Instead of the GNU extension `SHF_GNU_RETAIN`, Solaris provides equivalent
functionality with `SHF_SUNW_NODISCARD`. This patch implements the necessary
support.

Tested on `sparcv9-sun-solaris2.11`, `amd64-pc-solaris2.11`, and
`x86_64-pc-linux-gnu`.

Differential Revision: https://reviews.llvm.org/D107955

Rather than queuing up actions, have one function that does the
log2() fold in the obvious way, but with a flag that allows us
to check whether the fold will succeed without actually performing
it.

In D111530, I suggested that we add some relatively basic pattern-matching
folds for shifts and funnel shifts and avoid a more specialized solution
if possible.

We can start by implementing at least one of these in IR because it's
easier to write the code and verify with Alive2:
https://alive2.llvm.org/ce/z/qHpmNn

This will need to be adapted/extended for SDAG to handle the motivating
bug ( #49541 ) because the patterns only appear later with that example
(added some tests: bb850d42)

This can be extended within InstSimplify to handle cases where we 'and'
with a shift too (in that case, kill the funnel shift).
We could also handle patterns where the shift and funnel shift directions
are inverted, but I think it's better to canonicalize that instead to
avoid pattern-match case explosion.

Differential Revision: https://reviews.llvm.org/D120253

This mirrors the code structure in `lld/ELF`. It also paves the way for
an upcoming diff where I templatize things.

Reviewed By: #lld-macho, thakis

Differential Revision: https://reviews.llvm.org/D120376

This mirrors the code organization in `lld/ELF`.

Reviewed By: #lld-macho, thakis

Differential Revision: https://reviews.llvm.org/D120378

This reverts commit 169e1aba.

It also fixes an incorrect assumption in `initGlobalVars`.

What we're really doing here is converting Op0 udiv Op1 into
Op0 lshr log2(Op1), so phrase it in that way. Actually pushing
the lshr into the log2(Op1) expression should be seen as a separate
transform.

This patch enables inferring framework modules in explicit builds in all contexts. Until now, inferring framework modules only worked with `-fimplicit-module-maps` due to this block of code:

```
// HeaderSearch::loadFrameworkModule
  case LMM_InvalidModuleMap:
    // Try to infer a module map from the framework directory.
    if (HSOpts->ImplicitModuleMaps)
      ModMap.inferFrameworkModule(Dir, IsSystem, /*Parent=*/nullptr);
    break;
```

Reviewed By: Bigcheese

Differential Revision: https://reviews.llvm.org/D113880

The triple goes after the gcc version, not before. Also add the
/backward version.

Differential Revision: https://reviews.llvm.org/D120251

As a follow-up to eb4c8608
Should fix http://45.33.8.238/win/53749/step_4.txt

Related to https://reviews.llvm.org/D120195

The class is using an incredibly elaborate setup to create and destroy
an NSAutoreleasePool object. We can do it in a much simpler way by
making those calls inside our thread startup function.

The only effect of this patch is that the pool gets released at the end
of the ThreadCreateTrampoline function, instead of slightly later, when
pthreads begin thread-specific cleanup. However, the key destruction
order is unspecified, so nothing should be relying on that.

I didn't find a specific reason for why this would have to be done that
way in git history. It seems that before D5198, this was thread-specific
keys were the only way an os implementation (in Host::ThreadCreated)
could attach some value to a thread.

Differential Revision: https://reviews.llvm.org/D120322

Accept a function object instead of a raw pointer. This avoids a bunch
of boilerplate typically needed to pass arguments to the thread
functions.

Differential Revision: https://reviews.llvm.org/D120321

LLVM tools do not emit `DT_MIPS_XHASH` dynamic table tag. But now
`llvm-objdump` and `llvm-readelf` recognize this tag and print it.

Fixes https://github.com/llvm/llvm-project/issues/53996

This reverts commit f84d732f.
Breakage of "sanitizer-x86_64-linux-fast"

We were producing invalid instructions like "ubfx w0, w0, #20, #16".

Until now, subgroup builtins are available with `opencl-c.h` when at
least one of `cl_intel_subgroups`, `cl_khr_subgroups`, or
`__opencl_c_subgroups` is defined.  With `-fdeclare-opencl-builtins`,
subgroup builtins are conditionalized on `cl_khr_subgroups` only.

Align `-fdeclare-opencl-builtins` to `opencl-c.h` by introducing the
internal `__opencl_subgroup_builtins` macro.

Differential Revision: https://reviews.llvm.org/D120254

This function only works on the RHS operand.

This reverts commit 2acead35.

Let's try `REQUIRES: asserts`.

This reverts commit d16c5f41.

Let's try `REQUIRES: asserts`.

Before this change without any good reason Lexer::LangOpts is sometimes accessed by getter and another time read directly in Lexer functions.
Since getLangOpts is a bit more verbose prefer direct access to LangOpts member when possible.

Differential Revision: https://reviews.llvm.org/D120333

This is the same special logic we apply for SPF signed clamps
when computing the number of sign bits, just for intrinsics.

This just uses the same logic as the select case, but there's
multiple directions this could be improved in: We could also use
the num sign bits from the clamped value, we could do this during
constant range calculation, and there's probably unsigned analogues
for the constant range case at least.

Add intrinsic versions of existing SPF tests.

This switches the testcase committed for initial C++20 modules import tracking to
use split-file rather than preprocessor directives.

Differential Revision: https://reviews.llvm.org/D120352

[RISCV] Avoid infinite loop between DAGCombiner::visitMUL and RISCVISelLowering::transformAddImmMulImm

See https://github.com/llvm/llvm-project/issues/53831 for a full discussion.

The basic issue is that DAGCombiner::visitMUL and
RISCVISelLowering;:transformAddImmMullImm get stuck in a loop, as the
current checks in transformAddImmMulImm aren't sufficient to avoid all
cases where DAGCombiner::isMulAddWithConstProfitable might trigger a
transformation. This patch makes transformAddImmMulImm bail out if C0
(the constant used for multiplication) has more than one use.

Differential Revision: https://reviews.llvm.org/D120332

Expand `TruncInstCombine` to handle loops by adding `phi` nodes
to expression graph.

Reviewed by: RKSimon, lebedev.ri

Differential Revision: https://reviews.llvm.org/D109817

We were handling sub of umax, but not the conjugated umin case.

https://alive2.llvm.org/ce/z/4fdZfy
https://alive2.llvm.org/ce/z/BhUQBM

We should be converting these into usub.sat.

add –allow-unused-prefixes.

This test has two runs that differ in what functions are left after the inliner,
for example: barney exists on OLDPM path but don’t exist on NEWPM path.
I restored prefixes this test had had after automatic checks were introduced
for this test.

For now there are no checks left for ALL_NEWPM path, but the behavior seem to
change over time so I added –allow-unused-prefixes to ease following check updates.

Renamed %tmp => %temp IR values to avoid update warning.

Differential revision: https://reviews.llvm.org/D120207

This reverts commit 7ea103de.

It is useful to be able to visualise the C++20 modules content of a PCM file
both for inspection and for testing.  In particular, when adding more module
types to support C++20 Partitions and Header Units, we would like to be able
to confirm that the output PCM has the intended structure.

The existing scheme for dumping data is restricted to the content of the AST
file control block, which does not include structural data beyond imports.

The change here makes use of the AST unit that is set up by BeginSourceFile
to query for the information on the primary and sub-modules.  We can then
inspect each of these in turn, accounting for Global, Private, Imported and
Exported modules/fragments and then showing the sub-stucture of the main
module(s).

The disadvantage of this mechanism is that it has no easy method to control
the granularity of the output.  Perhaps more detailed inspection would be
better handled by a stand-alone module inspection tool.

Differential Revision: https://reviews.llvm.org/D119823

The pointers are used immediately, so assert the cast is correct instead of returning nullptr

Currently, the driver generates the tables with "run-time type
information for derived types" only when specific actions are run.
However, the corresponding data might be required by the subsequent
compilation stages (e.g. lowering, code-gen) and should be generated
unconditionally. Note that this is only possible once the semantic
checks have been run.

Note that when generating these tables, extra semantic errors might be
generated. The driver will always report these and in most cases such
semantic errors will cause the driver to exit immediately. The only
exception are actions inheriting from `PrescanAndSemaDebugAction`.
Currently, there's only one such action: `DebugDumpAllAction`
(corresponds to `-fdebug-dump-all` command-line flag). I've updated the
comments for this action to clarify this.

This change will mostly affect lowering, which currently is only
available for most basic examples (e.g. empty programs). I wasn't able
to find a working case that would demonstrate the new behaviour. I
hope that this change is straightforward enough and am submitting it
without a test.

Differential Revision: https://reviews.llvm.org/D120051

libtool uses file names to name members of an static library.
Files, located in different directories and having matching name,
would have the same name inside an archive. This is not a problem
for ld, but may be a problem for ar. This patch renames files
from ObjCopy library to avoid names clashing.

See https://reviews.llvm.org/D88827#3335814

Differential Revision: https://reviews.llvm.org/D120345