Otherwise callers of these functions have to check both the return value
for and the contents of the returned llvm::Optional.
Fixes#53742
Differential Revision: https://reviews.llvm.org/D119525
In 529aa4b011
by setting the identifier info to nullptr, we started to subtly
interfere with the parts in the beginning of the function,
529aa4b011/clang/lib/Format/UnwrappedLineParser.cpp (L991)
causing the preprocessor nesting to change in some cases. E.g., for the
added regression test, clang-format started incorrectly guessing the
language as C++.
This tries to address this by introducing an internal identifier info
element to use instead.
Reviewed By: curdeius, MyDeveloperDay
Differential Revision: https://reviews.llvm.org/D120315
'streaming-sve' is not a feature that users should be able to set,
hence why it shouldn't show up in user-diagnostics. The only
flag that end-users should be able to set is '+sme'.
Reviewed By: paulwalker-arm
Differential Revision: https://reviews.llvm.org/D120256
This patch is the first in a series of patches to upstream the support for Apple's DriverKit. Once complete, it will allow targeting DriverKit platform with Clang similarly to AppleClang.
This code was originally authored by JF Bastien.
Differential Revision: https://reviews.llvm.org/D118046
Split v512.32 binary ops into two v256.32 ops using packing support
opcodes (vec_unpack_lo|hi, vec_pack).
Depends on D120053 for packing opcodes.
Reviewed By: kaz7
Differential Revision: https://reviews.llvm.org/D120146
With only a load-fold the diffs look neutral. If there's a load and store (rmw)
fold opportunity as shown in the test based on #53862, then we end up with an
extra instruction.
Fixes#53862
Differential Revision: https://reviews.llvm.org/D120281
This patch brings in some initial changes for lowering Fortran
intrinsics. Intrinsics are generally lowered to a mix of FIR and
MLIR operations, runtime calls or LLVM intrinsics. This patch
particularly brings in the lowering of the Fortran `andi` intrinsic
to `arith.andi` in MLIR.
The significant changes are in ConvertExpr.cpp and IntrinsicCall.cpp.
Intrinsic functions occur as part of expressions. Lowering deals with this
in ConvertExpr.cpp in `genval(const Fortran::evaluate::FunctionRef<A> &funcRef)`.
The code in the above mentioned function kicks of a sequence of calls
that ultimately results in a call to the `genIand ` function in
IntrinsicCall.cpp which creates the MLIR `arith.andi` operation.
A few tests are also included.
Note: Generally intrinsics like `iand` can occur in array (elemental)
context, but since that part is not fully supported in lowering, tests
are only added for the scalar context.
This patch is part of upstreaming from the fir-dev branch of
https://github.com/flang-compiler/f18-llvm-project.
Reviewed By: clementval
Differential Revision: https://reviews.llvm.org/D119990
Co-authored-by: Jean Perier <jperier@nvidia.com>
Co-authored-by: Eric Schweitz <eschweitz@nvidia.com>
Co-authored-by: zacharyselk <zrselk@gmail.com>
Co-authored-by: V Donaldson <vdonaldson@nvidia.com>
Co-authored-by: Valentin Clement <clementval@gmail.com>
Extends getReductionOpChain to look through Phis which may be part of
the reduction chain. adjustRecipesForReductions will now also create a
CondOp for VPReductionRecipe if the block is predicated and not only if
foldTailByMasking is true.
Changes were required in tryToBlend to ensure that we don't attempt
to convert the reduction Phi into a select by returning a VPBlendRecipe.
The VPReductionRecipe will create a select between the Phi and the reduction.
Reviewed By: david-arm
Differential Revision: https://reviews.llvm.org/D117580
The race is between these two pieces of code that are executed in two separate
lldb-vscode threads (the first is in the main thread and another is in the
event-handling thread):
```
// lldb-vscode.cpp
g_vsc.debugger.SetAsync(false);
g_vsc.target.Launch(launch_info, error);
g_vsc.debugger.SetAsync(true);
```
```
// Target.cpp
bool old_async = debugger.GetAsyncExecution();
debugger.SetAsyncExecution(true);
debugger.GetCommandInterpreter().HandleCommands(GetCommands(), exc_ctx,
options, result);
debugger.SetAsyncExecution(old_async);
```
The sequence that leads to the bug is this one:
1. Main thread enables synchronous mode and launches the process.
2. When the process is launched, it generates the first stop event.
3. This stop event is catched by the event-handling thread and DoOnRemoval
is invoked.
4. Inside DoOnRemoval, this thread runs stop hooks. And before running stop
hooks, the current synchronization mode is stored into old_async (and
right now it is equal to "false").
5. The main thread finishes the launch and returns to lldb-vscode, the
synchronization mode is restored to asynchronous by lldb-vscode.
6. Event-handling thread finishes stop hooks processing and restores the
synchronization mode according to old_async (i.e. makes the mode synchronous)
7. And now the mode is synchronous while lldb-vscode expects it to be
asynchronous. Synchronous mode forbids the process to broadcast public stop
events, so, VS Code just hangs because lldb-vscode doesn't notify it about
stops.
So, this diff makes the target intercept the first stop event if the process is
launched in the synchronous mode, thus preventing stop hooks execution.
The bug is only present on Windows because other platforms already
intercept this event using their own hijacking listeners.
So, this diff also fixes some problems with lldb-vscode tests on Windows to make
it possible to run the related test. Other tests still can't be enabled because
the debugged program prints something into stdout and LLDB can't intercept this
output and redirect it to lldb-vscode properly.
Reviewed By: jingham
Differential Revision: https://reviews.llvm.org/D119548
Header class in SPIR-V HTML spec has changed. Update script to reflect that.
Reviewed By: antiagainst
Differential Revision: https://reviews.llvm.org/D120179
This patch fixes a logical error in how we work with `LoopUsers` map.
It maps a loop onto a set of AddRecs that depend on it. The Addrecs
are added to this map only once when they are created and put to
the UniqueSCEVs` map.
The only purpose of this map is to make sure that, whenever we forget
a loop, all (directly or indirectly) dependent SCEVs get forgotten too.
Current code erases SCEVs from dependent set of a given loop whenever
we forget this loop. This is not a correct behavior due to the following scenario:
1. We have a loop `L` and an AddRec `AR` that depends on it;
2. We modify something in the loop, but don't destroy it. We still call forgetLoop on it;
3. `AR` is no longer dependent on `L` according to `LoopUsers`. It is erased from
ValueExprMap` and `ExprValue map, but still exists in UniqueSCEVs;
4. We can later request the very same AddRec for the very same loop again, and get existing
SCEV `AR`.
5. Now, `AR` exists and is used again, but its notion that it depends on `L` is lost;
6. Then we decide to delete `L`. `AR` will not be forgotten because we have lost it;
7. Just you wait when you run into a dangling pointer problem, or any other kind of problem
because an active SCEV is now referecing a non-existent loop.
The solution to this is to stop erasing values from `LoopUsers`. Yes, we will maybe forget something
that is already not used, but it's cheap.
This fixes a functional bug and potentially may have negative compile time impact on methods with
huge or numerous loops.
Differential Revision: https://reviews.llvm.org/D120303
Reviewed By: nikic
Most places already seem to use the short spelling instead of
'unsigned int/long', so perform the following substitutions:
s/unsigned int /uint /g
s/unsigned long /ulong /g
This simplifies completeness comparisons against OpenCLBuiltins.td.
Differential Revision: https://reviews.llvm.org/D120032
This is an initial enabling patch for module partition support.
We add enumerations for partition interfaces/implementations.
This means that the module kind enumeration now occupies three
bits, so the AST streamer is adjusted for this. Adding one bit there
seems preferable to trying to overload the meanings of existing
kinds (and we will also want to add a C++20 header unit case later).
Differential Revision: https://reviews.llvm.org/D114714
As suggested by the cmake warning:
CMake Warning at <...>/llvm-project/libcxx-ci/libcxx/CMakeLists.txt:289 (message):
LIBCXX_TARGET_TRIPLE is deprecated, please use CMAKE_CXX_COMPILER_TARGET instead
Depends on D119948
Differential Revision: https://reviews.llvm.org/D120038
This patch fixes an invalid TypeSize->uint64_t implicit conversion in
FoldReinterpretLoadFromConst. If the size of the constant is scalable
we bail out of the optimisation for now.
Tests added here:
Transforms/InstCombine/load-store-forward.ll
Differential Revision: https://reviews.llvm.org/D120240
Summary:
This patch adds checks that were missing in clang for Armv8.5/6/7-A. These include:
* ACLE macro defines for AArch32.
* Handling of crypto and SM4, SHA and AES feature flags on clang's driver.
Reviewers: dmgreen, SjoerdMeijer, tmatheson
Differential Revision: https://reviews.llvm.org/D116153
Constants cannot be cyclic, but they can be tree-like. Keep a
visited set to ensure we do not degenerate to exponential run-time.
This fixes the problem reported in https://reviews.llvm.org/D117223#3335482,
though I haven't been able to construct a concise test case for
the issue. This requires a combination of dead constants and the
kind of constant expression tree that textual IR cannot represent
(because the textual representation, unlike the in-memory
representation, is also exponential in size).
The related functionality is moved over to the bufferization dialect. Test cases are cleaned up a bit.
Differential Revision: https://reviews.llvm.org/D120191
Adds a new option InsertBraces to insert the optional braces after
if, else, for, while, and do in C++.
Differential Revision: https://reviews.llvm.org/D120217
We use offloading sections in the new Clang driver scheme to embed
device code into the host. We later use these sections to link the
device image, after which point they are completely unused and should
not be loaded into memory if they are still in the executable.
Reviewed By: JonChesterfield
Differential Revision: https://reviews.llvm.org/D120275
We found a case in the Swift benchmarks where the MachineOutliner introduces
about a 20% compile time overhead in comparison to building without the
MachineOutliner.
The origin of this slowdown is that the benchmark has long blocks which incur
lots of LRU checks for lots of candidates.
Imagine a case like this:
```
bb:
i1
i2
i3
...
i123456
```
Now imagine that all of the outlining candidates appear early in the block, and
that something like, say, NZCV is defined at the end of the block.
The outliner has to check liveness for certain registers across all candidates,
because outlining from areas where those registers are used is unsafe at call
boundaries.
This is fairly wasteful because in the previously-described case, the outlining
candidates will never appear in an area where those registers are live.
To avoid this, precalculate areas where we will consider outlining from.
Anything outside of these areas is mapped to illegal and not included in the
outlining search space. This allows us to reduce the size of the outliner's
suffix tree as well, giving us a potential memory win.
By precalculating areas, we can also optimize other checks too, like whether
or not LR is live across an outlining candidate.
Doing all of this is about a 16% compile time improvement on the case.
This is likely useful for other targets (e.g. ARM + RISCV) as well, but for now,
this only implements the AArch64 path. The original "is the MBB safe" method
still works as before.
Otherwise this is can be sensitive in the face of changes in register names.
I also gardened the test case a little to make it look a little nicer.
Reviewed By: fhahn
Differential Revision: https://reviews.llvm.org/D120276
Timm Bäder