Since some profiling tools, such as gprof, ftrace, and uftrace, use
-pg option to generate a mcount function call at the entry of each
function. Function invocation can be detected by this hook function.
But mcount insertion is done before function inlining phase in clang,
sometime a function that already has a mcount call can be inlined in the
middle of another function.
This patch adds an attribute "counting-function" to each function
rather than emitting the mcount call directly in frontend so that this
attribute can be processed in backend. Then the mcount calls can be
properly inserted in backend after all the other optimizations are
completed.
Link: https://llvm.org/bugs/show_bug.cgi?id=28660
Reviewers: hans, rjmccall, hfinkel, rengolin, compnerd
Subscribers: shenhan, cfe-commits
Differential Revision: https://reviews.llvm.org/D22666
llvm-svn: 280355
-fprofile-dir=path allows the user to specify where .gcda files should be
emitted when the program is run. In particular, this is the first flag that
causes the .gcno and .o files to have different paths, LLVM is extended to
support this. -fprofile-dir= does not change the file name in the .gcno (and
thus where lcov looks for the source) but it does change the name in the .gcda
(and thus where the runtime library writes the .gcda file). It's different from
a GCOV_PREFIX because a user can observe that the GCOV_PREFIX_STRIP will strip
paths off of -fprofile-dir= but not off of a supplied GCOV_PREFIX.
To implement this we split -coverage-file into -coverage-data-file and
-coverage-notes-file to specify the two different names. The !llvm.gcov
metadata node grows from a 2-element form {string coverage-file, node dbg.cu}
to 3-elements, {string coverage-notes-file, string coverage-data-file, node
dbg.cu}. In the 3-element form, the file name is already "mangled" with
.gcno/.gcda suffixes, while the 2-element form left that to the middle end
pass.
llvm-svn: 280306
Classes with no virtual methods or whose virtual methods were all
inherited from virtual bases don't have a vfptr at offset zero. We were
crashing attempting to get the layout of that non-existent vftable.
We don't need any vshape info in this case because the debugger can
infer it from the base class information. The current class may not
introduce any virtual methods if we are in this situation.
llvm-svn: 280287
The shape is really just the number of methods in the vftable, since we
don't support 16 bit far calls. All calls are near. Encode this number
in the size of the artificial __vtbl_ptr_type DIDerivedType that we
generate. For DWARF, this will be a normal pointer, but for codeview
this will be a wide pointer that gets pattern matched into a
VFTableShape record. Insert this type into the element list of all
dynamic classes when emitting CodeView, so that the backend can emit the
shape even if the vptr lives in a primary base class.
Fixes PR28150
llvm-svn: 280255
In most cases these code regions are just redundant, but sometimes they
could be assigned to the counter of the parent code region instead of
the counter of the nested block.
Differential Revision: https://reviews.llvm.org/D23987
llvm-svn: 280199
to CC1, which are translated to function attributes and can e.g. be mapped on
build attributes FP_exceptions and FP_denormal. Setting these build attributes
allows better selection of floating point libraries.
Differential Revision: https://reviews.llvm.org/D23840
llvm-svn: 280064
The PPC64 DWARF register-size table did not match the ABI specification (or
GCC, for that matter). Fix that, and add a regression test.
Fixes PR27931.
llvm-svn: 280053
If there were several nested statements arranged in a way that all of them
end up with the same macro, then the expansion of this macro was assigned
with all the corresponding counters of these statements.
As a result, the wrong counter value was shown for the macro in llvm-cov.
This patch fixes the issue by preventing adding a counter for an expanded
source range if it already has an assigned counter, which is expected
to come from the most specific statement.
Differential Revision: https://reviews.llvm.org/D23160
llvm-svn: 279962
Otherwise we can't handle secondary base classes at offsets greater than
2**24. This agrees with libstdc++abi.
We could extend this change to other LLP64 platforms, but then we would
want to update libc++abi and it would require additional review.
Fixes PR29116
llvm-svn: 279786
This isn't exactly what MSVC does, unfortunately. MSVC does not pass
objects with destructors but no copy constructors by address. More ARM
expertise is required to really understand what should be done here.
Fixes PR29136.
llvm-svn: 279764
In cases where .dwo/.dwp files are guaranteed to be available, skipping
the extra online (in the .o file) inline info can save a substantial
amount of space - see the original r221306 for more details there.
llvm-svn: 279651
in isDefinedInClangModule() and assume that the incomplete definition
is not defined in the module.
This broke the -gmodules self host recently.
rdar://problem/27894367
llvm-svn: 279485
Structs are currently handled as pointer + byval, which makes AMDGPU
LLVM backend generate incorrect code when structs are used. This patch
changes struct argument to be handled directly and without flattening,
which Clover (Mesa 3D Gallium OpenCL state tracker) will be able to
handle. Flattening would expand the struct to individual elements and
pass each as a separate argument, which Clover can not
handle. Furthermore, such expansion does not fit the OpenCL
programming model which requires to explicitely specify each argument
index, size and memory location.
Patch by Vedran Miletić
llvm-svn: 279463
Summary:
int __builtin_amdgcn_ds_swizzle (int a, int imm);
while imm is a constant.
Differential Revision:
http://reviews.llvm.org/D23682
llvm-svn: 279165
Pointers of certain GPUs in AMDGCN target in private address space is 32 bit but pointers in other address spaces are 64 bit. size_t type should be defined as 64 bit for these GPUs so that it could hold pointers in all address spaces. Also fixed issues in pointer arithmetic codegen by using pointer specific intptr type.
Differential Revision: https://reviews.llvm.org/D23361
llvm-svn: 279121
This reverts commit r279003 as it breaks some of our buildbots (e.g.
clang-cmake-aarch64-quick, clang-x86_64-linux-selfhost-modules).
The error is in OpenMP/teams_distribute_simd_ast_print.cpp:
clang: /home/buildslave/buildslave/clang-cmake-aarch64-quick/llvm/include/llvm/ADT/DenseMap.h:527:
bool llvm::DenseMapBase<DerivedT, KeyT, ValueT, KeyInfoT, BucketT>::LookupBucketFor(const LookupKeyT&, const BucketT*&) const
[with LookupKeyT = clang::Stmt*; DerivedT = llvm::DenseMap<clang::Stmt*, long unsigned int>;
KeyT = clang::Stmt*; ValueT = long unsigned int;
KeyInfoT = llvm::DenseMapInfo<clang::Stmt*>;
BucketT = llvm::detail::DenseMapPair<clang::Stmt*, long unsigned int>]:
Assertion `!KeyInfoT::isEqual(Val, EmptyKey) && !KeyInfoT::isEqual(Val, TombstoneKey) &&
"Empty/Tombstone value shouldn't be inserted into map!"' failed.
llvm-svn: 279045
This patch is to implement sema and parsing for 'teams distribute simd’ pragma.
This patch is originated by Carlo Bertolli.
Differential Revision: https://reviews.llvm.org/D23528
llvm-svn: 279003
The previous condition would erroneously mark all CXXRecordDecls
that didn't have any fields as being defined in a clang module.
This patch fixes the condition to only apply to explicit template
instantiations.
<rdar://problem/27771823>
llvm-svn: 278952
This affects functions with the C++11 [[ noreturn ]] and C11 _Noreturn
specifiers.
Patch by Victor Leschuk!
https://reviews.llvm.org/D23168
llvm-svn: 278942
Use BB.getNextNode(), which returns nullptr on end(), instead of
&*BB.getIterator(), which is UB on end().
CodeGenFunction::createBasicBlock expects nullptr in this case already.
llvm-svn: 278898
With -debug-info-kind=limited, we omit debug info for dynamic classes that live in other TUs. This reduces duplicate type information. When statically linked, the type information comes together. But if your binary has a class derived from a base in a DLL, the base class info is not available to the debugger.
The decision is made in shouldOmitDefinition (CGDebugInfo.cpp). Per a suggestion from rnk, I've tweaked the decision so that we do include definitions for classes marked as DLL imports. This should be a relatively small number of classes, so we don't pay a large price for duplication of the type info, yet it should cover most cases on Windows.
Essentially this makes debug info for DLLs independent, but we still assume that all TUs within the same DLL will be consistently built with (or without) debug info and the debugger will be able to search across the debug info within that scope to resolve any declarations into definitions, etc.
llvm-svn: 278861
Summary:
This patch lets you create diagnostics that are emitted if and only if a
particular FunctionDecl is codegen'ed.
This is necessary for CUDA, where some constructs -- e.g. calls from
host+device functions to host functions when compiling for device -- are
allowed to appear in semantically-correct programs, but only if they're
never codegen'ed.
Reviewers: rnk
Subscribers: cfe-commits, tra
Differential Revision: https://reviews.llvm.org/D23241
llvm-svn: 278735
We processed unnamed bitfields after our logic for non-vector field
elements in records larger than 128 bits. The vector logic would
determine that the bit-field disqualifies the record from occupying a
register despite the unnamed bit-field not participating in the record
size nor its alignment.
N.B. This behavior matches GCC and ICC.
llvm-svn: 278656
An __m512 vector type wrapped in a structure should be passed in a
vector register.
Our prior implementation was based on a draft version of the psABI.
This fixes PR28975.
N.B. The update to the ABI was made here:
https://github.com/hjl-tools/x86-psABI/commit/30f9c9
llvm-svn: 278655
This matches the way nvcc encapsulates GPU binaries into host object file.
Now cuobjdump can deal with clang-compiled object files.
Differential Revision: https://reviews.llvm.org/D23429
llvm-svn: 278549
Summary:
This changes clang to use the llvm::lto::thinBackend function instead of
its own less comprehensive ThinLTO backend implementation.
Patch by Peter Collingbourne
Reviewers: tejohnson, mehdi_amini
Subscribers: cfe-commits, mehdi_amini
Differential Revision: https://reviews.llvm.org/D21545
llvm-svn: 278541
constraints were added to _mm256_broadcast_{pd,ps} intel intrinsics.
The spec for these intrinics is ... pretty much silent on alignment.
This is especially frustrating considering the amount of discussion of
alignment in the load and store instrinsics. So I was forced to rely on
the specification for the VBROADCASTF128 instruction.
That instruction's spec is *also* completely silent on alignment.
Fortunately, when it comes to the instruction's spec, silence is enough.
There is no #GP fault option for an underaligned address so this
instruction, and by inference the intrinsic, can read any alignment.
As it happens, the old code worked exactly this way and in fact we have
plenty of code that hands pointers with less than 16-byte alignment to
these intrinsics. This code broke pretty spectacularly with this commit.
Fortunately, the fix is super simple! Change a 16 to a 1, and ta da!
Anyways, a lot of debugging for a really boring fix. =]
llvm-svn: 278202
Let the driver pass the option to frontend. Do not set precision metadata for division instructions when this option is set. Set function attribute "correctly-rounded-divide-sqrt-fp-math" based on this option.
Differential Revision: https://reviews.llvm.org/D22940
llvm-svn: 278155
Summary:
Based on a patch by Michael Mueller.
This attribute specifies that a function can be hooked or patched. This
mechanism was originally devised by Microsoft for hotpatching their
binaries (which they're constantly updating to stay ahead of crackers,
script kiddies, and other ne'er-do-wells on the Internet), but it's now
commonly abused by Windows programs that want to hook API functions. It
is for this reason that this attribute was added to GCC--hence the name,
`ms_hook_prologue`.
Depends on D19908.
Reviewers: rnk, aaron.ballman
Subscribers: cfe-commits
Differential Revision: https://reviews.llvm.org/D19909
llvm-svn: 278050
This patch (with the corresponding ARM backend patch) adds support for
some new relocation models:
* Read-only position independence (ROPI): Code and read-only data is accessed
PC-relative. The offsets between all code and RO data sections are known at
static link time.
* Read-write position independence (RWPI): Read-write data is accessed relative
to a static base register. The offsets between all writeable data sections
are known at static link time.
These two modes are independent (they specify how different objects
should be addressed), so they can be used individually or together.
These modes are intended for bare-metal systems or systems with small
real-time operating systems. They are designed to avoid the need for a
dynamic linker, the only initialisation required is setting the static
base register to an appropriate value for RWPI code.
There is one C construct not currently supported by these modes: global
variables initialised to the address of another global variable or
function, where that address is not known at static-link time. There are
a few possible ways to solve this:
* Disallow this, and require the user to write their own initialisation
function if they need variables like this.
* Emit dynamic initialisers for these variables in the compiler, called from
the .init_array section (as is currently done for C++ dynamic initialisers).
We have a patch to do this, described in my original RFC email
(http://lists.llvm.org/pipermail/llvm-dev/2015-December/093022.html), but the
feedback from that RFC thread was that this is not something that belongs in
clang.
* Use a small dynamic loader to fix up these variables, by adding the
difference between the load and execution address of the relevant section.
This would require linker co-operation to generate a table of addresses that
need fixing up.
Differential Revision: https://reviews.llvm.org/D23196
llvm-svn: 278016
For now just disregard the using declaration in this case. Suboptimal,
but wiring up the ability to have declarations of functions that are
separate from their definition (we currently only do that for member
functions) and have differing return types (we don't have any support
for that) is more work than seems reasonable to at least fix this crash.
llvm-svn: 277852
