Currently s_getreg_b32 is missing the possible mode use. Really we
need separate pseudos for mode-only accesses, but leave this as a
pre-existing issue.

https://reviews.llvm.org/D152710

When there are multiple catch-all patterns (i.e. a single `*`), GNU ld and gold
select the last pattern. Match their behavior.

This change was inspired by a correction made by Michael Kerrisk to a blog post
I wrote at https://maskray.me/blog/2020-11-26-all-about-symbol-versioning ,
following the current lld rules.

Note: GNU ld prefers global: patterns to local: patterns, which might seem
awkward (https://www.airs.com/blog/archives/300). gold doesn't follow this
behavior, and we do not either.

This test fails with `CLANG_DEFAULT_CXX_STDLIB=libc++`

The default constructor for SBCommandInterpreter was (unintentionally)
made protected in 27b6a4e6. The goal of the patch was to make the
constructor taking an lldb_private type protected, but due to the
presence of a default argument, this ctor also served as the default
constructor. The latter should remain public.

This commit reinstates the original behavior by removing the default
argument and having an explicit, public default constructor.

Change to use VE_LD_LIBRARY_PATH for VE instead of LD_LIBRARY_PATH. The
VE is connected to the host, and compiled test programs for VE is
invoked on the host and transferred to the VE. If programs are compiled
for the host, we use LD_LIBRARY_PATH. Otherwise, we use
VE_LD_LIBRARY_PATH.

This patch adds the option of building an optional symbol table for the
top operation in the `gpu-module-to-binary` pass. The table is not
created by default as most targets don't need it; instead, it is lazily
built. The table is passed through a callback in `TargetOptions`.

This patch is required to integrate #65539 .

Support OpenMP runtime library on VE.  This patch makes OpenMP compilable
for VE architecture.  Almost all tests run correctly on VE.

Reviewed By: tianshilei1992

Differential Revision: https://reviews.llvm.org/D159401

The front end doesn't create captured stmt for unroll directive. This
leads to
a crash when `-fopenmp-simd` is used, as reported in #63570.

Fix #63570.

Fix-forward for a9f30097

Use fallback name only on the top level, in Symbolizer::Demangle or
DlAddrSymbolizer. This makes PlatformDemangle to be more consistent with
SymbolizerTool and the loop in Symbolizer::Demangle which iterates over
all availible options.

It really bugs me that it breaks to
``` c++
...) noexcept(
    noexcept(condition)...
```

This is a fix for people like me.

https://reviews.llvm.org/D157911

This re-applies 4b17c81d, "[jitlink/rtdydl][checker] Add TargetFlag
dependent disassembler switching support", which was reverted in
4871a9ca due to bot failures.

The patch has been updated to add missing plumbing for Subtarget Features and
a CPU string, which should fix the failing tests.

https://reviews.llvm.org/D158280

After 98e6deb6 the 'HeadersForSymbolTest.IWYUTransitiveExportWithPrivate' test in 'ClangIncludeCleanerTest' started failing. This is most likely because `FileEntryRef::getName()` now starts with ".\" on Windows, whereas `FileEntry::getName()` did not. This commit fixes assumption of forward slash separators.

Summary:
This patch implements fwrite, putc, putchar, and fputc on the GPU. These
are very straightforward, the main difference for the GPU implementation
is that we are currently ignoring `errno`. This patch also introduces a
minimal smoke test for `putc` that is an exact copy of the `puts` test
except we print the string char by char. This also modifies the `fopen`
test to use `fwrite` to mirror its use of `fread` so that it is tested
as well.

Add a CODE_OF_CONDUCT.md file to the root of the repository. The file
itself references the LLVM Community Code of Conduct. GitHub will
recognize this file and put a link to it to the right of the repository,
similar to the license and security policy, making the CoC easier to
discover.

Fix crash on RAUW due to locals and globals having different address
spaces. This is the intent of the original code, but it assumes the
alloca address space is 0. This patch fixes the code to check that the
global's address space matches `DL.getAllocaAddrSpace()` instead.

Fixes #65155

The revert happened due to a build bot failure that threw 'CUDA_ERROR_UNSUPPORTED_PTX_VERSION'.
The failure's root cause was a pass using "+ptx76" for compilation and an old CUDA driver
on the bot. This commit relands the patch with "+ptx60".

Original Gh PR: #65768
Original commit message:
    Migrate tests referencing `gpu-to-cubin` to the new compilation workflow
    using `TargetAttrs`. The `test-lower-to-nvvm` pass pipeline was modified
    to use the new compilation workflow to simplify the introduction of
    future tests.

    The `createLowerGpuOpsToNVVMOpsPass` function was removed, as it didn't
    allow for passing all options available in the `ConvertGpuOpsToNVVMOp`
    pass.

According to the manual, cas, casl, casx and casxl are synthetic
instructions. They map to casa and casxa with certain ASI tags.

Revert "[mlir][test][gpu] Migrate CUDA tests to the TargetAttr compilation workflow (#65768) (#65848)

This reverts commit d21b6729.

Migrate tests referencing `gpu-to-cubin` to the new compilation workflow
using `TargetAttrs`. The `test-lower-to-nvvm` pass pipeline was modified
to use the new compilation workflow to simplify the introduction of
future tests.

The `createLowerGpuOpsToNVVMOpsPass` function was removed, as it didn't
allow for passing all options available in the `ConvertGpuOpsToNVVMOp`
pass.

Linker relaxation may change relocations (offsets and types). However,
when --emit-relocs is used, relocations are simply copied from the input
section causing a mismatch with the corresponding (relaxed) code
section.

This patch fixes this as follows: for non-relocatable RISC-V binaries,
`InputSection::copyRelocations` reads relocations from the relocated
section's `relocations` array (since this gets updated by the relaxation
code). For all other cases, relocations are read from the input section
directly as before.

In order to reuse as much code as possible, and to keep the diff small,
the original `InputSection::copyRelocations` is changed to accept the
relocations as a range of `Relocation` objects. This means that, in the
general case when reading from the input section, raw relocations need
to be converted to `Relocation`s first, which introduces quite a bit of
boiler plate. It also means there's a slight code size increase due to
the extra instantiations of `copyRelocations` (for both range types).

Reviewed By: MaskRay

Differential Revision: https://reviews.llvm.org/D159082

This enables canonicalization to fold away unnecessary tensor.dim ops
which in turn enables folding away of other operations, as can be seen
in conv_tensors_dynamic where affine.min operations were folded away.

Now what we have delegate() we can use it here.