The instruction itself can be considered good for minbitwidth casting,
even if one of the operand checks returns false.

Reviewers: RKSimon

Reviewed By: RKSimon

Pull Request: https://github.com/llvm/llvm-project/pull/84363

Reverts llvm/llvm-project#82763 because caused a regressions with
inlining.
See
https://github.com/llvm/llvm-project/pull/84494#issuecomment-1996047458

Cutting down on diff in an upcoming change.

Avoid referencing executionEnvironment in the device code, since
environment.cpp is not part of the CUDA build yet.
This is a temporary fix before #85182 is merged.

Add a clspv switch in gen_convert.cl
This is needed as Vulkan SPIR-V does not respect the assumptions
needed to have the generic convert.cl compliant on many platforms.

It is needed because of the conversion of TYPE_MAX and
TYPE_MIN. Depending on the platform the behaviour can vary, but most
of them just do not convert correctly those 2 values.

Because of that, we also need to avoid having explicit function for
simple conversions because it allows llvm to optimise the code, thus
removing some of the added checks that are in fact needed.

We can treat a shift by constant as a multiply by a power of 2
and we can treat an or disjoint as a 'add nsw nuw'.

I've added a helper struct similar to a struct used in
ScalarEvolution.cpp
to represent the opcode, operands, and NSW/NUW flags for normal
add/sub/mul
and shl/or that are being treated as mul/add.

I don't think we need to teach cloneIVUser about this. It will continue
to clone them using cloneBitwiseIVUser. After the cloning we will ask
for the SCEV expression for the cloned IV user and verify that it
matches
the AddRec returned by getExtendedOperandRecurrence. Since SCEV also
knows how to convert shl to mul and or disjoint to add nsw nuw, this
should
usually match. If it doesn't match, the cloned IV user will be deleted.

Use a normal min/max reduction that doesn't propagate nans and force
the result to nan at the end if any elements were nan.

This reverts commit 6f3f659c.

This change is causing TSAN failures on a bot as reported in #85188:
https://lab.llvm.org/buildbot/#/builders/247/builds/15279

This also moves `__all` from `sfinae_helpers.h` to `conjunction.h`.

Reverts llvm/llvm-project#85135

There is an issue with module file generation in flang build.

Dexter currently supports Visual Studio 2015/2017/2019, but not 2022;
this patch adds support for 2022.

Reset operand list whenever we create a new instruction via a parameter
passed by reference. Most functions were already doing this, but there
are several places missing the reset. Potentially, if we don not clear
the list it could lead to invalid instruction operands. But the existing
code is unaffected.

This member switched from an optional to a vector in 0f252e7b.

This reverts commit 611c62b3.

Removing the extension to FullWidth should make them much more efficient
in the 64-bit case, because 65-bit APInts use a separate allocation for
their bits.

Comparing c_ptr type for equality or inequality is raising an error. 

```
not yet implemented: intrinsic module procedure: c_ptr_eq
```
or this one for inequality
```
not yet implemented: intrinsic module procedure: c_ptr_ne
```

This patch adds a lowering for them and fix the `__fortran_builtins.f90`
module for inequality.

…ctive

The function `ActOnOpenMPTargetParallelForSimdDirective` gets the number
of capture levels for OMPD_target_parallel_for, whereas the intended
directive is OMPD_target_parallel_for_simd.

getArithmeticInstrCost is used by both LoopVectorizer and SLPVectorizer
to compute the cost of frem, which becomes a call cost on AArch64 when
TLI has a vector library function.

Add tests that do SLP vectorization for code that contains 2x double and
4x float frem instructions.

Code in plugins-nextgen reading ELF files is currently hard-coded to
assume a 64-bit little-endian ELF format. Unfortunately, this assumption
is even embedded in the interface between GlobalHandler and Utils/ELF
routines, which use ELF64LE types.

To fix this, I've refactored the interface to use generic types, in
particular by using (a unique_ptr to) ObjectFile instead of
ELF64LEObjectFile, and ELFSymbolRef instead of ELF64LE::Sym.

This allows properly templating over multiple ELF format variants inside
Utils/ELF; specifically, this patch adds support for 64-bit big-endian
ELF files in addition to 64-bit little-endian files.

These are Wasm only functions so they are better be within `WebAssembly`
namespace rather than the `llvm` namespace which includes the whole
LLVM.

Also this removes `extern` keywords which are not strictly necessary.

The calls are already musttail.

When calling setbuf(nullptr, 0) before performing file operations it
should set the file to unbuffered mode. Currently the code avoids
buffering internally, but the underlying stream still can buffer.

This is addressed by disabling the buffering of the underlying stream.

Fixes: https://github.com/llvm/llvm-project/issues/60509

This reverts commit bbc0f99f
(https://reviews.llvm.org/D157663).

With this change, `-g` for the next major release 19.1 will generate
R_RISCV_SET_ULEB128/R_RISCV_SUB_ULEB128 relocations, which require
lld>=18 or binutils>=2.41.
binutils 2.41 is relatively new, but GCC has been producing
R_RISCV_SET_ULEB128/R_RISCV_SUB_ULEB128 for some time now.

#69295 demoted Defined symbols relative to discarded sections.
If such a symbol is unreferenced, the desired behavior is to
eliminate it from .symtab just like --gc-sections discarded
definitions.
Linux kernel's CONFIG_DEBUG_FORCE_WEAK_PER_CPU=y configuration expects
that the unreferenced `unused` is not emitted to .symtab
(https://github.com/ClangBuiltLinux/linux/issues/2006).

For relocations referencing demoted symbols, the symbol index restores
to 0 like older lld (`R_X86_64_64 0` in `discard-section.s`).

Fix #85048

No testcase: I'm still thinking about the best way to test this.

Some languages may create artificial functions that have no real user
code, even though there is line table information for them. One such
case is with coroutine code that receives the CoroSplitter
transformation in LLVM IR. That code transformation creates many
different Functions, cloning one Instruction into many Instructions in
many different Functions and copying the associated debug locations.

It would be difficult to make that pass delete debug locations of cloned
instructions in a language agnostic way (is it even possible?), but LLDB
can ignore certain locations by querying its Language APIs and having it
decide based on, for example, mangling information.

Expose patterns so they can be reused in other passes.

There were two problems:
* The `%s` argument to `FileCheck` was repeated.
* A single dash for `-check-prefix` was used but we need two dashes.

In file included from ../llvm-project/flang/lib/Lower/OpenMP/Clauses.cpp:9:
../llvm-project/flang/lib/Lower/OpenMP/Clauses.h:195:17: error: suggest braces around initialization of subobject [-Werror,-Wmissing-braces]
  195 |   return Clause{id, specific, source};
      |                 ^~~~~~~~~~~~
      |                 {           }

Darwin AArch64 application processors are run with Top Byte Ignore mode
enabled so metadata may be stored in the top byte, it needs to be
ignored when reading/writing memory. David Spickett handled this already
in the base class Process::ReadMemory but ProcessMachCore overrides that
method (to avoid the memory cache) and did not pick up the same change.
I add a test case that creates a pointer with metadata in the top byte
and dereferences it with a live process and with a corefile.

rdar://123784501

... when doing binary operations on them.

The depths of the Root and the NewRoot are to be compared in
MachineCombiner::improvesCriticalPathLen(), and while the call to
BlockTrace.getInstrCycles(*Root) includes the Depth of a PHI, for some
reason PHI nodes have been ignored in getOperandDef(). 

This patch removes the special handling of PHIs in getOperandDef() so that
Root and NewRoot get a fair comparison. This does not affect loop headers
as MachineTraceMetrics handles that case by ignoring incoming PHI edges.

This enables clang and external nullability checkers to make use of
these annotations on nullable C++ class types like unique_ptr.

These types are recognized by the presence of the _Nullable attribute.
Nullable standard library types implicitly receive this attribute.

Existing static warnings for raw pointers are extended to smart
pointers:

- nullptr used as return value or argument for non-null functions
  (`-Wnonnull`)
- assigning or initializing nonnull variables with nullable values
  (`-Wnullable-to-nonnull-conversion`)

It doesn't implicitly add these attributes based on the assume_nonnull
pragma, nor warn on missing attributes where the pragma would apply
them.
I'm not confident that the pragma's current behavior will work well for
C++ (where type-based metaprogramming is much more common than C/ObjC).
We'd like to revisit this once we have more implementation experience.

Support can be detected as `__has_feature(nullability_on_classes)`.
This is needed for back-compatibility, as previously clang would issue a
hard error when _Nullable appears on a smart pointer.

UBSan's `-fsanitize=nullability` will not check smart-pointer types.
It can be made to do so by synthesizing calls to `operator bool`, but
that's left for future work.

Discussion:
https://discourse.llvm.org/t/rfc-allowing-nonnull-etc-on-smart-pointers/77201/26