[clang] 1b7db40 - [HLSL][SPIR-V] Target `directx` is required
Author: Michael Liao Date: 2024-04-26T15:07:13-04:00 New Revision: 1b7db405b97cc350e2de243683273e9946fc0bd0 URL: https://github.com/llvm/llvm-project/commit/1b7db405b97cc350e2de243683273e9946fc0bd0 DIFF: https://github.com/llvm/llvm-project/commit/1b7db405b97cc350e2de243683273e9946fc0bd0.diff LOG: [HLSL][SPIR-V] Target `directx` is required - One of tests needs target directx Added: Modified: clang/test/Driver/hlsl-lang-targets-spirv.hlsl Removed: diff --git a/clang/test/Driver/hlsl-lang-targets-spirv.hlsl b/clang/test/Driver/hlsl-lang-targets-spirv.hlsl index b86c2e01f8d80e..61b10e1648c52b 100644 --- a/clang/test/Driver/hlsl-lang-targets-spirv.hlsl +++ b/clang/test/Driver/hlsl-lang-targets-spirv.hlsl @@ -1,4 +1,5 @@ // REQUIRES: spirv-registered-target +// REQUIRES: directx-registered-target // Supported targets // ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] 4bb5d48 - [clang][NFC] Fix CUDA clang-cl tests
Author: Michael Liao Date: 2024-04-09T11:55:31-04:00 New Revision: 4bb5d48584818646a31a1ba4bfbbd658b7dfbe67 URL: https://github.com/llvm/llvm-project/commit/4bb5d48584818646a31a1ba4bfbbd658b7dfbe67 DIFF: https://github.com/llvm/llvm-project/commit/4bb5d48584818646a31a1ba4bfbbd658b7dfbe67.diff LOG: [clang][NFC] Fix CUDA clang-cl tests - Add '--' argument to prevent interpreting intput files as options starting with '/'. Fix test failure after 2921a0928c71f4ee652a2478283e47ab5ffebf58. Added: Modified: clang/test/Driver/cuda-external-tools.cu Removed: diff --git a/clang/test/Driver/cuda-external-tools.cu b/clang/test/Driver/cuda-external-tools.cu index d9564d026b4faa..9ada0cf8595dc6 100644 --- a/clang/test/Driver/cuda-external-tools.cu +++ b/clang/test/Driver/cuda-external-tools.cu @@ -89,7 +89,7 @@ // Check -Xcuda-ptxas with clang-cl // RUN: %clang_cl -### -c -Xcuda-ptxas -foo1 \ // RUN: --offload-arch=sm_35 --cuda-path=%S/Inputs/CUDA/usr/local/cuda \ -// RUN: -Xcuda-ptxas -foo2 %s 2>&1 \ +// RUN: -Xcuda-ptxas -foo2 -- %s 2>&1 \ // RUN: | FileCheck -check-prefixes=CHECK,SM35,PTXAS-EXTRA %s // MacOS spot-checks ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] 81ae2a8 - [clang] Fix '-Wunused-variable' warnings. NFC
Author: Michael Liao
Date: 2023-12-24T22:00:57-05:00
New Revision: 81ae2a8bb01d38162e0269fc6819584af6d60b03
URL:
https://github.com/llvm/llvm-project/commit/81ae2a8bb01d38162e0269fc6819584af6d60b03
DIFF:
https://github.com/llvm/llvm-project/commit/81ae2a8bb01d38162e0269fc6819584af6d60b03.diff
LOG: [clang] Fix '-Wunused-variable' warnings. NFC
Added:
Modified:
clang/lib/Driver/ToolChains/Clang.cpp
Removed:
diff --git a/clang/lib/Driver/ToolChains/Clang.cpp
b/clang/lib/Driver/ToolChains/Clang.cpp
index 4783affd3220bc..70dc7e54aca125 100644
--- a/clang/lib/Driver/ToolChains/Clang.cpp
+++ b/clang/lib/Driver/ToolChains/Clang.cpp
@@ -3203,13 +3203,13 @@ static void RenderFloatingPointOptions(const ToolChain
, const Driver ,
options::OPT_fstrict_float_cast_overflow, false))
CmdArgs.push_back("-fno-strict-float-cast-overflow");
- if (const Arg *A = Args.getLastArg(options::OPT_fcx_limited_range))
+ if (Args.hasArg(options::OPT_fcx_limited_range))
CmdArgs.push_back("-fcx-limited-range");
- if (const Arg *A = Args.getLastArg(options::OPT_fcx_fortran_rules))
+ if (Args.hasArg(options::OPT_fcx_fortran_rules))
CmdArgs.push_back("-fcx-fortran-rules");
- if (const Arg *A = Args.getLastArg(options::OPT_fno_cx_limited_range))
+ if (Args.hasArg(options::OPT_fno_cx_limited_range))
CmdArgs.push_back("-fno-cx-limited-range");
- if (const Arg *A = Args.getLastArg(options::OPT_fno_cx_fortran_rules))
+ if (Args.hasArg(options::OPT_fno_cx_fortran_rules))
CmdArgs.push_back("-fno-cx-fortran-rules");
}
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] 7b12d8b - [clang][Tests] Fix shared build. NFC
Author: Michael Liao Date: 2023-10-12T12:24:18-04:00 New Revision: 7b12d8bf8a1ff1540e32345b045f813644708a71 URL: https://github.com/llvm/llvm-project/commit/7b12d8bf8a1ff1540e32345b045f813644708a71 DIFF: https://github.com/llvm/llvm-project/commit/7b12d8bf8a1ff1540e32345b045f813644708a71.diff LOG: [clang][Tests] Fix shared build. NFC Added: Modified: clang/unittests/AST/Interp/CMakeLists.txt Removed: diff --git a/clang/unittests/AST/Interp/CMakeLists.txt b/clang/unittests/AST/Interp/CMakeLists.txt index e8d41091af40cda..8fa5c85064dbce5 100644 --- a/clang/unittests/AST/Interp/CMakeLists.txt +++ b/clang/unittests/AST/Interp/CMakeLists.txt @@ -5,7 +5,10 @@ add_clang_unittest(InterpTests clang_target_link_libraries(InterpTests PRIVATE clangAST + clangASTMatchers clangBasic + clangFrontend + clangSerialization clangTooling ) ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] 4edde41 - [clang][APINotes] Fix build error due to `-fpermissive` on GCC. NFC
Author: Michael Liao
Date: 2023-08-17T15:00:39-04:00
New Revision: 4edde41daed5e5e0b9aab2322215ddc2535f4cfd
URL:
https://github.com/llvm/llvm-project/commit/4edde41daed5e5e0b9aab2322215ddc2535f4cfd
DIFF:
https://github.com/llvm/llvm-project/commit/4edde41daed5e5e0b9aab2322215ddc2535f4cfd.diff
LOG: [clang][APINotes] Fix build error due to `-fpermissive` on GCC. NFC
Added:
Modified:
clang/lib/APINotes/APINotesWriter.cpp
Removed:
diff --git a/clang/lib/APINotes/APINotesWriter.cpp
b/clang/lib/APINotes/APINotesWriter.cpp
index 1a3d66a547f6ad..aad4c886bdd66d 100644
--- a/clang/lib/APINotes/APINotesWriter.cpp
+++ b/clang/lib/APINotes/APINotesWriter.cpp
@@ -891,7 +891,7 @@ unsigned getFunctionInfoSize(const FunctionInfo ) {
}
/// Emit a serialized representation of the function information.
-static void emitFunctionInfo(raw_ostream , const FunctionInfo ) {
+void emitFunctionInfo(raw_ostream , const FunctionInfo ) {
emitCommonEntityInfo(OS, FI);
uint8_t flags = 0;
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] 2daf91d - Fix shared library build again from 1c9a800. NFC
Author: Michael Liao Date: 2023-05-24T14:09:38-04:00 New Revision: 2daf91dae3bc25d2ffb869d9781d9f4496a27d02 URL: https://github.com/llvm/llvm-project/commit/2daf91dae3bc25d2ffb869d9781d9f4496a27d02 DIFF: https://github.com/llvm/llvm-project/commit/2daf91dae3bc25d2ffb869d9781d9f4496a27d02.diff LOG: Fix shared library build again from 1c9a800. NFC Added: Modified: clang/unittests/Serialization/CMakeLists.txt Removed: diff --git a/clang/unittests/Serialization/CMakeLists.txt b/clang/unittests/Serialization/CMakeLists.txt index 6b82ad91e5ec..44e4ecb31436 100644 --- a/clang/unittests/Serialization/CMakeLists.txt +++ b/clang/unittests/Serialization/CMakeLists.txt @@ -1,6 +1,7 @@ set(LLVM_LINK_COMPONENTS BitReader BitstreamReader + FrontendOpenMP Support ) ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] 058f04e - [clang] Fix another case where CPlusPlus2b is still used.
Author: Michael Liao
Date: 2023-05-04T14:27:33-04:00
New Revision: 058f04ea7dcbafbeed271fa75ee65e41409b4479
URL:
https://github.com/llvm/llvm-project/commit/058f04ea7dcbafbeed271fa75ee65e41409b4479
DIFF:
https://github.com/llvm/llvm-project/commit/058f04ea7dcbafbeed271fa75ee65e41409b4479.diff
LOG: [clang] Fix another case where CPlusPlus2b is still used.
Added:
Modified:
clang/lib/Sema/SemaDeclCXX.cpp
Removed:
diff --git a/clang/lib/Sema/SemaDeclCXX.cpp b/clang/lib/Sema/SemaDeclCXX.cpp
index 5b7ee09ac4c7..208e34a40e48 100644
--- a/clang/lib/Sema/SemaDeclCXX.cpp
+++ b/clang/lib/Sema/SemaDeclCXX.cpp
@@ -8829,7 +8829,7 @@ bool Sema::CheckExplicitlyDefaultedComparison(Scope *S,
FunctionDecl *FD,
CheckConstexprParameterTypes(*this, FD, CheckConstexprKind::Diagnose)
&&
!Info.Constexpr) {
Diag(FD->getBeginLoc(),
- getLangOpts().CPlusPlus2b
+ getLangOpts().CPlusPlus23
?
diag::warn_cxx2b_compat_defaulted_comparison_constexpr_mismatch
: diag::ext_defaulted_comparison_constexpr_mismatch)
<< FD->isImplicit() << (int)DCK << FD->isConsteval();
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] b323b40 - [clang] Fix build after https://reviews.llvm.org/D149553
Author: Michael Liao
Date: 2023-05-04T14:23:46-04:00
New Revision: b323b407f76d22bfc08b1430f7952c03eb504288
URL:
https://github.com/llvm/llvm-project/commit/b323b407f76d22bfc08b1430f7952c03eb504288
DIFF:
https://github.com/llvm/llvm-project/commit/b323b407f76d22bfc08b1430f7952c03eb504288.diff
LOG: [clang] Fix build after https://reviews.llvm.org/D149553
- `CXXPre2bCompat` is referenced somewhere after being removed.
- More warning messages on c++2b need refining
Added:
Modified:
clang/include/clang/Basic/DiagnosticSemaKinds.td
Removed:
diff --git a/clang/include/clang/Basic/DiagnosticSemaKinds.td
b/clang/include/clang/Basic/DiagnosticSemaKinds.td
index f0ff7d0f3169..7df963f7ba05 100644
--- a/clang/include/clang/Basic/DiagnosticSemaKinds.td
+++ b/clang/include/clang/Basic/DiagnosticSemaKinds.td
@@ -9440,7 +9440,7 @@ def
warn_cxx2b_compat_defaulted_comparison_constexpr_mismatch : Warning<
"%select{|for which the corresponding implicit 'operator==' }0 "
"invokes a non-constexpr comparison function is incompatible with C++ "
"standards before C++2b">,
- InGroup, DefaultIgnore;
+ InGroup, DefaultIgnore;
def note_defaulted_comparison_not_constexpr : Note<
"non-constexpr comparison function would be used to compare "
"%select{|member %1|base class %1}0">;
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] 3c83480 - [clang][AST] Fix `-Wuninitialized`. NFC
Author: Michael Liao
Date: 2023-04-09T15:58:10-04:00
New Revision: 3c83480ae95dde9b5d45b6fd7cdb1c64332531d7
URL:
https://github.com/llvm/llvm-project/commit/3c83480ae95dde9b5d45b6fd7cdb1c64332531d7
DIFF:
https://github.com/llvm/llvm-project/commit/3c83480ae95dde9b5d45b6fd7cdb1c64332531d7.diff
LOG: [clang][AST] Fix `-Wuninitialized`. NFC
- Adjust the declaration order as non-static member are initialized in
order of declaration in the class definition.
Added:
Modified:
clang/lib/AST/MicrosoftMangle.cpp
Removed:
diff --git a/clang/lib/AST/MicrosoftMangle.cpp
b/clang/lib/AST/MicrosoftMangle.cpp
index d8c837bcade02..e0fd8abe5e3b8 100644
--- a/clang/lib/AST/MicrosoftMangle.cpp
+++ b/clang/lib/AST/MicrosoftMangle.cpp
@@ -327,8 +327,8 @@ class MicrosoftCXXNameMangler {
typedef llvm::DenseMap TemplateArgStringMap;
TemplateArgStringMap TemplateArgStrings;
- llvm::StringSaver TemplateArgStringStorage;
llvm::BumpPtrAllocator TemplateArgStringStorageAlloc;
+ llvm::StringSaver TemplateArgStringStorage;
typedef std::set> PassObjectSizeArgsSet;
PassObjectSizeArgsSet PassObjectSizeArgs;
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] cd61d2a - [clang][CodeGen][NFC] Fix `llvm-else-after-return`
Author: Michael Liao
Date: 2023-01-25T13:53:35-05:00
New Revision: cd61d2abe0fdfcee52e16998f7f3fda82572cd6f
URL:
https://github.com/llvm/llvm-project/commit/cd61d2abe0fdfcee52e16998f7f3fda82572cd6f
DIFF:
https://github.com/llvm/llvm-project/commit/cd61d2abe0fdfcee52e16998f7f3fda82572cd6f.diff
LOG: [clang][CodeGen][NFC] Fix `llvm-else-after-return`
Added:
Modified:
clang/lib/CodeGen/CodeGenModule.cpp
Removed:
diff --git a/clang/lib/CodeGen/CodeGenModule.cpp
b/clang/lib/CodeGen/CodeGenModule.cpp
index 12d602fed6932..38e6ec6634e88 100644
--- a/clang/lib/CodeGen/CodeGenModule.cpp
+++ b/clang/lib/CodeGen/CodeGenModule.cpp
@@ -3318,7 +3318,8 @@ void CodeGenModule::EmitGlobal(GlobalDecl GD) {
if (MustBeEmitted(Global))
EmitOMPDeclareReduction(DRD);
return;
-} else if (auto *DMD = dyn_cast(Global)) {
+}
+if (auto *DMD = dyn_cast(Global)) {
if (MustBeEmitted(Global))
EmitOMPDeclareMapper(DMD);
return;
@@ -4687,16 +4688,17 @@ LangAS CodeGenModule::GetGlobalVarAddressSpace(const
VarDecl *D) {
return LangAS::sycl_global;
if (LangOpts.CUDA && LangOpts.CUDAIsDevice) {
-if (D && D->hasAttr())
- return LangAS::cuda_constant;
-else if (D && D->hasAttr())
- return LangAS::cuda_shared;
-else if (D && D->hasAttr())
- return LangAS::cuda_device;
-else if (D && D->getType().isConstQualified())
- return LangAS::cuda_constant;
-else
- return LangAS::cuda_device;
+if (D) {
+ if (D->hasAttr())
+return LangAS::cuda_constant;
+ if (D->hasAttr())
+return LangAS::cuda_shared;
+ if (D->hasAttr())
+return LangAS::cuda_device;
+ if (D->getType().isConstQualified())
+return LangAS::cuda_constant;
+}
+return LangAS::cuda_device;
}
if (LangOpts.OpenMP) {
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] 036aeac - [Testing] Fix the shared build. NFC.
Author: Michael Liao Date: 2022-04-22T02:46:54-04:00 New Revision: 036aeac36c00f4390e861118f536150b366beaaf URL: https://github.com/llvm/llvm-project/commit/036aeac36c00f4390e861118f536150b366beaaf DIFF: https://github.com/llvm/llvm-project/commit/036aeac36c00f4390e861118f536150b366beaaf.diff LOG: [Testing] Fix the shared build. NFC. Added: Modified: clang/lib/Testing/CMakeLists.txt Removed: diff --git a/clang/lib/Testing/CMakeLists.txt b/clang/lib/Testing/CMakeLists.txt index 68ed32ba85b1e..49b6787959bc1 100644 --- a/clang/lib/Testing/CMakeLists.txt +++ b/clang/lib/Testing/CMakeLists.txt @@ -16,8 +16,11 @@ add_llvm_library(clangTesting clang_target_link_libraries(clangTesting PRIVATE + clangAST clangBasic clangFrontend + clangLex + clangSerialization ) target_link_libraries(clangTesting ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] 73ab5fd - [clang] Fix shared build. NFC.
Author: Michael Liao Date: 2022-03-30T14:05:14-04:00 New Revision: 73ab5fd17b5726543554621410124ebae953dc6b URL: https://github.com/llvm/llvm-project/commit/73ab5fd17b5726543554621410124ebae953dc6b DIFF: https://github.com/llvm/llvm-project/commit/73ab5fd17b5726543554621410124ebae953dc6b.diff LOG: [clang] Fix shared build. NFC. Added: Modified: clang/lib/ExtractAPI/CMakeLists.txt Removed: diff --git a/clang/lib/ExtractAPI/CMakeLists.txt b/clang/lib/ExtractAPI/CMakeLists.txt index 044caa0922483..f194ae342c20f 100644 --- a/clang/lib/ExtractAPI/CMakeLists.txt +++ b/clang/lib/ExtractAPI/CMakeLists.txt @@ -14,4 +14,5 @@ add_clang_library(clangExtractAPI clangBasic clangFrontend clangIndex + clangLex ) ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] 7505aee - [clang] Pacify GCC warning. NFC.
Author: Michael Liao
Date: 2022-01-03T11:05:36-05:00
New Revision: 7505aeefc4e615520e2c822b9647dad4a48276b9
URL:
https://github.com/llvm/llvm-project/commit/7505aeefc4e615520e2c822b9647dad4a48276b9
DIFF:
https://github.com/llvm/llvm-project/commit/7505aeefc4e615520e2c822b9647dad4a48276b9.diff
LOG: [clang] Pacify GCC warning. NFC.
- This partially reverts d677a7cb056b17145a50ec8ca2ab6d5f4c494749 to
pacify GCC warnings like
```
base class should be explicitly initialized in the copy constructor
```
- Shall we keep turning on option `IgnoreBaseInCopyConstructors` when
enabling `readability-redundant-member-init` check?
Added:
Modified:
clang/include/clang/Basic/Diagnostic.h
clang/include/clang/Basic/PartialDiagnostic.h
Removed:
diff --git a/clang/include/clang/Basic/Diagnostic.h
b/clang/include/clang/Basic/Diagnostic.h
index 6a80823d12422..e5577e74fa639 100644
--- a/clang/include/clang/Basic/Diagnostic.h
+++ b/clang/include/clang/Basic/Diagnostic.h
@@ -1326,7 +1326,7 @@ class DiagnosticBuilder : public StreamingDiagnostic {
public:
/// Copy constructor. When copied, this "takes" the diagnostic info from the
/// input and neuters it.
- DiagnosticBuilder(const DiagnosticBuilder ) {
+ DiagnosticBuilder(const DiagnosticBuilder ) : StreamingDiagnostic() {
DiagObj = D.DiagObj;
DiagStorage = D.DiagStorage;
IsActive = D.IsActive;
diff --git a/clang/include/clang/Basic/PartialDiagnostic.h
b/clang/include/clang/Basic/PartialDiagnostic.h
index 217441979869b..9fb70bff7fee1 100644
--- a/clang/include/clang/Basic/PartialDiagnostic.h
+++ b/clang/include/clang/Basic/PartialDiagnostic.h
@@ -49,7 +49,8 @@ class PartialDiagnostic : public StreamingDiagnostic {
PartialDiagnostic(unsigned DiagID, DiagStorageAllocator _)
: StreamingDiagnostic(Allocator_), DiagID(DiagID) {}
- PartialDiagnostic(const PartialDiagnostic ) : DiagID(Other.DiagID) {
+ PartialDiagnostic(const PartialDiagnostic )
+ : StreamingDiagnostic(), DiagID(Other.DiagID) {
Allocator = Other.Allocator;
if (Other.DiagStorage) {
DiagStorage = getStorage();
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] 17414b6 - Fix shared build of unittests.
Author: Michael Liao Date: 2021-12-10T15:33:56-05:00 New Revision: 17414b61245dcd3ae96c447762e8f776856a733c URL: https://github.com/llvm/llvm-project/commit/17414b61245dcd3ae96c447762e8f776856a733c DIFF: https://github.com/llvm/llvm-project/commit/17414b61245dcd3ae96c447762e8f776856a733c.diff LOG: Fix shared build of unittests. Added: Modified: clang/unittests/Analysis/FlowSensitive/CMakeLists.txt Removed: diff --git a/clang/unittests/Analysis/FlowSensitive/CMakeLists.txt b/clang/unittests/Analysis/FlowSensitive/CMakeLists.txt index d6f38c9404abc..f651cdeff1b55 100644 --- a/clang/unittests/Analysis/FlowSensitive/CMakeLists.txt +++ b/clang/unittests/Analysis/FlowSensitive/CMakeLists.txt @@ -1,4 +1,5 @@ set(LLVM_LINK_COMPONENTS + FrontendOpenMP Support ) ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] bf22593 - [InferAddressSpaces] Support assumed addrspaces from addrspace predicates.
Author: Michael Liao
Date: 2021-11-08T16:51:57-05:00
New Revision: bf225939bc3acf936c962f24423d3bb5ddd4c93f
URL:
https://github.com/llvm/llvm-project/commit/bf225939bc3acf936c962f24423d3bb5ddd4c93f
DIFF:
https://github.com/llvm/llvm-project/commit/bf225939bc3acf936c962f24423d3bb5ddd4c93f.diff
LOG: [InferAddressSpaces] Support assumed addrspaces from addrspace predicates.
- CUDA cannot associate memory space with pointer types. Even though Clang
could add extra attributes to specify the address space explicitly on a pointer
type, it breaks the portability between Clang and NVCC.
- This change proposes to assume the address space from a pointer from the
assumption built upon target-specific address space predicates, such as
`__isGlobal` from CUDA. E.g.,
```
foo(float *p) {
__builtin_assume(__isGlobal(p));
// From there, we could assume p is a global pointer instead of a
// generic one.
}
```
This makes the code portable without introducing the implementation-specific
features.
Note that NVCC starts to support __builtin_assume from version 11.
Reviewed By: arsenm
Differential Revision: https://reviews.llvm.org/D112041
Added:
llvm/test/Transforms/InferAddressSpaces/AMDGPU/builtin-assumed-addrspace.ll
llvm/test/Transforms/InferAddressSpaces/NVPTX/builtin-assumed-addrspace.ll
Modified:
clang/test/CodeGen/thinlto-distributed-newpm.ll
llvm/include/llvm/Analysis/AssumptionCache.h
llvm/include/llvm/Analysis/TargetTransformInfo.h
llvm/include/llvm/Analysis/TargetTransformInfoImpl.h
llvm/include/llvm/CodeGen/BasicTTIImpl.h
llvm/include/llvm/Target/TargetMachine.h
llvm/lib/Analysis/AssumptionCache.cpp
llvm/lib/Analysis/TargetTransformInfo.cpp
llvm/lib/Target/AMDGPU/AMDGPUTargetMachine.cpp
llvm/lib/Target/AMDGPU/AMDGPUTargetMachine.h
llvm/lib/Target/NVPTX/NVPTXTargetMachine.cpp
llvm/lib/Target/NVPTX/NVPTXTargetMachine.h
llvm/lib/Transforms/Scalar/InferAddressSpaces.cpp
llvm/test/Other/loop-pm-invalidation.ll
llvm/test/Other/new-pass-manager.ll
llvm/test/Other/new-pm-lto-defaults.ll
llvm/test/Other/new-pm-thinlto-prelink-pgo-defaults.ll
llvm/test/Transforms/LoopRotate/pr35210.ll
llvm/unittests/Analysis/AssumeBundleQueriesTest.cpp
Removed:
diff --git a/clang/test/CodeGen/thinlto-distributed-newpm.ll
b/clang/test/CodeGen/thinlto-distributed-newpm.ll
index 8f7fc5e9b8411..87dc19f29e1ba 100644
--- a/clang/test/CodeGen/thinlto-distributed-newpm.ll
+++ b/clang/test/CodeGen/thinlto-distributed-newpm.ll
@@ -47,11 +47,11 @@
; CHECK-O: Running pass: PromotePass
; CHECK-O: Running analysis: DominatorTreeAnalysis on main
; CHECK-O: Running analysis: AssumptionAnalysis on main
+; CHECK-O: Running analysis: TargetIRAnalysis on main
; CHECK-O: Running pass: DeadArgumentEliminationPass
; CHECK-O: Running pass: InstCombinePass on main
; CHECK-O: Running analysis: TargetLibraryAnalysis on main
; CHECK-O: Running analysis: OptimizationRemarkEmitterAnalysis on main
-; CHECK-O: Running analysis: TargetIRAnalysis on main
; CHECK-O: Running analysis: AAManager on main
; CHECK-O: Running analysis: BasicAA on main
; CHECK-O: Running analysis: ScopedNoAliasAA on main
diff --git a/llvm/include/llvm/Analysis/AssumptionCache.h
b/llvm/include/llvm/Analysis/AssumptionCache.h
index 51d04bd8cf022..12dd9b04c9323 100644
--- a/llvm/include/llvm/Analysis/AssumptionCache.h
+++ b/llvm/include/llvm/Analysis/AssumptionCache.h
@@ -29,6 +29,7 @@ namespace llvm {
class AssumeInst;
class Function;
class raw_ostream;
+class TargetTransformInfo;
class Value;
/// A cache of \@llvm.assume calls within a function.
@@ -59,6 +60,8 @@ class AssumptionCache {
/// We track this to lazily populate our assumptions.
Function
+ TargetTransformInfo *TTI;
+
/// Vector of weak value handles to calls of the \@llvm.assume
/// intrinsic.
SmallVector AssumeHandles;
@@ -103,7 +106,8 @@ class AssumptionCache {
public:
/// Construct an AssumptionCache from a function by scanning all of
/// its instructions.
- AssumptionCache(Function ) : F(F) {}
+ AssumptionCache(Function , TargetTransformInfo *TTI = nullptr)
+ : F(F), TTI(TTI) {}
/// This cache is designed to be self-updating and so it should never be
/// invalidated.
@@ -174,9 +178,7 @@ class AssumptionAnalysis : public
AnalysisInfoMixin {
public:
using Result = AssumptionCache;
- AssumptionCache run(Function , FunctionAnalysisManager &) {
-return AssumptionCache(F);
- }
+ AssumptionCache run(Function , FunctionAnalysisManager &);
};
/// Printer pass for the \c AssumptionAnalysis results.
diff --git a/llvm/include/llvm/Analysis/TargetTransformInfo.h
b/llvm/include/llvm/Analysis/TargetTransformInfo.h
index e3cf87612e9c3..4312c2ae0de63 100644
--- a/llvm/include/llvm/Analysis/TargetTransformInfo.h
+++ b/llvm/include/llvm/Analysis/TargetTransformInfo.h
[clang] 6fe902d - [cuda] Add address space predicate funuctions.
Author: Michael Liao
Date: 2021-10-19T16:20:14-04:00
New Revision: 6fe902daf931dedf6e958b43c043cb57bb612daf
URL:
https://github.com/llvm/llvm-project/commit/6fe902daf931dedf6e958b43c043cb57bb612daf
DIFF:
https://github.com/llvm/llvm-project/commit/6fe902daf931dedf6e958b43c043cb57bb612daf.diff
LOG: [cuda] Add address space predicate funuctions.
- Add the missing NVVM predicate builtins on address space checking
- Redefine them as pure functions so that they could be used in
__builtin_assume.
Reviewed By: tra
Differential Revision: https://reviews.llvm.org/D112053
Added:
Modified:
clang/include/clang/Basic/BuiltinsNVPTX.def
clang/lib/Headers/__clang_cuda_runtime_wrapper.h
Removed:
diff --git a/clang/include/clang/Basic/BuiltinsNVPTX.def
b/clang/include/clang/Basic/BuiltinsNVPTX.def
index 907a99af532c3..7afee4dbc80bc 100644
--- a/clang/include/clang/Basic/BuiltinsNVPTX.def
+++ b/clang/include/clang/Basic/BuiltinsNVPTX.def
@@ -687,6 +687,12 @@ BUILTIN(__nvvm_ldg_f2, "E2fE2fC*", "")
BUILTIN(__nvvm_ldg_f4, "E4fE4fC*", "")
BUILTIN(__nvvm_ldg_d2, "E2dE2dC*", "")
+// Address space predicates.
+BUILTIN(__nvvm_isspacep_const, "bvC*", "nc")
+BUILTIN(__nvvm_isspacep_global, "bvC*", "nc")
+BUILTIN(__nvvm_isspacep_local, "bvC*", "nc")
+BUILTIN(__nvvm_isspacep_shared, "bvC*", "nc")
+
// Builtins to support WMMA instructions on sm_70
TARGET_BUILTIN(__hmma_m16n16k16_ld_a, "vi*iC*UiIi", "", AND(SM_70,PTX60))
TARGET_BUILTIN(__hmma_m16n16k16_ld_b, "vi*iC*UiIi", "", AND(SM_70,PTX60))
diff --git a/clang/lib/Headers/__clang_cuda_runtime_wrapper.h
b/clang/lib/Headers/__clang_cuda_runtime_wrapper.h
index 33aa25fb2d73c..512fc300fc344 100644
--- a/clang/lib/Headers/__clang_cuda_runtime_wrapper.h
+++ b/clang/lib/Headers/__clang_cuda_runtime_wrapper.h
@@ -271,7 +271,38 @@ static inline __device__ void __brkpt(int __c) {
__brkpt(); }
#undef __CUDABE__
#endif
#include "sm_20_atomic_functions.hpp"
+// Predicate functions used in `__builtin_assume` need to have no side effect.
+// However, sm_20_intrinsics.hpp doesn't define them with neither pure nor
+// const attribute. Rename definitions from sm_20_intrinsics.hpp and re-define
+// them as pure ones.
+#pragma push_macro("__isGlobal")
+#pragma push_macro("__isShared")
+#pragma push_macro("__isConstant")
+#pragma push_macro("__isLocal")
+#define __isGlobal __ignored_cuda___isGlobal
+#define __isShared __ignored_cuda___isShared
+#define __isConstant __ignored_cuda___isConstant
+#define __isLocal __ignored_cuda___isLocal
#include "sm_20_intrinsics.hpp"
+#pragma pop_macro("__isGlobal")
+#pragma pop_macro("__isShared")
+#pragma pop_macro("__isConstant")
+#pragma pop_macro("__isLocal")
+#pragma push_macro("__DEVICE__")
+#define __DEVICE__ static __device__ __forceinline__ __attribute__((const))
+__DEVICE__ unsigned int __isGlobal(const void *p) {
+ return __nvvm_isspacep_global(p);
+}
+__DEVICE__ unsigned int __isShared(const void *p) {
+ return __nvvm_isspacep_shared(p);
+}
+__DEVICE__ unsigned int __isConstant(const void *p) {
+ return __nvvm_isspacep_const(p);
+}
+__DEVICE__ unsigned int __isLocal(const void *p) {
+ return __nvvm_isspacep_local(p);
+}
+#pragma pop_macro("__DEVICE__")
#include "sm_32_atomic_functions.hpp"
// Don't include sm_30_intrinsics.h and sm_32_intrinsics.h. These define the
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] 6ec36d1 - [cuda] Mark builtin texture/surface reference variable as 'externally_initialized'.
Author: Michael Liao
Date: 2021-08-09T13:27:40-04:00
New Revision: 6ec36d18ec7b29b471bbe50502beb5b35de3975c
URL:
https://github.com/llvm/llvm-project/commit/6ec36d18ec7b29b471bbe50502beb5b35de3975c
DIFF:
https://github.com/llvm/llvm-project/commit/6ec36d18ec7b29b471bbe50502beb5b35de3975c.diff
LOG: [cuda] Mark builtin texture/surface reference variable as
'externally_initialized'.
- They need to be preserved even if there's no reference within the
device code as the host code may need to initialize them based on the
application logic.
Reviewed By: tra
Differential Revision: https://reviews.llvm.org/D107718
Added:
Modified:
clang/lib/CodeGen/CodeGenModule.cpp
clang/test/CodeGenCUDA/surface.cu
clang/test/CodeGenCUDA/texture.cu
Removed:
diff --git a/clang/lib/CodeGen/CodeGenModule.cpp
b/clang/lib/CodeGen/CodeGenModule.cpp
index 49a1396b58e3a..13520861fe9b6 100644
--- a/clang/lib/CodeGen/CodeGenModule.cpp
+++ b/clang/lib/CodeGen/CodeGenModule.cpp
@@ -4438,7 +4438,9 @@ void CodeGenModule::EmitGlobalVarDefinition(const VarDecl
*D,
if (GV && LangOpts.CUDA) {
if (LangOpts.CUDAIsDevice) {
if (Linkage != llvm::GlobalValue::InternalLinkage &&
- (D->hasAttr() || D->hasAttr()))
+ (D->hasAttr() || D->hasAttr() ||
+ D->getType()->isCUDADeviceBuiltinSurfaceType() ||
+ D->getType()->isCUDADeviceBuiltinTextureType()))
GV->setExternallyInitialized(true);
} else {
getCUDARuntime().internalizeDeviceSideVar(D, Linkage);
diff --git a/clang/test/CodeGenCUDA/surface.cu
b/clang/test/CodeGenCUDA/surface.cu
index 0bf17091081b1..eedae5473fcfc 100644
--- a/clang/test/CodeGenCUDA/surface.cu
+++ b/clang/test/CodeGenCUDA/surface.cu
@@ -19,7 +19,7 @@ struct __attribute__((device_builtin_surface_type))
surface : public
};
// On the device side, surface references are represented as `i64` handles.
-// DEVICE: @surf ={{.*}} addrspace(1) global i64 undef, align 4
+// DEVICE: @surf ={{.*}} addrspace(1) externally_initialized global i64 undef,
align 4
// On the host side, they remain in the original type.
// HOST: @surf = internal global %struct.surface
// HOST: @0 = private unnamed_addr constant [5 x i8] c"surf\00"
diff --git a/clang/test/CodeGenCUDA/texture.cu
b/clang/test/CodeGenCUDA/texture.cu
index 8a966194340aa..0bb8cd48dcaa7 100644
--- a/clang/test/CodeGenCUDA/texture.cu
+++ b/clang/test/CodeGenCUDA/texture.cu
@@ -19,8 +19,8 @@ struct __attribute__((device_builtin_texture_type)) texture :
public textureRefe
};
// On the device side, texture references are represented as `i64` handles.
-// DEVICE: @tex ={{.*}} addrspace(1) global i64 undef, align 4
-// DEVICE: @norm ={{.*}} addrspace(1) global i64 undef, align 4
+// DEVICE: @tex ={{.*}} addrspace(1) externally_initialized global i64 undef,
align 4
+// DEVICE: @norm ={{.*}} addrspace(1) externally_initialized global i64 undef,
align 4
// On the host side, they remain in the original type.
// HOST: @tex = internal global %struct.texture
// HOST: @norm = internal global %struct.texture
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] 4e5d9c8 - [Internalize] Preserve variables externally initialized.
Author: Michael Liao
Date: 2021-07-08T10:48:19-04:00
New Revision: 4e5d9c88033f1fc5d5206a02d8303bc6de43cf2b
URL:
https://github.com/llvm/llvm-project/commit/4e5d9c88033f1fc5d5206a02d8303bc6de43cf2b
DIFF:
https://github.com/llvm/llvm-project/commit/4e5d9c88033f1fc5d5206a02d8303bc6de43cf2b.diff
LOG: [Internalize] Preserve variables externally initialized.
- ``externally_initialized`` variables would be initialized or modified
elsewhere. Particularly, CUDA or HIP may have host code to initialize
or modify ``externally_initialized`` device variables, which may not
be explicitly referenced on the device side but may still be used
through the host side interfaces. Not preserving them triggers the
elimination of them in the GlobalDCE and breaks the user code.
Reviewed By: yaxunl
Differential Revision: https://reviews.llvm.org/D105135
Added:
llvm/test/Transforms/Internalize/externally-initialized.ll
Modified:
clang/test/CodeGenCUDA/host-used-device-var.cu
clang/test/CodeGenCUDA/unused-global-var.cu
llvm/lib/Transforms/IPO/Internalize.cpp
Removed:
diff --git a/clang/test/CodeGenCUDA/host-used-device-var.cu
b/clang/test/CodeGenCUDA/host-used-device-var.cu
index b94ef689b3162..6bb5757052946 100644
--- a/clang/test/CodeGenCUDA/host-used-device-var.cu
+++ b/clang/test/CodeGenCUDA/host-used-device-var.cu
@@ -15,14 +15,14 @@
#include "Inputs/cuda.h"
-// Check device variables used by neither host nor device functioins are not
kept.
-
-// DEV-NEG-NOT: @v1
+// DEV-DAG: @v1
__device__ int v1;
-// DEV-NEG-NOT: @v2
+// DEV-DAG: @v2
__constant__ int v2;
+// Check device variables used by neither host nor device functioins are not
kept.
+
// DEV-NEG-NOT: @_ZL2v3
static __device__ int v3;
diff --git a/clang/test/CodeGenCUDA/unused-global-var.cu
b/clang/test/CodeGenCUDA/unused-global-var.cu
index 1dbb3a22563c8..c091e83eda70a 100644
--- a/clang/test/CodeGenCUDA/unused-global-var.cu
+++ b/clang/test/CodeGenCUDA/unused-global-var.cu
@@ -15,14 +15,14 @@
// DCE before internalization. This test makes sure unused global variables
// are eliminated.
-// Check unused device/constant variables are eliminated.
-
-// NEGCHK-NOT: @v1
+// CHECK-DAG: @v1
__device__ int v1;
-// NEGCHK-NOT: @v2
+// CHECK-DAG: @v2
__constant__ int v2;
+// Check unused device/constant variables are eliminated.
+
// NEGCHK-NOT: @_ZL2v3
constexpr int v3 = 1;
diff --git a/llvm/lib/Transforms/IPO/Internalize.cpp
b/llvm/lib/Transforms/IPO/Internalize.cpp
index 008712c87988b..cf8da0baebe41 100644
--- a/llvm/lib/Transforms/IPO/Internalize.cpp
+++ b/llvm/lib/Transforms/IPO/Internalize.cpp
@@ -101,6 +101,12 @@ bool InternalizePass::shouldPreserveGV(const GlobalValue
) {
if (GV.hasDLLExportStorageClass())
return true;
+ // As the name suggests, externally initialized variables need preserving as
+ // they would be initialized elsewhere externally.
+ if (const auto *G = dyn_cast())
+if (G->isExternallyInitialized())
+ return true;
+
// Already local, has nothing to do.
if (GV.hasLocalLinkage())
return false;
diff --git a/llvm/test/Transforms/Internalize/externally-initialized.ll
b/llvm/test/Transforms/Internalize/externally-initialized.ll
new file mode 100644
index 0..4c24e53543db9
--- /dev/null
+++ b/llvm/test/Transforms/Internalize/externally-initialized.ll
@@ -0,0 +1,7 @@
+; RUN: opt < %s -internalize -S | FileCheck %s
+; RUN: opt < %s -passes=internalize -S | FileCheck %s
+
+; CHECK: @G0
+; CHECK-NOT: internal
+; CHECK-SAME: global i32
+@G0 = protected externally_initialized global i32 0, align 4
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] 948308e - Fix `-Wunused-variable` warning. NFC.
Author: Michael Liao
Date: 2021-06-28T22:50:36-04:00
New Revision: 948308ef34dc7da8bb741a85eb9941cc2b05d227
URL:
https://github.com/llvm/llvm-project/commit/948308ef34dc7da8bb741a85eb9941cc2b05d227
DIFF:
https://github.com/llvm/llvm-project/commit/948308ef34dc7da8bb741a85eb9941cc2b05d227.diff
LOG: Fix `-Wunused-variable` warning. NFC.
Added:
Modified:
clang/lib/CodeGen/CGCall.cpp
Removed:
diff --git a/clang/lib/CodeGen/CGCall.cpp b/clang/lib/CodeGen/CGCall.cpp
index 35b34179cc23..4ff6c632b61d 100644
--- a/clang/lib/CodeGen/CGCall.cpp
+++ b/clang/lib/CodeGen/CGCall.cpp
@@ -2173,7 +2173,7 @@ void CodeGenModule::ConstructAttributeList(StringRef Name,
// Add "sample-profile-suffix-elision-policy" attribute for internal linkage
// functions with -funique-internal-linkage-names.
if (TargetDecl && CodeGenOpts.UniqueInternalLinkageNames) {
-if (auto *Fn = dyn_cast(TargetDecl)) {
+if (isa(TargetDecl)) {
if (this->getFunctionLinkage(CalleeInfo.getCalleeDecl()) ==
llvm::GlobalValue::InternalLinkage)
FuncAttrs.addAttribute("sample-profile-suffix-elision-policy",
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] 01bf529 - Recommit of a2fdf9d4d734732a6fa9288f1ffdf12bf8618123.
Author: Michael Liao
Date: 2021-02-05T11:27:30-05:00
New Revision: 01bf529db2cf465b029e29e537807576bfcbc452
URL:
https://github.com/llvm/llvm-project/commit/01bf529db2cf465b029e29e537807576bfcbc452
DIFF:
https://github.com/llvm/llvm-project/commit/01bf529db2cf465b029e29e537807576bfcbc452.diff
LOG: Recommit of a2fdf9d4d734732a6fa9288f1ffdf12bf8618123.
- The failures are all cc1-based tests due to the missing `-aux-triple` options,
which is always prepared by the driver in CUDA/HIP compilation.
- Add extra check on the missing aux-targetinfo to prevent crashing.
[hip][cuda] Enable extended lambda support on Windows.
- On Windows, extended lambda has extra issues due to the numbering
schemes are different between the host compilation (Microsoft C++ ABI)
and the device compilation (Itanium C++ ABI. Additional device side
lambda number is required per lambda for the host compilation to
correctly mangle the device-side lambda name.
- A hybrid numbering context `MSHIPNumberingContext` is introduced to
number a lambda for both host- and device-compilations.
Reviewed By: rnk
Differential Revision: https://reviews.llvm.org/D69322
This reverts commit 4874ff02417916cc9ff994b34abcb5e563056546.
Added:
Modified:
clang/include/clang/AST/ASTContext.h
clang/include/clang/AST/DeclCXX.h
clang/include/clang/AST/Mangle.h
clang/include/clang/AST/MangleNumberingContext.h
clang/include/clang/Sema/Sema.h
clang/lib/AST/ASTImporter.cpp
clang/lib/AST/CXXABI.h
clang/lib/AST/DeclCXX.cpp
clang/lib/AST/ItaniumCXXABI.cpp
clang/lib/AST/ItaniumMangle.cpp
clang/lib/AST/MicrosoftCXXABI.cpp
clang/lib/CodeGen/CGCUDANV.cpp
clang/lib/Sema/SemaLambda.cpp
clang/lib/Sema/TreeTransform.h
clang/lib/Serialization/ASTReaderDecl.cpp
clang/lib/Serialization/ASTWriter.cpp
clang/test/CodeGenCUDA/unnamed-types.cu
Removed:
diff --git a/clang/include/clang/AST/ASTContext.h
b/clang/include/clang/AST/ASTContext.h
index ce47d54e44b0..ae69a68608b7 100644
--- a/clang/include/clang/AST/ASTContext.h
+++ b/clang/include/clang/AST/ASTContext.h
@@ -538,6 +538,9 @@ class ASTContext : public RefCountedBase {
/// need them (like static local vars).
llvm::MapVector MangleNumbers;
llvm::MapVector StaticLocalNumbers;
+ /// Mapping the associated device lambda mangling number if present.
+ mutable llvm::DenseMap
+ DeviceLambdaManglingNumbers;
/// Mapping that stores parameterIndex values for ParmVarDecls when
/// that value exceeds the bitfield size of ParmVarDeclBits.ParameterIndex.
diff --git a/clang/include/clang/AST/DeclCXX.h
b/clang/include/clang/AST/DeclCXX.h
index e32101bb2276..89006b1cfa7f 100644
--- a/clang/include/clang/AST/DeclCXX.h
+++ b/clang/include/clang/AST/DeclCXX.h
@@ -1735,6 +1735,12 @@ class CXXRecordDecl : public RecordDecl {
getLambdaData().HasKnownInternalLinkage = HasKnownInternalLinkage;
}
+ /// Set the device side mangling number.
+ void setDeviceLambdaManglingNumber(unsigned Num) const;
+
+ /// Retrieve the device side mangling number.
+ unsigned getDeviceLambdaManglingNumber() const;
+
/// Returns the inheritance model used for this record.
MSInheritanceModel getMSInheritanceModel() const;
diff --git a/clang/include/clang/AST/Mangle.h
b/clang/include/clang/AST/Mangle.h
index 6506ad542cc3..13b436cdca3e 100644
--- a/clang/include/clang/AST/Mangle.h
+++ b/clang/include/clang/AST/Mangle.h
@@ -107,6 +107,9 @@ class MangleContext {
virtual bool shouldMangleCXXName(const NamedDecl *D) = 0;
virtual bool shouldMangleStringLiteral(const StringLiteral *SL) = 0;
+ virtual bool isDeviceMangleContext() const { return false; }
+ virtual void setDeviceMangleContext(bool) {}
+
// FIXME: consider replacing raw_ostream & with something like SmallString &.
void mangleName(GlobalDecl GD, raw_ostream &);
virtual void mangleCXXName(GlobalDecl GD, raw_ostream &) = 0;
diff --git a/clang/include/clang/AST/MangleNumberingContext.h
b/clang/include/clang/AST/MangleNumberingContext.h
index f1ca6a05dbaf..eb33759682d6 100644
--- a/clang/include/clang/AST/MangleNumberingContext.h
+++ b/clang/include/clang/AST/MangleNumberingContext.h
@@ -52,6 +52,11 @@ class MangleNumberingContext {
/// this context.
virtual unsigned getManglingNumber(const TagDecl *TD,
unsigned MSLocalManglingNumber) = 0;
+
+ /// Retrieve the mangling number of a new lambda expression with the
+ /// given call operator within the device context. No device number is
+ /// assigned if there's no device numbering context is associated.
+ virtual unsigned getDeviceManglingNumber(const CXXMethodDecl *) { return 0; }
};
} // end namespace clang
diff --git a/clang/include/clang/Sema/Sema.h b/clang/include/clang/Sema/Sema.h
index ea20ada56abc..68420fcbb85f 100644
--- a/clang/include/clang/Sema/Sema.h
+++
[clang] a2fdf9d - [hip][cuda] Enable extended lambda support on Windows.
Author: Michael Liao
Date: 2021-02-04T01:38:29-05:00
New Revision: a2fdf9d4d734732a6fa9288f1ffdf12bf8618123
URL:
https://github.com/llvm/llvm-project/commit/a2fdf9d4d734732a6fa9288f1ffdf12bf8618123
DIFF:
https://github.com/llvm/llvm-project/commit/a2fdf9d4d734732a6fa9288f1ffdf12bf8618123.diff
LOG: [hip][cuda] Enable extended lambda support on Windows.
- On Windows, extended lambda has extra issues due to the numbering
schemes are different between the host compilation (Microsoft C++ ABI)
and the device compilation (Itanium C++ ABI. Additional device side
lambda number is required per lambda for the host compilation to
correctly mangle the device-side lambda name.
- A hybrid numbering context `MSHIPNumberingContext` is introduced to
number a lambda for both host- and device-compilations.
Reviewed By: rnk
Differential Revision: https://reviews.llvm.org/D69322
Added:
Modified:
clang/include/clang/AST/ASTContext.h
clang/include/clang/AST/DeclCXX.h
clang/include/clang/AST/Mangle.h
clang/include/clang/AST/MangleNumberingContext.h
clang/include/clang/Sema/Sema.h
clang/lib/AST/ASTImporter.cpp
clang/lib/AST/CXXABI.h
clang/lib/AST/DeclCXX.cpp
clang/lib/AST/ItaniumCXXABI.cpp
clang/lib/AST/ItaniumMangle.cpp
clang/lib/AST/MicrosoftCXXABI.cpp
clang/lib/CodeGen/CGCUDANV.cpp
clang/lib/Sema/SemaLambda.cpp
clang/lib/Sema/TreeTransform.h
clang/lib/Serialization/ASTReaderDecl.cpp
clang/lib/Serialization/ASTWriter.cpp
clang/test/CodeGenCUDA/ms-linker-options.cu
clang/test/CodeGenCUDA/unnamed-types.cu
Removed:
diff --git a/clang/include/clang/AST/ASTContext.h
b/clang/include/clang/AST/ASTContext.h
index ce47d54e44b0..ae69a68608b7 100644
--- a/clang/include/clang/AST/ASTContext.h
+++ b/clang/include/clang/AST/ASTContext.h
@@ -538,6 +538,9 @@ class ASTContext : public RefCountedBase {
/// need them (like static local vars).
llvm::MapVector MangleNumbers;
llvm::MapVector StaticLocalNumbers;
+ /// Mapping the associated device lambda mangling number if present.
+ mutable llvm::DenseMap
+ DeviceLambdaManglingNumbers;
/// Mapping that stores parameterIndex values for ParmVarDecls when
/// that value exceeds the bitfield size of ParmVarDeclBits.ParameterIndex.
diff --git a/clang/include/clang/AST/DeclCXX.h
b/clang/include/clang/AST/DeclCXX.h
index e32101bb2276..89006b1cfa7f 100644
--- a/clang/include/clang/AST/DeclCXX.h
+++ b/clang/include/clang/AST/DeclCXX.h
@@ -1735,6 +1735,12 @@ class CXXRecordDecl : public RecordDecl {
getLambdaData().HasKnownInternalLinkage = HasKnownInternalLinkage;
}
+ /// Set the device side mangling number.
+ void setDeviceLambdaManglingNumber(unsigned Num) const;
+
+ /// Retrieve the device side mangling number.
+ unsigned getDeviceLambdaManglingNumber() const;
+
/// Returns the inheritance model used for this record.
MSInheritanceModel getMSInheritanceModel() const;
diff --git a/clang/include/clang/AST/Mangle.h
b/clang/include/clang/AST/Mangle.h
index 6506ad542cc3..13b436cdca3e 100644
--- a/clang/include/clang/AST/Mangle.h
+++ b/clang/include/clang/AST/Mangle.h
@@ -107,6 +107,9 @@ class MangleContext {
virtual bool shouldMangleCXXName(const NamedDecl *D) = 0;
virtual bool shouldMangleStringLiteral(const StringLiteral *SL) = 0;
+ virtual bool isDeviceMangleContext() const { return false; }
+ virtual void setDeviceMangleContext(bool) {}
+
// FIXME: consider replacing raw_ostream & with something like SmallString &.
void mangleName(GlobalDecl GD, raw_ostream &);
virtual void mangleCXXName(GlobalDecl GD, raw_ostream &) = 0;
diff --git a/clang/include/clang/AST/MangleNumberingContext.h
b/clang/include/clang/AST/MangleNumberingContext.h
index f1ca6a05dbaf..eb33759682d6 100644
--- a/clang/include/clang/AST/MangleNumberingContext.h
+++ b/clang/include/clang/AST/MangleNumberingContext.h
@@ -52,6 +52,11 @@ class MangleNumberingContext {
/// this context.
virtual unsigned getManglingNumber(const TagDecl *TD,
unsigned MSLocalManglingNumber) = 0;
+
+ /// Retrieve the mangling number of a new lambda expression with the
+ /// given call operator within the device context. No device number is
+ /// assigned if there's no device numbering context is associated.
+ virtual unsigned getDeviceManglingNumber(const CXXMethodDecl *) { return 0; }
};
} // end namespace clang
diff --git a/clang/include/clang/Sema/Sema.h b/clang/include/clang/Sema/Sema.h
index 2fca81d25345..1c4942a37112 100644
--- a/clang/include/clang/Sema/Sema.h
+++ b/clang/include/clang/Sema/Sema.h
@@ -6558,7 +6558,7 @@ class Sema final {
/// Number lambda for linkage purposes if necessary.
void handleLambdaNumbering(
CXXRecordDecl *Class, CXXMethodDecl *Method,
- Optional> Mangling = None);
+ Optional> Mangling = None);
[clang] 7b5d7c7 - [hip] Fix `` compilation on Windows with VS2019.
Author: Michael Liao
Date: 2021-01-20T16:43:44-05:00
New Revision: 7b5d7c7b0a2479de007ad18b947459b71667
URL:
https://github.com/llvm/llvm-project/commit/7b5d7c7b0a2479de007ad18b947459b71667
DIFF:
https://github.com/llvm/llvm-project/commit/7b5d7c7b0a2479de007ad18b947459b71667.diff
LOG: [hip] Fix `` compilation on Windows with VS2019.
Differential Revision: https://reviews.llvm.org/D95075
Added:
Modified:
clang/lib/Headers/__clang_hip_cmath.h
Removed:
diff --git a/clang/lib/Headers/__clang_hip_cmath.h
b/clang/lib/Headers/__clang_hip_cmath.h
index 128d64e271b8..cd22a2df954b 100644
--- a/clang/lib/Headers/__clang_hip_cmath.h
+++ b/clang/lib/Headers/__clang_hip_cmath.h
@@ -626,6 +626,13 @@ _GLIBCXX_END_NAMESPACE_VERSION
// Define device-side math functions from on MSVC.
#if defined(_MSC_VER)
+
+// Before VS2019, `` is also included in `` and other headers.
+// But, from VS2019, it's only included in ``. Need to include
+// `` here to ensure C functions declared there won't be markded as
+// `__host__` and `__device__` through `` wrapper.
+#include
+
#if defined(__cplusplus)
extern "C" {
#endif // defined(__cplusplus)
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] f78d6af - [hip] Enable HIP compilation with ` on MSVC.
Author: Michael Liao
Date: 2021-01-07T17:41:28-05:00
New Revision: f78d6af7319aa676a0f9f6cbb982f21c96e9aac5
URL:
https://github.com/llvm/llvm-project/commit/f78d6af7319aa676a0f9f6cbb982f21c96e9aac5
DIFF:
https://github.com/llvm/llvm-project/commit/f78d6af7319aa676a0f9f6cbb982f21c96e9aac5.diff
LOG: [hip] Enable HIP compilation with ` on MSVC.
- MSVC has different `` implementation which calls into functions
declared in ``. Provide their device-side implementation to enable
`` compilation on HIP Windows.
Differential Revision: https://reviews.llvm.org/D93638
Added:
Modified:
clang/lib/Headers/__clang_hip_cmath.h
Removed:
diff --git a/clang/lib/Headers/__clang_hip_cmath.h
b/clang/lib/Headers/__clang_hip_cmath.h
index 3a702587ee17..128d64e271b8 100644
--- a/clang/lib/Headers/__clang_hip_cmath.h
+++ b/clang/lib/Headers/__clang_hip_cmath.h
@@ -624,6 +624,34 @@ _GLIBCXX_END_NAMESPACE_VERSION
} // namespace std
#endif
+// Define device-side math functions from on MSVC.
+#if defined(_MSC_VER)
+#if defined(__cplusplus)
+extern "C" {
+#endif // defined(__cplusplus)
+__DEVICE__ __attribute__((overloadable)) double _Cosh(double x, double y) {
+ return cosh(x) * y;
+}
+__DEVICE__ __attribute__((overloadable)) float _FCosh(float x, float y) {
+ return coshf(x) * y;
+}
+__DEVICE__ __attribute__((overloadable)) short _Dtest(double *p) {
+ return fpclassify(*p);
+}
+__DEVICE__ __attribute__((overloadable)) short _FDtest(float *p) {
+ return fpclassify(*p);
+}
+__DEVICE__ __attribute__((overloadable)) double _Sinh(double x, double y) {
+ return sinh(x) * y;
+}
+__DEVICE__ __attribute__((overloadable)) float _FSinh(float x, float y) {
+ return sinhf(x) * y;
+}
+#if defined(__cplusplus)
+}
+#endif // defined(__cplusplus)
+#endif // defined(_MSC_VER)
+
#pragma pop_macro("__DEVICE__")
#endif // __CLANG_HIP_CMATH_H__
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] 2a29ce3 - [hip] Fix HIP version parsing.
Author: Michael Liao
Date: 2021-01-06T17:00:14-05:00
New Revision: 2a29ce303451375bbf1de7c971296553ef5d9beb
URL:
https://github.com/llvm/llvm-project/commit/2a29ce303451375bbf1de7c971296553ef5d9beb
DIFF:
https://github.com/llvm/llvm-project/commit/2a29ce303451375bbf1de7c971296553ef5d9beb.diff
LOG: [hip] Fix HIP version parsing.
- Need trimming before parsing major or minor version numbers. This's required
due to the different line ending on Windows.
- In addition, the integer conversion may fail due to invalid char. Return that
parsing function return `true` when the parsing fails.
Differential Revision: https://reviews.llvm.org/D93587
Added:
Modified:
clang/lib/Driver/ToolChains/AMDGPU.cpp
clang/lib/Driver/ToolChains/ROCm.h
clang/test/Driver/Inputs/rocm/bin/.hipVersion
Removed:
diff --git a/clang/lib/Driver/ToolChains/AMDGPU.cpp
b/clang/lib/Driver/ToolChains/AMDGPU.cpp
index 565a77e07fd8..0971a2da62a3 100644
--- a/clang/lib/Driver/ToolChains/AMDGPU.cpp
+++ b/clang/lib/Driver/ToolChains/AMDGPU.cpp
@@ -88,23 +88,30 @@ void
RocmInstallationDetector::scanLibDevicePath(llvm::StringRef Path) {
}
}
-void RocmInstallationDetector::ParseHIPVersionFile(llvm::StringRef V) {
+// Parse and extract version numbers from `.hipVersion`. Return `true` if
+// the parsing fails.
+bool RocmInstallationDetector::parseHIPVersionFile(llvm::StringRef V) {
SmallVector VersionParts;
V.split(VersionParts, '\n');
- unsigned Major;
- unsigned Minor;
+ unsigned Major = ~0U;
+ unsigned Minor = ~0U;
for (auto Part : VersionParts) {
-auto Splits = Part.split('=');
-if (Splits.first == "HIP_VERSION_MAJOR")
- Splits.second.getAsInteger(0, Major);
-else if (Splits.first == "HIP_VERSION_MINOR")
- Splits.second.getAsInteger(0, Minor);
-else if (Splits.first == "HIP_VERSION_PATCH")
+auto Splits = Part.rtrim().split('=');
+if (Splits.first == "HIP_VERSION_MAJOR") {
+ if (Splits.second.getAsInteger(0, Major))
+return true;
+} else if (Splits.first == "HIP_VERSION_MINOR") {
+ if (Splits.second.getAsInteger(0, Minor))
+return true;
+} else if (Splits.first == "HIP_VERSION_PATCH")
VersionPatch = Splits.second.str();
}
+ if (Major == ~0U || Minor == ~0U)
+return true;
VersionMajorMinor = llvm::VersionTuple(Major, Minor);
DetectedVersion =
(Twine(Major) + "." + Twine(Minor) + "." + VersionPatch).str();
+ return false;
}
// For candidate specified by --rocm-path we do not do strict check.
@@ -290,7 +297,8 @@ void RocmInstallationDetector::detectHIPRuntime() {
continue;
if (HIPVersionArg.empty() && VersionFile)
- ParseHIPVersionFile((*VersionFile)->getBuffer());
+ if (parseHIPVersionFile((*VersionFile)->getBuffer()))
+continue;
HasHIPRuntime = true;
return;
diff --git a/clang/lib/Driver/ToolChains/ROCm.h
b/clang/lib/Driver/ToolChains/ROCm.h
index 27c7d8b0ee54..21e62a465d7b 100644
--- a/clang/lib/Driver/ToolChains/ROCm.h
+++ b/clang/lib/Driver/ToolChains/ROCm.h
@@ -103,7 +103,7 @@ class RocmInstallationDetector {
}
void scanLibDevicePath(llvm::StringRef Path);
- void ParseHIPVersionFile(llvm::StringRef V);
+ bool parseHIPVersionFile(llvm::StringRef V);
SmallVector getInstallationPathCandidates();
public:
diff --git a/clang/test/Driver/Inputs/rocm/bin/.hipVersion
b/clang/test/Driver/Inputs/rocm/bin/.hipVersion
index 48ee6f10c3e4..677293c09139 100644
--- a/clang/test/Driver/Inputs/rocm/bin/.hipVersion
+++ b/clang/test/Driver/Inputs/rocm/bin/.hipVersion
@@ -1,4 +1,6 @@
# Auto-generated by cmake
-HIP_VERSION_MAJOR=3
+# NOTE: The trailing whitespace is added on purpose to verify that these
+# whitespaces are trimmed before paring.
+HIP_VERSION_MAJOR=3
HIP_VERSION_MINOR=6
HIP_VERSION_PATCH=20214-a2917cd
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] bb8d20d - [cuda][hip] Fix typoes in header wrappers.
Author: Michael Liao Date: 2020-12-21T13:02:47-05:00 New Revision: bb8d20d9f3bb955ae6f6143d24749faf61d573a9 URL: https://github.com/llvm/llvm-project/commit/bb8d20d9f3bb955ae6f6143d24749faf61d573a9 DIFF: https://github.com/llvm/llvm-project/commit/bb8d20d9f3bb955ae6f6143d24749faf61d573a9.diff LOG: [cuda][hip] Fix typoes in header wrappers. Added: Modified: clang/lib/Headers/cuda_wrappers/algorithm clang/lib/Headers/cuda_wrappers/new Removed: diff --git a/clang/lib/Headers/cuda_wrappers/algorithm b/clang/lib/Headers/cuda_wrappers/algorithm index 01af18360d8d..f14a0b00bb04 100644 --- a/clang/lib/Headers/cuda_wrappers/algorithm +++ b/clang/lib/Headers/cuda_wrappers/algorithm @@ -1,4 +1,4 @@ -/*=== complex - CUDA wrapper for === +/*=== algorithm - CUDA wrapper for -=== * * Permission is hereby granted, free of charge, to any person obtaining a copy * of this software and associated documentation files (the "Software"), to deal diff --git a/clang/lib/Headers/cuda_wrappers/new b/clang/lib/Headers/cuda_wrappers/new index 7f255314056a..d5fb3b7011de 100644 --- a/clang/lib/Headers/cuda_wrappers/new +++ b/clang/lib/Headers/cuda_wrappers/new @@ -1,4 +1,4 @@ -/*=== complex - CUDA wrapper for --=== +/*=== new - CUDA wrapper for -=== * * Permission is hereby granted, free of charge, to any person obtaining a copy * of this software and associated documentation files (the "Software"), to deal ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] d8949a8 - [hip] Fix host object creation from fatbin
Author: Michael Liao Date: 2020-12-02T10:36:01-05:00 New Revision: d8949a8ad3ca2a39ffe69df76e2c3f5fd73efec0 URL: https://github.com/llvm/llvm-project/commit/d8949a8ad3ca2a39ffe69df76e2c3f5fd73efec0 DIFF: https://github.com/llvm/llvm-project/commit/d8949a8ad3ca2a39ffe69df76e2c3f5fd73efec0.diff LOG: [hip] Fix host object creation from fatbin - `__hip_fatbin` should a symbol in `.hip_fatbin` section. Differential Revision: https://reviews.llvm.org/D92418 Added: Modified: clang/lib/Driver/ToolChains/HIP.cpp Removed: diff --git a/clang/lib/Driver/ToolChains/HIP.cpp b/clang/lib/Driver/ToolChains/HIP.cpp index a06835eee024..fc1103b48a99 100644 --- a/clang/lib/Driver/ToolChains/HIP.cpp +++ b/clang/lib/Driver/ToolChains/HIP.cpp @@ -178,8 +178,7 @@ void AMDGCN::Linker::constructGenerateObjFileFromHIPFatBinary( ObjStream << "# HIP Object Generator\n"; ObjStream << "# *** Automatically generated by Clang ***\n"; ObjStream << " .type __hip_fatbin,@object\n"; - ObjStream << " .section .hip_fatbin,\"aMS\",@progbits,1\n"; - ObjStream << " .data\n"; + ObjStream << " .section .hip_fatbin,\"a\",@progbits\n"; ObjStream << " .globl __hip_fatbin\n"; ObjStream << " .p2align " << llvm::Log2(llvm::Align(HIPCodeObjectAlign)) << "\n"; ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] f375885 - [InferAddrSpace] Teach to handle assumed address space.
Author: Michael Liao
Date: 2020-11-16T17:06:33-05:00
New Revision: f375885ab86d1b3e82269725c8e9aa49f347b4a7
URL:
https://github.com/llvm/llvm-project/commit/f375885ab86d1b3e82269725c8e9aa49f347b4a7
DIFF:
https://github.com/llvm/llvm-project/commit/f375885ab86d1b3e82269725c8e9aa49f347b4a7.diff
LOG: [InferAddrSpace] Teach to handle assumed address space.
- In certain cases, a generic pointer could be assumed as a pointer to
the global memory space or other spaces. With a dedicated target hook
to query that address space from a given value, infer-address-space
pass could infer and propagate that to all its users.
Differential Revision: https://reviews.llvm.org/D91121
Added:
llvm/test/Transforms/InferAddressSpaces/AMDGPU/assumed-addrspace.ll
Modified:
clang/test/CodeGenCUDA/amdgpu-kernel-arg-pointer-type.cu
llvm/docs/AMDGPUUsage.rst
llvm/include/llvm/Analysis/TargetTransformInfo.h
llvm/include/llvm/Analysis/TargetTransformInfoImpl.h
llvm/include/llvm/CodeGen/BasicTTIImpl.h
llvm/include/llvm/Target/TargetMachine.h
llvm/lib/Analysis/TargetTransformInfo.cpp
llvm/lib/Target/AMDGPU/AMDGPUTargetMachine.cpp
llvm/lib/Target/AMDGPU/AMDGPUTargetMachine.h
llvm/lib/Transforms/Scalar/InferAddressSpaces.cpp
llvm/test/CodeGen/AMDGPU/GlobalISel/divergent-control-flow.ll
Removed:
diff --git a/clang/test/CodeGenCUDA/amdgpu-kernel-arg-pointer-type.cu
b/clang/test/CodeGenCUDA/amdgpu-kernel-arg-pointer-type.cu
index dc4659856026..da1f4b65f719 100644
--- a/clang/test/CodeGenCUDA/amdgpu-kernel-arg-pointer-type.cu
+++ b/clang/test/CodeGenCUDA/amdgpu-kernel-arg-pointer-type.cu
@@ -56,20 +56,24 @@ struct S {
int *x;
float *y;
};
-// `by-val` struct will be coerced into a similar struct with all generic
-// pointers lowerd into global ones.
+// `by-val` struct is passed by-indirect-alias (a mix of by-ref and indirect
+// by-val). However, the enhanced address inferring pass should be able to
+// assume they are global pointers.
+//
// HOST: define void @_Z22__device_stub__kernel41S(i32* %s.coerce0, float*
%s.coerce1)
// COMMON-LABEL: define amdgpu_kernel void @_Z7kernel41S(%struct.S
addrspace(4)*{{.*}} byref(%struct.S) align 8 %0)
// OPT: [[R0:%.*]] = getelementptr inbounds %struct.S, %struct.S addrspace(4)*
%0, i64 0, i32 0
// OPT: [[P0:%.*]] = load i32*, i32* addrspace(4)* [[R0]], align 8
+// OPT: [[G0:%.*]] = addrspacecast i32* [[P0]] to i32 addrspace(1)*
// OPT: [[R1:%.*]] = getelementptr inbounds %struct.S, %struct.S addrspace(4)*
%0, i64 0, i32 1
// OPT: [[P1:%.*]] = load float*, float* addrspace(4)* [[R1]], align 8
-// OPT: [[V0:%.*]] = load i32, i32* [[P0]], align 4
+// OPT: [[G1:%.*]] = addrspacecast float* [[P1]] to float addrspace(1)*
+// OPT: [[V0:%.*]] = load i32, i32 addrspace(1)* [[G0]], align 4
// OPT: [[INC:%.*]] = add nsw i32 [[V0]], 1
-// OPT: store i32 [[INC]], i32* [[P0]], align 4
-// OPT: [[V1:%.*]] = load float, float* [[P1]], align 4
+// OPT: store i32 [[INC]], i32 addrspace(1)* [[G0]], align 4
+// OPT: [[V1:%.*]] = load float, float addrspace(1)* [[G1]], align 4
// OPT: [[ADD:%.*]] = fadd contract float [[V1]], 1.00e+00
-// OPT: store float [[ADD]], float* [[P1]], align 4
+// OPT: store float [[ADD]], float addrspace(1)* [[G1]], align 4
// OPT: ret void
__global__ void kernel4(struct S s) {
s.x[0]++;
@@ -87,19 +91,24 @@ __global__ void kernel5(struct S *s) {
struct T {
float *x[2];
};
-// `by-val` array is also coerced.
+// `by-val` array is passed by-indirect-alias (a mix of by-ref and indirect
+// by-val). However, the enhanced address inferring pass should be able to
+// assume they are global pointers.
+//
// HOST: define void @_Z22__device_stub__kernel61T(float* %t.coerce0, float*
%t.coerce1)
// COMMON-LABEL: define amdgpu_kernel void @_Z7kernel61T(%struct.T
addrspace(4)*{{.*}} byref(%struct.T) align 8 %0)
// OPT: [[R0:%.*]] = getelementptr inbounds %struct.T, %struct.T addrspace(4)*
%0, i64 0, i32 0, i64 0
// OPT: [[P0:%.*]] = load float*, float* addrspace(4)* [[R0]], align 8
+// OPT: [[G0:%.*]] = addrspacecast float* [[P0]] to float addrspace(1)*
// OPT: [[R1:%.*]] = getelementptr inbounds %struct.T, %struct.T addrspace(4)*
%0, i64 0, i32 0, i64 1
// OPT: [[P1:%.*]] = load float*, float* addrspace(4)* [[R1]], align 8
-// OPT: [[V0:%.*]] = load float, float* [[P0]], align 4
+// OPT: [[G1:%.*]] = addrspacecast float* [[P1]] to float addrspace(1)*
+// OPT: [[V0:%.*]] = load float, float addrspace(1)* [[G0]], align 4
// OPT: [[ADD0:%.*]] = fadd contract float [[V0]], 1.00e+00
-// OPT: store float [[ADD0]], float* [[P0]], align 4
-// OPT: [[V1:%.*]] = load float, float* [[P1]], align 4
+// OPT: store float [[ADD0]], float addrspace(1)* [[G0]], align 4
+// OPT: [[V1:%.*]] = load float, float addrspace(1)* [[G1]], align 4
// OPT: [[ADD1:%.*]] = fadd contract float [[V1]], 2.00e+00
-// OPT: store float
[clang] 8920ef0 - [hip] Remove the coercion on aggregate kernel arguments.
Author: Michael Liao
Date: 2020-11-12T21:19:30-05:00
New Revision: 8920ef06a138c46b208fb6471d500261c4b9bacc
URL:
https://github.com/llvm/llvm-project/commit/8920ef06a138c46b208fb6471d500261c4b9bacc
DIFF:
https://github.com/llvm/llvm-project/commit/8920ef06a138c46b208fb6471d500261c4b9bacc.diff
LOG: [hip] Remove the coercion on aggregate kernel arguments.
- If an aggregate argument is indirectly accessed within kernels, direct
passing results in unpromotable `alloca`, which degrade performance
significantly. InferAddrSpace pass is enhanced in
[D91121](https://reviews.llvm.org/D91121) to take the assumption that
generic pointers loaded from the constant memory could be regarded
global ones. The need for the coercion on aggregate arguments is
mitigated.
Differential Revision: https://reviews.llvm.org/D89980
Added:
Modified:
clang/lib/CodeGen/TargetInfo.cpp
clang/test/CodeGenCUDA/amdgpu-kernel-arg-pointer-type.cu
clang/test/CodeGenCUDA/kernel-args.cu
Removed:
diff --git a/clang/lib/CodeGen/TargetInfo.cpp
b/clang/lib/CodeGen/TargetInfo.cpp
index 63502ccf7a38..1e5920322ecd 100644
--- a/clang/lib/CodeGen/TargetInfo.cpp
+++ b/clang/lib/CodeGen/TargetInfo.cpp
@@ -8712,35 +8712,9 @@ class AMDGPUABIInfo final : public DefaultABIInfo {
bool isHomogeneousAggregateSmallEnough(const Type *Base,
uint64_t Members) const override;
- // Coerce HIP pointer arguments from generic pointers to global ones.
+ // Coerce HIP scalar pointer arguments from generic pointers to global ones.
llvm::Type *coerceKernelArgumentType(llvm::Type *Ty, unsigned FromAS,
unsigned ToAS) const {
-// Structure types.
-if (auto STy = dyn_cast(Ty)) {
- SmallVector EltTys;
- bool Changed = false;
- for (auto T : STy->elements()) {
-auto NT = coerceKernelArgumentType(T, FromAS, ToAS);
-EltTys.push_back(NT);
-Changed |= (NT != T);
- }
- // Skip if there is no change in element types.
- if (!Changed)
-return STy;
- if (STy->hasName())
-return llvm::StructType::create(
-EltTys, (STy->getName() + ".coerce").str(), STy->isPacked());
- return llvm::StructType::get(getVMContext(), EltTys, STy->isPacked());
-}
-// Array types.
-if (auto ATy = dyn_cast(Ty)) {
- auto T = ATy->getElementType();
- auto NT = coerceKernelArgumentType(T, FromAS, ToAS);
- // Skip if there is no change in that element type.
- if (NT == T)
-return ATy;
- return llvm::ArrayType::get(NT, ATy->getNumElements());
-}
// Single value types.
if (Ty->isPointerTy() && Ty->getPointerAddressSpace() == FromAS)
return llvm::PointerType::get(
diff --git a/clang/test/CodeGenCUDA/amdgpu-kernel-arg-pointer-type.cu
b/clang/test/CodeGenCUDA/amdgpu-kernel-arg-pointer-type.cu
index 2660a5f14f90..dc4659856026 100644
--- a/clang/test/CodeGenCUDA/amdgpu-kernel-arg-pointer-type.cu
+++ b/clang/test/CodeGenCUDA/amdgpu-kernel-arg-pointer-type.cu
@@ -9,8 +9,6 @@
// Coerced struct from `struct S` without all generic pointers lowered into
// global ones.
-// COMMON: %struct.S.coerce = type { i32 addrspace(1)*, float addrspace(1)* }
-// COMMON: %struct.T.coerce = type { [2 x float addrspace(1)*] }
// On the host-side compilation, generic pointer won't be coerced.
// HOST-NOT: %struct.S.coerce
@@ -61,15 +59,17 @@ struct S {
// `by-val` struct will be coerced into a similar struct with all generic
// pointers lowerd into global ones.
// HOST: define void @_Z22__device_stub__kernel41S(i32* %s.coerce0, float*
%s.coerce1)
-// COMMON-LABEL: define amdgpu_kernel void @_Z7kernel41S(%struct.S.coerce
%s.coerce)
-// OPT: [[P0:%.*]] = extractvalue %struct.S.coerce %s.coerce, 0
-// OPT: [[P1:%.*]] = extractvalue %struct.S.coerce %s.coerce, 1
-// OPT: [[V0:%.*]] = load i32, i32 addrspace(1)* [[P0]], align 4
+// COMMON-LABEL: define amdgpu_kernel void @_Z7kernel41S(%struct.S
addrspace(4)*{{.*}} byref(%struct.S) align 8 %0)
+// OPT: [[R0:%.*]] = getelementptr inbounds %struct.S, %struct.S addrspace(4)*
%0, i64 0, i32 0
+// OPT: [[P0:%.*]] = load i32*, i32* addrspace(4)* [[R0]], align 8
+// OPT: [[R1:%.*]] = getelementptr inbounds %struct.S, %struct.S addrspace(4)*
%0, i64 0, i32 1
+// OPT: [[P1:%.*]] = load float*, float* addrspace(4)* [[R1]], align 8
+// OPT: [[V0:%.*]] = load i32, i32* [[P0]], align 4
// OPT: [[INC:%.*]] = add nsw i32 [[V0]], 1
-// OPT: store i32 [[INC]], i32 addrspace(1)* [[P0]], align 4
-// OPT: [[V1:%.*]] = load float, float addrspace(1)* [[P1]], align 4
+// OPT: store i32 [[INC]], i32* [[P0]], align 4
+// OPT: [[V1:%.*]] = load float, float* [[P1]], align 4
// OPT: [[ADD:%.*]] = fadd contract float [[V1]], 1.00e+00
-// OPT: store float [[ADD]], float addrspace(1)* [[P1]], align 4
+// OPT: store float
[clang] 23c6d15 - [amdgpu] Add `llvm.amdgcn.endpgm` support.
Author: Michael Liao
Date: 2020-11-05T19:06:50-05:00
New Revision: 23c6d1501d80073784cab367d30d50419ffa5706
URL:
https://github.com/llvm/llvm-project/commit/23c6d1501d80073784cab367d30d50419ffa5706
DIFF:
https://github.com/llvm/llvm-project/commit/23c6d1501d80073784cab367d30d50419ffa5706.diff
LOG: [amdgpu] Add `llvm.amdgcn.endpgm` support.
- `llvm.amdgcn.endpgm` is added to enable "abort" support.
Differential Revision: https://reviews.llvm.org/D90809
Added:
llvm/test/CodeGen/AMDGPU/amd.endpgm.ll
Modified:
clang/include/clang/Basic/BuiltinsAMDGPU.def
clang/test/CodeGenCUDA/builtins-amdgcn.cu
llvm/include/llvm/IR/IntrinsicsAMDGPU.td
llvm/lib/Target/AMDGPU/SOPInstructions.td
Removed:
diff --git a/clang/include/clang/Basic/BuiltinsAMDGPU.def
b/clang/include/clang/Basic/BuiltinsAMDGPU.def
index f5901e6f8f3b..123a7ad212da 100644
--- a/clang/include/clang/Basic/BuiltinsAMDGPU.def
+++ b/clang/include/clang/Basic/BuiltinsAMDGPU.def
@@ -214,6 +214,8 @@ BUILTIN(__builtin_amdgcn_read_exec, "LUi", "nc")
BUILTIN(__builtin_amdgcn_read_exec_lo, "Ui", "nc")
BUILTIN(__builtin_amdgcn_read_exec_hi, "Ui", "nc")
+BUILTIN(__builtin_amdgcn_endpgm, "v", "nr")
+
//===--===//
// R600-NI only builtins.
//===--===//
diff --git a/clang/test/CodeGenCUDA/builtins-amdgcn.cu
b/clang/test/CodeGenCUDA/builtins-amdgcn.cu
index 1c3a79064595..8f0d0d0801bd 100644
--- a/clang/test/CodeGenCUDA/builtins-amdgcn.cu
+++ b/clang/test/CodeGenCUDA/builtins-amdgcn.cu
@@ -16,3 +16,9 @@ void test_ds_fmax(float src) {
__shared__ float shared;
volatile float x = __builtin_amdgcn_ds_fmaxf(, src, 0, 0, false);
}
+
+// CHECK-LABEL: @_Z6endpgmv(
+// CHECK: call void @llvm.amdgcn.endpgm()
+__global__ void endpgm() {
+ __builtin_amdgcn_endpgm();
+}
diff --git a/llvm/include/llvm/IR/IntrinsicsAMDGPU.td
b/llvm/include/llvm/IR/IntrinsicsAMDGPU.td
index 304377ce28ab..bc04fa40f2a8 100644
--- a/llvm/include/llvm/IR/IntrinsicsAMDGPU.td
+++ b/llvm/include/llvm/IR/IntrinsicsAMDGPU.td
@@ -1577,6 +1577,10 @@ def int_amdgcn_wqm_vote : Intrinsic<[llvm_i1_ty],
// FIXME: Should this be IntrNoMem, IntrHasSideEffects, or IntrWillReturn?
def int_amdgcn_kill : Intrinsic<[], [llvm_i1_ty], []>;
+def int_amdgcn_endpgm : GCCBuiltin<"__builtin_amdgcn_endpgm">,
+ Intrinsic<[], [], [IntrNoReturn, IntrCold, IntrNoMem, IntrHasSideEffects]
+>;
+
// Copies the active channels of the source value to the destination value,
// with the guarantee that the source value is computed as if the entire
// program were executed in Whole Wavefront Mode, i.e. with all channels
diff --git a/llvm/lib/Target/AMDGPU/SOPInstructions.td
b/llvm/lib/Target/AMDGPU/SOPInstructions.td
index 08966d7d62eb..00527171ff11 100644
--- a/llvm/lib/Target/AMDGPU/SOPInstructions.td
+++ b/llvm/lib/Target/AMDGPU/SOPInstructions.td
@@ -1118,6 +1118,7 @@ let isTerminator = 1 in {
def S_ENDPGM : SOPP_Pseudo<"s_endpgm", (ins EndpgmImm:$simm16), "$simm16"> {
let isBarrier = 1;
let isReturn = 1;
+ let hasSideEffects = 1;
}
def S_ENDPGM_SAVED : SOPP_Pseudo<"s_endpgm_saved", (ins)> {
@@ -1328,6 +1329,11 @@ def : GCNPat <
(S_ENDPGM (i16 0))
>;
+def : GCNPat <
+ (int_amdgcn_endpgm),
+(S_ENDPGM (i16 0))
+>;
+
def : GCNPat <
(i64 (ctpop i64:$src)),
(i64 (REG_SEQUENCE SReg_64,
diff --git a/llvm/test/CodeGen/AMDGPU/amd.endpgm.ll
b/llvm/test/CodeGen/AMDGPU/amd.endpgm.ll
new file mode 100644
index ..ac9cd0699118
--- /dev/null
+++ b/llvm/test/CodeGen/AMDGPU/amd.endpgm.ll
@@ -0,0 +1,50 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: llc -march=amdgcn -mcpu=gfx900 -verify-machineinstrs < %s | FileCheck %s
+
+define amdgpu_kernel void @test0() {
+; CHECK-LABEL: test0:
+; CHECK: ; %bb.0:
+; CHECK-NEXT:s_endpgm
+ tail call void @llvm.amdgcn.endpgm()
+ unreachable
+}
+
+define void @test1() {
+; CHECK-LABEL: test1:
+; CHECK: ; %bb.0:
+; CHECK-NEXT:s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
+; CHECK-NEXT:s_endpgm
+ tail call void @llvm.amdgcn.endpgm()
+ unreachable
+}
+
+define amdgpu_kernel void @test2(i32* %p, i32 %x) {
+; CHECK-LABEL: test2:
+; CHECK: ; %bb.0:
+; CHECK-NEXT:s_load_dword s2, s[0:1], 0x2c
+; CHECK-NEXT:s_waitcnt lgkmcnt(0)
+; CHECK-NEXT:s_cmp_lt_i32 s2, 1
+; CHECK-NEXT:s_cbranch_scc0 BB2_2
+; CHECK-NEXT: ; %bb.1: ; %else
+; CHECK-NEXT:s_load_dwordx2 s[0:1], s[0:1], 0x24
+; CHECK-NEXT:v_mov_b32_e32 v2, s2
+; CHECK-NEXT:s_waitcnt lgkmcnt(0)
+; CHECK-NEXT:v_mov_b32_e32 v0, s0
+; CHECK-NEXT:v_mov_b32_e32 v1, s1
+; CHECK-NEXT:flat_store_dword v[0:1], v2
+; CHECK-NEXT:s_endpgm
+; CHECK-NEXT: BB2_2: ; %then
+; CHECK-NEXT:s_endpgm
+ %cond = icmp sgt i32 %x, 0
+ br i1 %cond, label
[clang] 1bcec29 - Only run when `arm` is registered. NFC.
Author: Michael Liao Date: 2020-10-21T09:30:07-04:00 New Revision: 1bcec29afb321976cdcaa632ee6a47567dd651a7 URL: https://github.com/llvm/llvm-project/commit/1bcec29afb321976cdcaa632ee6a47567dd651a7 DIFF: https://github.com/llvm/llvm-project/commit/1bcec29afb321976cdcaa632ee6a47567dd651a7.diff LOG: Only run when `arm` is registered. NFC. Added: Modified: clang/test/Driver/arm-float-abi.c Removed: diff --git a/clang/test/Driver/arm-float-abi.c b/clang/test/Driver/arm-float-abi.c index 74ba3fd3bc57..294f02444769 100644 --- a/clang/test/Driver/arm-float-abi.c +++ b/clang/test/Driver/arm-float-abi.c @@ -1,3 +1,4 @@ +// REQUIRES: arm-registered-target // RUN: not %clang %s -target armv7-apple-ios -mfloat-abi=hard 2>&1 | FileCheck -check-prefix=ARMV7-ERROR %s // RUN: %clang %s -target armv7-apple-ios -mfloat-abi=softfp -### 2>&1 | FileCheck -check-prefix=NOERROR %s // RUN: %clang %s -arch armv7 -target thumbv7-apple-darwin-eabi -mfloat-abi=hard -### 2>&1 | FileCheck -check-prefix=NOERROR %s ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] e7a6915 - Revert "[clang] Fix warnings on the missing of explicitly copy constructor on the base class. NFC."
Author: Michael Liao
Date: 2020-10-20T10:25:20-04:00
New Revision: e7a69158635a30cb673e443a3b95ece359c72cc1
URL:
https://github.com/llvm/llvm-project/commit/e7a69158635a30cb673e443a3b95ece359c72cc1
DIFF:
https://github.com/llvm/llvm-project/commit/e7a69158635a30cb673e443a3b95ece359c72cc1.diff
LOG: Revert "[clang] Fix warnings on the missing of explicitly copy constructor
on the base class. NFC."
This reverts commit 1ed506deaddb41870d22f5b48d52ba710e8d6c00.
Added:
Modified:
clang/include/clang/Basic/Diagnostic.h
clang/include/clang/Basic/PartialDiagnostic.h
Removed:
diff --git a/clang/include/clang/Basic/Diagnostic.h
b/clang/include/clang/Basic/Diagnostic.h
index 3895e1f458948..f17b98f740385 100644
--- a/clang/include/clang/Basic/Diagnostic.h
+++ b/clang/include/clang/Basic/Diagnostic.h
@@ -1284,7 +1284,7 @@ class DiagnosticBuilder : public StreamingDiagnostic {
public:
/// Copy constructor. When copied, this "takes" the diagnostic info from the
/// input and neuters it.
- DiagnosticBuilder(const DiagnosticBuilder ) : StreamingDiagnostic(D) {
+ DiagnosticBuilder(const DiagnosticBuilder ) {
DiagObj = D.DiagObj;
DiagStorage = D.DiagStorage;
IsActive = D.IsActive;
diff --git a/clang/include/clang/Basic/PartialDiagnostic.h
b/clang/include/clang/Basic/PartialDiagnostic.h
index 9ddf64d2de2c5..9e017902b1205 100644
--- a/clang/include/clang/Basic/PartialDiagnostic.h
+++ b/clang/include/clang/Basic/PartialDiagnostic.h
@@ -49,8 +49,7 @@ class PartialDiagnostic : public StreamingDiagnostic {
PartialDiagnostic(unsigned DiagID, DiagStorageAllocator _)
: StreamingDiagnostic(Allocator_), DiagID(DiagID) {}
- PartialDiagnostic(const PartialDiagnostic )
- : StreamingDiagnostic(Other), DiagID(Other.DiagID) {
+ PartialDiagnostic(const PartialDiagnostic ) : DiagID(Other.DiagID) {
Allocator = Other.Allocator;
if (Other.DiagStorage) {
DiagStorage = getStorage();
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] 1ed506d - [clang] Fix warnings on the missing of explicitly copy constructor on the base class. NFC.
Author: Michael Liao
Date: 2020-10-20T10:06:24-04:00
New Revision: 1ed506deaddb41870d22f5b48d52ba710e8d6c00
URL:
https://github.com/llvm/llvm-project/commit/1ed506deaddb41870d22f5b48d52ba710e8d6c00
DIFF:
https://github.com/llvm/llvm-project/commit/1ed506deaddb41870d22f5b48d52ba710e8d6c00.diff
LOG: [clang] Fix warnings on the missing of explicitly copy constructor on the
base class. NFC.
Added:
Modified:
clang/include/clang/Basic/Diagnostic.h
clang/include/clang/Basic/PartialDiagnostic.h
Removed:
diff --git a/clang/include/clang/Basic/Diagnostic.h
b/clang/include/clang/Basic/Diagnostic.h
index f17b98f74038..3895e1f45894 100644
--- a/clang/include/clang/Basic/Diagnostic.h
+++ b/clang/include/clang/Basic/Diagnostic.h
@@ -1284,7 +1284,7 @@ class DiagnosticBuilder : public StreamingDiagnostic {
public:
/// Copy constructor. When copied, this "takes" the diagnostic info from the
/// input and neuters it.
- DiagnosticBuilder(const DiagnosticBuilder ) {
+ DiagnosticBuilder(const DiagnosticBuilder ) : StreamingDiagnostic(D) {
DiagObj = D.DiagObj;
DiagStorage = D.DiagStorage;
IsActive = D.IsActive;
diff --git a/clang/include/clang/Basic/PartialDiagnostic.h
b/clang/include/clang/Basic/PartialDiagnostic.h
index 9e017902b120..9ddf64d2de2c 100644
--- a/clang/include/clang/Basic/PartialDiagnostic.h
+++ b/clang/include/clang/Basic/PartialDiagnostic.h
@@ -49,7 +49,8 @@ class PartialDiagnostic : public StreamingDiagnostic {
PartialDiagnostic(unsigned DiagID, DiagStorageAllocator _)
: StreamingDiagnostic(Allocator_), DiagID(DiagID) {}
- PartialDiagnostic(const PartialDiagnostic ) : DiagID(Other.DiagID) {
+ PartialDiagnostic(const PartialDiagnostic )
+ : StreamingDiagnostic(Other), DiagID(Other.DiagID) {
Allocator = Other.Allocator;
if (Other.DiagStorage) {
DiagStorage = getStorage();
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] b21ad3b - Fix `-Wparentheses` warnings. NFC.
Author: Michael Liao
Date: 2020-10-14T10:11:19-04:00
New Revision: b21ad3b66bce942ee6e0f5b1fcfdea31928005a7
URL:
https://github.com/llvm/llvm-project/commit/b21ad3b66bce942ee6e0f5b1fcfdea31928005a7
DIFF:
https://github.com/llvm/llvm-project/commit/b21ad3b66bce942ee6e0f5b1fcfdea31928005a7.diff
LOG: Fix `-Wparentheses` warnings. NFC.
Added:
Modified:
clang/lib/Sema/SemaExpr.cpp
Removed:
diff --git a/clang/lib/Sema/SemaExpr.cpp b/clang/lib/Sema/SemaExpr.cpp
index a02db2293bcc..0407d5bb7f6c 100644
--- a/clang/lib/Sema/SemaExpr.cpp
+++ b/clang/lib/Sema/SemaExpr.cpp
@@ -6384,10 +6384,10 @@ ExprResult Sema::BuildCallExpr(Scope *Scope, Expr *Fn,
SourceLocation LParenLoc,
if (Context.isDependenceAllowed() &&
(Fn->isTypeDependent() || Expr::hasAnyTypeDependentArguments(ArgExprs)))
{
assert(!getLangOpts().CPlusPlus);
-assert(Fn->containsErrors() ||
- llvm::any_of(ArgExprs,
-[](clang::Expr *E) { return E->containsErrors(); }) &&
- "should only occur in error-recovery path.");
+assert((Fn->containsErrors() ||
+llvm::any_of(ArgExprs,
+ [](clang::Expr *E) { return E->containsErrors(); }))
&&
+ "should only occur in error-recovery path.");
QualType ReturnType =
llvm::isa_and_nonnull(NDecl)
? dyn_cast(NDecl)->getCallResultType()
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] 8c36eaf - [clang][opencl][codegen] Remove the insertion of `correctly-rounded-divide-sqrt-fp-math` fn-attr.
Author: Michael Liao
Date: 2020-10-01T11:07:39-04:00
New Revision: 8c36eaf0377285acb89c319582d9666e60f42007
URL:
https://github.com/llvm/llvm-project/commit/8c36eaf0377285acb89c319582d9666e60f42007
DIFF:
https://github.com/llvm/llvm-project/commit/8c36eaf0377285acb89c319582d9666e60f42007.diff
LOG: [clang][opencl][codegen] Remove the insertion of
`correctly-rounded-divide-sqrt-fp-math` fn-attr.
- `-cl-fp32-correctly-rounded-divide-sqrt` is already handled in a
per-instruction manner by annotating the accuracy required. There's no
need to add that fn-attr. So far, there's no in-tree backend handling
that attr and that OpenCL specific option.
- In case that out-of-tree backends are broken, this change could be
reverted if those backends could not be fixed.
Differential Revision: https://reviews.llvm.org/D88424
Added:
Modified:
clang/lib/CodeGen/CGCall.cpp
clang/test/CodeGenOpenCL/amdgpu-attrs.cl
clang/test/CodeGenOpenCL/fpmath.cl
Removed:
diff --git a/clang/lib/CodeGen/CGCall.cpp b/clang/lib/CodeGen/CGCall.cpp
index ec7ddf8b5d9e..cb03e025e19e 100644
--- a/clang/lib/CodeGen/CGCall.cpp
+++ b/clang/lib/CodeGen/CGCall.cpp
@@ -1794,11 +1794,6 @@ void
CodeGenModule::getDefaultFunctionAttributes(StringRef Name,
llvm::utostr(CodeGenOpts.SSPBufferSize));
FuncAttrs.addAttribute("no-signed-zeros-fp-math",
llvm::toStringRef(LangOpts.NoSignedZero));
-if (getLangOpts().OpenCL) {
- FuncAttrs.addAttribute(
- "correctly-rounded-divide-sqrt-fp-math",
- llvm::toStringRef(CodeGenOpts.CorrectlyRoundedDivSqrt));
-}
// TODO: Reciprocal estimate codegen options should apply to instructions?
const std::vector = CodeGenOpts.Reciprocals;
diff --git a/clang/test/CodeGenOpenCL/amdgpu-attrs.cl
b/clang/test/CodeGenOpenCL/amdgpu-attrs.cl
index 13f8b1191c2b..9156c45f4939 100644
--- a/clang/test/CodeGenOpenCL/amdgpu-attrs.cl
+++ b/clang/test/CodeGenOpenCL/amdgpu-attrs.cl
@@ -190,5 +190,5 @@ kernel void default_kernel() {
// CHECK-DAG: attributes
[[FLAT_WORK_GROUP_SIZE_32_64_WAVES_PER_EU_2_NUM_SGPR_32_NUM_VGPR_64]] = {{.*}}
"amdgpu-flat-work-group-size"="32,64" "amdgpu-implicitarg-num-bytes"="56"
"amdgpu-num-sgpr"="32" "amdgpu-num-vgpr"="64" "amdgpu-waves-per-eu"="2"
// CHECK-DAG: attributes
[[FLAT_WORK_GROUP_SIZE_32_64_WAVES_PER_EU_2_4_NUM_SGPR_32_NUM_VGPR_64]] =
{{.*}} "amdgpu-flat-work-group-size"="32,64"
"amdgpu-implicitarg-num-bytes"="56" "amdgpu-num-sgpr"="32"
"amdgpu-num-vgpr"="64" "amdgpu-waves-per-eu"="2,4"
-// CHECK-DAG: attributes [[A_FUNCTION]] = {{.*}}
"correctly-rounded-divide-sqrt-fp-math"="false"
+// CHECK-DAG: attributes [[A_FUNCTION]] = {{.*}}
// CHECK-DAG: attributes [[DEFAULT_KERNEL_ATTRS]] = {{.*}}
"amdgpu-flat-work-group-size"="1,256" "amdgpu-implicitarg-num-bytes"="56"
diff --git a/clang/test/CodeGenOpenCL/fpmath.cl
b/clang/test/CodeGenOpenCL/fpmath.cl
index 0108d909c94e..36cb8e68ea7c 100644
--- a/clang/test/CodeGenOpenCL/fpmath.cl
+++ b/clang/test/CodeGenOpenCL/fpmath.cl
@@ -7,7 +7,6 @@ typedef __attribute__(( ext_vector_type(4) )) float float4;
float spscalardiv(float a, float b) {
// CHECK: @spscalardiv
- // CHECK: #[[ATTR:[0-9]+]]
// CHECK: fdiv{{.*}},
// NODIVOPT: !fpmath ![[MD:[0-9]+]]
// DIVOPT-NOT: !fpmath ![[MD:[0-9]+]]
@@ -16,7 +15,6 @@ float spscalardiv(float a, float b) {
float4 spvectordiv(float4 a, float4 b) {
// CHECK: @spvectordiv
- // CHECK: #[[ATTR2:[0-9]+]]
// CHECK: fdiv{{.*}},
// NODIVOPT: !fpmath ![[MD]]
// DIVOPT-NOT: !fpmath ![[MD]]
@@ -38,18 +36,9 @@ void testdbllit(long *val) {
#pragma OPENCL EXTENSION cl_khr_fp64 : enable
double dpscalardiv(double a, double b) {
// CHECK: @dpscalardiv
- // CHECK: #[[ATTR]]
// CHECK-NOT: !fpmath
return a / b;
}
#endif
-// CHECK: attributes #[[ATTR]] = {
-// NODIVOPT-SAME: "correctly-rounded-divide-sqrt-fp-math"="false"
-// DIVOPT-SAME: "correctly-rounded-divide-sqrt-fp-math"="true"
-// CHECK-SAME: }
-// CHECK: attributes #[[ATTR2]] = {
-// NODIVOPT-SAME: "correctly-rounded-divide-sqrt-fp-math"="false"
-// DIVOPT-SAME: "correctly-rounded-divide-sqrt-fp-math"="true"
-// CHECK-SAME: }
// NODIVOPT: ![[MD]] = !{float 2.50e+00}
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] 5dbf80c - [clang][codegen] Annotate `correctly-rounded-divide-sqrt-fp-math` fn-attr for OpenCL only.
Author: Michael Liao
Date: 2020-09-28T11:40:32-04:00
New Revision: 5dbf80cad9556e222c4383960007fc0b27ea9541
URL:
https://github.com/llvm/llvm-project/commit/5dbf80cad9556e222c4383960007fc0b27ea9541
DIFF:
https://github.com/llvm/llvm-project/commit/5dbf80cad9556e222c4383960007fc0b27ea9541.diff
LOG: [clang][codegen] Annotate `correctly-rounded-divide-sqrt-fp-math` fn-attr
for OpenCL only.
- `-cl-fp32-correctly-rounded-divide-sqrt` is an OpenCL-specific option
and `correctly-rounded-divide-sqrt-fp-math` should be added for OpenCL
at most.
Differential revision: https://reviews.llvm.org/D88303
Added:
Modified:
clang/lib/CodeGen/CGCall.cpp
clang/test/CodeGen/complex-builtins.c
clang/test/CodeGen/complex-libcalls.c
clang/test/CodeGen/math-builtins.c
clang/test/CodeGen/math-libcalls.c
Removed:
diff --git a/clang/lib/CodeGen/CGCall.cpp b/clang/lib/CodeGen/CGCall.cpp
index ff41764a4a4d..9ccbe87fab66 100644
--- a/clang/lib/CodeGen/CGCall.cpp
+++ b/clang/lib/CodeGen/CGCall.cpp
@@ -1794,9 +1794,11 @@ void
CodeGenModule::getDefaultFunctionAttributes(StringRef Name,
llvm::utostr(CodeGenOpts.SSPBufferSize));
FuncAttrs.addAttribute("no-signed-zeros-fp-math",
llvm::toStringRef(LangOpts.NoSignedZero));
-FuncAttrs.addAttribute(
-"correctly-rounded-divide-sqrt-fp-math",
-llvm::toStringRef(CodeGenOpts.CorrectlyRoundedDivSqrt));
+if (getLangOpts().OpenCL) {
+ FuncAttrs.addAttribute(
+ "correctly-rounded-divide-sqrt-fp-math",
+ llvm::toStringRef(CodeGenOpts.CorrectlyRoundedDivSqrt));
+}
// TODO: Reciprocal estimate codegen options should apply to instructions?
const std::vector = CodeGenOpts.Reciprocals;
diff --git a/clang/test/CodeGen/complex-builtins.c
b/clang/test/CodeGen/complex-builtins.c
index 7ee2d6d84857..96c0e7117016 100644
--- a/clang/test/CodeGen/complex-builtins.c
+++ b/clang/test/CodeGen/complex-builtins.c
@@ -197,10 +197,8 @@ void foo(float f) {
// HAS_ERRNO: declare { x86_fp80, x86_fp80 } @ctanhl({ x86_fp80, x86_fp80 }*
byval({ x86_fp80, x86_fp80 }) align 16) [[NOT_READNONE]]
};
-
// NO__ERRNO: attributes [[READNONE]] = { {{.*}}readnone{{.*}} }
-// NO__ERRNO: attributes [[NOT_READNONE]] = { nounwind "correctly{{.*}} }
+// NO__ERRNO: attributes [[NOT_READNONE]] = { nounwind {{.*}} }
-// HAS_ERRNO: attributes [[NOT_READNONE]] = { nounwind "correctly{{.*}} }
+// HAS_ERRNO: attributes [[NOT_READNONE]] = { nounwind {{.*}} }
// HAS_ERRNO: attributes [[READNONE]] = { {{.*}}readnone{{.*}} }
-
diff --git a/clang/test/CodeGen/complex-libcalls.c
b/clang/test/CodeGen/complex-libcalls.c
index 248041788293..9bd419a83821 100644
--- a/clang/test/CodeGen/complex-libcalls.c
+++ b/clang/test/CodeGen/complex-libcalls.c
@@ -197,10 +197,8 @@ void foo(float f) {
// HAS_ERRNO: declare { x86_fp80, x86_fp80 } @ctanhl({ x86_fp80, x86_fp80 }*
byval({ x86_fp80, x86_fp80 }) align 16) [[NOT_READNONE]]
};
-
// NO__ERRNO: attributes [[READNONE]] = { {{.*}}readnone{{.*}} }
-// NO__ERRNO: attributes [[NOT_READNONE]] = { nounwind "correctly{{.*}} }
+// NO__ERRNO: attributes [[NOT_READNONE]] = { nounwind {{.*}} }
-// HAS_ERRNO: attributes [[NOT_READNONE]] = { nounwind "correctly{{.*}} }
+// HAS_ERRNO: attributes [[NOT_READNONE]] = { nounwind {{.*}} }
// HAS_ERRNO: attributes [[READNONE]] = { {{.*}}readnone{{.*}} }
-
diff --git a/clang/test/CodeGen/math-builtins.c
b/clang/test/CodeGen/math-builtins.c
index 13e9c13096f2..8aadd050ee89 100644
--- a/clang/test/CodeGen/math-builtins.c
+++ b/clang/test/CodeGen/math-builtins.c
@@ -577,13 +577,12 @@ void foo(double *d, float f, float *fp, long double *l,
int *i, const char *c) {
// HAS_ERRNO: declare x86_fp80 @llvm.trunc.f80(x86_fp80) [[READNONE_INTRINSIC]]
};
-
// NO__ERRNO: attributes [[READNONE]] = { {{.*}}readnone{{.*}} }
// NO__ERRNO: attributes [[READNONE_INTRINSIC]] = { {{.*}}readnone{{.*}} }
-// NO__ERRNO: attributes [[NOT_READNONE]] = { nounwind "correctly{{.*}} }
+// NO__ERRNO: attributes [[NOT_READNONE]] = { nounwind {{.*}} }
// NO__ERRNO: attributes [[PURE]] = { {{.*}}readonly{{.*}} }
-// HAS_ERRNO: attributes [[NOT_READNONE]] = { nounwind "correctly{{.*}} }
+// HAS_ERRNO: attributes [[NOT_READNONE]] = { nounwind {{.*}} }
// HAS_ERRNO: attributes [[READNONE_INTRINSIC]] = { {{.*}}readnone{{.*}} }
// HAS_ERRNO: attributes [[PURE]] = { {{.*}}readonly{{.*}} }
// HAS_ERRNO: attributes [[READNONE]] = { {{.*}}readnone{{.*}} }
diff --git a/clang/test/CodeGen/math-libcalls.c
b/clang/test/CodeGen/math-libcalls.c
index 97a87beb12ec..51bdc5218fde 100644
--- a/clang/test/CodeGen/math-libcalls.c
+++ b/clang/test/CodeGen/math-libcalls.c
@@ -532,13 +532,12 @@ void foo(double *d, float f, float *fp, long double *l,
int *i, const char *c) {
// HAS_ERRNO: declare x86_fp80 @llvm.trunc.f80(x86_fp80)
[clang] 4d4f092 - [clang][codegen] Skip adding default function attributes on intrinsics.
Author: Michael Liao
Date: 2020-09-16T14:10:05-04:00
New Revision: 4d4f0922837de3f1aa9862ae8a8d941b3b6e5f78
URL:
https://github.com/llvm/llvm-project/commit/4d4f0922837de3f1aa9862ae8a8d941b3b6e5f78
DIFF:
https://github.com/llvm/llvm-project/commit/4d4f0922837de3f1aa9862ae8a8d941b3b6e5f78.diff
LOG: [clang][codegen] Skip adding default function attributes on intrinsics.
- After loading builtin bitcode for linking, skip adding default
function attributes on LLVM intrinsics as their attributes are
well-defined and retrieved directly from internal definitions. Adding
extra attributes on intrinsics results in inconsistent result when
`-save-temps` is present. Also, that makes few optimizations
conservative.
Differential Revision: https://reviews.llvm.org/D87761
Added:
clang/test/CodeGenCUDA/Inputs/device-lib-code.ll
clang/test/CodeGenCUDA/dft-func-attr-skip-intrinsic.hip
Modified:
clang/lib/CodeGen/CodeGenAction.cpp
Removed:
diff --git a/clang/lib/CodeGen/CodeGenAction.cpp
b/clang/lib/CodeGen/CodeGenAction.cpp
index 5a6ce0f5dbd5..eda4beff78b7 100644
--- a/clang/lib/CodeGen/CodeGenAction.cpp
+++ b/clang/lib/CodeGen/CodeGenAction.cpp
@@ -245,8 +245,13 @@ namespace clang {
bool LinkInModules() {
for (auto : LinkModules) {
if (LM.PropagateAttrs)
- for (Function : *LM.Module)
+ for (Function : *LM.Module) {
+// Skip intrinsics. Keep consistent with how intrinsics are created
+// in LLVM IR.
+if (F.isIntrinsic())
+ continue;
Gen->CGM().addDefaultFunctionDefinitionAttributes(F);
+ }
CurLinkModule = LM.Module.get();
diff --git a/clang/test/CodeGenCUDA/Inputs/device-lib-code.ll
b/clang/test/CodeGenCUDA/Inputs/device-lib-code.ll
new file mode 100644
index ..43ec911fb02c
--- /dev/null
+++ b/clang/test/CodeGenCUDA/Inputs/device-lib-code.ll
@@ -0,0 +1,5 @@
+define linkonce_odr protected float @__ocml_fma_f32(float %0, float %1, float
%2) local_unnamed_addr {
+ %4 = tail call float @llvm.fma.f32(float %0, float %1, float %2)
+ ret float %4
+}
+declare float @llvm.fma.f32(float, float, float)
diff --git a/clang/test/CodeGenCUDA/dft-func-attr-skip-intrinsic.hip
b/clang/test/CodeGenCUDA/dft-func-attr-skip-intrinsic.hip
new file mode 100644
index ..9e3e436200fc
--- /dev/null
+++ b/clang/test/CodeGenCUDA/dft-func-attr-skip-intrinsic.hip
@@ -0,0 +1,18 @@
+// REQUIRES: amdgpu-registered-target
+// RUN: %clang_cc1 -x ir -fcuda-is-device -triple amdgcn-amd-amdhsa
-emit-llvm-bc -disable-llvm-passes -o %t.bc %S/Inputs/device-lib-code.ll
+// RUN: %clang_cc1 -x hip -fcuda-is-device -triple amdgcn-amd-amdhsa
-mlink-builtin-bitcode %t.bc -emit-llvm -disable-llvm-passes -o - %s |
FileCheck %s
+
+#include "Inputs/cuda.h"
+
+extern "C" __device__ float __ocml_fma_f32(float x, float y, float z);
+
+__device__ float foo(float x) {
+ return __ocml_fma_f32(x, x, x);
+}
+
+// CHECK: {{^}}define{{.*}} @__ocml_fma_f32{{.*}} [[ATTR1:#[0-9]+]]
+// CHECK: {{^}}declare{{.*}} @llvm.fma.f32{{.*}} [[ATTR2:#[0-9]+]]
+// CHECK: attributes [[ATTR1]] = { convergent
+// CHECK: attributes [[ATTR2]] = {
+// CHECK-NOT: convergent
+// CHECK: }
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
Re: r364428 - Make CodeGen depend on ASTMatchers
b22d4504968 was committed last night. On Fri, Sep 11, 2020 at 9:30 AM Vassil Vassilev wrote: > > On 9/11/20 5:13 AM, Michael LIAO wrote: > > That change was added long ago to fix the shared library build. > > Possibly, there are changes removing that dependency then. Just > > verified that removing that dependency is just fine. > > >That's great! Would you commit that change or should I? > > > > > > On Thu, Sep 10, 2020 at 6:48 AM Vassil Vassilev > > wrote: > >> Hello, > >> > >> IIUC, clang's CodeGen does not immediately depend on ASTMatchers. I > >> was wondering what is the reason for inserting such a dependency to fix > >> the shared library builds? > >> > >> Can you give more details about the failure you are fixing? > >> > >> Sorry for the late question. > >> > >> Best, Vassil > >> On 6/26/19 5:13 PM, Michael Liao via cfe-commits wrote: > >>> Author: hliao > >>> Date: Wed Jun 26 07:13:43 2019 > >>> New Revision: 364428 > >>> > >>> URL: http://llvm.org/viewvc/llvm-project?rev=364428=rev > >>> Log: > >>> Make CodeGen depend on ASTMatchers > >>> > >>> - Shared library builds are broken due to the missing dependency. > >>> > >>> Modified: > >>> cfe/trunk/lib/CodeGen/CMakeLists.txt > >>> > >>> Modified: cfe/trunk/lib/CodeGen/CMakeLists.txt > >>> URL: > >>> http://llvm.org/viewvc/llvm-project/cfe/trunk/lib/CodeGen/CMakeLists.txt?rev=364428=364427=364428=diff > >>> == > >>> --- cfe/trunk/lib/CodeGen/CMakeLists.txt (original) > >>> +++ cfe/trunk/lib/CodeGen/CMakeLists.txt Wed Jun 26 07:13:43 2019 > >>> @@ -101,6 +101,7 @@ add_clang_library(clangCodeGen > >>> LINK_LIBS > >>> clangAnalysis > >>> clangAST > >>> + clangASTMatchers > >>> clangBasic > >>> clangFrontend > >>> clangLex > >>> > >>> > >>> ___ > >>> cfe-commits mailing list > >>> cfe-commits@lists.llvm.org > >>> https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits > >> > ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] b22d450 - Remove dependency on clangASTMatchers.
Author: Michael Liao Date: 2020-09-10T22:17:48-04:00 New Revision: b22d45049682d1461b6b786f159681e2e5c2ce24 URL: https://github.com/llvm/llvm-project/commit/b22d45049682d1461b6b786f159681e2e5c2ce24 DIFF: https://github.com/llvm/llvm-project/commit/b22d45049682d1461b6b786f159681e2e5c2ce24.diff LOG: Remove dependency on clangASTMatchers. - It seems no long required for shared library builds. Added: Modified: clang/lib/CodeGen/CMakeLists.txt Removed: diff --git a/clang/lib/CodeGen/CMakeLists.txt b/clang/lib/CodeGen/CMakeLists.txt index f47ecd9bf846..4039277707c5 100644 --- a/clang/lib/CodeGen/CMakeLists.txt +++ b/clang/lib/CodeGen/CMakeLists.txt @@ -92,7 +92,6 @@ add_clang_library(clangCodeGen LINK_LIBS clangAnalysis clangAST - clangASTMatchers clangBasic clangFrontend clangLex ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
Re: r364428 - Make CodeGen depend on ASTMatchers
That change was added long ago to fix the shared library build. Possibly, there are changes removing that dependency then. Just verified that removing that dependency is just fine. On Thu, Sep 10, 2020 at 6:48 AM Vassil Vassilev wrote: > > Hello, > >IIUC, clang's CodeGen does not immediately depend on ASTMatchers. I > was wondering what is the reason for inserting such a dependency to fix > the shared library builds? > >Can you give more details about the failure you are fixing? > >Sorry for the late question. > > Best, Vassil > On 6/26/19 5:13 PM, Michael Liao via cfe-commits wrote: > > Author: hliao > > Date: Wed Jun 26 07:13:43 2019 > > New Revision: 364428 > > > > URL: http://llvm.org/viewvc/llvm-project?rev=364428=rev > > Log: > > Make CodeGen depend on ASTMatchers > > > > - Shared library builds are broken due to the missing dependency. > > > > Modified: > > cfe/trunk/lib/CodeGen/CMakeLists.txt > > > > Modified: cfe/trunk/lib/CodeGen/CMakeLists.txt > > URL: > > http://llvm.org/viewvc/llvm-project/cfe/trunk/lib/CodeGen/CMakeLists.txt?rev=364428=364427=364428=diff > > == > > --- cfe/trunk/lib/CodeGen/CMakeLists.txt (original) > > +++ cfe/trunk/lib/CodeGen/CMakeLists.txt Wed Jun 26 07:13:43 2019 > > @@ -101,6 +101,7 @@ add_clang_library(clangCodeGen > > LINK_LIBS > > clangAnalysis > > clangAST > > + clangASTMatchers > > clangBasic > > clangFrontend > > clangLex > > > > > > ___ > > cfe-commits mailing list > > cfe-commits@lists.llvm.org > > https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits > > ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] c7b683c - [PGO][CUDA][HIP] Skip generating profile on the device stub and wrong-side functions.
Author: Michael Liao
Date: 2020-08-10T11:01:46-04:00
New Revision: c7b683c126b849dab5c81e7deecfc1e61f8563a0
URL:
https://github.com/llvm/llvm-project/commit/c7b683c126b849dab5c81e7deecfc1e61f8563a0
DIFF:
https://github.com/llvm/llvm-project/commit/c7b683c126b849dab5c81e7deecfc1e61f8563a0.diff
LOG: [PGO][CUDA][HIP] Skip generating profile on the device stub and wrong-side
functions.
- Skip generating profile data on `__global__` function in the host
compilation. It's a host-side stub function only and don't have
profile instrumentation generated on the real function body. The extra
profile data results in the malformed instrumentation profile data.
- Skip generating region mapping on functions in the wrong-side, i.e.,
+ For the device compilation, skip host-only functions; and,
+ For the host compilation, skip device-only functions (including
`__global__` functions.)
- As the device-side profiling is not ready yet, only host-side profile
code generation is checked.
Differential Revision: https://reviews.llvm.org/D85276
Added:
clang/test/CodeGenCUDA/profile-coverage-mapping.cu
Modified:
clang/lib/CodeGen/CodeGenPGO.cpp
Removed:
diff --git a/clang/lib/CodeGen/CodeGenPGO.cpp
b/clang/lib/CodeGen/CodeGenPGO.cpp
index e810f608ab78..be3c50b99f30 100644
--- a/clang/lib/CodeGen/CodeGenPGO.cpp
+++ b/clang/lib/CodeGen/CodeGenPGO.cpp
@@ -773,6 +773,11 @@ void CodeGenPGO::assignRegionCounters(GlobalDecl GD,
llvm::Function *Fn) {
if (!D->hasBody())
return;
+ // Skip CUDA/HIP kernel launch stub functions.
+ if (CGM.getLangOpts().CUDA && !CGM.getLangOpts().CUDAIsDevice &&
+ D->hasAttr())
+return;
+
bool InstrumentRegions = CGM.getCodeGenOpts().hasProfileClangInstr();
llvm::IndexedInstrProfReader *PGOReader = CGM.getPGOReader();
if (!InstrumentRegions && !PGOReader)
@@ -831,6 +836,18 @@ bool CodeGenPGO::skipRegionMappingForDecl(const Decl *D) {
if (!D->getBody())
return true;
+ // Skip host-only functions in the CUDA device compilation and device-only
+ // functions in the host compilation. Just roughly filter them out based on
+ // the function attributes. If there are effectively host-only or device-only
+ // ones, their coverage mapping may still be generated.
+ if (CGM.getLangOpts().CUDA &&
+ ((CGM.getLangOpts().CUDAIsDevice && !D->hasAttr() &&
+!D->hasAttr()) ||
+ (!CGM.getLangOpts().CUDAIsDevice &&
+(D->hasAttr() ||
+ (!D->hasAttr() && D->hasAttr())
+return true;
+
// Don't map the functions in system headers.
const auto = CGM.getContext().getSourceManager();
auto Loc = D->getBody()->getBeginLoc();
diff --git a/clang/test/CodeGenCUDA/profile-coverage-mapping.cu
b/clang/test/CodeGenCUDA/profile-coverage-mapping.cu
new file mode 100644
index ..5eae6f10e0ea
--- /dev/null
+++ b/clang/test/CodeGenCUDA/profile-coverage-mapping.cu
@@ -0,0 +1,20 @@
+// RUN: echo "GPU binary would be here" > %t
+// RUN: %clang_cc1 -fprofile-instrument=clang -triple x86_64-linux-gnu
-target-sdk-version=8.0 -fcuda-include-gpubinary %t -emit-llvm -o - %s |
FileCheck --check-prefix=PGOGEN %s
+// RUN: %clang_cc1 -fprofile-instrument=clang -fcoverage-mapping -triple
x86_64-linux-gnu -target-sdk-version=8.0 -fcuda-include-gpubinary %t -emit-llvm
-o - %s | FileCheck --check-prefix=COVMAP %s
+// RUN: %clang_cc1 -fprofile-instrument=clang -fcoverage-mapping
-dump-coverage-mapping -triple x86_64-linux-gnu -target-sdk-version=8.0
-fcuda-include-gpubinary %t -emit-llvm-only -o - %s | FileCheck
--check-prefix=MAPPING %s
+
+#include "Inputs/cuda.h"
+
+// PGOGEN-NOT: @__profn_{{.*kernel.*}} =
+// COVMAP-COUNT-2: section "__llvm_covfun", comdat
+// COVMAP-NOT: section "__llvm_covfun", comdat
+// MAPPING-NOT: {{.*dfn.*}}:
+// MAPPING-NOT: {{.*kernel.*}}:
+
+__device__ void dfn(int i) {}
+
+__global__ void kernel(int i) { dfn(i); }
+
+void host(void) {
+ kernel<<<1, 1>>>(1);
+}
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] b8409c0 - Fix `-Wreturn-type` warning. NFC.
Author: Michael Liao
Date: 2020-07-11T16:20:41-04:00
New Revision: b8409c03ed90807f3d49c7d98dceea98cf461f7a
URL:
https://github.com/llvm/llvm-project/commit/b8409c03ed90807f3d49c7d98dceea98cf461f7a
DIFF:
https://github.com/llvm/llvm-project/commit/b8409c03ed90807f3d49c7d98dceea98cf461f7a.diff
LOG: Fix `-Wreturn-type` warning. NFC.
Added:
Modified:
clang/lib/Tooling/Syntax/BuildTree.cpp
Removed:
diff --git a/clang/lib/Tooling/Syntax/BuildTree.cpp
b/clang/lib/Tooling/Syntax/BuildTree.cpp
index 6d13f1ace83b..1f192180ec45 100644
--- a/clang/lib/Tooling/Syntax/BuildTree.cpp
+++ b/clang/lib/Tooling/Syntax/BuildTree.cpp
@@ -750,6 +750,7 @@ class BuildTreeVisitor : public
RecursiveASTVisitor {
return new (allocator()) syntax::FloatUserDefinedLiteralExpression;
}
}
+llvm_unreachable("Unknown literal operator kind.");
}
bool WalkUpFromUserDefinedLiteral(UserDefinedLiteral *S) {
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] 471c806 - [hip] Refine `clang/test/CodeGenCUDA/amdgpu-kernel-arg-pointer-type.cu`
Author: Michael Liao Date: 2020-06-25T23:57:08-04:00 New Revision: 471c806a45bbac2f0f4274d8bea383d06d397a84 URL: https://github.com/llvm/llvm-project/commit/471c806a45bbac2f0f4274d8bea383d06d397a84 DIFF: https://github.com/llvm/llvm-project/commit/471c806a45bbac2f0f4274d8bea383d06d397a84.diff LOG: [hip] Refine `clang/test/CodeGenCUDA/amdgpu-kernel-arg-pointer-type.cu` - Require target x86 being enabled as well. Added: Modified: clang/test/CodeGenCUDA/amdgpu-kernel-arg-pointer-type.cu Removed: diff --git a/clang/test/CodeGenCUDA/amdgpu-kernel-arg-pointer-type.cu b/clang/test/CodeGenCUDA/amdgpu-kernel-arg-pointer-type.cu index 99284c04e5cc..2660a5f14f90 100644 --- a/clang/test/CodeGenCUDA/amdgpu-kernel-arg-pointer-type.cu +++ b/clang/test/CodeGenCUDA/amdgpu-kernel-arg-pointer-type.cu @@ -1,4 +1,6 @@ +// REQUIRES: x86-registered-target // REQUIRES: amdgpu-registered-target + // RUN: %clang_cc1 -triple amdgcn-amd-amdhsa -fcuda-is-device -emit-llvm -x hip %s -o - | FileCheck --check-prefixes=COMMON,CHECK %s // RUN: %clang_cc1 -triple amdgcn-amd-amdhsa -fcuda-is-device -emit-llvm -x hip %s -disable-O0-optnone -o - | opt -S -O2 | FileCheck %s --check-prefixes=COMMON,OPT // RUN: %clang_cc1 -triple x86_64-unknown-linux-gnu -emit-llvm -x hip %s -o - | FileCheck -check-prefix=HOST %s ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] 0723b18 - [hip] Re-enable `clang/test/CodeGenCUDA/amdgpu-kernel-arg-pointer-type.cu`
Author: Michael Liao Date: 2020-06-25T22:29:27-04:00 New Revision: 0723b1891fac8f79f92549e3bcac9112be4ebd43 URL: https://github.com/llvm/llvm-project/commit/0723b1891fac8f79f92549e3bcac9112be4ebd43 DIFF: https://github.com/llvm/llvm-project/commit/0723b1891fac8f79f92549e3bcac9112be4ebd43.diff LOG: [hip] Re-enable `clang/test/CodeGenCUDA/amdgpu-kernel-arg-pointer-type.cu` - Require amdgpu target being enabled. Added: Modified: clang/test/CodeGenCUDA/amdgpu-kernel-arg-pointer-type.cu Removed: diff --git a/clang/test/CodeGenCUDA/amdgpu-kernel-arg-pointer-type.cu b/clang/test/CodeGenCUDA/amdgpu-kernel-arg-pointer-type.cu index 3021e73780f4..99284c04e5cc 100644 --- a/clang/test/CodeGenCUDA/amdgpu-kernel-arg-pointer-type.cu +++ b/clang/test/CodeGenCUDA/amdgpu-kernel-arg-pointer-type.cu @@ -1,4 +1,4 @@ -// XFAIL: * +// REQUIRES: amdgpu-registered-target // RUN: %clang_cc1 -triple amdgcn-amd-amdhsa -fcuda-is-device -emit-llvm -x hip %s -o - | FileCheck --check-prefixes=COMMON,CHECK %s // RUN: %clang_cc1 -triple amdgcn-amd-amdhsa -fcuda-is-device -emit-llvm -x hip %s -disable-O0-optnone -o - | opt -S -O2 | FileCheck %s --check-prefixes=COMMON,OPT // RUN: %clang_cc1 -triple x86_64-unknown-linux-gnu -emit-llvm -x hip %s -o - | FileCheck -check-prefix=HOST %s ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] d3f437d - [hip] Disable test temporarily due to failures on build servers.
Author: Michael Liao Date: 2020-06-25T22:04:20-04:00 New Revision: d3f437d35189f7567294daf3e60e08326e64994a URL: https://github.com/llvm/llvm-project/commit/d3f437d35189f7567294daf3e60e08326e64994a DIFF: https://github.com/llvm/llvm-project/commit/d3f437d35189f7567294daf3e60e08326e64994a.diff LOG: [hip] Disable test temporarily due to failures on build servers. Added: Modified: clang/test/CodeGenCUDA/amdgpu-kernel-arg-pointer-type.cu Removed: diff --git a/clang/test/CodeGenCUDA/amdgpu-kernel-arg-pointer-type.cu b/clang/test/CodeGenCUDA/amdgpu-kernel-arg-pointer-type.cu index 8c102d339863..3021e73780f4 100644 --- a/clang/test/CodeGenCUDA/amdgpu-kernel-arg-pointer-type.cu +++ b/clang/test/CodeGenCUDA/amdgpu-kernel-arg-pointer-type.cu @@ -1,3 +1,4 @@ +// XFAIL: * // RUN: %clang_cc1 -triple amdgcn-amd-amdhsa -fcuda-is-device -emit-llvm -x hip %s -o - | FileCheck --check-prefixes=COMMON,CHECK %s // RUN: %clang_cc1 -triple amdgcn-amd-amdhsa -fcuda-is-device -emit-llvm -x hip %s -disable-O0-optnone -o - | opt -S -O2 | FileCheck %s --check-prefixes=COMMON,OPT // RUN: %clang_cc1 -triple x86_64-unknown-linux-gnu -emit-llvm -x hip %s -o - | FileCheck -check-prefix=HOST %s ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] dccfaac - [InferAddressSpaces] Handle the pair of `ptrtoint`/`inttoptr`.
Author: Michael Liao
Date: 2020-06-25T20:46:56-04:00
New Revision: dccfaacf93e1c4801cbcc4686f64eb8a35564ff7
URL:
https://github.com/llvm/llvm-project/commit/dccfaacf93e1c4801cbcc4686f64eb8a35564ff7
DIFF:
https://github.com/llvm/llvm-project/commit/dccfaacf93e1c4801cbcc4686f64eb8a35564ff7.diff
LOG: [InferAddressSpaces] Handle the pair of `ptrtoint`/`inttoptr`.
Summary:
- `ptrtoint` and `inttoptr` are defined as no-op casts if the integer
value as the same size as the pointer value. The pair of
`ptrtoint`/`inttoptr` is in fact a no-op cast sequence between
different address spaces. Teach `infer-address-spaces` to handle them
like a `bitcast`.
Reviewers: arsenm, chandlerc
Subscribers: jvesely, wdng, nhaehnle, hiraditya, kerbowa, cfe-commits,
llvm-commits
Tags: #clang, #llvm
Differential Revision: https://reviews.llvm.org/D81938
Added:
llvm/test/Transforms/InferAddressSpaces/AMDGPU/noop-ptrint-pair.ll
Modified:
clang/test/CodeGenCUDA/amdgpu-kernel-arg-pointer-type.cu
llvm/lib/Transforms/Scalar/InferAddressSpaces.cpp
Removed:
diff --git a/clang/test/CodeGenCUDA/amdgpu-kernel-arg-pointer-type.cu
b/clang/test/CodeGenCUDA/amdgpu-kernel-arg-pointer-type.cu
index 73ab9edf318e..8c102d339863 100644
--- a/clang/test/CodeGenCUDA/amdgpu-kernel-arg-pointer-type.cu
+++ b/clang/test/CodeGenCUDA/amdgpu-kernel-arg-pointer-type.cu
@@ -1,37 +1,52 @@
-// RUN: %clang_cc1 -triple amdgcn-amd-amdhsa -fcuda-is-device -emit-llvm -x
hip %s -o - | FileCheck %s
+// RUN: %clang_cc1 -triple amdgcn-amd-amdhsa -fcuda-is-device -emit-llvm -x
hip %s -o - | FileCheck --check-prefixes=COMMON,CHECK %s
+// RUN: %clang_cc1 -triple amdgcn-amd-amdhsa -fcuda-is-device -emit-llvm -x
hip %s -disable-O0-optnone -o - | opt -S -O2 | FileCheck %s
--check-prefixes=COMMON,OPT
// RUN: %clang_cc1 -triple x86_64-unknown-linux-gnu -emit-llvm -x hip %s -o -
| FileCheck -check-prefix=HOST %s
#include "Inputs/cuda.h"
// Coerced struct from `struct S` without all generic pointers lowered into
// global ones.
-// CHECK: %struct.S.coerce = type { i32 addrspace(1)*, float addrspace(1)* }
-// CHECK: %struct.T.coerce = type { [2 x float addrspace(1)*] }
+// COMMON: %struct.S.coerce = type { i32 addrspace(1)*, float addrspace(1)* }
+// COMMON: %struct.T.coerce = type { [2 x float addrspace(1)*] }
// On the host-side compilation, generic pointer won't be coerced.
// HOST-NOT: %struct.S.coerce
// HOST-NOT: %struct.T.coerce
-// CHECK: define amdgpu_kernel void @_Z7kernel1Pi(i32 addrspace(1)* %x.coerce)
// HOST: define void @_Z22__device_stub__kernel1Pi(i32* %x)
+// COMMON-LABEL: define amdgpu_kernel void @_Z7kernel1Pi(i32
addrspace(1)*{{.*}} %x.coerce)
+// CHECK: = addrspacecast [[TYPE:.*]] addrspace(1)* %{{.*}} to [[TYPE]]*
+// CHECK-NOT: = addrspacecast [[TYPE:.*]] addrspace(1)* %{{.*}} to [[TYPE]]*
+// OPT: [[VAL:%.*]] = load i32, i32 addrspace(1)* %x.coerce, align 4
+// OPT: [[INC:%.*]] = add nsw i32 [[VAL]], 1
+// OPT: store i32 [[INC]], i32 addrspace(1)* %x.coerce, align 4
+// OPT: ret void
__global__ void kernel1(int *x) {
x[0]++;
}
-// CHECK: define amdgpu_kernel void @_Z7kernel2Ri(i32 addrspace(1)* nonnull
align 4 dereferenceable(4) %x.coerce)
// HOST: define void @_Z22__device_stub__kernel2Ri(i32* nonnull align 4
dereferenceable(4) %x)
+// COMMON-LABEL: define amdgpu_kernel void @_Z7kernel2Ri(i32
addrspace(1)*{{.*}} nonnull align 4 dereferenceable(4) %x.coerce)
+// CHECK: = addrspacecast [[TYPE:.*]] addrspace(1)* %{{.*}} to [[TYPE]]*
+// CHECK-NOT: = addrspacecast [[TYPE:.*]] addrspace(1)* %{{.*}} to [[TYPE]]*
+// OPT: [[VAL:%.*]] = load i32, i32 addrspace(1)* %x.coerce, align 4
+// OPT: [[INC:%.*]] = add nsw i32 [[VAL]], 1
+// OPT: store i32 [[INC]], i32 addrspace(1)* %x.coerce, align 4
+// OPT: ret void
__global__ void kernel2(int ) {
x++;
}
-// CHECK: define amdgpu_kernel void @_Z7kernel3PU3AS2iPU3AS1i(i32
addrspace(2)* %x, i32 addrspace(1)* %y)
// HOST: define void @_Z22__device_stub__kernel3PU3AS2iPU3AS1i(i32
addrspace(2)* %x, i32 addrspace(1)* %y)
+// CHECK-LABEL: define amdgpu_kernel void @_Z7kernel3PU3AS2iPU3AS1i(i32
addrspace(2)*{{.*}} %x, i32 addrspace(1)*{{.*}} %y)
+// CHECK-NOT: = addrspacecast [[TYPE:.*]] addrspace(1)* %{{.*}} to [[TYPE]]*
__global__ void kernel3(__attribute__((address_space(2))) int *x,
__attribute__((address_space(1))) int *y) {
y[0] = x[0];
}
-// CHECK: define void @_Z4funcPi(i32* %x)
+// COMMON-LABEL: define void @_Z4funcPi(i32*{{.*}} %x)
+// CHECK-NOT: = addrspacecast [[TYPE:.*]] addrspace(1)* %{{.*}} to [[TYPE]]*
__device__ void func(int *x) {
x[0]++;
}
@@ -42,16 +57,25 @@ struct S {
};
// `by-val` struct will be coerced into a similar struct with all generic
// pointers lowerd into global ones.
-// CHECK: define amdgpu_kernel void @_Z7kernel41S(%struct.S.coerce %s.coerce)
// HOST: define void
[clang] ebc9e0f - Fix coding style. NFC.
Author: Michael Liao
Date: 2020-06-24T13:13:42-04:00
New Revision: ebc9e0f1f0786b892b4a6eaf50013a18aed31aa5
URL:
https://github.com/llvm/llvm-project/commit/ebc9e0f1f0786b892b4a6eaf50013a18aed31aa5
DIFF:
https://github.com/llvm/llvm-project/commit/ebc9e0f1f0786b892b4a6eaf50013a18aed31aa5.diff
LOG: Fix coding style. NFC.
- Remove `else` after `return`.
Added:
Modified:
clang/lib/CodeGen/CodeGenModule.cpp
Removed:
diff --git a/clang/lib/CodeGen/CodeGenModule.cpp
b/clang/lib/CodeGen/CodeGenModule.cpp
index f0ab5165584c..5e902aa15bcf 100644
--- a/clang/lib/CodeGen/CodeGenModule.cpp
+++ b/clang/lib/CodeGen/CodeGenModule.cpp
@@ -3656,26 +3656,29 @@ CodeGenModule::GetOrCreateLLVMGlobal(StringRef
MangledName,
}
llvm::Constant *
-CodeGenModule::GetAddrOfGlobal(GlobalDecl GD,
- ForDefinition_t IsForDefinition) {
+CodeGenModule::GetAddrOfGlobal(GlobalDecl GD, ForDefinition_t IsForDefinition)
{
const Decl *D = GD.getDecl();
+
if (isa(D) || isa(D))
return getAddrOfCXXStructor(GD, /*FnInfo=*/nullptr, /*FnType=*/nullptr,
/*DontDefer=*/false, IsForDefinition);
- else if (isa(D)) {
-auto FInfo = ().arrangeCXXMethodDeclaration(
-cast(D));
+
+ if (isa(D)) {
+auto FInfo =
+().arrangeCXXMethodDeclaration(cast(D));
auto Ty = getTypes().GetFunctionType(*FInfo);
return GetAddrOfFunction(GD, Ty, /*ForVTable=*/false, /*DontDefer=*/false,
IsForDefinition);
- } else if (isa(D)) {
+ }
+
+ if (isa(D)) {
const CGFunctionInfo = getTypes().arrangeGlobalDeclaration(GD);
llvm::FunctionType *Ty = getTypes().GetFunctionType(FI);
return GetAddrOfFunction(GD, Ty, /*ForVTable=*/false, /*DontDefer=*/false,
IsForDefinition);
- } else
-return GetAddrOfGlobalVar(cast(D), /*Ty=*/nullptr,
- IsForDefinition);
+ }
+
+ return GetAddrOfGlobalVar(cast(D), /*Ty=*/nullptr, IsForDefinition);
}
llvm::GlobalVariable *CodeGenModule::CreateOrReplaceCXXRuntimeVariable(
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] e830fa2 - [clang][amdgpu] Prefer not using `fp16` conversion intrinsics.
Author: Michael Liao
Date: 2020-06-16T10:21:56-04:00
New Revision: e830fa260da9d3bbe99c4176b4ddb6aa5e6229dd
URL:
https://github.com/llvm/llvm-project/commit/e830fa260da9d3bbe99c4176b4ddb6aa5e6229dd
DIFF:
https://github.com/llvm/llvm-project/commit/e830fa260da9d3bbe99c4176b4ddb6aa5e6229dd.diff
LOG: [clang][amdgpu] Prefer not using `fp16` conversion intrinsics.
Reviewers: yaxunl, arsenm
Subscribers: kzhuravl, jvesely, wdng, nhaehnle, dstuttard, tpr, t-tye, kerbowa,
cfe-commits
Tags: #clang
Differential Revision: https://reviews.llvm.org/D81849
Added:
clang/test/CodeGenHIP/half.hip
Modified:
clang/lib/Basic/Targets/AMDGPU.h
Removed:
diff --git a/clang/lib/Basic/Targets/AMDGPU.h
b/clang/lib/Basic/Targets/AMDGPU.h
index e4194a881e3f..387b91abb537 100644
--- a/clang/lib/Basic/Targets/AMDGPU.h
+++ b/clang/lib/Basic/Targets/AMDGPU.h
@@ -219,6 +219,8 @@ class LLVM_LIBRARY_VISIBILITY AMDGPUTargetInfo final :
public TargetInfo {
ArrayRef getTargetBuiltins() const override;
+ bool useFP16ConversionIntrinsics() const override { return false; }
+
void getTargetDefines(const LangOptions ,
MacroBuilder ) const override;
diff --git a/clang/test/CodeGenHIP/half.hip b/clang/test/CodeGenHIP/half.hip
new file mode 100644
index ..d5dd43d51fdf
--- /dev/null
+++ b/clang/test/CodeGenHIP/half.hip
@@ -0,0 +1,16 @@
+// REQUIRES: amdgpu-registered-target
+// RUN: %clang_cc1 -triple amdgcn-amd-amdhsa -x hip -emit-llvm
-fcuda-is-device -o - %s | FileCheck %s
+
+#define __device__ __attribute__((device))
+
+// CHECK-LABEL: @_Z2d0DF16_
+// CHECK: fpext
+__device__ float d0(_Float16 x) {
+ return x;
+}
+
+// CHECK-LABEL: @_Z2d1f
+// CHECK: fptrunc
+__device__ _Float16 d1(float x) {
+ return x;
+}
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] 6dd0580 - [hip] Fix the failed test case due to the additional backend phase.
Author: Michael Liao
Date: 2020-06-10T15:06:06-04:00
New Revision: 6dd058083208d58c6a7005bfb092cee085fd9a48
URL:
https://github.com/llvm/llvm-project/commit/6dd058083208d58c6a7005bfb092cee085fd9a48
DIFF:
https://github.com/llvm/llvm-project/commit/6dd058083208d58c6a7005bfb092cee085fd9a48.diff
LOG: [hip] Fix the failed test case due to the additional backend phase.
Added:
Modified:
clang/test/Driver/cuda-phases.cu
Removed:
diff --git a/clang/test/Driver/cuda-phases.cu
b/clang/test/Driver/cuda-phases.cu
index 58be50ae2e12..acbf345f85c6 100644
--- a/clang/test/Driver/cuda-phases.cu
+++ b/clang/test/Driver/cuda-phases.cu
@@ -49,9 +49,10 @@
// BIN_AMD_NRDC-DAG: [[P12:[0-9]+]]: backend, {[[P11]]}, assembler,
(host-[[T]])
// BIN-DAG: [[P13:[0-9]+]]: assembler, {[[P12]]}, object, (host-[[T]])
// BIN-DAG: [[P14:[0-9]+]]: linker, {[[P13]]}, image, (host-[[T]])
-// BIN_AMD_RDC-DAG: [[P15:[0-9]+]]: linker, {[[P5]]}, image, (device-[[T]],
[[ARCH]])
-// BIN_AMD_RDC-DAG: [[P16:[0-9]+]]: offload, "host-[[T]]
(powerpc64le-ibm-linux-gnu)" {[[P14]]},
-// BIN_AMD_RDC-DAG-SAME: "device-[[T]]
([[TRIPLE:amdgcn-amd-amdhsa]]:[[ARCH]])" {[[P15]]}, object
+// BIN_AMD_RDC-DAG: [[P15:[0-9]+]]: backend, {[[P5]]}, ir, (device-[[T]],
[[ARCH]])
+// BIN_AMD_RDC-DAG: [[P16:[0-9]+]]: linker, {[[P15]]}, image, (device-[[T]],
[[ARCH]])
+// BIN_AMD_RDC-DAG: [[P17:[0-9]+]]: offload, "host-[[T]]
(powerpc64le-ibm-linux-gnu)" {[[P14]]},
+// BIN_AMD_RDC-DAG-SAME: "device-[[T]]
([[TRIPLE:amdgcn-amd-amdhsa]]:[[ARCH]])" {[[P16]]}, object
//
// Test single gpu architecture up to the assemble phase.
@@ -109,11 +110,13 @@
// BIN2_AMD-DAG: [[P19:[0-9]+]]: backend, {[[P2]]}, assembler, (host-[[T]])
// BIN2-DAG: [[P20:[0-9]+]]: assembler, {[[P19]]}, object, (host-[[T]])
// BIN2-DAG: [[P21:[0-9]+]]: linker, {[[P20]]}, image, (host-[[T]])
-// BIN2_AMD-DAG: [[P22:[0-9]+]]: linker, {[[P5]]}, image, (device-[[T]],
[[ARCH1]])
-// BIN2_AMD-DAG: [[P23:[0-9]+]]: linker, {[[P12]]}, image, (device-[[T]],
[[ARCH2]])
-// BIN2_AMD-DAG: [[P24:[0-9]+]]: offload, "host-[[T]]
(powerpc64le-ibm-linux-gnu)" {[[P21]]},
-// BIN2_AMD-DAG-SAME: "device-[[T]] ([[TRIPLE:amdgcn-amd-amdhsa]]:[[ARCH1]])"
{[[P22]]},
-// BIN2_AMD-DAG-SAME: "device-[[T]] ([[TRIPLE:amdgcn-amd-amdhsa]]:[[ARCH2]])"
{[[P23]]}, object
+// BIN2_AMD-DAG: [[P22:[0-9]+]]: backend, {[[P5]]}, ir, (device-[[T]],
[[ARCH1]])
+// BIN2_AMD-DAG: [[P23:[0-9]+]]: backend, {[[P12]]}, ir, (device-[[T]],
[[ARCH2]])
+// BIN2_AMD-DAG: [[P24:[0-9]+]]: linker, {[[P22]]}, image, (device-[[T]],
[[ARCH1]])
+// BIN2_AMD-DAG: [[P25:[0-9]+]]: linker, {[[P23]]}, image, (device-[[T]],
[[ARCH2]])
+// BIN2_AMD-DAG: [[P26:[0-9]+]]: offload, "host-[[T]]
(powerpc64le-ibm-linux-gnu)" {[[P21]]},
+// BIN2_AMD-DAG-SAME: "device-[[T]] ([[TRIPLE:amdgcn-amd-amdhsa]]:[[ARCH1]])"
{[[P24]]},
+// BIN2_AMD-DAG-SAME: "device-[[T]] ([[TRIPLE:amdgcn-amd-amdhsa]]:[[ARCH2]])"
{[[P25]]}, object
//
// Test two gpu architecturess up to the assemble phase.
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] 8b6821a - [hip] Fix device-only relocatable code compilation.
Author: Michael Liao
Date: 2020-06-10T14:10:41-04:00
New Revision: 8b6821a5843bb321b3738e2519beae7142e62928
URL:
https://github.com/llvm/llvm-project/commit/8b6821a5843bb321b3738e2519beae7142e62928
DIFF:
https://github.com/llvm/llvm-project/commit/8b6821a5843bb321b3738e2519beae7142e62928.diff
LOG: [hip] Fix device-only relocatable code compilation.
Summary:
- In HIP, just as the regular device-only compilation, the device-only
relocatable code compilation should not involve offload bundle.
- In addition, that device-only relocatable code compilation should have
the similar 3 steps, namely preprocessor, compile, and backend, to the
regular code generation with `-emit-llvm`.
Reviewers: yaxunl, tra
Subscribers: cfe-commits
Tags: #clang
Differential Revision: https://reviews.llvm.org/D81427
Added:
clang/test/Driver/hip-rdc-device-only.hip
Modified:
clang/lib/Driver/Driver.cpp
Removed:
diff --git a/clang/lib/Driver/Driver.cpp b/clang/lib/Driver/Driver.cpp
index 8cc5eceaa512..19faf0968e08 100644
--- a/clang/lib/Driver/Driver.cpp
+++ b/clang/lib/Driver/Driver.cpp
@@ -2705,9 +2705,7 @@ class OffloadingActionBuilder final {
// backend and assemble phases to output LLVM IR. Except for generating
// non-relocatable device coee, where we generate fat binary for device
// code and pass to host in Backend phase.
- if (CudaDeviceActions.empty() ||
- (CurPhase == phases::Backend && Relocatable) ||
- CurPhase == phases::Assemble)
+ if (CudaDeviceActions.empty())
return ABRT_Success;
assert(((CurPhase == phases::Link && Relocatable) ||
@@ -2781,9 +2779,11 @@ class OffloadingActionBuilder final {
}
// By default, we produce an action for each device arch.
- for (Action * : CudaDeviceActions)
-A = C.getDriver().ConstructPhaseAction(C, Args, CurPhase, A,
- AssociatedOffloadKind);
+ if (!Relocatable || CurPhase <= phases::Backend) {
+for (Action * : CudaDeviceActions)
+ A = C.getDriver().ConstructPhaseAction(C, Args, CurPhase, A,
+ AssociatedOffloadKind);
+ }
return (CompileDeviceOnly && CurPhase == FinalPhase) ? ABRT_Ignore_Host
: ABRT_Success;
@@ -3668,7 +3668,10 @@ Action *Driver::ConstructPhaseAction(
Args.hasArg(options::OPT_S) ? types::TY_LTO_IR : types::TY_LTO_BC;
return C.MakeAction(Input, Output);
}
-if (Args.hasArg(options::OPT_emit_llvm)) {
+if (Args.hasArg(options::OPT_emit_llvm) ||
+(TargetDeviceOffloadKind == Action::OFK_HIP &&
+ Args.hasFlag(options::OPT_fgpu_rdc, options::OPT_fno_gpu_rdc,
+ false))) {
types::ID Output =
Args.hasArg(options::OPT_S) ? types::TY_LLVM_IR : types::TY_LLVM_BC;
return C.MakeAction(Input, Output);
@@ -4588,8 +4591,19 @@ const char *Driver::GetNamedOutputPath(Compilation ,
const JobAction ,
// When using both -save-temps and -emit-llvm, use a ".tmp.bc" suffix for
// the unoptimized bitcode so that it does not get overwritten by the ".bc"
// optimized bitcode output.
-if (!AtTopLevel && C.getArgs().hasArg(options::OPT_emit_llvm) &&
-JA.getType() == types::TY_LLVM_BC)
+auto IsHIPRDCInCompilePhase = [](const JobAction ,
+ const llvm::opt::DerivedArgList ) {
+ // The relocatable compilation in HIP implies -emit-llvm. Similarly, use
a
+ // ".tmp.bc" suffix for the unoptimized bitcode (generated in the compile
+ // phase.)
+ return isa(JA) &&
+ JA.getOffloadingDeviceKind() == Action::OFK_HIP &&
+ Args.hasFlag(options::OPT_fgpu_rdc, options::OPT_fno_gpu_rdc,
+ false);
+};
+if (!AtTopLevel && JA.getType() == types::TY_LLVM_BC &&
+(C.getArgs().hasArg(options::OPT_emit_llvm) ||
+ IsHIPRDCInCompilePhase(JA, C.getArgs(
Suffixed += ".tmp";
Suffixed += '.';
Suffixed += Suffix;
diff --git a/clang/test/Driver/hip-rdc-device-only.hip
b/clang/test/Driver/hip-rdc-device-only.hip
new file mode 100644
index ..4bdff628466a
--- /dev/null
+++ b/clang/test/Driver/hip-rdc-device-only.hip
@@ -0,0 +1,144 @@
+// REQUIRES: clang-driver
+// REQUIRES: x86-registered-target
+// REQUIRES: amdgpu-registered-target
+
+// RUN: %clang -### -target x86_64-linux-gnu \
+// RUN: -x hip --cuda-gpu-arch=gfx803 --cuda-gpu-arch=gfx900 \
+// RUN: -c -nogpuinc -nogpulib --cuda-device-only -fgpu-rdc \
+// RUN: %S/Inputs/hip_multiple_inputs/a.cu \
+// RUN: %S/Inputs/hip_multiple_inputs/b.hip \
+// RUN: 2>&1 | FileCheck -check-prefixes=COMMON,EMITBC %s
+
+// With `-emit-llvm`, the output should be the same as the aforementioned line
+// as
[clang] 276c8dd - [clang][codegen] Refactor argument loading in function prolog. NFC.
Author: Michael Liao
Date: 2020-05-05T15:31:51-04:00
New Revision: 276c8dde0b58cfe29035448a27e16eff9fcf2a5a
URL:
https://github.com/llvm/llvm-project/commit/276c8dde0b58cfe29035448a27e16eff9fcf2a5a
DIFF:
https://github.com/llvm/llvm-project/commit/276c8dde0b58cfe29035448a27e16eff9fcf2a5a.diff
LOG: [clang][codegen] Refactor argument loading in function prolog. NFC.
Summary:
- Skip copying function arguments and unnecessary casting by using them
directly.
Reviewers: rjmccall, kerbowa, yaxunl
Subscribers: cfe-commits
Tags: #clang
Differential Revision: https://reviews.llvm.org/D79394
Added:
Modified:
clang/lib/CodeGen/CGCall.cpp
clang/lib/CodeGen/CodeGenFunction.h
Removed:
diff --git a/clang/lib/CodeGen/CGCall.cpp b/clang/lib/CodeGen/CGCall.cpp
index 55f106e7c300..44f298892ecf 100644
--- a/clang/lib/CodeGen/CGCall.cpp
+++ b/clang/lib/CodeGen/CGCall.cpp
@@ -1016,8 +1016,8 @@ static void forConstantArrayExpansion(CodeGenFunction
,
}
}
-void CodeGenFunction::ExpandTypeFromArgs(
-QualType Ty, LValue LV, SmallVectorImpl::iterator ) {
+void CodeGenFunction::ExpandTypeFromArgs(QualType Ty, LValue LV,
+ llvm::Function::arg_iterator ) {
assert(LV.isSimple() &&
"Unexpected non-simple lvalue during struct expansion.");
@@ -1046,17 +1046,17 @@ void CodeGenFunction::ExpandTypeFromArgs(
ExpandTypeFromArgs(FD->getType(), SubLV, AI);
}
} else if (isa(Exp.get())) {
-auto realValue = *AI++;
-auto imagValue = *AI++;
+auto realValue = &*AI++;
+auto imagValue = &*AI++;
EmitStoreOfComplex(ComplexPairTy(realValue, imagValue), LV, /*init*/ true);
} else {
// Call EmitStoreOfScalar except when the lvalue is a bitfield to emit a
// primitive store.
assert(isa(Exp.get()));
if (LV.isBitField())
- EmitStoreThroughLValue(RValue::get(*AI++), LV);
+ EmitStoreThroughLValue(RValue::get(&*AI++), LV);
else
- EmitStoreOfScalar(*AI++, LV);
+ EmitStoreOfScalar(&*AI++, LV);
}
}
@@ -2323,19 +2323,13 @@ void CodeGenFunction::EmitFunctionProlog(const
CGFunctionInfo ,
// simplify.
ClangToLLVMArgMapping IRFunctionArgs(CGM.getContext(), FI);
- // Flattened function arguments.
- SmallVector FnArgs;
- FnArgs.reserve(IRFunctionArgs.totalIRArgs());
- for (auto : Fn->args()) {
-FnArgs.push_back();
- }
- assert(FnArgs.size() == IRFunctionArgs.totalIRArgs());
+ assert(Fn->arg_size() == IRFunctionArgs.totalIRArgs());
// If we're using inalloca, all the memory arguments are GEPs off of the last
// parameter, which is a pointer to the complete memory area.
Address ArgStruct = Address::invalid();
if (IRFunctionArgs.hasInallocaArg()) {
-ArgStruct = Address(FnArgs[IRFunctionArgs.getInallocaArgNo()],
+ArgStruct = Address(Fn->getArg(IRFunctionArgs.getInallocaArgNo()),
FI.getArgStructAlignment());
assert(ArgStruct.getType() == FI.getArgStruct()->getPointerTo());
@@ -2343,7 +2337,7 @@ void CodeGenFunction::EmitFunctionProlog(const
CGFunctionInfo ,
// Name the struct return parameter.
if (IRFunctionArgs.hasSRetArg()) {
-auto AI = cast(FnArgs[IRFunctionArgs.getSRetArgNo()]);
+auto AI = Fn->getArg(IRFunctionArgs.getSRetArgNo());
AI->setName("agg.result");
AI->addAttr(llvm::Attribute::NoAlias);
}
@@ -2394,7 +2388,8 @@ void CodeGenFunction::EmitFunctionProlog(const
CGFunctionInfo ,
case ABIArgInfo::Indirect: {
assert(NumIRArgs == 1);
- Address ParamAddr = Address(FnArgs[FirstIRArg], ArgI.getIndirectAlign());
+ Address ParamAddr =
+ Address(Fn->getArg(FirstIRArg), ArgI.getIndirectAlign());
if (!hasScalarEvaluationKind(Ty)) {
// Aggregates and complex variables are accessed by reference. All we
@@ -2436,8 +2431,7 @@ void CodeGenFunction::EmitFunctionProlog(const
CGFunctionInfo ,
ArgI.getCoerceToType() == ConvertType(Ty) &&
ArgI.getDirectOffset() == 0) {
assert(NumIRArgs == 1);
-llvm::Value *V = FnArgs[FirstIRArg];
-auto AI = cast(V);
+auto AI = Fn->getArg(FirstIRArg);
if (const ParmVarDecl *PVD = dyn_cast(Arg)) {
if (getNonNullAttr(CurCodeDecl, PVD, PVD->getType(),
@@ -2499,6 +2493,7 @@ void CodeGenFunction::EmitFunctionProlog(const
CGFunctionInfo ,
// LLVM expects swifterror parameters to be used in very restricted
// ways. Copy the value into a less-restricted temporary.
+llvm::Value *V = AI;
if (FI.getExtParameterInfo(ArgNo).getABI()
== ParameterABI::SwiftErrorResult) {
QualType pointeeTy = Ty->getPointeeType();
@@ -2560,7 +2555,7 @@ void CodeGenFunction::EmitFunctionProlog(const
CGFunctionInfo ,
assert(STy->getNumElements() == NumIRArgs);
for (unsigned i = 0, e = STy->getNumElements();
[clang] 9142c0b - [clang][codegen] Hoist parameter attribute setting in function prolog.
Author: Michael Liao
Date: 2020-05-05T15:31:51-04:00
New Revision: 9142c0b46bfea13d9348ab3d1d706a10ad9e5c8e
URL:
https://github.com/llvm/llvm-project/commit/9142c0b46bfea13d9348ab3d1d706a10ad9e5c8e
DIFF:
https://github.com/llvm/llvm-project/commit/9142c0b46bfea13d9348ab3d1d706a10ad9e5c8e.diff
LOG: [clang][codegen] Hoist parameter attribute setting in function prolog.
Summary:
- If the coerced type is still a pointer, it should be set with proper
parameter attributes, such as `noalias`, `nonnull`, and etc. Hoist
that (pointer) parameter attribute setting so that the coerced pointer
parameter could be marked properly.
Depends on D79394
Reviewers: rjmccall, kerbowa, yaxunl
Subscribers: jvesely, nhaehnle, cfe-commits
Tags: #clang
Differential Revision: https://reviews.llvm.org/D79395
Added:
Modified:
clang/lib/CodeGen/CGCall.cpp
clang/test/CodeGenCUDA/amdgpu-kernel-arg-pointer-type.cu
Removed:
diff --git a/clang/lib/CodeGen/CGCall.cpp b/clang/lib/CodeGen/CGCall.cpp
index 44f298892ecf..e336741d9111 100644
--- a/clang/lib/CodeGen/CGCall.cpp
+++ b/clang/lib/CodeGen/CGCall.cpp
@@ -2425,15 +2425,18 @@ void CodeGenFunction::EmitFunctionProlog(const
CGFunctionInfo ,
case ABIArgInfo::Extend:
case ABIArgInfo::Direct: {
-
- // If we have the trivial case, handle it with no muss and fuss.
- if (!isa(ArgI.getCoerceToType()) &&
- ArgI.getCoerceToType() == ConvertType(Ty) &&
- ArgI.getDirectOffset() == 0) {
+ auto AI = Fn->getArg(FirstIRArg);
+ llvm::Type *LTy = ConvertType(Arg->getType());
+
+ // Prepare parameter attributes. So far, only attributes for pointer
+ // parameters are prepared. See
+ // http://llvm.org/docs/LangRef.html#paramattrs.
+ if (ArgI.getDirectOffset() == 0 && LTy->isPointerTy() &&
+ ArgI.getCoerceToType()->isPointerTy()) {
assert(NumIRArgs == 1);
-auto AI = Fn->getArg(FirstIRArg);
if (const ParmVarDecl *PVD = dyn_cast(Arg)) {
+ // Set `nonnull` attribute if any.
if (getNonNullAttr(CurCodeDecl, PVD, PVD->getType(),
PVD->getFunctionScopeIndex()) &&
!CGM.getCodeGenOpts().NullPointerIsValid)
@@ -2471,6 +2474,7 @@ void CodeGenFunction::EmitFunctionProlog(const
CGFunctionInfo ,
AI->addAttr(llvm::Attribute::NonNull);
}
+ // Set `align` attribute if any.
const auto *AVAttr = PVD->getAttr();
if (!AVAttr)
if (const auto *TOTy = dyn_cast(OTy))
@@ -2488,8 +2492,17 @@ void CodeGenFunction::EmitFunctionProlog(const
CGFunctionInfo ,
}
}
+// Set 'noalias' if an argument type has the `restrict` qualifier.
if (Arg->getType().isRestrictQualified())
AI->addAttr(llvm::Attribute::NoAlias);
+ }
+
+ // Prepare the argument value. If we have the trivial case, handle it
+ // with no muss and fuss.
+ if (!isa(ArgI.getCoerceToType()) &&
+ ArgI.getCoerceToType() == ConvertType(Ty) &&
+ ArgI.getDirectOffset() == 0) {
+assert(NumIRArgs == 1);
// LLVM expects swifterror parameters to be used in very restricted
// ways. Copy the value into a less-restricted temporary.
diff --git a/clang/test/CodeGenCUDA/amdgpu-kernel-arg-pointer-type.cu
b/clang/test/CodeGenCUDA/amdgpu-kernel-arg-pointer-type.cu
index 6e4de1f0f5c3..8aeb0f759e6c 100644
--- a/clang/test/CodeGenCUDA/amdgpu-kernel-arg-pointer-type.cu
+++ b/clang/test/CodeGenCUDA/amdgpu-kernel-arg-pointer-type.cu
@@ -67,3 +67,10 @@ __global__ void kernel6(struct T t) {
t.x[0][0] += 1.f;
t.x[1][0] += 2.f;
}
+
+// Check that coerced pointers retain the noalias attribute when qualified
with __restrict.
+// CHECK: define amdgpu_kernel void @_Z7kernel7Pi(i32 addrspace(1)* noalias
%x.coerce)
+// HOST: define void @_Z22__device_stub__kernel7Pi(i32* noalias %x)
+__global__ void kernel7(int *__restrict x) {
+ x[0]++;
+}
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] f3a3db8 - Add the missing '='. NFC.
Author: Michael Liao
Date: 2020-05-02T01:07:44-04:00
New Revision: f3a3db8627e9fa673fa413a2e41fe5443db7c6c3
URL:
https://github.com/llvm/llvm-project/commit/f3a3db8627e9fa673fa413a2e41fe5443db7c6c3
DIFF:
https://github.com/llvm/llvm-project/commit/f3a3db8627e9fa673fa413a2e41fe5443db7c6c3.diff
LOG: Add the missing '='. NFC.
Added:
Modified:
clang/lib/Format/UnwrappedLineParser.cpp
Removed:
diff --git a/clang/lib/Format/UnwrappedLineParser.cpp
b/clang/lib/Format/UnwrappedLineParser.cpp
index 7456e6d5fb48..96e0bd2276fa 100644
--- a/clang/lib/Format/UnwrappedLineParser.cpp
+++ b/clang/lib/Format/UnwrappedLineParser.cpp
@@ -1471,7 +1471,7 @@ void UnwrappedLineParser::parseStructuralElement() {
} else if (Style.Language == FormatStyle::LK_Proto &&
FormatTok->Tok.is(tok::less)) {
nextToken();
-parseBracedList(/*ContinueOnSemicolons=*/false, /*IsEnum*/false,
+parseBracedList(/*ContinueOnSemicolons=*/false, /*IsEnum=*/false,
/*ClosingBraceKind=*/tok::greater);
}
break;
@@ -1824,7 +1824,7 @@ bool UnwrappedLineParser::parseBracedList(bool
ContinueOnSemicolons,
case tok::less:
if (Style.Language == FormatStyle::LK_Proto) {
nextToken();
-parseBracedList(/*ContinueOnSemicolons=*/false, /*IsEnum*/false,
+parseBracedList(/*ContinueOnSemicolons=*/false, /*IsEnum=*/false,
/*ClosingBraceKind=*/tok::greater);
} else {
nextToken();
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] d1c4361 - [clang-format] Add the missing default argument.
Author: Michael Liao
Date: 2020-04-30T17:36:43-04:00
New Revision: d1c43615ed068f2f915ccdd959ef583cd5177929
URL:
https://github.com/llvm/llvm-project/commit/d1c43615ed068f2f915ccdd959ef583cd5177929
DIFF:
https://github.com/llvm/llvm-project/commit/d1c43615ed068f2f915ccdd959ef583cd5177929.diff
LOG: [clang-format] Add the missing default argument.
Added:
Modified:
clang/lib/Format/UnwrappedLineParser.cpp
Removed:
diff --git a/clang/lib/Format/UnwrappedLineParser.cpp
b/clang/lib/Format/UnwrappedLineParser.cpp
index c9528188c61c..7456e6d5fb48 100644
--- a/clang/lib/Format/UnwrappedLineParser.cpp
+++ b/clang/lib/Format/UnwrappedLineParser.cpp
@@ -1471,7 +1471,7 @@ void UnwrappedLineParser::parseStructuralElement() {
} else if (Style.Language == FormatStyle::LK_Proto &&
FormatTok->Tok.is(tok::less)) {
nextToken();
-parseBracedList(/*ContinueOnSemicolons=*/false,
+parseBracedList(/*ContinueOnSemicolons=*/false, /*IsEnum*/false,
/*ClosingBraceKind=*/tok::greater);
}
break;
@@ -1824,7 +1824,7 @@ bool UnwrappedLineParser::parseBracedList(bool
ContinueOnSemicolons,
case tok::less:
if (Style.Language == FormatStyle::LK_Proto) {
nextToken();
-parseBracedList(/*ContinueOnSemicolons=*/false,
+parseBracedList(/*ContinueOnSemicolons=*/false, /*IsEnum*/false,
/*ClosingBraceKind=*/tok::greater);
} else {
nextToken();
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] 612720d - [hip] Remove test using `hip_pinned_shadow` attribute. NFC.
Author: Michael Liao
Date: 2020-04-27T16:44:59-04:00
New Revision: 612720db874d06a50b793c301e5b3b857e3e7c70
URL:
https://github.com/llvm/llvm-project/commit/612720db874d06a50b793c301e5b3b857e3e7c70
DIFF:
https://github.com/llvm/llvm-project/commit/612720db874d06a50b793c301e5b3b857e3e7c70.diff
LOG: [hip] Remove test using `hip_pinned_shadow` attribute. NFC.
Added:
Modified:
Removed:
clang/test/CodeGenCUDA/hip-pinned-shadow.hip
diff --git a/clang/test/CodeGenCUDA/hip-pinned-shadow.hip
b/clang/test/CodeGenCUDA/hip-pinned-shadow.hip
deleted file mode 100644
index 7f0e7544d828..
--- a/clang/test/CodeGenCUDA/hip-pinned-shadow.hip
+++ /dev/null
@@ -1,27 +0,0 @@
-// REQUIRES: amdgpu-registered-target
-
-// RUN: %clang_cc1 -triple amdgcn -fcuda-is-device -std=c++11 -fvisibility
hidden -fapply-global-visibility-to-externs \
-// RUN: -emit-llvm -o - %s | FileCheck -check-prefixes=HIPDEV %s
-// RUN: %clang_cc1 -triple x86_64 -std=c++11 \
-// RUN: -emit-llvm -o - %s | FileCheck -check-prefixes=HIPHOST %s
-// RUN: %clang_cc1 -triple amdgcn -fcuda-is-device -std=c++11 -fvisibility
hidden -fapply-global-visibility-to-externs \
-// RUN: -O3 -emit-llvm -o - %s | FileCheck -check-prefixes=HIPDEVUNSED %s
-
-struct textureReference {
- int a;
-};
-
-template
-struct texture : public textureReference {
-texture() { a = 1; }
-};
-
-__attribute__((hip_pinned_shadow)) texture tex;
-// CUDADEV-NOT: @tex
-// CUDAHOST-NOT: call i32 @__hipRegisterVar{{.*}}@tex
-// HIPDEV: @tex = external addrspace(1) global %struct.texture
-// HIPDEV-NOT: declare{{.*}}void @_ZN7textureIfLi2ELi1EEC1Ev
-// HIPHOST: define{{.*}}@_ZN7textureIfLi2ELi1EEC1Ev
-// HIPHOST: call i32 @__hipRegisterVar{{.*}}@tex{{.*}}i32 0, i32 4, i32 0,
i32 0)
-// HIPDEVUNSED: @tex = external addrspace(1) global %struct.texture
-// HIPDEVUNSED-NOT: declare{{.*}}void @_ZN7textureIfLi2ELi1EEC1Ev
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] 86e3b73 - [hip] Claim builtin type `__float128` supported if the host target supports it.
Author: Michael Liao
Date: 2020-04-21T15:56:40-04:00
New Revision: 86e3b735cd803cc22c9eae15d99ce9df5956aae6
URL:
https://github.com/llvm/llvm-project/commit/86e3b735cd803cc22c9eae15d99ce9df5956aae6
DIFF:
https://github.com/llvm/llvm-project/commit/86e3b735cd803cc22c9eae15d99ce9df5956aae6.diff
LOG: [hip] Claim builtin type `__float128` supported if the host target
supports it.
Reviewers: tra, yaxunl
Subscribers: jvesely, nhaehnle, kerbowa, cfe-commits
Tags: #clang
Differential Revision: https://reviews.llvm.org/D78513
Added:
clang/test/SemaCUDA/amdgpu-f128.cu
Modified:
clang/lib/Basic/Targets/AMDGPU.cpp
Removed:
diff --git a/clang/lib/Basic/Targets/AMDGPU.cpp
b/clang/lib/Basic/Targets/AMDGPU.cpp
index 3fd9008e4660..b9d7640a10b8 100644
--- a/clang/lib/Basic/Targets/AMDGPU.cpp
+++ b/clang/lib/Basic/Targets/AMDGPU.cpp
@@ -363,4 +363,17 @@ void AMDGPUTargetInfo::setAuxTarget(const TargetInfo *Aux)
{
copyAuxTarget(Aux);
LongDoubleFormat = SaveLongDoubleFormat;
Float128Format = SaveFloat128Format;
+ // For certain builtin types support on the host target, claim they are
+ // support to pass the compilation of the host code during the device-side
+ // compilation.
+ // FIXME: As the side effect, we also accept `__float128` uses in the device
+ // code. To rejct these builtin types supported in the host target but not in
+ // the device target, one approach would support `device_builtin` attribute
+ // so that we could tell the device builtin types from the host ones. The
+ // also solves the
diff erent representations of the same builtin type, such
+ // as `size_t` in the MSVC environment.
+ if (Aux->hasFloat128Type()) {
+HasFloat128 = true;
+Float128Format = DoubleFormat;
+ }
}
diff --git a/clang/test/SemaCUDA/amdgpu-f128.cu
b/clang/test/SemaCUDA/amdgpu-f128.cu
new file mode 100644
index ..9a0212cdb93c
--- /dev/null
+++ b/clang/test/SemaCUDA/amdgpu-f128.cu
@@ -0,0 +1,4 @@
+// RUN: %clang_cc1 -triple amdgcn-amd-amdhsa -aux-triple
x86_64-unknown-linux-gnu -fcuda-is-device -fsyntax-only -verify %s
+
+// expected-no-diagnostics
+typedef __float128 f128_t;
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] 50472c4 - Remove extra ‘;’. NFC.
Author: Michael Liao
Date: 2020-04-15T17:22:03-04:00
New Revision: 50472c422cc6d27a4532a4025c4339fb6920
URL:
https://github.com/llvm/llvm-project/commit/50472c422cc6d27a4532a4025c4339fb6920
DIFF:
https://github.com/llvm/llvm-project/commit/50472c422cc6d27a4532a4025c4339fb6920.diff
LOG: Remove extra ‘;’. NFC.
Added:
Modified:
clang/include/clang/AST/RecursiveASTVisitor.h
Removed:
diff --git a/clang/include/clang/AST/RecursiveASTVisitor.h
b/clang/include/clang/AST/RecursiveASTVisitor.h
index 10ea91ea9cfa..85eb6259a419 100644
--- a/clang/include/clang/AST/RecursiveASTVisitor.h
+++ b/clang/include/clang/AST/RecursiveASTVisitor.h
@@ -1993,7 +1993,7 @@ DEF_TRAVERSE_DECL(BindingDecl, {
DEF_TRAVERSE_DECL(MSPropertyDecl, { TRY_TO(TraverseDeclaratorHelper(D)); })
-DEF_TRAVERSE_DECL(MSGuidDecl, {});
+DEF_TRAVERSE_DECL(MSGuidDecl, {})
DEF_TRAVERSE_DECL(FieldDecl, {
TRY_TO(TraverseDeclaratorHelper(D));
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] 96c4ec8 - Remove extra whitespace. NFC.
Author: Michael Liao Date: 2020-04-10T03:22:01-04:00 New Revision: 96c4ec8fdbd95048114cf058679bd8fc08ab76b3 URL: https://github.com/llvm/llvm-project/commit/96c4ec8fdbd95048114cf058679bd8fc08ab76b3 DIFF: https://github.com/llvm/llvm-project/commit/96c4ec8fdbd95048114cf058679bd8fc08ab76b3.diff LOG: Remove extra whitespace. NFC. Added: Modified: clang/include/clang/Basic/DiagnosticDriverKinds.td Removed: diff --git a/clang/include/clang/Basic/DiagnosticDriverKinds.td b/clang/include/clang/Basic/DiagnosticDriverKinds.td index e35ca843ff56..cba59cb3b66d 100644 --- a/clang/include/clang/Basic/DiagnosticDriverKinds.td +++ b/clang/include/clang/Basic/DiagnosticDriverKinds.td @@ -50,14 +50,14 @@ def warn_drv_avr_stdlib_not_linked: Warning< InGroup; def err_drv_cuda_bad_gpu_arch : Error<"Unsupported CUDA gpu architecture: %0">; def err_drv_no_cuda_installation : Error< - "cannot find CUDA installation. Provide its path via --cuda-path, or pass " + "cannot find CUDA installation. Provide its path via --cuda-path, or pass " "-nocudainc to build without CUDA includes.">; def err_drv_no_cuda_libdevice : Error< "cannot find libdevice for %0. Provide path to diff erent CUDA installation " "via --cuda-path, or pass -nocudalib to build without linking with libdevice.">; def err_drv_cuda_version_unsupported : Error< "GPU arch %0 is supported by CUDA versions between %1 and %2 (inclusive), " - "but installation at %3 is %4. Use --cuda-path to specify a diff erent CUDA " + "but installation at %3 is %4. Use --cuda-path to specify a diff erent CUDA " "install, pass a diff erent GPU arch with --cuda-gpu-arch, or pass " "--no-cuda-version-check.">; def warn_drv_unknown_cuda_version: Warning< ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] c97be2c - [hip] Remove `hip_pinned_shadow`.
Author: Michael Liao
Date: 2020-04-07T09:51:49-04:00
New Revision: c97be2c377852fad7eb38409aae5692fa417e49b
URL:
https://github.com/llvm/llvm-project/commit/c97be2c377852fad7eb38409aae5692fa417e49b
DIFF:
https://github.com/llvm/llvm-project/commit/c97be2c377852fad7eb38409aae5692fa417e49b.diff
LOG: [hip] Remove `hip_pinned_shadow`.
Summary:
- Use `device_builtin_surface` and `device_builtin_texture` for
surface/texture reference support. So far, both the host and device
use the same reference type, which could be revised later when
interface/implementation is stablized.
Reviewers: yaxunl
Subscribers: cfe-commits
Tags: #clang
Differential Revision: https://reviews.llvm.org/D77583
Added:
Modified:
clang/include/clang/Basic/Attr.td
clang/include/clang/Basic/AttrDocs.td
clang/lib/CodeGen/CodeGenModule.cpp
clang/lib/CodeGen/CodeGenModule.h
clang/lib/CodeGen/TargetInfo.cpp
clang/lib/Driver/ToolChains/HIP.cpp
clang/lib/Sema/SemaDeclAttr.cpp
clang/test/Driver/hip-toolchain-no-rdc.hip
clang/test/Driver/hip-toolchain-rdc.hip
clang/test/Misc/pragma-attribute-supported-attributes-list.test
Removed:
clang/test/AST/ast-dump-hip-pinned-shadow.cu
clang/test/SemaCUDA/hip-pinned-shadow.cu
diff --git a/clang/include/clang/Basic/Attr.td
b/clang/include/clang/Basic/Attr.td
index f55ce2cc84dd..c586f9b9466a 100644
--- a/clang/include/clang/Basic/Attr.td
+++ b/clang/include/clang/Basic/Attr.td
@@ -322,7 +322,6 @@ class LangOpt {
def MicrosoftExt : LangOpt<"MicrosoftExt">;
def Borland : LangOpt<"Borland">;
def CUDA : LangOpt<"CUDA">;
-def HIP : LangOpt<"HIP">;
def SYCL : LangOpt<"SYCLIsDevice">;
def COnly : LangOpt<"", "!LangOpts.CPlusPlus">;
def CPlusPlus : LangOpt<"CPlusPlus">;
@@ -1052,13 +1051,6 @@ def CUDADevice : InheritableAttr {
let Documentation = [Undocumented];
}
-def HIPPinnedShadow : InheritableAttr {
- let Spellings = [GNU<"hip_pinned_shadow">,
Declspec<"__hip_pinned_shadow__">];
- let Subjects = SubjectList<[Var]>;
- let LangOpts = [HIP];
- let Documentation = [HIPPinnedShadowDocs];
-}
-
def CUDADeviceBuiltin : IgnoredAttr {
let Spellings = [GNU<"device_builtin">, Declspec<"__device_builtin__">];
let LangOpts = [CUDA];
diff --git a/clang/include/clang/Basic/AttrDocs.td
b/clang/include/clang/Basic/AttrDocs.td
index fb1c82a80115..36561c04d395 100644
--- a/clang/include/clang/Basic/AttrDocs.td
+++ b/clang/include/clang/Basic/AttrDocs.td
@@ -4613,18 +4613,6 @@ only call one function.
}];
}
-def HIPPinnedShadowDocs : Documentation {
- let Category = DocCatType;
- let Content = [{
-The GNU style attribute __attribute__((hip_pinned_shadow)) or MSVC style
attribute
-__declspec(hip_pinned_shadow) can be added to the definition of a global
variable
-to indicate it is a HIP pinned shadow variable. A HIP pinned shadow variable
can
-be accessed on both device side and host side. It has external linkage and is
-not initialized on device side. It has internal linkage and is initialized by
-the initializer on host side.
- }];
-}
-
def CUDADeviceBuiltinSurfaceTypeDocs : Documentation {
let Category = DocCatType;
let Content = [{
diff --git a/clang/lib/CodeGen/CodeGenModule.cpp
b/clang/lib/CodeGen/CodeGenModule.cpp
index 1645a9eb17de..8b7d52b88146 100644
--- a/clang/lib/CodeGen/CodeGenModule.cpp
+++ b/clang/lib/CodeGen/CodeGenModule.cpp
@@ -1955,9 +1955,9 @@ void CodeGenModule::SetFunctionAttributes(GlobalDecl GD,
llvm::Function *F,
}
}
-void CodeGenModule::addUsedGlobal(llvm::GlobalValue *GV, bool SkipCheck) {
- assert(SkipCheck || (!GV->isDeclaration() &&
- "Only globals with definition can force usage."));
+void CodeGenModule::addUsedGlobal(llvm::GlobalValue *GV) {
+ assert(!GV->isDeclaration() &&
+ "Only globals with definition can force usage.");
LLVMUsed.emplace_back(GV);
}
@@ -2520,7 +2520,6 @@ void CodeGenModule::EmitGlobal(GlobalDecl GD) {
!Global->hasAttr() &&
!Global->hasAttr() &&
!Global->hasAttr() &&
- !(LangOpts.HIP && Global->hasAttr()) &&
!Global->getType()->isCUDADeviceBuiltinSurfaceType() &&
!Global->getType()->isCUDADeviceBuiltinTextureType())
return;
@@ -3928,10 +3927,8 @@ void CodeGenModule::EmitGlobalVarDefinition(const
VarDecl *D,
D->getType()->isCUDADeviceBuiltinTextureType());
// HIP pinned shadow of initialized host-side global variables are also
// left undefined.
- bool IsHIPPinnedShadowVar =
- getLangOpts().CUDAIsDevice && D->hasAttr();
- if (getLangOpts().CUDA && (IsCUDASharedVar || IsCUDAShadowVar ||
- IsCUDADeviceShadowVar || IsHIPPinnedShadowVar))
+ if (getLangOpts().CUDA &&
+ (IsCUDASharedVar || IsCUDAShadowVar || IsCUDADeviceShadowVar))
Init = llvm::UndefValue::get(getTypes().ConvertType(ASTTy));
else if
[clang] b952d79 - [cuda][hip] Fix `RegisterVar` function prototype.
Author: Michael Liao
Date: 2020-04-03T12:57:09-04:00
New Revision: b952d799cacdb7efd44c1c9468bb11471cc16874
URL:
https://github.com/llvm/llvm-project/commit/b952d799cacdb7efd44c1c9468bb11471cc16874
DIFF:
https://github.com/llvm/llvm-project/commit/b952d799cacdb7efd44c1c9468bb11471cc16874.diff
LOG: [cuda][hip] Fix `RegisterVar` function prototype.
Summary:
- `RegisterVar` has `void` return type and `size_t` in its variable size
parameter in HIP or CUDA 9.0+.
Reviewers: tra, yaxunl
Subscribers: cfe-commits
Tags: #clang
Differential Revision: https://reviews.llvm.org/D77398
Added:
Modified:
clang/include/clang/Basic/Cuda.h
clang/lib/Basic/Cuda.cpp
clang/lib/CodeGen/CGCUDANV.cpp
clang/test/CodeGenCUDA/device-stub.cu
Removed:
diff --git a/clang/include/clang/Basic/Cuda.h
b/clang/include/clang/Basic/Cuda.h
index da572957d10d..c2ebf8734304 100644
--- a/clang/include/clang/Basic/Cuda.h
+++ b/clang/include/clang/Basic/Cuda.h
@@ -117,6 +117,7 @@ enum class CudaFeature {
CUDA_USES_FATBIN_REGISTER_END,
};
+CudaVersion ToCudaVersion(llvm::VersionTuple);
bool CudaFeatureEnabled(llvm::VersionTuple, CudaFeature);
bool CudaFeatureEnabled(CudaVersion, CudaFeature);
diff --git a/clang/lib/Basic/Cuda.cpp b/clang/lib/Basic/Cuda.cpp
index e06d120c58bf..74eb5473b71d 100644
--- a/clang/lib/Basic/Cuda.cpp
+++ b/clang/lib/Basic/Cuda.cpp
@@ -362,7 +362,7 @@ CudaVersion MaxVersionForCudaArch(CudaArch A) {
}
}
-static CudaVersion ToCudaVersion(llvm::VersionTuple Version) {
+CudaVersion ToCudaVersion(llvm::VersionTuple Version) {
int IVer =
Version.getMajor() * 10 + Version.getMinor().getValueOr(0);
switch(IVer) {
diff --git a/clang/lib/CodeGen/CGCUDANV.cpp b/clang/lib/CodeGen/CGCUDANV.cpp
index 6d92ef33b885..351c5058aa4c 100644
--- a/clang/lib/CodeGen/CGCUDANV.cpp
+++ b/clang/lib/CodeGen/CGCUDANV.cpp
@@ -440,13 +440,19 @@ llvm::Function *CGNVCUDARuntime::makeRegisterGlobalsFn() {
Builder.CreateCall(RegisterFunc, Args);
}
+ llvm::Type *VarSizeTy = IntTy;
+ // For HIP or CUDA 9.0+, device variable size is type of `size_t`.
+ if (CGM.getLangOpts().HIP ||
+ ToCudaVersion(CGM.getTarget().getSDKVersion()) >= CudaVersion::CUDA_90)
+VarSizeTy = SizeTy;
+
// void __cudaRegisterVar(void **, char *, char *, const char *,
//int, int, int, int)
llvm::Type *RegisterVarParams[] = {VoidPtrPtrTy, CharPtrTy, CharPtrTy,
- CharPtrTy,IntTy, IntTy,
+ CharPtrTy,IntTy, VarSizeTy,
IntTy,IntTy};
llvm::FunctionCallee RegisterVar = CGM.CreateRuntimeFunction(
- llvm::FunctionType::get(IntTy, RegisterVarParams, false),
+ llvm::FunctionType::get(VoidTy, RegisterVarParams, false),
addUnderscoredPrefixToName("RegisterVar"));
// void __cudaRegisterSurface(void **, const struct surfaceReference *,
//const void **, const char *, int, int);
@@ -476,7 +482,7 @@ llvm::Function *CGNVCUDARuntime::makeRegisterGlobalsFn() {
VarName,
VarName,
llvm::ConstantInt::get(IntTy, Info.Flags.isExtern()),
- llvm::ConstantInt::get(IntTy, VarSize),
+ llvm::ConstantInt::get(VarSizeTy, VarSize),
llvm::ConstantInt::get(IntTy, Info.Flags.isConstant()),
llvm::ConstantInt::get(IntTy, 0)};
Builder.CreateCall(RegisterVar, Args);
diff --git a/clang/test/CodeGenCUDA/device-stub.cu
b/clang/test/CodeGenCUDA/device-stub.cu
index 9db5738cdede..0f4a5644fd48 100644
--- a/clang/test/CodeGenCUDA/device-stub.cu
+++ b/clang/test/CodeGenCUDA/device-stub.cu
@@ -181,10 +181,10 @@ void hostfunc(void) { kernelfunc<<<1, 1>>>(1, 1, 1); }
// Test that we've built a function to register kernels and global vars.
// ALL: define internal void @__[[PREFIX]]_register_globals
// ALL: call{{.*}}[[PREFIX]]RegisterFunction(i8** %0,
{{.*}}kernelfunc{{[^,]*}}, {{[^@]*}}@0
-// ALL-DAG: call{{.*}}[[PREFIX]]RegisterVar(i8** %0,
{{.*}}device_var{{[^,]*}}, {{[^@]*}}@1, {{.*}}i32 0, i32 4, i32 0, i32 0
-// ALL-DAG: call{{.*}}[[PREFIX]]RegisterVar(i8** %0,
{{.*}}constant_var{{[^,]*}}, {{[^@]*}}@2, {{.*}}i32 0, i32 4, i32 1, i32 0
-// ALL-DAG: call{{.*}}[[PREFIX]]RegisterVar(i8** %0,
{{.*}}ext_device_var_def{{[^,]*}}, {{[^@]*}}@3, {{.*}}i32 0, i32 4, i32 0, i32 0
-// ALL-DAG: call{{.*}}[[PREFIX]]RegisterVar(i8** %0,
{{.*}}ext_constant_var_def{{[^,]*}}, {{[^@]*}}@4, {{.*}}i32 0, i32 4, i32 1,
i32 0
+// ALL-DAG: call void {{.*}}[[PREFIX]]RegisterVar(i8** %0,
{{.*}}device_var{{[^,]*}}, {{[^@]*}}@1, {{.*}}i32 0, {{i32|i64}} 4, i32 0, i32 0
+// ALL-DAG: call void {{.*}}[[PREFIX]]RegisterVar(i8** %0,
{{.*}}constant_var{{[^,]*}}, {{[^@]*}}@2, {{.*}}i32 0, {{i32|i64}} 4, i32 1,
i32 0
+// ALL-DAG: call void {{.*}}[[PREFIX]]RegisterVar(i8** %0,
[clang] cb63893 - Fix GCC warning on enum class bitfield. NFC.
Author: Michael Liao
Date: 2020-03-28T10:20:34-04:00
New Revision: cb6389360b05e8f89d09ff133a4ba1fd011866c5
URL:
https://github.com/llvm/llvm-project/commit/cb6389360b05e8f89d09ff133a4ba1fd011866c5
DIFF:
https://github.com/llvm/llvm-project/commit/cb6389360b05e8f89d09ff133a4ba1fd011866c5.diff
LOG: Fix GCC warning on enum class bitfield. NFC.
Added:
Modified:
clang/lib/CodeGen/CGCUDANV.cpp
clang/lib/CodeGen/CGCUDARuntime.h
Removed:
diff --git a/clang/lib/CodeGen/CGCUDANV.cpp b/clang/lib/CodeGen/CGCUDANV.cpp
index ed02a7dc9173..6d92ef33b885 100644
--- a/clang/lib/CodeGen/CGCUDANV.cpp
+++ b/clang/lib/CodeGen/CGCUDANV.cpp
@@ -466,18 +466,19 @@ llvm::Function *CGNVCUDARuntime::makeRegisterGlobalsFn() {
for (auto & : DeviceVars) {
llvm::GlobalVariable *Var = Info.Var;
llvm::Constant *VarName = makeConstantString(getDeviceSideName(Info.D));
-switch (Info.Flags.Kind) {
+switch (Info.Flags.getKind()) {
case DeviceVarFlags::Variable: {
uint64_t VarSize =
CGM.getDataLayout().getTypeAllocSize(Var->getValueType());
- llvm::Value *Args[] = {,
- Builder.CreateBitCast(Var, VoidPtrTy),
- VarName,
- VarName,
- llvm::ConstantInt::get(IntTy, Info.Flags.Extern),
- llvm::ConstantInt::get(IntTy, VarSize),
- llvm::ConstantInt::get(IntTy,
Info.Flags.Constant),
- llvm::ConstantInt::get(IntTy, 0)};
+ llvm::Value *Args[] = {
+ ,
+ Builder.CreateBitCast(Var, VoidPtrTy),
+ VarName,
+ VarName,
+ llvm::ConstantInt::get(IntTy, Info.Flags.isExtern()),
+ llvm::ConstantInt::get(IntTy, VarSize),
+ llvm::ConstantInt::get(IntTy, Info.Flags.isConstant()),
+ llvm::ConstantInt::get(IntTy, 0)};
Builder.CreateCall(RegisterVar, Args);
break;
}
@@ -485,16 +486,16 @@ llvm::Function *CGNVCUDARuntime::makeRegisterGlobalsFn() {
Builder.CreateCall(
RegisterSurf,
{, Builder.CreateBitCast(Var, VoidPtrTy), VarName,
- VarName, llvm::ConstantInt::get(IntTy, Info.Flags.SurfTexType),
- llvm::ConstantInt::get(IntTy, Info.Flags.Extern)});
+ VarName, llvm::ConstantInt::get(IntTy, Info.Flags.getSurfTexType()),
+ llvm::ConstantInt::get(IntTy, Info.Flags.isExtern())});
break;
case DeviceVarFlags::Texture:
Builder.CreateCall(
RegisterTex,
{, Builder.CreateBitCast(Var, VoidPtrTy), VarName,
- VarName, llvm::ConstantInt::get(IntTy, Info.Flags.SurfTexType),
- llvm::ConstantInt::get(IntTy, Info.Flags.Normalized),
- llvm::ConstantInt::get(IntTy, Info.Flags.Extern)});
+ VarName, llvm::ConstantInt::get(IntTy, Info.Flags.getSurfTexType()),
+ llvm::ConstantInt::get(IntTy, Info.Flags.isNormalized()),
+ llvm::ConstantInt::get(IntTy, Info.Flags.isExtern())});
break;
}
}
diff --git a/clang/lib/CodeGen/CGCUDARuntime.h
b/clang/lib/CodeGen/CGCUDARuntime.h
index b26132420d65..19e70a2022a5 100644
--- a/clang/lib/CodeGen/CGCUDARuntime.h
+++ b/clang/lib/CodeGen/CGCUDARuntime.h
@@ -42,17 +42,30 @@ class CGCUDARuntime {
public:
// Global variable properties that must be passed to CUDA runtime.
- struct DeviceVarFlags {
-enum DeviceVarKind : unsigned {
+ class DeviceVarFlags {
+ public:
+enum DeviceVarKind {
Variable, // Variable
Surface, // Builtin surface
Texture, // Builtin texture
};
-DeviceVarKind Kind : 2;
+
+ private:
+unsigned Kind : 2;
unsigned Extern : 1;
unsigned Constant : 1; // Constant variable.
unsigned Normalized : 1; // Normalized texture.
int SurfTexType; // Type of surface/texutre.
+
+ public:
+DeviceVarFlags(DeviceVarKind K, bool E, bool C, bool N, int T)
+: Kind(K), Extern(E), Constant(C), Normalized(N), SurfTexType(T) {}
+
+DeviceVarKind getKind() const { return static_cast(Kind); }
+bool isExtern() const { return Extern; }
+bool isConstant() const { return Constant; }
+bool isNormalized() const { return Normalized; }
+int getSurfTexType() const { return SurfTexType; }
};
CGCUDARuntime(CodeGenModule ) : CGM(CGM) {}
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] 5be9b8c - [cuda][hip] Add CUDA builtin surface/texture reference support.
Author: Michael Liao
Date: 2020-03-27T17:18:49-04:00
New Revision: 5be9b8cbe2b2253f78a09a863bef18e574737465
URL:
https://github.com/llvm/llvm-project/commit/5be9b8cbe2b2253f78a09a863bef18e574737465
DIFF:
https://github.com/llvm/llvm-project/commit/5be9b8cbe2b2253f78a09a863bef18e574737465.diff
LOG: [cuda][hip] Add CUDA builtin surface/texture reference support.
Summary: - Re-commit after fix Sema checks on partial template specialization.
Reviewers: tra, rjmccall, yaxunl, a.sidorin
Subscribers: cfe-commits
Tags: #clang
Differential Revision: https://reviews.llvm.org/D76365
Added:
clang/test/CodeGenCUDA/surface.cu
clang/test/CodeGenCUDA/texture.cu
Modified:
clang/include/clang/AST/Type.h
clang/include/clang/Basic/Attr.td
clang/include/clang/Basic/AttrDocs.td
clang/include/clang/Basic/DiagnosticSemaKinds.td
clang/lib/AST/Type.cpp
clang/lib/CodeGen/CGCUDANV.cpp
clang/lib/CodeGen/CGCUDARuntime.h
clang/lib/CodeGen/CGExprAgg.cpp
clang/lib/CodeGen/CodeGenModule.cpp
clang/lib/CodeGen/CodeGenTypes.cpp
clang/lib/CodeGen/TargetInfo.cpp
clang/lib/CodeGen/TargetInfo.h
clang/lib/Headers/__clang_cuda_runtime_wrapper.h
clang/lib/Sema/SemaDeclAttr.cpp
clang/lib/Sema/SemaDeclCXX.cpp
clang/test/Misc/pragma-attribute-supported-attributes-list.test
clang/test/SemaCUDA/attr-declspec.cu
clang/test/SemaCUDA/attributes-on-non-cuda.cu
clang/test/SemaCUDA/bad-attributes.cu
llvm/include/llvm/IR/Operator.h
Removed:
diff --git a/clang/include/clang/AST/Type.h b/clang/include/clang/AST/Type.h
index 3a2411b4ed29..6b46fc5ad312 100644
--- a/clang/include/clang/AST/Type.h
+++ b/clang/include/clang/AST/Type.h
@@ -2111,6 +2111,11 @@ class alignas(8) Type : public ExtQualsTypeCommonBase {
/// than implicitly __strong.
bool isObjCARCImplicitlyUnretainedType() const;
+ /// Check if the type is the CUDA device builtin surface type.
+ bool isCUDADeviceBuiltinSurfaceType() const;
+ /// Check if the type is the CUDA device builtin texture type.
+ bool isCUDADeviceBuiltinTextureType() const;
+
/// Return the implicit lifetime for this type, which must not be dependent.
Qualifiers::ObjCLifetime getObjCARCImplicitLifetime() const;
diff --git a/clang/include/clang/Basic/Attr.td
b/clang/include/clang/Basic/Attr.td
index 5a90b2be2cbf..96bfdd313f47 100644
--- a/clang/include/clang/Basic/Attr.td
+++ b/clang/include/clang/Basic/Attr.td
@@ -1064,16 +1064,20 @@ def CUDADeviceBuiltin : IgnoredAttr {
let LangOpts = [CUDA];
}
-def CUDADeviceBuiltinSurfaceType : IgnoredAttr {
+def CUDADeviceBuiltinSurfaceType : InheritableAttr {
let Spellings = [GNU<"device_builtin_surface_type">,
Declspec<"__device_builtin_surface_type__">];
let LangOpts = [CUDA];
+ let Subjects = SubjectList<[CXXRecord]>;
+ let Documentation = [CUDADeviceBuiltinSurfaceTypeDocs];
}
-def CUDADeviceBuiltinTextureType : IgnoredAttr {
+def CUDADeviceBuiltinTextureType : InheritableAttr {
let Spellings = [GNU<"device_builtin_texture_type">,
Declspec<"__device_builtin_texture_type__">];
let LangOpts = [CUDA];
+ let Subjects = SubjectList<[CXXRecord]>;
+ let Documentation = [CUDADeviceBuiltinTextureTypeDocs];
}
def CUDAGlobal : InheritableAttr {
diff --git a/clang/include/clang/Basic/AttrDocs.td
b/clang/include/clang/Basic/AttrDocs.td
index a1cf25ed3929..2c89dc6f4952 100644
--- a/clang/include/clang/Basic/AttrDocs.td
+++ b/clang/include/clang/Basic/AttrDocs.td
@@ -4624,6 +4624,28 @@ the initializer on host side.
}];
}
+def CUDADeviceBuiltinSurfaceTypeDocs : Documentation {
+ let Category = DocCatType;
+ let Content = [{
+The ``device_builtin_surface_type`` attribute can be applied to a class
+template when declaring the surface reference. A surface reference variable
+could be accessed on the host side and, on the device side, might be translated
+into an internal surface object, which is established through surface bind and
+unbind runtime APIs.
+ }];
+}
+
+def CUDADeviceBuiltinTextureTypeDocs : Documentation {
+ let Category = DocCatType;
+ let Content = [{
+The ``device_builtin_texture_type`` attribute can be applied to a class
+template when declaring the texture reference. A texture reference variable
+could be accessed on the host side and, on the device side, might be translated
+into an internal texture object, which is established through texture bind and
+unbind runtime APIs.
+ }];
+}
+
def LifetimeOwnerDocs : Documentation {
let Category = DocCatDecl;
let Content = [{
diff --git a/clang/include/clang/Basic/DiagnosticSemaKinds.td
b/clang/include/clang/Basic/DiagnosticSemaKinds.td
index 762dd1469236..c642c2ba36c8 100644
--- a/clang/include/clang/Basic/DiagnosticSemaKinds.td
+++ b/clang/include/clang/Basic/DiagnosticSemaKinds.td
@@ -7967,6 +7967,22 @@ def err_cuda_ovl_target : Error<
def
Re: [clang] d264f02 - Fix `-Wreturn-type` warning. NFC.
Thanks for catching that.
On Thu, Mar 26, 2020 at 11:14 PM David Blaikie wrote:
>
> Usually this sort of thing is addressed with llvm_unreachable, rather than a
> return statement that's not expected to be reached by any valid execution of
> LLVM (it'd require a carefully hand-crafted CPU kind to reach that return
> (since all the actual enumerators result in returns earlier, in the switch
> statement above), which probably isn't an intended code path?)
>
> I've made that change in 819e540208d5d62e7841d0dbdef3580eecc2c2b6
>
> On Wed, Mar 25, 2020 at 9:59 PM Michael Liao via cfe-commits
> wrote:
>>
>>
>> Author: Michael Liao
>> Date: 2020-03-26T00:53:24-04:00
>> New Revision: d264f02c6f502960e2bcdd332f250efc702d09f2
>>
>> URL:
>> https://github.com/llvm/llvm-project/commit/d264f02c6f502960e2bcdd332f250efc702d09f2
>> DIFF:
>> https://github.com/llvm/llvm-project/commit/d264f02c6f502960e2bcdd332f250efc702d09f2.diff
>>
>> LOG: Fix `-Wreturn-type` warning. NFC.
>>
>> Added:
>>
>>
>> Modified:
>> clang/lib/Basic/Targets/X86.cpp
>>
>> Removed:
>>
>>
>>
>>
>> diff --git a/clang/lib/Basic/Targets/X86.cpp
>> b/clang/lib/Basic/Targets/X86.cpp
>> index f35b520de657..8a7d0f17760e 100644
>> --- a/clang/lib/Basic/Targets/X86.cpp
>> +++ b/clang/lib/Basic/Targets/X86.cpp
>> @@ -1842,6 +1842,7 @@ Optional
>> X86TargetInfo::getCPUCacheLineSize() const {
>> case CK_Generic:
>>return None;
>>}
>> + return None;
>> }
>>
>> bool X86TargetInfo::validateOutputSize(const llvm::StringMap
>> ,
>>
>>
>>
>> ___
>> cfe-commits mailing list
>> cfe-commits@lists.llvm.org
>> https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] 6a9ad5f - [cuda][hip] Add CUDA builtin surface/texture reference support.
Author: Michael Liao
Date: 2020-03-26T14:44:52-04:00
New Revision: 6a9ad5f3f4ac66f0cae592e911f4baeb6ee5eca6
URL:
https://github.com/llvm/llvm-project/commit/6a9ad5f3f4ac66f0cae592e911f4baeb6ee5eca6
DIFF:
https://github.com/llvm/llvm-project/commit/6a9ad5f3f4ac66f0cae592e911f4baeb6ee5eca6.diff
LOG: [cuda][hip] Add CUDA builtin surface/texture reference support.
Summary:
- Even though the bindless surface/texture interfaces are promoted,
there are still code using surface/texture references. For example,
[PR#26400](https://bugs.llvm.org/show_bug.cgi?id=26400) reports the
compilation issue for code using `tex2D` with texture references. For
better compatibility, this patch proposes the support of
surface/texture references.
- Due to the absent documentation and magic headers, it's believed that
`nvcc` does use builtins for texture support. From the limited NVVM
documentation[^nvvm] and NVPTX backend texture/surface related
tests[^test], it's believed that surface/texture references are
supported by replacing their reference types, which are annotated with
`device_builtin_surface_type`/`device_builtin_texture_type`, with the
corresponding handle-like object types, `cudaSurfaceObject_t` or
`cudaTextureObject_t`, in the device-side compilation. On the host
side, that global handle variables are registered and will be
established and updated later when corresponding binding/unbinding
APIs are called[^bind]. Surface/texture references are most like
device global variables but represented in different types on the host
and device sides.
- In this patch, the following changes are proposed to support that
behavior:
+ Refine `device_builtin_surface_type` and
`device_builtin_texture_type` attributes to be applied on `Type`
decl only to check whether a variable is of the surface/texture
reference type.
+ Add hooks in code generation to replace that reference types with
the correponding object types as well as all accesses to them. In
particular, `nvvm.texsurf.handle.internal` should be used to load
object handles from global reference variables[^texsurf] as well as
metadata annotations.
+ Generate host-side registration with proper template argument
parsing.
---
[^nvvm]: https://docs.nvidia.com/cuda/pdf/NVVM_IR_Specification.pdf
[^test]:
https://raw.githubusercontent.com/llvm/llvm-project/master/llvm/test/CodeGen/NVPTX/tex-read-cuda.ll
[^bind]: See section 3.2.11.1.2 ``Texture reference API` in [CUDA C Programming
Guide](https://docs.nvidia.com/cuda/pdf/CUDA_C_Programming_Guide.pdf).
[^texsurf]: According to NVVM IR, `nvvm.texsurf.handle` should be used. But,
the current backend doesn't have that supported. We may revise that later.
Reviewers: tra, rjmccall, yaxunl, a.sidorin
Subscribers: cfe-commits
Tags: #clang
Differential Revision: https://reviews.llvm.org/D76365
Added:
clang/test/CodeGenCUDA/surface.cu
clang/test/CodeGenCUDA/texture.cu
Modified:
clang/include/clang/AST/Type.h
clang/include/clang/Basic/Attr.td
clang/include/clang/Basic/AttrDocs.td
clang/include/clang/Basic/DiagnosticSemaKinds.td
clang/lib/AST/Type.cpp
clang/lib/CodeGen/CGCUDANV.cpp
clang/lib/CodeGen/CGCUDARuntime.h
clang/lib/CodeGen/CGExprAgg.cpp
clang/lib/CodeGen/CodeGenModule.cpp
clang/lib/CodeGen/CodeGenTypes.cpp
clang/lib/CodeGen/TargetInfo.cpp
clang/lib/CodeGen/TargetInfo.h
clang/lib/Headers/__clang_cuda_runtime_wrapper.h
clang/lib/Sema/SemaDeclAttr.cpp
clang/lib/Sema/SemaDeclCXX.cpp
clang/test/Misc/pragma-attribute-supported-attributes-list.test
clang/test/SemaCUDA/attr-declspec.cu
clang/test/SemaCUDA/attributes-on-non-cuda.cu
clang/test/SemaCUDA/bad-attributes.cu
llvm/include/llvm/IR/Operator.h
Removed:
diff --git a/clang/include/clang/AST/Type.h b/clang/include/clang/AST/Type.h
index b8f49127bbd0..673d37109eb6 100644
--- a/clang/include/clang/AST/Type.h
+++ b/clang/include/clang/AST/Type.h
@@ -2111,6 +2111,11 @@ class alignas(8) Type : public ExtQualsTypeCommonBase {
/// than implicitly __strong.
bool isObjCARCImplicitlyUnretainedType() const;
+ /// Check if the type is the CUDA device builtin surface type.
+ bool isCUDADeviceBuiltinSurfaceType() const;
+ /// Check if the type is the CUDA device builtin texture type.
+ bool isCUDADeviceBuiltinTextureType() const;
+
/// Return the implicit lifetime for this type, which must not be dependent.
Qualifiers::ObjCLifetime getObjCARCImplicitLifetime() const;
diff --git a/clang/include/clang/Basic/Attr.td
b/clang/include/clang/Basic/Attr.td
index 5a90b2be2cbf..96bfdd313f47 100644
--- a/clang/include/clang/Basic/Attr.td
+++ b/clang/include/clang/Basic/Attr.td
@@ -1064,16 +1064,20 @@ def CUDADeviceBuiltin : IgnoredAttr {
let LangOpts = [CUDA];
}
-def CUDADeviceBuiltinSurfaceType : IgnoredAttr {
+def
[clang] d264f02 - Fix `-Wreturn-type` warning. NFC.
Author: Michael Liao
Date: 2020-03-26T00:53:24-04:00
New Revision: d264f02c6f502960e2bcdd332f250efc702d09f2
URL:
https://github.com/llvm/llvm-project/commit/d264f02c6f502960e2bcdd332f250efc702d09f2
DIFF:
https://github.com/llvm/llvm-project/commit/d264f02c6f502960e2bcdd332f250efc702d09f2.diff
LOG: Fix `-Wreturn-type` warning. NFC.
Added:
Modified:
clang/lib/Basic/Targets/X86.cpp
Removed:
diff --git a/clang/lib/Basic/Targets/X86.cpp b/clang/lib/Basic/Targets/X86.cpp
index f35b520de657..8a7d0f17760e 100644
--- a/clang/lib/Basic/Targets/X86.cpp
+++ b/clang/lib/Basic/Targets/X86.cpp
@@ -1842,6 +1842,7 @@ Optional X86TargetInfo::getCPUCacheLineSize()
const {
case CK_Generic:
return None;
}
+ return None;
}
bool X86TargetInfo::validateOutputSize(const llvm::StringMap ,
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] 4f4e687 - [test][clang][driver] Add required features.
Author: Michael Liao Date: 2020-03-24T17:08:21-04:00 New Revision: 4f4e68799fd55c7023e685161de6f6bb1ada16d5 URL: https://github.com/llvm/llvm-project/commit/4f4e68799fd55c7023e685161de6f6bb1ada16d5 DIFF: https://github.com/llvm/llvm-project/commit/4f4e68799fd55c7023e685161de6f6bb1ada16d5.diff LOG: [test][clang][driver] Add required features. - to avoid false alarms on builds without that features. Added: Modified: clang/test/Driver/save-temps.c Removed: diff --git a/clang/test/Driver/save-temps.c b/clang/test/Driver/save-temps.c index b0cfa4fd814a..a26ba9f4ec0d 100644 --- a/clang/test/Driver/save-temps.c +++ b/clang/test/Driver/save-temps.c @@ -1,3 +1,6 @@ +// REQUIRES: x86-registered-target +// REQUIRES: arm-registered-target + // RUN: %clang -target x86_64-apple-darwin -save-temps -arch x86_64 %s -### 2>&1 \ // RUN: | FileCheck %s // CHECK: "-o" "save-temps.i" ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] a4edea2 - Fix `-Wunused-variable` warning. NFC.
Author: Michael Liao Date: 2020-03-20T09:31:58-04:00 New Revision: a4edea29be2a77a8c8c237d75563a09a61791442 URL: https://github.com/llvm/llvm-project/commit/a4edea29be2a77a8c8c237d75563a09a61791442 DIFF: https://github.com/llvm/llvm-project/commit/a4edea29be2a77a8c8c237d75563a09a61791442.diff LOG: Fix `-Wunused-variable` warning. NFC. Added: Modified: clang/lib/Sema/SemaExpr.cpp Removed: diff --git a/clang/lib/Sema/SemaExpr.cpp b/clang/lib/Sema/SemaExpr.cpp index eaded8e92d7c..137aae883aa6 100644 --- a/clang/lib/Sema/SemaExpr.cpp +++ b/clang/lib/Sema/SemaExpr.cpp @@ -15490,8 +15490,7 @@ static void RemoveNestedImmediateInvocation( /// nowhere in the expression being transformed therefore will not be rebuilt. /// Setting AllowSkippingFirstCXXConstructExpr to false will prevent from /// skipping the first CXXConstructExpr. - if (auto *OldExpr = - dyn_cast(It->getPointer()->IgnoreImplicit())) + if (isa(It->getPointer()->IgnoreImplicit())) Transformer.AllowSkippingFirstCXXConstructExpr = false; ExprResult Res = Transformer.TransformExpr(It->getPointer()->getSubExpr()); ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] 4cf01ed - [hip] Revise `GlobalDecl` constructors. NFC.
Author: Michael Liao
Date: 2020-03-18T09:33:39-04:00
New Revision: 4cf01ed75e35e7bd3ef8ef1a2192c7f4656ab545
URL:
https://github.com/llvm/llvm-project/commit/4cf01ed75e35e7bd3ef8ef1a2192c7f4656ab545
DIFF:
https://github.com/llvm/llvm-project/commit/4cf01ed75e35e7bd3ef8ef1a2192c7f4656ab545.diff
LOG: [hip] Revise `GlobalDecl` constructors. NFC.
Summary:
- https://reviews.llvm.org/D68578 revises the `GlobalDecl` constructors
to ensure all GPU kernels have `ReferenceKenelKind` initialized
properly with an explicit constructor and static one. But, there are
lots of places using the implicit constructor triggering the assertion
on non-GPU kernels. That's found in compilation of many tests and
workloads.
- Fixing all of them may change more code and, more importantly, all of
them assumes the default kernel reference kind. This patch changes
that constructor to tell `CUDAGlobalAttr` and construct `GlobalDecl`
properly.
Reviewers: yaxunl
Subscribers: cfe-commits
Tags: #clang
Differential Revision: https://reviews.llvm.org/D76344
Added:
Modified:
clang/include/clang/AST/GlobalDecl.h
clang/lib/AST/Expr.cpp
clang/lib/AST/ItaniumMangle.cpp
clang/lib/AST/Mangle.cpp
clang/lib/CodeGen/CGDebugInfo.cpp
clang/lib/CodeGen/CGDecl.cpp
clang/lib/CodeGen/CGExpr.cpp
clang/lib/CodeGen/CodeGenModule.cpp
clang/lib/CodeGen/CodeGenModule.h
Removed:
diff --git a/clang/include/clang/AST/GlobalDecl.h
b/clang/include/clang/AST/GlobalDecl.h
index d2b5566a4cfa..bf30d9b94235 100644
--- a/clang/include/clang/AST/GlobalDecl.h
+++ b/clang/include/clang/AST/GlobalDecl.h
@@ -68,7 +68,15 @@ class GlobalDecl {
GlobalDecl(const VarDecl *D) { Init(D);}
GlobalDecl(const FunctionDecl *D, unsigned MVIndex = 0)
: MultiVersionIndex(MVIndex) {
-Init(D);
+if (!D->hasAttr()) {
+ Init(D);
+ return;
+}
+Value.setPointerAndInt(D, unsigned(getDefaultKernelReference(D)));
+ }
+ GlobalDecl(const FunctionDecl *D, KernelReferenceKind Kind)
+ : Value(D, unsigned(Kind)) {
+assert(D->hasAttr() && "Decl is not a GPU kernel!");
}
GlobalDecl(const NamedDecl *D) { Init(D); }
GlobalDecl(const BlockDecl *D) { Init(D); }
@@ -80,10 +88,6 @@ class GlobalDecl {
GlobalDecl(const CXXDestructorDecl *D, CXXDtorType Type) : Value(D, Type) {}
GlobalDecl(const VarDecl *D, DynamicInitKind StubKind)
: Value(D, unsigned(StubKind)) {}
- GlobalDecl(const FunctionDecl *D, KernelReferenceKind Kind)
- : Value(D, unsigned(Kind)) {
-assert(D->hasAttr() && "Decl is not a GPU kernel!");
- }
GlobalDecl getCanonicalDecl() const {
GlobalDecl CanonGD;
@@ -145,10 +149,10 @@ class GlobalDecl {
return GD;
}
- static GlobalDecl getDefaultKernelReference(const FunctionDecl *D) {
-return GlobalDecl(D, D->getASTContext().getLangOpts().CUDAIsDevice
- ? KernelReferenceKind::Kernel
- : KernelReferenceKind::Stub);
+ static KernelReferenceKind getDefaultKernelReference(const FunctionDecl *D) {
+return D->getASTContext().getLangOpts().CUDAIsDevice
+ ? KernelReferenceKind::Kernel
+ : KernelReferenceKind::Stub;
}
GlobalDecl getWithDecl(const Decl *D) {
diff --git a/clang/lib/AST/Expr.cpp b/clang/lib/AST/Expr.cpp
index 1eb56c30283c..6591b0481d4b 100644
--- a/clang/lib/AST/Expr.cpp
+++ b/clang/lib/AST/Expr.cpp
@@ -567,7 +567,7 @@ std::string PredefinedExpr::ComputeName(IdentKind IK, const
Decl *CurrentDecl) {
else if (const CXXDestructorDecl *DD = dyn_cast(ND))
GD = GlobalDecl(DD, Dtor_Base);
else if (ND->hasAttr())
- GD = GlobalDecl::getDefaultKernelReference(cast(ND));
+ GD = GlobalDecl(cast(ND));
else
GD = GlobalDecl(ND);
MC->mangleName(GD, Out);
diff --git a/clang/lib/AST/ItaniumMangle.cpp b/clang/lib/AST/ItaniumMangle.cpp
index 6deab2a95494..12e1fc589fdc 100644
--- a/clang/lib/AST/ItaniumMangle.cpp
+++ b/clang/lib/AST/ItaniumMangle.cpp
@@ -1575,14 +1575,8 @@ static GlobalDecl getParentOfLocalEntity(const
DeclContext *DC) {
GD = GlobalDecl(CD, Ctor_Complete);
else if (auto *DD = dyn_cast(DC))
GD = GlobalDecl(DD, Dtor_Complete);
- else {
-auto *FD = cast(DC);
-// Local variables can only exist in real kernels.
-if (FD->hasAttr())
- GD = GlobalDecl(FD, KernelReferenceKind::Kernel);
-else
- GD = GlobalDecl(FD);
- }
+ else
+GD = GlobalDecl(cast(DC));
return GD;
}
diff --git a/clang/lib/AST/Mangle.cpp b/clang/lib/AST/Mangle.cpp
index cc46994c1003..30078fcb243d 100644
--- a/clang/lib/AST/Mangle.cpp
+++ b/clang/lib/AST/Mangle.cpp
@@ -444,7 +444,7 @@ class ASTNameGenerator::Implementation {
else if (const auto *DtorD = dyn_cast(D))
GD = GlobalDecl(DtorD, Dtor_Complete);
else if (D->hasAttr())
-GD =
[clang] a2920c4 - [codegen] Fix one more case where `getGlobalDecl` should be used. NFC.
Author: Michael Liao
Date: 2020-03-17T17:56:47-04:00
New Revision: a2920c4ea9971cc38cbca3d6e10ccb10ab83a462
URL:
https://github.com/llvm/llvm-project/commit/a2920c4ea9971cc38cbca3d6e10ccb10ab83a462
DIFF:
https://github.com/llvm/llvm-project/commit/a2920c4ea9971cc38cbca3d6e10ccb10ab83a462.diff
LOG: [codegen] Fix one more case where `getGlobalDecl` should be used. NFC.
- After https://reviews.llvm.org/D68578, the implicit conversion from
`FunctionDecl` to `GlobalDecl` needs replacing with `getGlobalDecl`;
otherwise, assertion is triggered.
Added:
Modified:
clang/lib/CodeGen/CGDebugInfo.cpp
Removed:
diff --git a/clang/lib/CodeGen/CGDebugInfo.cpp
b/clang/lib/CodeGen/CGDebugInfo.cpp
index 94dab4c85614..a1ecddd3da7f 100644
--- a/clang/lib/CodeGen/CGDebugInfo.cpp
+++ b/clang/lib/CodeGen/CGDebugInfo.cpp
@@ -3833,7 +3833,8 @@ void CGDebugInfo::EmitFuncDeclForCallSite(llvm::CallBase
*CallOrInvoke,
// create the one describing the function in order to have complete
// call site debug info.
if (!CalleeDecl->isStatic() && !CalleeDecl->isInlined())
-EmitFunctionDecl(CalleeDecl, CalleeDecl->getLocation(), CalleeType, Func);
+EmitFunctionDecl(CGM.getGlobalDecl(CalleeDecl), CalleeDecl->getLocation(),
+ CalleeType, Func);
}
void CGDebugInfo::EmitInlineFunctionStart(CGBuilderTy , GlobalDecl GD)
{
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] 073dbaa - Fix GCC warnings. NFC.
Author: Michael Liao
Date: 2020-03-08T13:00:36-04:00
New Revision: 073dbaae39724ea860b5957fe47ecc1c2a84b197
URL:
https://github.com/llvm/llvm-project/commit/073dbaae39724ea860b5957fe47ecc1c2a84b197
DIFF:
https://github.com/llvm/llvm-project/commit/073dbaae39724ea860b5957fe47ecc1c2a84b197.diff
LOG: Fix GCC warnings. NFC.
Added:
Modified:
clang/lib/AST/ItaniumMangle.cpp
clang/lib/Index/USRGeneration.cpp
Removed:
diff --git a/clang/lib/AST/ItaniumMangle.cpp b/clang/lib/AST/ItaniumMangle.cpp
index 5cc66a0a5778..63e34653637e 100644
--- a/clang/lib/AST/ItaniumMangle.cpp
+++ b/clang/lib/AST/ItaniumMangle.cpp
@@ -641,7 +641,7 @@ void CXXNameMangler::mangle(GlobalDecl GD) {
//::=
//::=
Out << "_Z";
- if (const FunctionDecl *FD = dyn_cast(GD.getDecl()))
+ if (isa(GD.getDecl()))
mangleFunctionEncoding(GD);
else if (const VarDecl *VD = dyn_cast(GD.getDecl()))
mangleName(VD);
diff --git a/clang/lib/Index/USRGeneration.cpp
b/clang/lib/Index/USRGeneration.cpp
index f3eb653f10fa..7972d0a047c2 100644
--- a/clang/lib/Index/USRGeneration.cpp
+++ b/clang/lib/Index/USRGeneration.cpp
@@ -388,7 +388,7 @@ static const ObjCCategoryDecl *getCategoryContext(const
NamedDecl *D) {
if (auto *ICD = dyn_cast(D->getDeclContext()))
return ICD->getCategoryDecl();
return nullptr;
-};
+}
void USRGenerator::VisitObjCMethodDecl(const ObjCMethodDecl *D) {
const DeclContext *container = D->getDeclContext();
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] f6a3ac1 - Fix `-Wunused-variable` warning. NFC.
Author: Michael Liao
Date: 2020-02-12T12:45:14-05:00
New Revision: f6a3ac150b8d9f3458f526cf76ebcd545bfc1898
URL:
https://github.com/llvm/llvm-project/commit/f6a3ac150b8d9f3458f526cf76ebcd545bfc1898
DIFF:
https://github.com/llvm/llvm-project/commit/f6a3ac150b8d9f3458f526cf76ebcd545bfc1898.diff
LOG: Fix `-Wunused-variable` warning. NFC.
Added:
Modified:
clang/lib/CodeGen/TargetInfo.cpp
Removed:
diff --git a/clang/lib/CodeGen/TargetInfo.cpp
b/clang/lib/CodeGen/TargetInfo.cpp
index 905440febb51..e40f24d0ca4d 100644
--- a/clang/lib/CodeGen/TargetInfo.cpp
+++ b/clang/lib/CodeGen/TargetInfo.cpp
@@ -7578,7 +7578,7 @@ ABIArgInfo HexagonABIInfo::classifyReturnType(QualType
RetTy) const {
const TargetInfo = CGT.getTarget();
uint64_t Size = getContext().getTypeSize(RetTy);
- if (const auto *VecTy = RetTy->getAs()) {
+ if (RetTy->getAs()) {
// HVX vectors are returned in vector registers or register pairs.
if (T.hasFeature("hvx")) {
assert(T.hasFeature("hvx-length64b") || T.hasFeature("hvx-length128b"));
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] a067891 - [clang][codegen] Fix another lifetime emission on alloca on non-default address space.
Author: Michael Liao
Date: 2020-02-10T00:15:56-05:00
New Revision: a06789138987d1f64bb2f97d3a5c0f39eaf94715
URL:
https://github.com/llvm/llvm-project/commit/a06789138987d1f64bb2f97d3a5c0f39eaf94715
DIFF:
https://github.com/llvm/llvm-project/commit/a06789138987d1f64bb2f97d3a5c0f39eaf94715.diff
LOG: [clang][codegen] Fix another lifetime emission on alloca on non-default
address space.
- Lifetime intrinsics expect the pointer directly from alloca. Need
extra handling for targets with alloca on non-default (or non-zero)
address space.
Added:
clang/test/CodeGenCXX/amdgcn-call-with-aggarg.cpp
Modified:
clang/lib/CodeGen/CGCall.cpp
clang/lib/CodeGen/CodeGenFunction.h
Removed:
diff --git a/clang/lib/CodeGen/CGCall.cpp b/clang/lib/CodeGen/CGCall.cpp
index 3edcfb21ef34..9ef2a3b3d099 100644
--- a/clang/lib/CodeGen/CGCall.cpp
+++ b/clang/lib/CodeGen/CGCall.cpp
@@ -3690,8 +3690,9 @@ void CodeGenFunction::EmitCallArg(CallArgList ,
const Expr *E,
}
AggValueSlot ArgSlot = AggValueSlot::ignored();
+ Address ArgSlotAlloca = Address::invalid();
if (hasAggregateEvaluationKind(E->getType())) {
-ArgSlot = CreateAggTemp(E->getType(), "agg.tmp");
+ArgSlot = CreateAggTemp(E->getType(), "agg.tmp", );
// Emit a lifetime start/end for this temporary. If the type has a
// destructor, then we need to keep it alive. FIXME: We should still be
able
@@ -3699,8 +3700,9 @@ void CodeGenFunction::EmitCallArg(CallArgList ,
const Expr *E,
if (!E->getType().isDestructedType()) {
uint64_t size =
CGM.getDataLayout().getTypeAllocSize(ConvertTypeForMem(E->getType()));
- if (auto *lifetimeSize = EmitLifetimeStart(size, ArgSlot.getPointer()))
-args.addLifetimeCleanup({ArgSlot.getPointer(), lifetimeSize});
+ if (auto *lifetimeSize =
+ EmitLifetimeStart(size, ArgSlotAlloca.getPointer()))
+args.addLifetimeCleanup({ArgSlotAlloca.getPointer(), lifetimeSize});
}
}
diff --git a/clang/lib/CodeGen/CodeGenFunction.h
b/clang/lib/CodeGen/CodeGenFunction.h
index f48d8a4cc366..7ddd38c7b262 100644
--- a/clang/lib/CodeGen/CodeGenFunction.h
+++ b/clang/lib/CodeGen/CodeGenFunction.h
@@ -2264,8 +2264,9 @@ class CodeGenFunction : public CodeGenTypeCache {
/// CreateAggTemp - Create a temporary memory object for the given
/// aggregate type.
- AggValueSlot CreateAggTemp(QualType T, const Twine = "tmp") {
-return AggValueSlot::forAddr(CreateMemTemp(T, Name),
+ AggValueSlot CreateAggTemp(QualType T, const Twine = "tmp",
+ Address *Alloca = nullptr) {
+return AggValueSlot::forAddr(CreateMemTemp(T, Name, Alloca),
T.getQualifiers(),
AggValueSlot::IsNotDestructed,
AggValueSlot::DoesNotNeedGCBarriers,
diff --git a/clang/test/CodeGenCXX/amdgcn-call-with-aggarg.cpp
b/clang/test/CodeGenCXX/amdgcn-call-with-aggarg.cpp
new file mode 100644
index ..e9d3683cfaa2
--- /dev/null
+++ b/clang/test/CodeGenCXX/amdgcn-call-with-aggarg.cpp
@@ -0,0 +1,19 @@
+// RUN: %clang_cc1 -triple amdgcn-amd-amdhsa -emit-llvm -O3
-disable-llvm-passes -o - %s | FileCheck %s
+
+struct A {
+ float x, y, z, w;
+};
+
+void foo(A a);
+
+// CHECK-LABEL: @_Z4testv
+// CHECK: %[[lvar:.*]] = alloca %struct.A, align 4, addrspace(5)
+// CHECK: %[[atmp:.*]] = alloca %struct.A, align 4, addrspace(5)
+// CHECK: %[[lcst:.*]] = bitcast %struct.A addrspace(5)* %[[lvar]] to i8
addrspace(5)*
+// CHECK: call void @llvm.lifetime.start.p5i8(i64 16, i8 addrspace(5)*
%[[lcst]]
+// CHECK: %[[acst:.*]] = bitcast %struct.A addrspace(5)* %[[atmp]] to i8
addrspace(5)*
+// CHECK: call void @llvm.lifetime.start.p5i8(i64 16, i8 addrspace(5)*
%[[acst]]
+void test() {
+ A a;
+ foo(a);
+}
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] 2926917 - [clang] Fix linkage of nested lambdas.
Author: Michael Liao
Date: 2020-02-07T13:24:21-05:00
New Revision: 2926917f430d705f084813b63a40fafc61872524
URL:
https://github.com/llvm/llvm-project/commit/2926917f430d705f084813b63a40fafc61872524
DIFF:
https://github.com/llvm/llvm-project/commit/2926917f430d705f084813b63a40fafc61872524.diff
LOG: [clang] Fix linkage of nested lambdas.
patch from Philippe Daouadi
This is an attempt to fix
[PR#44368](https://bugs.llvm.org/show_bug.cgi?id=44368)
This effectively reverts [D1783](https://reviews.llvm.org/D1783). It
doesn't break the current tests and fixes the test that this commit
adds.
We now decide of a lambda linkage only depending on the visibility of
its parent context.
Differential Revision: https://reviews.llvm.org/D73701
Added:
Modified:
clang/lib/AST/Decl.cpp
clang/test/CodeGenCXX/lambda-expressions-nested-linkage.cpp
Removed:
diff --git a/clang/lib/AST/Decl.cpp b/clang/lib/AST/Decl.cpp
index 0d30f64b992e..216137bf74f9 100644
--- a/clang/lib/AST/Decl.cpp
+++ b/clang/lib/AST/Decl.cpp
@@ -1318,19 +1318,6 @@ LinkageInfo LinkageComputer::getLVForLocalDecl(const
NamedDecl *D,
LV.isVisibilityExplicit());
}
-static inline const CXXRecordDecl*
-getOutermostEnclosingLambda(const CXXRecordDecl *Record) {
- const CXXRecordDecl *Ret = Record;
- while (Record && Record->isLambda()) {
-Ret = Record;
-if (!Record->getParent()) break;
-// Get the Containing Class of this Lambda Class
-Record = dyn_cast_or_null(
- Record->getParent()->getParent());
- }
- return Ret;
-}
-
LinkageInfo LinkageComputer::computeLVForDecl(const NamedDecl *D,
LVComputationKind computation,
bool IgnoreVarTypeLinkage) {
@@ -1396,25 +1383,9 @@ LinkageInfo LinkageComputer::computeLVForDecl(const
NamedDecl *D,
return getInternalLinkageFor(D);
}
-// This lambda has its linkage/visibility determined:
-// - either by the outermost lambda if that lambda has no mangling
-//number.
-// - or by the parent of the outer most lambda
-// This prevents infinite recursion in settings such as nested lambdas
-// used in NSDMI's, for e.g.
-// struct L {
-//int t{};
-//int t2 = ([](int a) { return [](int b) { return b; };})(t)(t);
-// };
-const CXXRecordDecl *OuterMostLambda =
-getOutermostEnclosingLambda(Record);
-if (OuterMostLambda->hasKnownLambdaInternalLinkage() ||
-!OuterMostLambda->getLambdaManglingNumber())
- return getInternalLinkageFor(D);
-
return getLVForClosure(
- OuterMostLambda->getDeclContext()->getRedeclContext(),
- OuterMostLambda->getLambdaContextDecl(), computation);
+ Record->getDeclContext()->getRedeclContext(),
+ Record->getLambdaContextDecl(), computation);
}
break;
diff --git a/clang/test/CodeGenCXX/lambda-expressions-nested-linkage.cpp
b/clang/test/CodeGenCXX/lambda-expressions-nested-linkage.cpp
index 9a449874cb85..6b45645c9e2d 100644
--- a/clang/test/CodeGenCXX/lambda-expressions-nested-linkage.cpp
+++ b/clang/test/CodeGenCXX/lambda-expressions-nested-linkage.cpp
@@ -1,4 +1,5 @@
// RUN: %clang_cc1 -triple x86_64-apple-darwin10.0.0 -fblocks -emit-llvm -o -
%s -fexceptions -std=c++11 | FileCheck %s
+// RUN: %clang_cc1 -triple x86_64-apple-darwin10.0.0 -fblocks -emit-llvm -o -
%s -fexceptions -std=c++14 | FileCheck --check-prefixes=CHECK,CXX14 %s
// CHECK-LABEL: define void @_ZN19non_inline_function3fooEv()
// CHECK-LABEL: define internal void
@"_ZZN19non_inline_function3fooEvENK3$_0clEi"(%class.anon
@@ -51,3 +52,18 @@ inline int foo() {
}
int use = foo();
}
+
+#if __cplusplus >= 201402L
+// CXX14-LABEL: define internal void
@"_ZZZN32lambda_capture_in_generic_lambda3fooIiEEDavENKUlT_E_clIZNS_L1fEvE3$_1EEDaS1_ENKUlvE_clEv"
+namespace lambda_capture_in_generic_lambda {
+template auto foo() {
+ return [](auto func) {
+[func] { func(); }();
+ };
+}
+static void f() {
+ foo()([] { });
+}
+void f1() { f(); }
+}
+#endif
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] 6f5a159 - [clang][driver] Clean up unnecessary reference to TC. NFC.
Author: Michael Liao
Date: 2020-02-06T15:14:21-05:00
New Revision: 6f5a159eab8d3fecdbbc741a38c970c0149b3c96
URL:
https://github.com/llvm/llvm-project/commit/6f5a159eab8d3fecdbbc741a38c970c0149b3c96
DIFF:
https://github.com/llvm/llvm-project/commit/6f5a159eab8d3fecdbbc741a38c970c0149b3c96.diff
LOG: [clang][driver] Clean up unnecessary reference to TC. NFC.
Added:
Modified:
clang/lib/Driver/ToolChains/Clang.cpp
Removed:
diff --git a/clang/lib/Driver/ToolChains/Clang.cpp
b/clang/lib/Driver/ToolChains/Clang.cpp
index 37adad152c56..65039ac64b5a 100644
--- a/clang/lib/Driver/ToolChains/Clang.cpp
+++ b/clang/lib/Driver/ToolChains/Clang.cpp
@@ -307,10 +307,9 @@ static void getWebAssemblyTargetFeatures(const ArgList
,
handleTargetFeaturesGroup(Args, Features,
options::OPT_m_wasm_Features_Group);
}
-static void getTargetFeatures(const ToolChain , const llvm::Triple ,
+static void getTargetFeatures(const Driver , const llvm::Triple ,
const ArgList , ArgStringList ,
bool ForAS, bool IsAux = false) {
- const Driver = TC.getDriver();
std::vector Features;
switch (Triple.getArch()) {
default:
@@ -1594,7 +1593,7 @@ void Clang::RenderTargetOptions(const llvm::Triple
,
const ToolChain = getToolChain();
// Add the target features
- getTargetFeatures(TC, EffectiveTriple, Args, CmdArgs, false);
+ getTargetFeatures(TC.getDriver(), EffectiveTriple, Args, CmdArgs, false);
// Add target specific flags.
switch (TC.getArch()) {
@@ -4643,7 +4642,7 @@ void Clang::ConstructJob(Compilation , const JobAction
,
CmdArgs.push_back("-aux-target-cpu");
CmdArgs.push_back(Args.MakeArgString(HostCPU));
}
-getTargetFeatures(TC, *TC.getAuxTriple(), HostArgs, CmdArgs,
+getTargetFeatures(D, *TC.getAuxTriple(), HostArgs, CmdArgs,
/*ForAS*/ false, /*IsAux*/ true);
}
@@ -6679,7 +6678,7 @@ void ClangAs::ConstructJob(Compilation , const
JobAction ,
}
// Add the target features
- getTargetFeatures(getToolChain(), Triple, Args, CmdArgs, true);
+ getTargetFeatures(D, Triple, Args, CmdArgs, true);
// Ignore explicit -force_cpusubtype_ALL option.
(void)Args.hasArg(options::OPT_force__cpusubtype__ALL);
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] 318d0ed - Fix warning on unused variables. NFC.
Author: Michael Liao
Date: 2020-02-06T12:21:20-05:00
New Revision: 318d0ede572080f18d0106dbc354e11c88329a84
URL:
https://github.com/llvm/llvm-project/commit/318d0ede572080f18d0106dbc354e11c88329a84
DIFF:
https://github.com/llvm/llvm-project/commit/318d0ede572080f18d0106dbc354e11c88329a84.diff
LOG: Fix warning on unused variables. NFC.
Added:
Modified:
clang/lib/CodeGen/CGDebugInfo.cpp
Removed:
diff --git a/clang/lib/CodeGen/CGDebugInfo.cpp
b/clang/lib/CodeGen/CGDebugInfo.cpp
index 0e54e9419356..e171082942f6 100644
--- a/clang/lib/CodeGen/CGDebugInfo.cpp
+++ b/clang/lib/CodeGen/CGDebugInfo.cpp
@@ -3667,7 +3667,7 @@ void CGDebugInfo::EmitFunctionStart(GlobalDecl GD,
SourceLocation Loc,
} else {
Name = Fn->getName();
-if (const auto *BD = dyn_cast(D))
+if (isa(D))
LinkageName = Name;
Flags |= llvm::DINode::FlagPrototyped;
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] 09a8812 - [clang][driver][ARM] Clean up ARM target & feature checking in clang driver.
Author: Michael Liao
Date: 2020-02-06T08:57:52-05:00
New Revision: 09a88120c9269a9af0d80bc59afb2cb5806140ff
URL:
https://github.com/llvm/llvm-project/commit/09a88120c9269a9af0d80bc59afb2cb5806140ff
DIFF:
https://github.com/llvm/llvm-project/commit/09a88120c9269a9af0d80bc59afb2cb5806140ff.diff
LOG: [clang][driver][ARM] Clean up ARM target & feature checking in clang
driver.
Summary:
- Similar to other targets, instead of passing a toolchain, a driver
argument should be passed into `arm::getARMTargetFeatures`. Aslo, that
routine should honor the specified triple. Refactor
`arm::getARMFloatABI` with 2 separate interfaces. One has the original
parameters and the other uses the driver and the specified triple.
- That fixes an issue when target & features are queried during the
offload compilation, where the specified triple should be checked
instead of a effective triple. A previously failed test is re-enabled.
Subscribers: kristof.beyls, cfe-commits
Tags: #clang
Differential Revision: https://reviews.llvm.org/D74020
Added:
Modified:
clang/lib/Driver/ToolChains/Arch/ARM.cpp
clang/lib/Driver/ToolChains/Arch/ARM.h
clang/lib/Driver/ToolChains/Clang.cpp
clang/test/Driver/cuda-simple.cu
Removed:
diff --git a/clang/lib/Driver/ToolChains/Arch/ARM.cpp
b/clang/lib/Driver/ToolChains/Arch/ARM.cpp
index ce3990038a4b..18bd1317fbc2 100644
--- a/clang/lib/Driver/ToolChains/Arch/ARM.cpp
+++ b/clang/lib/Driver/ToolChains/Arch/ARM.cpp
@@ -137,9 +137,8 @@ bool arm::useAAPCSForMachO(const llvm::Triple ) {
}
// Select mode for reading thread pointer (-mtp=soft/cp15).
-arm::ReadTPMode arm::getReadTPMode(const ToolChain , const ArgList ) {
+arm::ReadTPMode arm::getReadTPMode(const Driver , const ArgList ) {
if (Arg *A = Args.getLastArg(options::OPT_mtp_mode_EQ)) {
-const Driver = TC.getDriver();
arm::ReadTPMode ThreadPointer =
llvm::StringSwitch(A->getValue())
.Case("cp15", ReadTPMode::Cp15)
@@ -156,11 +155,14 @@ arm::ReadTPMode arm::getReadTPMode(const ToolChain ,
const ArgList ) {
return ReadTPMode::Soft;
}
+arm::FloatABI arm::getARMFloatABI(const ToolChain , const ArgList ) {
+ return arm::getARMFloatABI(TC.getDriver(), TC.getEffectiveTriple(), Args);
+}
+
// Select the float ABI as determined by -msoft-float, -mhard-float, and
// -mfloat-abi=.
-arm::FloatABI arm::getARMFloatABI(const ToolChain , const ArgList ) {
- const Driver = TC.getDriver();
- const llvm::Triple = TC.getEffectiveTriple();
+arm::FloatABI arm::getARMFloatABI(const Driver , const llvm::Triple ,
+ const ArgList ) {
auto SubArch = getARMSubArchVersionNumber(Triple);
arm::FloatABI ABI = FloatABI::Invalid;
if (Arg *A =
@@ -276,18 +278,13 @@ arm::FloatABI arm::getARMFloatABI(const ToolChain ,
const ArgList ) {
return ABI;
}
-void arm::getARMTargetFeatures(const ToolChain ,
- const llvm::Triple ,
- const ArgList ,
- ArgStringList ,
- std::vector ,
- bool ForAS) {
- const Driver = TC.getDriver();
-
+void arm::getARMTargetFeatures(const Driver , const llvm::Triple ,
+ const ArgList , ArgStringList ,
+ std::vector , bool ForAS) {
bool KernelOrKext =
Args.hasArg(options::OPT_mkernel, options::OPT_fapple_kext);
- arm::FloatABI ABI = arm::getARMFloatABI(TC, Args);
- arm::ReadTPMode ThreadPointer = arm::getReadTPMode(TC, Args);
+ arm::FloatABI ABI = arm::getARMFloatABI(D, Triple, Args);
+ arm::ReadTPMode ThreadPointer = arm::getReadTPMode(D, Args);
const Arg *WaCPU = nullptr, *WaFPU = nullptr;
const Arg *WaHDiv = nullptr, *WaArch = nullptr;
diff --git a/clang/lib/Driver/ToolChains/Arch/ARM.h
b/clang/lib/Driver/ToolChains/Arch/ARM.h
index 5640f8371262..0ba1a59852aa 100644
--- a/clang/lib/Driver/ToolChains/Arch/ARM.h
+++ b/clang/lib/Driver/ToolChains/Arch/ARM.h
@@ -48,13 +48,15 @@ enum class FloatABI {
};
FloatABI getARMFloatABI(const ToolChain , const llvm::opt::ArgList );
-ReadTPMode getReadTPMode(const ToolChain , const llvm::opt::ArgList );
+FloatABI getARMFloatABI(const Driver , const llvm::Triple ,
+const llvm::opt::ArgList );
+ReadTPMode getReadTPMode(const Driver , const llvm::opt::ArgList );
bool useAAPCSForMachO(const llvm::Triple );
void getARMArchCPUFromArgs(const llvm::opt::ArgList ,
llvm::StringRef , llvm::StringRef ,
bool FromAs = false);
-void getARMTargetFeatures(const ToolChain , const llvm::Triple ,
+void getARMTargetFeatures(const Driver , const llvm::Triple ,
const llvm::opt::ArgList ,
llvm::opt::ArgStringList ,
[clang] b642e03 - [cuda][hip] Temporarily XFAIL on arm
Author: Michael Liao Date: 2020-02-04T20:25:12-05:00 New Revision: b642e0348512a83505900ae00844f5f60ebeac45 URL: https://github.com/llvm/llvm-project/commit/b642e0348512a83505900ae00844f5f60ebeac45 DIFF: https://github.com/llvm/llvm-project/commit/b642e0348512a83505900ae00844f5f60ebeac45.diff LOG: [cuda][hip] Temporarily XFAIL on arm Added: Modified: clang/test/Driver/cuda-simple.cu Removed: diff --git a/clang/test/Driver/cuda-simple.cu b/clang/test/Driver/cuda-simple.cu index 54e18403108b..d2daf88a68de 100644 --- a/clang/test/Driver/cuda-simple.cu +++ b/clang/test/Driver/cuda-simple.cu @@ -1,6 +1,7 @@ // Verify that we can parse a simple CUDA file with or without -save-temps // http://llvm.org/PR22936 // RUN: %clang -nocudainc -nocudalib -Werror -fsyntax-only -c %s +// XFAIL: arm // // Verify that we pass -x cuda-cpp-output to compiler after // preprocessing a CUDA file ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] ccac6b2 - [hip] Properly populate macros based on host processor.
Author: Michael Liao
Date: 2020-02-04T15:36:14-05:00
New Revision: ccac6b2bf877337a883c3763e41a529d8f9cc1ff
URL:
https://github.com/llvm/llvm-project/commit/ccac6b2bf877337a883c3763e41a529d8f9cc1ff
DIFF:
https://github.com/llvm/llvm-project/commit/ccac6b2bf877337a883c3763e41a529d8f9cc1ff.diff
LOG: [hip] Properly populate macros based on host processor.
Summary:
- The device compilation needs to have a consistent source code compared
to the corresponding host compilation. If macros based on the
host-specific target processor is not properly populated, the device
compilation may fail due to the inconsistent source after the
preprocessor. So far, only the host triple is used to build the
macros. If a detailed host CPU target or certain features are
specified, macros derived from them won't be populated properly, e.g.
`__SSE3__` won't be added unless `+sse3` feature is present. On
Windows compilation compatible with MSVC, that missing macros result
in that intrinsics are not included and cause device compilation
failure on the host-side source.
- This patch addresses this issue by introducing two `cc1` options,
i.e., `-aux-target-cpu` and `-aux-target-feature`. If a specific host
CPU target or certain features are specified, the compiler driver will
append them during the construction of the offline compilation
actions. Then, the toolchain in `cc1` phase will populate macros
accordingly.
- An internal option `--gpu-use-aux-triple-only` is added to fall back
the original behavior to help diagnosing potential issues from the new
behavior.
Reviewers: tra, yaxunl
Subscribers: cfe-commits
Tags: #clang
Differential Revision: https://reviews.llvm.org/D73942
Added:
clang/test/Driver/hip-host-cpu-features.hip
clang/test/Preprocessor/hip-host-cpu-macros.cu
Modified:
clang/include/clang/Driver/CC1Options.td
clang/include/clang/Driver/Options.td
clang/include/clang/Frontend/FrontendOptions.h
clang/lib/Driver/ToolChains/Clang.cpp
clang/lib/Frontend/CompilerInstance.cpp
clang/lib/Frontend/CompilerInvocation.cpp
Removed:
diff --git a/clang/include/clang/Driver/CC1Options.td
b/clang/include/clang/Driver/CC1Options.td
index f535d86d9b5e..0d0b05f8961c 100644
--- a/clang/include/clang/Driver/CC1Options.td
+++ b/clang/include/clang/Driver/CC1Options.td
@@ -482,6 +482,10 @@ def cc1as : Flag<["-"], "cc1as">;
def ast_merge : Separate<["-"], "ast-merge">,
MetaVarName<"">,
HelpText<"Merge the given AST file into the translation unit being
compiled.">;
+def aux_target_cpu : Separate<["-"], "aux-target-cpu">,
+ HelpText<"Target a specific auxiliary cpu type">;
+def aux_target_feature : Separate<["-"], "aux-target-feature">,
+ HelpText<"Target specific auxiliary attributes">;
def aux_triple : Separate<["-"], "aux-triple">,
HelpText<"Auxiliary target triple.">;
def code_completion_at : Separate<["-"], "code-completion-at">,
diff --git a/clang/include/clang/Driver/Options.td
b/clang/include/clang/Driver/Options.td
index 388ff094ae44..2c925d018da7 100644
--- a/clang/include/clang/Driver/Options.td
+++ b/clang/include/clang/Driver/Options.td
@@ -549,6 +549,9 @@ def c : Flag<["-"], "c">, Flags<[DriverOption]>,
Group,
def fconvergent_functions : Flag<["-"], "fconvergent-functions">,
Group, Flags<[CC1Option]>,
HelpText<"Assume functions may be convergent">;
+def gpu_use_aux_triple_only : Flag<["--"], "gpu-use-aux-triple-only">,
+ InternalDriverOpt, HelpText<"Prepare '-aux-triple' only without populating "
+ "'-aux-target-cpu' and '-aux-target-feature'.">;
def cuda_device_only : Flag<["--"], "cuda-device-only">,
HelpText<"Compile CUDA code for device only">;
def cuda_host_only : Flag<["--"], "cuda-host-only">,
diff --git a/clang/include/clang/Frontend/FrontendOptions.h
b/clang/include/clang/Frontend/FrontendOptions.h
index 09969b596d63..2adc6319810c 100644
--- a/clang/include/clang/Frontend/FrontendOptions.h
+++ b/clang/include/clang/Frontend/FrontendOptions.h
@@ -426,9 +426,15 @@ class FrontendOptions {
/// (in the format produced by -fdump-record-layouts).
std::string OverrideRecordLayoutsFile;
- /// Auxiliary triple for CUDA compilation.
+ /// Auxiliary triple for CUDA/HIP compilation.
std::string AuxTriple;
+ /// Auxiliary target CPU for CUDA/HIP compilation.
+ Optional AuxTargetCPU;
+
+ /// Auxiliary target features for CUDA/HIP compilation.
+ Optional> AuxTargetFeatures;
+
/// Filename to write statistics to.
std::string StatsFile;
diff --git a/clang/lib/Driver/ToolChains/Clang.cpp
b/clang/lib/Driver/ToolChains/Clang.cpp
index 6f092ca274c0..ccdfbe8c604f 100644
--- a/clang/lib/Driver/ToolChains/Clang.cpp
+++ b/clang/lib/Driver/ToolChains/Clang.cpp
@@ -309,7 +309,7 @@ static void getWebAssemblyTargetFeatures(const ArgList
,
static void getTargetFeatures(const ToolChain ,
[clang] 268e57b - [clang][driver] Remove an unused parameter. NFC.
Author: Michael Liao
Date: 2020-02-01T16:18:05-05:00
New Revision: 268e57bd35d7e05928820ad90f325e19e7a809d0
URL:
https://github.com/llvm/llvm-project/commit/268e57bd35d7e05928820ad90f325e19e7a809d0
DIFF:
https://github.com/llvm/llvm-project/commit/268e57bd35d7e05928820ad90f325e19e7a809d0.diff
LOG: [clang][driver] Remove an unused parameter. NFC.
- Group relevant code together.
Added:
Modified:
clang/lib/Driver/ToolChains/Clang.cpp
Removed:
diff --git a/clang/lib/Driver/ToolChains/Clang.cpp
b/clang/lib/Driver/ToolChains/Clang.cpp
index 510dc19f7a90..c8195f0f4ccf 100644
--- a/clang/lib/Driver/ToolChains/Clang.cpp
+++ b/clang/lib/Driver/ToolChains/Clang.cpp
@@ -3607,8 +3607,7 @@ static DwarfFissionKind getDebugFissionKind(const Driver
,
static void RenderDebugOptions(const ToolChain , const Driver ,
const llvm::Triple , const ArgList ,
- bool EmitCodeView, bool IsWindowsMSVC,
- ArgStringList ,
+ bool EmitCodeView, ArgStringList ,
codegenoptions::DebugInfoKind ,
DwarfFissionKind ) {
if (Args.hasFlag(options::OPT_fdebug_info_for_profiling,
@@ -4651,8 +4650,8 @@ void Clang::ConstructJob(Compilation , const JobAction
,
AddClangCLArgs(Args, InputType, CmdArgs, , );
DwarfFissionKind DwarfFission;
- RenderDebugOptions(TC, D, RawTriple, Args, EmitCodeView, IsWindowsMSVC,
- CmdArgs, DebugInfoKind, DwarfFission);
+ RenderDebugOptions(TC, D, RawTriple, Args, EmitCodeView, CmdArgs,
+ DebugInfoKind, DwarfFission);
// Add the split debug info name to the command lines here so we
// can propagate it to the backend.
@@ -5352,16 +5351,16 @@ void Clang::ConstructJob(Compilation , const
JobAction ,
RawTriple.isOSDarwin() && !KernelOrKext))
CmdArgs.push_back("-fregister-global-dtors-with-atexit");
- // -fms-extensions=0 is default.
- if (Args.hasFlag(options::OPT_fms_extensions, options::OPT_fno_ms_extensions,
- IsWindowsMSVC))
-CmdArgs.push_back("-fms-extensions");
-
// -fno-use-line-directives is default.
if (Args.hasFlag(options::OPT_fuse_line_directives,
options::OPT_fno_use_line_directives, false))
CmdArgs.push_back("-fuse-line-directives");
+ // -fms-extensions=0 is default.
+ if (Args.hasFlag(options::OPT_fms_extensions, options::OPT_fno_ms_extensions,
+ IsWindowsMSVC))
+CmdArgs.push_back("-fms-extensions");
+
// -fms-compatibility=0 is default.
bool IsMSVCCompat = Args.hasFlag(
options::OPT_fms_compatibility, options::OPT_fno_ms_compatibility,
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] 49f7bc9 - [hip] Remove `-Werror=format-nonliteral`
Author: Michael Liao
Date: 2020-01-23T11:02:11-05:00
New Revision: 49f7bc9e1e50eb8f6e065f97585b3bf0bcc23d5c
URL:
https://github.com/llvm/llvm-project/commit/49f7bc9e1e50eb8f6e065f97585b3bf0bcc23d5c
DIFF:
https://github.com/llvm/llvm-project/commit/49f7bc9e1e50eb8f6e065f97585b3bf0bcc23d5c.diff
LOG: [hip] Remove `-Werror=format-nonliteral`
Summary:
- It won't distinguish host and device code and trigger compilation
failure on irrelevant code.
Reviewers: sameerds, yaxunl
Subscribers: cfe-commits
Tags: #clang
Differential Revision: https://reviews.llvm.org/D73224
Added:
Modified:
clang/lib/Driver/ToolChains/HIP.cpp
clang/test/Driver/hip-printf.hip
Removed:
diff --git a/clang/lib/Driver/ToolChains/HIP.cpp
b/clang/lib/Driver/ToolChains/HIP.cpp
index 4772a6fb6b10..7039ddeabd57 100644
--- a/clang/lib/Driver/ToolChains/HIP.cpp
+++ b/clang/lib/Driver/ToolChains/HIP.cpp
@@ -425,7 +425,6 @@ Tool *HIPToolChain::buildLinker() const {
void HIPToolChain::addClangWarningOptions(ArgStringList ) const {
HostTC.addClangWarningOptions(CC1Args);
- CC1Args.push_back("-Werror=format-nonliteral");
}
ToolChain::CXXStdlibType
diff --git a/clang/test/Driver/hip-printf.hip
b/clang/test/Driver/hip-printf.hip
index 2df344f8fb2e..ada6c651ddb7 100644
--- a/clang/test/Driver/hip-printf.hip
+++ b/clang/test/Driver/hip-printf.hip
@@ -6,4 +6,4 @@
// RUN: %s 2>&1 | FileCheck %s
// CHECK: [[CLANG:".*clang.*"]] "-cc1"
-// CHECK-SAME: "-Werror=format-nonliteral"
+// CHECK-NOT: "-Werror=format-nonliteral"
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] 70b53a3 - Fix gcc `-Wunused-variable` warning. NFC.
Author: Michael Liao
Date: 2020-01-19T12:24:21-05:00
New Revision: 70b53a301888fe2be36996b41a7dd5aa7c256dc9
URL:
https://github.com/llvm/llvm-project/commit/70b53a301888fe2be36996b41a7dd5aa7c256dc9
DIFF:
https://github.com/llvm/llvm-project/commit/70b53a301888fe2be36996b41a7dd5aa7c256dc9.diff
LOG: Fix gcc `-Wunused-variable` warning. NFC.
Added:
Modified:
clang/lib/Serialization/ASTReaderStmt.cpp
Removed:
diff --git a/clang/lib/Serialization/ASTReaderStmt.cpp
b/clang/lib/Serialization/ASTReaderStmt.cpp
index 5dd0ef9d43c3..99dc1e9172c4 100644
--- a/clang/lib/Serialization/ASTReaderStmt.cpp
+++ b/clang/lib/Serialization/ASTReaderStmt.cpp
@@ -732,7 +732,7 @@ readConstraintSatisfaction(ASTRecordReader ) {
unsigned NumDetailRecords = Record.readInt();
for (unsigned i = 0; i != NumDetailRecords; ++i) {
Expr *ConstraintExpr = Record.readExpr();
- if (bool IsDiagnostic = Record.readInt()) {
+ if (/* IsDiagnostic */Record.readInt()) {
SourceLocation DiagLocation = Record.readSourceLocation();
std::string DiagMessage = Record.readString();
Satisfaction.Details.emplace_back(
@@ -822,7 +822,7 @@ void ASTStmtReader::VisitRequiresExpr(RequiresExpr *E) {
Req.emplace();
} else {
NoexceptLoc = Record.readSourceLocation();
- switch (auto returnTypeRequirementKind = Record.readInt()) {
+ switch (/* returnTypeRequirementKind */Record.readInt()) {
case 0:
// No return type requirement.
Req.emplace();
@@ -853,7 +853,7 @@ void ASTStmtReader::VisitRequiresExpr(RequiresExpr *E) {
std::move(*Req));
} break;
case concepts::Requirement::RK_Nested: {
-if (bool IsSubstitutionDiagnostic = Record.readInt()) {
+if (/* IsSubstitutionDiagnostic */Record.readInt()) {
R = new (Record.getContext()) concepts::NestedRequirement(
readSubstitutionDiagnostic(Record));
break;
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang-tools-extra] 40514a7 - [clangd] Add workaround for GCC5 host compilers. NFC.
Author: Michael Liao
Date: 2020-01-16T16:05:22-05:00
New Revision: 40514a7d7a3b745ba43c2d014e54a0d78d65d957
URL:
https://github.com/llvm/llvm-project/commit/40514a7d7a3b745ba43c2d014e54a0d78d65d957
DIFF:
https://github.com/llvm/llvm-project/commit/40514a7d7a3b745ba43c2d014e54a0d78d65d957.diff
LOG: [clangd] Add workaround for GCC5 host compilers. NFC.
Added:
Modified:
clang-tools-extra/clangd/Hover.cpp
Removed:
diff --git a/clang-tools-extra/clangd/Hover.cpp
b/clang-tools-extra/clangd/Hover.cpp
index cfa5e3bf93fb..ad715db4d5eb 100644
--- a/clang-tools-extra/clangd/Hover.cpp
+++ b/clang-tools-extra/clangd/Hover.cpp
@@ -439,7 +439,13 @@ bool isLiteral(const Expr *E) {
llvm::StringLiteral getNameForExpr(const Expr *E) {
// FIXME: Come up with names for `special` expressions.
- return "expression";
+ //
+ // It's an known issue for GCC5, https://godbolt.org/z/Z_tbgi. Work around
+ // that by using explicit conversion constructor.
+ //
+ // TODO: Once GCC5 is fully retired and not the minimal requirement as stated
+ // in `GettingStarted`, please remove the explicit conversion constructor.
+ return llvm::StringLiteral("expression");
}
// Generates hover info for evaluatable expressions.
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] a3490e3 - Remove trailing `;`. NFC.
Author: Michael Liao
Date: 2020-01-14T16:52:20-05:00
New Revision: a3490e3e3d38d502179329f76138d96c5b2bab88
URL:
https://github.com/llvm/llvm-project/commit/a3490e3e3d38d502179329f76138d96c5b2bab88
DIFF:
https://github.com/llvm/llvm-project/commit/a3490e3e3d38d502179329f76138d96c5b2bab88.diff
LOG: Remove trailing `;`. NFC.
Added:
Modified:
clang/lib/Tooling/Syntax/Tree.cpp
Removed:
diff --git a/clang/lib/Tooling/Syntax/Tree.cpp
b/clang/lib/Tooling/Syntax/Tree.cpp
index 9f028c0be3b4..9a6270ec4cce 100644
--- a/clang/lib/Tooling/Syntax/Tree.cpp
+++ b/clang/lib/Tooling/Syntax/Tree.cpp
@@ -29,7 +29,7 @@ static void traverse(syntax::Node *N,
traverse(static_cast(N), [&](const syntax::Node *N) {
Visit(const_cast(N));
});
-};
+}
} // namespace
syntax::Arena::Arena(SourceManager , const LangOptions ,
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] 7cee288 - Fix `-Wunused-variable` warning. NFC.
Author: Michael Liao Date: 2019-12-21T11:10:35-05:00 New Revision: 7cee28858674d233903e92b7a0c49b07b05ed3d3 URL: https://github.com/llvm/llvm-project/commit/7cee28858674d233903e92b7a0c49b07b05ed3d3 DIFF: https://github.com/llvm/llvm-project/commit/7cee28858674d233903e92b7a0c49b07b05ed3d3.diff LOG: Fix `-Wunused-variable` warning. NFC. Added: Modified: clang/lib/Sema/SemaDeclObjC.cpp Removed: diff --git a/clang/lib/Sema/SemaDeclObjC.cpp b/clang/lib/Sema/SemaDeclObjC.cpp index 4fdddfbb7a7e..5fdf6aeed5b4 100644 --- a/clang/lib/Sema/SemaDeclObjC.cpp +++ b/clang/lib/Sema/SemaDeclObjC.cpp @@ -4776,7 +4776,7 @@ Decl *Sema::ActOnMethodDeclaration( if (auto *Cat = dyn_cast(IMD->getDeclContext())) decl = Cat->IsClassExtension() ? 1 : 2; -if (auto *Cat = dyn_cast(ImpDecl)) +if (isa(ImpDecl)) impl = 1 + (decl != 0); Diag(ObjCMethod->getLocation(), ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] 6626e5a - Fix compilation warning from GCC7. NFC.
Author: Michael Liao
Date: 2019-12-09T10:11:27-05:00
New Revision: 6626e5a06a99b29b388f2dffde2c16f8eb5ded46
URL:
https://github.com/llvm/llvm-project/commit/6626e5a06a99b29b388f2dffde2c16f8eb5ded46
DIFF:
https://github.com/llvm/llvm-project/commit/6626e5a06a99b29b388f2dffde2c16f8eb5ded46.diff
LOG: Fix compilation warning from GCC7. NFC.
Added:
Modified:
clang/lib/Sema/SemaDeclCXX.cpp
Removed:
diff --git a/clang/lib/Sema/SemaDeclCXX.cpp b/clang/lib/Sema/SemaDeclCXX.cpp
index c8b95983f03c..d0857a5de817 100644
--- a/clang/lib/Sema/SemaDeclCXX.cpp
+++ b/clang/lib/Sema/SemaDeclCXX.cpp
@@ -7103,6 +7103,7 @@ class DefaultedComparisonVisitor {
ResultList Results;
switch (DCK) {
+default:
case DefaultedComparisonKind::None:
llvm_unreachable("not a defaulted comparison");
@@ -7592,6 +7593,7 @@ class DefaultedComparisonSynthesizer
return StmtError();
switch (DCK) {
+default:
case DefaultedComparisonKind::None:
llvm_unreachable("not a defaulted comparison");
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] f2ace9d - Add `QualType::hasAddressSpace`. NFC.
Author: Michael Liao
Date: 2019-12-06T13:08:55-05:00
New Revision: f2ace9d6005b4ffc6f6fc068c1aac897d871df7a
URL:
https://github.com/llvm/llvm-project/commit/f2ace9d6005b4ffc6f6fc068c1aac897d871df7a
DIFF:
https://github.com/llvm/llvm-project/commit/f2ace9d6005b4ffc6f6fc068c1aac897d871df7a.diff
LOG: Add `QualType::hasAddressSpace`. NFC.
- Add that as a shorthand of .getQualifiers().hasAddressSpace().
- Simplify related code.
Added:
Modified:
clang/include/clang/AST/Type.h
clang/lib/Sema/SemaDecl.cpp
clang/lib/Sema/SemaExpr.cpp
clang/lib/Sema/SemaInit.cpp
clang/lib/Sema/SemaOverload.cpp
clang/lib/Sema/SemaType.cpp
clang/lib/StaticAnalyzer/Checkers/DereferenceChecker.cpp
Removed:
diff --git a/clang/include/clang/AST/Type.h b/clang/include/clang/AST/Type.h
index 02c9aa403b5a..caf2a3dd79a3 100644
--- a/clang/include/clang/AST/Type.h
+++ b/clang/include/clang/AST/Type.h
@@ -1046,6 +1046,9 @@ class QualType {
ID.AddPointer(getAsOpaquePtr());
}
+ /// Check if this type has any address space qualifier.
+ inline bool hasAddressSpace() const;
+
/// Return the address space of this type.
inline LangAS getAddressSpace() const;
@@ -6276,6 +6279,11 @@ inline void QualType::removeLocalCVRQualifiers(unsigned
Mask) {
removeLocalFastQualifiers(Mask);
}
+/// Check if this type has any address space qualifier.
+inline bool QualType::hasAddressSpace() const {
+ return getQualifiers().hasAddressSpace();
+}
+
/// Return the address space of this type.
inline LangAS QualType::getAddressSpace() const {
return getQualifiers().getAddressSpace();
diff --git a/clang/lib/Sema/SemaDecl.cpp b/clang/lib/Sema/SemaDecl.cpp
index 660be458a698..0e38d6bfaf93 100644
--- a/clang/lib/Sema/SemaDecl.cpp
+++ b/clang/lib/Sema/SemaDecl.cpp
@@ -6118,7 +6118,7 @@ bool Sema::inferObjCARCLifetime(ValueDecl *decl) {
}
void Sema::deduceOpenCLAddressSpace(ValueDecl *Decl) {
- if (Decl->getType().getQualifiers().hasAddressSpace())
+ if (Decl->getType().hasAddressSpace())
return;
if (VarDecl *Var = dyn_cast(Decl)) {
QualType Type = Var->getType();
@@ -6132,7 +6132,7 @@ void Sema::deduceOpenCLAddressSpace(ValueDecl *Decl) {
// type has no address space yet, deduce it now.
if (auto DT = dyn_cast(Type)) {
auto OrigTy = DT->getOriginalType();
- if (!OrigTy.getQualifiers().hasAddressSpace() && OrigTy->isArrayType()) {
+ if (!OrigTy.hasAddressSpace() && OrigTy->isArrayType()) {
// Add the address space to the original array type and then propagate
// that to the element type through `getAsArrayType`.
OrigTy = Context.getAddrSpaceQualType(OrigTy, ImplAS);
@@ -16094,7 +16094,7 @@ FieldDecl *Sema::CheckFieldDecl(DeclarationName Name,
QualType T,
}
// TR 18037 does not allow fields to be declared with address space
- if (T.getQualifiers().hasAddressSpace() || T->isDependentAddressSpaceType()
||
+ if (T.hasAddressSpace() || T->isDependentAddressSpaceType() ||
T->getBaseElementTypeUnsafe()->isDependentAddressSpaceType()) {
Diag(Loc, diag::err_field_with_address_space);
Record->setInvalidDecl();
diff --git a/clang/lib/Sema/SemaExpr.cpp b/clang/lib/Sema/SemaExpr.cpp
index c53a4b789bed..e2c37f8f5238 100644
--- a/clang/lib/Sema/SemaExpr.cpp
+++ b/clang/lib/Sema/SemaExpr.cpp
@@ -5445,15 +5445,15 @@ static FunctionDecl *rewriteBuiltinFunctionDecl(Sema
*Sema, ASTContext ,
Expr *Arg = ArgRes.get();
QualType ArgType = Arg->getType();
if (!ParamType->isPointerType() ||
-ParamType.getQualifiers().hasAddressSpace() ||
+ParamType.hasAddressSpace() ||
!ArgType->isPointerType() ||
-!ArgType->getPointeeType().getQualifiers().hasAddressSpace()) {
+!ArgType->getPointeeType().hasAddressSpace()) {
OverloadParams.push_back(ParamType);
continue;
}
QualType PointeeType = ParamType->getPointeeType();
-if (PointeeType.getQualifiers().hasAddressSpace())
+if (PointeeType.hasAddressSpace())
continue;
NeedsNewDecl = true;
diff --git a/clang/lib/Sema/SemaInit.cpp b/clang/lib/Sema/SemaInit.cpp
index 7421754d95ca..cc9d1a4f6256 100644
--- a/clang/lib/Sema/SemaInit.cpp
+++ b/clang/lib/Sema/SemaInit.cpp
@@ -7853,9 +7853,8 @@ ExprResult InitializationSequence::Perform(Sema ,
// OpenCL v2.0 s6.13.11.1. atomic variables can be initialized in global
scope
QualType ETy = Entity.getType();
- Qualifiers TyQualifiers = ETy.getQualifiers();
- bool HasGlobalAS = TyQualifiers.hasAddressSpace() &&
- TyQualifiers.getAddressSpace() == LangAS::opencl_global;
+ bool HasGlobalAS = ETy.hasAddressSpace() &&
+ ETy.getAddressSpace() == LangAS::opencl_global;
if (S.getLangOpts().OpenCLVersion >= 200 &&
ETy->isAtomicType() && !HasGlobalAS &&
diff --git a/clang/lib/Sema/SemaOverload.cpp
[clang] fa9dd41 - [opencl] Fix address space deduction on array variables.
Author: Michael Liao
Date: 2019-12-04T09:37:50-05:00
New Revision: fa9dd410a9a9aa65ce6731cbe1ee12c5941eb3e8
URL:
https://github.com/llvm/llvm-project/commit/fa9dd410a9a9aa65ce6731cbe1ee12c5941eb3e8
DIFF:
https://github.com/llvm/llvm-project/commit/fa9dd410a9a9aa65ce6731cbe1ee12c5941eb3e8.diff
LOG: [opencl] Fix address space deduction on array variables.
Summary:
- The deduced address space needs applying to its element type as well.
Reviewers: Anastasia
Subscribers: yaxunl, cfe-commits
Tags: #clang
Differential Revision: https://reviews.llvm.org/D70981
Added:
Modified:
clang/lib/Sema/SemaDecl.cpp
clang/test/SemaOpenCL/address-spaces.cl
Removed:
diff --git a/clang/lib/Sema/SemaDecl.cpp b/clang/lib/Sema/SemaDecl.cpp
index d35037273106..660be458a698 100644
--- a/clang/lib/Sema/SemaDecl.cpp
+++ b/clang/lib/Sema/SemaDecl.cpp
@@ -6128,7 +6128,26 @@ void Sema::deduceOpenCLAddressSpace(ValueDecl *Decl) {
if ((getLangOpts().OpenCLCPlusPlus || getLangOpts().OpenCLVersion >= 200)
&&
Var->hasGlobalStorage())
ImplAS = LangAS::opencl_global;
+// If the original type from a decayed type is an array type and that array
+// type has no address space yet, deduce it now.
+if (auto DT = dyn_cast(Type)) {
+ auto OrigTy = DT->getOriginalType();
+ if (!OrigTy.getQualifiers().hasAddressSpace() && OrigTy->isArrayType()) {
+// Add the address space to the original array type and then propagate
+// that to the element type through `getAsArrayType`.
+OrigTy = Context.getAddrSpaceQualType(OrigTy, ImplAS);
+OrigTy = QualType(Context.getAsArrayType(OrigTy), 0);
+// Re-generate the decayed type.
+Type = Context.getDecayedType(OrigTy);
+ }
+}
Type = Context.getAddrSpaceQualType(Type, ImplAS);
+// Apply any qualifiers (including address space) from the array type to
+// the element type. This implements C99 6.7.3p8: "If the specification of
+// an array type includes any type qualifiers, the element type is so
+// qualified, not the array type."
+if (Type->isArrayType())
+ Type = QualType(Context.getAsArrayType(Type), 0);
Decl->setType(Type);
}
}
diff --git a/clang/test/SemaOpenCL/address-spaces.cl
b/clang/test/SemaOpenCL/address-spaces.cl
index 55a55dc75050..09a6dd0ba53f 100644
--- a/clang/test/SemaOpenCL/address-spaces.cl
+++ b/clang/test/SemaOpenCL/address-spaces.cl
@@ -241,3 +241,10 @@ void func_multiple_addr(void) {
__private private_int_t var5; // expected-warning {{multiple identical
address spaces specified for type}}
__private private_int_t *var6;// expected-warning {{multiple identical
address spaces specified for type}}
}
+
+void func_with_array_param(const unsigned data[16]);
+
+__kernel void k() {
+ unsigned data[16];
+ func_with_array_param(data);
+}
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] 59312cb - Fix warning on unused variable. NFC.
Author: Michael Liao
Date: 2019-12-03T21:16:10-05:00
New Revision: 59312cb0b81ca13f0674dde66b8e87a8d51d4dda
URL:
https://github.com/llvm/llvm-project/commit/59312cb0b81ca13f0674dde66b8e87a8d51d4dda
DIFF:
https://github.com/llvm/llvm-project/commit/59312cb0b81ca13f0674dde66b8e87a8d51d4dda.diff
LOG: Fix warning on unused variable. NFC.
Added:
Modified:
clang/lib/AST/Expr.cpp
Removed:
diff --git a/clang/lib/AST/Expr.cpp b/clang/lib/AST/Expr.cpp
index a73531ad5fad..3bc2ea60aa14 100644
--- a/clang/lib/AST/Expr.cpp
+++ b/clang/lib/AST/Expr.cpp
@@ -1678,7 +1678,7 @@ MemberExpr *MemberExpr::Create(
MemberExpr *E = new (Mem) MemberExpr(Base, IsArrow, OperatorLoc, MemberDecl,
NameInfo, T, VK, OK, NOUR);
- if (FieldDecl *Field = dyn_cast(MemberDecl)) {
+ if (isa(MemberDecl)) {
DeclContext *DC = MemberDecl->getDeclContext();
// dyn_cast_or_null is used to handle objC variables which do not
// have a declaration context.
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] 59e69fe - Fix warning on extra ';'. NFC.
Author: Michael Liao
Date: 2019-12-03T16:02:55-05:00
New Revision: 59e69fefab883984e81c77aef58ba587060e87f2
URL:
https://github.com/llvm/llvm-project/commit/59e69fefab883984e81c77aef58ba587060e87f2
DIFF:
https://github.com/llvm/llvm-project/commit/59e69fefab883984e81c77aef58ba587060e87f2.diff
LOG: Fix warning on extra ';'. NFC.
Added:
Modified:
clang/lib/Format/TokenAnnotator.cpp
Removed:
diff --git a/clang/lib/Format/TokenAnnotator.cpp
b/clang/lib/Format/TokenAnnotator.cpp
index 93cb36961ee5..d5d394e61926 100644
--- a/clang/lib/Format/TokenAnnotator.cpp
+++ b/clang/lib/Format/TokenAnnotator.cpp
@@ -2597,7 +2597,7 @@ bool TokenAnnotator::spaceRequiredBeforeParens(const
FormatToken ) const {
static bool isKeywordWithCondition(const FormatToken ) {
return Tok.isOneOf(tok::kw_if, tok::kw_for, tok::kw_while, tok::kw_switch,
tok::kw_constexpr);
-};
+}
bool TokenAnnotator::spaceRequiredBetween(const AnnotatedLine ,
const FormatToken ,
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] c4afc65 - Fix compilation warning. NFC.
Author: Michael Liao
Date: 2019-11-21T12:07:13-05:00
New Revision: c4afc6566a64e6be3f77271781a147bb5ff98b0c
URL:
https://github.com/llvm/llvm-project/commit/c4afc6566a64e6be3f77271781a147bb5ff98b0c
DIFF:
https://github.com/llvm/llvm-project/commit/c4afc6566a64e6be3f77271781a147bb5ff98b0c.diff
LOG: Fix compilation warning. NFC.
Added:
Modified:
clang/lib/Driver/Driver.cpp
Removed:
diff --git a/clang/lib/Driver/Driver.cpp b/clang/lib/Driver/Driver.cpp
index 90f3cea5b2af..c1173e3ddbf0 100644
--- a/clang/lib/Driver/Driver.cpp
+++ b/clang/lib/Driver/Driver.cpp
@@ -3498,7 +3498,7 @@ void Driver::BuildActions(Compilation , DerivedArgList
,
Actions.push_back(
C.MakeAction(MergerInputs, types::TY_Image));
- if (Arg *A = Args.getLastArg(options::OPT_emit_interface_stubs)) {
+ if (Args.hasArg(options::OPT_emit_interface_stubs)) {
llvm::SmallVector PhaseList;
if (Args.hasArg(options::OPT_c)) {
llvm::SmallVector
CompilePhaseList;
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] 0a220de - [HIP] Fix visibility for 'extern' device variables.
Author: Michael Liao
Date: 2019-11-05T14:19:32-05:00
New Revision: 0a220de9e9ca3e6786df6c03fd37668815805c62
URL:
https://github.com/llvm/llvm-project/commit/0a220de9e9ca3e6786df6c03fd37668815805c62
DIFF:
https://github.com/llvm/llvm-project/commit/0a220de9e9ca3e6786df6c03fd37668815805c62.diff
LOG: [HIP] Fix visibility for 'extern' device variables.
Summary:
- Fix a bug which misses the change for a variable to be set with
target-specific attributes.
Reviewers: yaxunl
Subscribers: jvesely, nhaehnle, cfe-commits
Tags: #clang
Differential Revision: https://reviews.llvm.org/D63020
Added:
Modified:
clang/lib/CodeGen/CodeGenModule.cpp
clang/test/CodeGenCUDA/amdgpu-visibility.cu
Removed:
diff --git a/clang/lib/CodeGen/CodeGenModule.cpp
b/clang/lib/CodeGen/CodeGenModule.cpp
index 49df82cea42b..be8f389e1809 100644
--- a/clang/lib/CodeGen/CodeGenModule.cpp
+++ b/clang/lib/CodeGen/CodeGenModule.cpp
@@ -3575,6 +3575,9 @@ CodeGenModule::GetOrCreateLLVMGlobal(StringRef
MangledName,
}
}
+ if (GV->isDeclaration())
+getTargetCodeGenInfo().setTargetAttributes(D, GV, *this);
+
LangAS ExpectedAS =
D ? D->getType().getAddressSpace()
: (LangOpts.OpenCL ? LangAS::opencl_global : LangAS::Default);
@@ -3584,9 +3587,6 @@ CodeGenModule::GetOrCreateLLVMGlobal(StringRef
MangledName,
return getTargetCodeGenInfo().performAddrSpaceCast(*this, GV, AddrSpace,
ExpectedAS, Ty);
- if (GV->isDeclaration())
-getTargetCodeGenInfo().setTargetAttributes(D, GV, *this);
-
return GV;
}
diff --git a/clang/test/CodeGenCUDA/amdgpu-visibility.cu
b/clang/test/CodeGenCUDA/amdgpu-visibility.cu
index 9f44eb047f82..f23e562a4f29 100644
--- a/clang/test/CodeGenCUDA/amdgpu-visibility.cu
+++ b/clang/test/CodeGenCUDA/amdgpu-visibility.cu
@@ -13,6 +13,16 @@
__constant__ int c;
__device__ int g;
+// CHECK-DEFAULT: @e = external addrspace(1) global
+// CHECK-PROTECTED: @e = external protected addrspace(1) global
+// CHECK-HIDDEN: @e = external protected addrspace(1) global
+extern __device__ int e;
+
+// dummy one to hold reference to `e`.
+__device__ int f() {
+ return e;
+}
+
// CHECK-DEFAULT: define amdgpu_kernel void @_Z3foov()
// CHECK-PROTECTED: define protected amdgpu_kernel void @_Z3foov()
// CHECK-HIDDEN: define protected amdgpu_kernel void @_Z3foov()
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] 15140e4 - [hip] Enable pointer argument lowering through coercing type.
Author: Michael Liao
Date: 2019-11-05T13:05:05-05:00
New Revision: 15140e4bacf94fbc509e5a139909aefcd1cc3363
URL:
https://github.com/llvm/llvm-project/commit/15140e4bacf94fbc509e5a139909aefcd1cc3363
DIFF:
https://github.com/llvm/llvm-project/commit/15140e4bacf94fbc509e5a139909aefcd1cc3363.diff
LOG: [hip] Enable pointer argument lowering through coercing type.
Reviewers: tra, rjmccall, yaxunl
Subscribers: jvesely, nhaehnle, cfe-commits
Tags: #clang
Differential Revision: https://reviews.llvm.org/D69826
Added:
clang/test/CodeGenCUDA/amdgpu-kernel-arg-pointer-type.cu
Modified:
clang/lib/CodeGen/CGCall.cpp
clang/lib/CodeGen/TargetInfo.cpp
Removed:
diff --git a/clang/lib/CodeGen/CGCall.cpp b/clang/lib/CodeGen/CGCall.cpp
index 62e8fa037013..e832e4c28334 100644
--- a/clang/lib/CodeGen/CGCall.cpp
+++ b/clang/lib/CodeGen/CGCall.cpp
@@ -1305,6 +1305,15 @@ static void CreateCoercedStore(llvm::Value *Src,
DstTy = Dst.getType()->getElementType();
}
+ llvm::PointerType *SrcPtrTy = llvm::dyn_cast(SrcTy);
+ llvm::PointerType *DstPtrTy = llvm::dyn_cast(DstTy);
+ if (SrcPtrTy && DstPtrTy &&
+ SrcPtrTy->getAddressSpace() != DstPtrTy->getAddressSpace()) {
+Src = CGF.Builder.CreatePointerBitCastOrAddrSpaceCast(Src, DstTy);
+CGF.Builder.CreateStore(Src, Dst, DstIsVolatile);
+return;
+ }
+
// If the source and destination are integer or pointer types, just do an
// extension or truncation to the desired type.
if ((isa(SrcTy) || isa(SrcTy)) &&
diff --git a/clang/lib/CodeGen/TargetInfo.cpp
b/clang/lib/CodeGen/TargetInfo.cpp
index e33d69c86b3c..26c527d7c983 100644
--- a/clang/lib/CodeGen/TargetInfo.cpp
+++ b/clang/lib/CodeGen/TargetInfo.cpp
@@ -7685,6 +7685,42 @@ class AMDGPUABIInfo final : public DefaultABIInfo {
bool isHomogeneousAggregateSmallEnough(const Type *Base,
uint64_t Members) const override;
+ // Coerce HIP pointer arguments from generic pointers to global ones.
+ llvm::Type *coerceKernelArgumentType(llvm::Type *Ty, unsigned FromAS,
+ unsigned ToAS) const {
+// Structure types.
+if (auto STy = dyn_cast(Ty)) {
+ SmallVector EltTys;
+ bool Changed = false;
+ for (auto T : STy->elements()) {
+auto NT = coerceKernelArgumentType(T, FromAS, ToAS);
+EltTys.push_back(NT);
+Changed |= (NT != T);
+ }
+ // Skip if there is no change in element types.
+ if (!Changed)
+return STy;
+ if (STy->hasName())
+return llvm::StructType::create(
+EltTys, (STy->getName() + ".coerce").str(), STy->isPacked());
+ return llvm::StructType::get(getVMContext(), EltTys, STy->isPacked());
+}
+// Arrary types.
+if (auto ATy = dyn_cast(Ty)) {
+ auto T = ATy->getElementType();
+ auto NT = coerceKernelArgumentType(T, FromAS, ToAS);
+ // Skip if there is no change in that element type.
+ if (NT == T)
+return ATy;
+ return llvm::ArrayType::get(NT, ATy->getNumElements());
+}
+// Single value types.
+if (Ty->isPointerTy() && Ty->getPointerAddressSpace() == FromAS)
+ return llvm::PointerType::get(
+ cast(Ty)->getElementType(), ToAS);
+return Ty;
+ }
+
public:
explicit AMDGPUABIInfo(CodeGen::CodeGenTypes ) :
DefaultABIInfo(CGT) {}
@@ -7812,14 +7848,22 @@ ABIArgInfo
AMDGPUABIInfo::classifyKernelArgumentType(QualType Ty) const {
// TODO: Can we omit empty structs?
- // Coerce single element structs to its element.
+ llvm::Type *LTy = nullptr;
if (const Type *SeltTy = isSingleElementStruct(Ty, getContext()))
-return ABIArgInfo::getDirect(CGT.ConvertType(QualType(SeltTy, 0)));
+LTy = CGT.ConvertType(QualType(SeltTy, 0));
+
+ if (getContext().getLangOpts().HIP) {
+if (!LTy)
+ LTy = CGT.ConvertType(Ty);
+LTy = coerceKernelArgumentType(
+LTy, /*FromAS=*/getContext().getTargetAddressSpace(LangAS::Default),
+/*ToAS=*/getContext().getTargetAddressSpace(LangAS::cuda_device));
+ }
// If we set CanBeFlattened to true, CodeGen will expand the struct to its
// individual elements, which confuses the Clover OpenCL backend; therefore
we
// have to set it to false here. Other args of getDirect() are just defaults.
- return ABIArgInfo::getDirect(nullptr, 0, nullptr, false);
+ return ABIArgInfo::getDirect(LTy, 0, nullptr, false);
}
ABIArgInfo AMDGPUABIInfo::classifyArgumentType(QualType Ty,
diff --git a/clang/test/CodeGenCUDA/amdgpu-kernel-arg-pointer-type.cu
b/clang/test/CodeGenCUDA/amdgpu-kernel-arg-pointer-type.cu
new file mode 100644
index ..cb8a75882d4d
--- /dev/null
+++ b/clang/test/CodeGenCUDA/amdgpu-kernel-arg-pointer-type.cu
@@ -0,0 +1,69 @@
+// RUN: %clang_cc1 -triple amdgcn-amd-amdhsa -fcuda-is-device -emit-llvm -x
hip %s -o - | FileCheck %s
+// RUN: %clang_cc1
[clang] d142ec6 - Fix compilation warning. NFC.
Author: Michael Liao
Date: 2019-11-04T10:01:50-05:00
New Revision: d142ec6fef9a053c9fd9edb5a388203cdb121e65
URL:
https://github.com/llvm/llvm-project/commit/d142ec6fef9a053c9fd9edb5a388203cdb121e65
DIFF:
https://github.com/llvm/llvm-project/commit/d142ec6fef9a053c9fd9edb5a388203cdb121e65.diff
LOG: Fix compilation warning. NFC.
Added:
Modified:
clang/lib/Driver/ToolChains/Darwin.cpp
Removed:
diff --git a/clang/lib/Driver/ToolChains/Darwin.cpp
b/clang/lib/Driver/ToolChains/Darwin.cpp
index d8c18effd62c..d550eea94670 100644
--- a/clang/lib/Driver/ToolChains/Darwin.cpp
+++ b/clang/lib/Driver/ToolChains/Darwin.cpp
@@ -1527,8 +1527,8 @@ getDeploymentTargetFromEnvironmentVariables(const Driver
,
Targets[Darwin::TvOS] = "";
} else {
// Don't allow conflicts in any other platform.
-int FirstTarget = llvm::array_lengthof(Targets);
-for (int I = 0; I != llvm::array_lengthof(Targets); ++I) {
+unsigned FirstTarget = llvm::array_lengthof(Targets);
+for (unsigned I = 0; I != llvm::array_lengthof(Targets); ++I) {
if (Targets[I].empty())
continue;
if (FirstTarget == llvm::array_lengthof(Targets))
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] 45787e5 - Fix compilation warning. NFC.
Author: Michael Liao
Date: 2019-10-25T01:06:52-04:00
New Revision: 45787e56829f47e45d127882b1cd1821e7022e68
URL:
https://github.com/llvm/llvm-project/commit/45787e56829f47e45d127882b1cd1821e7022e68
DIFF:
https://github.com/llvm/llvm-project/commit/45787e56829f47e45d127882b1cd1821e7022e68.diff
LOG: Fix compilation warning. NFC.
Added:
Modified:
clang/lib/CodeGen/CGBuiltin.cpp
Removed:
diff --git a/clang/lib/CodeGen/CGBuiltin.cpp b/clang/lib/CodeGen/CGBuiltin.cpp
index 2caa8509ea06..9a56116173ec 100644
--- a/clang/lib/CodeGen/CGBuiltin.cpp
+++ b/clang/lib/CodeGen/CGBuiltin.cpp
@@ -6874,6 +6874,7 @@ Value *CodeGenFunction::EmitARMMVEBuiltinExpr(unsigned
BuiltinID,
return ToReturn;
}
}
+ llvm_unreachable("unknown custom codegen type.");
}
static Value *EmitAArch64TblBuiltinExpr(CodeGenFunction , unsigned
BuiltinID,
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang] 5a48678 - [hip] Allow the declaration of functions with variadic arguments in HIP.
Author: Michael Liao
Date: 2019-10-25T00:39:24-04:00
New Revision: 5a48678a6a1619fada23641a68c2d95ee57806b1
URL:
https://github.com/llvm/llvm-project/commit/5a48678a6a1619fada23641a68c2d95ee57806b1
DIFF:
https://github.com/llvm/llvm-project/commit/5a48678a6a1619fada23641a68c2d95ee57806b1.diff
LOG: [hip] Allow the declaration of functions with variadic arguments in HIP.
Summary:
- As variadic parameters have the lowest rank in overload resolution,
without real usage of `va_arg`, they are commonly used as the
catch-all fallbacks in SFINAE. As the front-end still reports errors
on calls to `va_arg`, the declaration of functions with variadic
arguments should be allowed in general.
Reviewers: jlebar, tra, yaxunl
Subscribers: cfe-commits
Tags: #clang
Differential Revision: https://reviews.llvm.org/D69389
Added:
Modified:
clang/lib/CodeGen/TargetInfo.cpp
clang/lib/Driver/ToolChains/Clang.cpp
clang/lib/Driver/ToolChains/HIP.cpp
clang/test/Driver/hip-toolchain-no-rdc.hip
clang/test/Driver/hip-toolchain-rdc.hip
Removed:
diff --git a/clang/lib/CodeGen/TargetInfo.cpp
b/clang/lib/CodeGen/TargetInfo.cpp
index c2c7b8bf653b..e33d69c86b3c 100644
--- a/clang/lib/CodeGen/TargetInfo.cpp
+++ b/clang/lib/CodeGen/TargetInfo.cpp
@@ -7694,6 +7694,8 @@ class AMDGPUABIInfo final : public DefaultABIInfo {
ABIArgInfo classifyArgumentType(QualType Ty, unsigned ) const;
void computeInfo(CGFunctionInfo ) const override;
+ Address EmitVAArg(CodeGenFunction , Address VAListAddr,
+QualType Ty) const override;
};
bool AMDGPUABIInfo::isHomogeneousAggregateBaseType(QualType Ty) const {
@@ -7757,6 +7759,11 @@ void AMDGPUABIInfo::computeInfo(CGFunctionInfo )
const {
}
}
+Address AMDGPUABIInfo::EmitVAArg(CodeGenFunction , Address VAListAddr,
+ QualType Ty) const {
+ llvm_unreachable("AMDGPU does not support varargs");
+}
+
ABIArgInfo AMDGPUABIInfo::classifyReturnType(QualType RetTy) const {
if (isAggregateTypeForABI(RetTy)) {
// Records with non-trivial destructors/copy-constructors should not be
diff --git a/clang/lib/Driver/ToolChains/Clang.cpp
b/clang/lib/Driver/ToolChains/Clang.cpp
index 70c70dcdbd4d..ae12465d3f8b 100644
--- a/clang/lib/Driver/ToolChains/Clang.cpp
+++ b/clang/lib/Driver/ToolChains/Clang.cpp
@@ -5334,6 +5334,9 @@ void Clang::ConstructJob(Compilation , const JobAction
,
CmdArgs.push_back("-fcuda-short-ptr");
}
+ if (IsHIP)
+CmdArgs.push_back("-fcuda-allow-variadic-functions");
+
// OpenMP offloading device jobs take the argument -fopenmp-host-ir-file-path
// to specify the result of the compile phase on the host, so the meaningful
// device declarations can be identified. Also, -fopenmp-is-device is passed
diff --git a/clang/lib/Driver/ToolChains/HIP.cpp
b/clang/lib/Driver/ToolChains/HIP.cpp
index d84a454359ad..1053a1a60978 100644
--- a/clang/lib/Driver/ToolChains/HIP.cpp
+++ b/clang/lib/Driver/ToolChains/HIP.cpp
@@ -296,6 +296,8 @@ void HIPToolChain::addClangTargetOptions(
options::OPT_fno_gpu_allow_device_init, false))
CC1Args.push_back("-fgpu-allow-device-init");
+ CC1Args.push_back("-fcuda-allow-variadic-functions");
+
// Default to "hidden" visibility, as object level linking will not be
// supported for the foreseeable future.
if (!DriverArgs.hasArg(options::OPT_fvisibility_EQ,
diff --git a/clang/test/Driver/hip-toolchain-no-rdc.hip
b/clang/test/Driver/hip-toolchain-no-rdc.hip
index 540b93286053..74b53cfb 100644
--- a/clang/test/Driver/hip-toolchain-no-rdc.hip
+++ b/clang/test/Driver/hip-toolchain-no-rdc.hip
@@ -20,7 +20,7 @@
// CHECK-SAME: "-aux-triple" "x86_64-unknown-linux-gnu"
// CHECK-SAME: "-emit-llvm-bc"
// CHECK-SAME: {{.*}} "-main-file-name" "a.cu" {{.*}} "-target-cpu" "gfx803"
-// CHECK-SAME: "-fcuda-is-device" "-fvisibility" "hidden"
+// CHECK-SAME: "-fcuda-is-device" "-fcuda-allow-variadic-functions"
"-fvisibility" "hidden"
// CHECK-SAME: "-fapply-global-visibility-to-externs"
// CHECK-SAME: "{{.*}}lib1.bc" "{{.*}}lib2.bc"
// CHECK-SAME: {{.*}} "-o" [[A_BC_803:".*bc"]] "-x" "hip"
@@ -48,7 +48,7 @@
// CHECK-SAME: "-aux-triple" "x86_64-unknown-linux-gnu"
// CHECK-SAME: "-emit-llvm-bc"
// CHECK-SAME: {{.*}} "-main-file-name" "a.cu" {{.*}} "-target-cpu" "gfx900"
-// CHECK-SAME: "-fcuda-is-device" "-fvisibility" "hidden"
+// CHECK-SAME: "-fcuda-is-device" "-fcuda-allow-variadic-functions"
"-fvisibility" "hidden"
// CHECK-SAME: "-fapply-global-visibility-to-externs"
// CHECK-SAME: "{{.*}}lib1.bc" "{{.*}}lib2.bc"
// CHECK-SAME: {{.*}} "-o" [[A_BC_900:".*bc"]] "-x" "hip"
@@ -92,7 +92,7 @@
// CHECK-SAME: "-aux-triple" "x86_64-unknown-linux-gnu"
// CHECK-SAME: "-emit-llvm-bc"
// CHECK-SAME: {{.*}} "-main-file-name" "b.hip" {{.*}} "-target-cpu" "gfx803"
-// CHECK-SAME: "-fcuda-is-device" "-fvisibility"
[clang] 114de1e - Minor coding style fix. NFC.
Author: Michael Liao
Date: 2019-10-22T04:32:30Z
New Revision: 114de1eab29c06ac097c0e97feb713d616798f7a
URL:
https://github.com/llvm/llvm-project/commit/114de1eab29c06ac097c0e97feb713d616798f7a
DIFF:
https://github.com/llvm/llvm-project/commit/114de1eab29c06ac097c0e97feb713d616798f7a.diff
LOG: Minor coding style fix. NFC.
llvm-svn: 375478
Added:
Modified:
clang/lib/Sema/SemaLambda.cpp
Removed:
diff --git a/clang/lib/Sema/SemaLambda.cpp b/clang/lib/Sema/SemaLambda.cpp
index 749b0f2caaa0..c6b19a0b195c 100644
--- a/clang/lib/Sema/SemaLambda.cpp
+++ b/clang/lib/Sema/SemaLambda.cpp
@@ -444,7 +444,8 @@ void Sema::handleLambdaNumbering(
}
auto getMangleNumberingContext =
- [this](CXXRecordDecl *Class, Decl *ManglingContextDecl) ->
MangleNumberingContext * {
+ [this](CXXRecordDecl *Class,
+ Decl *ManglingContextDecl) -> MangleNumberingContext * {
// Get mangle numbering context if there's any extra decl context.
if (ManglingContextDecl)
return (
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
r375310 - [clang][driver] Print compilation phases with indentation.
Author: hliao
Date: Fri Oct 18 17:17:00 2019
New Revision: 375310
URL: http://llvm.org/viewvc/llvm-project?rev=375310=rev
Log:
[clang][driver] Print compilation phases with indentation.
Reviewers: tra, sfantao, echristo
Subscribers: cfe-commits
Tags: #clang
Differential Revision: https://reviews.llvm.org/D69124
Modified:
cfe/trunk/lib/Driver/Driver.cpp
Modified: cfe/trunk/lib/Driver/Driver.cpp
URL:
http://llvm.org/viewvc/llvm-project/cfe/trunk/lib/Driver/Driver.cpp?rev=375310=375309=375310=diff
==
--- cfe/trunk/lib/Driver/Driver.cpp (original)
+++ cfe/trunk/lib/Driver/Driver.cpp Fri Oct 18 17:17:00 2019
@@ -1802,23 +1802,36 @@ bool Driver::HandleImmediateArgs(const C
return true;
}
+enum {
+ TopLevelAction = 0,
+ HeadSibAction = 1,
+ OtherSibAction = 2,
+};
+
// Display an action graph human-readably. Action A is the "sink" node
// and latest-occuring action. Traversal is in pre-order, visiting the
// inputs to each action before printing the action itself.
static unsigned PrintActions1(const Compilation , Action *A,
- std::map ) {
+ std::map ,
+ Twine Indent = {}, int Kind = TopLevelAction) {
if (Ids.count(A)) // A was already visited.
return Ids[A];
std::string str;
llvm::raw_string_ostream os(str);
+ auto getSibIndent = [](int K) -> Twine {
+return (K == HeadSibAction) ? " " : (K == OtherSibAction) ? "| " : "";
+ };
+
+ Twine SibIndent = Indent + getSibIndent(Kind);
+ int SibKind = HeadSibAction;
os << Action::getClassName(A->getKind()) << ", ";
if (InputAction *IA = dyn_cast(A)) {
os << "\"" << IA->getInputArg().getValue() << "\"";
} else if (BindArchAction *BIA = dyn_cast(A)) {
os << '"' << BIA->getArchName() << '"' << ", {"
- << PrintActions1(C, *BIA->input_begin(), Ids) << "}";
+ << PrintActions1(C, *BIA->input_begin(), Ids, SibIndent, SibKind) <<
"}";
} else if (OffloadAction *OA = dyn_cast(A)) {
bool IsFirst = true;
OA->doOnEachDependence(
@@ -1841,8 +1854,9 @@ static unsigned PrintActions1(const Comp
os << ":" << BoundArch;
os << ")";
os << '"';
- os << " {" << PrintActions1(C, A, Ids) << "}";
+ os << " {" << PrintActions1(C, A, Ids, SibIndent, SibKind) << "}";
IsFirst = false;
+ SibKind = OtherSibAction;
});
} else {
const ActionList *AL = >getInputs();
@@ -1850,8 +1864,9 @@ static unsigned PrintActions1(const Comp
if (AL->size()) {
const char *Prefix = "{";
for (Action *PreRequisite : *AL) {
-os << Prefix << PrintActions1(C, PreRequisite, Ids);
+os << Prefix << PrintActions1(C, PreRequisite, Ids, SibIndent,
SibKind);
Prefix = ", ";
+SibKind = OtherSibAction;
}
os << "}";
} else
@@ -1872,9 +1887,13 @@ static unsigned PrintActions1(const Comp
}
}
+ auto getSelfIndent = [](int K) -> Twine {
+return (K == HeadSibAction) ? "+- " : (K == OtherSibAction) ? "|- " : "";
+ };
+
unsigned Id = Ids.size();
Ids[A] = Id;
- llvm::errs() << Id << ": " << os.str() << ", "
+ llvm::errs() << Indent + getSelfIndent(Kind) << Id << ": " << os.str() << ",
"
<< types::getTypeName(A->getType()) << offload_os.str() << "\n";
return Id;
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
r375309 - [hip][cuda] Fix the extended lambda name mangling issue.
Author: hliao
Date: Fri Oct 18 17:15:19 2019
New Revision: 375309
URL: http://llvm.org/viewvc/llvm-project?rev=375309=rev
Log:
[hip][cuda] Fix the extended lambda name mangling issue.
Summary:
- HIP/CUDA host side needs to use device kernel symbol name to match the
device side binaries. Without a consistent naming between host- and
device-side compilations, it's risky that wrong device binaries are
executed. Consistent naming is usually not an issue until unnamed
types are used, especially the lambda. In this patch, the consistent
name mangling is addressed for the extended lambdas, i.e. the lambdas
annotated with `__device__`.
- In [Itanium C++ ABI][1], the mangling of the lambda is generally
unspecified unless, in certain cases, ODR rule is required to ensure
consisent naming cross TUs. The extended lambda is such a case as its
name may be part of a device kernel function, e.g., the extended
lambda is used as a template argument and etc. Thus, we need to force
ODR for extended lambdas as they are referenced in both device- and
host-side TUs. Furthermore, if a extended lambda is nested in other
(extended or not) lambdas, those lambdas are required to follow ODR
naming as well. This patch revises the current lambda mangle numbering
to force ODR from an extended lambda to all its parent lambdas.
- On the other side, the aforementioned ODR naming should not change
those lambdas' original linkages, i.e., we cannot replace the original
`internal` with `linkonce_odr`; otherwise, we may violate ODR in
general. This patch introduces a new field `HasKnownInternalLinkage`
in lambda data to decouple the current linkage calculation based on
mangling number assigned.
[1]: https://itanium-cxx-abi.github.io/cxx-abi/abi.html
Reviewers: tra, rsmith, yaxunl, martong, shafik
Subscribers: cfe-commits
Tags: #clang
Differential Revision: https://reviews.llvm.org/D68818
Added:
cfe/trunk/test/CodeGenCUDA/unnamed-types.cu
Modified:
cfe/trunk/include/clang/AST/DeclCXX.h
cfe/trunk/include/clang/Sema/Sema.h
cfe/trunk/lib/AST/ASTImporter.cpp
cfe/trunk/lib/AST/Decl.cpp
cfe/trunk/lib/Sema/SemaLambda.cpp
cfe/trunk/lib/Sema/TreeTransform.h
cfe/trunk/lib/Serialization/ASTReaderDecl.cpp
cfe/trunk/lib/Serialization/ASTWriter.cpp
Modified: cfe/trunk/include/clang/AST/DeclCXX.h
URL:
http://llvm.org/viewvc/llvm-project/cfe/trunk/include/clang/AST/DeclCXX.h?rev=375309=375308=375309=diff
==
--- cfe/trunk/include/clang/AST/DeclCXX.h (original)
+++ cfe/trunk/include/clang/AST/DeclCXX.h Fri Oct 18 17:15:19 2019
@@ -389,9 +389,12 @@ class CXXRecordDecl : public RecordDecl
/// The number of explicit captures in this lambda.
unsigned NumExplicitCaptures : 13;
+/// Has known `internal` linkage.
+unsigned HasKnownInternalLinkage : 1;
+
/// The number used to indicate this lambda expression for name
/// mangling in the Itanium C++ ABI.
-unsigned ManglingNumber = 0;
+unsigned ManglingNumber : 31;
/// The declaration that provides context for this lambda, if the
/// actual DeclContext does not suffice. This is used for lambdas that
@@ -406,12 +409,12 @@ class CXXRecordDecl : public RecordDecl
/// The type of the call method.
TypeSourceInfo *MethodTyInfo;
-LambdaDefinitionData(CXXRecordDecl *D, TypeSourceInfo *Info,
- bool Dependent, bool IsGeneric,
- LambdaCaptureDefault CaptureDefault)
- : DefinitionData(D), Dependent(Dependent), IsGenericLambda(IsGeneric),
-CaptureDefault(CaptureDefault), NumCaptures(0), NumExplicitCaptures(0),
-MethodTyInfo(Info) {
+LambdaDefinitionData(CXXRecordDecl *D, TypeSourceInfo *Info, bool
Dependent,
+ bool IsGeneric, LambdaCaptureDefault CaptureDefault)
+: DefinitionData(D), Dependent(Dependent), IsGenericLambda(IsGeneric),
+ CaptureDefault(CaptureDefault), NumCaptures(0),
+ NumExplicitCaptures(0), HasKnownInternalLinkage(0),
ManglingNumber(0),
+ MethodTyInfo(Info) {
IsLambda = true;
// C++1z [expr.prim.lambda]p4:
@@ -1705,6 +1708,13 @@ public:
return getLambdaData().ManglingNumber;
}
+ /// The lambda is known to has internal linkage no matter whether it has name
+ /// mangling number.
+ bool hasKnownLambdaInternalLinkage() const {
+assert(isLambda() && "Not a lambda closure type!");
+return getLambdaData().HasKnownInternalLinkage;
+ }
+
/// Retrieve the declaration that provides additional context for a
/// lambda, when the normal declaration context is not specific enough.
///
@@ -1718,9 +1728,12 @@ public:
/// Set the mangling number and context declaration for a lambda
/// class.
- void setLambdaMangling(unsigned ManglingNumber, Decl *ContextDecl) {
+ void setLambdaMangling(unsigned ManglingNumber, Decl
r375245 - [tooling] Relax an assert when multiple GPU targets are specified.
Author: hliao Date: Fri Oct 18 08:03:34 2019 New Revision: 375245 URL: http://llvm.org/viewvc/llvm-project?rev=375245=rev Log: [tooling] Relax an assert when multiple GPU targets are specified. Modified: cfe/trunk/lib/Tooling/Tooling.cpp Modified: cfe/trunk/lib/Tooling/Tooling.cpp URL: http://llvm.org/viewvc/llvm-project/cfe/trunk/lib/Tooling/Tooling.cpp?rev=375245=375244=375245=diff == --- cfe/trunk/lib/Tooling/Tooling.cpp (original) +++ cfe/trunk/lib/Tooling/Tooling.cpp Fri Oct 18 08:03:34 2019 @@ -105,7 +105,7 @@ static const llvm::opt::ArgStringList *g // tooling will consider host-compilation only. For tooling on device // compilation, device compilation only option, such as // `--cuda-device-only`, needs specifying. -assert(Actions.size() == 2); +assert(Actions.size() > 1); assert( isa(Actions.front()) || // On MacOSX real actions may end up being wrapped in ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
[clang-tools-extra] r375039 - [clangd] Add the missing dependency on `clangLex`.
Author: hliao Date: Wed Oct 16 13:22:54 2019 New Revision: 375039 URL: http://llvm.org/viewvc/llvm-project?rev=375039=rev Log: [clangd] Add the missing dependency on `clangLex`. Modified: clang-tools-extra/trunk/clangd/refactor/tweaks/CMakeLists.txt clang-tools-extra/trunk/clangd/tool/CMakeLists.txt Modified: clang-tools-extra/trunk/clangd/refactor/tweaks/CMakeLists.txt URL: http://llvm.org/viewvc/llvm-project/clang-tools-extra/trunk/clangd/refactor/tweaks/CMakeLists.txt?rev=375039=375038=375039=diff == --- clang-tools-extra/trunk/clangd/refactor/tweaks/CMakeLists.txt (original) +++ clang-tools-extra/trunk/clangd/refactor/tweaks/CMakeLists.txt Wed Oct 16 13:22:54 2019 @@ -26,6 +26,7 @@ add_clang_library(clangDaemonTweaks OBJE clangAST clangBasic clangDaemon + clangLex clangToolingCore clangToolingRefactoring clangToolingSyntax Modified: clang-tools-extra/trunk/clangd/tool/CMakeLists.txt URL: http://llvm.org/viewvc/llvm-project/clang-tools-extra/trunk/clangd/tool/CMakeLists.txt?rev=375039=375038=375039=diff == --- clang-tools-extra/trunk/clangd/tool/CMakeLists.txt (original) +++ clang-tools-extra/trunk/clangd/tool/CMakeLists.txt Wed Oct 16 13:22:54 2019 @@ -21,6 +21,7 @@ clang_target_link_libraries(clangd clangBasic clangFormat clangFrontend + clangLex clangSema clangTooling clangToolingCore ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
r374478 - [tooling] Fix assertion on MacOSX.
Author: hliao Date: Thu Oct 10 16:45:20 2019 New Revision: 374478 URL: http://llvm.org/viewvc/llvm-project?rev=374478=rev Log: [tooling] Fix assertion on MacOSX. Modified: cfe/trunk/lib/Tooling/Tooling.cpp Modified: cfe/trunk/lib/Tooling/Tooling.cpp URL: http://llvm.org/viewvc/llvm-project/cfe/trunk/lib/Tooling/Tooling.cpp?rev=374478=374477=374478=diff == --- cfe/trunk/lib/Tooling/Tooling.cpp (original) +++ cfe/trunk/lib/Tooling/Tooling.cpp Thu Oct 10 16:45:20 2019 @@ -106,7 +106,12 @@ static const llvm::opt::ArgStringList *g // compilation, device compilation only option, such as // `--cuda-device-only`, needs specifying. assert(Actions.size() == 2); -assert(isa(Actions.front())); +assert( +isa(Actions.front()) || +// On MacOSX real actions may end up being wrapped in +// BindArchAction. +(isa(Actions.front()) && + isa(*Actions.front()->input_begin(; OffloadCompilation = true; break; } ___ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
