Skip to content

Arm64/Sve: Do not mark *Prefetch* APIs as MEMORY_LOAD #104396

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 3 commits into from
Jul 4, 2024
Merged

Arm64/Sve: Do not mark *Prefetch* APIs as MEMORY_LOAD #104396

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
eff666d
fix prefetch in proper way
kunalspathak Jul 3, 2024
73e20ba
jit format
kunalspathak Jul 3, 2024
1df8ffb
review comment
kunalspathak Jul 4, 2024
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
8 changes: 8 additions & 0 deletions src/coreclr/jit/fgdiagnostic.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -3459,6 +3459,14 @@ void Compiler::fgDebugCheckFlags(GenTree* tree, BasicBlock* block)

#if defined(TARGET_ARM64)
case NI_ArmBase_Yield:
case NI_Sve_PrefetchBytes:
case NI_Sve_PrefetchInt16:
case NI_Sve_PrefetchInt32:
case NI_Sve_PrefetchInt64:
case NI_Sve_GatherPrefetch16Bit:
case NI_Sve_GatherPrefetch32Bit:
case NI_Sve_GatherPrefetch64Bit:
case NI_Sve_GatherPrefetch8Bit:
{
assert(tree->OperRequiresCallFlag(this));
expectedFlags |= GTF_GLOB_REF;
Expand Down
24 changes: 16 additions & 8 deletions src/coreclr/jit/gentree.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -27459,17 +27459,9 @@ bool GenTreeHWIntrinsic::OperIsMemoryLoad(GenTree** pAddr) const
case NI_Sve_Load2xVectorAndUnzip:
case NI_Sve_Load3xVectorAndUnzip:
case NI_Sve_Load4xVectorAndUnzip:
case NI_Sve_PrefetchBytes:
case NI_Sve_PrefetchInt16:
case NI_Sve_PrefetchInt32:
case NI_Sve_PrefetchInt64:
addr = Op(2);
break;

case NI_Sve_GatherPrefetch8Bit:
case NI_Sve_GatherPrefetch16Bit:
case NI_Sve_GatherPrefetch32Bit:
case NI_Sve_GatherPrefetch64Bit:
case NI_Sve_GatherVector:
case NI_Sve_GatherVectorByteZeroExtend:
case NI_Sve_GatherVectorInt16SignExtend:
Expand Down Expand Up @@ -27963,6 +27955,14 @@ bool GenTreeHWIntrinsic::OperRequiresCallFlag() const

#if defined(TARGET_ARM64)
case NI_ArmBase_Yield:
case NI_Sve_PrefetchBytes:
case NI_Sve_PrefetchInt16:
case NI_Sve_PrefetchInt32:
case NI_Sve_PrefetchInt64:
case NI_Sve_GatherPrefetch16Bit:
case NI_Sve_GatherPrefetch32Bit:
case NI_Sve_GatherPrefetch64Bit:
case NI_Sve_GatherPrefetch8Bit:
{
return true;
}
Expand Down Expand Up @@ -28145,6 +28145,14 @@ void GenTreeHWIntrinsic::Initialize(NamedIntrinsic intrinsicId)

#if defined(TARGET_ARM64)
case NI_ArmBase_Yield:
case NI_Sve_PrefetchBytes:
case NI_Sve_PrefetchInt16:
case NI_Sve_PrefetchInt32:
case NI_Sve_PrefetchInt64:
case NI_Sve_GatherPrefetch16Bit:
case NI_Sve_GatherPrefetch32Bit:
case NI_Sve_GatherPrefetch64Bit:
case NI_Sve_GatherPrefetch8Bit:
{
// Mark as a call and global reference, much as is done for GT_KEEPALIVE
gtFlags |= (GTF_CALL | GTF_GLOB_REF);
Expand Down
24 changes: 0 additions & 24 deletions src/coreclr/jit/hwintrinsic.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1856,14 +1856,6 @@ GenTree* Compiler::impHWIntrinsic(NamedIntrinsic intrinsic,
break;

#elif defined(TARGET_ARM64)
case NI_Sve_GatherPrefetch8Bit:
case NI_Sve_GatherPrefetch16Bit:
case NI_Sve_GatherPrefetch32Bit:
case NI_Sve_GatherPrefetch64Bit:
assert(varTypeIsSIMD(op2->TypeGet()));
retNode->AsHWIntrinsic()->SetAuxiliaryJitType(getBaseJitTypeOfSIMDType(sigReader.op2ClsHnd));
break;

case NI_Sve_GatherVector:
case NI_Sve_GatherVectorByteZeroExtend:
case NI_Sve_GatherVectorInt16SignExtend:
Expand Down Expand Up @@ -1893,22 +1885,6 @@ GenTree* Compiler::impHWIntrinsic(NamedIntrinsic intrinsic,
assert(!isScalar);
retNode =
gtNewSimdHWIntrinsicNode(nodeRetType, op1, op2, op3, op4, intrinsic, simdBaseJitType, simdSize);

#if defined(TARGET_ARM64)
switch (intrinsic)
{
case NI_Sve_GatherPrefetch8Bit:
case NI_Sve_GatherPrefetch16Bit:
case NI_Sve_GatherPrefetch32Bit:
case NI_Sve_GatherPrefetch64Bit:
assert(varTypeIsSIMD(op3->TypeGet()));
retNode->AsHWIntrinsic()->SetAuxiliaryJitType(getBaseJitTypeOfSIMDType(sigReader.op3ClsHnd));
break;

default:
break;
}
#endif
break;
}

Expand Down
76 changes: 76 additions & 0 deletions src/coreclr/jit/hwintrinsicarm64.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -2725,6 +2725,82 @@ GenTree* Compiler::impSpecialIntrinsic(NamedIntrinsic intrinsic,
retNode->AsHWIntrinsic()->SetAuxiliaryJitType(op1BaseJitType);
break;
}
case NI_Sve_GatherPrefetch8Bit:
case NI_Sve_GatherPrefetch16Bit:
case NI_Sve_GatherPrefetch32Bit:
case NI_Sve_GatherPrefetch64Bit:
case NI_Sve_PrefetchBytes:
case NI_Sve_PrefetchInt16:
case NI_Sve_PrefetchInt32:
case NI_Sve_PrefetchInt64:
{
assert((sig->numArgs == 3) || (sig->numArgs == 4));
assert(!isScalar);

var_types argType = TYP_UNKNOWN;
CORINFO_CLASS_HANDLE argClass = NO_CLASS_HANDLE;
int immLowerBound = 0;
int immUpperBound = 0;

CORINFO_ARG_LIST_HANDLE arg1 = sig->args;
CORINFO_ARG_LIST_HANDLE arg2 = info.compCompHnd->getArgNext(arg1);
CORINFO_ARG_LIST_HANDLE arg3 = info.compCompHnd->getArgNext(arg2);

HWIntrinsicInfo::lookupImmBounds(intrinsic, simdSize, simdBaseType, 1, &immLowerBound, &immUpperBound);

if (sig->numArgs == 3)
{
argType = JITtype2varType(strip(info.compCompHnd->getArgType(sig, arg3, &argClass)));
op3 = getArgForHWIntrinsic(argType, argClass);

assert(HWIntrinsicInfo::isImmOp(intrinsic, op3));
op3 = addRangeCheckIfNeeded(intrinsic, op3, mustExpand, immLowerBound, immUpperBound);

argType = JITtype2varType(strip(info.compCompHnd->getArgType(sig, arg2, &argClass)));
op2 = getArgForHWIntrinsic(argType, argClass);
CorInfoType op2BaseJitType = getBaseJitTypeOfSIMDType(argClass);
argType = JITtype2varType(strip(info.compCompHnd->getArgType(sig, arg1, &argClass)));
op1 = impPopStack().val;

#ifdef DEBUG

if ((intrinsic == NI_Sve_GatherPrefetch8Bit) || (intrinsic == NI_Sve_GatherPrefetch16Bit) ||
(intrinsic == NI_Sve_GatherPrefetch32Bit) || (intrinsic == NI_Sve_GatherPrefetch64Bit))
{
assert(varTypeIsSIMD(op2->TypeGet()));
}
else
{
assert(varTypeIsIntegral(op2->TypeGet()));
}
#endif
retNode = gtNewSimdHWIntrinsicNode(retType, op1, op2, op3, intrinsic, simdBaseJitType, simdSize);
retNode->AsHWIntrinsic()->SetAuxiliaryJitType(op2BaseJitType);
}
else
{
CORINFO_ARG_LIST_HANDLE arg4 = info.compCompHnd->getArgNext(arg3);
argType = JITtype2varType(strip(info.compCompHnd->getArgType(sig, arg4, &argClass)));
op4 = getArgForHWIntrinsic(argType, argClass);

assert(HWIntrinsicInfo::isImmOp(intrinsic, op4));
op3 = addRangeCheckIfNeeded(intrinsic, op4, mustExpand, immLowerBound, immUpperBound);

argType = JITtype2varType(strip(info.compCompHnd->getArgType(sig, arg3, &argClass)));
op3 = getArgForHWIntrinsic(argType, argClass);
CorInfoType op3BaseJitType = getBaseJitTypeOfSIMDType(argClass);
argType = JITtype2varType(strip(info.compCompHnd->getArgType(sig, arg2, &argClass)));
op2 = getArgForHWIntrinsic(argType, argClass);
argType = JITtype2varType(strip(info.compCompHnd->getArgType(sig, arg1, &argClass)));
op1 = impPopStack().val;

assert(varTypeIsSIMD(op3->TypeGet()));
retNode = gtNewSimdHWIntrinsicNode(retType, op1, op2, op3, op4, intrinsic, simdBaseJitType, simdSize);
retNode->AsHWIntrinsic()->SetAuxiliaryJitType(op3BaseJitType);
}

break;
}

default:
{
Expand Down
Loading
Loading