For large integers (for example, magic numbers generated by
TargetLowering::BuildSDIV when dividing by constant), we may
need about 4~8 instructions to build them.
In the same time, it just takes two instructions to load
constants (with extra cycles to access memory), so it may be
profitable to put these integers into constant pool.
Reviewed By: asb, craig.topper
Differential Revision: https://reviews.llvm.org/D114950
https://reviews.llvm.org/D109555 added support to APInt for this, so the special case to disable it is no longer valid. It is in fact legal to construct these programmatically today, and they print properly but do not parse.
Justification: zero bit integers arise naturally in various bit reduction optimization problems, and having them defined for MLIR reduces special casing.
I think there is a solid case for i0 and ui0 being supported. I'm less convinced about si0 and opted to just allow the parser to round-trip values that already verify. The counter argument is that the proper singular value for an si0 is -1. But the counter to this counter is that the sign bit is N-1, which does not exist for si0 and it is not unreasonable to consider this non-existent bit to be 0. Various sources consider it having the singular value "0" to be the least surprising.
Reviewed By: lattner
Differential Revision: https://reviews.llvm.org/D116413
The 'New' only makes sense in the context of these being
output arguments, but they are also used as inputs first.
Drop the 'New' and just call them LHS/RHS.
Factored out of D116421.
In debug info correlation mode we do not emit the data globals so we
need to explicitly mark the counter globals as used so they don't get
stripped.
Reviewed By: kyulee
Differential Revision: https://reviews.llvm.org/D115981
These method currently takes a SmallVector<AffineExpr> & as an
argument to return the dims as AffineExpr. This creation of
AffineExpr objects is unnecessary.
Differential Revision: https://reviews.llvm.org/D116422
Need to clear CurrentOrder order mask if it is determined that
extractelements form identity order and need to use a vector-like
construct when iterating over ordered entries in the reorderTopToBottom
function.
Similar to D62188: a BitcodeFile's symbol table may be iterated twice, once in
--start-lib (lazy) state, and once in the non-lazy state. This patch
makes `parseLazy` save `symbols[i]` so that the non-lazy state does not need to
re-insert to the global symbol table. Avoiding a redundant `saver.save` may save
memory.
`Maximum resident set size (kbytes)` for a large --thinlto-index-only link:
* without the patch: 10164000
* with the patch: 10095716 (0.6% decrease)
Note: we can remove `saver.save` if `BitcodeCompiler::add` does not transfer the ownership
of `f.obj` in `checkError(ltoObj->add(std::move(f.obj), resols));`.
Reviewed By: tejohnson
Differential Revision: https://reviews.llvm.org/D116390
This patch adds isel support for STRICT_LRINT/LLRINT/LROUND/LLROUND.
It also adds test cases for f32 and f64 constrained intrinsics that
correspond to the intrinsics in float-intrinsics.ll and
double-intrinsics.ll. Support for promoting the integer argument of
STRICT_FPOWI was added.
I've skipped adding tests for f16 intrinsics, since we don't have libcalls
for them and we have inconsistent support for promoting them in LegalizeDAG.
This will need to be examined more closely.
Reviewed By: asb
Differential Revision: https://reviews.llvm.org/D116323
Both of DenseElementsAttr and ElementsAttrTrait define the method of
getElementType, this commit makes it available on
DenseIntOrFPElementsAttr and DenseStringElementsAttr.
Reviewed By: mehdi_amini
Differential Revision: https://reviews.llvm.org/D116389
@tejohnson noticed that freeing MemoryBuffer instances right before
`lto->compile` can save RSS, likely because the memory can be reused by
LTO indexing (e.g. ThinLTO import/export lists).).
For ELFFileBase instances, symbol and section names are backed by MemoryBuffer,
so destroying MemoryBuffer would make some infrequent passes (parseSymbolVersion,
reportBackrefs) crash and make debugging difficult.
For a BitcodeFile, its content is completely unused, but destroying its
MemoryBuffer makes the buffer identifier inaccessible and may introduce
constraints for future changes.
This patch leverages madvise(MADV_DONTNEED) which achieves the major gain
without the latent issues.
`Maximum resident set size (kbytes): ` for a large --thinlto-index-only link:
* current behavior: 10146104KiB
* destroy MemoryBuffer instances: 8555240KiB
* madvise(MADV_DONTNEED) just bitcodeFiles and lazyBitcodeFiles: 8737372KiB
* madvise(MADV_DONTNEED) all MemoryBuffers: 8739796KiB (16% decrease)
Depends on D116366
Reviewed By: tejohnson
Differential Revision: https://reviews.llvm.org/D116367
Per the discussion in https://reviews.llvm.org/D116345 it makes sense
to move AtomicRMWOp out of the standard dialect. This was accentuated by the
need to add a fold op with a memref::cast. The only dialect
that would permit this is the memref dialect (keeping it in the standard dialect
or moving it to the arithmetic dialect would require those dialects to have a
dependency on the memref dialect, which breaks linking).
As the AtomicRMWKind enum is used throughout, this has been moved to Arith.
Reviewed By: Mogball
Differential Revision: https://reviews.llvm.org/D116392
Extends response file expansion to recognize `<CFGDIR>` and expand to the
current file's directory. This makes it much easier to author clang config
files rooted in portable, potentially not-installed SDK directories.
A typical use case may be something like the following:
```
# sample_sdk.cfg
--target=sample
-isystem <CFGDIR>/include
-L <CFGDIR>/lib
-T <CFGDIR>/ldscripts/link.ld
```
Reviewed By: sepavloff
Differential Revision: https://reviews.llvm.org/D115604
On *NIX systems, this API calls madvise(MADV_DONTNEED) on read-only file mappings.
It should not be used on a writable buffer.
The API is used to implement ld.lld LTO memory saving trick (D116367).
Note: on read-only file mappings, Linux's MADV_DONTNEED semantics match POSIX
POSIX_MADV_DONTNEED and BSD systems' MADV_DONTNEED.
On Windows, VirtualAllocEx MEM_COMMIT/MEM_RESET have similar semantics
but are unfortunately not drop-in replacements. dontNeedIfMmap is currently a no-op.
Reviewed By: aganea
Differential Revision: https://reviews.llvm.org/D116366
To match the `--help` message and most other utilities.
While here, change `option:: -output=output` to `option:: --output=<output>` and
omit the value name for the short options (convention of other utilities).
Reviewed By: snehasish
Differential Revision: https://reviews.llvm.org/D116353
This reverts commit e60d6dfd5a.
clang-ppc64le-rhel buildbot failed (https://lab.llvm.org/buildbot#builders/57/builds/13424):
tools/lld/MachO/CMakeFiles/lldMachO.dir/Symbols.cpp.o: In function `lld::demangle(llvm::StringRef, bool)':
Symbols.cpp:(.text._ZN3lld8demangleEN4llvm9StringRefEb[_ZN3lld8demangleEN4llvm9StringRefEb]+0x90): undefined reference to `llvm::demangle(std::string const&)'
LLVM core library supports demangling other mangled symbols other than itanium,
such as D and Rust. LLD should use those demanglers in order to output pretty
demangled symbols on error messages.
Reviewed By: MaskRay
Differential Revision: https://reviews.llvm.org/D116279
((X << C) + Y) >>u C --> (X + (Y >>u C)) & (-1 >>u C)
https://alive2.llvm.org/ce/z/DY9DPg
This replaces a shift with an 'and', and in the case
where the add has a constant operand, it eliminates
both shifts.
As noted in the TODO comment, we already have this fold when
the shifts are in the opposite order (and that code handles
bitwise logic ops too).
Fixes#52851
((X << C) + Y) >>u C --> (X + (Y >>u C)) & (-1 >>u C)
https://alive2.llvm.org/ce/z/DY9DPg
This replaces a shift with an 'and', and in the case
where the add has a constant operand, it eliminates
both shifts.
As noted in the TODO comment, we already have this fold when
the shifts are in the opposite order (and that code handles
bitwise logic ops too).
Fixes#52851
This reverts commit ba07f300c6.
A build-vector sequence is made of pairs: rotate+insert. When constructing
a single vector, this results in a chain of 2*N instructions. The rotate
operation is a permute operation, but the insert uses a multiplication
resource: insert and rotate can execute in the same cycle, but obviously
they cannot operate on the same vector. The original halving idea is still
beneficial since it does allow for insert/rotate overlap, and for hiding
insert's latency.
vector.transfer operations do not have rank-reducing semantics.
Bail on illegal rank-reduction: we need to check that the rank-reduced
dims are exactly the leading dims. I.e. the following is illegal:
```
%0 = vector.transfer_write %v, %t[0,0], %cst :
vector<2x4xf32>, tensor<2x4xf32>
%1 = tensor.insert_slice %0 into %tt[0,0,0][2,1,4][1,1,1] :
tensor<2x4xf32> into tensor<2x1x4xf32>
```
Cannot fold into:
```
%0 = vector.transfer_write %v, %t[0,0,0], %cst :
vector<2x4xf32>, tensor<2x1x4xf32>
```
For this, check the trailing `vectorRank` dims of the insert_slice result
tensor match the trailing dims of the inferred result tensor.
Differential Revision: https://reviews.llvm.org/D116409
This small patch adds two useful improvements:
- allows one to specify the emulator path as a bare filename, and have
it be looked up in the PATH
- allows one to leave the path empty and have the filename be derived
from the architecture.
In D109958 it was noticed that we could optimise the pipeline and avoid
rerunning LoopSimplify/LCSSA for LoopFlatten by moving it to a LoopPassManager.
Differential Revision: https://reviews.llvm.org/D110057
It was being used only in some very old tests (which pass even without
it) and its implementation is highly questionable.
These days we have different mechanisms for requesting a build with a
particular kind of c++ library (USE_LIB(STD)CPP in the makefile).
The bitcode reader expected that the pointers are typed,
so that it can extract the function type for the assembly
so `bitc::CST_CODE_INLINEASM` did not explicitly store said function type.
I'm not really sure how the upgrade path will look for existing bitcode,
but i think we can easily support opaque pointers going forward,
by simply storing the function type.
Reviewed By: #opaque-pointers, nikic
Differential Revision: https://reviews.llvm.org/D116341
Craig Topper