At the moment, proveNoUnsignedWrapViaInduction may be called for the
same AddRec a large number of times via getZeroExtendExpr. This can have
a severe compile-time impact for very loop-heavy code. One one
particular workload, LSR takes ~51s without this patch, almost
exlusively in proveNoUnsignedWrapViaInduction. With this patch, the time
in LSR drops to ~0.4s.
If proveNoUnsignedWrapViaInduction failed to prove NUW the first time,
it is unlikely to succeed on subsequent tries and the cost doesn't seem
to be justified.
Besides drastically improving compile-time in some excessive cases, this
also has a slightly positive compile-time impact on CTMark:
NewPM-O3: -0.07%
NewPM-ReleaseThinLTO: -0.08%
NewPM-ReleaseLTO-g: -0.06
https://llvm-compile-time-tracker.com/compare.php?from=b435da027d7774c24cdb8c88d09f6b771e07fb14&to=f2729e33e8284b502f6c35a43345272252f35d12&stat=instructions
Reviewed By: nikic
Differential Revision: https://reviews.llvm.org/D130648
The solution is to favor the longest possible nest-name-specifier, and
drop other alternatives by using the guard, per per C++ [basic.lookup.qual.general].
Motivated cases:
```
Foo::Foo() {};
// the constructor can be parsed as:
// - Foo ::Foo(); // where the first Foo is return-type, and ::Foo is the function declarator
// + Foo::Foo(); // where Foo::Foo is the function declarator
```
```
void test() {
// a very slow parsing case when there are many qualifers!
X::Y::Z;
// The statement can be parsed as:
// - X ::Y::Z; // ::Y::Z is the declarator
// - X::Y ::Z; // ::Z is the declarator
// + X::Y::Z; // a declaration without declarator (X::Y::Z is decl-specifier-seq)
// + X::Y::Z; // a qualifed-id expression
}
```
Differential Revision: https://reviews.llvm.org/D130511
This is similar to the LLVM_TABLEGEN, CLANG_TABLEGEN and
CLANG_PSEUDO_GEN cmake cache variables.
Differential Revision: https://reviews.llvm.org/D129799
Rename it to clang-tidy-confusable-chars-gen, to make its role
clearer in a wider context.
In cross builds, the caller might want to provide this tool
externally (to avoid needing to rebuild it in the cross build).
In such a case, having the tool properly namespaced makes its role
clearer.
This matches how the clang-pseudo-gen tool was renamed in
a43fef05d4 / D126725.
Differential Revision: https://reviews.llvm.org/D129798
New exp2 function algorithm:
1) Improved performance: 8.176 vs 15.270 by core-math perf tool.
2) Improved accuracy. Only two special values left.
3) Lookup table size reduced twice.
Differential Revision: https://reviews.llvm.org/D129005
DebugLocEntry assumes that it either contains 1 item that has no fragment
or many items that all have fragments (see the assert in addValues).
When EXPENSIVE_CHECKS is enabled, _GLIBCXX_DEBUG is defined. On a few machines
I've checked, this causes std::sort to call the comparator even
if there is only 1 item to sort. Perhaps to check that it is implemented
properly ordering wise, I didn't find out exactly why.
operator< for a DbgValueLoc will crash if this happens because the
optional Fragment is empty.
Compiler/linker/optimisation level seems to make this happen
or not. So I've seen this happen on x86 Ubuntu but the buildbot
for release EXPENSIVE_CHECKS did not have this issue.
Add an explicit check whether we have 1 item.
Reviewed By: aprantl
Differential Revision: https://reviews.llvm.org/D130156
This hint instructs the linker to optimize an adrp+add+ldr sequence used
for loading from a local symbol's address by loading directly if it's
close enough, or with an adrp(p)+ldr sequence if it's not.
This transformation is the same as what's done for ADRP_LDR_GOT_LDR when
the symbol is local. The logic for acting on this hint is therefore
moved to a new function which will be called from the existing
applyAdrpLdrGotLdr() function.
Differential Revision: https://reviews.llvm.org/D130505
In DWARF5, the `DW_AT_name` and `DW_AT_comp_dir` attributes are encoded
using the `strx*` forms, which specify an index into `__debug_str_offs`.
This commit adds that section to DwarfObject, so the debug info parser
can resolve these references.
The test case was manually adapted from stabs-icf.s.
Fixes#51668
Differential Revision: https://reviews.llvm.org/D130559
This commit folds a `tensor.cast` op into a `tensor.collapse_shape` op
when following two conditions meet:
1. the `tensor.collapse_shape` op consumes result of the `tensor.cast` op.
2. `tensor.cast` op casts to a more dynamic version of the source tensor.
This is added as a canonicalization pattern in `tensor.collapse_shape` op.
Signed-Off-By: Gaurav Shukla <gaurav@nod-labs.com>
Reviewed By: mravishankar
Differential Revision: https://reviews.llvm.org/D130650
As test in PR56672 shows, LAA produces different results which lead to either
positive or negative vectorization decisions depending on the order of blocks
in loop. The exact reason of this is not clear to me, however this makes investigation
of related bugs extremely complex.
Current order of blocks in the loop is arbitrary. It may change, for example, if loop
info analysis is dropped and recomputed. Seems that it interferes with LAA's logic.
This patch chooses fixed traversal order of blocks in loops, making it RPOT.
Note: this is *not* a fix for bug with incorrect analysis result. It just makes
the answer more robust to make the investigation easier.
Differential Revision: https://reviews.llvm.org/D130482
Reviewed By: aeubanks, fhahn
We will insert a new operand which is identical to the Dest for complex
FMUL with a mask. https://godbolt.org/z/eTEdnYv3q
Complex FMA and FMUL with maskz don't have this problem.
Reviewed By: LuoYuanke, skan
Differential Revision: https://reviews.llvm.org/D130638
By not clustering loads and adjusting heuristics to more aggressively reduce
register pressure we may be able to increase occupancy for the function if it
was dropped in a first pass scheduling.
Similarly, try to reduce spilling if register usage exceeds lower bound
occupancy.
Reviewed By: rampitec
Differential Revision: https://reviews.llvm.org/D130329
In diff and diffstat modes, the return code is != 0 even when there are no
changes between commits. This issue can be fixed by passing --exit-code to
git-diff command that returns 0 when there are no changes and using that as
the return code for git-clang-format.
Fixes#56736.
Differential Revision: https://reviews.llvm.org/D129311
The checkout action will hard-code the default github actions token in
the git config so that all pushes use it. We need to set
persist-credentials=false so we can use a token that has permission
to push to the llvm-project-release-prs repo.
Clear all kill flags on source register when folding a COPY.
This is necessary because the kills may now be out of order with the uses.
Reviewed By: foad
Differential Revision: https://reviews.llvm.org/D130622
* https://discourse.llvm.org/t/rfc-removing-the-quant-dialect/3643/8
* Removes most ops. Leaves casts given final comment (can remove more in a followup).
* There are a few uses in Tosa keeping some of the utilities alive. In a followup, I will probably elect to just move simplified versions of them into Tosa itself vs having this quasi-library dependency.
Differential Revision: https://reviews.llvm.org/D120204
InstCombine and DAGCombine prefer to keep shl before binops.
This patch teaches isel to convert to (shl (and/or/xor X, C1 >> C2), C2)
if (C1 >> C2) is a simm12. The idea was taken from X86's isel code.
There's a special case implemented for a sext_inreg between the
shift and the binop.
Differential Revision: https://reviews.llvm.org/D130610
Vtables will be emitted in fewer places than ctors (every ctor
references the vtable, so at worst it's the same places - but at best
the type has a non-inline key function and the vtable is emitted in one
place)
Pulling this fix out of 517bbc64db which
was reverted in 4821508d4d
This commit extends UnifyAliasedResourcePass to handle the case
where aliased resources have different vector sizes. (It still
requires all scalar types to be of the same bitwidth.) This is
effectively reusing the code for handling different-bitwidth
scalar types.
Reviewed By: ThomasRaoux
Differential Revision: https://reviews.llvm.org/D130671
This commit fixes spv.CompositeConstruct to assembly to list
operand types to enable vector construction out of smaller vectors.
Validation is also fixed to properly check the cases for vector
construction.
Reviewed By: ThomasRaoux
Differential Revision: https://reviews.llvm.org/D130669
tryLastChanceRecoloring iterates over the set of LiveInterval pointers
and used that to seed the recoloring stack, which was
nondeterministic. Fixes a future test failing about 20% of the time.
This just takes the order the interfering vreg was encountered. Not
sure if we should try to order this more intelligently.
Xcode 14 no longer puts the Rosetta expanded shared cache in a directory
named "16.0". Instead, it includes the real version number (e.g. 13.0),
the build string and the architecture, similar to the device support
directory names for iOS, tvOS and watchOS.
Currently, when there are multiple directories, we might end up picking
the wrong one in GetSDKDirectoryForCurrentOSVersion. The problem is that
without the build string we have no way to differentiate between
multiple directories with the same version number. This patch fixes the
problem by using GetOSBuildString which, as the name implies, returns
the build string if known.
This also adds a test for Rosetta debugging on Apple Silicon. Depending
on whether the Rosetta expanded shared cache is present, the test
ensures that there is or isn't a diagnostic about reading out of memory.
rdar://97576121
Differential revision: https://reviews.llvm.org/D130540
I noticed that the test TestSetWatchpoint.py was failing every so often
on macOS. The failure was in the last assert, that after destroying the
SBTarget containing it, the SBWatchpoint was still saying it was valid.
IsValid in this case just meant the watchpoint weak pointer could be turned
into a shared pointer. The watchpoint shared pointers have two strong references
in general, one to the "Target::m_last_created_watchpoint", and one in the
Target::m_watchpoint_list. Target::Destroy reset the last created watchpoint
but neglected to call RemoveAll on the watchpoint list (it does the analogous
work for the internal & external breakpoint lists...) This patch does the
equivalent cleanup for the watchpoint list.
DR2338 clarified that it was undefined behavior to set the value outside the
range of the enumerations values for an enum without a fixed underlying type.
We should diagnose this with a constant expression context.
Differential Revision: https://reviews.llvm.org/D130058
Florian Hahn