Currently, variadic dbg.values (i.e. those using a DIArgList as part of
their location) are not handled properly by FastISel or GlobalISel, and
will produce invalid DBG_VALUE instructions if they encounter them. This
patch fixes this issue by emitting undef DBG_VALUE instructions for
variadic dbg.values, so that no incorrect instruction is produced and
any prior variable location is terminated.

This is simply a quick-fix to prevent errors; a correct implementation
should come later for these ISel pipelines to ensure that we do not drop
debug information unnecessarily.

Differential Revision: https://reviews.llvm.org/D102500

- Replace class(*) member by a c_ptr member to avoid having to handle
  polymorphic components in the type info table generation. Polymorphic
  entity handling will require these very tables to be lowered properly.
  Note: keep the init as NullPointer/Designators. This is technically
  invalid Fortran, the init should have c_ptr type. But wrapping this
  in a C_LOC intrinsic call would make runtime generation and lowering
  more complex with no real benefits.

- ComponentIterator is crashing when used on the generated derived
  types in GetScope. This patch makes GetScope more robust, but it
  is not entirely clear to me why this is only happening with the
  generated derived types.

- The type of generated character globals was incorrect because
  Scope::FindType was matching character types with different
  length. Add a CharacterTypeSpec == operator to fix this.

Differential Revision: https://reviews.llvm.org/D102768

Follow up to D101357 / 3fa6510f.
Supersedes D102330.

Goal: Use flags setting rdffrs instead of rdffr + ptest.

Problem: RDFFR_P doesn't have have a flags setting equivalent.

Solution: in instcombine, canonicalize to RDFFR_PP at the IR level, and
rely on RDFFR_PP+PTEST => RDFFRS_PP optimization in
AArch64InstrInfo::optimizePTestInstr.

While here:

* Test that rdffr.z+ptest generates a rdffrs.
* Use update_{test,llc}_checks.py on the tests.
* Use sve attribute on functions.

Differential Revision: https://reviews.llvm.org/D102623

GlobalOpt can slice structs/arrays and change GEPs in the process,
but it was not updating alignments for load/store users. This
eventually causes the crashing seen in:
https://llvm.org/PR49661
https://llvm.org/PR50253

On x86, this required SLP+codegen to create an aligned vector
store on an invalid address. The bugs would be easier to
demonstrate on a target with stricter alignment requirements.

I'm not sure if this is a complete solution. The alignment
updating code is adapted from InstCombine, so I assume that
part is tested and good.

Differential Revision: https://reviews.llvm.org/D102552

Goes with D102552

Not sure if that's the ostringstream or our conversion code, but this is
returning the wrong results on Windows.

If we gather extract elements and they actually are just shuffles, it
might be profitable to vectorize them even if the tree is tiny.

Differential Revision: https://reviews.llvm.org/D101460

Drop the Linalg dialect EDSC subdirectory and update all uses.

Differential Revision: https://reviews.llvm.org/D102848

The diagnostics were not following the usual style rules.

[libomptarget][amdgpu] Remove majority of fatal errors

Replaces most calls to exit() with returning an error to the library entry
point. Minor changes to error handling for clear bugs, remove some dead code.

Each exit() call site replaced is either in a library entry point or a
function that already returns error codes on some paths. The existing handling
is not well tested but replacing exit() with a fallback path should be a strict
improvement.

Remaining two early exit points are an abort() from a callback and exit() from
within msgpack. Fixes for those are less obvious and left for a later patch.

Reviewed By: pdhaliwal

Differential Revision: https://reviews.llvm.org/D102346

Linker scripts might not handle COMDAT sections. SLSHardeing adds
new section for each __llvm_slsblr_thunk_xN. This new option allows
the generation of the thunks into the normal text section to handle these
exceptional cases.
,comdat or ,noncomdat can be added to harden-sls to control the codegen.
-mharden-sls=[all|retbr|blr],nocomdat.

Reviewed By: kristof.beyls

Differential Revision: https://reviews.llvm.org/D100546

This is not an issue anymore since we don't build the libc++ dylib with
C++14 anymore (see https://llvm.org/PR28391) for details.

Differential Revision: https://reviews.llvm.org/D102106

[libomptarget] Disable test bug49334 on amdgpu

Hangs on amdgpu, do not know why. Disable to unblock build.

Reviewed By: ye-luo

Differential Revision: https://reviews.llvm.org/D102017

Eventually, this should become the default way of running the tests.
For now, only move a few CI nodes to it, and keep a node that runs the
legacy configuration.

Differential Revision: https://reviews.llvm.org/D97565

The old invitation has expired, so I've created a new one and update the
link in the docs accordingly.

We were always storing a regular ValueDecl* as decomposition declaration
and haven't been using the opportunity to initialize it lazily.

Reviewed By: aaron.ballman

Differential Revision: https://reviews.llvm.org/D99455

This reverts commit 76f375f3.

This will be pushed again, after investigating a test failure:
https://lab.llvm.org/buildbot/#/builders/16/builds/11254

The test previous accidentally passed because it was looking for a lack
of specific input from the binary(!) output being sent to stdout.

This reverts commit f7c53515.

To investigate a test failure.

This is an improvement of [0]. This adds checking of
original llvm.dbg.values()/declares() instructions in
optimizations.

We have picked a real issue that has been found with
this (actually, picked one variable location missing
from [1] and resolved the issue), and the result is
the fix for that -- D100844.

Before applying the D100844, using the options from [0]
(but with this patch applied) on the compilation of GDB 7.11,
the final HTML report for the debug-info issues can be found
at [1] (please scroll down, and look for
"Summary of Variable Location Bugs"). After applying
the D100844, the numbers has improved a bit -- please take
a look into [2].

[0] https://llvm.org/docs/HowToUpdateDebugInfo.html\
[1] https://djolertrk.github.io/di-check-before-adce-fix/
[2] https://djolertrk.github.io/di-check-after-adce-fix/

Differential Revision: https://reviews.llvm.org/D100845

Differential Revision: https://reviews.llvm.org/D102238

On AIX, char bitfields have the same alignment as unsigned int.
Reference: https://reviews.llvm.org/D87029

Differential Revision: https://reviews.llvm.org/D102715

[clang] Invalidate a non-dependent-type RecordDecl when it has any dependent-type base class specifier.

This happens during the error-recovery, and it would esacpe all
dependent-type check guards in getTypeInfo/constexpr-evaluator code
paths, which lead to crashes.

Differential Revision: https://reviews.llvm.org/D102773

Strictly speaking, the architecture manual no longer uses the st
mnemonic, but that's a much more intrusive change for little gain.

Differential Revision: https://reviews.llvm.org/D96313

This adds the straightforward conversion for EqualOp
(two complex numbers are equal if both the real and the imaginary part are equal).

Differential Revision: https://reviews.llvm.org/D102840

When trying to return a type such as <vscale x 1 x i32> from a
function we crash in DAGTypeLegalizer::WidenVecRes_EXTRACT_SUBVECTOR
when attempting to get the fixed number of elements in the vector.

For the simple case we are dealing with, i.e. extracting
<vscale x 1 x i32> from index 0 of input vector <vscale x 4 x i32>
we can simply rely upon existing code that just returns the input.

Differential Revision: https://reviews.llvm.org/D102605

Haswell, Excavator and early Ryzen all have slower 256-bit non-uniform vector shifts (confirmed on AMDSoG/Agner/instlatx64 and llvm models) - so bump the worst case costs accordingly.

Noticed while investigating PR50364

This adds custom lowering for the MLOAD and MSTORE ISD nodes when
passed fixed length vectors in SVE. This is done by converting the
vectors to VLA vectors and using the VLA code generation.

Fixed length extending loads and truncating stores currently produce
correct code, but do not use the built in extend/truncate in the
load and store instructions. This will be fixed in a future patch.

Differential Revision: https://reviews.llvm.org/D101834

This reverts commit 76b8754d and ensures the PCM files are created in the correct directory (not in the current working directory).

This will allow to use llvm-strip with file names that begin with dashes.

Reviewed By: jhenderson

Differential Revision: https://reviews.llvm.org/D102825

This adds some extra codegen tests for abs and hadd, regenerating some
of the existing tests with updated check lines.

This patch prepares llvm-objcopy to move its implementation
into a separate library. To make it possible it is necessary
to minimize internal dependencies.

Differential Revision: https://reviews.llvm.org/D99055