This method previously always recursively checked both the left-hand
side and right-hand side of binary operations for splatted (broadcast)
vector values to determine if the parent DAG node is a splat.

Like several other SelectionDAG methods, limit the recursion depth to
MaxRecursionDepth (6). This prevents stack overflow.
See also https://issuetracker.google.com/173785481

Patch by Nicolas Capens. Thanks!

Differential Revision: https://reviews.llvm.org/D92421

To support llorg builds this patch provides the following changes:

1)  Added cmake variable ITTAPI_GIT_REPOSITORY to control the location of ITTAPI repository.
     Default value of ITTAPI_GIT_REPOSITORY is github location: https://github.com/intel/ittapi.git
     Also, the separate cmake variable ITTAPI_GIT_TAG was added for repo tag.
2)  Added cmake variable ITTAPI_SOURCE_DIR to control the place where the repo will be cloned.
     Default value of ITTAPI_SOURCE_DIR is build area: PROJECT_BINARY_DIR

Reviewed By: etyurin, bader

Patch by ekovanov.

Differential Revision: https://reviews.llvm.org/D91935

This was accidentally reverted by a later change.

LSR currently only runs in the codegen pass manager.
There are a couple issues with LSR and the NPM.
1) Lots of tests assume that LCSSA isn't run before LSR. This breaks a
bunch of tests' expected output. This is fixable with some time put in.
2) LSR doesn't preserve LCSSA. See
llvm/test/Analysis/MemorySSA/update-remove-deadblocks.ll. LSR's use of
SCEVExpander is the only use of SCEVExpander where the PreserveLCSSA option is
off. Turning it on causes some code sinking out of loops to fail due to
SCEVExpander's inability to handle the newly created trivial PHI nodes in the
broken critical edge (I was looking at
llvm/test/Transforms/LoopStrengthReduce/X86/2011-11-29-postincphi.ll).
I also tried simply just calling formLCSSA() at the end of LSR, but the extra
PHI nodes cause regressions in codegen tests.

We'll delay figuring these issues out until later.

This causes the number of check-llvm failures with -enable-new-pm true
by default to go from 60 to 29.

Reviewed By: asbirlea

Differential Revision: https://reviews.llvm.org/D92796

Otherwise `check-llvm-*` may not rebuild llvm-profgen, causing llvm-profgen tests
to fail if llvm-profgen happens to be stale.

The reproducers currently use a static variable to track the API
boundary. This is obviously incorrect when the SB API is used
concurrently. While I do not plan to support that use-case (right now),
I do want to avoid us crashing. As a first step, correctly track API
boundaries across multiple threads.

Before this patch SB API calls made by the embedded script interpreter
would be considered "behind the API boundary" and correctly ignored.
After this patch, we need to tell the reproducers to ignore the
scripting thread as a "private thread".

Differential revision: https://reviews.llvm.org/D92811

Reviewed By: hans

Differential Revision: https://reviews.llvm.org/D92866

Explicitly opt-out llvm/test/Transforms/Attributor.

Verified by flipping the default value of allow-unused-prefixes and
observing that none of the failures were under llvm/test/Transforms.

Differential Revision: https://reviews.llvm.org/D92404

This is a step towards making compile_commands.json reloadable.

The idea is:
 - in addition to rare CDB loads we're soon going to have somewhat-rare CDB
   reloads and fairly-common stat() of files to validate the CDB
 - so stop doing all our work under a big global lock, instead using it to
   acquire per-directory structures with their own locks
 - each directory can be refreshed from disk every N seconds, like filecache
 - avoid locking these at all in the most common case: directory has no CDB

Differential Revision: https://reviews.llvm.org/D92381

The goal was to add coverage for back-deployment over the filesystem
library, but it was added in macOS 10.15, not 10.14.

Differential Revision: https://reviews.llvm.org/D92937

No changes to the tests themselves, other than some auto -> const auto
diagnostic fixes and formatting.

Differential Revision: https://reviews.llvm.org/D92939

Add vsum and vfsum intrinsic instructions and regression tests.

Reviewed By: simoll

Differential Revision: https://reviews.llvm.org/D92938

Differential Revision: https://reviews.llvm.org/D92674

This is an enhancement to load vectorization that is motivated by
a pattern in https://llvm.org/PR16739.
Unfortunately, it's still not enough to make a difference there.
We will have to handle multi-use cases in some better way to avoid
creating multiple overlapping loads.

Differential Revision: https://reviews.llvm.org/D92858

We could create uadd.sat under incorrect circumstances
if a select with -1 as the false value was canonicalized
by swapping the T/F values. Unlike the other transforms
in the same function, it is not invariant to equality.

Some alive proofs: https://alive2.llvm.org/ce/z/emmKKL

Based on original patch by David Green!

Fixes https://bugs.llvm.org/show_bug.cgi?id=48390

Differential Revision: https://reviews.llvm.org/D92717

The non-strict variants are already handled because they are canonicalized
to strict variants by swapping hands in both the select and icmp,
and the fold simply considers that strictness is irrelevant here.

But that isn't actually true for the last pattern, as PR48390 reports.

Add vfmk intrinsic instructions, a few pseudo instructions to expand
vfmk intrinsic using VM512 correctly, and regression tests.

Reviewed By: simoll

Differential Revision: https://reviews.llvm.org/D92758

Should fix non-x86 bot failures.

[X86] Fold CONCAT(VPERMV3(X,Y,M0),VPERMV3(Z,W,M1)) -> VPERMV3(CONCAT(X,Z),CONCAT(Y,W),CONCAT(M0,M1))

Further prep work toward supporting different subvector sizes in combineX86ShufflesRecursively

The wrapper clears shadow for any events written.

Reviewed By: stephan.yichao.zhao

Differential Revision: https://reviews.llvm.org/D92891

TestLldbGdbServer.py testcases are timing out on LLDB/AArch64 Linux
buildbot since recent changes. I am temporarily increasing
DEFAULT_TIMEOUT to 20 seconds to see impact.

For stores chain vectorization we choose the size of vector
elements to ensure we fit to minimum and maximum vector register
size for the number of elements given. This patch corrects vector
element size choosing the width of value truncated just before
storing instead of the width of value stored.

Fixes PR46983

Differential Revision: https://reviews.llvm.org/D92824

If a function parameter is marked as "undef", prevent creation
of CallSiteInfo for that parameter.
Without this patch, the parameter's call_site_value would be incorrect.
The incorrect call_value case reported in PR39716,
addressed in D85111.
​
Patch by Nikola Tesic
​
Differential revision: https://reviews.llvm.org/D92471

This patch adds the following DAGCombines, which apply if isVectorLoadExtDesirable() returns true:
 - fold (and (masked_gather x)) -> (zext_masked_gather x)
 - fold (sext_inreg (masked_gather x)) -> (sext_masked_gather x)

LowerMGATHER has also been updated to fetch the LoadExtType associated with the
gather and also use this value to determine the correct masked gather opcode to use.

Reviewed By: sdesmalen

Differential Revision: https://reviews.llvm.org/D92230

* Steps are scaled by `vscale`, a runtime value.
* Changes to circumvent the cost-model for now (temporary)
  so that the cost-model can be implemented separately.

This can vectorize the following loop [1]:

   void loop(int N, double *a, double *b) {
     #pragma clang loop vectorize_width(4, scalable)
     for (int i = 0; i < N; i++) {
       a[i] = b[i] + 1.0;
     }
   }

[1] This source-level example is based on the pragma proposed
separately in D89031. This patch only implements the LLVM part.

Reviewed By: dmgreen

Differential Revision: https://reviews.llvm.org/D91077

This patch removes a number of asserts that VF is not scalable, even though
the code where this assert lives does nothing that prevents VF being scalable.

Reviewed By: dmgreen

Differential Revision: https://reviews.llvm.org/D91060

Adds the ExtensionType flag, which reflects the LoadExtType of a MaskedGatherSDNode.
Also updated SelectionDAGDumper::print_details so that details of the gather
load (is signed, is scaled & extension type) are printed.

Reviewed By: sdesmalen

Differential Revision: https://reviews.llvm.org/D91084

[mlir] Use mlir::OpState::operator->() to get to methods of mlir::Operation. This is a preparation step to remove the corresponding methods from OpState.

Reviewed By: silvas, rriddle

Differential Revision: https://reviews.llvm.org/D92878

This commit adds two new intrinsics.

- llvm.experimental.vector.insert: used to insert a vector into another
  vector starting at a given index.

- llvm.experimental.vector.extract: used to extract a subvector from a
  larger vector starting from a given index.

The codegen work for these intrinsics has already been completed; this
commit is simply exposing the existing ISD nodes to LLVM IR.

Reviewed By: cameron.mcinally

Differential Revision: https://reviews.llvm.org/D91362

Reviewed By: sdesmalen

Differential Revision: https://reviews.llvm.org/D92482

Translate VP intrinsics to VP_* SDNodes.  The tests check whether a
matching vp_* SDNode is emitted.

Reviewed By: craig.topper

Differential Revision: https://reviews.llvm.org/D91441

The original code was inserting the barrier at the location given by the
caller. Make sure it is always inserted at the end of the loop exit block
instead.

Reviewed By: Meinersbur

Differential Revision: https://reviews.llvm.org/D92849

I accidentally pushed the wrong patch originally.

Fix failure introduced by 843f2dbf.

[NFC][Instructions] Refactor CmpInst::getFlippedStrictnessPredicate() in terms of is{,Non}StrictPredicate()/get{Non,}StrictPredicate()

In particular, this creates getStrictPredicate() method,
to be symmetrical with already-existing getNonStrictPredicate().

The register operand was not being marked as a def when it should be. No tests
for this in the main branch as there are not yet any pseudos without a
non-negative VLIndex.

Also change the type of a virtual register operand from unsigned to Register
and adjust formatting.

Reviewed By: craig.topper

Differential Revision: https://reviews.llvm.org/D92823