDAGInstrs.cpp Mon Jul 31
> >> 20:32:15 2017
> >> > > @@ -121,63 +121,6 @@ ScheduleDAGInstrs::ScheduleDAGInstrs(Mac
> >> > > SchedModel.init(ST.getSchedModel(), &ST, TII);
> >> > > }
> >> > >
> >> > > -/// This is the function that does the work of looking through
> >> basic
> >> > > -/// ptrtoint+arithmetic+inttoptr sequences.
> >> > > -static const Value *getUnderlyingObjectFromInt(const Value *V) {
> >> > > - do {
> >> > > - if (const Operator *U = dyn_cast<Operator>(V)) {
> >> > > - // If we find a ptrtoint, we can transfer control back to
the
> >> > > - // regular getUnderlyingObjectFromInt.
> >> > > - if (U->getOpcode() == Instruction::PtrToInt)
> >> > > - return U->getOperand(0);
> >> > > - // If we find an add of a constant, a multiplied value, or
> >> > a phi, it's
> >> > > - // likely that the other operand will lead us to the base
> >> > > - // object. We don't have to worry about the case where the
> >> > > - // object address is somehow being computed by the
multiply,
> >> > > - // because our callers only care when the result is an
> >> > > - // identifiable object.
> >> > > - if (U->getOpcode() != Instruction::Add ||
> >> > > - (!isa<ConstantInt>(U->getOperand(1)) &&
> >> > > - Operator::getOpcode(U->getOperand(1)) !=
> >> Instruction::Mul &&
> >> > > - !isa<PHINode>(U->getOperand(1))))
> >> > > - return V;
> >> > > - V = U->getOperand(0);
> >> > > - } else {
> >> > > - return V;
> >> > > - }
> >> > > - assert(V->getType()->isIntegerTy() && "Unexpected operand
> >> type!");
> >> > > - } while (true);
> >> > > -}
> >> > > -
> >> > > -/// This is a wrapper around GetUnderlyingObjects and adds
> >> > support for basic
> >> > > -/// ptrtoint+arithmetic+inttoptr sequences.
> >> > > -static void getUnderlyingObjects(const Value *V,
> >> > > - SmallVectorImpl<Value *>
&Objects,
> >> > > - const DataLayout &DL) {
> >> > > - SmallPtrSet<const Value *, 16> Visited;
> >> > > - SmallVector<const Value *, 4> Working(1, V);
> >> > > - do {
> >> > > - V = Working.pop_back_val();
> >> > > -
> >> > > - SmallVector<Value *, 4> Objs;
> >> > > - GetUnderlyingObjects(const_cast<Value *>(V), Objs, DL);
> >> > > -
> >> > > - for (Value *V : Objs) {
> >> > > - if (!Visited.insert(V).second)
> >> > > - continue;
> >> > > - if (Operator::getOpcode(V) == Instruction::IntToPtr) {
> >> > > - const Value *O =
> >> > > - getUnderlyingObjectFromInt(cast<User>(V)->getOperand
(0));
> >> > > - if (O->getType()->isPointerTy()) {
> >> > > - Working.push_back(O);
> >> > > - continue;
> >> > > - }
> >> > > - }
> >> > > - Objects.push_back(const_cast<Value *>(V));
> >> > > - }
> >> > > - } while (!Working.empty());
> >> > > -}
> >> > > -
> >> > > /// If this machine instr has memory reference information and
> >> > it can be tracked
> >> > > /// to a normal reference to a known object, return the Value
> >> > for that object.
> >> > > static void getUnderlyingObjectsForInstr(const MachineInstr
*MI,
> >> > > @@ -208,12 +151,10 @@ static void getUnderlyingObjectsForInstr
> >> > >
> >> > Objects.push_back(UnderlyingObjectsVector::value_type(PSV,
MayAlias));
> >> > > } else if (const Value *V = MMO->getValue()) {
> >> > > SmallVector<Value *, 4> Objs;
> >> > > - getUnderlyingObjects(V, Objs, DL);
> >> > > + getUnderlyingObjectsForCodeGen(V, Objs, DL);
> >> > >
> >> > > for (Value *V : Objs) {
> >> > > - if (!isIdentifiedObject(V))
> >> > > - return false;
> >> > > -
> >> > > + assert(isIdentifiedObject(V));
> >> > > Objects.push_back(UnderlyingObjectsVector::value_type(V, true));
> >> > > }
> >> > > } else
> >> > >
> >> > > Modified: llvm/trunk/lib/CodeGen/StackColoring.cpp
> >> > > URL: https://urldefense.proofpoint.com/v2/url?
> >> >
> >>
>
u=http-3A__llvm.org_viewvc_llvm-2Dproject_llvm_trunk_lib_CodeGen_StackColoring.cpp-3Frev-3D309651-26r1-3D309650-26r2-3D309651-26view-3Ddiff&d=DwICaQ&c=jf_iaSHvJObTbx-
> >>
> >> > siA1ZOg&r=m0X4DAZw5Qh-99BsgZP4_s5n1wQ8Sz4aTCPvRS-
> >> >
> >>
>
YkVI&m=mIxa9fsPE91JBDyRBM9jO3Gnmf1VjsdJcurWlZTiDmA&s=GtnObcIhIwZTCrC2KWaTzADR6e8JEcuoyWSr5_hAM1Q&e=
> >>
> >> > >
> >> >
> >>
>
==============================================================================
> >>
> >> > > --- llvm/trunk/lib/CodeGen/StackColoring.cpp (original)
> >> > > +++ llvm/trunk/lib/CodeGen/StackColoring.cpp Mon Jul 31 20:32:15
> >> 2017
> >> > > @@ -37,6 +37,7 @@
> >> > > #include "llvm/CodeGen/MachineRegisterInfo.h"
> >> > > #include "llvm/CodeGen/Passes.h"
> >> > > #include "llvm/CodeGen/PseudoSourceValue.h"
> >> > > +#include "llvm/CodeGen/SelectionDAGNodes.h"
> >> > > #include "llvm/CodeGen/SlotIndexes.h"
> >> > > #include "llvm/CodeGen/StackProtector.h"
> >> > > #include "llvm/CodeGen/WinEHFuncInfo.h"
> >> > > @@ -889,6 +890,10 @@ void StackColoring::remapInstructions(De
> >> > >
> >> > > // Keep a list of *allocas* which need to be remapped.
> >> > > DenseMap<const AllocaInst*, const AllocaInst*> Allocas;
> >> > > +
> >> > > + // Keep a list of allocas which has been affected by the
remap.
> >> > > + SmallPtrSet<const AllocaInst*, 32> MergedAllocas;
> >> > > +
> >> > > for (const std::pair<int, int> &SI : SlotRemap) {
> >> > > const AllocaInst *From = MFI->getObjectAllocation
(SI.first);
> >> > > const AllocaInst *To = MFI->getObjectAllocation(SI.second);
> >> > > @@ -908,6 +913,10 @@ void StackColoring::remapInstructions(De
> >> > > Inst = Cast;
> >> > > }
> >> > >
> >> > > + // We keep both slots to maintain AliasAnalysis metadata
later.
> >> > > + MergedAllocas.insert(From);
> >> > > + MergedAllocas.insert(To);
> >> > > +
> >> > > // Allow the stack protector to adjust its value map to
> >> > account for the
> >> > > // upcoming replacement.
> >> > > SP->adjustForColoring(From, To);
> >> > > @@ -939,13 +948,6 @@ void StackColoring::remapInstructions(De
> >> > >
> >> > > // Update the MachineMemOperand to use the new alloca.
> >> > > for (MachineMemOperand *MMO : I.memoperands()) {
> >> > > - // FIXME: In order to enable the use of TBAA when using
> >> > AA in CodeGen,
> >> > > - // we'll also need to update the TBAA nodes in MMOs with
> >> values
> >> > > - // derived from the merged allocas. When doing this,
> >> > we'll need to use
> >> > > - // the same variant of GetUnderlyingObjects that is used
> >> by the
> >> > > - // instruction scheduler (that can look through
> >> ptrtoint/inttoptr
> >> > > - // pairs).
> >> > > -
> >> > > // We've replaced IR-level uses of the remapped
> >> allocas,so we only
> >> > > // need to replace direct uses here.
> >> > > const AllocaInst *AI =
> >> > dyn_cast_or_null<AllocaInst>(MMO->getValue());
> >> > > @@ -997,6 +999,48 @@ void StackColoring::remapInstructions(De
> >> > > MO.setIndex(ToSlot);
> >> > > FixedInstr++;
> >> > > }
> >> > > +
> >> > > + // We adjust AliasAnalysis information for merged stack
> >> slots.
> >> > > + MachineSDNode::mmo_iterator MemOps =
> >> > > + MF->allocateMemRefsArray(I.getNumMemOperands());
> >> > > + unsigned MemOpIdx = 0;
> >> > > + bool ReplaceMemOps = false;
> >> > > + for (MachineMemOperand *MMO : I.memoperands()) {
> >> > > + // If this memory location can be a slot remapped here,
> >> > > + // we remove AA information.
> >> > > + bool MayHaveConflictingAAMD = false;
> >> > > + if (MMO->getAAInfo()) {
> >> > > + if (const Value *MMOV = MMO->getValue()) {
> >> > > + SmallVector<Value *, 4> Objs;
> >> > > + getUnderlyingObjectsForCodeGen(MMOV, Objs,
> >> > MF->getDataLayout());
> >> > > +
> >> > > + if (Objs.empty())
> >> > > + MayHaveConflictingAAMD = true;
> >> > > + else
> >> > > + for (Value *V : Objs) {
> >> > > + // If this memory location comes from a known
> >> stack slot
> >> > > + // that is not remapped, we continue checking.
> >> > > + // Otherwise, we need to invalidate AA
infomation.
> >> > > + const AllocaInst *AI =
> >> dyn_cast_or_null<AllocaInst>(V);
> >> > > + if (AI && MergedAllocas.count(AI)) {
> >> > > + MayHaveConflictingAAMD = true;
> >> > > + break;
> >> > > + }
> >> > > + }
> >> > > + }
> >> > > + }
> >> > > + if (MayHaveConflictingAAMD) {
> >> > > + MemOps[MemOpIdx++] = MF->getMachineMemOperand(MMO,
> >> AAMDNodes());
> >> > > + ReplaceMemOps = true;
> >> > > + }
> >> > > + else
> >> > > + MemOps[MemOpIdx++] = MMO;
> >> > > + }
> >> > > +
> >> > > + // If any memory operand is updated, set memory references
of
> >> > > + // this instruction.
> >> > > + if (ReplaceMemOps)
> >> > > + I.setMemRefs(std::make_pair(MemOps,
> >> I.getNumMemOperands()));
> >> > > }
> >> > >
> >> > > // Update the location of C++ catch objects for the MSVC
> >> > personality routine.
> >> > >
> >> > >
> >> > > _______________________________________________
> >> > > llvm-commits mailing list
> >> > > llvm-commits at lists.llvm.org
> >> > > https://urldefense.proofpoint.com/v2/url?
> >> >
> >>
>
u=http-3A__lists.llvm.org_cgi-2Dbin_mailman_listinfo_llvm-2Dcommits&d=DwICaQ&c=jf_iaSHvJObTbx-
> >>
> >> > siA1ZOg&r=m0X4DAZw5Qh-99BsgZP4_s5n1wQ8Sz4aTCPvRS-
> >> >
> >>
>
YkVI&m=mIxa9fsPE91JBDyRBM9jO3Gnmf1VjsdJcurWlZTiDmA&s=ybOAOXxp0X9Ut7wbe4GMXq94sxexsNZIn0-
> >>
> >> > _6pvOmdI&e=
> >> > >
> >> >
> >>
> >
> >
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20171004/76d29b96/attachment.html>
More information about the llvm-commits
mailing list