This reverts commit e77e891d.

Differential Revision: https://reviews.llvm.org/D156947

Lots of users use "po" as their default print command. If the type
doesn't implement the description function the output is often not what
the user wants. Print a hint telling the user that they might prefer
using "p" instead.

Differential Revision: https://reviews.llvm.org/D153489

For code like:
struct foo { ... };
struct bar { struct foo foo; };
const struct foo my_foo = { ... };
struct bar my_bar = { .foo = my_foo };

Eli Friedman points out the relevant part of the C standard seems to
have some flexibility in what is considered a constant expression:

6.6 paragraph 10:
An implementation may accept other forms of constant expressions.

GCC 8 added support for these, so clang not supporting them has been a
constant thorn in the side of source code portability within the Linux
kernel.

Fixes: https://github.com/llvm/llvm-project/issues/44502

Reviewed By: efriedma

Differential Revision: https://reviews.llvm.org/D76096

We previously defaulted to `sm_35` for the purpose of unspecified
architecture. This was removed in new CUDA versions so we should bump
this up.

Reviewed By: jdoerfert

Differential Revision: https://reviews.llvm.org/D156936

The first trivial example I tried failed to merge due to the user scan
logic. Remove the complicated scan of users handling with distance
thresholds, with a same block restriction. The actual expansion of
sincos is basically the same size as sin or cos individually. Copy the
technique the generic optimization uses, which is to just use the
input instruction as the insert point or just insert at the start of
the entry block.

https://reviews.llvm.org/D156706

Ran across this when making a change to RISCV memset lowering. Seems very odd that manually merging a store into a vector prevents it from being further merged.

Differential Revision: https://reviews.llvm.org/D156349

Return gracefully instead of crashing. Add missing type conversion
tests.

Fixes: https://github.com/llvm/llvm-project/issues/61044

Reviewed By: qedawkins

Differential Revision: https://reviews.llvm.org/D156942

We currently spell check options that are listed as unsupported, but
this doesn't make much sense. If an option is explicitly unsupported
why would one that's spelled similarly be useful?

It looks like the reason this was added was that we explicitly mark
all `--something` flags as Unsupported rather than just leaving them
undefined and treating them as unknown. Drop that handling so that we
don't regress on things like misspelling `--help`.

Differential Revision: https://reviews.llvm.org/D156925

Check the interrupt flag while interpreting IR expressions and allow the
user to interrupt them.

Differential revision: https://reviews.llvm.org/D156822

On Fedora, rocminfo is a fedora package and rocm_agent_enumberator is
installed to /usr/bin.  This causes this error when building.

CMake Error at external/llvm-project/mlir/lib/ExecutionEngine/CMakeLists.txt:232 (message):
  Could not run rocm_agent_enumerator and ROCM_TEST_CHIPSET is not defined

So use find_program() to look for rocm_agent_enumerator instead of assuming a
single location.

Signed-off-by: Tom Rix <trix@redhat.com>

Reviewed By: krzysz00

Differential Revision: https://reviews.llvm.org/D156826

This reverts commit 8b76b44e.

SUMMARY OF CHANGES
------------------
This patch aims to reduce test duplication and to improve code re-use in
SparseTensor integration tests for CPU. This is a direct follow-up of:
  1. https://reviews.llvm.org/D155403 (test duplication), and
  2. https://reviews.llvm.org/D155405 (code re-use),

The key logic for this patch is implemented in:
  * SparseTensor/CPU/lit.local.cfg.
Essentially, the set-up that used to be repeated across all test files
has been extracted into a common LIT configuration file. This makes code
re-use straightforward.

All SVE/VLA tests are now enabled _conditionally_ and refactored to use
`mlir-cpu-runner` rather than `lli`. The former helps with test
duplication and the latter with code re-use.

A few additional refactoring changes are included.

1. The reduce verbosity, long runtime library names like:

  %mlir_native_utils_lib_dir/libmlir_c_runner_utils%shlibext

are replaced with:

  %mlir_c_runner_utils

2. In order to keep the code and the comments in sync, and to maintain
   consistency across the tests, the following:

  enable-runtime-library=true

is swapped with (and vice-versa):

  enable-runtime-library=false

Note that this change won't affect test coverage. Only few tests
required such update.

3. A VLS vectorization `RUN` line is added in tests where there was a
   VLA/VLS `RUN` line, but no VLS `RUN` line (with a few exceptions of
   tests that only contained one `RUN` line to begin with).

4. A few test variables are renamed/added. Most notable example:
  * %{options}` --> %{sparse_compiler_opts}

TEST RUNTIME IMPROVEMENT
------------------------
Tl;Dr This change improves test execution time by ~25%.

At the moment, the following `llvm-lit` invocation takes ~7.30s on my
AArch64 workstation (with SVE):

  llvm-lit  <llvm-project>/mlir/test/Integration/Dialect/SparseTensor/CPU/

This timing doesn't change no matter what the value of the following
CMake variable is (that should disable some tests):

  MLIR_RUN_ARM_SVE_TESTS

With this patch, the execution time will indeed depend on the value of
the above CMake variable:
  * with `MLIR_RUN_ARM_SVE_TESTS=true` the timing remains intact,
  * with `MLIR_RUN_ARM_SVE_TESTS=false` the timing drops to ~5.40s (~25%
    improvement).
This is expected:
  * on average there are 4 `RUN` lines per test,
  * _without this change_ (and with `MLIR_RUN_ARM_SVE_TESTS=false`) the
    4th `RUN` line would in most cases duplicate the 3rd `RUN` line,
  * _with this change) (and with `MLIR_RUN_ARM_SVE_TESTS=false`) the
    4th `RUN` line becomes empty.

PATCH SIZE
----------
While rather large and touching many files, most changes in this patch
are rather mechanical. All test configurations have been preserved and
only in a handful of cases new `RUN` lines added.

Differential Revision: https://reviews.llvm.org/D156625

Two run lines fail due to missing device libs for gfx902, which are fixed by
using gfx1010 which has device lib.

The other two spack tests have to be removed since there is no reliable
way to make them pass/fail not depending on whether there is ROCm
installed on the system.

Reviewed by: Fangrui Song, Joseph Huber

Differential Revision: https://reviews.llvm.org/D156935

* Allow pull request for clang-format changes

Remove repo lockdown on clang-format specific directories

* Missing / from unittests

"Flags" usually refers to boolean options and is used as such in other
places of Options.td (e.g. "Target-dependent compilation options",
"<clang-cl ignored options>").
This patch changes some misnomer group DocName from "flags" to
"options".

Other libc implementations support underscores in NaN(n-char-sequence)
strings. Us not supporting that is causing fuzz failures, so this patch
solves the problem.

Reviewed By: lntue

Differential Revision: https://reviews.llvm.org/D156927

CodeGen/X86/pseudo_cmov_lower2.ll fails using libc++ debug mode
(D150264) without this change.

Reviewed By: MaskRay, aeubanks

Differential Revision: https://reviews.llvm.org/D155811

Breaks CSKY tests.

This reverts commit 948f205a.

This patch fixes failing tests after checking the return code from the
driver. This is mostly due to the ROCm libraries not being present
during most compilations. Passing `-nogpuinc` should allow us to compile
without it for tests that require it. Additionally, some old tests set
the architecture of Nvidia tests to `sm_35` which is officially
unsupported in CUDA 12+ so it prints an error. We just increase in this
case.

Reviewed By: MaskRay, yaxunl

Differential Revision: https://reviews.llvm.org/D156930

Revert "Reapply: [MemCpyOpt] implement single BB stack-move optimization which unify the static unescaped allocas"""

Breaks Asan and LTO.

This reverts commit ea72b513.

This simplifies the op and avoids unnecessary alias checks introduced during the lowering to memref.

Reviewed By: springerm

Differential Revision: https://reviews.llvm.org/D156807

These test cases previously caused an error. RISCVInstrInfo::copyPhysReg also needed a tweak in order to account for copying bf16 values in FPR16 registers.

Differential Revision: https://reviews.llvm.org/D156883

Otherwise if the default target triple has `HasNativeLLVMSupport() ==
false`, the %clang command will exit with code 1 due to
err_drv_no_linker_llvm_support.

The issue is exposed by D156363.

I want these to have RISC-V semantics not LLVM IR semantics. Specifically
that -0.0 comes before +0.0.

This is needed to emulate FMAXIMUM/FMINIMUM for vectors.

This reverts commit 145a9290.

Review requested by @ast and @yonghong-song.

This reverts commit 82bc1839.

ASan catches this easily enough.

This patch enables linking of the static archive when fsanitize-stable-abi is set and stops linking to the asan dylib.

To link to the Address Sanitizer stable abi static library use "-fsanitize=address -fsanitize-stable-abi"

Updates a test with these flags.

rdar://112480890

Differential Revision: https://reviews.llvm.org/D156718

Currently, the example workspace files do not work properly. The
llvm_disable_optional_support_deps option was removed in 7b5d6cd7
and zlib and zstd have been needed-by-default dependencies for a while,
so it would make sense to show definitions in the example bazel
workspace files, especially given the version sensitivity of the zlib-ng
build. This patch removes the use of dated build config flags and adds
dependency definitions to both of the example workspaces.

Reviewed By: aaronmondal

Differential Revision: https://reviews.llvm.org/D156654

When compiling Rust code we sometimes see incomplete debug info leading
to crashes. Narrow the interfaces so we can see where it happens.

Reviewed By: ajwerner

Differential Revision: https://reviews.llvm.org/D156443

This patch applies the semantic checks for executable allocation directives to the new allocators construct.  It also introduces a new check that ensures all items in the list appear in the corresponding Fortran allocate statement.

Reviewed By: kiranchandramohan

Differential Revision: https://reviews.llvm.org/D150428

For character type with unknown length we end up generating
a GEP with the base type `llvm.ptr<i[width]>`. The GEP produces
the address of the first element of the slice, and it should be
using the offset computed in the number of characters, while we were
providing the offset in bytes.

Simple reproducer fails with and w/o HLFIR:
```
program test
  integer,parameter :: ck = 4
  character(:,ck),allocatable :: res(:,:)
  allocate(character(3,ck) :: res(2,2))
  res(1,1) = ck_'111'
  res(1,2) = ck_'222'
  res(2,1) = ck_'333'
  res(2,2) = ck_'444'
  call check(res)
contains
  subroutine check(res)
    character(:,ck),allocatable :: res(:,:)
    print *, res(2,:)
  end subroutine check
end program test
```

Reviewed By: clementval

Differential Revision: https://reviews.llvm.org/D156849

A section of a parameter array may be non-contiguous,
so the current !IsVariable(expr) check is too optimistic
to claim contiguity.

This patch fixes issues with incorrect hlfir.designate op generated
during lowering: the lowering queries IsContiguous to decide whether
to use fir.box<fir.array> or plain fir.ref<fir.array> to represent
the designator result.

Reviewed By: klausler

Differential Revision: https://reviews.llvm.org/D156494

This patch improves the lowering by changing target LLVM intrinsics from
`reduce.fmax` and `reduce.fmin`,
which have different semantic for handling NaN,
to `reduce.fmaximum` and `reduce.fminimum` ones.

Fixes #63969

Depends on D155869

Reviewed By: dcaballe

Differential Revision: https://reviews.llvm.org/D155877

Depends on D141672

Differential Revision: https://reviews.llvm.org/D138809

There were some checks removed previously in
bc196970.

However, all these SPIs are actually defined in macOS 12 and onward, not
macOSX 10.12 as the previous commit would suggest. As a result, if you
have access to these SPIs lldb will fail to compile correctly.

Instead of adding back the __builtin_availability checks, it seems
easier just to check the minimum deployment target with Availability
macros.

Differential Revision: https://reviews.llvm.org/D156838

Add ExtractAPI support C++ classes, fields,  methods, and various qualifiers and specifiers

Differential Revision: https://reviews.llvm.org/D153557

This reverts commit d9b1036b.

Bots are showing breakage.