The static analyzer is warning about potential null dereferences, but in these cases we should be able to use cast<SCEVConstant> directly and if not assert will fire for us.
llvm-svn: 373465
Summary:
This extends the PeelingModuloScheduleExpander to generate prolog and epilog code,
and correctly stitch uses through the prolog, kernel, epilog DAG.
The key concept in this patch is to ensure that all transforms are *local*; only a
function of a block and its immediate predecessor and successor. By defining the problem in this way
we can inductively rewrite the entire DAG using only local knowledge that is easy to
reason about.
For example, we assume that all prologs and epilogs are near-perfect clones of the
steady-state kernel. This means that if a block has an instruction that is predicated out,
we can redirect all users of that instruction to that equivalent instruction in our
immediate predecessor. As all blocks are clones, every instruction must have an equivalent in
every other block.
Similarly we can make the assumption by construction that if a value defined in a block is used
outside that block, the only possible user is its immediate successors. We maintain this
even for values that are used outside the loop by creating a limited form of LCSSA.
This code isn't small, but it isn't complex.
Enabled a bunch of testing from Hexagon. There are a couple of tests not enabled yet;
I'm about 80% sure there isn't buggy codegen but the tests are checking for patterns
that we don't produce. Those still need a bit more investigation. In the meantime we
(Google) are happy with the code produced by this on our downstream SMS implementation,
and believe it generates correct code.
Subscribers: mgorny, hiraditya, jsji, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D68205
llvm-svn: 373462
Fixes PR43181. This option was recently added to GNU objcopy (binutils
PR24942).
`llvm-objcopy -I binary -O elf64-x86-64 --set-section-alignment .data=8` can set the alignment of .data.
Reviewed By: grimar, jhenderson, rupprecht
Differential Revision: https://reviews.llvm.org/D67656
llvm-svn: 373461
We now only use this function directly after initialization. As Clear()
resets the ASTContext back to its initial state, this is just a no-op.
There are no other users for this and we no longer can set the ASTContext
after construction, so Clear has no useful purpose anymore. It's also
mostly copy-pasted from Finalize().
llvm-svn: 373460
In similar fashion to D67721, we can simplify FMA multiplications if any
of the operands is NaN or undef. In instcombine, we will simplify the
FMA to an fadd with a NaN operand, which in turn gets folded to NaN.
Note that this just changes SimplifyFMAFMul, so we still not catch the
case where only the Add part of the FMA is Nan/Undef.
Reviewers: cameron.mcinally, mcberg2017, spatel, arsenm
Reviewed By: cameron.mcinally
Differential Revision: https://reviews.llvm.org/D68265
llvm-svn: 373459
Reason for this patch is the Ssame reason as for the previous patches:
Having a ClangASTContext and being able to switch the associated ASTContext isn't
a use case we have (or should have), so let's simplify all this code.
This way it becomes clearer in what order we initialize data structures.
The DWARFASTParserClangTests changes are necessary as the test is using
a ClangASTContext but relied on the fact that no called function ever calls
getASTContext() on our ClangASTContext (as that would create the ASTContext).
As we now always create the ASTContext the fact that we had an uninitialized
FileSystem made the test crash.
llvm-svn: 373457
This is intended to be similar to the constant folding results from
D67446
and earlier, but not all operands are constant in these tests, so the
responsibility for folding is left to InstSimplify.
Differential Revision: https://reviews.llvm.org/D67721
llvm-svn: 373455
The static analyzer is warning about a potential null dereference, but we should be able to use cast<Function> directly and if not assert will fire for us.
llvm-svn: 373449
The static analyzer is warning about a potential null dereference, but we know that the source won't be null so just use dyn_cast, which will assert if the value somehow is actually null.
llvm-svn: 373448
The static analyzer is warning about potential null dereferences, but we should be able to use castAs<> directly and if not assert will fire for us.
llvm-svn: 373447
Identity shuffles, of the form (0, 1, 2, 3, ...) are perfectly OK under MVE
(they essentially just become bitcasts). We were not catching that in the
existing set of what we considered legal though. On NEON, they would be covered
by vext's, but that is not generally available in MVE.
This uses ShuffleVectorInst::isIdentityMask which is a little odd to use here
but does what we want and prevents us from just rewriting what is the same
function.
Differential Revision: https://reviews.llvm.org/D68241
llvm-svn: 373446
Summary:
This would reduce the false positive when the static index is in an
unavailable state, e.g. background index is not finished.
Reviewers: sammccall
Subscribers: ilya-biryukov, MaskRay, jkorous, arphaman, kadircet, usaxena95, cfe-commits
Tags: #clang
Differential Revision: https://reviews.llvm.org/D68325
llvm-svn: 373444
Summary:
Previously they would be marked as partially-selected based on the early claim,
and never updated as no more tokens were claimed.
This affects simple VarDecls like "int x".
Reviewers: SureYeaah
Subscribers: ilya-biryukov, MaskRay, jkorous, arphaman, kadircet, cfe-commits
Tags: #clang
Differential Revision: https://reviews.llvm.org/D66872
llvm-svn: 373442
Summary:
when renaming a typedef decl, we used to rename the underlying decl of the
typedef, we should rename the typedef itself.
Subscribers: cfe-commits
Tags: #clang
Differential Revision: https://reviews.llvm.org/D68322
llvm-svn: 373440
Summary:
The historical context:
- clang-format was written when C++11 was current,
and the main language-version concern was >> vs > > template-closers.
An option was added to allow selection of the 03/11 behavior, or auto-detection.
- there was no option to choose simply "latest standard" so anyone who didn't
ever want 03 behavior or auto-detection specified Cpp11.
- In r185149 this option started to affect lexer mode.
- no options were added to cover c++14, as parsing/formatting
didn't change that much. The usage of Cpp11 to mean "latest" became
codified e.g. in r206263
- c++17 added some new constructs. These were mostly backwards-compatible and so
not used in old programs, so having no way to turn them off was OK.
- c++20 added some new constructs and keywords (e.g. co_*) that changed the
meaning of existing programs, and people started to complain that
the c++20 parsing couldn't be turned off.
New plan:
- Default ('Auto') behavior remains unchanged: parse as latest, format
template-closers based on input.
- Add new 'Latest' option that more clearly expresses the intent "use
modern features" that many projects have chosen for their .clang-format files.
- Allow pinning to *any* language version, using the same name as clang -std:
c++03, c++11, c++14 etc. These set precise lexer options, and any
clang-format code depending on these can use a >= check.
- For backwards compatibility, `Cpp11` is an alias for `Latest`, not `c++11`.
This matches the historical documented semantics of this option.
This spelling (and `Cpp03`) are deprecated.
Reviewers: klimek, modocache
Subscribers: cfe-commits
Tags: #clang
Differential Revision: https://reviews.llvm.org/D67541
llvm-svn: 373439
Summary:
This allows intrinsics such as the following to be defined:
- declare <n x 4 x i32> @llvm.something.nxv4f32(<n x 4 x i32>, <n x 4 x i1>, <n x 4 x float>)
...where <n x 4 x i32> is derived from <n x 4 x float>, but
the element needs bitcasting to int.
Reviewers: c-rhodes, sdesmalen, rovka
Reviewed By: c-rhodes
Subscribers: tschuett, hiraditya, jdoerfert, llvm-commits, cfe-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D68021
llvm-svn: 373437
Summary:
Editors only know about file URIs, make sure we do not use any custom
schemes while sending edits.
Reviewers: sammccall
Subscribers: ilya-biryukov, MaskRay, jkorous, arphaman, usaxena95, cfe-commits
Tags: #clang
Differential Revision: https://reviews.llvm.org/D68324
llvm-svn: 373435
Summary:
Printf lowering unconditionally visited every instruction in the module.
To make it faster in the common case where there are no printfs, look up
the printf function (if any) and iterate over its users instead.
Reviewers: rampitec, kzhuravl, alex-t, arsenm
Subscribers: jvesely, wdng, nhaehnle, yaxunl, dstuttard, tpr, t-tye, hiraditya, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D68145
llvm-svn: 373433
This is modeled after the same functionality for jump tables, which was
added in r357067.
Differential revision: https://reviews.llvm.org/D68131
llvm-svn: 373431
removeUnreachableBlocks knows how to preserve the DomTree, so make use
of it instead of re-computing the DT.
Reviewers: davide, kuhar, brzycki
Reviewed By: davide, kuhar
Differential Revision: https://reviews.llvm.org/D68298
llvm-svn: 373430
Two small changes in llvm::removeUnreachableBlocks() to avoid unnecessary (re-)computation.
First, replace the use of count() with find(), which has better time complexity.
Second, because we have already computed the set of dead blocks, replace the second loop over all basic blocks to a loop only over the already computed dead blocks. This simplifies the loop and avoids recomputation.
Patch by Rodrigo Caetano Rocha <rcor.cs@gmail.com>
Reviewers: efriedma, spatel, fhahn, xbolva00
Reviewed By: fhahn, xbolva00
Differential Revision: https://reviews.llvm.org/D68191
llvm-svn: 373429
The patch committed was not the accepted version but the
previous one. This commit fixes this issue.
Differential Revision: https://reviews.llvm.org/D64736
llvm-svn: 373428
The tool reports verbose output for the DWARF debug location coverage.
The llvm-locstats for each variable or formal parameter DIE computes what
percentage from the code section bytes, where it is in scope, it has
location description. The line 0 shows the number (and the percentage) of
DIEs with no location information, but the line 100 shows the number (and
the percentage) of DIEs where there is location information in all code
section bytes (where the variable or parameter is in the scope). The line
50..59 shows the number (and the percentage) of DIEs where the location
information is in between 50 and 59 percentage of its scope covered.
Differential Revision: https://reviews.llvm.org/D66526
The cause of the test failure was resolved.
llvm-svn: 373427
If archive files are passed as input files, llvm-lib needs to append
the members of the input archive files to the output file. This patch
implements that behavior.
This patch splits an existing function into smaller functions.
Effectively, the new code is only `if (Magic == file_magic::archive)
{ ... }` part.
Fixes https://bugs.llvm.org/show_bug.cgi?id=32674
Differential Revision: https://reviews.llvm.org/D68204
llvm-svn: 373424
These patterns use zmm registers for 128/256-bit compares when
the VLX instructions aren't available. Previously we only
supported registers, but as PR36191 notes we can fold broadcast
loads, but not regular loads.
llvm-svn: 373423
r368237 attempted to improve fix-its for move warnings, but introduced some
regressions to -Wpessimizing-move. Revert that change and add the missing
test cases to the pessimizing move test to prevent future regressions.
llvm-svn: 373421
We currently just look for files named in the modulemap in its
associated source directory. This means that we can't name
generated files, like TypeNodes.def now is, which means we can't
explicitly mark it as textual. But fortunately that's okay
because (as I understand it) the most important purpose of naming
the header in the modulemap is to ensure that it's not treated as
public, and the search for public headers also only considers
files in the associated source directory. This isn't an elegant
solution, since among other things it means that a build which
wrote the generated files directly into the source directory would
result in something that wouldn't build as a module, but that's
a problem for all our other generated files as well.
llvm-svn: 373416
In principle this should behave as any other constant. However
eliminateFrameIndex currently assumes a VALU use and uses a vector
shift. Work around this by selecting to VGPR for now until
eliminateFrameIndex is fixed.
llvm-svn: 373415
Simon Pilgrim