Instead of expliciting emitting a setc in the inline asm instructions,
we can use flag output. This allows the backend to use the flag
directly if it is needed by a branch. Previously we needed a test
instruction to convert the register back to a flag.
If the flag can't be used directly, the backend will emit a setcc.
Differential Revision: https://reviews.llvm.org/D87888
This patch legalizes the v256i1 and v512i1 types that will be used for MMA.
It implements loads and stores of these types.
v256i1 is a pair of VSX registers, so for this type, we load/store the two
underlying registers. v512i1 is used for MMA accumulators. So in addition to
loading and storing the 4 associated VSX registers, we generate instructions to
prime (copy the VSX registers to the accumulator) after loading and unprime
(copy the accumulator back to the VSX registers) before storing.
This patch also adds the UACC register class that is necessary to implement the
loads and stores. This class represents accumulator in their unprimed form and
allow the distinction between primed and unprimed accumulators to avoid invalid
copies of the VSX registers associated with primed accumulators.
Differential Revision: https://reviews.llvm.org/D84968
Extend -fsanitize=nullability-arg to handle call sites which accept C++
member pointers.
rdar://62476022
Differential Revision: https://reviews.llvm.org/D88336
- `-cl-fp32-correctly-rounded-divide-sqrt` is an OpenCL-specific option
and `correctly-rounded-divide-sqrt-fp-math` should be added for OpenCL
at most.
Differential revision: https://reviews.llvm.org/D88303
After some recent upstream discussion we decided that it was best
to avoid having the / operator for both ElementCount and TypeSize,
since this could give the impression that these classes can be used
in the same way as basic integer integer types. However, division
for scalable types is a bit odd because we are only dividing the
minimum quantity by a value, as opposed to something like:
(MinSize * Vscale) / SomeValue
This is why when performing division it's important the caller
first establishes whether the operation makes sense, perhaps by
calling isKnownMultipleOf() prior to division. The caller must now
explictly call divideCoefficientBy() on the class to perform the
operation.
Differential Revision: https://reviews.llvm.org/D87700
References to different declarations of the same entity aren't different
values, so shouldn't have different representations.
Recommit of e6393ee813 with fixed handling
for weak declarations. We now look for attributes on the most recent
declaration when determining whether a declaration is weak. (Second
recommit with further fixes for mishandling of weak declarations. Our
behavior here is fundamentally unsound -- see PR47663 -- but this
approach attempts to not make things worse.)
Previous description didn't actually state the effect the attribute has on
thread safety analysis (causing analysis to assume the capability is held).
Previous description was also ambiguous about (or slightly overstated) the
noreturn assumption made by thread safety analysis, implying the assumption had
to be true about the function's behavior in general, and not just its behavior
in places where it's used. Stating the assumption specifically should avoid a
perceived need to disable thread safety analysis in places where only asserting
that a specific capability is held would be better.
Reviewed By: aaronpuchert, vasild
Differential Revision: https://reviews.llvm.org/D87629
There appears to be a mis-compile with MemorySSA-backed DSE in
combination with llvm.lifetime.end. It currently appears like
DSE is doing the right thing and the llvm.lifetime.end markers
are incorrect. The reverted patch uncovers the mis-compile.
This patch temporarily switches back to the legacy DSE
implementation, while we investigate.
This reverts commit 9d172c8e9c.
The change implements evaluation of constant floating point expressions
under non-default rounding modes. The main objective was to support
evaluation of global variable initializers, where constant rounding mode
may be specified by `#pragma STDC FENV_ROUND`.
Differential Revision: https://reviews.llvm.org/D87822
Previously for nowait target, CG emitted a function call to `__tgt_target_nowait`, etc. However, in OpenMP RTL, these functions just directly call the no-nowait version, which means nowait is not working as expected.
OpenMP specification says a target is acutally a target task, which is an untied and detachable task. It is natural to go to the direction that generates a task for a nowait target. However, OpenMP task has a problem that it must be within to a parallel region; otherwise the task will be executed immediately. As a result, if we directly wrap to a regular task, the `target nowait` outside of a parallel region is still a synchronous version.
In D77609, I added the support for unshackled task in OpenMP RTL. Basically, unshackled task is a task that is not bound to any parallel region. So all nowait target will be tranformed into an unshackled task. In order to distinguish from regular task, a new flag bit is set for unshackled task. This flag will be used by RTL for later process.
Since all target tasks are allocated via `__kmpc_omp_target_task_alloc`, and in current `libomptarget`, `__kmpc_omp_target_task_alloc` just calls `__kmpc_omp_task_alloc`. Therefore, we can modify the flag in `__kmpc_omp_target_task_alloc` so that we don't need to modify the FE too much. If users choose to opt out the feature, they just need to use a RTL w/o support of unshackled threads.
As a result, in this patch, the `target nowait` region is simply wrapped into a regular task. Later once we have RTL support for unshackled tasks, the wrapped tasks can be executed by unshackled threads w/o changes in the FE.
Reviewed By: jdoerfert
Differential Revision: https://reviews.llvm.org/D78075
This attribute allows declarations to be restricted to the framework
itself, enabling Swift to remove the declarations when importing
libraries. This is useful in the case that the functions can be
implemented in a more natural way for Swift.
This is based on the work of the original changes in
8afaf3aad2
Differential Revision: https://reviews.llvm.org/D87720
Reviewed By: Aaron Ballman
Make the corresponding change that was made for byval in
b7141207a4. Like byval, this requires a
bulk update of the test IR tests to include the type before this can
be mandatory.
This code never actually did anything in the implementation.
`mergeDeclAttribute` is declared as `static`, and referenced exactly
once in the file: from `Sema::mergeDeclAttributes`.
`Sema::mergeDeclAttributes` sets `LocalAMK` to `AMK_None`. If the
attribute is `DeprecatedAttr`, `UnavailableAttr`, or `AvailabilityAttr`
then the `LocalAMK` is updated. However, because we are dealing with a
`SwiftNameDeclAttr` here, `LocalAMK` remains `AMK_None`. This is then
passed to the function which will as a result pass the value of
`AMK_None == AMK_Override` aka `false`. Simply propagate the value
through and erase the dead codepath.
Thanks to Aaron Ballman for flagging the use of the availability merge
kind here leading to this simplification!
Differential Revision: https://reviews.llvm.org/D88263
Reviewed By: Aaron Ballman
Add support for expanding the %t filename specifier in LLVM_PROFILE_FILE
to the TMPDIR environment variable. This is supported on all platforms.
On Darwin, TMPDIR is used to specify a temporary application-specific
scratch directory. When testing apps on remote devices, it can be
challenging for the host device to determine the correct TMPDIR, so it's
helpful to have the runtime do this work.
rdar://68524185
Differential Revision: https://reviews.llvm.org/D87332
Summary:
The MacroExpander allows to expand simple (non-resursive) macro
definitions from a macro identifier token and macro arguments. It
annotates the tokens with a newly introduced MacroContext that keeps
track of the role a token played in expanding the macro in order to
be able to reconstruct the macro expansion from an expanded (formatted)
token stream.
Made Token explicitly copy-able to enable copying tokens from the parsed
macro definition.
Reviewers: sammccall
Subscribers: mgorny, cfe-commits
Tags: #clang
Differential Revision: https://reviews.llvm.org/D83296
For /C++/ constructor initializers `ExprEngine:computeUnderConstruction()`
asserts that they are all member initializers. This is not neccessarily
true when this function is used to get the return value for the
construction context thus attempts to fetch return values of base and
delegating constructor initializers result in assertions. This small
patch fixes this issue.
Differential Revision: https://reviews.llvm.org/D85351
PAC/BTI-related codegen in the AArch64 backend is controlled by a set
of LLVM IR function attributes, added to the function by Clang, based
on command-line options and GCC-style function attributes. However,
functions, generated in the LLVM middle end (for example,
asan.module.ctor or __llvm_gcov_write_out) do not get any attributes
and the backend incorrectly does not do any PAC/BTI code generation.
This patch record the default state of PAC/BTI codegen in a set of
LLVM IR module-level attributes, based on command-line options:
* "sign-return-address", with non-zero value means generate code to
sign return addresses (PAC-RET), zero value means disable PAC-RET.
* "sign-return-address-all", with non-zero value means enable PAC-RET
for all functions, zero value means enable PAC-RET only for
functions, which spill LR.
* "sign-return-address-with-bkey", with non-zero value means use B-key
for signing, zero value mean use A-key.
This set of attributes are always added for AArch64 targets (as
opposed, for example, to interpreting a missing attribute as having a
value 0) in order to be able to check for conflicts when combining
module attributed during LTO.
Module-level attributes are overridden by function level attributes.
All the decision making about whether to not to generate PAC and/or
BTI code is factored out into AArch64FunctionInfo, there shouldn't be
any places left, other than AArch64FunctionInfo, which directly
examine PAC/BTI attributes, except AArch64AsmPrinter.cpp, which
is/will-be handled by a separate patch.
Differential Revision: https://reviews.llvm.org/D85649
Adding this test so that I can extend it in a follow on patch with
expected IR for AIX when I implement complex handling in
AIXABIInfo.
Reviewed By: daltenty, ZarkoCA
Differential Revision: https://reviews.llvm.org/D88105
Add the ability to selectively instrument a subset of functions by dividing the functions into N logical groups and then selecting a group to cover. By selecting different groups over time you could cover the entire application incrementally with lower overhead than instrumenting the entire application at once.
Differential Revision: https://reviews.llvm.org/D87953
Passing them directly is likely to be non-conforming, since it usually
involves copying the bytes of the record. For unknown architectures, we
don't know what MSVC does or will do, but we should at least try to
conform as well as we can.
Regardless of the target architecture, we should always use the C rules
(RAA_Default) for records that "canBePassedInRegisters". Those are
trivially copyable things, and things marked with [[trivial_abi]].
This should be NFC, although it changes where the final decision about
x86_32 overaligned records is made. The current x86_32 C rules say that
overaligned things are passed indirectly, so there is no functional
difference.
constructors.
This changes the code to avoid using constructor homing for aggregate
classes and classes with trivial default constructors, instead of trying
to loop through the constructors.
Differential Revision: https://reviews.llvm.org/D87808
This is needed to compile some headers in CUDA-11 that assume that threadIdx is
implicitly convertible to dim3. With NVCC, threadIdx is uint3 and there's
dim3(uint3) constructor. Clang uses a special type for the builtin variables, so
that path does not work. Instead, this patch adds conversion function to the
builtin variable classes. that will allow them to be converted to dim3 and uint3.
Differential Revision: https://reviews.llvm.org/D88250
This is triggered during serialization. The test is for modules, but
will occur for any serialization effort using asm goto.
Reviewed By: nickdesaulniers, jyknight
Differential Revision: https://reviews.llvm.org/D88195
The following crashes on my system before this patch, but not after:
void foo(int i) {
switch (i) {
case 1:
case 2:
... 100000 cases ...
;
}
}
clang-query -c="match stmt(hasAncestor(stmt()))" deep.c
I'm not sure it's actually a sane testcase to run though, it's pretty slow :-)
Differential Revision: https://reviews.llvm.org/D88222
Need to fix a check for the variable if it is declared in the inner
OpenMP region to be able to firstprivatize it.
Reviewed By: jdoerfert
Differential Revision: https://reviews.llvm.org/D88240
This reverts commit c4bacc3c9b.
Test "LLVM :: ThinLTO/X86/funcimport-stats.ll" is failing. Reverting now
and will recommit after making the test not fail with the added stats.
Measure amount of high-level or fixed-cost operations performed during
building/loading modules and during header search. High-level operations
like building a module or processing a .pcm file are motivated by
previous issues where clang was re-building modules or re-reading .pcm
files unnecessarily. Fixed-cost operations like `stat` calls are tracked
because clang cannot change how long each operation takes but it can
perform fewer of such operations to improve the compile time.
Also tracking such stats over time can help us detect compile-time
regressions. Added stats are more stable than the actual measured
compilation time, so expect the detected regressions to be less noisy.
rdar://problem/55715134
Reviewed By: aprantl, bruno
Differential Revision: https://reviews.llvm.org/D86895
As mentioned in the bug report, tryEmitPrivate chokes on the
MaterializeTemporaryExpr in the reproducers, since it assumes that if
there are elements, than it must be a ConstantArrayType. However, the
MaterializeTemporaryExpr (which matches exactly the AST when it is NOT a
global/static) has an incomplete array type.
This changes the section where the number-of-elements is non-zero to
properly handle non-CAT types by just extracting it as an array type
(since all we needed was the element type out of it).
This reverts commit 8e780a1653.
DiagnosticBuilder is a value type, created on the stack everywhere. IMO
we should not be adding a vtable to it, and making very operator<< use a
virtual interface. There are other feasible designs for implementing
this. The original review, D84362, was approved by @tra, who is
responsible for Clang's CUDA support, but it wasn't reviewed by @rsmith
or anyone responsible for clang's diagnostic library.
Need to fix a check for the variable if it is declared in the inner
OpenMP region to be able to firstprivatize it.
Differential Revision: https://reviews.llvm.org/D88240
No need to make final copy from the firsptrivate/lastprivate copy to the original item if the item is a data memeber.
Firstprivate copy creates a copy by reference and the original item gets
updated correctly when updating the lastprivate shared variable.
Differential Revision: https://reviews.llvm.org/D88179
Add the `swift_newtype` attribute which allows a type definition to be
imported into Swift as a new type. The imported type must be either an
enumerated type (enum) or an object type (struct).
This is based on the work of the original changes in
8afaf3aad2
Differential Revision: https://reviews.llvm.org/D87652
Reviewed By: Aaron Ballman
Using a pointer for the description string in StaticDiagInfoRec causes
several problems:
1. We don't need to use a whole pointer to represent the string;
2. The use of pointers incurs runtime relocations for those pointers;
the relocations take up space on disk and represent runtime overhead;
3. The need to relocate data implies that, on some platforms, the entire
array containing StaticDiagInfoRecs cannot be shared between processes.
This patch changes the storage scheme for the diagnostic descriptions to
avoid these problems. We instead generate (effectively) one large
string and then StaticDiagInfoRec conceptually holds offsets into the
string. We elected to also move the storage of those offsets into a
separate array to further reduce the space required.
On x86-64 Linux, this change removes about 120KB of relocations and
moves about 60KB from the non-shareable .data.rel.ro section to
shareable .rodata. (The array is about 80KB before this, but we
eliminated 4 bytes/entry by using offsets rather than pointers.) We
actually reap this benefit twice, because these tables show up in both
libclang.so and libclang-cpp.so and we get the reduction in both places.
Differential Revision: https://reviews.llvm.org/D81865
Before this patch, the CrashRecoveryContext was returning -2 upon a signal, like ExecuteAndWait does. This didn't match the behavior on Windows, where the the exception code was returned.
We now return the signal's code, which optionally allows for re-throwing the signal later. Doing so requires all custom handlers to be removed first, through llvm::sys::unregisterHandlers() which we made a public API.
This is part of https://reviews.llvm.org/D70378
Sema::DiagnoseSwiftName uses the constant 12 instead of the
corresponding enumerator ExpectedFunctionWithProtoType. This is
fragile and will fail if a new value gets added in the middle of the
enum.
Reviewed By: aaron.ballman
Differential Revision: https://reviews.llvm.org/D88164
This patch implements custom codegen for the vec_replace_elt and
vec_replace_unaligned builtins.
These builtins map to the @llvm.ppc.altivec.vinsw and @llvm.ppc.altivec.vinsd
intrinsics depending on the arguments. The main motivation for doing custom
codegen for these intrinsics is because there are float and double versions of
the builtin. Normally, the converting the float to an integer would be done via
fptoui in the IR. This is incorrect as fptoui truncates the value and we must
ensure the value is not truncated. Therefore, we provide custom codegen to utilize
bitcast instead as bitcasts do not truncate.
Differential Revision: https://reviews.llvm.org/D83500
I believe the inline asm emitted here should have a memory clobber since it writes to memory.
It was also missing the dirflag clobber that we use by default along with flags and fpsr. To avoid missing defaults in the future, get the default list from the target
Differential Revision: https://reviews.llvm.org/D88121
This recommits 829d14ee0a.
The patch was reverted due to a regression in some CUDA app
which was thought to be caused by this patch. However, investigation
showed that the regression was due to some other issues, therefore
recommit this patch.
This patch implements the vec_[all|any]_[eq | ne | lt | gt | le | ge] builtins for vector signed/unsigned __int128.
Differential Revision: https://reviews.llvm.org/D87910
D87921 was reverted in commit b89059a313
as it was causing an unknown llvm PPC bot failure. Reapplying the patch
after confirming that this is not responsible. Build bot failure:
https://reviews.llvm.org/D87921#2286644 which caused the revert.
The wrong placement of add pass with optimizations led to
-funique-internal-linkage-names being disabled.
Fixed the placement of the MPM.addpass for UniqueInternalLinkageNames to make it
work correctly with -O2 and new pass manager. Updated the tests to explicitly
check O0 and O1.
Differential Revision: https://reviews.llvm.org/D87921
Check whether /etc/env.d/gcc exists before trying to read from any
file from there. This saves a few OS calls on a non-Gentoo system.
Differential Revision: https://reviews.llvm.org/D87143
This is important to not regress because it allows us to capture pre-optimization
bitcode and options, and replay the full optimization pipeline.
Differential Revision: https://reviews.llvm.org/D88114
Currently, when marshaling a dynamic AST matchers, we check for the type
and value validity of matcher arguments at the same time for some matchers.
For instance, when marshaling hasAttr("foo"), the argument is first type
checked to ensure it's a string and then checked to see if that string can
locate an attribute with that name. Similar happens for other enumeration
conversions like cast kinds or unary operator kinds. If the type is
correct but the value cannot be looked up, we make a best-effort attempt
to find a nearby name that the user might have meant, but if one cannot
be found, we throw our hands up and claim the types don't match.
This has an unfortunate behavior that when the user enters something of
the correct type but a best guess cannot be located, you get confusing
error messages like:
Incorrect type for arg 1. (Expected = string) != (Actual = String).
This patch splits the argument check into two parts: if the types don't
match, give a type diagnostic. If the type matches but the value cannot
be converted, give a best guess diagnostic or a value could not be
located diagnostic. This addresses PR47057.
recommit e50465ecef with fix for
regression in lldb tests.
Two issues:
1. the directory part of original .dwo file was dropped
2. if the stem of the .dwo file contains '.', the last dot
and strings after that were removed
This recommit fixes those two issues.
A static device variable may be accessed in host code through
cudaMemCpyFromSymbol etc. Currently clang does not
emit the static device variable if it is only referenced by
host code, which causes host code to fail at run time.
This patch fixes that.
Differential Revision: https://reviews.llvm.org/D88115
The signature should not be part of the summaries as many FIXME comments
suggests. By separating the signature, we open up the way to a generic
matching implementation which could be used later under the hoods of
CallDescriptionMap.
Differential Revision: https://reviews.llvm.org/D88100
It is no longer needed to add summaries of 'getline' for different
possible underlying types of ssize_t. We can just simply lookup the
type.
Differential Revision: https://reviews.llvm.org/D88092
Some of the predicates can't always be decided - for example when a type
definition isn't available. At the same time it's necessary to let
client code decide what to do about such cases - specifically we can't
just use true or false values as there are callees with
conflicting strategies how to handle this.
This is a speculative fix for PR47276.
Differential Revision: https://reviews.llvm.org/D88133
This completes the circle, complementing -lto-embed-bitcode
(specifically, post-merge-pre-opt). Using -thinlto-assume-merged skips
function importing. The index file is still needed for the other data it
contains.
Differential Revision: https://reviews.llvm.org/D87949
We leave a dangling TypoExpr when typo-correction is performed
successfully in `checkArgsForPlaceholders`, which leads a crash in the
later TypoCorrection.
This code was added in 1586782767,
and it didn't seem to have enough test coverage.
The fix is to remove this part, and no failuer tests.
Reviewed By: rsmith
Differential Revision: https://reviews.llvm.org/D87815
A recent change increased the stack size of memoizedMatchesAncestorOfRecursively
leading to stack overflows on real code involving large fold expressions.
It's not totally unreasonable to choke on very deep ASTs, but as common
infrastructure it's be nice if ASTMatchFinder is more robust.
(It already uses data recursion for the regular "downward" traversal.)
Differential Revision: https://reviews.llvm.org/D86964
This patch implements the vector string isolate (predicate and non-predicate
versions) builtins. The predicate builtins are custom selected within PPCISelDAGToDAG.
Differential Revision: https://reviews.llvm.org/D87671
This patch implements the 128-bit vector divide extended builtins in Clang/LLVM.
These builtins map to the vdivesq and vdiveuq instructions respectively.
Differential Revision: https://reviews.llvm.org/D87729
This introduces the new `swift_name` attribute that allows annotating
APIs with an alternate spelling for Swift. This is used as part of the
importing mechanism to allow interfaces to be imported with a new name
into Swift. It takes a parameter which is the Swift function name.
This parameter is validated to check if it matches the possible
transformed signature in Swift.
This is based on the work of the original changes in
8afaf3aad2
Differential Revision: https://reviews.llvm.org/D87534
Reviewed By: Aaron Ballman, Dmitri Gribenko
since crti is required for functional static initialization.
Reviewed By: hubert.reinterpretcast
Differential Revision: https://reviews.llvm.org/D87927
Set the default wchar_t type on z/OS, and unsigned as the default.
Reviewed By: hubert.reinterpretcast, fanbo-meng
Differential Revision: https://reviews.llvm.org/D87624
There can be Macros that are tagged with `modifiable`. Thus verifying
`canModifyAllDescendants` is not sufficient to avoid macros when deep
copying.
We think the `TokenBuffer` could inform us whether a `Token` comes from
a macro. We'll look into that when we can surface this information
easily, for instance in unit tests for `ComputeReplacements`.
Differential Revision: https://reviews.llvm.org/D88034
When cross compiling with clang-cl, clang splits the INCLUDE env
variable around semicolons (clang/lib/Driver/ToolChains/MSVC.cpp,
MSVCToolChain::AddClangSystemIncludeArgs) and lld splits the
LIB variable similarly (lld/COFF/Driver.cpp,
LinkerDriver::addLibSearchPaths). Therefore, the consensus for
cross compilation with clang-cl and lld-link seems to be to use
semicolons, despite path lists normally being separated by colons
on unix and EnvPathSeparator being set to that.
Therefore, handle the LIB variable similarly in Clang, when
handling lib file arguments when driving linking via Clang.
This fixes commands like "clang-cl test.c -Fetest.exe kernel32.lib" in
a cross compilation setting. Normally, most users call (lld-)link
directly, but meson happens to use this command syntax for
has_function() tests.
Reapply: Change Program.h to define procid_t as ::pid_t. When included
in lldb/unittests/Host/NativeProcessProtocolTest.cpp, it is included
after an lldb namespace containing an lldb::pid_t typedef, followed
later by a "using namespace lldb;". Previously, Program.h wasn't
included in this translation unit, but now it ends up included
transitively from Process.h.
Differential Revision: https://reviews.llvm.org/D88002
* Introduce `TreeTest.cpp` to unit test `Tree.h`
* Add `generateAllTreesWithShape` to generating test cases
* Add tests for `findFirstLeaf` and `findLastLeaf`
* Fix implementations of `findFirstLeaf` and `findLastLeaf` that had
been broken when empty `Tree` were present.
Differential Revision: https://reviews.llvm.org/D87779
Aaron Ballman