| Age | Commit message (Collapse) | Author |
|---|
|
SemaInit.cpp to a new place. NFC
This is a refactoring change for better code isolation and reuse, a
first step to extend it for assignments.
|
|
This patch covers
[CWG2811](https://cplusplus.github.io/CWG/issues/2811.html) "Clarify
"use" of main", basically adding a test for `-Wmain`, focusing on usages
of `main` in unevaluated contexts.
To my understanding, the diagnostic message is based on the wording, so
I updated it based on the new wording. I also replaces "ISO C++
requires" with a phrasing that explicitly says "extension".
|
|
This patch implements 2024-05-31 resolution of a tentatively ready issue
[CWG2877](https://cplusplus.github.io/CWG/issues/2877.html) "Type-only
lookup for using-enum-declarator", which supersedes earlier
[CWG2621](https://cplusplus.github.io/CWG/issues/2621.html) "Kind of
lookup for `using enum` declarations".
Now we perform type-only lookup (not to be confused with type-only
context) for `elaborated-enum-declarator`. This is the same kind of
lookup that elaborated type specifiers and base specifiers undergo.
I also found out (and fixed) that one of our existing tests claimed that
a dependent type can be used in `elaborated-enum-declarator`, but that's
not the case:
> The
[using-enum-declarator](http://eel.is/c++draft/enum.udecl#nt:using-enum-declarator)
shall designate a non-dependent type with a reachable
[enum-specifier](http://eel.is/c++draft/dcl.enum#nt:enum-specifier)[.](http://eel.is/c++draft/enum.udecl#1.sentence-2)
|
|
This PR covers the following Core issues:
[CWG204](https://cplusplus.github.io/CWG/issues/204.html) "Exported
class templates"
[CWG323](https://cplusplus.github.io/CWG/issues/323.html) "Where must
`export` appear?"
[CWG335](https://cplusplus.github.io/CWG/issues/335.html) "Allowing
`export` on template members of nontemplate classes"
[CWG820](https://cplusplus.github.io/CWG/issues/820.html) "Deprecation
of `export`"
I believe the list above is entirety of Core issues that are dedicated
solely to `export template`.
I believe we have two main points of view here, which command what this
PR should do:
1. (easy) Removal of `export template` was done as a defect report in
CWG820, and the rest are effectively superseded by it, because we apply
defect reports retroactively.
2. (harder) Those Core issues are testable individually, so we should
test them for the behavior Core wanted at the time.
This PR implements the first option, making our C++ DR status page
greener.
I think I can be persuaded to go with the second option, if reviewers
have strong preference for it.
|
|
To avoid a duplicate hash map lookup.
|
|
Any fp128 need to end up as libcall, as will f32->i128 and f64->i128.
f16 are a bit special as the maximum range of the result fits in a i17,
so can be shrank to an i64. Vector with i128/fp128 types are scalarized.
|
|
@kiranchandramohan mentioned
[here](https://github.com/llvm/llvm-project/pull/91582#discussion_r1606046605)
that LLVM IR tests should go in the Integration folder. He also
mentioned
[here](https://github.com/llvm/llvm-project/pull/91582#discussion_r1606684034)
that tests for `add-debug-info` pass should test that pass only. There
were some tests which were added before his comments so I have cleaned
them in this PR. The following changes were made.
1. Move LLVM IR tests to `Integration` folder.
2. Change tests from f90 to fir and only test changes done by
`add-debug-info` pass.
|
|
Not sure why Github precommit tests didn't show this. Do they not build
for all targets?
|
|
|
|
Syntacore SCR3 is a microcontroller-class processor core. Overview:
https://syntacore.com/products/scr3
This PR introduces two CPUs:
* 'syntacore-scr3-rv32' which is rv32imc
* 'syntacore-scr3-rv64' which is rv64imac
---------
Co-authored-by: Dmitrii Petrov <dmitrii.petrov@syntacore.com>
|
|
Fixes https://github.com/llvm/llvm-project/issues/96197.
A global alias should always point to a definition. Ifuncs are
definitions, so far so good. However an ifunc may be statically resolved
to a function that is declared but not defined in the translation unit.
With this patch we perform static resolution if:
* the resolvee is defined, else if
* none of the ifunc users is a global alias
|
|
Fixes:
llvm/lib/Target/AArch64/AArch64GenRegisterBankInfo.def:185:12: warning: suggest parentheses around ‘&&’ within ‘||’ [-Wparentheses]
assert(!Size.isScalable() || MinSize >= 128
~~~~~~~~~~~~~~
&& "Scalable vector types should have size of at least 128 bits");
|
|
|
|
|
|
before inlining" (2nd Try)
In the re-commit, just dropping the propagation of `writeonly` as that
is the only attribute that can play poorly with call slot optimization
(see issue: #95152 for more details).
Closes #95888
|
|
This doesn't seem to crash anymore.
|
|
We handle them like arrays but still need to differentiate between
array/vector/complex types when dealing with QualTypes.
|
|
This commit removes a `FIXME` in the code base that was in place because
of patterns that used the dialect conversion API incorrectly. Those
patterns have been fixed and the workaround is no longer needed.
|
|
This allows code with SVE intrinsics to be compiled with +sme,+nosve,
assuming the encompassing function is in the correct mode (see #93802)
|
|
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, `@outer_dyn_drop_inner_most_dim` is replaced with:
* `@contiguous_inner_most_dynamic_outer`
I am also adding a similar test for scalable vectors. In addition,
* `@drop_two_inner_most_dim` and
`@drop_two_inner_most_dim_scalable_inner_dim`,
are renamed as `@contiguous_inner_most` and
`@contiguous_inner_most_scalable_inner_dim`, respectively, to match
their counterpart for `xfer_read`.
NOTE: This PR is limited to tests for `vector.transfer_write`
This is a follow-up for: #94490, #94604, #94906
|
|
clangSerialization currently uses hash_combine/hash_value from
Hashing.h, which are not guaranteed to be deterministic.
Replace these uses with xxh3_64bits.
Pull Request: https://github.com/llvm/llvm-project/pull/96136
|
|
Move PassInstrumentationAnalysis into PassInstrumentation.h and stop
including it in PassManager.h (effectively inverting the direction of
the dependency).
Most places using PassManager are not interested in PassInstrumentation,
and we no longer have any uses of it in PassManager.h itself (only in
PassManagerImpl.h).
|
|
As far as I can tell, this pull request was not approved, and
did not go through an RFC on discourse.
This reverts commit 89881480030f48f83af668175b70a9798edca2fb.
This reverts commit 225d8fc8eb24fb797154c1ef6dcbe5ba033142da.
|
|
|
|
This fixes a test failure introduced in 9f71a6b.
|
|
|
|
DWARF-4 introduced DWARF discriminators but both, the classic and the
parallel DWARF linkers discarded them so far.
After applying this patch dsymutil, which uses the DWARF linkers,
correctly preserves discriminator information.
|
|
(#96277)
This change removes the hard coded opcode values and replaces them with
regexes so that the test can deal with different opcode numbers being
assigned and unbreak a ton of bots.
|
|
This patch teaches addValueData to honor Site for verification
purposes. It does not affect the profile data in any manner.
|
|
A previous commit implemented _inp/w/d and renaming that to
__inbyte/word/dword
These intrinsics were declared in the header but never implemented.
|
|
At the same time, rename `PostVisitCFG` to the more descriptive
`PostAnalysisCallbacks` (which emphasizes the fact that these callbacks
are run
after the dataflow analysis itself has converged).
Before this patch, it was only possible to run a callback on the state
_after_
the transfer function had been applied, but for many analyses, it's more
natural
to to check the state _before_ the transfer function has been applied,
because we
are usually checking the preconditions for some operation. Some checks
are
impossible to perform on the "after" state because we can no longer
check the
precondition; for example, the `++` / `--` operators on raw pointers
require the
operand to be nonnull, but after the transfer function for the operator
has been
applied, the original value of the pointer can no longer be accessed.
`UncheckedOptionalAccessModelTest` has been modified to run the
diagnosis
callback on the "before" state. In this particular case, diagnosis can
be run
unchanged on either the "before" or "after" state, but we want this test
to
demonstrate that running diagnosis on the "before" state is usually the
preferred approach.
This change is backwards-compatible; all existing analyses will continue
to run
the callback on the "after" state.
|
|
Add missing flags in cmake.
Follow-up to cae6d458a.
|
|
Otherwise hip-phases.hip could fail when llvm::hash_value(StringRef)
changes.
|
|
"shortest" would be an empty sequence and the rule would never match.
|
|
Change adds parsing support for omp loop, omp target loop, omp target
parallel loop and omp target teams loop.
|
|
Currently, on different platform, the behaivor of llvm.minnum is
different if one operand is sNaN:
When we compare sNaN vs NUM:
ARM/AArch64/PowerPC: follow the IEEE754-2008's minNUM: return qNaN.
RISC-V/Hexagon follow the IEEE754-2019's minimumNumber: return NUM. X86:
Returns NUM but not same with IEEE754-2019's minimumNumber as
+0.0 is not always greater than -0.0.
MIPS/LoongArch/Generic: return NUM.
LIBCALL: returns qNaN.
So, let's introduce llvm.minmumnum/llvm.maximumnum, which always follow
IEEE754-2019's minimumNumber/maximumNumber.
Half-fix: #93033
|
|
When `!isInt<32>(x.va)`, `addq x@GOTPCREL(%rip), %rax` (possibly due to
getelementptr of a global variable) should not be optimized to
`addq offset, %rax`. We currently have either an assertion failure or an
incorrect optimization.
Fix it using the relaxOnce framework.
|
|
According to the spec, M implies Zmmul.
|
|
In no transitive change mode, the unresolved ADL may not
be referenced, then it is problematic if the change didn't get
propagated correctly. This patch add references to all the possible
module files for the unresolved ADL to solve the problem.
|
|
https://godbolt.org/z/G1K8Wszn9 for reference.
Starting with MSVC 1920+, VS2019, C++17 auto NTTP now adds `M <type>` to
the mangled name to avoid name collisions with different deduced types.
This PR fixes pointers. Pointers to members will be fixed in an upcoming
PR.
Here is a small example. The godbolt has more thorough examples.
```
template<auto>
struct AutoParmTemplate
{
AutoParmTemplate() {}
};
int i;
int main()
{
// MSVC 1916: ??0?$AutoParmTemplate@$1?i@@3HA@@QEAA@XZ
// MSVC 1929: ??0?$AutoParmTemplate@$MPEAH1?i@@3HA@@QEAA@XZ
// Clang: ??0?$AutoParmTemplate@$1?i@@3HA@@QEAA@XZ
AutoParmTemplate<&i> x;
}
```
|
|
On PPC, when there is dynamic alloca, only r1 points to the backchain.
|
|
|
|
|
|
(#96188)
In `computeExitLimit()`, we use `getExitingBlock()` to check if loop has
exactly one exiting block.
Since `computeExitLimit()` is only used in
`computeBackedgeTakenCount()`, and `getExitingBlocks()` is called to get
all exiting blocks in `computeBackedgeTakenCount()`, we can simply check
if loop has exactly one exiting block by checking if the number of
exiting blocks equals 1 in `computeBackedgeTakenCount()` and pass it as
an argument to `computeExitLimit()`.
This change helps to improve the compile time for files containing large
loops.
|
|
Reverts llvm/llvm-project#94575 since introduces test failure:
https://green.lab.llvm.org/job/llvm.org/view/LLDB/job/as-lldb-cmake/6166/
|
|
Following of https://github.com/llvm/llvm-project/pull/92085.
#### motivation
The motivation is still cutting of the unnecessary change in the
dependency chain. See the above link (recursively) for details.
And this will be the last patch of the `no-transitive-*-change` series.
If there are any following patches, they might be C++20 Named modules
specific to handle special grammars like `ADL` (See the reply in
https://discourse.llvm.org/t/rfc-c-20-modules-introduce-thin-bmi-and-decls-hash/74755/53
for example). So they won't affect the whole serialization part as the
series patch did.
#### example
After this patch, finally we are able to cut of unnecessary change of
types. For example,
```
//--- m-partA.cppm
export module m:partA;
//--- m-partA.v1.cppm
export module m:partA;
namespace NS {
class A {
public:
int getValue() {
return 43;
}
};
}
//--- m-partB.cppm
export module m:partB;
export inline int getB() {
return 430;
}
//--- m.cppm
export module m;
export import :partA;
export import :partB;
//--- useBOnly.cppm
export module useBOnly;
import m;
export inline int get() {
return getB();
}
```
The BMI of `useBOnly.cppm` is expected to not change if we only add a
new class in `m:partA`. This will be pretty useful in practice.
#### implementation details
The key idea of this patch is similar with the previous patches: extend
the 32bits type ID to 64bits so that we can store the module file index
in the higher bits. Then the encoding of the type ID is independent on
the imported modules.
But there are two differences from the previous patches:
- TypeID is not completely an index of serialized types. We used the
lower 3 bits to store the qualifiers.
- TypeID won't take part in any lookup process. So the uses of TypeID is
much less than the previous patches.
The first difference make we have some more slightly complex bit
operations. And the second difference makes the patch much simpler than
the previous ones.
|
|
Tighten the reserve() to `Record.size() / 2` instead of `Record.size()`
in the HasProfile/HasRelBF cases. For the uncommon old profile format
cases we leave it as is, but those should be rare and not worth
optimizing.
This reduces peak memory during ThinLTO indexing by ~3% in one example.
Alternatively, we could make the branching for reserve more complex and
try to cover every case.
|
|
I've made this small cosmetic changes,
`s/CountersOffsetInBiasMode/FileOffsetToCounters/`
|
|
This patch makes interactive mode as the default when using the crashlog
command. It replaces the existing `-i|--interactive` flag with a new
`-m|--mode` option, that can either be `interactive` or `batch`.
By default, when the option is not explicitely set by the user, the
interactive mode is selected, however, lldb will fallback to batch mode
if the command interpreter is not interactive or if stdout is not a tty.
This also adds some railguards to prevent users from using interactive
only options with the batch mode and updates the tests accordingly.
rdar://97801509
Differential Revision: https://reviews.llvm.org/D141658
Signed-off-by: Med Ismail Bennani <ismail@bennani.ma>
|
|
Rather than propgating a compile define, add an explicit cmake flag for
controlling the size. The default for baremetal is 100KB and the default
for others is 1GB.
|
