.altinstructions section contains a list of structures where fields can
have different sizes while other fields could be present or not
depending on the kernel version. Add automatic detection of such
variations and use it by default. The user can still overwrite the
automatic detection with `--alt-inst-has-padlen` and
`--alt-inst-feature-size` options.
`GetDeclContextDIEs` and `DIEDeclContextsMatch` are unused (possibly
since we added support for simplified template names, but I haven't
checked). `GetDeclContextDIEs` is also very similar (but subtly
different) from `GetDeclContext` and `GetTypeLookupContext`.
I am keeping `GetParentDeclContextDIE` as that one still has some
callers, but I want to look into the possibility of merging it with at
least one of the functions mentioned above.
Before this PR, clangd forcefully disabled misc-const-correctness in
disableUnusableChecks().
Now we have a FastCheckFilter configuration whose default value
(Strict) also disables it. This patch removes misc-const-correctness
from disableUnusableChecks() so it's possible to enable by setting
FastCheckFilter to None.
Fixes https://github.com/llvm/llvm-project/issues/89758
The character reduce runtime functions expect a pointer to a scalar
character of the correct length for the result of character reduce. A
descriptor was passed so far. Fix the lowering so a proper temporary is
created and passed to the runtime.
In some modules, e.g. Kotlin-generated IR, we end up with a huge RefSCC
and the call graph updates done as a result of the inliner take a long
time. This is due to RefSCC::removeInternalRefEdges() getting called
many times, each time removing one function from the RefSCC, but each
call to removeInternalRefEdges() is proportional to the size of the
RefSCC.
There are two places that call removeInternalRefEdges(), in
updateCGAndAnalysisManagerForPass() and
LazyCallGraph::removeDeadFunction().
1) Since LazyCallGraph can deal with spurious (edges that exist in the
graph but not in the IR) ref edges, we can simply not call
removeInternalRefEdges() in updateCGAndAnalysisManagerForPass().
2) LazyCallGraph::removeDeadFunction() still ends up taking the brunt of
compile time with the above change for the original reason. So instead
we batch all the dead function removals so we can call
removeInternalRefEdges() just once. This requires some changes to
callers of removeDeadFunction() to not actually erase the function from
the module, but defer it to when we batch delete dead functions at the
end of the CGSCC run, leaving the function body as "unreachable" in the
meantime. We still need to ensure that call edges are accurate. I had
also tried deleting dead functions after visiting a RefSCC, but deleting
them all at once at the end was simpler.
Many test changes are due to not performing unnecessary revisits of an
SCC (the CGSCC infrastructure deems ref edge refinements as unimportant
when it comes to revisiting SCCs, although that seems to not be
consistently true given these changes) because we don't remove some ref
edges. Specifically for devirt-invalidated.ll this seems to expose an
inlining order issue with the inliner. Probably unimportant for this
type of intentionally weird call graph.
Compile time:
https://llvm-compile-time-tracker.com/compare.php?from=6f2c61071c274a1b5e212e6ad4114641ec7c7fc3&to=b08c90d05e290dd065755ea776ceaf1420680224&stat=instructions:u
Fragments are allocated with `operator new` and stored in an ilist with
Prev/Next/Parent pointers. A more efficient representation would be an
array of fragments without the overhead of Prev/Next pointers.
As the first step, replace ilist with singly-linked lists.
* `getPrevNode` uses have been eliminated by previous changes.
* The last use of the `Prev` pointer remains: for each subsection, there is an insertion point and
the current insertion point is stored at `CurInsertionPoint`.
* `HexagonAsmBackend::finishLayout` needs a backward iterator. Save all
fragments within `Frags`. Hexagon programs are usually small, and the
performance does not matter that much.
To eliminate `Prev`, change the subsection representation to
singly-linked lists for subsections and a pointer to the active
singly-linked list. The fragments from all subsections will be chained
together at layout time.
Since fragment lists are disconnected before layout time, we can remove
`MCFragment::SubsectionNumber` (https://reviews.llvm.org/D69411). The
current implementation of `AttemptToFoldSymbolOffsetDifference` requires
future improvement for robustness.
Pull Request: https://github.com/llvm/llvm-project/pull/95077
For i64 this avoids loading a 64-bit value into register, for smaller registers this just avoids an immediate operand.
For i8+i16, limit to one use case as we save fewer bytes and these can be wasted entirely on extra register moves.
Fixes#67709
Following a rather direct approach to expose PDL usage from C and then
Python. This doesn't yes plumb through adding support for custom
matchers through this interface, so constrained to basics initially.
This also exposes greedy rewrite driver. Only way currently to define
patterns is via PDL (just to keep small). The creation of the PDL
pattern module could be improved to avoid folks potentially accessing
the module used to construct it post construction. No ergonomic work
done yet.
---------
Signed-off-by: Jacques Pienaar <jpienaar@google.com>
The wasm backend fetches the tan runtime lib call in
`llvm/include/llvm/IR/RuntimeLibcalls.def` via `StaticLibcallNameMap()`,
but ignores the runtime function because a function sinature mapping is
not specified in RuntimeLibcallSignatureTable(). The fix is to specify
the function signatures for float32-128.
This is a fix for a build break reported on PR
https://github.com/llvm/llvm-project/pull/94559#issuecomment-2159923215.
Handling parallel region RaW conflicts should usually be the
responsibility of the source program, rather than bufferization
analysis. However, to preserve current functionality, checks on parallel
regions is put behind a bufferization in this PR, which is on by
default. Default functionality will not change, but this PR enables the
option to leave parallelism checks out of the bufferization analysis.
This makes sure we try to process declaration DIEs that are erroneously
present in the index. Until bd5c6367bd, clang was emitting index
entries for declaration DIEs with DW_AT_signature attributes. This makes
sure to avoid returning those DIEs as the definitions of a type, but
also makes sure to pass through DIEs referring to static constexpr
member variables, which is a (probably nonconforming) extension used by
dsymutil.
It adds test cases for both of the scenarios. It is essentially a
recommit of #91808.
Also reverts "[MLIR][Flang][DebugInfo] Convert debug format in MLIR translators"
The patch above introduces behaviour controlled by an LLVM flag into the
Flang driver, which is incorrect behaviour.
This reverts commits:
3cc2710e0d.
460408f78b.
Prepare for new pass manager version of `MachineDominatorTreeAnalysis`.
We may need a machine dominator tree version of `DomTreeUpdater` to
handle `SplitCriticalEdge` in some CodeGen passes.
This is a printf style variadic function. If using a "%s" format, we
should pass "const char *" rather than "StringRef".
The use of data() here is safe because we know that the StringRef was
originally derived from a null-terminated string.
As we approach the state where support for debug intrinsics is dropping and
we print and use debug records by default, the documentation should be updated
to refer to debug records as the primary debug info representation, with
debug intrinsics being relegated to an optional alternative.
This patch performs a few updates:
- Replace references to intrinsics with references to records across all
the documentation.
- Replace intrinsics with records in code examples.
- Move debug records prior to debug intrinsics in the
SourceLevelDebugging document, and change text to refer to them as the
primary representation.
- Add release notes describing the change.
The main goal of this PR (and subsequent PRs), is to add more tests with
scalable vectors to:
* vector-transfer-collapse-inner-most-dims.mlir
There's quite a few cases to consider, hence this is split into multiple
PRs. In this PR, the very first test is complemented with all the
possible combinations:
* scalable (rather than fixed) unit trailing dim,
* dynamic (rather than static) trailing dim in the source memref.
Also,
* `@leading_scalable_dimension_transfer_read` and
`@trailing_scalable_one_dim_transfer_read`,
are replaced with:
* `@contiguous_inner_most_scalable_inner_dim` and
`@negative_scalable_unit_dim`,
respectively, and added to the list above (i.e. alongside other
variations for the very first test).
In addition:
* "_view" is removed from function names (it's not clear to me what it
was meant to signify)
* extra comments are added to separate tests for vector.transfer_read
and vector.transfer_write
NOTE: This PR is limited to tests for `vector.transfer_read`.
Reapplies the original patch with some additional conversion layers added
to the MLIR translator, to ensure that we don't write the new debug info
format unless WriteNewDbgInfoFormat is set.
This reverts commit 8c5d9c79b9.
Those BitVectors get expensive on targets like AMDGPU with thousands of
registers, and RegAliasIterator is also expensive.
We can move all liveness calculations to use RegUnits instead to speed
it up for targets where RegAliasIterator is expensive, like AMDGPU.
On targets where RegAliasIterator is cheap, this alternative can be a little more expensive, but I believe the tradeoff is worth it.
This commit reimplements the functionality of the Clang Static Analyzer
checker `alpha.core.SizeofPointer` within clang-tidy by adding a new
(off-by-default) option to bugprone-sizeof-expression which activates
reporting all the `sizeof(ptr)` expressions (where ptr is an expression
that produces a pointer).
The main motivation for this change is that `alpha.core.SizeofPointer`
was an AST-based checker, which did not rely on the path sensitive
capabilities of the Static Analyzer, so there was no reason to keep it
in the Static Analyzer instead of the more lightweight clang-tidy.
After this commit I'm planning to create a separate commit that deletes
`alpha.core.SizeofPointer` from Clang Static Analyzer.
It was natural to place this moved logic in bugprone-sizeof-expression,
because that check already provided several heuristics that reported
various especially suspicious classes of `sizeof(ptr)` expressions.
The new mode `WarnOnSizeOfPointer` is off-by-default, so it won't
surprise the existing users; but it can provide a more through coverage
for the vulnerability CWE-467 ("Use of sizeof() on a Pointer Type") than
the existing partial heuristics.
Previously this checker had an exception that the RHS of a
`sizeof(array) / sizeof(array[0])` expression is not reported; I
generalized this to an exception that the check doesn't report
`sizeof(expr[0])` and `sizeof(*expr)`. This idea is taken from the
Static Analyzer checker `alpha.core.SizeofPointer` (which had an
exception for `*expr`), but analysis of open source projects confirmed
that this indeed eliminates lots of unwanted results.
Note that the suppression of `sizeof(expr[0])` and `sizeof(*expr)`
reports also affects the "old" mode `WarnOnSizeOfPointerToAggregate`
which is enabled by default.
This commit also replaces the old message "suspicious usage of
'sizeof(A*)'; pointer to aggregate" with two more concrete messages; but
I feel that this tidy check would deserve a through cleanup of all the
diagnostic messages that it can produce. (I added a FIXME to mark one
outright misleading message.)
MLIR's LLVM dialect does not internally support debug records, only
converting to/from debug intrinsics. To smooth the transition from
intrinsics to records, there is a step prior to IR->MLIR translation
that switches the IR module to intrinsic-form; this patch adds the
equivalent conversion to record-form at MLIR->IR translation, and also
modifies the flang front end to use the WriteNewDbgInfoFormat flag when
it is emitting LLVM IR.
The case-list of the switches generated by this pass were not
"deterministic" (based on allocation patterns).
This is because the CaseList order relied on an unordered_set order.
Using the sorted exit target list for those should solve the problem.
Fixes#94961
Signed-off-by: Nathan Gauër <brioche@google.com>
There were four tests in Driver that actually tested bits of Driver and
bits of CodeGen, and therefore had target restrictions. Rework those
four tests into one Driver test (with no target restrictions) and two
target-specific CodeGen tests.
The current logic tests for inf/inf and 0/0 inputs using a NaN check.
This doesn't work with all fastmath flags. With nnan and ninf, we can
just check for a 0 maximum. With only nnan, we have to check for both
cases separately.
Function `getNeutralElement` already indicates "cannot find neutral
element" by returning nullptr through the return value, and no
additional error log needs to be output.
