This patch adds the functionality to compare BFI counts with real
profile
counts right after reading the profile. It will print remarks under
-Rpass-analysis=pgo, or the internal option -pass-remarks-analysis=pgo.
Differential Revision: https://reviews.llvm.org/D91813
The X86-64 ABI defines va_list as
typedef struct {
unsigned int gp_offset;
unsigned int fp_offset;
void *overflow_arg_area;
void *reg_save_area;
} va_list[1];
This means the size, alignment, and reg_save_area offset will depend on
whether we are in LP64 or in ILP32 mode, so this commit adds the checks.
Additionally, the VAARG_64 pseudo-instruction assumed 64-bit pointers, so
this commit adds a VAARG_X32 pseudo-instruction that behaves just like
VAARG_64, except for assuming 32-bit pointers.
Some of these changes were originally done by
Michael Liao <michael.hliao@gmail.com>.
Fixes https://bugs.llvm.org/show_bug.cgi?id=48428.
Reviewed By: RKSimon
Differential Revision: https://reviews.llvm.org/D93160
Make these arguments named constants in the Config class instead
of being positional arguments to MapAllocatorCache. This makes the
configuration easier to follow.
Eventually we should follow suit with the other classes but this is
a start.
Differential Revision: https://reviews.llvm.org/D93251
Checking that `T` is constructible from `Args...` is technically not
required by the Standard, although any implementation will obviously
error out if that's not satisfied. However, this check is incompatible
with using Allocator construction in the control block (upcoming change
as part of implementing P0674), so I'm removing it now to reduce the
upcoming diff as much as possible.
Differential Revision: https://reviews.llvm.org/D93246
As noted in D93229, the transform from scalar load to vector load
potentially leaks poison from the extra vector elements that are
being loaded.
We could use freeze here (and x86 codegen at least appears to be
the same either way), but we already have a shuffle in this logic
to optionally change the vector size, so let's allow that
instruction to serve both purposes.
Differential Revision: https://reviews.llvm.org/D93238
Migrate `HeaderSearch::LoadedModuleMaps` and a number of APIs over to
`FileEntryRef`. This should have no functionality change. Note that two
`FileEntryRef`s hash the same if they point at the same `FileEntry`.
Differential Revision: https://reviews.llvm.org/D92975
This adds support for loops like
unsigned clz(unsigned x) {
unsigned w = sizeof (x) * CHAR_BIT;
while (x) {
w--;
x >>= 1;
}
return w;
}
and
unsigned clz(unsigned x) {
unsigned w = sizeof (x) * CHAR_BIT - 1;
while (x >>= 1) {
w--;
}
return w;
}
To support these we look for add x, -1 as well as add x, 1 that
we already matched. If the value was -1 we need to subtract from
the initial counter value instead of adding to it.
Fixes PR48404.
Differential Revision: https://reviews.llvm.org/D92745
This revision adds a new `StaticVerifierFunctionEmitter` class that emits local static functions in the .cpp file for shared operation verification. This class deduplicates shared operation verification code by emitting static functions alongside the op definitions. These methods are local to the definition file, and are invoked within the operation verify methods. The first bit of shared verification is for the type constraints used when verifying operands and results. An example is shown below:
```
static LogicalResult localVerify(...) {
...
}
LogicalResult OpA::verify(...) {
if (failed(localVerify(...)))
return failure();
...
}
LogicalResult OpB::verify(...) {
if (failed(localVerify(...)))
return failure();
...
}
```
This allowed for saving >400kb of code size from a downstream TensorFlow project (~15% of MLIR code size).
Differential Revision: https://reviews.llvm.org/D91381
This commit is a step towards making it easier to add support for arrays
in allocate_shared. Adding support for arrays will require writing multiple
functions, and the current complexity of writing allocate_shared is
prohibitive for understanding.
Differential Revision: https://reviews.llvm.org/D93130
(The clang build fails for me locally, so this is based on built bot output and a guess as to root cause.)
f5fe849 made the execution of LAA conditional, so I'm guessing that's the root cause.
This patch implements a DDG printer pass that generates a graph in
the DOT description language, providing a more visually appealing
representation of the DDG. Similar to the CFG DOT printer, this
functionality is provided under an option called -dot-ddg and can
be generated in a less verbose mode under -dot-ddg-only option.
Differential Revision: https://reviews.llvm.org/D90159
Currently the backend special cases x86_intrcc and treats the first
parameter as byval. Make the IR require byval for this parameter to
remove this special case, and avoid the dependence on the pointee
element type.
Fixes bug 46672.
I'm not sure the IR is enforcing all the calling convention
constraints. clang seems to ignore the attribute for empty parameter
lists, but the IR tolerates it.
In addition to making the code a lot easier to grasp by localizing many
helper functions to the only file where they are actually needed, this
will allow creating helper functions that depend on allocator_traits
outside of <memory>.
This is done as part of implementing array support in allocate_shared,
which requires non-trivial array initialization algorithms that would be
better to keep out of <memory> for sanity. It's also a first step towards
splitting up our monolithic headers into finer grained ones, which will
make it easier to reuse functionality across the library. For example,
it's just weird that we had to define `addressof` inside <type_traits>
to avoid circular dependencies -- instead it's better to implement those
in true helper headers.
Differential Revision: https://reviews.llvm.org/D93074
This patch add support of riscv multilibs in the Baremetal toolchain. It is
a bit different to what is done in GNU.cpp as we are not iterating a
GNU sysroot to find the multilibs. This is intended for an llvm only
toolchain. We are not checking for the presence of any runtime bits to
enable a specific multilib.
I have structured the patch so that other targets for which
there is no multilibs support yet in Baremetal.cpp (e.g. arm-none-eabi)
will not be affected. Patch also allows some multilibs reuse.
Long term, I would like to go in the direction of data-driven specification of
multilib directories and flags.
Reviewed By: jroelofs
Differential Revision: https://reviews.llvm.org/D93138
The entry block should always be the first BB in a function.
So we should not rotate a chain contains the entry block.
Differential Revision: https://reviews.llvm.org/D92882
his is a preparation patch for supporting multiple exits in the loop vectorizer, by itself it should be mostly NFC. This patch moves the loop structure checks from LAA to their respective consumers (where duplicates don't already exist). Moving the checks does end up changing some of the optimization warnings and debug output slightly, but nothing that appears to be a regression.
Why do this? Well, after auditing the code, I can't actually find anything in LAA itself which relies on having all instructions within a loop execute an equal number of times. This patch simply makes this explicit so that if one consumer - say LV in the near future (hopefully) - wants to handle a broader class of loops, it can do so.
Differential Revision: https://reviews.llvm.org/D92066
This will allow for caching pattern lists across multiple pass instances, such as when multithreading. This is an extremely important invariant for PDL patterns, which are compiled at runtime when the FrozenRewritePatternList is built.
Differential Revision: https://reviews.llvm.org/D93146
Before this patch, the Restorer depended on copy elision to happen.
Without copy elision, the function ScopedSet calls the move constructor
before its dtor. The dtor will prematurely restore the reference to the
original value.
Instead of relying the compiler to not use the Restorer's copy
constructor, delete its copy and assign operators. Hence, callers cannot
move or copy a Restorer object anymore, and have to explicitly provide
the reset state. ScopedSet avoids calling move/copy operations by
relying on unnamed return value optimization, which is mandatory in
C++17.
Reviewed By: klausler
Differential Revision: https://reviews.llvm.org/D88797
Some operations use integer literals as part of their custom format that don't necessarily map to an internal IntegerAttr. This revision exposes the same `parseInteger` functions as the DialectAsmParser to allow for these operations to parse integer literals without incurring the otherwise unnecessary roundtrip through IntegerAttr.
Differential Revision: https://reviews.llvm.org/D93152
This revision adds a new `printNewline` hook to OpAsmPrinter that allows for printing a newline within the custom format of an operation, that is then indented to the start of the operation. Support for the declarative assembly format is also added, in the form of a `\n` literal.
Differential Revision: https://reviews.llvm.org/D93151
`isCUDADeviceBuiltinSurfaceType()`/`isCUDADeviceBuiltinTextureType()` do not
work on dependent types as they rely on specific type attributes.
Differential Revision: https://reviews.llvm.org/D92893
My bot runs VS 2019, but it could not compile this code.
Message:
[55/2465] Building CXX object lib\Target\Hexagon\CMakeFiles\LLVMHexagonCodeGen.dir\HexagonVectorCombine.cpp.obj
FAILED: lib/Target/Hexagon/CMakeFiles/LLVMHexagonCodeGen.dir/HexagonVectorCombine.cpp.obj
...
C:\Program Files (x86)\Microsoft Visual Studio\2019\Professional\VC\Tools\MSVC\14.23.28105\include\map(71): error C2976: 'std::map': too few template arguments
C:\Program Files (x86)\Microsoft Visual Studio\2019\Professional\VC\Tools\MSVC\14.23.28105\include\map(71): note: see declaration of 'std::map'
The version in the path, 14.23, corresponds to _MSC_VER 1923, so raise
the version floor to 1924.
I have not tested with versions between 1924 and 1928 (latest), but the
latest works with the variadic version.
This moves the vtype decoding and printing to RISCVBaseInfo. This keeps all of
the decoding code in the same area as the encoding code. This will make it
easier to change the decoding for the 1.0 spec in the future.
We're now sharing the printing with the debug output for operands in the
assembler. This also fixes that debug output to include the tail and mask
agnostic bits. Since the printing code works on the vtype immediate value, we
now encode the immediate during parsing and store just the immediate in the
operand.
We currently do this for SANITIZER_IOS, which includes devices *and* simulators. This change opts out the check for simulators to unify the behavior with macOS, because VM size is really a property of the host OS, and not the simulator.
<rdar://problem/72129387>
Differential Revision: https://reviews.llvm.org/D93140
Fix a bug causing to pick the wrong vector size to broadcast to when the source
vectors have different ranks.
Differential Revision: https://reviews.llvm.org/D93118
- New function SDValue getBackchainAddress() used by
lowerDYNAMIC_STACKALLOC() and lowerSTACKRESTORE() to properly handle the
backchain offset also with packed-stack.
- Make a common function getBackchainOffset() for the computation of the
backchain offset and use in some places (NFC).
Review: Ulrich Weigand
Differential Revision: https://reviews.llvm.org/D93171
- Once an instruction is simplified, foldable candidates from it should
be invalidated or skipped as the operand index is no longer valid.
Differential Revision: https://reviews.llvm.org/D93174
Rong Xu