Dynamic tls access model will be lowered to MI which clobbers CTR in
the loop in ISEL(ADDItlsgdLADDR) and post-isel CTR loop pass will revert
the loop to a normal compare + branch form.

So no need to add this clobber check in hardware loop insertion pass now.

Reviewed By: nemanjai

Differential revision: https://reviews.llvm.org/D140367

Fuchsia already implements AddClangSystemIncludeArgs(). So it looks like we
just have to switch over to using it.

Reviewed By: phosek

Differential Revision: https://reviews.llvm.org/D141073

Passes before hardware loop insertion change the loop to a form which
is not a hardware loop candidate (return early before checking the ctr clobbers).
And the PHI in the loop exit block is also optimized away. This breaks the
previous test point when the case was committed. Fixing this by running this
case just before hardware loop insertion pass.

Reviewed By: nemanjai

Differential revision: https://reviews.llvm.org/D140366

There are a few intrinsics or instructions on LoongArch
that are only appropriate for loongarch32 target. So the
feature "32bit" is added to implement it.

Reviewed By: SixWeining

Differential Revision: https://reviews.llvm.org/D140872

Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP macros

Reviewed By: SixWeining, MaskRay

Differential Revision: https://reviews.llvm.org/D141070

We need this because WGSL does not support extended multiplication ops.

Fixes: https://github.com/llvm/llvm-project/issues/59563

Reviewed By: antiagainst

Differential Revision: https://reviews.llvm.org/D141096

This reverts commit 5e8f597c. It caused msan and ubsan breakages.

The instruction name is x.f with the destination type first. The
template name was intended as "convert F to X". So the F comes first.

It is only used in the LSAN specific part.

Reviewed By: vitalybuka

Differential Revision: https://reviews.llvm.org/D140109

Differential Revision: https://reviews.llvm.org/D141097

Add codegen pattern for sparse_tensor.indices_buffer.

Reviewed By: aartbik

Differential Revision: https://reviews.llvm.org/D140871

Fix an off-by-one error in extended umul extension for WebGPU.
Revert to the long multiplication algorithm originally added to wide
integer emulation, which was deleted in D139776. It is much easier
to see why it is correct.

Add runtime tests based on the mlir-vulkan-runner. These run both with
and without umul extension.

Issue: https://github.com/llvm/llvm-project/issues/59563

Reviewed By: antiagainst

Differential Revision: https://reviews.llvm.org/D141085

Closes #38042.

Differential Revision: https://reviews.llvm.org/D141035

Differential Revision: https://reviews.llvm.org/D140188

This teaches LVI (and thus CVP) to extract range information
from branches whose condition is negated using (`xor %c, true`).
On the implementation side, we switch the cache to additionally
track whether we're looking for the inverted value or not and
otherwise using the existing support for computing inverted
conditions.

I think the biggest question here is why this negation shows up
here at all. After all, it should always be possible for some
other pass to fold such a negation into a branch, comparison or
some other logical operation. Indeed, instcombine does just that.
However, these negations can be otherwise fairly persistent, e.g.
instsimplify is not able to exchange branch conditions from
negations. In addition, jumpthreading, which sits at the same
point in default pass pipeline also handles this pattern, which
adds further evidence that we might expect these negations to
not have been canonicalized away yet at this point in the pass
pipeline.

In the particular case I was looking at there was a bit of a
circular dependency where flags computed by cvp were needed
by instcombine, and incstombine's folding of the negation was
needed for cvp. Adding a second instombine pass would have
worked of course, but instcombine can be somewhat expensive,
so it appeared desirable to not require it to have run
before cvp (as is the case in the default pass pipeline).

Reviewed By: nikic

Differential Revision: https://reviews.llvm.org/D140933

In C mode, if e1 has __attribute__((noreturn)) but e2 doesn't, `(c ? e1 : e2)`
is incorrectly noreturn and Clang codegen produces `unreachable`
which may lead to miscompiles (see [1] `gawk/support/dfa.c`).
This problem has been known since
8c6b56f3 (2010) or earlier.

Fix this by making the result type noreturn only if both e1 and e2 are
noreturn, matching GCC.

`_Noreturn` and `[[noreturn]]` do not have the aforementioned problem.

Fix https://github.com/llvm/llvm-project/issues/59792 [1]

Reviewed By: aaron.ballman

Differential Revision: https://reviews.llvm.org/D140868

This rewrite fixes https://github.com/llvm/llvm-project/issues/59316.

Previously LowerSwitch uses int64_t, which will crash on case branches using integers with more than 64 bits.
Using APInt fixes this problem. This patch also includes a test

Reviewed By: RKSimon

Differential Revision: https://reviews.llvm.org/D140747

This was added to the static matchers in
125ccd37, but the dynamic matcher was
missed. This adds the dynamic matcher to the list.

[Support] Do not run test on z/OS

A part of the unit test CommandLineTest/BadResponseFile, added in
the commit fd3d7a9f need to be disable for z/OS as it was already done for AIX platform.

Reviewed By: fanbo-meng

Differential Revision: https://reviews.llvm.org/D141084

Reviewed By: nikic, hctim

Differential Revision: https://reviews.llvm.org/D141083

Previously, the model for structs modeled all fields in a struct when
`createValue` was called for that type. This patch adds a prepass on the
function under analysis to discover the fields referenced in the scope and then
limits modeling to only those fields.  This reduces wasted memory usage
(modeling unused fields) which can be important for programss that use large
structs.

Note: This patch obviates the need for https://reviews.llvm.org/D123032.

Differential Revision: https://reviews.llvm.org/D140694

lldb::LanguageType is used as a parameter in
SBExpressionOptions::SetLanguage(), which actually makes this type API
too. Commit 6eaedbb5 added a
`: uint16_t` to it, which broke binary compatibility for the SBAPI. This
patch reverts to the original enum.

I tried moving the entire enum into include/API, but that created a
cyclic module dependency between API and Utility. To keep things
simple, this just reverts to the original code and adds a warning.

rdar://103415402

Differential Revision: https://reviews.llvm.org/D141087

This was only used for checking if is_shared/is_private were legal,
which we're not bothering to do anymore.

This is apparently visible to more than the target attribute (which
seems to silently ignore unrecognized features), so this has the
potential to break something (i.e. see the OpenMP test change)

Change clang::annotate into clang::annotate_type.

The example will get error like
error: 'annotate' attribute cannot be applied to types
int* [[clang::annotate("category1", "foo", 1)]] f(int[[clang::annotate("category2")]] *);

Reviewed By: aaron.ballman

Differential Revision: https://reviews.llvm.org/D139935

Amdgpu kernel with function attribute "uniform-work-group-size"="true" requires
uniform work group size (i.e. each dimension of global size is a multiple of
corresponding dimension of work group size). hipExtModuleLaunchKernel allows to
launch HIP kernel with non-uniform workgroup size, which makes it necessary for
runtime to check and enforce uniform workgroup size if kernel requires it. To
let runtime be able to enforce that, this metadata is needed to indicate that
the kernel requires uniform workgroup size.

Reviewed By: kzhuravl, arsenm

Differential Revision: https://reviews.llvm.org/D141012

MC and lld/ELF defaults were flipped in 2016. For Clang: CMake
ENABLE_X86_RELAX_RELOCATIONS defaults to on in 2020. It makes sense for
the TargetOptions default to be true now.

R_X86_64_GOTPCRELX/R_X86_64_REX_GOTPCRELX require GNU ld newer than 2015-10
(subsumed by the current requirement of -fbinutils-version=).

This should fix `rustc -Z plt=no` PIC relocatable files with GNU ld.
(See https://github.com/rust-lang/rust/pull/106380)

Account for an r_brace that precedes an "else if" statement when
calculating whether the line might fit on one line if the r_brace
is removed.

Fixes #59778.

Differential Revision: https://reviews.llvm.org/D140835

clang would improperly disallow GNU attributes before C++ standard
attributes when a declaration had a linkage specifier.  Handle this
similarly to the previous case of invalid parsing.  We now better match
the parsing rules from GCC.

Differential Revision: https://reviews.llvm.org/D140507
Reviewed By: aaron.ballman

Since the divergence-driven ISel was fully enabled we have more VGPRs available.
         MachineScheduler trying to take advantage of that bumps up the occupancy sacrificing
         the hiding of memory access latency.  This really spoils the initially good schedule.
         A new metric that reflects the latency hiding quality of the schedule has been created
         to make it to balance between occupancy and latency. The metric is based on the latency
         model which computes the bubble to working cycles ratio. Then we use this ratio to decide
         if the higher occupancy schedule is profitable as follows:

             Profit = NewOccupancy/OldOccupancy * OldMetric/NewMetric

Reviewed By: rampitec

Differential Revision: https://reviews.llvm.org/D139710

[Fix]"[-Wunsafe-buffer-usage] Add a new `forEachDescendant` matcher that skips callable declarations"

The original patch in commit b2ac5fd7
causes compilation errors which can be reproduced by the
`-fdelayed-template-parsing` flag.  This commit fixes the problem.

Related differential revision: https://reviews.llvm.org/D138329

Revert "Revert "[-Wunsafe-buffer-usage] Add a new `forEachDescendant` matcher that skips callable declarations""

This reverts commit f58b0253.

 The previous revert reverts a patch that causes compilation problem on
 windows which can be reproduced using `-fdelayed-template-parsing`.
 I'm now to revert the patch back and commit a fix next.

appendToGlobalCtors implicitly assumes this is the case, since it
deletes and recreates without trying to update any uses.

This ran into an interesting problem in a few linker tests. During the
link, ConstantExpr casts were speculatively created to replace any
uses that might need them for replacement. These unused ConstantExprs
would hang around and still appear on the use list. It seems like a
bug that those stick around, but I'm not sure where those are supposed
to be cleaned up. Avoid this by not creating the casts for appending
linkage.

Delete one of the casts entirely as it breaks no tests. The verifier
has enforced a specific type for these since 2011, so I don't see why
we would need to handle linking modules with a wrong types. One test
does fail without the second cast (Linker/appending-global-proto.ll,
added by D95126). This test looks contrived; it's using appending
linkage with a regular variable. The LangRef suggests this is illegal
(and suggests another missing verifier check).

We removed the Go bindings in https://reviews.llvm.org/D135436 but
missed documentation that talks about the bindings.

Reviewed By: reames

Differential Revision: https://reviews.llvm.org/D140928

We can technically handle them, but since they shouldn't come up in any
real-world programs (since ld64 dedups strings unconditionally), there's
no reason to support them.

It's a thoroughly untested code path too -- as evidenced by the fact
that the only test this change breaks is one that verifies that we
reject relocations when dedup'ing. There is no test that covers the case
where we handle relocations in cstring sections when dedup is disabled.

Reviewed By: #lld-macho, oontvoo, keith, thakis

Differential Revision: https://reviews.llvm.org/D141025

The tests contain a redundant condition in the else if branch that can
be simplified with fb13dcf3.

Update the condition used to prevent it from getting removed.

Alive2 proof for check removal:
https://alive2.llvm.org/ce/z/iFBnsy

I found one typo of "implemnt", then some more.
s/implemnt/implement/g

This was directly considering the pointee type, and also applying
special semantics to constant address space.