As the test case shows if there is an ExtractValueInst in the Ret block, function dupRetToEnableTailCallOpts can't duplicate it into the block containing call. So later no tail call is generated in CodeGen.
This patch adds the ExtractValueInst handling code in function dupRetToEnableTailCallOpts and FoldReturnIntoUncondBranch, and later tail call can be generated for this case.
Differential Revision: https://reviews.llvm.org/D74242
They need to appear before any member functions are ODR used, because
they change the visibility of many of these functions and otherwise
they could end up with hidden visibility in the DSO.
This teaches Loop Strength Reduction the details about masked load and
store address operands, so that it can have a better time optimising
them as it would for normal loads and stores.
Differential Revision: https://reviews.llvm.org/D75371
Summary:
[libomptarget] Implement hip atomic functions in terms of intrinsics
All but atomicInc can be implemented using type generic clang intrinsics.
There is not yet a corresponding intrinsic for atomicInc in clang, only one in
LLVM. This patch leaves atomicInc as an unresolved symbol.
Reviewers: jdoerfert, ABataev, hfinkel, grokos, arsenm
Reviewed By: arsenm
Subscribers: sri, saiislam, wdng, jvesely, mgorny, jfb, openmp-commits
Tags: #openmp
Differential Revision: https://reviews.llvm.org/D73076
Summary:
The order of the operations has fallen out of sync as operations have been renamed and new ones have been added.
Differential Revision: https://reviews.llvm.org/D75540
This makes sure that the constant expression bitcast goes through
target-dependent constant folding, and thus avoids an additional
iteration of InstCombine.
Spin-off from D75407. As described there, ConstantFoldConstant()
currently returns null for non-ConstantExpr/ConstantVector inputs,
but otherwise always returns non-null, independently of whether
any folding has happened or not.
This is confusing and makes consumer code more complicated.
I would expect either that ConstantFoldConstant() returns only if
it actually folded something, or that it always returns non-null.
I'm going to the latter possibility here, which appears to be more
useful considering existing usage.
Differential Revision: https://reviews.llvm.org/D75543
The output of subprocess.check_output is decode()'d through the rest of
this python program, but one appears to have been missed for the output
of the call to "clang -print-file-name=include".
On Windows, with Python 3.6, this leads to the 'args' array being a mix of
bytes and strings, which causes exceptions later down the line.
I can't easily test with python2 on Windows, but python2 on Ubuntu 18.04
was happy with this change.
If we would emit a VBROADCAST node, we can instead directly emit
a VBROADCAST_LOAD. This allows us to get rid of the special case
to use an f64 load on 32-bit targets for vXi64.
I believe there is more cleanup we can do later in this function,
but I'll do that in follow ups.
The addition of MatrixBuilder.h broke the modules build:
```
While building module 'LLVM_intrinsic_gen' imported from llvm-project/llvm/lib/IR/AbstractCallSite.cpp:19:
While building module 'LLVM_IR' imported from llvm-project/llvm/include/llvm/IR/Argument.h:19:
In file included from <module-includes>:6:
llvm-project/llvm/include/llvm/IR/MatrixBuilder.h:19:10: fatal error: cyclic dependency in module 'LLVM_intrinsic_gen': LLVM_intrinsic_gen -> LLVM_IR -> LLVM_intrinsic_gen
^
While building module 'LLVM_intrinsic_gen' imported from llvm-project/llvm/lib/IR/AbstractCallSite.cpp:19:
In file included from <module-includes>:1:
llvm-project/llvm/include/llvm/IR/Argument.h:19:10: fatal error: could not build module 'LLVM_IR'
~~~~~~~~^~~~~~~~~~~~~~~~~
llvm-project/llvm/lib/IR/AbstractCallSite.cpp:19:10: fatal error: could not build module 'LLVM_intrinsic_gen'
~~~~~~~~^~~~~~~~~~~~~~~~~~~~
```
The goal of the test was only to check that we could access the
`this->current` member of std::reverse_iterator from a derived
class, but in doing so we incremented a null iterator, which is UB.
The initial placement of vector-combine in the opt pipeline revealed phase ordering bugs:
https://bugs.llvm.org/show_bug.cgi?id=45015https://bugs.llvm.org/show_bug.cgi?id=42022
This patch contains a few independent changes:
1. Move the pass up in the pipeline, so it happens just after loop-vectorization.
This is only to keep vectorization passes together in the pipeline at the moment.
I don't have evidence of interaction between these yet.
2. Add an -early-cse pass after -vector-combine to clean up redundant ops. This was
partly proposed as far back as rL219644 (which is why it's effectively being moved
in the old PM code). This is important because the subsequent -instcombine doesn't
work as well without EarlyCSE. With the CSE, -instcombine is able to squash
shuffles together in 1 of the tests (because those are simple "select" shuffles).
3. Remove the -vector-combine pass that was running after SLP. We may want to do that
eventually, but I don't have a test case to support it yet.
Differential Revision: https://reviews.llvm.org/D75145
Summary:
Intended to be a non-functional change but it turned out CallEvent handles
constructor calls unlike CallExpr which doesn't triggered for constructors.
All in all, this change shouldn't be observable since constructors are not
yet propagating taintness like functions.
In the future constructors should propagate taintness as well.
This change includes:
- NFCi change all uses of the CallExpr to CallEvent
- NFC rename some functions, mark static them etc.
- NFC omit explicit TaintPropagationRule type in switches
- NFC apply some clang-tidy fixits
Reviewers: NoQ, Szelethus, boga95
Reviewed By: Szelethus
Subscribers: martong, whisperity, xazax.hun, baloghadamsoftware, szepet,
a.sidorin, mikhail.ramalho, donat.nagy, dkrupp, Charusso, cfe-commits
Tags: #clang
Differential Revision: https://reviews.llvm.org/D72035
As discussed in the commit thread for rGa253a2a and D73978, we can do more undef folding for FP ops.
The nnan and ninf fast-math-flags specify that if an operand is the disallowed value, the result is
poison, so we can produce an undef result.
But this doesn't work as expected (the undef operand cases remain) because of a Flags propagation
problem in SelectionDAGBuilder.
I've added DAGCombiner calls to enable these for the other cases because we've shown in other
patches that (because of the limited way that SDAG iterates), it is possible to miss simplifications
like this if they are done only at node creation time.
Several potential follow-ups to expand on this patch are possible.
Differential Revision: https://reviews.llvm.org/D75576
Recently introduced support for converting sequential reduction loops to
CFG of basic blocks in the Standard dialect makes it possible to perform
a staged conversion of parallel reduction loops into a similar CFG by
using sequential loops as an intermediate step. This is already the case
for parallel loops without reduction, so extend the pattern to support
an additional use case.
Differential Revision: https://reviews.llvm.org/D75599
We need to make sure that PluginAttrInstances is deleted before shared libraries
are unloaded, because otherwise when deleting its contents we'll try to access
a virtual destructor which no longer exists.
As shared libraries are managed using ManagedStatic we can do this by also using
ManagedStatic for PluginAttrInstances as ManagedStatics are deleted in reverse
order of construction and we know that PluginAttrInstances will only be
accessed, and thus constructed, after shared libraries have been loaded.
This reverts commit d93371238e.
The commit broke the build in several configurations (including
Windows and bare-metal). For details see comments in
https://reviews.llvm.org/D75480
This prevents calling Breakpoint::shared_from_this of an object that is not owned by any shared_ptr.
Differential Revision: https://reviews.llvm.org/D74557
Summary:
A test is passing `-o -` to lld in the hope of writing the output to
standard out but that is not the case. Instead it creates a file named
`-.lto.o`. This fixes it by creating a temporary file in the work
directory.
Differential Revision: https://reviews.llvm.org/D75605
This command had nearly identical code for the "then" and "else"
branches of the "if (m_options.num_instructions != 0)" condition.
This patch factors out the common parts of the two blocks to reduce
duplication.
Summary:
This adds an rsqrt op to the standard dialect, and lowers
it as 1 / sqrt to the LLVM dialect.
Differential Revision: https://reviews.llvm.org/D75353
Summary:
getInitialLength is a *DWARF*DataExtractor method so I had to "upgrade"
some DataExtractors to be able to make use of it.
Reviewers: ikudrin, jhenderson, probinson
Subscribers: aprantl, hiraditya, llvm-commits, dblaikie
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D75535
Guozhi Wei