There were basically two bugs here:
When C++20 `to_address` is called on `int arr[10]`, then `const _Ptr&` becomes
a reference to a const array, and then we dispatch to `__to_address<const int(&)[10]>`,
which, oops, gives us a `const int*` result instead of an `int*` result.
Solution: We need to provide the two standard-specified overloads of
`std::to_address` in exactly the same way that we provide two overloads
of `__to_address`.
When `__to_address` is called on a pointer type, `__to_address(const _Ptr&)`
is disabled so we successfully avoid trying to instantiate pointer_traits of
that pointer type. But when it's called on an array type, it's not disabled
for array types, so we go ahead and instantiate pointer_traits<int[10]>,
which goes boom. Solution: We need to disable `__to_address(const _Ptr&)`
for both pointer and array types. Also disable it for function types,
so that they get the nice error message; and put a test on it.
Differential Revision: https://reviews.llvm.org/D109331
Add tests showing `span` is trivially_destructible and nothrow_destructible.
Note that we do not need to explicitly default the destructor in `span`.
Reviewed By: ldionne, Mordante, #libc
Differential Revision: https://reviews.llvm.org/D109286
Similar to D108842, D108844, and D108926.
__has_builtin(builtin_mul_overflow) returns true for 32b x86 targets,
but Clang is deferring to compiler RT when encountering long long types.
This breaks ARCH=i386 + CONFIG_BLK_DEV_NBD=y builds of the Linux kernel
that are using builtin_mul_overflow with these types for these targets.
If the semantics of __has_builtin mean "the compiler resolves these,
always" then we shouldn't conditionally emit a libcall.
This will still need to be worked around in the Linux kernel in order to
continue to support these builds of the Linux kernel for this
target with older releases of clang.
Link: https://bugs.llvm.org/show_bug.cgi?id=28629
Link: https://bugs.llvm.org/show_bug.cgi?id=35922
Link: https://github.com/ClangBuiltLinux/linux/issues/1438
Reviewed By: lebedev.ri, RKSimon
Differential Revision: https://reviews.llvm.org/D108928
Implement IEEE Real::SQRT() operation, then use it to
also implement Real::HYPOT(), which can then be used directly
to implement Complex::ABS().
Differential Revision: https://reviews.llvm.org/D109250
The lowering has been incorrectly using the operands of the original op instead
of rewritten operands provided to matchAndRewrite call. This may lead to
spurious materializations and generally invalid IR.
Reviewed By: aartbik
Differential Revision: https://reviews.llvm.org/D109355
Use the `HBuilder` interface to provide default implementations of `llvm::hash_value`.
Reviewed By: dexonsmith
Differential Revision: https://reviews.llvm.org/D109024
Extend the existing single-pass algorithm for `Merger::Merge` with an algorithm that gives better results. This new implementation can be used with a new **set_cover_merge=1** flag.
This greedy set cover implementation gives a substantially smaller final corpus (40%-80% less testcases) while preserving the same features/coverage. At the same time, the execution time penalty is not that significant (+50% for ~1M corpus files and far less for smaller corpora). These results were obtained by comparing several targets with varying size corpora.
Change `Merger::CrashResistantMergeInternalStep` to collect all features from each file and not just unique ones. This is needed for the set cover algorithm to work correctly. The implementation of the algorithm in `Merger::SetCoverMerge` uses a bitvector to store features that are covered by a file while performing the pass. Collisions while indexing the bitvector are ignored similarly to the fuzzer.
Reviewed By: morehouse
Differential Revision: https://reviews.llvm.org/D105284
Extract `HashBuilder::IsHashableData` out of class; it does not depend on
template parametres.
Reviewed By: dexonsmith
Differential Revision: https://reviews.llvm.org/D109205
The new interface only marks begin/end of a scope construct for
corresponding OMPT events, and we can use existing interfaces for
reduction operations.
Differential Revision: https://reviews.llvm.org/D108062
It only worked for internal procedures of subprograms,
but must also allow for internal procedures of the
main program. This broke the use of host-associated
implicitly-typed symbols in specification expressions
of internal procedures.
Differential Revision: https://reviews.llvm.org/D109262
Using the similarity found from the IRSimilarity Identifier, we take regions with structural similarity, and deduplicate them into a separate function. The Code Extractor is able to provide most of this functionality.
For simplicity, we start by only outlining regions with a single entry and single exit branch, this reduces the complexity in handling phi nodes outside the region, and handling many sets of outputs for each of the different exit blocks.
Reviewer: paquette
Differential Revision: https://reviews.llvm.org/D106990
This patch adds a fix to do early if conversion to select when
conditional branch not using physical register to prevent the crash when
expanding ISEL instruction.
Reviewed By: lei, kamaub, PowerPC
Differential revision: https://reviews.llvm.org/D108302
The target platform could differ from the host platform for the cross
platform builds. Some tests are depended on the build host features and
they need to determine a proper platform environment.
This commit adds a build host platform name feature for the libc++ tests
in format `buildhost=<platform>`, such as `buildhost=linux`, `buildhost=darwin`,
`buildhost=windows`, etc.
The Windows host gets two features: one `buildhost=windows` and another based
on Windows "sub-system", such as `buildhost=win32`, `buildhost=cygwin`, etc.
Differential Revision: https://reviews.llvm.org/D102045
mdfind can return multiple results, some of which are not even dSYM
bundles, but Xcode archives (.xcrachive).
Currently, we end up concatenating the paths, which is obviously bogus.
This patch not only fixes that, but now also skips paths that don't have
a Contents/Resources/DWARF subdirectory.
rdar://81270312
Differential revision: https://reviews.llvm.org/D109263
It doesn't appear to be possible to generate this from tests atm, but it matches what we do in sse12_ord_cmp
Fixes unused template arg identified in D109359
This is exposed by enabling FastIsel on 64bit AIX.
We are generating XSRSP regardless of the arch,
which may be wrong when -mcpu=pwr7.
The fix is to guard the generation in P8 only.
Reviewed By: qiucf
Differential Revision: https://reviews.llvm.org/D109365
so that it gets installed in LLVM_INSTALL_TOOLCHAIN_ONLY builds,
such as used by the Windows installer.
Differential revision: https://reviews.llvm.org/D109358
Currently native clusterization simply groups all benchmarks
by the opcode of key instruction, but that is suboptimal in certain cases,
e.g. where we can already tell that the particular instructions
already resolve into different sched classes.
The conversion has been incorrectly using the operands of the original
operation instead of the converted operands provided to the matchAndRewrite
call. This may lead to spurious materializations and generally invalid IR if
the producer of the original operands is deleted in the process of conversion.
Reviewed By: csigg
Differential Revision: https://reviews.llvm.org/D109356
Using PUNPKLO/HI instead of ZIP1/ZIP2, because that avoids
having to generate a predicate with all lanes inactive (PFALSE).
Reviewed By: CarolineConcatto
Differential Revision: https://reviews.llvm.org/D109312
On some architectures such as Arm and X86 the encoding for a nop may
change depending on the subtarget in operation at the time of
encoding. This change replaces the per module MCSubtargetInfo retained
by the targets AsmBackend in favour of passing through the local
MCSubtargetInfo in operation at the time.
On Arm using the architectural NOP instruction can have a performance
benefit on some implementations.
For Arm I've deleted the copy of the AsmBackend's MCSubtargetInfo to
limit the chances of this causing problems in the future. I've not
done this for other targets such as X86 as there is more frequent use
of the MCSubtargetInfo and it looks to be for stable properties that
we would not expect to vary per function.
This change required threading STI through MCNopsFragment and
MCBoundaryAlignFragment.
I've attempted to take into account the in tree experimental backends.
Differential Revision: https://reviews.llvm.org/D45962
In preparation for passing the MCSubtargetInfo (STI) through to writeNops
so that it can use the STI in operation at the time, we need to record the
STI in operation when a MCAlignFragment may write nops as padding. The
STI is currently unused, a further patch will pass it through to
writeNops.
There are many places that can create an MCAlignFragment, in most cases
we can find out the STI in operation at the time. In a few places this
isn't possible as we are in initialisation or finalisation, or are
emitting constant pools. When possible I've tried to find the most
appropriate existing fragment to obtain the STI from, when none is
available use the per module STI.
For constant pools we don't actually need to use EmitCodeAlign as the
constant pools are data anyway so falling through into it via an
executable NOP is no better than falling through into data padding.
This is a prerequisite for D45962 which uses the STI to emit the
appropriate NOP for the STI. Which can differ per fragment.
Note that involves an interface change to InitSections. It is now
called initSections and requires a SubtargetInfo as a parameter.
Differential Revision: https://reviews.llvm.org/D45961
Legalizing G_MUL for non-standard types (like i33) generated an error. Putting
minScalar and maxScalar instead of clampScalar. Also using new rule, instead
of widening to the next power of 2, widen to the next multiple of the passed
argument (32 in this case), so instead of widening i65 to i128, we widen it to
i96.
Patch by: Mateja Marjanovic
Differential Revision: https://reviews.llvm.org/D109228
Add implementation for the legalization of G_ROTL and G_ROTR machine
instructions. They are very similar to funnel shift instructions, the only
difference is funnel shifts have 3 operands, whereas rotate instructions have
two operands, the first being the register that is being rotated and the second
being the number of shifts. The legalization of G_ROTL/G_ROTR is just lowering
them into funnel shift instructions if they are legal.
Patch by: Mateja Marjanovic
Differential Revision: https://reviews.llvm.org/D105347
This was an accidental behaviour change in D106789 and this patch
restores it back to original state.
Differential Revision: https://reviews.llvm.org/D109361
This patch changes the default monotonicity of dynamic schedule from
monotonic to non-monotonic when no modifier is specified.
Differential Revision: https://reviews.llvm.org/D109026
We previously didn't have any tests to defend the cost model
for gathers and scatters using SVE without a vscale_range
attribute. I've added tests to existing files:
Analysis/CostModel/AArch64/sve-gather.ll
Analysis/CostModel/AArch64/sve-scatter.ll
Differential Revision: https://reviews.llvm.org/D109055
Move inverse_throughput, latency and uops to sub-directories (like we already do for lbr), which require libpfm, so we can relax the lit limits for analysis tests in the x86 root directory.
Differential Revision: https://reviews.llvm.org/D109353
Add KnownBits handling and unit tests for X*X self-multiplication cases which guarantee that bit1 of their results will be zero - see PR48683.
https://alive2.llvm.org/ce/z/NN_eaR
The next step will be to add suitable test coverage so this can be enabled in ValueTracking/DAG/GlobalISel - currently only a single Analysis/ScalarEvolution test is affected.
Differential Revision: https://reviews.llvm.org/D108992
Legalize G_MEMCPY, G_MEMMOVE, G_MEMSET and G_MEMCPY_INLINE.
Corresponding intrinsics are replaced by a loop that uses loads/stores in
AMDGPULowerIntrinsics pass unless their length is a constant lower then
MemIntrinsicExpandSizeThresholdOpt (default 1024). Any G_MEM* instruction that
reaches legalizer should have a const length argument and should be expanded
into appropriate number of loads + stores.
Differential Revision: https://reviews.llvm.org/D108357
Arthur O'Dwyer