After refactoring the phi recipes, we can now iterate over all header
phis in a VPlan to detect reductions when it comes to fixing them up
when tail folding.

This reduces the coupling with the cost model & legal by using the
information directly available in VPlan. It also removes a call to
getOrAddVPValue, which references the original IR value which may
become outdated after VPlan transformations.

Reviewed By: Ayal

Differential Revision: https://reviews.llvm.org/D100102

This is related to PR51392.

Before this patch, the timeline view was rounding doubles to the first decimal,
using a logic similar to this:

```
  double AverageTime = (double)Input / CumulativeExecutions;
  double Result = floor((AverageTime * 10) + 0.5) / 10
```

Here, Input and CumulativeExecutions are both unsigned integers.
The last operation is what effectively performs the rounding of AverageTime.

PR51392 has been raised because - under specific -m32 configurations of GCC -
one of the timeline tests reports slighlty different values (due to a different
rounding choice).

This patch tries to minimise the propagation of floating-point error by
hoisting the multiply by 10, so that it is performed on the unsigned.

```
  double AverageTime = (double)(Input * 10) / CumulativeExecutions;
  floor(AverageTime + 0.5) / 10
```

So we are trading a floating point multiply for a integer multiply (which can be
expanded using a simple MUL or using an `ADD + LEA` sequence). This decrease in
floating point operations executed should also help with decreasing the error in
the computation..

Strictly speaking, that computation will always be potentially subject to error
(depending on what values are passed in input). However, this patch should
improve the situation and make bug like PR51392 less frequent.

This patch enables compressed input sections on big-endian targets by
checking the target endianness and selecting an appropriate `Chdr`
structure.

Fixes PR51369

Differential Revision: https://reviews.llvm.org/D107635

We should use MachineInstr::eraseFromParentAndMarkDBGValuesForRemoval()
instead of eraseFromParent().

We should probably use that in other places too but fix this issue which
affects clang bootstrap builds for now.

We were using an OpaqueValueExpr allocated on the stack to store
the size of a VLA. Because the VLASizeMap in CodegenFunction
uses the address of the expression to avoid recomputing VLAs,
we were accidentally reusing an earlier llvm::Value. This led to
invalid LLVM IR.

This is a temporary solution until VLASizeMap can be pushed and popped
based on the context.

Differential Revision: https://reviews.llvm.org/D107666

This commit adds the isnan intrinsic and provides a default expansion
for it in the SDAG. However, it makes the assumption that types
it operates on are IEEE-compliant types. This is not always the case.
An example of that is PPC "double double" which has a representation
that
- Does not need to conform to IEEE requirements for isnan as it is
  not an IEEE-compliant type
- Does not have a representation that allows for straightforward
  reinterpreting as an integer and use of integer operations

The result was that this commit broke __builtin_isnan for ppc_fp128
making many valid numeric values report a NaN.

This patch simply changes the expansion to always expand to unordered
comparison (regardless of whether FP exceptions are tracked). This
is inline with previous semantics.

Reviewed by: benshi001

Differential Revision: https://reviews.llvm.org/D107672

The _zx_vmar_root_self function is not a system call but
a libc function declared in a separate header.

Reviewed By: gulfem

Differential Revision: https://reviews.llvm.org/D107616

Upstream support for NSConstantArray, NSConstantIntegerNumber,
NSConstant{Float,Double}Number and NSConstantDictionary.

We would've upstreamed this earlier but testing it requires
-fno-constant-nsnumber-literals, -fno-constant-nsarray-literals and
-fno-constant-nsdictionary-literals which haven't been upstreamed yet.
As a temporary workaround use the system compiler (xcrun clang) for the
constant variant of the tests.

I'm just upstreaming this. The patch and the tests were all authored by
Fred Riss.

Differential revision: https://reviews.llvm.org/D107660

This reverts commit 34d78b6a.

This breaks build bots witha  missing file:
/home/worker/2.0.1/lldb-x86_64-debian/llvm-project/lldb/source/Plugins/Language/ObjC/Cocoa.cpp:10:10: fatal error: 'objc/runtime.h' file not found

The pointer to the dyld trie data structure which lldb needs to parse to get
"trampoline kinds" on Darwin used to be a field in the LC_DYLD_INFO load command. A
new load command was added recently dedicated to this purpose: LC_DYLD_EXPORTS_TRIE.
The format of the trie did not change, however. So all we have to do is use the new
command if present. The commands are supposed to be mutually exclusive, so I added
an lldb_assert to warn if they are not.

Differential Revision: https://reviews.llvm.org/D107673

Perform scalar constant propagation for FPTruncOp only if the resulting value can be represented without precision loss or rounding.

Example:
%cst = constant 1.000000e+00 : f32
%0 = fptrunc %cst : f32 to bf16
-->
%cst = constant 1.000000e+00 : bf16

Reviewed By: mehdi_amini

Differential Revision: https://reviews.llvm.org/D107518

These 3 are fairly simple, pipes, workgroups and subgroups.

Reviewed By: Anastasia

Differential Revision: https://reviews.llvm.org/D105858

Adds NVPTX intrinsics for the CUDA PTX `ldmatrix.sync.aligned` instructions added in PTX 6.5.

PTX ISA description of `ldmatrix.sync.aligned`: https://docs.nvidia.com/cuda/parallel-thread-execution/index.html#warp-level-matrix-instructions-ldmatrix



Authored-by: Steffen Larsen <steffen.larsen@codeplay.com>

Reviewed By: tra

Differential Revision: https://reviews.llvm.org/D107046

Upstream support for NSConstantArray, NSConstantIntegerNumber,
NSConstant{Float,Double}Number and NSConstantDictionary.

We would've upstreamed this earlier but testing it requires
-fno-constant-nsnumber-literals, -fno-constant-nsarray-literals and
-fno-constant-nsdictionary-literals which haven't been upstreamed yet.
As a temporary workaround use the system compiler (xcrun clang) for the
constant variant of the tests.

I'm just upstreaming this. The patch and the tests were all authored by
Fred Riss.

Differential revision: https://reviews.llvm.org/D107660

The test shows excessive register pressure after the MachineLICM.
This is a pre-commit for the patch fixing it.

Differential Revision:

This provides a convenient way to limit a breakpoint
to the current thread when setting it from the command line w/o
having to figure out what the current thread is.

Differential Revision: https://reviews.llvm.org/D107015

This reverts commit 100a7b61.

Speculating that this is the reason behind a sanitizer failure:
https://lab.llvm.org/buildbot/#/builders/37/builds/5945

After D94315 we add the `NoInline` attribute to the outlined function to handle
data environments in the OpenMP if clause. This conflicted with the `AlwaysInline`
attribute added to the outlined function. for better performance in D106799.
The data environments should ideally not require NoInline, but for now this
fixes PR51349.

Reviewed By: mikerice

Differential Revision: https://reviews.llvm.org/D107649

There may be some generalizations (see test comments) of these patterns,
but this should handle the cases motivated by:
https://llvm.org/PR51315
https://llvm.org/PR51259

The backend may want to transform differently, but at least for
the x86 examples that I looked at, there does not appear to be
any significant perf diff either way.

https://llvm.org/PR51315

- Replace 'cmp+sel' with 'umin' if possible.

Reviewed By: foad

Differential Revision: https://reviews.llvm.org/D107507

Merge two versions of `initializeProfileForContinuousMode` function into one.

Differential Revision: https://reviews.llvm.org/D107591

The diagnostic texts for warning on attributes that don't appear on the
initial declaration is generally useful.  We'd like to re-use it in
D106030, but first let's combine two that already are very similar so we
may re-use it a third time in that commit.

Also, fix a few places that were using notePreviousDefinition to point
to declarations, to instead use diag::note_previous_declaration.

Reviewed By: aaron.ballman

Differential Revision: https://reviews.llvm.org/D107613

These aren't terribly common, but we currently mishandle them badly.
Not only do we not recogize the attributes themselves, but we often end up
selecting some node other than the parent (because source ranges aren't accurate
in the presence of attributes).

Differential Revision: https://reviews.llvm.org/D89785

This reverts commit 3241680f.
Fixed mangled post-test formatting :-(

This reverts commit a4bdcdad.

Fails bots:
https://lab.llvm.org/buildbot/#/builders/109/builds/20231/steps/6/logs/stdio

Differential Revision: https://reviews.llvm.org/D89743

It looks like recent CoreFoundation builds strip the non-public symbol
that we were looking for to find the 2 boolean "classes". The public
symbol is of course there, and it contains the address of the private
one. If we don't find the private symbol directly, go through a memory
read at the public symbol's location instead.

DAG is unnecessary and we need this hook to implement store merging on GlobalISel too.

On FreeBSD, the system `<libelf.h>` already declares `struct Elf_Note`
indirectly (via `<sys/elf_common.h>`). This results in compile errors
when building the libomptarget amdgpu plugin. Avoid redeclaring `struct
Elf_Note` on FreeBSD to fix the errors.

Reviewed By: JonChesterfield

Differential Revision: https://reviews.llvm.org/D107661

    Differential Revision: https://reviews.llvm.org/D107611

This reverts commit 48ad446a.

Lift checks for -Werror into local functions.

Reviewed By: awarzynski

Differential Revision: https://reviews.llvm.org/D101261

Include the vaue of LLVM_ENABLE_NEW_PASS_MANAGER in generated
LLVMConfig.cmake since it is needed by clang's build system.  This fixes
test failures when the new pass manager is enabled (i.e. by default)
by having clang's CMake files correctly detect that and skip relevant
tests.

Differential Revision: https://reviews.llvm.org/D107628

Differential Revision: https://reviews.llvm.org/D107643

Attempt to enable MemCpyOpt unconditionally in D104801 uncovered the fact that
there are users that do not expect LLVM to materialize `memset` intrinsic.

While other passes can do that, too, MemCpyOpt triggers it more frequently and
breaks sanitizers and some downstream users.

For now introduce a flag to force-enable the flag and opt-in only CUDA
compilation with NVPTX back-end.

Differential Revision: https://reviews.llvm.org/D106401