This patch implements codegen for the reduction clause on
any teams construct for elementary data types. It builds
on parallel reductions on the GPU. Subsequently,
the team master writes to a unique location in a global
memory scratchpad. The last team to do so loads and
reduces this array to calculate the final result.
This patch emits two helper functions that are used by
the OpenMP runtime on the GPU to perform reductions across
teams.
Patch by Tian Jin in collaboration with Arpith Jacob
Reviewers: ABataev
Differential Revision: https://reviews.llvm.org/D29879
llvm-svn: 295335
This patch implements codegen for the reduction clause on
any parallel construct for elementary data types. An efficient
implementation requires hierarchical reduction within a
warp and a threadblock. It is complicated by the fact that
variables declared in the stack of a CUDA thread cannot be
shared with other threads.
The patch creates a struct to hold reduction variables and
a number of helper functions. The OpenMP runtime on the GPU
implements reduction algorithms that uses these helper
functions to perform reductions within a team. Variables are
shared between CUDA threads using shuffle intrinsics.
An implementation of reductions on the NVPTX device is
substantially different to that of CPUs. However, this patch
is written so that there are minimal changes to the rest of
OpenMP codegen.
The implemented design allows the compiler and runtime to be
decoupled, i.e., the runtime does not need to know of the
reduction operation(s), the type of the reduction variable(s),
or the number of reductions. The design also allows reuse of
host codegen, with appropriate specialization for the NVPTX
device.
While the patch does introduce a number of abstractions, the
expected use case calls for inlining of the GPU OpenMP runtime.
After inlining and optimizations in LLVM, these abstractions
are unwound and performance of OpenMP reductions is comparable
to CUDA-canonical code.
Patch by Tian Jin in collaboration with Arpith Jacob
Reviewers: ABataev
Differential Revision: https://reviews.llvm.org/D29758
llvm-svn: 295333
This patch implements codegen for the reduction clause on
any parallel construct for elementary data types. An efficient
implementation requires hierarchical reduction within a
warp and a threadblock. It is complicated by the fact that
variables declared in the stack of a CUDA thread cannot be
shared with other threads.
The patch creates a struct to hold reduction variables and
a number of helper functions. The OpenMP runtime on the GPU
implements reduction algorithms that uses these helper
functions to perform reductions within a team. Variables are
shared between CUDA threads using shuffle intrinsics.
An implementation of reductions on the NVPTX device is
substantially different to that of CPUs. However, this patch
is written so that there are minimal changes to the rest of
OpenMP codegen.
The implemented design allows the compiler and runtime to be
decoupled, i.e., the runtime does not need to know of the
reduction operation(s), the type of the reduction variable(s),
or the number of reductions. The design also allows reuse of
host codegen, with appropriate specialization for the NVPTX
device.
While the patch does introduce a number of abstractions, the
expected use case calls for inlining of the GPU OpenMP runtime.
After inlining and optimizations in LLVM, these abstractions
are unwound and performance of OpenMP reductions is comparable
to CUDA-canonical code.
Patch by Tian Jin in collaboration with Arpith Jacob
Reviewers: ABataev
Differential Revision: https://reviews.llvm.org/D29758
llvm-svn: 295319
Removed ndrange_t as Clang builtin type and added
as a struct type in the OpenCL header.
Use type name to do the Sema checking in enqueue_kernel
and modify IR generation accordingly.
Review: D28058
Patch by Dmitry Borisenkov!
llvm-svn: 295311
Recommit r293585 that was reverted in r293611 with new fixes. The previous
issue was determined to be an overly aggressive AST visitor from forward
declared objects. The visitor will now only deeply visit certain Decl's and
only do a shallow information extraction from all other Decl's.
When objects are imported for modules, there is a chance that a name collision
will cause an ODR violation. Previously, only a small number of such
violations were detected. This patch provides a stronger check based on
AST nodes.
The information needed to uniquely identify an object is taken from the AST and
put into a one-dimensional byte stream. This stream is then hashed to give
a value to represent the object, which is stored with the other object data
in the module.
When modules are loaded, and Decl's are merged, the hash values of the two
Decl's are compared. Only Decl's with matched hash values will be merged.
Mismatch hashes will generate a module error, and if possible, point to the
first difference between the two objects.
The transform from AST to byte stream is a modified depth first algorithm.
Due to references between some AST nodes, a pure depth first algorithm could
generate loops. For Stmt nodes, a straight depth first processing occurs.
For Type and Decl nodes, they are replaced with an index number and only on
first visit will these nodes be processed. As an optimization, boolean
values are saved and stored together in reverse order at the end of the
byte stream to lower the ammount of data that needs to be hashed.
Compile time impact was measured at 1.5-2.0% during module building, and
negligible during builds without module building.
Differential Revision: https://reviews.llvm.org/D21675
llvm-svn: 295284
Enable evaluation of captures within constexpr lambdas by using a strategy similar to that used in CodeGen:
- when starting evaluation of a lambda's call operator, create a map from VarDecl's to a closure's FieldDecls
- every time a VarDecl (or '*this) that represents a capture is encountered while evaluating the expression via the expression evaluator (specifically the LValueEvaluator) in ExprConstant.cpp - it is replaced by the corresponding FieldDecl LValue (an Lvalue-to-Rvalue conversion on this LValue representation then determines the right rvalue when needed).
Thanks to Richard Smith and Hubert Tong for their review and feedback!
https://reviews.llvm.org/D29748
llvm-svn: 295279
This appears to be the only template argument deduction context where we were
missing this check. Surprisingly, other implementations also appear to miss
the check in this case; it may turn out that important code is relying on
the widespread non-conformance here, in which case we'll need to reconsider.
llvm-svn: 295277
After r264564, we allowed direct-list-initialization of an enum from an
integral value in C++1z mode, so long as that value can convert to the
enum's underlying type.
In this kind of initialization, we need a lvalue-to-rvalue conversion
for the initializer value if it is not a rvalue. This lets us accept the
following code:
enum class A : unsigned {};
A foo(unsigned x) { return A{x}; }
Differential Revision: https://reviews.llvm.org/D29723
llvm-svn: 295266
Destructor references are not modelled explicitly in the AST. This adds
checks for destructor calls due to variable definitions and temporaries.
If a dllimport function references a non-dllimport destructor, it must
not be emitted available_externally, as the referenced destructor might
live across the DLL boundary and isn't exported.
llvm-svn: 295258
The function is used to check whether a type is a class with
non-dllimport destructor. It needs to look through typedefs and array
types.
llvm-svn: 295257
Two functions that differ only in their enable_if attributes are
considered overloads, so we should check for those when we're trying to
figure out if two functions are mergeable.
We need to do the same thing for pass_object_size, as well. Looks like
that'll be a bit less trivial, since we sometimes do these merging
checks before we have pass_object_size attributes available (see the
merge checks in ASTDeclReader::VisitFunctionDecl that happen before we
read parameters, and merge checks in calls to ReadDeclAs<>()).
llvm-svn: 295252
r274291 made changes to prefer calling a move constructor to calling a
copy constructor when returning from a function. This caused programs to
crash when a __block variable in the heap was moved out and used later.
This commit fixes the bug by disallowing moving out of __block variables
implicitly.
rdar://problem/28181080
Differential Revision: https://reviews.llvm.org/D29908
llvm-svn: 295150
Summary: Previously the cleanups (e.g. dtor calls) are inserted into the
outer scope (e.g. function body scope), instead of it's own scope. After
the fix, the cleanups are inserted right after getting the size value.
This fixes pr30306.
Reviewers: rsmith
Subscribers: cfe-commits
Differential Revision: https://reviews.llvm.org/D24333
llvm-svn: 295123
that has been explicitly specialized!
We assume in various places that we can tell the template specialization kind
of a class type by looking at the declaration produced by TagType::getDecl.
That was previously not quite true: for an explicit specialization, we could
have first seen a template-id denoting the specialization (with a use that does
not trigger an implicit instantiation of the defintiion) and then seen the
first explicit specialization declaration. TagType::getDecl would previously
return an arbitrary declaration when called on a not-yet-defined class; it
now consistently returns the most recent declaration in that case.
llvm-svn: 295118
In case user did not provide valid standard name for -std option, available
values (with short description) will be reported.
Patch by Paweł Żukowski!
llvm-svn: 295113
This is a re-try of r295085: fix up some test cases that assume that
profile name variables are preserved by the instrprof pass.
This catches one additional case in test/CoverageMapping/unused_names.c.
llvm-svn: 295101
block or lambda.
This is a follow-up to r281682, which fixed a bug in computeBlockInfo
where the captured VarDecl's type, rather than the captured field type
of the enclosing lambda or block, was used to compute the layout of a
block.
This commit makes similar changes to enterBlockScope. This is necessary
to correctly determine whether a block capture requires cleanup.
rdar://problem/30388124
llvm-svn: 295034
guide from a constructor.
The purpose of this change is to avoid triggering instantiation of the class
when substituting back into the deduction guide if it uses a typedef member.
We will still instantiate the class if the constructor (explicitly or
implicitly, directly or indirectly) uses the current instantiation in a way
that we can't canonicalize out, but that seems unavoidable.
llvm-svn: 295016
They are a little bit of a special case in the mangling. They are always
mangled without taking into account their virtual-ness of the
destructor. They are also mangled to return void, unlike the actual
destructor.
This fixes PR31931.
Differential Revision: https://reviews.llvm.org/D29912
llvm-svn: 295010
such guides below explicit ones, and ensure that references to the class's
template parameters are not treated as forwarding references.
We make a few tweaks to the wording in the current standard:
1) The constructor parameter list is copied faithfully to the deduction guide,
without losing default arguments or a varargs ellipsis (which the standard
wording loses by omission).
2) If the class template declares no constructors, we add a T() -> T<...> guide
(which will only ever work if T has default arguments for all non-pack
template parameters).
3) If the class template declares nothing that looks like a copy or move
constructor, we add a T(T<...>) -> T<...> guide.
#2 and #3 follow from the "pretend we had a class type with these constructors"
philosophy for deduction guides.
llvm-svn: 295007
This bypasses integer sanitization checks which are redundant on the expression since it's been checked by Sema. Fixes a clang codegen assertion on "void test() { new int[0+1]{0}; }" when building with -fsanitize=signed-integer-overflow.
llvm-svn: 295006
This commit adds context sensitive code completion support for the C++11
keywords that currently don't have completion results.
The following keywords are supported by this patch:
alignas
constexpr
static_assert
noexcept (as a function/method qualifier)
thread_local
The following special identifiers are also supported:
final (as a method qualifier or class qualifier)
override
rdar://29219185
Differential Revision: https://reviews.llvm.org/D28286
llvm-svn: 295001
Different diagnostics when format string does not match
actual arg type.
This commit contains the first 2 of 3 tests reviewed in D29685
llvm-svn: 294979
If the preamble had diagnostic state this would leave behind invalid
state in the DiagnosticsEngine and crash later. The test case runs into
an assertion in DiagnosticsEngine::setSourceManager.
llvm-svn: 294963
Summary:
Sema::CheckCompletedCoroutineBody was growing unwieldy with building all of the substatements. Also, constructors for CoroutineBodyStmt had way too many parameters.
Instead, CoroutineBodyStmt now defines CtorArgs structure with all of the required construction parameters.
CheckCompleteCoroutineBody delegates construction of individual substatements to short functions one per each substatement.
Also, added a drive-by fix of initializing CoroutinePromise to nullptr in ScopeInfo.h.
And addressed the FIXME that wanted to tail allocate extra room at the end of the CoroutineBodyStmt to hold parameter move expressions. (The comment was longer that the code that implemented tail allocation).
Reviewers: rsmith, EricWF
Subscribers: mehdi_amini, cfe-commits
Differential Revision: https://reviews.llvm.org/D28835
llvm-svn: 294933
...function type with a redeclaration having the same attribute. Fixing this
introduced a secondary problem where we were assuming that K&R functions
could not be attributed types when reporting old-style function definitions
that are not preceded by a prototype."
Also Revert "Hopefully fixes a compile error introduced by r294861."
This reverts commit r294862, r294861, as they bork the ARM builds and
haven't fix it back.
Also, please, short commit titles, long commit decsriptions...
llvm-svn: 294910
Certain ARC runtime functions have an ABI contract of being forwarding.
Annotate the functions with the appropriate `returned` attribute on the
arguments. This hoists some of the runtime ABI contract information
into the frontend rather than the backend transformations.
The test adjustments are to mark the returned function parameter as
such. The minor change to the IR output is due to the fact that the
returned reference of the object causes it to extend the lifetime of the
object by returning an autoreleased return value. The result is that
the explicit objc_autorelease call is no longer formed, as autorelease
elision is now possible on the return.
llvm-svn: 294872
Summary:
The -mmcu option for GCC sets macros like __AVR_ATmega328P__ (with the trailing
underscores), be sure to include these underscores for Clangs -mcpu option.
See "AVR Built-in Macros" in https://gcc.gnu.org/onlinedocs/gcc/AVR-Options.html
Reviewers: jroelofs, dylanmckay
Reviewed By: jroelofs, dylanmckay
Subscribers: efriedma, cfe-commits
Differential Revision: https://reviews.llvm.org/D29817
llvm-svn: 294869
What we want to actually control this behavior is something more local
than an EvalutationMode. Please see the linked revision for more
discussion on why/etc.
This fixes PR31843.
Differential Revision: https://reviews.llvm.org/D29469
llvm-svn: 294800
It's actually meaningful and useful to allow such variables to have no
initializer, but we are strictly following the standard here until the C++
committee reaches consensus on allowing this.
llvm-svn: 294785
Expose the half type (fp16) through libclang and the python bindings.
It seems CXType_LastBuiltin was not updated in b2ea6d9 ("Enable
support for __float128 in Clang", 2016-04-13), so update it now.
Add an Index test for OpenCL types; in the future we will add other
OpenCL types such as images to this test.
Patch by Sven van Haastregt.
Differential Revision: https://reviews.llvm.org/D29718
llvm-svn: 294754
until we can get better TargetMachine::isCompatibleDataLayout to compare - otherwise
we can't code generate existing bitcode without a string equality data layout.
This reverts commit r294703.
llvm-svn: 294708
For other platforms we should find out what they need and likely
make the same change, however, a smaller additional change is easier
for platforms we know have it specified in the ABI.
clang support for r294702
llvm-svn: 294703
Summary:
This adds associated constraints as a property of class templates.
An error is produced if redeclarations are not similarly constrained.
Reviewers: rsmith, faisalv, aaron.ballman
Reviewed By: rsmith
Subscribers: cfe-commits, nwilson
Differential Revision: https://reviews.llvm.org/D25674
llvm-svn: 294697
This is a followup change to add v7ve support to clang for gcc
compatibility. Please see r294661.
Patch by Manoj Gupta.
Differential Revision: https://reviews.llvm.org/D29773
llvm-svn: 294662
__fastfail terminates the process immediately with a special system
call. It does not run any process shutdown code or exception recovery
logic.
Fixes PR31854
llvm-svn: 294606
1. Adds the command line flag for clzero.
2. Includes the clzero flag under znver1.
3. Defines the macro for clzero.
4. Adds a new file which has the intrinsic definition for clzero instruction.
Patch by Ganesh Gopalasubramanian with some additional tests from me.
Differential revision: https://reviews.llvm.org/D29386
llvm-svn: 294559
nested-name-specifier (as the standard appears to require), treat it as the
type specifier 'decltype(auto)' followed by a nested-name-specifier starting
with '::'.
llvm-svn: 294506
Sometimes the MS ABI needs to emit thunks for declarations that don't
have bodies. Destructor thunks make calls to inlinable functions, so
they need line info or LLVM will complain.
Fixes PR31893
llvm-svn: 294465
Summary:
This teaches clang how to parse and lower the 'interrupt' and 'naked'
attributes.
This allows interrupt signal handlers to be written.
Reviewers: aaron.ballman
Subscribers: malcolm.parsons, cfe-commits
Differential Revision: https://reviews.llvm.org/D28451
llvm-svn: 294402
Add a warning for shadowed variables across records. Referencing a
shadow'ed variable may not give the desired variable. Add an optional
warning for the shadowing.
Patch by James Sun!
llvm-svn: 294401
Following up on r291465 after a regression in r276159. When we use
-fmodule-name=X while building a PCH, modular headers in X will be
textually included and the compiler knows that we are not building
module X, so don't serialize such headers in the PCH as being part of a
module, because at this point they are not.
This was causing subtle bugs and malformed AST crashes, for instance,
when using the PCH in subsequent compiler invocation with -fmodules, the
HFI for a modular header would map to the PCH, which would force a
module load of and unexistent module ID.
rdar://problem/30171164
llvm-svn: 294361
Fix for bug 30217 - incorrect error given for logical
NOT operation with a pointer type: corrected sema check
and improved related tests.
Review: D29038
llvm-svn: 294313
We model deduction-guides as functions with a new kind of name that identifies
the template whose deduction they guide; the bulk of this patch is adding the
new name kind. This gives us a clean way to attach an extensible list of guides
to a class template in a way that doesn't require any special handling in AST
files etc (and we're going to need these functions we come to performing
deduction).
llvm-svn: 294266
GCC 7 will predefine two new macros on s390x:
- __ARCH__ indicates the ISA architecture level
- __VX__ indicates that the vector facility is available
This adds those macros to clang as well to ensure continued
compatibility with GCC.
llvm-svn: 294197
https://reviews.llvm.org/D29501
It looks like I forgot to remove a FIXME comment with the associated statement. The test does not need it and it gives the wrong impression of being an incomplete test.
llvm-svn: 294195
Summary:
This tells clang about all of the different AVR microcontrollers.
It also adds code to define the correct preprocessor macros for each
device.
Reviewers: jroelofs, asl
Reviewed By: asl
Subscribers: asl, cfe-commits
Differential Revision: https://reviews.llvm.org/D28346
llvm-svn: 294177
Summary:
Previously the method would simply return false, causing every single
inline assembly constraint to trigger a compile error.
This adds inline assembly constraint support for the AVR target.
This patch is derived from the code in
AVRISelLowering::getConstraintType.
More details can be found on the AVR-GCC reference wiki
http://www.nongnu.org/avr-libc/user-manual/inline_asm.html
Reviewers: jroelofs, asl
Reviewed By: asl
Subscribers: asl, ahatanak, saaadhu, cfe-commits
Differential Revision: https://reviews.llvm.org/D28344
llvm-svn: 294176
GCC does not generate `__unix` nor `unix` macros. The latter already
intrudes into the user's namespace and should be avoided. Use the
canonical spelling of `__unix__` across all the targets.
llvm-svn: 294148
While there is nothing to do at link time to get pthreads support on
darwin, specifying the argument is fine and we should not warn about
unused arguments.
llvm-svn: 294065
After r260016 and r260017 disabled typo correction for ivars and properties
clang didn't report errors about unresolved identifier in the base of ivar and
property ref expressions. This meant that clang invoked CodeGen on invalid AST
which then caused a crash.
This commit re-enables typo correction for ivars and properites, and fixes the
PR25113 & PR26486 (that were originally fixed in r260017 and r260016) in a
different manner by transforming the Objective-C ivar reference expression with
'IsFreeIvar' preserved.
rdar://30310772
llvm-svn: 294008
Guard against a null pointer dereference that caused Clang to crash
when processing a class containing an _Atomic qualified data member,
and that is tagged with 'dllexport'.
Differential Revision: https://reviews.llvm.org/D29208
llvm-svn: 293911
name. If the dependent name happened to end in a template-id (X<T>::Y<U>), we
would fail to notice that the 'typename' keyword is missing when resolving it
to a type.
It turns out that GCC has a similar bug. If this shows up in much real code, we
can easily downgrade this to an ExtWarn.
llvm-svn: 293815
For non-template dllimport functions, MSVC allows providing an inline
definition without spelling out the attribute again. In the example below, f
remains a dllimport function.
__declspec(dllimport) int f();
inline int f() { return 42; }
int useit() {
return f();
}
However, for a function template, not putting dllimport on the redeclaration
causes it to be dropped. In the example below, f is not dllimport.
template <typename> __declspec(dllimport) int f();
template <typename> inline int f() { return 42; }
int useit() {
return f<int>();
}
This patch makes Clang match MSVC for the second example.
MSVC does not warn about the attribute being dropped in the example above, but
I think we should. (MSVC does warn if the inline keyword isn't used.)
Differential Revision: https://reviews.llvm.org/D29152
llvm-svn: 293800
clang-cl would evaluate the arguments right-to-left (see PR), and for
non-Windows targets I suppose we only got it because we were already
emitting left-to-right in CodeGenFunction::EmitCallArgs.
Differential Revision: https://reviews.llvm.org/D29350
llvm-svn: 293732
The Module::WithCodegen flag was only being set when the module was
parsed from a ModuleMap. Instead set it late, in the ASTWriter to match
the layer where the MODULAR_CODEGEN_DECLs list is determined (the
WithCodegen flag essentially means "are this module's decls in
MODULAR_CODEGEN_DECLs").
When simultaneous emission of AST file and modular object is implemented
this may need to change - the Module::WithCodegen flag will need to be
set earlier, and ideally the MODULAR_CODEGEN_DECLs gathering will
consult this flag (that's not possible right now since Decls destined
for an AST File don't have a Module - only if they're /read/ from a
Module is that true - I expect that would need to change as well).
llvm-svn: 293692
FindInstantiatedDecl or passing it to RebuildMemberExpr.
This fixes PR30361.
rdar://problem/17341274
Differential Revision: https://reviews.llvm.org/D24969
llvm-svn: 293678
My original warning was very conservative and I never revisited the
heuristics that were used.
This would have caught http://crbug.com/687251 at compile time.
llvm-svn: 293677
We're seeing what we believe are false positives. (It's hard to tell with the
available diagnostics, and I'm not sure how to reduce them yet).
I'll send Richard reproduction details offline.
djasper/chandlerc suggested this should be a warning for now, to make rolling it
out feasible.
llvm-svn: 293611
Summary:
In VirtualCallChecker, handle indirect calls.
getDirectCallee() can be nullptr, and dyn_cast(nullptr) is UB
Reviewers: bkramer
Subscribers: cfe-commits
Differential Revision: https://reviews.llvm.org/D29303
llvm-svn: 293604
This fixes an assertion failure that occurs later in the function when
an ObjCEncodeExpr is cast to StringLiteral.
rdar://problem/30111207
llvm-svn: 293596
Don't try to map an APSInt addend to an int64_t in pointer arithmetic before
bounds-checking it. This gives more consistent behavior (outside C++11, we
consistently use 2s complement semantics for both pointer and integer overflow
in constant expressions) and fixes some cases where in C++11 we would fail to
properly check for out-of-bounds pointer arithmetic (if the 2s complement
64-bit overflow landed us back in-bounds).
In passing, also fix some cases where we'd perform possibly-overflowing
arithmetic on CharUnits (which have a signed underlying type) during constant
expression evaluation.
llvm-svn: 293595
When objects are imported for modules, there is a chance that a name collision
will cause an ODR violation. Previously, only a small number of such
violations were detected. This patch provides a stronger check based on
AST nodes.
The information needed to uniquely identify an object is taked from the AST and
put into a one-dimensional byte stream. This stream is then hashed to give
a value to represent the object, which is stored with the other object data
in the module.
When modules are loaded, and Decl's are merged, the hash values of the two
Decl's are compared. Only Decl's with matched hash values will be merged.
Mismatch hashes will generate a module error, and if possible, point to the
first difference between the two objects.
The transform from AST to byte stream is a modified depth first algorithm.
Due to references between some AST nodes, a pure depth first algorithm could
generate loops. For Stmt nodes, a straight depth first processing occurs.
For Type and Decl nodes, they are replaced with an index number and only on
first visit will these nodes be processed. As an optimization, boolean
values are saved and stored together in reverse order at the end of the
byte stream to lower the ammount of data that needs to be hashed.
Compile time impact was measured at 1.5-2.0% during module building, and
negligible during builds without module building.
Differential Revision: https://reviews.llvm.org/D21675
llvm-svn: 293585
This re-applies r293343 (reverts commit r293475) with a fix for an
assertion failure caused by a missing integer cast. I tested this patch
by using the built compiler to compile X86FastISel.cpp.o with ubsan.
Original commit message:
Ubsan does not report UB shifts in some cases where the shift exponent
needs to be truncated to match the type of the shift base. We perform a
range check on the truncated shift amount, leading to false negatives.
Fix the issue (PR27271) by performing the range check on the original
shift amount.
Differential Revision: https://reviews.llvm.org/D29234
llvm-svn: 293572
This fixes various ways to tickle an assertion in constant expression
evaluation when using __int128. Longer term, we need to figure out what should
happen here: either any kind of overflow in offset calculation should result in
a non-constant value or we should truncate to 64 bits. In C++11 onwards, we're
effectively already checking for overflow because we strictly enforce array
bounds checks, but even there some forms of overflow can slip past undetected.
llvm-svn: 293568
Accounts for a case that caused an assertion failure by attempting to
query for the vtable linkage of a non-dynamic type.t
This reverts commit r292801.
llvm-svn: 293462
First pass at generating weak definitions of inline functions from module files
(& skipping (-O0) or emitting available_externally (optimizations)
definitions where those modules are used).
External functions defined in modules are emitted into the modular
object file as well (this may turn an existing ODR violation (if that
module were imported into multiple translations) into valid/linkable
code).
Internal symbols (static functions, for example) are not correctly
supported yet. The symbol will be produced, internal, in the modular
object - unreferenceable from the users.
Reviewers: rsmith
Differential Revision: https://reviews.llvm.org/D28845
llvm-svn: 293456
Support for CUDA printf is exploited to support printf for
an NVPTX OpenMP device.
To reflect the support of both programming models, the file
CGCUDABuiltin.cpp has been renamed to CGGPUBuiltin.cpp, and
the call EmitCUDADevicePrintfCallExpr has been renamed to
EmitGPUDevicePrintfCallExpr.
Reviewers: jlebar
Differential Revision: https://reviews.llvm.org/D17890
llvm-svn: 293444
I modify clang driver for windows to include:
"-wholearchive:asan_dynamic_runtime_thunk", so all object files in the
static library: asan_dynamic_runtime_thunk are considered by the linker.
This is necessary, because some object files only include linker pragmas,
and doesn't resolve any symbol. If we don't include that flag, the
linker will ignore them, and won't read the linker pragmas.
Differential Revision: https://reviews.llvm.org/D29159
llvm-svn: 293420
r293360 broke some ARM bots, because size_t on those targets is
apparently `unsigned int`, not `unsigned long`. `sizeof(whatever)`
should to give us a `size_t`, so we can just use the type of that
instead.
llvm-svn: 293369
This patch changes how we handle argument-dependent `diagnose_if`
attributes. In particular, we now check them in the same place that we
check for things like passing NULL to Nonnull args, etc. This is
basically better in every way than how we were handling them before. :)
This fixes PR31638, PR31639, and PR31640.
Differential Revision: https://reviews.llvm.org/D28889
llvm-svn: 293360
As Mehdi put it, entities should either be
available_externally+weak_odr, or linkonce_odr+linkonce_odr. While some
functions are emitted a_e/weak, their local variables were emitted
a_e/linkonce_odr.
While it might be nice to emit them a_e/weak, the Itanium ABI (& best
guess at MSVC's behavior as well) requires the local to be
linkonce/linkonce.
Reviewers: rsmith, mehdi_amini
Differential Revision: https://reviews.llvm.org/D29233
llvm-svn: 293344
Ubsan does not report UB shifts in some cases where the shift exponent
needs to be truncated to match the type of the shift base. We perform a
range check on the truncated shift amount, leading to false negatives.
Fix the issue (PR27271) by performing the range check on the original
shift amount.
Differential Revision: https://reviews.llvm.org/D29234
llvm-svn: 293343
In r279546 I disabled all frame pointer elimination at the front-end on
ARM-Darwin (and warned about it) because before that the backend had been
silently ignoring these options. It turns out we didn't ignore
-momit-leaf-frame-pointer though, just the more general -fomit-frame-pointer.
So this re-enables passing that down to CodeGen so that everything really does
continue working as before (with better diagnostics).
llvm-svn: 293311
Both on Mac and Windows, it's common to have a 'Users' directory in the
root of the filesystem, so one might specify a filename as
'/Users/me/myfile.c'. clang-cl (as well as MSVC's cl.exe) will interpret
that as invoking the '/U' option, which is probably not what the user
wanted. Add a warning about this.
Differential Revision: https://reviews.llvm.org/D29198
llvm-svn: 293305
Modify ObjC blocks impl wrt address spaces as follows:
- keep default private address space for blocks generated
as local variables (with captures);
- add global address space for global block literals (no captures);
- make the block invoke function and enqueue_kernel prototype with
the generic AS block pointer parameter to accommodate both
private and global AS cases from above;
- add block handling into default AS because it's implemented as
a special pointer type (BlockPointer) in the frontend and therefore
it is used as a pointer everywhere. This is also needed to accommodate
both private and global AS blocks for the two cases above.
- removes ObjC RT specific symbols (NSConcreteStackBlock and
NSConcreteGlobalBlock) in the OpenCL mode.
Review: https://reviews.llvm.org/D28814
llvm-svn: 293286
The patch teaches the Clang driver how to handle the N64 static
relocation model properly. It enforces the correct target feature
(+noabicalls) when -fno-pic is used. This is required as non-pic
N64 code as the abi extension to call PIC code (CPIC) is unsupported.
Make PIC the default for mips64 and mips64el, this affects both N32
& N64 ABIs, to better match GCC.
As part of this effort, clean up the assembler invocation command
builder, so the correct flags are used.
This and r293279 in LLVM resolves PR/23485.
Thanks to Brooks Davis for reporting the issue!
Reviewers: slthakur, seanbruno
Differential Revision: https://reviews.llvm.org/D29031
llvm-svn: 293285
Function call can appear in the arguments of another function call, eg.:
foo(bar());
This patch adds support for such cases.
Patch by Ivan Sidorenko!
Differential revision: https://reviews.llvm.org/D28905
llvm-svn: 293280
another declaration, ensure we actually serialize / deserialize that
declaration.
Before this patch, if another copy of the typedef were merged with the parsed
version, we would emit type information referring to the merged version and
consequently emit nothing about the parsed anonymous struct. This resulted in
us losing information, particularly the visible merged module set for the
parsed definition. Force that information to be emitted and to be loaded when
the typedef is used.
llvm-svn: 293219
This change adds a new type node, DeducedTemplateSpecializationType, to
represent a type template name that has been used as a type. This is modeled
around AutoType, and shares a common base class for representing a deduced
placeholder type.
We allow deduced class template types in a few more places than the standard
does: in conditions and for-range-declarators, and in new-type-ids. This is
consistent with GCC and with discussion on the core reflector. This patch
does not yet support deduced class template types being named in typename
specifiers.
llvm-svn: 293207
Turning on the warning by default helps the users as it's a common
mistake to capture out-parameters in a block without ensuring the object
assigned doesn't get released.
rdar://problem/30200058
llvm-svn: 293199
even in the presence of nullability qualifiers.
This commit fixes bugs in r285031 where -Wblock-capture-autoreleasing
wouldn't issue warnings when the function parameters were annotated
with nullability qualifiers. Specifically, look through the sugar and
see if there is an AttributedType of kind attr_objc_ownership to
determine whether __autoreleasing was explicitly specified or implicitly
added by the compiler.
rdar://problem/30193488
llvm-svn: 293194
This is a simple patch to teach OpenMP codegen to emit the construct
in Generic mode.
Reviewers: ABataev
Differential Revision: https://reviews.llvm.org/D29143
llvm-svn: 293183
The handler that deals with IR passed/missed/analysis remarks is extended to
also handle the corresponding MIR remarks.
The more thorough testing in done via llc (rL293113, rL293121). Here we just
make sure that the functionality is accessible through clang.
llvm-svn: 293146
Instead of using the location of the beginning '-'/'+'.
This is consistent with location used for function decls and ObjC method calls where we use the base name as the location as well.
llvm-svn: 293134
Rather than storing a single flat list of SourceLocations where the diagnostic
state changes (in source order), we now store a separate list for each FileID
in which there is a diagnostic state transition. (State for other files is
built and cached lazily, on demand.) This has two consequences:
1) We can now sensibly support modules, and properly track the diagnostic state
for modular headers (this matters when, for instance, triggering instantiation
of a template defined within a module triggers diagnostics).
2) It's much faster than the old approach, since we can now just do a binary
search on the offsets within the FileID rather than needing to call
isBeforeInTranslationUnit to determine source order (which is surprisingly
slow). For some pathological (but real world) files, this reduces total
compilation time by more than 10%.
For now, the diagnostic state points for modules are loaded eagerly. It seems
feasible to defer this until diagnostic state information for one of the
module's files is needed, but that's not part of this patch.
llvm-svn: 293123
in the current lexical scope.
clang currently emits the lifetime.start marker of a variable when the
variable comes into scope even though a variable's lifetime starts at
the entry of the block with which it is associated, according to the C
standard. This normally doesn't cause any problems, but in the rare case
where a goto jumps backwards past the variable declaration to an earlier
point in the block (see the test case added to lifetime2.c), it can
cause mis-compilation.
To prevent such mis-compiles, this commit conservatively disables
emitting lifetime variables when a label has been seen in the current
block.
This problem was discussed on cfe-dev here:
http://lists.llvm.org/pipermail/cfe-dev/2016-July/050066.html
rdar://problem/30153946
Differential Revision: https://reviews.llvm.org/D27680
llvm-svn: 293106
Summary:
Now when you ask clang to link in a bitcode module, you can tell it to
set attributes on that module's functions to match what we would have
set if we'd emitted those functions ourselves.
This is particularly important for fast-math attributes in CUDA
compilations.
Each CUDA compilation links in libdevice, a bitcode library provided by
nvidia as part of the CUDA distribution. Without this patch, if we have
a user-function F that is compiled with -ffast-math that calls a
function G from libdevice, F will have the unsafe-fp-math=true (etc.)
attributes, but G will have no attributes.
Since F calls G, the inliner will merge G's attributes into F's. It
considers the lack of an unsafe-fp-math=true attribute on G to be
tantamount to unsafe-fp-math=false, so it "merges" these by setting
unsafe-fp-math=false on F.
This then continues up the call graph, until every function that
(transitively) calls something in libdevice gets unsafe-fp-math=false
set, thus disabling fastmath in almost all CUDA code.
Reviewers: echristo
Subscribers: hfinkel, llvm-commits, mehdi_amini
Differential Revision: https://reviews.llvm.org/D28538
llvm-svn: 293097
This patch adds support for the proc_bind clause on the Spmd construct
'target parallel' on the NVPTX device. Since the parallel region is created
upon kernel launch, this clause can be safely ignored on the NVPTX device at
codegen time for level 0 parallelism.
Reviewers: ABataev
Differential Revision: https://reviews.llvm.org/D29128
llvm-svn: 293069
Use the new llvm_canonicalize_cmake_booleans() function to canonicalize
booleans for lit tests. Replace the duplicate ENABLE_CLANG* variables
used to hold canonicalized values with in-place canonicalization. Use
implicit logic in Python code to avoid overrelying on exact 0/1 values.
Differential Revision: https://reviews.llvm.org/D28529
llvm-svn: 293052
Summary: This enables the test to run on systems where output cannot be written.
Reviewers: compnerd
Subscribers: cfe-commits
Differential Revision: https://reviews.llvm.org/D29123
llvm-svn: 293051
Prior to OpenCL 2.0, image3d_t can only be used with the write_only
access qualifier when the cl_khr_3d_image_writes extension is enabled,
see e.g. OpenCL 1.1 s6.8b.
Require the extension for write_only image3d_t types and guard uses of
write_only image3d_t in the OpenCL header.
Patch by Sven van Haastregt!
Review: https://reviews.llvm.org/D28860
llvm-svn: 293050
The thread_limit-clause on the combined directive applies to the
'teams' region of this construct. We modify the ThreadLimitClause
class to capture the clause expression within the 'target' region.
Reviewers: ABataev
Differential Revision: https://reviews.llvm.org/D29087
llvm-svn: 293049
The num_teams-clause on the combined directive applies to the
'teams' region of this construct. We modify the NumTeamsClause
class to capture the clause expression within the 'target' region.
Reviewers: ABataev
Differential Revision: https://reviews.llvm.org/D29085
llvm-svn: 293048
This is an attempt to avoid new false positives caused by the reverted r292800,
however the scope of the fix is significantly reduced - some variables are still
in incorrect memory spaces.
Relevant test cases added.
rdar://problem/30105546
rdar://problem/30156693
Differential revision: https://reviews.llvm.org/D28946
llvm-svn: 293043
This reverts commit r293004 because it broke the buildbots with "unknown CPU"
errors. I tried to fix it in r293026, but that broke on Green Dragon with this
kind of error:
error: expected string not found in input
// CHECK: l{{ +}}df{{ +}}*ABS*{{ +}}{{0+}}{{.+}}preprocessed-input.c{{$}}
^
<stdin>:2:1: note: scanning from here
/Users/buildslave/jenkins/sharedspace/incremental@2/clang-build/tools/clang/test/Frontend/Output/preprocessed-input.c.tmp.o: file format Mach-O 64-bit x86-64
^
<stdin>:2:67: note: possible intended match here
/Users/buildslave/jenkins/sharedspace/incremental@2/clang-build/tools/clang/test/Frontend/Output/preprocessed-input.c.tmp.o: file format Mach-O 64-bit x86-64
I suppose this means that llvm-objdump doesn't support Mach-O, so the test
should indeed check for linux (but not for x86). I'll leave it to someone that
knows better.
llvm-svn: 293032
This test broke on a lot of non-x86 buildbots with "unknowm CPU" errors. I don't
see anything platform-specific about this test, and it seems to work fine on ARM
if we just remove the -triple i686 flags from the run line.
llvm-svn: 293026
Sometime clang would be supplied -fobjc-arc -f(no)objc-arc-exceptions
and then later disable ARC with -fno-objc-arc, which only negate first
option, but not the latter, resulting usused argument warning. Silence
this warning only when -fno-objc-arc option is present.
Patch by Onha Choe!
llvm-svn: 293014
This patch adds support for codegen of 'target teams' on the host.
This combined directive has two captured statements, one for the
'teams' region, and the other for the 'parallel'.
This target teams region is offloaded using the __tgt_target_teams()
call. The patch sets the number of teams as an argument to
this call.
Reviewers: ABataev
Differential Revision: https://reviews.llvm.org/D29084
llvm-svn: 293005
Summary:
Clang appears to always use name as specified on the command
line, whereas gcc uses the name as specified in the linemarker at the
first line when compiling a preprocessed source. This results mismatch
between two compilers in FILE symbol table entry. This patch makes clang
to resemble gcc's behavior in finding the original source file name and
use it as an input file name.
Even with this patch, values of FILE symbol table entry may still be
different because clang uses dirname+basename for the entry whlie gcc
uses basename only. I'll write a patch for that once this patch is
committed.
Reviewers: dblaikie, inglorion
Reviewed By: inglorion
Subscribers: inglorion, aprantl, bruno
Differential Revision: https://reviews.llvm.org/D28796
llvm-svn: 293004
Arpith Chacko Jacob