This the second time I've stepped on this landmine, I'll look at setting
a lit local config. All the tests in this dir are going to be X86 for now.

This patch touches two optimizations, TwoAddressInstruction and X86's
FixupLEAs pass, both of which optimize by re-creating instructions. For
LEAs, various bits of arithmetic are better represented as LEAs on X86,
while TwoAddressInstruction sometimes converts instrs into three address
instructions if it's profitable.

For debug instruction referencing, both of these require substitutions to
be created -- the old instruction number must be pointed to the new
instruction number, as illustrated in the added test. If this isn't done,
any variable locations based on the optimized instruction are
conservatively dropped.

Differential Revision: https://reviews.llvm.org/D85756

There are bunch of optimization opportunities right now in the vector
popcnt code gen when doing simple less-than/greater-than comparisons, so
let's examine them all to ensure that things don't regress as different
scenarios are fixed. We can always delete some later once some fixes are
made.

Please note: the new files were auto-generated. If people want, I can
commit the short C code that printed out the various combinations.

https://llvm.discourse.group/t/rfc-standard-memref-cast-ops/1454/15

Differential Revision: https://reviews.llvm.org/D89784

The existing tests were mostly for 64-bit constants.

Differential Revision: https://reviews.llvm.org/D83210

This better reflects what this variable is about.

For more details about these instructions, please refer to the latest
ISE document:
https://software.intel.com/en-us/download/intel-architecture-instruction-set-extensions-programming-reference.

Reviewed By: craig.topper

Differential Revision: https://reviews.llvm.org/D89301

Now libc++ pipeline will be triggered from the "premerge-checks" and the
combined result are going to be returned to Harbormaster.

Reviewed-by: ldionne

Differential Revision: https://reviews.llvm.org/D89113

This better reflects what this logic actually does.

This reverts commit 3fce5ea7.

`make check` broken.

This improves simplifications for pattern `icmp (X+Y), (X+Z)` -> `icmp Y,Z`
if only one of the operands has NSW set, e.g.:

    icmp slt (x + 0), (x +nsw 1)

We can still safely rewrite this to:

    icmp slt 0, 1

because we know that the LHS can't overflow if the RHS has NSW set and
C1 < C2 && C1 >= 0, or C2 < C1 && C1 <= 0

This simplification is useful because ScalarEvolutionExpander which is used to
generate code for SCEVs in different loop optimisers is not always able to put
back NSW flags across control-flow, thus inhibiting CFG simplifications.

Differential Revision: https://reviews.llvm.org/D89317

findNearestCommonDominator never returns nullptr.

Make these types conform to the LLVM Coding Standards:

> Type names (including classes, structs, enums, typedefs, etc) should
> be nouns and start with an upper-case letter.

target Apple Silicon macOS machines

Differential Revision: https://reviews.llvm.org/D82699

Implement the corresponding thing using windows functions as well.

Differential Revision: https://reviews.llvm.org/D89864

The individual enum values in copy_options and file_type aren't
specified in the standard.

The standard doesn't require fs::path::format to be a scoped enum.

Differential Revision: https://reviews.llvm.org/D89866

Differential Revision: https://reviews.llvm.org/D89867

Copy over the compiler detection structure from libcxx, and set
_LIBCXXABI_WEAK like _LIBCPP_WEAK is set in libcxx.

This allows users to override operator new/delete, if using those
operators from libcxxabi instead of from libcxx.

Differential Revision: https://reviews.llvm.org/D89863

- Make the SIMemoryLegalizer insertAcquire function be in the same
  order for each target to be consistent.

Differential Revision: https://reviews.llvm.org/D89880

A recent commit to revert llvm-symbolizer changes forgot to revert this test fix.

This reverts commit 5e656ee4.

Many of these tests don't use the output of -analyze.

Use BFI if it is available and BPI otherwise.
This is a promised follow-up after D89541.

Reviewers: ebrevnov, mkazantsev
Reviewed By: ebrevnov
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D89773

Differential Revision: https://reviews.llvm.org/D89616

-analyze does not work with the NPM.
'print<foo>' passes should be used instead.

Split out of D89086 as suggested.

Change the default of the log_path flag to nullptr, and the code
consuming that flag (ReportFile::SetReportPath), to treat nullptr as
stderr (so no change to the behavior of existing users). This allows
code to distinguish between the log_path being specified explicitly as
stderr vs the default.

This is so the flag can be used to override the new report path variable
that will be encoded in the binary for memprof for runtime testing.

Differential Revision: https://reviews.llvm.org/D89629

Reviewed By: nickdesaulniers, MaskRay

Differential Revision: https://reviews.llvm.org/D88631

account when determining the identity of a class NTTP.

[FPEnv][X86][SystemZ] Use different algorithms for i64->double uint_to_fp under strictfp to avoid producing -0.0 when rounding toward negative infinity

Some of our conversion algorithms produce -0.0 when converting unsigned i64 to double when the rounding mode is round toward negative. This switches them to other algorithms that don't have this problem. Since it is undefined behavior to change rounding mode with the non-strict nodes, this patch only changes the behavior for strict nodes.

There are still problems with unsigned i32 conversions too which I'll try to fix in another patch.

Fixes part of PR47393

Reviewed By: efriedma

Differential Revision: https://reviews.llvm.org/D87115

Split off from D89251

Reviewed By: vitalybuka

Differential Revision: https://reviews.llvm.org/D89884

This reverts commit 716f7636.