We do the same for the analogous transform in DAGCombine, but this case
was missed in the recent patch which added support for zext nneg.
Sorry for the lack of test coverage. Not sure how to exercise this piece
of logic. It appears to have only minimal impact on LIT tests (only
test/CodeGen/X86/wide-scalar-shift-by-byte-multiple-legalization.ll),
and even then, the changes without it appear uninteresting. Maybe we
should remove this transform instead?
When merging blocks, if the previous block has no any branch instruction
and has one successor, the successor may be SEH landing pad and the
block will always raise exception and nerver fall through to next block.
We can not merge them in such case. isSuccessor should be used to
confirm it can fall through to next block.
Replacing a free extension with 2 or more extensions unnecessarily
increases the number of IR instructions without providing any benefits.
It also unnecessarily causes operations to be performed on wider types
than necessary.
In some cases, the extra extensions also pessimize codegen (see
bfis-in-loop.ll).
The changes in arm64-codegen-prepare-extload.ll also show that we avoid
promotions that should only be performed in stress mode.
PR: https://github.com/llvm/llvm-project/pull/77094
Without this gcc warned
../lib/CodeGen/AsmPrinter/DwarfDebug.cpp:3585:70: warning: suggest parentheses around '&&' within '||' [-Wparentheses]
3584 | ((&Current == &AccelDebugNames) &&
| ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
3585 | (Unit.getUnitDie().getTag() != dwarf::DW_TAG_type_unit)) &&
| ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~^~
3586 | "Kind is CU but TU is being processed.");
| ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
../lib/CodeGen/AsmPrinter/DwarfDebug.cpp:3589:70: warning: suggest parentheses around '&&' within '||' [-Wparentheses]
3588 | ((&Current == &AccelTypeUnitsDebugNames) &&
| ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
3589 | (Unit.getUnitDie().getTag() == dwarf::DW_TAG_type_unit)) &&
| ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~^~
3590 | "Kind is TU but CU is being processed.");
| ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Relying on ComputeKnownBits to find a splat is causing miscompilations where a shift of zero is being assumed to give zero, but further simplification leads to a shift of zero by undef, resulting in an unexpected undef value.
Fixes#78109
Implement PhiLoweringHelper for GlobalISel in DivergenceLoweringHelper.
Use machine uniformity analysis to find divergent i1 phis and select
them as lane mask phis in same way SILowerI1Copies select VReg_1 phis.
Note that divergent i1 phis include phis created by LCSSA and all cases
of uses outside of cycle are actually covered by "lowering LCSSA phis".
GlobalISel lane masks are registers with sgpr register class and S1 LLT.
TODO: General goal is that instructions created in this pass are fully
instruction-selected so that selection of lane mask phis is not split
across multiple passes.
patch 3 from: https://github.com/llvm/llvm-project/pull/73337
Add a debug counter that allows forcing an sdag fallback after a certain
number of functions.
The intended use-case is to bisect which function gets miscompiled by
global isel using `-debug-counter=globalisel-count=N` (in cases where
sdag doesn't also miscompile it, of course).
The "falling back" debug line is printed unconditionally, because using
`-debug-only` is usually too spammy for the intended purpose.
…tructions into account
Hint instructions like G_ASSERT_ZEXT cann be viewed as a copy. Including
this fact into the combiner allows the match more patterns involving
such instructions.
Return true iff all of vector elements are constant AND not zero
Fixes#77805
Previously, it'd return `true` (as in - the value is known to be never
zero) for any build_vector/splat_vector with non-constant elements.
Fix a crash of `replace-with-veclib` pass, when the arguments of the TLI
mapping do not match the original call.
Now, it simply ignores such cases.
Test require assertions as it accesses programmatically the debug log.
Reapplies reverted PR #77112
Reordering based on the sort order of the MemOpInfo array was disabled
in <https://reviews.llvm.org/D72706>. However, it's not clear this is
desirable for al targets. It also makes it more difficult to compare the
incremental benefit of enabling load clustering in the selectiondag
scheduler as well was the machinescheduler, as the sdag scheduler does
seem to allow this reordering.
This patch adds a parameter that can control the behaviour on a
per-target basis.
Split out from #73789.
Moved extractParts() and extractVectorParts() from LegalizerHelper
to Utils to be able to use it in different passes.
extractParts() will also try to use unmerge when doing irregular
splits where possible, falling back to extract elements when not.
Fix a crash of `replace-with-veclib` pass, when the arguments of the TLI
mapping do not match the original call.
Now, it simply ignores such cases.
Test require assertions as it accesses programmatically the debug log.
Bug 1 is triggered when a TU is already created, and we process the same
DICompositeType at a top level. We would switch to TU accelerator table,
but
would not switch back on early exit. As the result we would add CU
entries to the TU
accelerator table. When we try to write out TUs and normalize entries,
the
offsets for DIEs that are part of a CU would not have been computed, and
it
would assert on getOffset().
Bug 2 is triggered when processing nested TUs. When we exit from
addDwarfTypeUnitType we switched back to CU accelerator table. If we
were processing nested TUs, the rest of the entries from TUs would be
added to CU accelerator table. When we write out TUs, all the DIE
pointers will become invalid. Eventually it will assert during
normalization step after CU is processed.
Force live interval recomputation for a register if its definition is
narrowed to become partial. The live interval repair process cannot
otherwise detect these changes.
Since we already know which register we want to extend, we don't have to
ask its defining MI about it
---------
Co-authored-by: Emil Tywoniak <Emil.Tywoniak@hightec-rt.com>
- Previously, 'assignCustomValue' requests the number of assigned VAs
minus 1 is returned and treats 0 as the assignment failure. However,
under that arrangment, we cannot tell a successful *single* VA custom
assignment from the failure case.
- This change requests that 'assignCustomValue' just return the number
of all VAs assigned, including the first WA so that it won't be ambigous
to tell the failure case from the single VA custom assignment.
We record the usage of each `Predicate` and sort them by usage.
For the top 8 `Predicate`s, we will emit a `PC_CheckPredicateN` to
save one byte.
Overall this reduces the llc binary size with all in-tree targets by
about 61K.
This is a recommit of 1a57927, which was reverted in bc98c31.
The CI failures occurred when doing expensive checks (with option
`LLVM_ENABLE_EXPENSIVE_CHECKS` being ON).
The key point here is that we need stable sorting result in the
test, but doing expensive checks uncovered the non-determinism of
`llvm::sort`. So `llvm::sort` is changed to `llvm::stable_sort`
in this revised patch.
And we use `llvm::MapVector` to keep insertion order.
This was planned since its introduction, but wasn't rolled out for a
little bit longer than intended (ahem...8 years).
All in-tree targets have now been adjusted to call
setMaxAtomicSizeInBitsSupported explicitly where required, so this
should be a no-op. The docs in docs/Atomics.rst already claimed the
default was 0, so that doesn't need updating.
- [DebugMetadata][DwarfDebug] Support function-local types in lexical
block scopes (4/7)
- [CloneFunction][DebugInfo] Avoid cloning DILocalVariables of inlined
functions
This is a follow-up for https://reviews.llvm.org/D144006, fixing a crash
reported
in Chromium (https://reviews.llvm.org/D144006#4651955).
The first commit is added for convenience, as it has already been
accepted.
If DISubpogram was not cloned (e.g. we are cloning a function that has
other
functions inlined into it, and subprograms of the inlined functions are
not supposed to be cloned), it doesn't make sense to clone its
DILocalVariables as well.
Otherwise get duplicated DILocalVariables not tracked in their
subprogram's retainedNodes, that crash LTO with Chromium.
This is meant to be committed along with
https://reviews.llvm.org/D144006.
When removing an empty machine basic block, all of its successors should
be inherited by its fall through MBB. This keeps CFG as only have one
entry which is required by LiveDebugValues.
Reland #77441 as LiveDebugValues test.
We previously had a heuristic that if a value V was used multiple times
in a single PHI, then to avoid potentially rematerializing into many predecessors
we bail out. The phi uses only counted as a single use in the shouldLocalize() hook
because it counted the PHI as a single instruction use, not factoring in it may
have many incoming edges.
It turns out this heuristic is slightly too pessimistic, and allowing a small number
of these uses to be localized can improve code size due to shortening live ranges,
especially if those ranges span a call.
This change results in some improvements in size on CTMark -Os:
```
Program size.__text
before after diff
kimwitu++/kc 451676.00 451860.00 0.0%
mafft/pairlocalalign 241460.00 241540.00 0.0%
tramp3d-v4/tramp3d-v4 389216.00 389208.00 -0.0%
7zip/7zip-benchmark 587528.00 587464.00 -0.0%
Bullet/bullet 457424.00 457348.00 -0.0%
consumer-typeset/consumer-typeset 405472.00 405376.00 -0.0%
SPASS/SPASS 410288.00 410120.00 -0.0%
lencod/lencod 426396.00 426108.00 -0.1%
ClamAV/clamscan 380108.00 379756.00 -0.1%
sqlite3/sqlite3 283664.00 283372.00 -0.1%
Geomean difference -0.0%
```
I experimented with different variations and thresholds. Using 3 instead
of 2 resulted in a further 0.1% improvement on ClamAV but also regressed
sqlite3 by the same %.
Instcombine canonicalizes selects to floating point and integer minmax.
This and the dag combiner canonicalize to floating point minmax. None of
them canonicalizes to integer minmax. On Neoverse V2 basic integer
arithmetic and integer minmax have the same costs.
We record the usage of each `Predicate` and sort them by usage.
For the top 8 `Predicate`s, we will emit a `PC_CheckPredicateN` to
save one byte.
Overall this reduces the llc binary size with all in-tree targets by
about 61K.
We record the usage of each `PatternPredicate` and sort them by
usage.
For the top 8 `PatternPredicate`s, we will emit a
`OPC_CheckPatternPredicateN` to save one byte.
The old `OPC_CheckPatternPredicate2` is renamed to
`OPC_CheckPatternPredicateTwoByte`.
Overall this reduces the llc binary size with all in-tree targets by
about 93K.
