This avoids unneeded copies when using a range-based for loops.
This avoids new warnings due to D68912 adds -Wrange-loop-analysis to -Wall.
Differential Revision: https://reviews.llvm.org/D71526
Summary:
We currently have some very basic LLVM-style RTTI support in the ExternalASTSource class hierarchy
based on the `SemaSource` bool( to discriminate it form the ExternalSemaSource). As ExternalASTSource
is supposed to be subclassed we should have extendable LLVM-style RTTI in this class hierarchy to make life easier
for projects building on top of Clang.
Most notably the current RTTI implementation forces LLDB to implement RTTI for its
own ExternalASTSource class (ClangExternalASTSourceCommon) by keeping a global set of
ExternalASTSources that are known to be ClangExternalASTSourceCommon. Projects
using Clang currently have to dosimilar workarounds to get RTTI support for their subclasses.
This patch turns this into full-fledged LLVM-style RTTI based on a static `ID` variable similar to
other LLVM class hierarchies. Also removes the friend declaration from ExternalASTSource to
its child class that was only used to grant access to the `SemaSource` member.
Reviewers: aprantl, dblaikie, rjmccall
Reviewed By: aprantl
Subscribers: riccibruno, labath, lhames, cfe-commits
Tags: #clang
Differential Revision: https://reviews.llvm.org/D71397
This patch introduces the namespaces for the configured functions and
also enables the use of the member functions.
I added an optional Scope field for every configured function. Functions
without Scope match for every function regardless of the namespace.
Functions with Scope will match if the full name of the function starts
with the Scope.
Multiple functions can exist with the same name.
Differential Revision: https://reviews.llvm.org/D70878
We may wish to keep the PGO training data outside the repository. Add a
CMake variable to allow referencing an external lit testsuite.
Differential Revision: https://reviews.llvm.org/D71507
In revision 139006ceb6 the Serialization
folder got its first def file 'TypeBitCodes.def'. This broke the
modules build as this .def file was not textually included but implicitly
converted into a module due to our umbrella directive.
This patch fixes this by explicitly marking the .def file as textual.
Similar motivations to the movement of ASTRecordReader:
AbstractBasicWriter.h already has quite a few dependencies,
and it's going to get pretty large as we generate more and more
into it. Meanwhile, most clients don't depend on this detail of
the implementation and shouldn't need to be recompiled.
I've also made OMPClauseWriter private, like it belongs.
AbstractBasicReader.h has quite a few dependencies already,
and that's only likely to increase. Meanwhile, ASTRecordReader
is really an implementation detail of the ASTReader that is only
used in a small number of places.
I've kept it in a public header for the use of projects like Swift
that might want to plug in to Clang's serialization framework.
I've also moved OMPClauseReader into an implementation file,
although it can't be made private because of friendship.
On MSVC, friend declarations are (incorrectly) visible even if
not otherwise declared, which causes them to interfere with
lookup. ASTTypeWriter is actually in an anonymous namespace
and cannot be ASTWriter's friend. The others simply don't need
to be anymore.
The basic technical design here is that we have three levels
of readers and writers:
- At the lowest level, there's a `Basic{Reader,Writer}` that knows
how to emit the basic structures of the AST. CRTP allows this to
be metaprogrammed so that the client only needs to support a handful
of primitive types (e.g. `uint64_t` and `IdentifierInfo*`) and more
complicated "inline" structures such as `DeclarationName` can just
be emitted in terms of those primitives.
In Clang's binary-serialization code, these are
`ASTRecord{Reader,Writer}`. For now, a large number of basic
structures are still emitted explicitly by code on those classes
rather than by either TableGen or CRTP metaprogramming, but I
expect to move more of these over.
- In the middle, there's a `Property{Reader,Writer}` which is
responsible for processing the properties of a larger object. The
object-level reader/writer asks the property-level reader/writer to
project out a particular property, yielding a basic reader/writer
which will be used to read/write the property's value, like so:
```
propertyWriter.find("count").writeUInt32(node->getCount());
```
Clang's binary-serialization code ignores this level (it uses
the basic reader/writer as the property reader/writer and has the
projection methods just return `*this`) and simply relies on the
roperties being read/written in a stable order.
- At the highest level, there's an object reader/writer (e.g.
`Type{Reader,Writer}` which emits a logical object with properties.
Think of this as writing something like a JSON dictionary literal.
I haven't introduced support for bitcode abbreviations yet --- it
turns out that there aren't any operative abbreviations for types
besides the QualType one --- but I do have some ideas of how they
should work. At any rate, they'll be necessary in order to handle
statements.
I'm sorry for not disentangling the patches that added basic and type
reader/writers; I made some effort to, but I ran out of energy after
disentangling a number of other patches from the work.
Negligible impact on module size, time to build a set of about 20
fairly large modules, or time to read a few declarations out of them.
There are three significant changes here:
- Most of the methods to read various embedded structures (`APInt`,
`NestedNameSpecifier`, `DeclarationName`, etc.) have been moved
from `ASTReader` to `ASTRecordReader`. This cleans up quite a
bit of code which was passing around `(F, Record, Idx)` arguments
everywhere or doing explicit indexing, and it nicely parallels
how it works on the writer side. It also sets us up to then move
most of these methods into the `BasicReader`s that I'm introducing
as part of abstract serialization.
As part of this, several of the top-level reader methods (e.g.
`readTypeRecord`) have been converted to use `ASTRecordReader`
internally, which is a nice readability improvement.
- I've standardized most of these method names on `readFoo` rather
than `ReadFoo` (used in some of the helper structures) or `GetFoo`
(used for some specific types for no apparent reason).
- I've changed a few of these methods to return their result instead
of reading into an argument passed by reference. This is partly
for general consistency and partly because it will make the
metaprogramming easier with abstract serialization.
the tblgen AST node hierarchies.
Not totally NFC because both of the emitters now emit in a different
order. The type-nodes emitter now visits nodes in hierarchy order,
which means we could use range checks in classof if we had any types
that would benefit from that; currently we do not. The AST-nodes
emitter now uses a multimap keyed by the name of the record; previously
it was using `Record*`, which of couse isn't stable across processes
and may have led to non-reproducible builds in some circumstances.
I want to pass some CMake cache files in CLANG_BOOTSTRAP_CMAKE_ARGS as
`-C <cache file>.cmake` arguments. I want to be able to use the values
of the bootstrap passthrough variables in the cache files, so the cache
file arguments need to be after passthrough variables. This should be
safe because the values of passthrough variables are all constants and
can't refer to values in CLANG_BOOTSTRAP_CMAKE_ARGS.
Differential Revision: https://reviews.llvm.org/D71428
This matches https://gcc.gnu.org/onlinedocs/gcc/AArch64-Options.html
> -momit-leaf-frame-pointer
> -mno-omit-leaf-frame-pointer
>
> Omit or keep the frame pointer in leaf functions. The former behavior is the default.
-mno-omit-leaf-frame-pointer is currently a no-op because
TargetOptions::DisableFramePointerElim is only considered for non-leaf
functions.
Reviewed By: efriedma
Differential Revision: https://reviews.llvm.org/D71167
D39317 made clang use .init_array when no gcc installations is found.
This change changes all gcc installations to use .init_array .
GCC 4.7 by default stopped providing .ctors/.dtors compatible crt files,
and stopped emitting .ctors for __attribute__((constructor)).
.init_array should always work.
FreeBSD rules are moved to FreeBSD.cpp to make Generic_ELF rules clean.
Reviewed By: rnk
Differential Revision: https://reviews.llvm.org/D71434
When parsing the code with OpenMP and the function's body must be
skipped, need to skip also OpenMP annotation tokens. Otherwise the
counters for braces/parens are unbalanced and parsing fails.
Copy the block to the heap before passing it to the callee in case the
block escapes in the callee.
rdar://problem/55683462
Differential Revision: https://reviews.llvm.org/D71431
In looking into some other code, I came across this issue where a
float converted to a gcc integer vector via a splat causes it to miss
the float-to-integral cast, which causes some REALLY strange codegen
bugs.
The AST looked like:
`-ImplicitCastExpr <col:13>
'gcc_int_2':'__attribute__((__vector_size__(2 * sizeof(int)))) int' <VectorSplat>
`-ImplicitCastExpr <col:13> 'float' <LValueToRValue>
`-DeclRefExpr <col:13> 'float' lvalue ParmVar
0x556f16a5dc90 'f' 'float'
Despite the type of the VectorSplat cast as printed, it ended up
becoming a vector of float, which caused non-matching instructions. For
example, IntVector + a float constant resulted in:
add <2 x i32> %8, <2 x float> <float 3.000000e+00, float 3.000000e+00>
This patch corrects the conversion so that the float is first converted
to an integral, THEN splatted.
Summary:
I overstepped my reach and generated too many intrinsics; these never
made it into the tests.
Remove these extras. Some needed to be signed-olny, and there were some
possible but unrequired _x variants that needed an extra argument to
IntrinsicMX to allow [de-]selection at compile-time.
Reviewers: simon_tatham
Subscribers: kristof.beyls, dmgreen, cfe-commits
Tags: #clang
Differential Revision: https://reviews.llvm.org/D71466
Summary: Useful when positions are used to target nodes, with before/after ambiguity.
Reviewers: ilya-biryukov, kbobyrev
Subscribers: cfe-commits
Tags: #clang
Differential Revision: https://reviews.llvm.org/D71356
Summary:
I overstepped my reach and generated too many intrinsics; these never
made it into the tests.
Remove these extras. Some needed to be signed-olny, and there were some
possible but unrequired _x variants that needed an extra argument to
IntrinsicMX to allow [de-]selection at compile-time.
Reviewers: simon_tatham
Subscribers: kristof.beyls, dmgreen, cfe-commits
Tags: #clang
Differential Revision: https://reviews.llvm.org/D71466
GEP index size can be specified in the DataLayout, introduced in D42123. However, there were still places
in which getIndexSizeInBits was used interchangeably with getPointerSizeInBits. This notably caused issues
with Instcombine's visitPtrToInt; but the unit tests was incorrect, so this remained undiscovered.
This fixes the buildbot failures.
Differential Revision: https://reviews.llvm.org/D68328
Patch by Joseph Faulls!
Summary:
This patch adds intrinsics for the following MVE instructions:
* VABAV
* VMLADAV, VMLSDAV
* VMLALDAV, VMLSLDAV
* VRMLALDAVH, VRMLSLDAVH
Each of the above 4 groups has a corresponding new LLVM IR intrinsic,
since the instructions cannot be easily represented using
general-purpose IR operations.
Reviewers: simon_tatham, ostannard, dmgreen, MarkMurrayARM
Reviewed By: MarkMurrayARM
Subscribers: merge_guards_bot, kristof.beyls, hiraditya, cfe-commits, llvm-commits
Tags: #clang, #llvm
Differential Revision: https://reviews.llvm.org/D71062
Summary:
This fills in the remaining shift operations that take a single vector
input and an immediate shift count: the `vqshl`, `vqshlu`, `vrshr` and
`vshll[bt]` families.
`vshll[bt]` (which shifts each input lane left into a double-width
output lane) is the most interesting one. There are separate MC
instruction ids for shifting by exactly the input lane width and
shifting by less than that, because the instruction encoding is so
completely different for the lane-width special case. So I had to
write two sets of patterns to match based on the immediate shift
count, which involved adding a ComplexPattern matcher to avoid the
general-case pattern accidentally matching the special case too. For
that family I've made sure to add an llc codegen test for both
versions of each instruction.
I'm experimenting with a new strategy for parametrising the isel
patterns for all these instructions: adding extra fields to the
relevant `Instruction` subclass itself, which are ignored by the
Tablegen backends that generate the MC data, but can be retrieved from
each instance of that instruction subclass when it's passed as a
template parameter to the multiclass that generates its isel patterns.
A nice effect of that is that I can fill in those informational fields
using `let` blocks, rather than having to type them out once per
instruction at `defm` time.
(As a result, quite a lot of existing instruction `def`s are
reindented by this patch, so it's clearer to read with whitespace
changes ignored.)
Reviewers: dmgreen, MarkMurrayARM, miyuki, ostannard
Reviewed By: MarkMurrayARM
Subscribers: kristof.beyls, hiraditya, cfe-commits, llvm-commits
Tags: #clang, #llvm
Differential Revision: https://reviews.llvm.org/D71458
Allow sending address spaces into diagnostics to simplify and improve
error reporting. Improved wording of diagnostics for address spaces
in overloading.
Tags: #clang
Differential Revision: https://reviews.llvm.org/D71111
Summary:
Better use of multiclass is used, and this helped find some existing
bugs in the predicated VMULL* intrinsics, which are now fixed.
The refactored VMULL[TB]Q_(INT|POLY)_M() intrinsics were discovered
to have an argument ("inactive") with incorrect type, and this required
a fix that is included in this whole patch. The argument "inactive"
should have been the same width (per vector element) as the return
type of the intrinsic, but was not in the case where the return type
was double the element width of the input types.
To assist in testing the multiclassing , and to thwart further gremlins,
the unit tests are improved in scope.
The *.ll tests are all generated by a small bit of throw-away scripting
from the corresponding *.c tests, and as such the diffs are large and
nasty. Look at the file rather than the diff.
Reviewers: dmgreen, miyuki, ostannard, simon_tatham
Subscribers: kristof.beyls, hiraditya, cfe-commits, llvm-commits
Tags: #clang, #llvm
Differential Revision: https://reviews.llvm.org/D71421
Disable the instantiation-depth-default.cpp test on NetBSD since it
requires more stack space than we have by default on NetBSD.
Differential Revision: https://reviews.llvm.org/D71419
in declare variant.
If the types of the fnction are not equal, but match, at the codegen
thei may have different types. This may lead to compiler crash.
Summary:
This adds support for embedding bitcode in a binary during LTO. The libLTO gains supports the `-lto-embed-bitcode` flag. The option allows users of the LTO library to embed a bitcode section. For example, LLD can pass the option via `ld.lld -mllvm=-lto-embed-bitcode`.
This feature allows doing something comparable to `clang -c -fembed-bitcode`, but on the (LTO) linker level. Having bitcode alongside native code has many use-cases. To give an example, the MacOS linker can create a `-bitcode_bundle` section containing bitcode. Also, having this feature built into LLVM is an alternative to 3rd party tools such as [[ https://github.com/travitch/whole-program-llvm | wllvm ]] or [[ https://github.com/SRI-CSL/gllvm | gllvm ]]. As with these tools, this feature simplifies creating "whole-program" llvm bitcode files, but in contrast to wllvm/gllvm it does not rely on a specific llvm frontend/driver.
Patch by Josef Eisl <josef.eisl@oracle.com>
Reviewers: #llvm, #clang, rsmith, pcc, alexshap, tejohnson
Reviewed By: tejohnson
Subscribers: tejohnson, mehdi_amini, inglorion, hiraditya, aheejin, steven_wu, dexonsmith, dang, cfe-commits, llvm-commits, #llvm, #clang
Tags: #clang, #llvm
Differential Revision: https://reviews.llvm.org/D68213
This warning is supposed to be suppressed when the
constructor/destructor are non-trivial, since it might be a RAII type.
However, if the type has a trivial destructor and the constructor hasn't
been resolved (since it is called with dependent arguments), we were
still warning.
This patch suppresses the warning if the type could possibly have a
be a non-trivial constructor call. Note that this does not take the
arity of the constructors into consideration, so it might suppress
the warning in cases where it isn't possible to call a non-trivial
constructor.
Mark de Wever