Vedant Kumar [Fri, 5 Oct 2018 20:37:17 +0000 (20:37 +0000)]
[DebugInfo] Add support for DWARF5 call site-related attributes
DWARF v5 introduces DW_AT_call_all_calls, a subprogram attribute which
indicates that all calls (both regular and tail) within the subprogram
have call site entries. The information within these call site entries
can be used by a debugger to populate backtraces with synthetic tail
call frames.
Tail calling frames go missing in backtraces because the frame of the
caller is reused by the callee. Call site entries allow a debugger to
reconstruct a sequence of (tail) calls which led from one function to
another. This improves backtrace quality. There are limitations: tail
recursion isn't handled, variables within synthetic frames may not
survive to be inspected, etc. This approach is not novel, see:
This patch adds an IR-level flag (DIFlagAllCallsDescribed) which lowers
to DW_AT_call_all_calls. It adds the minimal amount of DWARF generation
support needed to emit standards-compliant call site entries. For easier
deployment, when the debugger tuning is LLDB, the DWARF requirement is
adjusted to v4.
Testing: Apart from check-{llvm, clang}, I built a stage2 RelWithDebInfo
clang binary. Its dSYM passed verification and grew by 1.4% compared to
the baseline. 151,879 call site entries were added.
Nico Weber [Fri, 5 Oct 2018 18:22:21 +0000 (18:22 +0000)]
clang-format: Don't insert spaces in front of :: for Java 8 Method References.
The existing code kept the space if it was there for identifiers, and it didn't
handle `this`. After this patch, for Java `this` is handled in addition to
identifiers, and existing space is always stripped between identifier and `::`.
Also accept `::` in addition to `.` in front of `<` in `foo::<T>bar` generic
calls.
James Y Knight [Fri, 5 Oct 2018 17:49:48 +0000 (17:49 +0000)]
Emit diagnostic note when calling an invalid function declaration.
The comment said it was intentionally not emitting any diagnostic
because the declaration itself was already diagnosed. However,
everywhere else that wants to not emit a diagnostic without an extra
note emits note_invalid_subexpr_in_const_expr instead, which gets
suppressed later.
This was the only place which did not emit a diagnostic note.
Alexey Bataev [Fri, 5 Oct 2018 15:08:53 +0000 (15:08 +0000)]
[OPENMP] Fix emission of the __kmpc_global_thread_num.
Fixed emission of the __kmpc_global_thread_num() so that it is not
messed up with alloca instructions anymore. Plus, fixes emission of the
__kmpc_global_thread_num() functions in the target outlined regions so
that they are not called before runtime is initialized.
Summary:
Instead of only examining call arguments, we also examine constructor
arguments applying the same rules.
That was an opportunity for refactoring the examination procedure to
work with iterators instead of integer indices. For the case of
CallExprs no functional change is intended.
Reid Kleckner [Thu, 4 Oct 2018 20:34:52 +0000 (20:34 +0000)]
Update Clang Windows getting started docs
Summary:
- Update the example VS project generation to use VS2017.
- Add docs for generating ninja build files, since they are popular.
- Remove reference to "make update" which no longer exists. Mention the
monorepo instead.
- Try to explain gnuwin32/coreutils requirements better.
- Use https:// links where possible
Shoaib Meenai [Thu, 4 Oct 2018 19:50:14 +0000 (19:50 +0000)]
[AST] Revert mangling changes from r339428
As discussed in https://reviews.llvm.org/D50144, we want Obj-C classes
to have the same mangling as C++ structs, to support headers like the
following:
```
@class I;
struct I;
void f(I *);
```
since the header can be used from both C++ and Obj-C++ TUs, and we want
a consistent mangling across the two to prevent link errors. Itanium
mangles both the same way, and so should the MS ABI.
The main concern with having the same mangling for C++ structs and Obj-C
classes was that we want to treat them differently for the purposes of
exception handling, e.g. we don't want a C++ catch statement for a
struct to be able to catch an Obj-C class with the same name as the
struct. We can accomplish this by mangling Obj-C class names differently
in their RTTI, which I'll do in a follow-up patch.
Louis Dionne [Thu, 4 Oct 2018 15:49:42 +0000 (15:49 +0000)]
[clang] Add the exclude_from_explicit_instantiation attribute
Summary:
This attribute allows excluding a member of a class template from being part
of an explicit template instantiation of that class template. This also makes
sure that code using such a member will not take for granted that an external
instantiation exists in another translation unit. The attribute was discussed
on cfe-dev at [1] and is primarily motivated by the removal of always_inline
in libc++ to control what's part of the ABI (see links in [1]).
Petr Pavlu [Thu, 4 Oct 2018 09:25:44 +0000 (09:25 +0000)]
[constexpr] Fix ICE when memcpy() is given a pointer to an incomplete array
Fix code for constant evaluation of __builtin_memcpy() and
__builtin_memmove() that would attempt to divide by zero when given two
pointers to an incomplete array.
[SEMA] split ExtWarn dupl-decl-spec's into Extension and ExtWarn
Summary:
For types deduced from typedef's and typeof's, don't warn for duplicate
declaration specifiers in C90 unless -pedantic.
Create a third diagnostic type for duplicate declaration specifiers.
Previously, we had an ExtWarn and a Warning. This change adds a third,
Extension, which only warns when -pedantic is set, staying silent
otherwise.
Artem Dergachev [Wed, 3 Oct 2018 22:48:00 +0000 (22:48 +0000)]
[analyzer] [tests] [quickfix] Make more test more resilient for non-defaut -std.
It is important to specify the version of the standard because tests should
test the same thing regardless of the current default version of the standard.
Matthew Voss [Wed, 3 Oct 2018 18:45:04 +0000 (18:45 +0000)]
Add template type and value parameter metadata nodes to template variable specializations
Summary: Add an optional attribute referring to a tuple of type and value template parameter nodes to the DIGlobalVariable node. This allows us to record the parameters of template variable specializations.
Martin Storsjo [Wed, 3 Oct 2018 18:24:05 +0000 (18:24 +0000)]
[test] Use --sysroot instead of -B in print-multi-directory.c
This avoids finding a similar matching GCC installation outside
of the test directory tree in the surrounding environment, which
would make the test fail. (This happened on Ubuntu 16.04.)
Aaron Puchert [Wed, 3 Oct 2018 11:58:19 +0000 (11:58 +0000)]
Thread safety analysis: Unwrap __builtin_expect in getTrylockCallExpr
Summary:
When people are really sure they'll get the lock they sometimes use
__builtin_expect. It's also used by some assertion implementations.
Asserting that try-lock succeeded is basically the same as asserting
that the lock is not held by anyone else (and acquiring it).
Reid Kleckner [Tue, 2 Oct 2018 22:23:27 +0000 (22:23 +0000)]
Remove spurious assertion in -Wdocumentation
Some code in OpenCV uses interesting doxygen directives that make it so
we don't see any tokens inside a @note, despite there definitely being
non-whitespace characters there. The consistency check isn't needed.
ParagraphComment supports receiving an empty list of comments.
Artem Dergachev [Tue, 2 Oct 2018 20:48:12 +0000 (20:48 +0000)]
[analyzer] Fix symbol discovery in nonloc::LocAsInteger values.
Doesn't do much despite sounding quite bad, but fixes an exotic test case where
liveness of a nonloc::LocAsInteger array index is now evaluated correctly.
Kristof Umann [Tue, 2 Oct 2018 19:27:34 +0000 (19:27 +0000)]
[analyzer][NFC] Refactor functions in PlistDiagnostics to take AnalyzerOptions as parameter
I intend to add a new flag macro-expnasions-as-events, and unfortunately
I'll only be able to convert the macro piece into an event one once I'm
about to emit it, due to the lack of an avaible Preprocessor object in
the BugReporter.
Jonas Hahnfeld [Tue, 2 Oct 2018 19:12:54 +0000 (19:12 +0000)]
[OpenMP][NVPTX] Simplify codegen for orphaned parallel, NFCI.
Worker threads fork off to the compiler generated worker function
directly after entering the kernel function. Hence, there is no
need to check whether the current thread is the master if we are
outside of a parallel region (neither SPMD nor parallel_level > 0).
Jonas Hahnfeld [Tue, 2 Oct 2018 19:12:47 +0000 (19:12 +0000)]
[OpenMP] Simplify code for reductions on distribute directives, NFC.
Only need to care about the 'distribute simd' case, all other composite
directives are handled elsewhere. This was already reflected in the
outer 'if' condition, so all other inner conditions could never be true.
Yaxun Liu [Tue, 2 Oct 2018 17:48:54 +0000 (17:48 +0000)]
[HIP] Support early finalization of device code for -fno-gpu-rdc
This patch renames -f{no-}cuda-rdc to -f{no-}gpu-rdc and keeps the original
options as aliases. When -fgpu-rdc is off,
clang will assume the device code in each translation unit does not call
external functions except those in the device library, therefore it is possible
to compile the device code in each translation unit to self-contained kernels
and embed them in the host object, so that the host object behaves like
usual host object which can be linked by lld.
The benefits of this feature is: 1. allow users to create static libraries which
can be linked by host linker; 2. amortized device code linking time.
This patch modifies HIP action builder to insert actions for linking device
code and generating HIP fatbin, and pass HIP fatbin to host backend action.
It extracts code for constructing command for generating HIP fatbin as
a function so that it can be reused by early finalization. It also modifies
codegen of HIP host constructor functions to embed the device fatbin
when it is available.
Summary:
Some macros are used in the body of function, and actually contain the trailing semicolon: they should thus be automatically followed by a new line, and not get merged with the next line. This is for example the case with Qt's Q_UNUSED macro:
void foo(int a, int b) {
Q_UNUSED(a)
return b;
}
This patch deals with these cases by introducing a new option to specify list of statement macros. This re-uses the system already in place for foreach macros, to ensure there is no impact on performance.
Haojian Wu [Tue, 2 Oct 2018 14:42:51 +0000 (14:42 +0000)]
[Preprocesssor] Filename should fall back to the written name when typo correction fails.
Summary:
The test is added in Testcase is at https://reviews.llvm.org/D52775. I tried to add the test to clang's code
completion test, it doesn't reproduce the crash.
r326937 ("[OpenCL] Remove block invoke function from emitted block
literal struct", 2018-03-07) broke block argument handling. In
particular the commit was causing a crash during code generation, see
the discussion in https://reviews.llvm.org/D43783 .
The offending commit has just been reverted; add a test to avoid
breaking this again in the future.
Summary:
Several improvements in preparation for the new backends.
Refactoring:
- Removed duplicated methods `fromBoolean`, `fromAPSInt`, `fromInt` and `fromAPFloat`. The methods `mkBoolean`, `mkBitvector` and `mkFloat` are now used instead.
- The names of the functions that convert BVs to FPs were swapped (`mkSBVtoFP`, `mkUBVtoFP`, `mkFPtoSBV`, `mkFPtoUBV`).
- Added a couple of comments in function calls.
Crosscheck encoding:
- Changed how constraints are encoded in the refutation manager so it doesn't start with (false OR ...). This change introduces one duplicated line (see file `BugReporterVisitors.cpp`, the `SMTConv::getRangeExpr is called twice, so I can remove this change if the duplication is a problem.
Bruno Ricci [Tue, 2 Oct 2018 11:46:38 +0000 (11:46 +0000)]
[AST] Pack the bit-fields of FunctionProtoType into Type.
Move the bit-fields of FunctionProtoType into FunctionTypeBitfields.
This cuts the size of FunctionProtoType by a pointer. Additionally use
llvm::TrailingObjects instead of manually doing the casts + arithmetic.
This patch is bigger then what could be expected for the following reasons:
1. As discussed before in D50631 it would be nice if there was some space left
in FunctionTypeBitfields for future additions. This patch introduces an
extra structure FunctionTypeExtraBitfields which is supposed to hold
uncommon bits and is stored in a trailing object. The number of exception
types NumExceptions is moved to this struct. As of this patch this trailing
struct will only be allocated if we have > 0 types in a dynamic exception
specification.
2. TrailingObjects cannot handle repeated types. Therefore the QualType
representing an exception type is wrapped in a struct ExceptionType.
The ExceptionType * is then reinterpret_cast'd to QualType *.
3. TrailingObjects needs the definition of the various trailing classes.
Therefore ExtParameterInfo, ExceptionType and FunctionTypeExtraBitfields
are put in FunctionType.
Eric Liu [Tue, 2 Oct 2018 10:29:00 +0000 (10:29 +0000)]
[CodeComplete] Re-fix accessibilty of protected members from base class.
Summary:
The initial fix (r337453) had bug and was partially reverted (r338255).
This simplies the original fix by explicitly passing the naming class to the
completion consumer.
Kadir Cetinkaya [Tue, 2 Oct 2018 09:42:31 +0000 (09:42 +0000)]
[clang] Implement Override Suggestions in Sema.
Summary:
In clangd we had a new type of completion suggestions for cpp
class/struct/unions that will show override signatures for virtual methods in
base classes. This patch implements it in sema because it is hard to deduce more
info about completion token outside of Sema and handle itchy cases.
See the patch D50898 for more info on the functionality.
In addition to above patch this one also converts the suggestion into a
CK_Pattern with whole insertion text as the name of the suggestion and factors
out CodeCompletionString generation for declerations so that it can be re-used
by others.
Akira Hatanaka [Tue, 2 Oct 2018 01:00:44 +0000 (01:00 +0000)]
[CodeGen] Before entering the loop that copies a non-trivial array field
of a non-trivial C struct, copy the preceding trivial fields that
haven't been copied.
This commit fixes a bug where the instructions used to copy the
preceding trivial fields were emitted inside the loop body.
Akira Hatanaka [Mon, 1 Oct 2018 21:51:28 +0000 (21:51 +0000)]
Distinguish `__block` variables that are captured by escaping blocks
from those that aren't.
This patch changes the way __block variables that aren't captured by
escaping blocks are handled:
- Since non-escaping blocks on the stack never get copied to the heap
(see https://reviews.llvm.org/D49303), Sema shouldn't error out when
the type of a non-escaping __block variable doesn't have an accessible
copy constructor.
- IRGen doesn't have to use the specialized byref structure (see
https://clang.llvm.org/docs/Block-ABI-Apple.html#id8) for a
non-escaping __block variable anymore. Instead IRGen can emit the
variable as a normal variable and copy the reference to the block
literal. Byref copy/dispose helpers aren't needed either.
This reapplies r343518 after fixing a use-after-free bug in function
Sema::ActOnBlockStmtExpr where the BlockScopeInfo was dereferenced after
it was popped and deleted.
Martin Storsjo [Mon, 1 Oct 2018 20:53:25 +0000 (20:53 +0000)]
[MinGW] Allow using ASan
Linking to ASan for MinGW is similar to MSVC, but MinGW always links
the MSVCRT dynamically, so there is only one of the MSVC cases to
consider.
When linking to a shared compiler runtime library on MinGW, the suffix
of the import library is .dll.a.
The existing case of .dll as suffix for windows in general doesn't
seem correct (since this is used for linking). As long as callers never
actually set the Shared flag, the default static suffix of .lib also
worked fine for import libraries as well.
Akira Hatanaka [Mon, 1 Oct 2018 18:50:14 +0000 (18:50 +0000)]
Distinguish `__block` variables that are captured by escaping blocks
from those that aren't.
This patch changes the way __block variables that aren't captured by
escaping blocks are handled:
- Since non-escaping blocks on the stack never get copied to the heap
(see https://reviews.llvm.org/D49303), Sema shouldn't error out when
the type of a non-escaping __block variable doesn't have an accessible
copy constructor.
- IRGen doesn't have to use the specialized byref structure (see
https://clang.llvm.org/docs/Block-ABI-Apple.html#id8) for a
non-escaping __block variable anymore. Instead IRGen can emit the
variable as a normal variable and copy the reference to the block
literal. Byref copy/dispose helpers aren't needed either.
This reapplies r341754, which was reverted in r341757 because it broke a
couple of bots. r341754 was calling markEscapingByrefs after the call to
PopFunctionScopeInfo, which caused the popped function scope to be
cleared out when the following code was compiled, for example:
$ cat test.m
struct A {
id data[10];
};
void foo() {
__block A v;
^{ (void)v; };
}
This commit calls markEscapingByrefs before calling PopFunctionScopeInfo
to prevent that from happening.
The initializer of `ParameterCount` was updated from 1 to 0 in r175165,
but the comment was never touched:
https://github.com/llvm-mirror/clang/commit/9fc56f2636137fcde8acb38865555ed6c7b84dfd
Haojian Wu [Mon, 1 Oct 2018 12:16:38 +0000 (12:16 +0000)]
Build clang-headers when building clang tools.
Summary:
clang tools require clang headers to work on real project, e.g. when we
build clangd via `ninja clangd`, we expect the binary can run on
real-world project (without running another command `ninja clang-headers`).
Fangrui Song [Sun, 30 Sep 2018 21:41:11 +0000 (21:41 +0000)]
Use the container form llvm::sort(C, ...)
There are a few leftovers of rC343147 that are not (\w+)\.begin but in
the form of ([-[:alnum:]>.]+)\.begin or spanning two lines. Change them
to use the container form in this commit. The 12 occurrences have been
inspected manually for safety.
Eric Fiselier [Sun, 30 Sep 2018 18:05:39 +0000 (18:05 +0000)]
Fix linkage error on ProgramPoint's dump method.
Currently, ProgramPoint::dump calls the out-of-line function ProgramPoint::print. This causes
libraries which include ProgramPoint.h to become dependent on libclangAnalysis, which in turn
causes missing symbol link error when building with -DBUILD_SHARED_LIBS=ON -DLLVM_ENABLE_MODULES=ON.
The breakage was introduced in r343160.
This patch fixes the issues by moving ProgramPoint::dump's declaration out of line.
[OPENMP]Fix PR39084: Check datasharing attributes of reduction variables only.
According to OpenMP, the reduction item must be shared in parent region.
But the item can be an array section or array subscript. In this case,
we should not check for the datasharing of the base declaration.
[X86] Add the movbe instruction intrinsics from icc.
These intrinsics exist in icc. They can be found on the Intel Intrinsics Guide website.
All the backend support is in place to pattern match a load+bswap or a bswap+store pattern to the MOVBE instructions. So we just need to get the frontend to emit the correct IR. The pointer arguments in icc are declared as void so I had to jump through a packed struct to forcing a specific alignment on the load/store. Same trick we use in the unaligned vector load/store intrinsics
projects / clang / log