r268131 - [CUDA] Copy host builtin types to NVPTXTargetInfo.
Justin Lebar via cfe-commits
cfe-commits at lists.llvm.org
Fri Apr 29 16:05:19 PDT 2016
Author: jlebar
Date: Fri Apr 29 18:05:19 2016
New Revision: 268131
URL: http://llvm.org/viewvc/llvm-project?rev=268131&view=rev
Log:
[CUDA] Copy host builtin types to NVPTXTargetInfo.
Summary:
Host and device types must match, otherwise when we pass values back and
forth between the host and device, we will get the wrong result.
This patch makes NVPTXTargetInfo inherit most of its type information
from the host's target info.
Reviewers: rsmith
Subscribers: cfe-commits, jhen, tra
Differential Revision: http://reviews.llvm.org/D19346
Added:
cfe/trunk/test/Preprocessor/cuda-types.cu
Modified:
cfe/trunk/include/clang/Basic/TargetOptions.h
cfe/trunk/lib/Basic/Targets.cpp
cfe/trunk/lib/Frontend/CompilerInstance.cpp
cfe/trunk/lib/Frontend/CompilerInvocation.cpp
Modified: cfe/trunk/include/clang/Basic/TargetOptions.h
URL: http://llvm.org/viewvc/llvm-project/cfe/trunk/include/clang/Basic/TargetOptions.h?rev=268131&r1=268130&r2=268131&view=diff
==============================================================================
--- cfe/trunk/include/clang/Basic/TargetOptions.h (original)
+++ cfe/trunk/include/clang/Basic/TargetOptions.h Fri Apr 29 18:05:19 2016
@@ -27,6 +27,10 @@ public:
/// target will be selected to match the host.
std::string Triple;
+ /// When compiling for the device side, contains the triple used to compile
+ /// for the host.
+ std::string HostTriple;
+
/// If given, the name of the target CPU to generate code for.
std::string CPU;
Modified: cfe/trunk/lib/Basic/Targets.cpp
URL: http://llvm.org/viewvc/llvm-project/cfe/trunk/lib/Basic/Targets.cpp?rev=268131&r1=268130&r2=268131&view=diff
==============================================================================
--- cfe/trunk/lib/Basic/Targets.cpp (original)
+++ cfe/trunk/lib/Basic/Targets.cpp Fri Apr 29 18:05:19 2016
@@ -65,6 +65,9 @@ static void defineCPUMacros(MacroBuilder
Builder.defineMacro("__tune_" + CPUName + "__");
}
+static TargetInfo *AllocateTarget(const llvm::Triple &Triple,
+ const TargetOptions &Opts);
+
//===----------------------------------------------------------------------===//
// Defines specific to certain operating systems.
//===----------------------------------------------------------------------===//
@@ -1627,7 +1630,7 @@ class NVPTXTargetInfo : public TargetInf
} GPU;
public:
- NVPTXTargetInfo(const llvm::Triple &Triple, const TargetOptions &)
+ NVPTXTargetInfo(const llvm::Triple &Triple, const TargetOptions &Opts)
: TargetInfo(Triple) {
BigEndian = false;
TLSSupported = false;
@@ -1639,6 +1642,65 @@ public:
NoAsmVariants = true;
// Set the default GPU to sm20
GPU = GK_SM20;
+
+ // If possible, get a TargetInfo for our host triple, so we can match its
+ // types.
+ llvm::Triple HostTriple(Opts.HostTriple);
+ if (HostTriple.isNVPTX())
+ return;
+ std::unique_ptr<TargetInfo> HostTarget(
+ AllocateTarget(llvm::Triple(Opts.HostTriple), Opts));
+ if (!HostTarget) {
+ return;
+ }
+
+ PointerWidth = HostTarget->getPointerWidth(/* AddrSpace = */ 0);
+ PointerAlign = HostTarget->getPointerAlign(/* AddrSpace = */ 0);
+ BoolWidth = HostTarget->getBoolWidth();
+ BoolAlign = HostTarget->getBoolAlign();
+ IntWidth = HostTarget->getIntWidth();
+ IntAlign = HostTarget->getIntAlign();
+ HalfWidth = HostTarget->getHalfWidth();
+ HalfAlign = HostTarget->getHalfAlign();
+ FloatWidth = HostTarget->getFloatWidth();
+ FloatAlign = HostTarget->getFloatAlign();
+ DoubleWidth = HostTarget->getDoubleWidth();
+ DoubleAlign = HostTarget->getDoubleAlign();
+ LongWidth = HostTarget->getLongWidth();
+ LongAlign = HostTarget->getLongAlign();
+ LongLongWidth = HostTarget->getLongLongWidth();
+ LongLongAlign = HostTarget->getLongLongAlign();
+ MinGlobalAlign = HostTarget->getMinGlobalAlign();
+ DefaultAlignForAttributeAligned =
+ HostTarget->getDefaultAlignForAttributeAligned();
+ SizeType = HostTarget->getSizeType();
+ IntMaxType = HostTarget->getIntMaxType();
+ PtrDiffType = HostTarget->getPtrDiffType(/* AddrSpace = */ 0);
+ IntPtrType = HostTarget->getIntPtrType();
+ WCharType = HostTarget->getWCharType();
+ WIntType = HostTarget->getWIntType();
+ Char16Type = HostTarget->getChar16Type();
+ Char32Type = HostTarget->getChar32Type();
+ Int64Type = HostTarget->getInt64Type();
+ SigAtomicType = HostTarget->getSigAtomicType();
+ ProcessIDType = HostTarget->getProcessIDType();
+
+ UseBitFieldTypeAlignment = HostTarget->useBitFieldTypeAlignment();
+ UseZeroLengthBitfieldAlignment =
+ HostTarget->useZeroLengthBitfieldAlignment();
+ UseExplicitBitFieldAlignment = HostTarget->useExplicitBitFieldAlignment();
+ ZeroLengthBitfieldBoundary = HostTarget->getZeroLengthBitfieldBoundary();
+
+ // Properties intentionally not copied from host:
+ // - LargeArrayMinWidth, LargeArrayAlign: Not visible across the
+ // host/device boundary.
+ // - SuitableAlign: Not visible across the host/device boundary, and may
+ // correctly be different on host/device, e.g. if host has wider vector
+ // types than device.
+ // - LongDoubleWidth, LongDoubleAlign: nvptx's long double type is the same
+ // as its double type, but that's not necessarily true on the host.
+ // TODO: nvcc emits a warning when using long double on device; we should
+ // do the same.
}
void getTargetDefines(const LangOptions &Opts,
MacroBuilder &Builder) const override {
Modified: cfe/trunk/lib/Frontend/CompilerInstance.cpp
URL: http://llvm.org/viewvc/llvm-project/cfe/trunk/lib/Frontend/CompilerInstance.cpp?rev=268131&r1=268130&r2=268131&view=diff
==============================================================================
--- cfe/trunk/lib/Frontend/CompilerInstance.cpp (original)
+++ cfe/trunk/lib/Frontend/CompilerInstance.cpp Fri Apr 29 18:05:19 2016
@@ -836,8 +836,9 @@ bool CompilerInstance::ExecuteAction(Fro
// Create TargetInfo for the other side of CUDA compilation.
if (getLangOpts().CUDA && !getFrontendOpts().AuxTriple.empty()) {
- std::shared_ptr<TargetOptions> TO(new TargetOptions);
+ auto TO = std::make_shared<TargetOptions>();
TO->Triple = getFrontendOpts().AuxTriple;
+ TO->HostTriple = getTarget().getTriple().str();
setAuxTarget(TargetInfo::CreateTargetInfo(getDiagnostics(), TO));
}
Modified: cfe/trunk/lib/Frontend/CompilerInvocation.cpp
URL: http://llvm.org/viewvc/llvm-project/cfe/trunk/lib/Frontend/CompilerInvocation.cpp?rev=268131&r1=268130&r2=268131&view=diff
==============================================================================
--- cfe/trunk/lib/Frontend/CompilerInvocation.cpp (original)
+++ cfe/trunk/lib/Frontend/CompilerInvocation.cpp Fri Apr 29 18:05:19 2016
@@ -2155,6 +2155,12 @@ bool CompilerInvocation::CreateFromArgs(
LangOpts.ObjCExceptions = 1;
}
+ // During CUDA device-side compilation, the aux triple is the triple used for
+ // host compilation.
+ if (LangOpts.CUDA && LangOpts.CUDAIsDevice) {
+ Res.getTargetOpts().HostTriple = Res.getFrontendOpts().AuxTriple;
+ }
+
// FIXME: Override value name discarding when asan or msan is used because the
// backend passes depend on the name of the alloca in order to print out
// names.
Added: cfe/trunk/test/Preprocessor/cuda-types.cu
URL: http://llvm.org/viewvc/llvm-project/cfe/trunk/test/Preprocessor/cuda-types.cu?rev=268131&view=auto
==============================================================================
--- cfe/trunk/test/Preprocessor/cuda-types.cu (added)
+++ cfe/trunk/test/Preprocessor/cuda-types.cu Fri Apr 29 18:05:19 2016
@@ -0,0 +1,27 @@
+// Check that types, widths, etc. match on the host and device sides of CUDA
+// compilations. Note that we filter out long double, as this is intentionally
+// different on host and device.
+
+// RUN: %clang --cuda-host-only -nocudainc -target i386-unknown-linux-gnu -x cuda -E -dM -o - /dev/null > %T/i386-host-defines
+// RUN: %clang --cuda-device-only -nocudainc -target i386-unknown-linux-gnu -x cuda -E -dM -o - /dev/null > %T/i386-device-defines
+// RUN: grep 'define __[^ ]*\(TYPE\|MAX\|SIZEOF|WIDTH\)' %T/i386-host-defines | grep -v '__LDBL\|_LONG_DOUBLE' > %T/i386-host-defines-filtered
+// RUN: grep 'define __[^ ]*\(TYPE\|MAX\|SIZEOF|WIDTH\)' %T/i386-device-defines | grep -v '__LDBL\|_LONG_DOUBLE' > %T/i386-device-defines-filtered
+// RUN: diff %T/i386-host-defines-filtered %T/i386-device-defines-filtered
+
+// RUN: %clang --cuda-host-only -nocudainc -target x86_64-unknown-linux-gnu -x cuda -E -dM -o - /dev/null > %T/x86_64-host-defines
+// RUN: %clang --cuda-device-only -nocudainc -target x86_64-unknown-linux-gnu -x cuda -E -dM -o - /dev/null > %T/x86_64-device-defines
+// RUN: grep 'define __[^ ]*\(TYPE\|MAX\|SIZEOF\|WIDTH\)' %T/x86_64-host-defines | grep -v '__LDBL\|_LONG_DOUBLE' > %T/x86_64-host-defines-filtered
+// RUN: grep 'define __[^ ]*\(TYPE\|MAX\|SIZEOF\|WIDTH\)' %T/x86_64-device-defines | grep -v '__LDBL\|_LONG_DOUBLE' > %T/x86_64-device-defines-filtered
+// RUN: diff %T/x86_64-host-defines-filtered %T/x86_64-device-defines-filtered
+
+// RUN: %clang --cuda-host-only -nocudainc -target powerpc64-unknown-linux-gnu -x cuda -E -dM -o - /dev/null > %T/powerpc64-host-defines
+// RUN: %clang --cuda-device-only -nocudainc -target powerpc64-unknown-linux-gnu -x cuda -E -dM -o - /dev/null > %T/powerpc64-device-defines
+// RUN: grep 'define __[^ ]*\(TYPE\|MAX\|SIZEOF\|WIDTH\)' %T/powerpc64-host-defines | grep -v '__LDBL\|_LONG_DOUBLE' > %T/powerpc64-host-defines-filtered
+// RUN: grep 'define __[^ ]*\(TYPE\|MAX\|SIZEOF\|WIDTH\)' %T/powerpc64-device-defines | grep -v '__LDBL\|_LONG_DOUBLE' > %T/powerpc64-device-defines-filtered
+// RUN: diff %T/powerpc64-host-defines-filtered %T/powerpc64-device-defines-filtered
+
+// RUN: %clang --cuda-host-only -nocudainc -target nvptx-nvidia-cuda -x cuda -E -dM -o - /dev/null > %T/nvptx-host-defines
+// RUN: %clang --cuda-device-only -nocudainc -target nvptx-nvidia-cuda -x cuda -E -dM -o - /dev/null > %T/nvptx-device-defines
+// RUN: grep 'define __[^ ]*\(TYPE\|MAX\|SIZEOF\|WIDTH\)' %T/nvptx-host-defines | grep -v '__LDBL\|_LONG_DOUBLE' > %T/nvptx-host-defines-filtered
+// RUN: grep 'define __[^ ]*\(TYPE\|MAX\|SIZEOF\|WIDTH\)' %T/nvptx-device-defines | grep -v '__LDBL\|_LONG_DOUBLE' > %T/nvptx-device-defines-filtered
+// RUN: diff %T/nvptx-host-defines-filtered %T/nvptx-device-defines-filtered
More information about the cfe-commits
mailing list