Previously, convertToMLIRSparseTensor assumes identity storage ordering and all
compressed dimensions. This change extends the function with two parameters for
users to specify the storage ordering and the sparsity of each dimension.
Modify PyTACO to reflect this change.
Reviewed By: aartbik
Differential Revision: https://reviews.llvm.org/D120643
Add an explicit LIBCXX_CXX_ABI=system-libcxxabi option for linking to
system-installed libc++abi. This fixes the ability to link against one
when building libcxx via the runtimes build, as otherwise the build
system insists on linking into in-tree targets.
Differential Revision: https://reviews.llvm.org/D119539
As discussed in https://reviews.llvm.org/D119743 scf.parallel would continuously stack allocate since the alloca op was placd in the wsloop rather than the omp.parallel. This PR is the second stage of the fix for that problem. Specifically, we now introduce an alloca scope around the inlined body of the scf.parallel and enable a canonicalization to hoist the allocations to the surrounding allocation scope (e.g. omp.parallel).
Reviewed By: ftynse
Differential Revision: https://reviews.llvm.org/D120423
This lowers VECTOR_SPLICE of scalable vectors to a slidedown follow by a slideup.
Fixed vectors are encouraged to use shufflevector instruction. The equivalent patch
for fixed vectors is D119039.
I've used a tail agnostic slidedown and limited the VL to only the
elements that will not be overwritten by the slideup. The slideup
uses VLMax for its VL. It unfortunately uses tail undisturbed policy
but it isn't required as there is no tail. We just need the merge
operand to carry the bits for the lower portion of the result.
Care was taken to ensure that either the slideup or slidedown will
be able to use a .vi instruction when the immediate is small. Which
one uses the immediate depends on the sign of the immediate.
Reviewed By: frasercrmck, ABataev
Differential Revision: https://reviews.llvm.org/D119303
Instead of passing an InstCmpInt * and a bool just pass the predicate
from the caller.
I'm considering moving the similar FCmp functions from InstCombine
over here and this makes the interface consistent with what is used
for FCmp.
Reviewed By: RKSimon
Differential Revision: https://reviews.llvm.org/D120609
Currently adding attribute no_sanitize("bounds") isn't disabling
-fsanitize=local-bounds (also enabled in -fsanitize=bounds). The Clang
frontend handles fsanitize=array-bounds which can already be disabled by
no_sanitize("bounds"). However, instrumentation added by the
BoundsChecking pass in the middle-end cannot be disabled by the
attribute.
The fix is very similar to D102772 that added the ability to selectively
disable sanitizer pass on certain functions.
In this patch, if no_sanitize("bounds") is provided, an additional
function attribute (NoSanitizeBounds) is attached to IR to let the
BoundsChecking pass know we want to disable local-bounds checking. In
order to support this feature, the IR is extended (similar to D102772)
to make Clang able to preserve the information and let BoundsChecking
pass know bounds checking is disabled for certain function.
Reviewed By: melver
Differential Revision: https://reviews.llvm.org/D119816
Essentially removed the "use omp_lib_kinds" statement and replaced it
with import to maintain consistency (and avoid compilation error
in case the omp_lib_kinds.mod file is not accessible) in header file.
The import is required to access entities in host scoping unit.
Differential Revision: https://reviews.llvm.org/D120707
New methods to the Atomic class have been added as required. Futex
related types have been consolidated at a common place.
Reviewed By: lntue
Differential Revision: https://reviews.llvm.org/D120705
Handle V_MAC_LEGACY_F32 and V_FMAC_LEGACY_F32 in
convertToThreeAddress, to avoid the need for an extra mov
instruction in some cases.
Differential Revision: https://reviews.llvm.org/D120704
Move MFMA handling to the top of convertToThreeAddress and pull
IsF16 calculation out of the switch. I think this makes it clearer
exactly which mac/fmac opcodes are handled, since they are now
listed in the switch with minimal extra clutter.
Differential Revision: https://reviews.llvm.org/D120703
This does tail merging (and deduplication) of the strings.
On a statically linked clang.exe, this shrinks the ~17 MB string
table by around 0.5 MB. This adds ~160 ms to the linking time
which originally was around 950 ms.
For cases where `-debug:symtab` or `-debug:dwarf` isn't set, the
string table is only used for long section names, where this
shouldn't make any difference at all.
Differential Revision: https://reviews.llvm.org/D120677
Judging by the name, and comparing with the f32 version, this was
supposed to be testing that FMAC with a non-inlinable constant
operand did not get converted to FMA.
This is used a few places in SemaTeplateInstantiateDecl, but is going
to be useful in SemaConcept.cpp as well. This patch switches it to be
a private function in Sema.
Differential Revision: https://reviews.llvm.org/D120729
Since CallDescriptions can only be matched against CallEvents that are created
during symbolic execution, it was not possible to use it in syntactic-only
contexts. For example, even though InnerPointerChecker can check with its set of
CallDescriptions whether a function call is interested during analysis, its
unable to check without hassle whether a non-analyzer piece of code also calls
such a function.
The patch adds the ability to use CallDescriptions in syntactic contexts as
well. While we already have that in Signature, we still want to leverage the
ability to use dynamic information when we have it (function pointers, for
example). This could be done with Signature as well (StdLibraryFunctionsChecker
does it), but it makes it even less of a drop-in replacement.
Differential Revision: https://reviews.llvm.org/D119004
Add LLVM_LIBC_CLANG_TIDY option and allow LLVM_LIBC_ENABLE_LINTING without full build.
Reviewed By: sivachandra
Differential Revision: https://reviews.llvm.org/D119180
If the types aren't legal, the expansions may get type legalized in a
different way preventing code sharing. If the type is legal, we will
share some instructions between the two expansions, but we will need an
extra register.
Since we don't appear to fold (neg (sub A, B)) if the sub has an
additional user, I think it makes sense not to expand NABS.
Reviewed By: RKSimon
Differential Revision: https://reviews.llvm.org/D120513
This transform can still be applied if there are more than two
phi inputs, as long as phi inputs with the same value are dominated
by the same idom edge.
There are two MMA patterns that have been added twice. This patch just removes
one set of petterns. Should not change the way MMA behaves.
Reviewed By: lei, #powerpc
Differential Revision: https://reviews.llvm.org/D120680
clang has support for lazy headers in module maps - if size and/or
modtime and provided in the cppmap file, headers are only resolved when
an include directive for a file with that size/modtime is encoutered.
Before this change, the lazy resolution was all-or-nothing per module.
That means as soon as even one file in that module potentially matched
an include, all lazy files in that module were resolved. With this
change, only files with matching size/modtime will be resolved.
The goal is to avoid unnecessary stat() calls on non-included files,
which is especially valuable on networked file systems, with higher
latency.
Differential Revision: https://reviews.llvm.org/D120569
Found by code inspection. I don't think it makes a difference with
current codegen, because if any source modifiers were present we
would have selected mad/fma instead of mac/fmac in the first place.
Differential Revision: https://reviews.llvm.org/D120709
Treat the icmp and sub symmetrically, and require that one of them
has one use, not the icmp in particular. This could be further
relaxed in the abs (but not nabs) case to not check one-use at
all.
FirstMI is only used to get the load/store operand and the machine
function. Pass the MF and register explicitly, so the helper can be used
to find rename registers for other instructions in the future.
This should no longer be necessary now that we canonicalize to
intrinsics. This may not be entirely NFC in practice if worklist
order gets inverted and we perform demanded bits simplification
of a select user before the select is canonicalized.
Instead, folks can use the equivalent variables provided by CMake
to set those. This removal aims to reduce complexity and potential
for confusion when setting the target triple for building the runtimes,
and make it correct when `CMAKE_OSX_ARCHITECTURES` is used (right now
both `-arch` and `--target=` will end up being passed, which is downright
incorrect).
Differential Revision: https://reviews.llvm.org/D112155
This commit reverts 5aaefa51 (and also partly 7f285f48e7 and b6d75682f9,
which were related to the original commit). As landed, 5aaefa51 had
unintended consequences on some downstream bots and didn't have proper
coverage upstream due to a few subtle things. Implementing this is
something we should do in libc++, however we'll first need to address
a few issues listed in https://reviews.llvm.org/D106124#3349710.
Differential Revision: https://reviews.llvm.org/D120683
The llvm.mlir.global operation accepts a region as initializer. This region
corresponds to an LLVM IR constant expression and therefore should not accept
operations with side effects. Add a corresponding verifier.
Reviewed By: wsmoses, bondhugula
Differential Revision: https://reviews.llvm.org/D120632
Breakpoint deletion in visual studio is currently implemented by
iterating over the breakpoints we want to delete, for each of which we
iterate over the complete set of breakpoints in the debugger instance
until we find the one we wish to delete. Ideally we would resolve this
by directly deleting each breakpoint by some ID rather than searching
through the full breakpoint list for them, but in the absence of such a
feature in VS we can instead invert the loop to improve performance.
This patch changes breakpoint deletion to iterate over the complete list
of breakpoints, deleting breakpoints that match the breakpoints we
expect to delete by checking set membership. This represents a
worst-case improvement from O(nm) to O(n), for 'm' breakpoints being
deleted out of 'n' total. In practise this is almost exactly 'm'-times
faster, as when we delete multiple breakpoints they are typically
adjacent in the full breakpoint list.
Differential Revision: https://reviews.llvm.org/D120658
It cannot match a `pure virtual function`. This patch fixes this behavior.
Reviewed By: aaron.ballman
Differential Revision: https://reviews.llvm.org/D116439
InstrRefBasedLDV allocates some big tables of ValueIDNum, to store live-in
and live-out block values in, that then get passed around as pointers
everywhere. This patch wraps the allocation in a std::unique_ptr, names
some types based on unique_ptr, and passes references to those around
instead. There's no functional change, but it makes it clearer to the
reader that references to these tables are borrowed rather than owned, and
we get some extra validity assertions too.
Differential Revision: https://reviews.llvm.org/D118774
The demangler doesn't understand 'aw' as an operator name. This adds
the necessary smarts -- you may use this as an operator functionname,
but not as an expression operator.
Reviewed By: ChuanqiXu
Differential Revision: https://reviews.llvm.org/D120143
We need to parse operator names in 3 places -- expressions, names &
fold expressions. Currently we have 3 separate pieces to do this, and a FIXME.
The operator name and expression parsing are implemented as
handwritten two-character nested switches, the fold expression is a
sequence of string comparisons.
This adds a new OperatorInfo class to encode the operator info
(encoding, kind, name), and has a table that it can binary search.
From that each of the above 3 uses are altered to use the new scheme.
Existing tests cover parsing operator encodings.
Reviewed By: ChuanqiXu
Differential Revision: https://reviews.llvm.org/D119467
