One AVX2+ targets we have a immediate VPERMQ/PD cross-lane permute thats better than relying on a pair of VPERM2F128 cross-lanes to feed a blend.

Reported on discourse

I incorrectly used freeze(undef) in rG1b07bd90

Differential Revision: https://reviews.llvm.org/D125593

`std::views::drops` and `std::views::join` have been implemented, but the tests
verifying the CPOs for them are still commented out. Uncomment the tests.

Differential Revision: https://reviews.llvm.org/D125618

Use early out for the i8 case.

I'm looking at avoiding MUL on targets that use libcalls for MUL.
So doing a little pre-refactoring.

Reviewed By: MaskRay

Differential Revision: https://reviews.llvm.org/D106888

This is an example usage of D120958.

After these patches are landed, we can strip off the codebeads officially.

Reviewed By: myhsu

Differential Revision: https://reviews.llvm.org/D120960

The name `MCFixedLenDisassembler.h` is out of date after D120958.

Rename it as `MCDecoderOps.h` to reflect the change.

Reviewed By: myhsu

Differential Revision: https://reviews.llvm.org/D124987

Add support for translating llvm::ShuffleVectorInst

Differential Revision: https://reviews.llvm.org/D125030

Add support for translating llvm::InsertValue and llvm::ExtractValue.

Differential Revision: https://reviews.llvm.org/D125028

ISD::VSCALE only allows constant operands.

Quite a few C++20 LWG issues/papers related to the One Ranges Proposal
were already effectively implemented (or contain semantic-only wording
changes that don't affect the implementation), mark them as such.

Differential Revision: https://reviews.llvm.org/D125065

This simplifies the string structs a bit more and the normal layout should not contain any undefined behaviour anymore. I don't think there is a way to achieve this in the alternate string mode without breaking the ABI.

Reviewed By: ldionne, #libc

Spies: libcxx-commits

Differential Revision: https://reviews.llvm.org/D125496

Clang erroneously flagged the function as "unused", but it is most
definitely used by gtest to pretty print the parameter value when
a test fails.

Make the pretty printing function a friend function in the parameter
class similar to other clang unit tests.

v_cttz_zero_undef_i64_with_select should be selecting '64' for the x != 0 case instead of '32' like we just did in the previous 'v_cttz_zero_undef_i32_with_select' test.

Noticed by accident because it was causing some weird regressions....

Differential Revision: https://reviews.llvm.org/D125612

Also, initialize entire mask to -1 to simplify undefined cases.

st_size may not be of importance to the abi if you are not using
copy relocations. This is helpful when you want to check the abi
of a shared object both when instrumented and not because asan
will increase the size of objects to include the redzone.

Differential revision: https://reviews.llvm.org/D124792

SelectionDAG::FoldConstantArithmetic determines if operands are foldable constants, so we don't need to bother with isConstantOrConstantVector / Opaque tests before calling it directly.

This reverts commit b6b0fd6a.

st_size may not be of importance to the abi if you are not using
copy relocations. This is helpful when you want to check the abi
of a shared object both when instrumented and not because asan
will increase the size of objects to include the redzone.

Differential revision: https://reviews.llvm.org/D124792

crashlog.py catches every exception in order to format them. This
results in both the exception name as well as the backtrace getting
swallowed.

Here's an example of the current output:

  error: python exception: in method 'SBTarget_ResolveLoadAddress', argument 2 of type 'lldb::addr_t'

Compare this to the output without the custom exception handling:

  Traceback (most recent call last):
    File "[...]/site-packages/lldb/macosx/crashlog.py", line 929, in __call__
      SymbolicateCrashLogs(debugger, shlex.split(command))
    File "[...]/site-packages/lldb/macosx/crashlog.py", line 1239, in SymbolicateCrashLogs
      SymbolicateCrashLog(crash_log, options)
    File "[...]/site-packages/lldb/macosx/crashlog.py", line 1006, in SymbolicateCrashLog
      thread.dump_symbolicated(crash_log, options)
    File "[...]/site-packages/lldb/macosx/crashlog.py", line 124, in dump_symbolicated
      symbolicated_frame_addresses = crash_log.symbolicate(
    File "[...]/site-p...

Previously, creating a zero floating-point constant used MOVID even when
NEON was disabled which resulted in the following fatal error:
`Attempting to emit MOVID instruction but the Feature_HasNEON predicate(s) are not met`

Reviewed By: dmgreen

Differential Revision: https://reviews.llvm.org/D125237

Variable captures such as `<MCInst #` can change based on unrelated changes
to the LLVM backends, to avoid the generated test cases being different
use an incrementing counter for variable names instead of using the
actual value from the output file.
This change may also be beneficial for some nameless IR variables
(especially when combined with filtering of output), but for now I've
restricted this change to the obvious candidates (--asm-show-inst output).

Reviewed By: MaskRay

Differential Revision: https://reviews.llvm.org/D125405

To avoid test churn when backends add/rename new instructions/registers,
it makes sense to use FileCheck captures for the exact MCInst/Reg number.
This is motivated by D125307, where I use --asm-show-inst to differentiate
the output for multiple instructions with the same mnemonic.

This does not quite fix the churn issue yet: While files with the generated
checks will be immune to the numbers changing, the update script test
still suffers from this problem since the number is encoded in the
FileCheck variable name. I plan to address this in a follow-up patch.

Reviewed By: MaskRay

Differential Revision: https://reviews.llvm.org/D125307

To avoid test churn when backends add/rename new instructions/registers,
it makes sense to scrub the exact MCInst/Reg number.

Differential Revision: https://reviews.llvm.org/D125305

This avoids repeated calls to get_idx_from_ir_value_match() and will make
it easier for a future patch that adds new assembly-level nameless values
in addition to the IR ones.

Differential Revision: https://reviews.llvm.org/D125390

I was trying to compile code with -march=+nosimd and hit various
instruction predicate verification errors, this patch should address the
ones I saw in integer to floating-pointer conversions.

I noticed that for signed conversions, some non-NEON instruction sequences
are shorter. I don't know if the longer one is still faster on current
architectures (the patterns date back to the initial backend import)

Reviewed By: dmgreen

Differential Revision: https://reviews.llvm.org/D125308

Differential Revision: https://reviews.llvm.org/D125240

Erase gpu.memcpy op when only uses of dest are
the memcpy op in question, its allocation and deallocation
ops.

Reviewed By: bondhugula, csigg

Differential Revision: https://reviews.llvm.org/D124257

SelectionDAG::FoldConstantArithmetic determines if operands are foldable constants, so we don't need to bother with isConstantOrConstantVector / Opaque tests before calling it directly.