This avoids some nasty issues with argument passing and lowering of
arbitrary v64i8 shuffles.

[X86] Only pass v64i8/v32i16 as v16i32 on non-avx512bw targets if the v16i32 type won't be split by prefer-vector-width=256

Otherwise just let the v64i8/v32i16 types be split to v32i8/v16i16.

In reality this shouldn't happen because it means we have a 512-bit
vector argument, but min-legal-vector-width says a value less than
512. But a 512-bit argument should have been factored into the
preferred vector width.

Summary:
This patch fixes two problems with the crtbegin.c as written:

1. In do_init, register_frame_info is not guarded by a #define, but in
do_fini, deregister_frame_info is guarded by #ifndef
CRT_HAS_INITFINI_ARRAY. Thus when CRT_HAS_INITFINI_ARRAY is not
defined, frames are registered but then never deregistered.

The frame registry mechanism builds a linked-list from the .so's
static variable do_init.object, and when the .so is unloaded, this
memory becomes invalid and should be deregistered.

Further, libgcc's crtbegin treats the frame registry as independent
from the initfini array mechanism.

This patch fixes this by adding a new #define,
"EH_USE_FRAME_INFO_REGISTRY", which is set by the cmake option
COMPILER_RT_CRT_USE_EH_FRAME_REGISTRY Currently, do_init calls
register_frame_info, and then calls the binary's constructors. This
allows constructors to safely use libunwind. However, do_fini calls
deregister_frame_info and then calls the binary's destructors. This
prevents destructors from safely using libunwind.

This patch also switches that ordering, so that destructors can safely
use libunwind. As it happens, this is a fairly common scenario for
thread sanitizer.

Some warnings in -Wtautological-compare subgroups are DefaultIgnore.
Adding this group to -Wmost, which is part of -Wall, will aid in their
discoverability.

Differential Revision: https://reviews.llvm.org/D69292

Enable to generate BTF_KIND_VARs for non-static
default-section globals which is not allowed previously.
Modified the existing test case to accommodate the new change.

Also removed unused linkage enum members VAR_GLOBAL_TENTATIVE and
VAR_GLOBAL_EXTERNAL.

Differential Revision: https://reviews.llvm.org/D70145

Summary:
If there are any internal methods whose address was taken, conclude there is nothing known in relation of any other internal method and a global.

Reviewers: nlopes, sanjoy.google

Subscribers: hiraditya, llvm-commits

Tags: #llvm

Differential Revision: https://reviews.llvm.org/D69690

This fixes the following warning for developers:

  Target 'liblldb' was changed to a FRAMEWORK sometime after install().  This
  may result in the wrong install DESTINATION.  Set the FRAMEWORK property
  earlier.

The solution is to pass the FRAMEWORK flag to add_lldb_library and set
the target property before install(). For now liblldb is the only
customer.

Resolves PR38906, PR40898.

Summary:
By additional regex match, grouping of main include can be enabled in files that are not normally considered as a C/C++ source code.
For example, this might be useful in templated code, where template implementations are being held in *Impl.hpp files.
On the occassion, 'assume-filename' option description was reworded as it was misleading. It has nothing to do with `style=file` option and it does not influence sourced style filename.

Reviewers: rsmith, ioeric, krasimir, sylvestre.ledru, MyDeveloperDay

Reviewed By: MyDeveloperDay

Subscribers: MyDeveloperDay, cfe-commits

Patch by:  furdyna

Tags: #clang

Differential Revision: https://reviews.llvm.org/D67750

Centralize the logic to remove debugserver from
LLVM_DISTRIBUTION_COMPONENTS when LLDB_USE_SYSTEM_DEBUGSERVER is
enabled. Now this happens regardless of whether the tests are enabled.

Fix the modeling for loads and stores using the register offset addresing mode.

Fix predicates related to the register offset addressing mode.

Root cause is VectorBlockGenerator::copyStmt iterates all instructions
in basic block, however some load instructions may be not unnecessary
thus removed by simplification. As a result, these load instructions
don't have a corresponding array.

Looking at BlockGenerator::copyBB, it only iterates instructions list
of ScopStmt. Given it must be a block type scop in case of
vectorization, I think we should do the same in
VectorBlockGenerator::copyStmt.

Patch by bin.narwal <bin.narwal@gmail.com>

Differential Revision: https://reviews.llvm.org/D70076

Let the checkers use a reference instead of a copy in a range-based
for loop.

This avoids new warnings due to D68912 adds -Wrange-loop-analysis to -Wall.

Differential Revision: https://reviews.llvm.org/D70047

The std::pair<const clang::ValueDecl *, llvm::ArrayRef<clang::OMPClauseMappableExprCommon::MappableComponent>>
type will be copied in a range-based for loop. Make the copy explicit to
avoid the -Wrange-loop-analysis warning.

This avoids new warnings due to D68912 adds -Wrange-loop-analysis to -Wall.

Differential Revision: https://reviews.llvm.org/D70046

The AssociationIteratorTy type will be copied in a range-based for loop.
Make the copy explicit to avoid the -Wrange-loop-analysis warning.

This avoids new warnings due to D68912 adds -Wrange-loop-analysis to -Wall.

Differential Revision: https://reviews.llvm.org/D70045

[LLDB][Formatters] Re-enable std::function formatter with fixes to improve non-cached lookup performance

Performance issues lead to the libc++ std::function formatter to be disabled. We addressed some of those performance issues by adding caching see D67111
This PR fixes the first lookup performance by not using FindSymbolsMatchingRegExAndType(...) and instead finding the compilation unit the std::function wrapped callable should be in and then searching for the callable directly in the CU.

Differential Revision: https://reviews.llvm.org/D69913

The parsing error tests in ELF/redefine-symbols.test are not specific to ELF.
Move them to redefine-symbols.test.
Add COFF/redefine-symbols.test for COFF specific tests.

Also fix the documentation regarding --redefine-syms: the old and new
names are separated by whitespace, not an equals sign.

Reviewed By: mstorsjo

Differential Revision: https://reviews.llvm.org/D70036

Catch up with an API change.

These relocations are specified by the ARM EHABI (section 6.3). As I understand
it, their purpose is to accommodate unwinder implementations that wish to
reduce code size by placing the implementations of the compact unwinding
decoders in a separate translation unit, and using extern weak symbols to
refer to them from the main unwinder implementation, so that they are only
linked when something in the binary needs them in order to unwind.

However, neither of the unwinders used on Android (libgcc, LLVM libunwind)
use this technique, and in fact emitting these relocations ends up being
counterproductive to code size because they cause a copy of the unwinder
to be statically linked into most binaries, regardless of whether it is
actually needed. Furthermore, these relocations create circular dependencies
(between libc and the unwinder) in cases where the unwinder is dynamically
linked and libc contains compact unwind info.

Therefore, deviate from the EHABI here and stop emitting these relocations
on Android.

Differential Revision: https://reviews.llvm.org/D70027

- Dependent components need linking directly.

reduction initializers.

Better to use copy constructor at the initialization of the declare
reduction construct rather than assignment operator.

We build with `-nostdinc++` and add our own header path via
`LIBCXXABI_LIBCXX_INCLUDES`.  However cmake tried to be clever and if
`LIBCXXABI_LIBCXX_INCLUDES` happens to match the compilers system path
it will remove the `-I` flag meaning we can't access any C++ headers.

Ideally cmake would be able see that we are using `-nostdinc++` and
disable this behaviour.

Differential Revision: https://reviews.llvm.org/D69973

This is basically the same bug as in r260434.

SymbolFileDWARF::FindTypes has exponential worst-case when digging
through dependency DAG of .pcm files because each object file and .pcm
file may depend on an already-visited .pcm file, which may again have
dependencies. Fixed here by carrying a set of already visited
SymbolFiles around.

rdar://problem/56993424

Differential Revision: https://reviews.llvm.org/D70106

Fail early, when we discover no tests at all, or filter out all of them.

There is also `--allow-empty-runs` to disable test to allow workflows
like `LIT_FILTER=abc ninja check-all`.  Apparently `check-all` invokes
lit multiple times if certain projects are enabled, which would produce
unwanted "empty runs". Specify via `LIT_OPTS=--allow-empty-runs`.

There are 3 causes for empty runs:
1) No tests discovered.  This is always an error.  Fix test suite config
   or command line.
2) All tests filtered out.  This is an error by default, but can be
   suppressed via `--alow-empty-runs`.  Should prevent accidentally
   passing empty runs, but allow the workflow above.
3) The number of shards is greater than the number of tests.  Currently,
   this is never an error.  Personally, I think we should consider
   making this an error by default; if this happens, you are doing
   something wrong. I added a warning but did not change the behavior,
   since this warrants more discussion.

Reviewed By: atrick, jdenny

Differential Revision: https://reviews.llvm.org/D70105

If ReadASTBlock does not find its top-level submodule, there's something
wrong the with the PCM.  Error in that case, to avoid hitting problems
further from the source.

Note that the Swift compiler sometimes hits a case in
CompilerInstance::loadModule where the top-level submodule mysteriously
does not have Module::IsFromModuleFile set.  That will emit a confusing
warn_missing_submodule, which was never intended for the main module.
The recent audit of error-handling in ReadAST may have rooted out the
real problem.  If not, this commit will help to clarify the real
problem, and replace a confusing warning with an error pointing at the
malformed PCM file.

We're specifically sniffing out whether the top-level submodule was
found/processed, in case there is a malformed module file that is
missing it.  If there is an error encountered during ReadSubmoduleBlock
the return status should already propagate through.  It would be nice to
detect other missing submodules around here to catch other instances of
warn_missing_submodule closer to the source, but that's left as a future
exercise.

https://reviews.llvm.org/D70063

1. Add pseudos PS_vloadrv_ai and PS_vstorerv_ai: those are now used
   for single vector registers in loadRegFromStackSlot (and store...).
2. Remove pseudos PS_vloadrwu_ai and PS_vstorerwu_ai. The alignment is
   now checked when expanding spill pseudos (both in frame lowering
   and in expand-post-ra-pseudos), and a proper instruction is generated.
3. Update MachineMemOperands when dealigning vector spill slots.
4. Return vector predicate registers in getCallerSavedRegs.

This will print [SP-56] instead of [SP+4294967240].

Summary:
This is found during review of https://reviews.llvm.org/D67088.

CHECK-DAG is non-overlapping after https://reviews.llvm.org/D47106.
-allow-deprecated-dag-overlap was introduced to temporary accept old
behavior.

But it actually hide some broken tests, eg: `test/CodeGen/PowerPC/swaps-le-1.ll`
The codegen has changed, but the CHECK-DAG still PASS due to allowing `overlap`.

This patch remove the deprecated options, and fix the broken tests.

Reviewers: #powerpc, hfinkel, nemanjai, steven.zhang, shchenz

Reviewed By: shchenz

Subscribers: shchenz, llvm-commits

Tags: #llvm

Differential Revision: https://reviews.llvm.org/D69733

[DBG][OPT] Attempt to salvage or undef debug info when removing trivially deletable instructions in the Reassociate Expression pass.

Reviewed By: aprantl, vsk

Differential revision: https://reviews.llvm.org/D69943

This patch adds a new IRTransformations directory to llvm/examples/. This is
intended to serve as a new home for example transformations/analysis
code used by various tutorials.

If LLVM_BUILD_EXAMPLES is enabled, the ExamplesIRTransforms library is
linked into the opt binary and the example passes become available.

To start off with, it contains the CFG simplifications used in the IR
part of the 'Getting Started With LLVM: Basics' tutorial at the US LLVM
Developers Meeting 2019.

Reviewers: paquette, jfb, meikeb, lhames, kbarton

Reviewed By: paquette

Differential Revision: https://reviews.llvm.org/D69416

This reverts commit 3f594ed1.

This change has cause LLDB expression evaluation to fail on Arm Linux.

Differential Revision: https://reviews.llvm.org/D63540

This warning triggers when a class defines a copy constructor but not a
copy-assignment operator (which then gets auto-generated by the
compiler). Fix the warning by deleting the other operator too, as the
default implementation works just fine.

Summary:
getAsInstruction is the only non-const member method.
It is impossible to enforce const-correctness because of it.

Reviewers: jmolloy, majnemer

Reviewed By: jmolloy

Subscribers: hiraditya, llvm-commits

Tags: #llvm

Differential Revision: https://reviews.llvm.org/D70113