[OPENMP] Simplified checks for declarations in declare target regions.
Sema analysis should not mark functions as an implicit declare target,
it may break codegen. Simplified semantic analysis and removed extra
code for implicit declare target functions.
Summary:
And add an option to disable this behavior. The option is only used in
AllTUsExecutor to avoid races when running concurrently on multiple
threads.
Erich Keane [Mon, 10 Sep 2018 21:12:21 +0000 (21:12 +0000)]
Move AESNI generation to Skylake and Goldmont
The instruction set first appeared with Westmere, but not all processors
in that and the next few generations have the instructions. According to
Wikipedia[1], the first generation in which all SKUs have AES
instructions are Skylake and Goldmont. I can't find any Skylake,
Kabylake, Kabylake-R or Cannon Lake currently listed at
https://ark.intel.com that says "Intel® AES New Instructions" "No".
This matches GCC commit
https://gcc.gnu.org/ml/gcc-patches/2018-08/msg01940.html
Summary:
_Atomic and __sync_* operations are implicitly sequentially-consistent. Some
codebases want to force explicit usage of memory order instead. This warning
allows them to know where implicit sequentially-consistent memory order is used.
The warning isn't on by default because _Atomic was purposefully designed to
have seq_cst as the default: the idea was that it's the right thing to use most
of the time. This warning allows developers who disagree to enforce explicit
usage instead.
A follow-up patch will take care of C++'s std::atomic. It'll be different enough
from this patch that I think it should be separate: for C++ the atomic
operations all have a memory order parameter (or two), but it's defaulted. I
believe this warning should trigger when the default is used, but not when
seq_cst is used explicitly (or implicitly as the failure order for cmpxchg).
This change allows usage of -march when using the clang-cl driver. This is similar to MSVC's /arch; however -march can target precisely all supported CPUs, while /arch has a more restricted set.
Adrian Prantl [Mon, 10 Sep 2018 16:14:28 +0000 (16:14 +0000)]
Remove all uses of DIFlagBlockByrefStruct
This patch removes the last reason why DIFlagBlockByrefStruct from
Clang by directly implementing the drilling into the member type done
in DwarfDebug::DbgVariable::getType() into the frontend.
[clang] Make sure codecompletion is called for calls even when inside a token.
Summary:
Currently CodeCompleteCall only gets called after a comma or parantheses. This
patch makes sure it is called even at the cases like:
```foo(1^);```
Adam Balogh [Mon, 10 Sep 2018 09:07:47 +0000 (09:07 +0000)]
[Analyzer] Iterator Checker - Part 8: Support for assign, clear, insert, emplace and erase operations
This patch adds support for the following operations in the iterator checkers: assign, clear, insert, insert_after, emplace, emplace_after, erase and erase_after. This affects mismatched iterator checks ("this" and parameter must match) and invalidation checks (according to the standard).
Adam Balogh [Mon, 10 Sep 2018 09:06:31 +0000 (09:06 +0000)]
[Analyzer] Iterator Checker - Part 7: Support for push and pop operations
This patch adds support for the following operations in the iterator checkers: push_back, push_front, emplace_back, emplace_front, pop_back and pop_front. This affects iterator range checks (range is extended after push and emplace and reduced after pop operations) and invalidation checks (according to the standard).
Adam Balogh [Mon, 10 Sep 2018 09:05:31 +0000 (09:05 +0000)]
[Analyzer] Iterator Checker - Part 6: Mismatched iterator checker for constructors and comparisons
Extension of the mismatched iterator checker for constructors taking range of first..last (first and last must be iterators of the same container) and also for comparisons of iterators of different containers (one does not compare iterators of different containers, since the set of iterators is partially ordered, there are no relations between iterators of different containers, except that they are always non-equal).
Adam Balogh [Mon, 10 Sep 2018 09:04:27 +0000 (09:04 +0000)]
[Analyzer] Iterator Checker - Part 5: Move Assignment of Containers
If a container is moved by its move assignment operator, according to the standard all their iterators except the past-end iterators remain valid but refer to the new container. This patch introduces support for this case in the iterator checkers.
Adam Balogh [Mon, 10 Sep 2018 09:03:22 +0000 (09:03 +0000)]
[Analyzer] Iterator Checker - Part 4: Mismatched iterator checker for function parameters
New check added to the checker which checks whether iterator parameters of template functions typed by the same template parameter refer to the same container.
Fangrui Song [Sun, 9 Sep 2018 17:20:03 +0000 (17:20 +0000)]
[Sema] Make typo correction slightly more efficient
edit_distance returns UpperBound+1 if the distance will exceed UpperBound. We can subtract 1 from UpperBound and change >= to > in the if condition. The threshold does not change but edit_distance will have more opportunity to bail out earlier.
Hamza Sood [Sun, 9 Sep 2018 12:06:35 +0000 (12:06 +0000)]
[Tooling] Improve handling of CL-style options
This patch fixes the handling of clang-cl options in InterpolatingCompilationDatabase.
They were previously ignored completely, which led to a lot of bugs:
Additional options were being added with the wrong syntax. E.g. a file was
specified as C++ by adding -x c++, which causes an error in CL mode.
The args were parsed and then rendered, which means that the aliasing information
was lost. E.g. /W4 was rendered to -Wall, which in CL mode means -Weverything.
CL options were ignored when checking things like -std=, so a lot of logic was
being bypassed.
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.
Richard Smith [Fri, 7 Sep 2018 23:57:54 +0000 (23:57 +0000)]
Do not use optimized atomic libcalls for misaligned atomics.
Summary:
The optimized (__atomic_foo_<n>) libcalls assume that the atomic object
is properly aligned, so should never be called on an underaligned
object.
This addresses one of several problems identified in PR38846.
[analyzer] Remove the "postponed" hack, deal with derived symbols using an extra map
The "derived" symbols indicate children fields of a larger symbol.
As parents do not have pointers to their children, the garbage
collection algorithm the analyzer currently uses adds such symbols into
a "postponed" category, and then keeps running through the worklist
until the fixed point is reached.
The current patch rectifies that by instead using a helper map which
stores pointers from parents to children, so that no fixed point
calculation is necessary.
The current patch yields ~5% improvement in running time on sqlite.
[X86] Custom emit __builtin_rdtscp so we can emit an explicit store for the out parameter
This is the clang side of D51803. The llvm intrinsic now returns two results. So we need to emit an explicit store in IR for the out parameter. This is similar to addcarry/subborrow/rdrand/rdseed.
[CodeComplete] Clearly distinguish signature help and code completion.
Summary:
Code completion in clang is actually a mix of two features:
- Code completion is a familiar feature. Results are exposed via the
CodeCompleteConsumer::ProcessCodeCompleteResults callback.
- Signature help figures out if the current expression is an argument of
some function call and shows corresponding signatures if so.
Results are exposed via CodeCompleteConsumer::ProcessOverloadCandidates.
This patch refactors the implementation to untangle those two from each
other and makes some naming tweaks to avoid confusion when reading the
code.
The refactoring is required for signature help fixes, see D51038.
The only intended behavior change is the order of callbacks.
ProcessOverloadCandidates is now called before ProcessCodeCompleteResults.
[RISCV] Add support for computing sysroot for riscv32-unknown-elf
Extends r338385 to allow the driver to compute the sysroot when an explicit path is not provided. This allows the linker to find C runtime files and the correct include directory for header files.
Reland r341390 clang-cl: Pass /Brepro to linker if it was passed to the compiler
The test was missing '--' on mac as pointed out by -Wslash-u-filename:
<stdin>:5:69: note: possible intended match here
clang: warning: '/Users/thakis/src/llvm-mono/clang/test/Driver/msvc-link.c' treated as the '/U' option [-Wslash-u-filename]
Boilerplate code for using KMSAN instrumentation in Clang.
We add a new command line flag, -fsanitize=kernel-memory, with a
corresponding SanitizerKind::KernelMemory, which, along with
SanitizerKind::Memory, maps to the memory_sanitizer feature.
KMSAN is only supported on x86_64 Linux.
It's incompatible with other sanitizers, but supports code coverage
instrumentation.
[analyzer] Skip printing trivial nodes in exploded graph
A node is considered to be trivial if it only has one successor, one
predecessor, and a state equal to the predecessor.
Can drastically (> 2x) reduce the size of the generated exploded
graph.
[analyzer] Remove traces of ubigraph visualization
Ubigraph project has been dead since about 2008, and to the best of my
knowledge, no one was using it.
Previously, I wasn't able to launch the existing binary at all.
[OPENMP] Fix PR38823: Do not emit vtable if it is not used in target
context.
If the explicit template instantiation definition defined outside of the
target context, its vtable should not be marked as used. This is true
for other situations where the compiler want to emit vtables
unconditionally.
Pavel Labath [Thu, 6 Sep 2018 17:01:45 +0000 (17:01 +0000)]
Re-commit "Enable DWARF accelerator tables by default when tuning for lldb (-glldb => -gpubnames)""
This recommits r341472, which was reverted due to test failures on macos bots.
The issue was that a macos target implies -glldb which, together with
this patch added a -gpubnames switch where there previously wasn't one.
The intentions of those checks was to check that -gpubnames is not
emitted by default so I add an explicit -ggdb arg to those command lines
to get same behavior on all platforms (the fact that -glldb *does* set
-gpubnames is tested by a separate test).
This is likely an error, so the patch forbids negative arguments in
several OpenCL attributes. Users who really want 4294967294 can still
use it as an unsigned representation.
Richard Smith [Wed, 5 Sep 2018 22:30:37 +0000 (22:30 +0000)]
PR38627: Fix handling of exception specification adjustment for
destructors.
We previously tried to patch up the exception specification after
completing the class, which went wrong when the exception specification
was needed within the class body (in particular, by a friend
redeclaration of the destructor in a nested class). We now mark the
destructor as having a not-yet-computed exception specification
immediately after creating it.
This requires delaying various checks against the exception
specification (where we'd previously have just got the wrong exception
specification, and now find we have an exception specification that we
can't compute yet) when those checks fire while the class is being
defined.
This also exposed an issue that we were missing a CodeSynthesisContext
for computation of exception specifications (otherwise we'd fail to make
the module containing the definition of the class visible when computing
its members' exception specs). Adding that incidentally also gives us a
diagnostic quality improvement.
This has also exposed an pre-existing problem: making the exception
specification evaluation context a non-SFINAE context (as it should be)
results in a bootstrap failure; PR38850 filed for this.
John McCall [Wed, 5 Sep 2018 19:02:00 +0000 (19:02 +0000)]
Add -Wobjc-property-assign-on-object-type.
This is a warning about using 'assign' instead of 'unsafe_unretained'
in Objective-C property declarations. It's off by default because there
isn't consensus in the Objective-C steering group that this is the right
thing to do, but we're nonetheless okay with adding it because there's a
substantial pool of Objective-C programmers who will appreciate the warning.
Fix arm_neon.h and arm_fp16.h generation for compiling with std=c89
Summary:
The inline attribute is not valid for C standard 89. Replace the argument in the generation of header files with __inline, as well adding tests for both header files.
Pavel Labath [Wed, 5 Sep 2018 14:38:44 +0000 (14:38 +0000)]
Enable DWARF accelerator tables by default when tuning for lldb (-glldb => -gpubnames)
Summary:
DWARF v5 accelerator tables provide a considerable performance
improvement for lldb and will make the default -glldb behavior same on
all targets (right now we emit apple tables on apple targets, but these
are not controlled by -gpubnames, only by -glldb).
Sam McCall [Wed, 5 Sep 2018 07:44:02 +0000 (07:44 +0000)]
clang-format: Fix formatting C++ namespaces with preceding 'inline' or 'export' specifier
This fixes formatting namespaces with preceding 'inline' and 'export' (Modules TS) specifiers.
This change fixes namespaces not being identified as such with preceding 'inline' or 'export' specifiers.
Motivation: I was experimenting with the Modules TS (-fmodules-ts) and found it would be useful if clang-format would correctly format 'export namespace'. While making the changes, I noticed that similar issues still exist with 'inline namespace', and addressed them as well.
Richard Smith [Wed, 5 Sep 2018 00:28:57 +0000 (00:28 +0000)]
Allow all supportable non-type attributes to be used with #pragma clang attribute.
Summary:
We previously disallowed use of undocumented attributes with #pragma clang
attribute, but the justification for doing so was weak and it prevented many
reasonable use cases.
Summary:
Added support of creating a hardlink from one file to another file.
After a hardlink is added between two files, both file will have the same:
1. UniqueID (inode)
2. Size
3. Buffer
This will bring replay of compilation closer to the actual compilation. There are instances where clang checks for the UniqueID of the file/header to be loaded which leads to a different behavior during replay as all files have different UniqueIDs.
[x86/SLH] Add a real Clang flag and LLVM IR attribute for Speculative
Load Hardening.
Wires up the existing pass to work with a proper IR attribute rather
than just a hidden/internal flag. The internal flag continues to work
for now, but I'll likely remove it soon.
Most of the churn here is adding the IR attribute. I talked about this
Kristof Beyls and he seemed at least initially OK with this direction.
The idea of using a full attribute here is that we *do* expect at least
some forms of this for other architectures. There isn't anything
*inherently* x86-specific about this technique, just that we only have
an implementation for x86 at the moment.
While we could potentially expose this as a Clang-level attribute as
well, that seems like a good question to defer for the moment as it
isn't 100% clear whether that or some other programmer interface (or
both?) would be best. We'll defer the programmer interface side of this
for now, but at least get to the point where the feature can be enabled
without relying on implementation details.
This also allows us to do something that was really hard before: we can
enable *just* the indirect call retpolines when using SLH. For x86, we
don't have any other way to mitigate indirect calls. Other architectures
may take a different approach of course, and none of this is surfaced to
user-level flags.
projects / clang / log