Exception handling SHOULD be possible depending on codegen (and if not,
we can make it trap and add a warning when we make that decision), but
throwing out of a compute construct is ill formed.

This patch adds an error for a 'throw' that isn't in a 'try' scope. This
error is imperfect as it won't diagnose a 'throw' that escapes its
'try', or one in a separate function, but it catches the obvious
mistakes. The other cases will need to be handled as runtime failures.

When the result of real input editing underflows to zero, return -0.
when the input field had a minus sign.

Move symbol names from directly under `objc` scope to
`objc::symbol_names`.
Ex: `objc::klass` -> `objc::symbol_names::klass`

Co-authored-by: Alex B <alexborcan@meta.com>

f18's module files are Fortran with a leading header comment containing
the module file format version and a hash of the following contents.
This hash is currently used only to protect module files against
corruption and truncation.

Extend the use of these hashes to catch or avoid some error cases. When
one module file depends upon another, note its hash in additional module
file header comments. This allows the compiler to detect when the module
dependency is on a module file that has been updated. Further, it allows
the compiler to find the right module file dependency when the same
module file name appears in multiple directories on the module search
path.

The order in which module files are written, when multiple modules
appear in a source file, is such that every dependency is written before
the module(s) that depend upon it, so that their hashes are known.

A warning is emitted when a module file is not the first hit on the
module file search path.

Further work is needed to add a compiler option that emits (larger)
stand-alone module files that incorporate copies of their dependencies
rather than relying on search paths. This will be desirable for
application libraries that want to ship only "top-level" module files
without needing to include their dependencies.

Another future work item would be to admit multiple modules in the same
compilation with the same name if they have distinct hashes.

Summary:
We use `add_llvm_library` as a shorthand for setting up all the
dependencies and libraries we need for the OpenMP offloading runtime as
they depend on a lot of the LLVM utilities. However, we always
explicitly installed these manually. Behind the scenes the function
would then install it again. This was unnoticed because until now the
destinations matched. Now that we want it to optionally go into the
other directory it is duplicating them. Fix this by stating that this is
a build tree only library so we can handle it ourselves.

It makes sense to split if the cost of computing `cond1` is high
(proportionally to how likely `cond0` is), but it doesn't really make
sense to introduce a second branch if its only a few instructions.

Splitting can also get in the way of potentially folding patterns.

This patch introduces some logic to try to check if the cost of
computing `cond1` is relatively low, and if so don't split the
branches.

Modest improvement on clang bootstrap build:
https://llvm-compile-time-tracker.com/compare.php?from=79ce933114e46c891a5632f7ad4a004b93a5b808&to=978278eabc0bafe2f390ca8fcdad24154f954020&stat=cycles
Average stage2-O3:   0.59% Improvement (cycles)
Average stage2-O0-g: 1.20% Improvement (cycles)

Likewise on llvm-test-suite on SKX saw a net 0.84% improvement  (cycles)

There is also a modest compile time improvement with this patch:
https://llvm-compile-time-tracker.com/compare.php?from=79ce933114e46c891a5632f7ad4a004b93a5b808&to=978278eabc0bafe2f390ca8fcdad24154f954020&stat=instructions%3Au

Note that the stage2 instruction count increases is expected, this
patch trades instructions for decreasing branch-misses (which is
proportionately lower):
https://llvm-compile-time-tracker.com/compare.php?from=79ce933114e46c891a5632f7ad4a004b93a5b808&to=978278eabc0bafe2f390ca8fcdad24154f954020&stat=branch-misses

NB: This will also likely help for APX targets with the new `CCMP` and
`CTEST` instructions.

Closes #81689

This patch allows using the name "rdma" as an alias for "rdm". The name
makes its way to target attributes as well as the command line via the
-march and -mcpu options. The motivation was originally to support this
in Function Multi Versioning but it also makes sense to align with GCC
on the command line.

Multiple formatting operator tests are commented. This change enables
them again.

[PR506629](https://llvm.org/PR50629) fixed by
[D153798](https://reviews.llvm.org/D153798).

Tests in `ConfigurableSpaceBeforeParens` have the same behavior as
before [D110833](https://reviews.llvm.org/D110833).

Update the test for function declaration and definition, as changed in
[D114696](https://reviews.llvm.org/D114696).

This is in preparation for
https://github.com/llvm/llvm-project/pull/77790 and makes it easy to add
other, non-power-of-2 VFs for processing.

PR: https://github.com/llvm/llvm-project/pull/82793

The code suggestion was neglected when the patch landed.

Previously, the function `install_bad_alloc_error_handler` was asserting that a bad_alloc error handler was not already installed. However, it was actually checking for the fatal-error handler, not for the bad_alloc error handler, causing an assertion failure if a fatal-error handler was already installed.

Fixes #83040

As discussed in https://github.com/llvm/llvm-project/pull/81427/files#r1509266817
rename the variable as a separate commit

- Add 'Init Undef Pass', which is target-independent now.

Exit early if MaxVF < MinVF. In that case, the loop body below will
never get entered. Note that this adjusts the condition from MaxVF <=
MinVF. If MaxVF == MinVF, vectorization may still be feasible (and the
loop below gets entered).

PR: https://github.com/llvm/llvm-project/pull/83283

Blank comment line afer \endverbatim malfunctions the doxygen parser.
  Check the "Detailed Description" section on
https://llvm.org/doxygen/classllvm_1_1vfs_1_1RedirectingFileSystem.html


Screenshot is attached below


![image](https://github.com/llvm/llvm-project/assets/22870466/35fc2444-d2f4-4638-990e-241cc14d0986)

Equivalent to "umax(A, B) - umin(A, B)"

This is just an initial correctness implementation, hopefully we can make this optimal in the future.

Shows cases where logical shifts of allsignbits values can be profitably converted to masks

g++ -flto has a diagnostic `-Wodr` about mismatched redeclarations,
which even apply to `enum`.

Fix #83529

Reviewers: thesamesam

Reviewed By: thesamesam

Pull Request: https://github.com/llvm/llvm-project/pull/83604

The help output for `thread backtrace` specifies that you can pass -1 to
`--count` to display all the frames.

```
-c <count> ( --count <count> )
            How many frames to display (-1 for all)
```

However, that doesn't work:

```
(lldb) thread backtrace --count -1
error: invalid integer value for option 'c'
```

The problem is that we store the option value as an unsigned and the
code to parse the string correctly rejects it. There's two ways to fix
this:

1. Make `m_count` a signed value so that it accepts negative values and
appease the parser. The function that prints the frames takes an
unsigned so a negative value will just become a really large positive
value, which is what the current implementation relies on.
2. Keep `m_count` unsigned and instead use 0 the magic value to show all
frames. I don't really see a point in not showing any frames at all,
plus that's already broken (`error: error displaying backtrace for
thread: "0x0001"`).

This patch implements (2) and at the same time improve the error
reporting so that we print the invalid value when we cannot parse it.

rdar://123881767

Summary:
These functions were implemented via builtins that aren't acually
supported. See https://godbolt.org/z/Wq6q6T1za. This caused the build to
crash if they were included. Remove these and replace with correct
implementations.

This commit adds the functionality to broadcast events using the
`Debugger::eBroadcastProgressCategory`
bit (https://github.com/llvm/llvm-project/pull/81169) by keeping track
of these reports with the `ProgressManager`
class (https://github.com/llvm/llvm-project/pull/81319). The new bit is
used in such a way that it will only broadcast the initial and final
progress reports for specific progress categories that are managed by
the progress manager.

This commit also adds a new test to the progress report unit test that
checks that only the initial/final reports are broadcasted when using
the new bit.

Masked gather is very expensive operation and sometimes better to
represent it as a serie of consecutive/strided loads + insertsubvectors
sequences. Patch adds some basic estimation and if loads+insertsubvector
is cheaper, decides to represent it in this way rather than masked
gather.

Reviewers: RKSimon

Reviewed By: RKSimon

Pull Request: https://github.com/llvm/llvm-project/pull/83481

This is in preparation for PR
https://github.com/llvm/llvm-project/pull/82533.

Here we Introduce a test for asan instrumentation where invalid code is
output (see bug https://github.com/llvm/llvm-project/issues/64990)
The `CHECK` lines are generated using `update_test_checks.py` script.
The actual fix PR will udpate this test to highlight the changes in the
generated code.

WaveSizePredicate is copied from pseudo to real

This change adds lowering for HLSL's frac intrinsic to DXIL.

This change should complete #70099

If the graph incudes only single "gather" node with only
extractelements/undefs, which used only in insertelement-based
buildvector sequences, it still might be profitable to vectorize it.
Need to rely on the cost model, not throw this graph away immediately.

Reviewers: RKSimon

Reviewed By: RKSimon

Pull Request: https://github.com/llvm/llvm-project/pull/83581

Similar to PR #83364.

For denormals we have to account for the exponent coming
from the significand.

hlsl_intrinsics.h - add the round  api
DXIL.td add the llvm intrinsic to DXIL lowering mapping 
This change reuses llvm's existing intrinsic
`__builtin_elementwise_round`\ `int_round`
This change implements: #70077

Broke a build bot: https://lab.llvm.org/buildbot/#/builders/124/builds/9846

This reverts commit e7c3cd24.

We can support these by changing the sext promotion to -zext(-C) and
replacing a sgt check with ugt. Reframing the logic in terms of how the
unsigned range are affected. More comments in the patch.

The new cases check isLegalAddImmediate to avoid some
regressions in lit tests.

If they are captured by value.

[Sema] -Wpointer-bool-conversion: suppress lambda function pointer conversion diagnostic during instantiation (#83497)

I have seen two internal pieces of code that uses a template type
parameter to accept any callable type (function pointer, std::function,
closure type, etc). The diagnostic added in #83152 would require
adaptation to the template, which is difficult and also seems
unnecessary. Example:

```cpp
template <typename... Ts>
static bool IsFalse(const Ts&...) { return false; }

template <typename T, typename... Ts,
          typename = typename std::enable_if<std::is_constructible<bool, const T&>::value>::type>
static bool IsFalse(const T& p, const Ts&...) {
  return p ? false : true;
}

template <typename... Args>
void Init(Args&&... args) {
  if (IsFalse(absl::implicit_cast<const typename std::decay<Args>::type&>(
              args)...)) {
    // A callable object convertible to false is either a null pointer or a
    // null functor (e.g., a default-constructed std::function).
    empty_ = true;
  } else {
    empty_ = false;
    new (&factory_) Factory(std::forward<Args>(args)...);
  }
}
```