platform/upstream/llvm.git
20 months ago[Fuchsia] Add other necessary components to LLDB install.
Daniel Thornburgh [Thu, 2 Mar 2023 19:53:09 +0000 (11:53 -0800)]
[Fuchsia] Add other necessary components to LLDB install.

Reviewed By: phosek

Differential Revision: https://reviews.llvm.org/D145187

20 months ago[Nomination] Adding Nvidia Compiler security representative to the LLVM security...
Ragavan Dasarathan [Fri, 3 Mar 2023 18:01:31 +0000 (10:01 -0800)]
[Nomination] Adding Nvidia Compiler security representative to the LLVM security group.

I'm from Nvidia Corporation. I work primarily on the Security aspect of Compiler toolchains based out of LLVM. I would like to nominate myself to join the LLVM security group as a vendor contact representative of Nvidia.

According to the information on the website, https://llvm.org/docs/Security.html, I fall under the following category: Vendor contacts.

We are particularly interested in the following:
  - Being aware of any security vulnerability that has been found in the compiler in a timely manner.
  - Correctly reporting any vulnerabilities or other security issues we have found in the compiler stack.
  - Discussing our use of static analysis / dynamic analysis / fuzzing / threat modelling with the committee and raising any concerns that have risen from said activities.

Looking forward to participating in further discussions and security sync ups with the rest of the committee
Please let me know if I need to provide any other information required for this nomination.
Thanks,
Ragavan

Reviewed By: ab, apilipenko, george.burgess.iv, kristof.beyls, mattdr, nikhgupt, probinson, peter.smith, pietroalbini

Differential Revision: https://reviews.llvm.org/D144585

20 months ago[lldb][TypeSystemClang] Use the CXXFunctionPointerSummaryProvider for member-function...
Michael Buch [Fri, 3 Mar 2023 14:18:21 +0000 (14:18 +0000)]
[lldb][TypeSystemClang] Use the CXXFunctionPointerSummaryProvider for member-function pointers

With this patch member-function pointers are formatted using
`CXXFunctionPointerSummaryProvider`.

This turns,
```
(lldb) v pointer_to_member_func
(void (Foo::*)()) ::pointer_to_member_func = 0x00000000000000000000000100003f94
```
into
```
(lldb) v pointer_to_member_func
(void (Foo::*)()) ::pointer_to_member_func = 0x00000000000000000000000100003f94 (a.out`Foo::member_func() at main.cpp:3)
```

Differential Revision: https://reviews.llvm.org/D145242

20 months ago[lldb][TypeSystemClang] Format pointers to member functions as eFormatHex
Michael Buch [Fri, 3 Mar 2023 14:21:12 +0000 (14:21 +0000)]
[lldb][TypeSystemClang] Format pointers to member functions as eFormatHex

Before this patch, LLDB used to format pointers to members, such as,
```
void (Foo::*pointer_to_member_func)() = &Foo::member_func;
```
as `eFormatBytes`. E.g.,
```
(lldb) v pointer_to_member_func
(void (Foo::*)()) $1 = 94 3f 00 00 01 00 00 00 00 00 00 00 00 00 00 00
```

This patch makes sure we format pointers to member functions the same
way we do regular function pointers.

After this patch we format member pointers as:
```
(lldb) v pointer_to_member_func
(void (Foo::*)()) ::pointer_to_member_func = 0x00000000000000000000000100003f94
```

Differential Revision: https://reviews.llvm.org/D145241

20 months ago[lldb][TypeSystemClang][NFC] Factor out l/r-value reference logic for IsXXXType APIs
Michael Buch [Fri, 3 Mar 2023 14:04:05 +0000 (14:04 +0000)]
[lldb][TypeSystemClang][NFC] Factor out l/r-value reference logic for IsXXXType APIs

This will be useful as we add more `IsXXXType` APIs for different
function types.

Differential Revision: https://reviews.llvm.org/D145240

20 months ago[lldb] Build fix: variable name typo
Michael Buch [Fri, 3 Mar 2023 17:41:30 +0000 (17:41 +0000)]
[lldb] Build fix: variable name typo

Build failure introduced in `de10c1a824405833a0f49b22e7fa3f32a1393cc3`

20 months ago[lldb] Ignore libcxx std::ranges global variables in frame var
Michael Buch [Fri, 3 Mar 2023 11:28:32 +0000 (11:28 +0000)]
[lldb] Ignore libcxx std::ranges global variables in frame var

The motivation is to avoid cluttering LLDB's global variable view for
std::ranges users.

Before:
```
(lldb) frame var -g
...
(const std::ranges::__end::__fn) std::__1::ranges::__cpo::end = {}
(const std::ranges::views::__all::__fn) std::__1::ranges::views::__cpo::all = {}
(const std::ranges::__begin::__fn) std::__1::ranges::__cpo::begin = {}
(const std::ranges::views::__take::__fn) std::__1::ranges::views::__cpo::take = {}
(const std::ranges::__max_element::__fn) std::__1::ranges::__cpo::max_element = {}
(const std::ranges::__size::__fn) std::__1::ranges::__cpo::size = {}
(const std::ranges::__data::__fn) std::__1::ranges::__cpo::data = {}
```

After this patch none of these __cpo variables would show up.

Differential Revision: https://reviews.llvm.org/D145245

20 months ago[HLSL] add max/min library functions
Joshua Batista [Fri, 3 Mar 2023 00:21:01 +0000 (16:21 -0800)]
[HLSL] add max/min library functions

This change exposes the max and min library functions for HLSL, excluding long, and long long doubles.
The max / min functions are supported for all scalar, vector, and matrix types.
Long and long long double support is missing in this patch because those types
don't exist in HLSL.

The full documentation of the HLSL max / min functions are available here:
https://docs.microsoft.com/en-us/windows/win32/direct3dhlsl/dx-graphics-hlsl-max
https://docs.microsoft.com/en-us/windows/win32/direct3dhlsl/dx-graphics-hlsl-min

Reviewed By: python3kgae

Differential Revision: https://reviews.llvm.org/D144309

20 months agoFix bots by adding a triple to the test
Aaron Ballman [Fri, 3 Mar 2023 17:26:04 +0000 (12:26 -0500)]
Fix bots by adding a triple to the test

Resolves the issue found by:
https://lab.llvm.org/buildbot/#/builders/245/builds/5384

20 months ago[BOLT] Remove dependency on StringMap iteration order
Maksim Panchenko [Wed, 1 Mar 2023 23:49:21 +0000 (15:49 -0800)]
[BOLT] Remove dependency on StringMap iteration order

Remove the usage of StringMap in places where the iteration order
affects the output since the iteration over StringMap is
non-deterministic.

Reviewed By: Amir

Differential Revision: https://reviews.llvm.org/D145194

20 months agoPrecommit tests for D145157.
Paul Walker [Fri, 3 Mar 2023 16:45:41 +0000 (16:45 +0000)]
Precommit tests for D145157.

20 months ago[lld-macho] Ignore -debug_variant flag
Keith Smiley [Thu, 2 Mar 2023 23:05:19 +0000 (15:05 -0800)]
[lld-macho] Ignore -debug_variant flag

As far as I can tell all this flag does in ld64 is warn about cases
where you're building dylibs that are going to be part of the shared
cache, which I assume only apple is doing. Strangely I have seen this
flag being passed in the wild so this silences the warning about it.

Differential Revision: https://reviews.llvm.org/D145200

20 months ago[mlir][vector] add support for printing f16 and bf16
Aart Bik [Fri, 3 Mar 2023 01:37:43 +0000 (17:37 -0800)]
[mlir][vector] add support for printing f16 and bf16

Love or hate it, but the vector.print operation was the very
first operation that actually made "end-to-end" CHECK integration
testing possible for MLIR. This revision adds support for
the -until recently- less common but important floating-point
types f16 and bf16.

This will become useful for accelerator specific testing (e.g. NVidia GPUs)

Reviewed By: wrengr

Differential Revision: https://reviews.llvm.org/D145207

20 months ago[C2x] Claim support for WG14 N2838
Aaron Ballman [Fri, 3 Mar 2023 16:56:22 +0000 (11:56 -0500)]
[C2x] Claim support for WG14 N2838

This paper clarifies that complete object types need to be smaller than
SIZE_MAX. We already conformed to that requirement, so this adds some
test coverage to prove it.

20 months ago[NFC][TargetParser] Remove Unused Includes
Archibald Elliott [Fri, 3 Feb 2023 14:48:05 +0000 (14:48 +0000)]
[NFC][TargetParser] Remove Unused Includes

20 months ago[NFC][ARM][AArch64] Cleanup TargetParser includes
Archibald Elliott [Fri, 3 Feb 2023 14:44:39 +0000 (14:44 +0000)]
[NFC][ARM][AArch64] Cleanup TargetParser includes

llvm/TargetParser/TargetParser.h now only includes AMDGPU-specific
functionality, the ARM- and AArch64-specific functionality is in other
headers.

20 months ago[Flang][OpenMP] Fix unused variable warning/error
Kiran Chandramohan [Fri, 3 Mar 2023 16:03:32 +0000 (16:03 +0000)]
[Flang][OpenMP] Fix unused variable warning/error

20 months ago[Flang][OpenMP] NFC: Refactor privatization code in Lowering 1/n
Kiran Chandramohan [Fri, 3 Mar 2023 13:57:41 +0000 (13:57 +0000)]
[Flang][OpenMP] NFC: Refactor privatization code in Lowering 1/n

The code for handing privatization code has grown in complexity.
This patch begins an effort to refactor and simplify the code.
The refactoring will hopefully will make it modular and
enable privatization to be invoked in different places. This
patch moves the privatization related code into a class called
dataSharingProcessor.

Reviewed By: DavidTruby

Differential Revision: https://reviews.llvm.org/D144766

20 months ago[LV] Allow scalarization of function calls when masking is required
Graham Hunter [Fri, 16 Sep 2022 14:23:18 +0000 (15:23 +0100)]
[LV] Allow scalarization of function calls when masking is required

This patch adds support for scalarizing calls to a function when
there is a vector variant that cannot be used, either because there
isn't a masked variant or because the cost model indicated a VF
without a masked variant was better.

Reviewed By: paulwalker-arm

Differential Revision: https://reviews.llvm.org/D134422

20 months ago[ReplaceConstant] Extract code for expanding users of constant (NFC)
Nikita Popov [Fri, 3 Mar 2023 15:07:43 +0000 (16:07 +0100)]
[ReplaceConstant] Extract code for expanding users of constant (NFC)

AMDGPU implements some handy code for expanding all constexpr
users of LDS globals. Extract the core logic into ReplaceConstant,
so that it can be reused elsewhere.

20 months ago[NFCI][AArch64] Use DAG Helper Functions
Archibald Elliott [Fri, 3 Mar 2023 13:17:00 +0000 (13:17 +0000)]
[NFCI][AArch64] Use DAG Helper Functions

This commit is a cleanup of a few things in the SelectionDAG
implementation:
- to have more places use `splitInt128`, rather than duplicating that
  functionality when they are turning a 128-bit value into two 64-bit
  values,
- to refactor `splitInt128` to use EXTRACT_ELEMENT which is for
  splitting wide integers in two (rather than the equivalent with shifts
  and truncs), and
- to use the concise helpers for INSERT_/EXTRACT_SUBREG in more places.

20 months ago[SLP][NFC]Update the test to avoid dead code elimination, NFC.
Alexey Bataev [Fri, 3 Mar 2023 14:09:47 +0000 (06:09 -0800)]
[SLP][NFC]Update the test to avoid dead code elimination, NFC.

20 months ago[mlir][IR] Hash result types in OperationFingerPrint
Matthias Springer [Fri, 3 Mar 2023 14:00:15 +0000 (15:00 +0100)]
[mlir][IR] Hash result types in OperationFingerPrint

Result types were not considered in OperationFingerPrint. This can lead to missing IR dumps when running with `shouldPrintAfterOnlyOnChange` and false positives in D144552.

Differential Revision: https://reviews.llvm.org/D145152

20 months ago[SCEV] Fix control flow warning (NFC)
Nikita Popov [Fri, 3 Mar 2023 14:03:51 +0000 (15:03 +0100)]
[SCEV] Fix control flow warning (NFC)

20 months ago[SCEV] Extract a helper to create a SCEV with new operands (NFC)
Nikita Popov [Fri, 3 Mar 2023 13:50:25 +0000 (14:50 +0100)]
[SCEV] Extract a helper to create a SCEV with new operands (NFC)

20 months ago[SCEV] Remove an unnecessary switch (NFC)
Nikita Popov [Fri, 3 Mar 2023 13:37:05 +0000 (14:37 +0100)]
[SCEV] Remove an unnecessary switch (NFC)

Just the scevUnconditionallyPropagatesPoisonFromOperands() check
is sufficient. Also rename the flag to be more in line with the
more general predicate.

20 months ago[MLIR][OpenMP] Add Lowering support for OpenMP Target Data with region
Akash Banerjee [Fri, 3 Mar 2023 13:04:39 +0000 (13:04 +0000)]
[MLIR][OpenMP] Add Lowering support for OpenMP Target Data with region

This patch adds Fortran Lowering support for the OpenMP Target Data Op with associated region.

Differential Revision: https://reviews.llvm.org/D144013

20 months ago[libc++][spaceship] Implement `operator<=>` for `forward_list`
Hristo Hristov [Fri, 3 Mar 2023 12:41:19 +0000 (13:41 +0100)]
[libc++][spaceship] Implement `operator<=>` for `forward_list`

Implemented `operator<=>` for `forward_list`

Reviewed By: #libc, philnik

Spies: philnik, libcxx-commits, yaxunl

Differential Revision: https://reviews.llvm.org/D145172

20 months ago[libc++][spaceship] Implement `operator<=>` for `deque`
Hristo Hristov [Fri, 3 Mar 2023 12:35:07 +0000 (13:35 +0100)]
[libc++][spaceship] Implement `operator<=>` for `deque`

Based on https://reviews.llvm.org/D132312
Dependes on https://reviews.llvm.org/D132312

Reviewed By: #libc, Mordante, philnik

Spies: philnik, Mordante, yaxunl, libcxx-commits

Differential Revision: https://reviews.llvm.org/D144821

20 months ago[InstCombine] Simplify separate_storage assumptions
David Goldblatt [Fri, 3 Mar 2023 12:23:27 +0000 (13:23 +0100)]
[InstCombine] Simplify separate_storage assumptions

Before this change, we call getUnderlyingObject on each separate_storage
operand on every alias() call (potentially requiring lots of pointer
chasing). Instead, we rewrite the assumptions in instcombine to do this
pointer-chasing once.

We still leave the getUnderlyingObject calls in alias(), just expecting
them to be no-ops much of the time. This is relatively fast (just a
couple dyn_casts with no pointer chasing) and avoids making alias
analysis results depend on whether or not instcombine has been run.

Differential Revision: https://reviews.llvm.org/D144933

20 months ago[Test] Add more tests with min/max loop guards (NFC)
Dmitry Makogon [Fri, 3 Mar 2023 11:11:44 +0000 (18:11 +0700)]
[Test] Add more tests with min/max loop guards (NFC)

This adds more test cases with loop guards involving min/max
and which should be covered by ScalarEvolution::applyLoopGuards.

20 months ago[ScalarEvolution] Factor out RewriteMap utilities in applyLoopGuards (NFC)
Dmitry Makogon [Fri, 3 Mar 2023 11:02:46 +0000 (18:02 +0700)]
[ScalarEvolution] Factor out RewriteMap utilities in applyLoopGuards (NFC)

This factors out two utilities used with RewriteMap in applyLoopGuards:
 - AddRewrite, which puts a rewrite rule in the map and if needed registers
   the rewrite in the list of rewritten expressions,
 - GetMaybeRewritten, which checks whether an expression has already been
   rewritten, and if so, returns the rewrite. Otherwise, returns the given
   expression.

This may be needed when adding new rewrite rules as not to copy-paste this
code.

20 months agoRevert "[ADCE] Only remove debug intrinsics if non debug instructions are removed"
Mikael Holmen [Fri, 3 Mar 2023 12:07:22 +0000 (13:07 +0100)]
Revert "[ADCE] Only remove debug intrinsics if non debug instructions are removed"

This reverts commit 8aa9ab336889ae2eb8e4188036faeb151379ab7b.

Reverting due to compile-time regressions as pointed out in
 https://reviews.llvm.org/D145051#4166656
E.g.
 "In particular tramp3d-v4 with debuginfo regressed by 15%."

20 months ago[NVPTX] Add f16 and v2f16 ldg builtins
Jakub Chlanda [Fri, 3 Mar 2023 07:36:51 +0000 (08:36 +0100)]
[NVPTX] Add f16 and v2f16 ldg builtins

Adds f16 and v2f16 ldg builtins and relevant tests.

Differential Revision: https://reviews.llvm.org/D144961

20 months ago[AMDGPU] Fix error message for illegal copy
Jay Foad [Fri, 3 Mar 2023 11:46:01 +0000 (11:46 +0000)]
[AMDGPU] Fix error message for illegal copy

20 months ago[mlir] Apply ClangTidy readability findings (NFC)
Adrian Kuegel [Fri, 3 Mar 2023 11:41:02 +0000 (12:41 +0100)]
[mlir] Apply ClangTidy readability findings (NFC)

20 months ago[AMDGPU] Add test case for AMDGPUInsertDelayAlu bug
Jay Foad [Fri, 3 Mar 2023 11:03:51 +0000 (11:03 +0000)]
[AMDGPU] Add test case for AMDGPUInsertDelayAlu bug

20 months ago[SimplifyCFG] Handle MD_noundef when hoisting common codes
luxufan [Fri, 3 Mar 2023 10:05:35 +0000 (18:05 +0800)]
[SimplifyCFG] Handle MD_noundef when hoisting common codes

Reviewed By: nikic

Differential Revision: https://reviews.llvm.org/D144939

20 months agoRevert "[libcxx] Temporarily skip Arm configs"
David Spickett [Fri, 3 Mar 2023 10:59:52 +0000 (10:59 +0000)]
Revert "[libcxx] Temporarily skip Arm configs"

This reverts commit 8a023fed2f848031769eae804e55eed719cdd12d.

The machine is back online.

20 months ago[IR][Legalization] Promote illegal deinterleave and interleave vectors
Caroline Concatto [Wed, 1 Mar 2023 13:03:28 +0000 (13:03 +0000)]
[IR][Legalization] Promote illegal deinterleave and interleave vectors

To make legalization easier, the operands and outputs have the same size for
these ISD Nodes. When legalizing the results in PromoteIntegerResult the operands
are legalized to the same size as the outputs.
The ISD Node has two output/results, therefore the legalizing functions update
both results/outputs.

Reviewed By: paulwalker-arm

Differential Revision: https://reviews.llvm.org/D144846

20 months ago[llvm][Uniformity] correctly use a vector as a set by uniqifying elements
Sameer Sahasrabuddhe [Fri, 3 Mar 2023 07:13:14 +0000 (12:43 +0530)]
[llvm][Uniformity] correctly use a vector as a set by uniqifying elements

The search for temporal divergence needs to determine a dominance frontier
defined for a cycle. The implementation uses a temporary vector to store a set
of newly discovered successors. Failing to uniqify the elements in this vector
causes a very large regression in compile time due to an exponential number of
redundant visits.

This fixes github issue #61123

Reviewed By: foad

Differential Revision: https://reviews.llvm.org/D145216

20 months ago[NFC] Refine tests by adding `:` to checks
Mariya Podchishchaeva [Fri, 3 Mar 2023 10:02:22 +0000 (05:02 -0500)]
[NFC] Refine tests by adding `:` to checks

The tests can fail if working directory where the tests were launched
has a `error` substring in its path.

Reviewed By: jhenderson, foad

Differential Revision: https://reviews.llvm.org/D144562

20 months agoFix crash in ConvertVectorToLLVM.cpp pattern
Mehdi Amini [Fri, 3 Mar 2023 10:14:17 +0000 (11:14 +0100)]
Fix crash in ConvertVectorToLLVM.cpp pattern

Fixes #61094

20 months agoRecommit [C++20] [Modules] Trying to compare the trailing require clause from the...
Chuanqi Xu [Fri, 3 Mar 2023 09:27:37 +0000 (17:27 +0800)]
Recommit [C++20] [Modules] Trying to compare the trailing require clause from the primary template function

Close https://github.com/llvm/llvm-project/issues/60890.

For the following example:

```
export module a;

export template<typename T>
struct a {
friend void aa(a) requires(true) {
}
};
```

```
export module b;

import a;

struct b {
a<int> m;
};
```

```
export module c;

import a;

struct c {
void f() const {
aa(a<int>());
}
};
```

```
import a;
import b;
import c;

void d() {
aa(a<int>());
}
```

The current clang will reject this incorrectly. The reason is that the
require clause  will be replaced with the evaluated version
(https://github.com/llvm/llvm-project/blob/efae3174f09560353fb0f3d528bcbffe060d5438/clang/lib/Sema/SemaConcept.cpp#L664-L665).
In module 'b', the friend function is instantiated but not used so the
require clause of the friend function is `(true)`. However, in module
'c', the friend function is used so the require clause is `true`. So
deserializer classify these two function to two different functions
instead of one. Then here is the bug report.

The proposed solution is to try to compare the trailing require clause
of the primary template when performing ODR checking.

Reviewed By: erichkeane

Differential Revision: https://reviews.llvm.org/D144626

20 months ago[libcxx] Temporarily skip Arm configs
David Spickett [Tue, 31 May 2022 10:10:12 +0000 (10:10 +0000)]
[libcxx] Temporarily skip Arm configs

The machine hosting these agents will be down
for maintenance today.

We (Linaro) will remove this once the agents are back online.

20 months agoRevert "Add the ability to segment GSYM files."
Douglas Yung [Fri, 3 Mar 2023 08:25:06 +0000 (00:25 -0800)]
Revert "Add the ability to segment GSYM files."

This reverts commit fe758254181a824d73ad960b651b42f671f8936b.

This change was causing several buildbot failures:
- https://lab.llvm.org/buildbot/#/builders/38/builds/10105
- https://lab.llvm.org/buildbot/#/builders/192/builds/562
- https://lab.llvm.org/buildbot/#/builders/109/builds/58893
- https://lab.llvm.org/buildbot/#/builders/16/builds/44360
- https://lab.llvm.org/buildbot/#/builders/247/builds/2095
- https://lab.llvm.org/buildbot/#/builders/196/builds/27236
- https://lab.llvm.org/buildbot/#/builders/54/builds/3714

20 months ago[clang][ASTImporter] Import TemplateName correctly
Balázs Kéri [Fri, 3 Mar 2023 07:46:06 +0000 (08:46 +0100)]
[clang][ASTImporter] Import TemplateName correctly

This is a fix for a problem when multiple template
specializations are created by ASTImporter for the
same specialization. The problem happens if a
TemplateName is imported that points to a template
delcaration (for a template template argument)
(specialization) that has multiple instances in the
declaration chain. If two TemplateName objects contain
different pointers to a template specialization,
these TemplateName objects will have different checksum
even if they point into the same declaration chain.
The problem is fixed if the canonical declaration is used.

Reviewed By: vabridgers, donat.nagy

Differential Revision: https://reviews.llvm.org/D144622

20 months ago[clang][RISCV] Enable -fasynchronous-unwind-tables by default on Linux
Kito Cheng [Fri, 3 Mar 2023 07:32:44 +0000 (15:32 +0800)]
[clang][RISCV] Enable -fasynchronous-unwind-tables by default on Linux

This could improve user experience for stack unwinding, and also this is
enabled by default by X86 and AArch64 and RISC-V GCC.

Reviewed By: luismarques, MaskRay

Differential Revision: https://reviews.llvm.org/D145164

20 months ago[CodeGen] guarantee templated static variables are initialized in the reverse instant...
Yuanfang Chen [Thu, 9 Jun 2022 19:48:23 +0000 (12:48 -0700)]
[CodeGen] guarantee templated static variables are initialized in the reverse instantiation order

Based on Richard's suggestion in D126341:
`If we can actually describe a rule that we provide for initialization
order of instantiated variables, and we can easily implement that rule
and be confident we won't want to substantially weaken it later, and we
can thereby assure our users that we will satisfy that rule, then I
think that could be interesting, but anything less than that doesn't
seem worthwhile to me.`

I'm giving it try here. IMHO the implementation is pretty simple and
does not change behavior for unrelated constructs like the timing when
instantiated variables are passed to CodeGen.

This is based on the same ordering guarantee needed for inline variables D127233.
To provide this guarantee, we also need to
emit DeferredDeclsToEmit in the DFS order. https://github.com/llvm/llvm-project/commit/e5df59ff78faebd897e81907606ce6074aac0df6
originally supported this but it is not exactly DFS order for cases like
the one in cwg362. For the example of Fib<5>, it triggers the instantiation
of Fib<4> and Fib<3>. However, due to the way GlobalEagerInstantiationScope
is implemented, Fib<4> does not actually trigger Fib<3> instantiation
since it is already triggered by Fib<5>. This breaks the guarantee.

This patch makes sure DeferredDeclsToEmit is emitted in DFS order by moving
DeferredDeclsToEmit storage from the call stack to an explicit stack-like
data structure. Then the DFS order could be enforced.

Differential Revision: https://reviews.llvm.org/D127259

20 months ago[TypePromotion] Deference pointer before printing it in a debug message.
Craig Topper [Fri, 3 Mar 2023 07:22:56 +0000 (23:22 -0800)]
[TypePromotion] Deference pointer before printing it in a debug message.

Without deferencing it just prints the value of the pointer which
isn't meaningful. Dereferencing prints the operand.

20 months ago[libc] Make errno an entrypoint.
Siva Chandra Reddy [Thu, 2 Mar 2023 05:53:38 +0000 (05:53 +0000)]
[libc] Make errno an entrypoint.

The entrypoint has been added to the various entrypoint lists. The libc
code style doc has been updated with information on how errno should be
set from the libc runtime code.

Reviewed By: lntue

Differential Revision: https://reviews.llvm.org/D145179

20 months ago[ADCE] Only remove debug intrinsics if non debug instructions are removed
Mikael Holmen [Thu, 2 Mar 2023 05:16:41 +0000 (06:16 +0100)]
[ADCE] Only remove debug intrinsics if non debug instructions are removed

We now limit ADCE to only remove debug intrinsics if it does something else
that would invalidate cached analyses anyway.
As we've seen in
 https://github.com/llvm/llvm-project/issues/58285
throwing away cached analysis info when only debug instructions are removed
can lead to different code when debug info is present or not present.

Differential Revision: https://reviews.llvm.org/D145051

20 months ago[C++20] [Modules] Deprecate to load C++20 Modules eagerly
Chuanqi Xu [Fri, 24 Feb 2023 08:41:52 +0000 (16:41 +0800)]
[C++20] [Modules] Deprecate to load C++20 Modules eagerly

Close https://github.com/llvm/llvm-project/issues/60824

The form -fmodule-file=<path-to-BMI> will load modules eagerly and the
form -fmodule-file=<module-name>=<path-to-BMI> will load modules lazily.
The inconsistency adds many additional burdens to the implementations.
And the inconsistency looks not helpful and necessary neither. So I want
to deprecate the form -fmodule-file=<path-to-BMI> for named modules.
This is pretty helpful for us (the developers).

Does this change make any regression from the perspective of the users?

To be honest, yes. But I think such regression is acceptable. Here is
the example:

```
// M.cppm
export module M;
export int m = 5;

// N.cpp
// import M; // woops, we forgot to import M.
int n = m;
```

In the original version, the compiler can diagnose the users to import
`M` since the compiler have already imported M. But in the later style,
the compiler can only say "unknown identifier `m`".

But I think such regression doesn't make a deal since it only works if
the user put `-fmodule-file=M.pcm` in the command line. But how can the
user put `-fmodule-file=M.pcm` in the command line without `import M;`?
Especially currently such options are generated by build systems. And
the build systems will only generate the command line from the source
file.

So I think this change is pretty pretty helpful for developers and
almost innocent for users and we should accept this one.

I'll add the release notes and edit the document after we land this.

Differential Revision: https://reviews.llvm.org/D144707

20 months agoAdd the ability to segment GSYM files.
Greg Clayton [Thu, 9 Feb 2023 18:16:34 +0000 (10:16 -0800)]
Add the ability to segment GSYM files.

Some workflows can generate large GSYM files and sharding GSYM files into segments can help some performant workflows that can take advantage of smaller GSYM files. This patch add a new --segment-size option to llvm-gsymutil. This option can specify a rough size in bytes of how large each segment should be.

Segmented GSYM files contain only the strings and files that are needed for the FunctionInfo objects that are added to each shard. The output file path gets the first address of the first contained function info appended as a suffix to the filename. If a base address of an image is set in the GsymCreator, then all segments will use this same base address which allows lookups for symbolication to happen correctly when the image has been slid in memory.

Code has been addeed to refactor and re-use methods within the GsymCreator to allow for segments to be created easily and tested.

Example of segmenting GSYM files:

$ llvm-gsymutil --convert llvm-gsymutil.dSYM -o llvm-gsymutil.gsym --segment-size 10485760
$ ls -l llvm-gsymutil.gsym-*
-rw-r--r--  1 gclayton  staff  10485839 Feb  9 10:45 llvm-gsymutil.gsym-0x1000030c0
-rw-r--r--  1 gclayton  staff  10485765 Feb  9 10:45 llvm-gsymutil.gsym-0x100668888
-rw-r--r--  1 gclayton  staff  10485881 Feb  9 10:45 llvm-gsymutil.gsym-0x100c948b8
-rw-r--r--  1 gclayton  staff  10485954 Feb  9 10:45 llvm-gsymutil.gsym-0x101659e70
-rw-r--r--  1 gclayton  staff  10485792 Feb  9 10:45 llvm-gsymutil.gsym-0x1022b1dc0
-rw-r--r--  1 gclayton  staff  10485889 Feb  9 10:45 llvm-gsymutil.gsym-0x102a18b10
-rw-r--r--  1 gclayton  staff  10485893 Feb  9 10:45 llvm-gsymutil.gsym-0x1030b05d0
-rw-r--r--  1 gclayton  staff  10485802 Feb  9 10:45 llvm-gsymutil.gsym-0x1037caaac
-rw-r--r--  1 gclayton  staff  10485781 Feb  9 10:45 llvm-gsymutil.gsym-0x103e767a0
-rw-r--r--  1 gclayton  staff  10485832 Feb  9 10:45 llvm-gsymutil.gsym-0x10452d0d4
-rw-r--r--  1 gclayton  staff  10485782 Feb  9 10:45 llvm-gsymutil.gsym-0x104b93310
-rw-r--r--  1 gclayton  staff   6255785 Feb  9 10:45 llvm-gsymutil.gsym-0x10526bf34

Differential Revision: https://reviews.llvm.org/D143793

20 months ago[PowerPC][NFC] cleanup isEligibleForTCO
Ting Wang [Fri, 3 Mar 2023 04:04:19 +0000 (23:04 -0500)]
[PowerPC][NFC] cleanup isEligibleForTCO

The input parameter IsByValArg to isEligibleForTCO() is false in all
cases, so it is considered redundant and should be removed.

Reviewed By: shchenz

Differential Revision: https://reviews.llvm.org/D145028

20 months ago[C++20] [Modules] Make TheImplicitGlobalModuleFragment and TheExportedImplicitGlobalM...
Chuanqi Xu [Fri, 3 Mar 2023 03:11:31 +0000 (11:11 +0800)]
[C++20] [Modules] Make TheImplicitGlobalModuleFragment and TheExportedImplicitGlobalModuleFragment to be useable modules

The unexported language linkage become unvisible to the current module
unit after the previous commit bf52ead24ca4. This patch fixes the issue.

20 months ago[C++20] [Modules] Support to export declarations in language linkage
Chuanqi Xu [Fri, 3 Mar 2023 02:31:48 +0000 (10:31 +0800)]
[C++20] [Modules] Support to export declarations in language linkage

Close https://github.com/llvm/llvm-project/issues/60405

See the discussion in the above link for the background.

What the patch does:
- Rename `Module::ModuleKind::GlobalModuleFragment` to
  `Module::ModuleKind::ExplicitGlobalModuleFragment`.
- Add another module kind `ImplicitGlobalModuleFragment` to
  `ModuleKind`.
- Create an implicit global module fragment for the language linkage
  declarations inside a module purview.
    - If the language linkage lives inside the scope of an export decl,
      the created modules is marked as exported to outer modules.
- In fact, Sema will only create at most 2 implicit global module
  fragments to avoid creating a lot of unnecessary modules in the edging
case.

Reviewed By: iains

Differential Revision: https://reviews.llvm.org/D144367

20 months ago[libc++] Temporarily not use compiler intrinsics for some type traits in Objective...
Konstantin Varlamov [Fri, 3 Mar 2023 01:35:03 +0000 (17:35 -0800)]
[libc++] Temporarily not use compiler intrinsics for some type traits in Objective-C++ mode.

Currently, there are bugs in Clang's intrinsics for type traits when
handling Objective-C++ `id` (e.g. in `add_pointer`). As a temporary
workaround, don't use these intrinsics in the Objective-C++ mode.

Differential Revision: https://reviews.llvm.org/D145186

20 months ago[NFC] Properly handle optional minor value for ArchInfo::Version
Zixu Wang [Fri, 3 Mar 2023 00:23:57 +0000 (16:23 -0800)]
[NFC] Properly handle optional minor value for ArchInfo::Version

Use `value_or` instead of `value` for checking minor versions in
`ArchInfo::implies`.

Differential Revision: https://reviews.llvm.org/D145206

20 months agoThis check-in makes the following improvements to the OpenMP Windows build:
Vadim Paretsky (Intel Americas Inc) [Thu, 2 Mar 2023 23:46:13 +0000 (15:46 -0800)]
This check-in makes the following improvements to the OpenMP Windows build:

Only generate the second def file when necessary (native Windows import
library builds).
Properly clean up .def file artifacts.
Reduce the re-generated import library build artifacts to the minimum.
Refactor the import library related portions of the script for clarity.

Tested with MSVC and MinWG/gcc12.0

Differential Revision:https://reviews.llvm.org/D144419

20 months agoRevert "[WebAssembly] Implement build-id feature"
Derek Schuff [Thu, 2 Mar 2023 23:28:56 +0000 (15:28 -0800)]
Revert "[WebAssembly] Implement build-id feature"

This reverts commit 41e31466af6a7feab82742bb01af43f4f3ae4ede
due to a build failure on Windows.

20 months ago[flang] Warn about inconsistent implicit interfaces
Peter Klausler [Fri, 17 Feb 2023 22:57:52 +0000 (14:57 -0800)]
[flang] Warn about inconsistent implicit interfaces

When a global procedure has no explicit interface, emit warnings
when its references are inconsistent implicit procedure interfaces.

Differential Revision: https://reviews.llvm.org/D145097

20 months ago[llvm] Add missing header guards in LogicCombine.h and LogicalExpr.h
Alex Langford [Thu, 2 Mar 2023 23:12:23 +0000 (15:12 -0800)]
[llvm] Add missing header guards in LogicCombine.h and LogicalExpr.h

These files were added in 97dcbea63e11d566cff0cd3a758cf1114cf1f633 but
it looks like they are missing header guards. This breaks module builds.

20 months ago[SelectionDAG] Fix missing lambda capture
Marco Elver [Thu, 2 Mar 2023 22:45:42 +0000 (23:45 +0100)]
[SelectionDAG] Fix missing lambda capture

Move MaxDepth into the lambda, since it is not needed outside. This
fixes some compilers that complain about missing capture:

  error C3493: 'MaxDepth' cannot be implicitly captured because no
  default capture mode has been specified

Fixes: f693932fbea7 ("[SelectionDAG] Transitively copy NodeExtraInfo on RAUW")

20 months ago[GISel][CSE][NFC]: Handle mutual recursion when inserting node
Aditya Nandakumar [Thu, 2 Mar 2023 22:06:29 +0000 (14:06 -0800)]
[GISel][CSE][NFC]: Handle mutual recursion when inserting node

GISel's CSE mechanism lazily inserts instructions into the CSE List
to improve on efficiency as well as efficacy of CSE
(for allowing partially built instructions to be fully built).

There's unfortunately a mutual recursion via
 `handleRecordedInsts -> handleRecordedInst -> insertNode-> handleRecordedInsts`.

So this change simply records that we're already draining this list so we can just bail out on the recursion.

No changes to codegen are expected as we're still draining/handling the temporary
list via pop_back and we should get the same sequence of instructions
whether we call pop_back in a loop at the top level or recursive.

https://reviews.llvm.org/D145006

reviewed by: dsanders

20 months ago[libc][NFC] Switch use of errno in src/time and test/src/time to libc_errno.
Raman Tenneti [Thu, 2 Mar 2023 22:24:42 +0000 (14:24 -0800)]
[libc][NFC] Switch use of errno in src/time and test/src/time to libc_errno.

Switch use of errno in src/time and test/src/time to libc_errno.

Reviewed By: sivachandra

Differential Revision: https://reviews.llvm.org/D145192

20 months agoRevert "Add SBValue::GetValueAsAddress API for removing non-addressing metadata"
Jason Molenda [Thu, 2 Mar 2023 22:36:37 +0000 (14:36 -0800)]
Revert "Add SBValue::GetValueAsAddress API for removing non-addressing metadata"

Revert while I investigate two CI bot failures;
the more important is the lldb-arm-ubuntu where
the FixAddress is removing the 0th bit so we're
adding the `actual=` decorator on a string pointer,

```
Got output:
(char *) strptr = 0x00400817 (actual=0x400816) ptr = [{ },{H}]
```

in TestDataFormatterSmartArray.py line 229.

This reverts commit 4d635be2dbadc77522eddc9668697385a3b9f8b4.

20 months ago[ASAN] Autogen masked load/store test coverage [nfc]
Philip Reames [Thu, 2 Mar 2023 22:16:19 +0000 (14:16 -0800)]
[ASAN] Autogen masked load/store test coverage [nfc]

I also removed two runlines which added no additional coverage.  No test in the file has both loads and stores, thus the two configurations duplicate the disabled configuration.

20 months ago[WebAssembly] Implement build-id feature
Derek Schuff [Wed, 4 Aug 2021 21:05:40 +0000 (14:05 -0700)]
[WebAssembly] Implement build-id feature

Implement the --build-id flag similarly to ELF, and generate a build_id
section according to the WebAssembly tool convention specified in
https://github.com/WebAssembly/tool-conventions/pull/183

The default style ("fast" aka "tree") hashes the contents of the output
and (unlike ELF) generates a v5 UUID based on the hash (using a random
namespace).
It also supports generating a random v4 UUID, a sha1 hash,
and a user-specified string (as ELF does).

Differential Revision: https://reviews.llvm.org/D107662

20 months ago[AArch64] Fix crash in LowerBUILD_VECTOR trying to create invalid EXTRACT_SUBVECTOR.
Amara Emerson [Thu, 2 Mar 2023 21:49:00 +0000 (13:49 -0800)]
[AArch64] Fix crash in LowerBUILD_VECTOR trying to create invalid EXTRACT_SUBVECTOR.

rdar://106096671

Differential Revision: https://reviews.llvm.org/D145185

20 months ago[SelectionDAG] Transitively copy NodeExtraInfo on RAUW
Marco Elver [Mon, 27 Feb 2023 11:05:02 +0000 (12:05 +0100)]
[SelectionDAG] Transitively copy NodeExtraInfo on RAUW

During legalization of the SelectionDAG, some nodes are replaced with
arch-specific nodes. These may be complex nodes, where the root node no
longer corresponds to the node that should carry the extra info.

Fix the issue by copying extra info to the new node and all its new
transitive operands during RAUW. See code comments for more details.

This fixes the remaining pcsections-atomics.ll tests on X86.

v2: Optimize copyExtraInfo() deep copy. For now we assume that only
NodeExtraInfo that have PCSections set require deep copy. Furthermore,
limit the depth of graph search while pre-populating the visited set,
assuming the to-be-replaced subgraph 'From' has limited complexity. An
assertion catches if the maximum depth needs to be increased.

Reviewed By: dvyukov

Differential Revision: https://reviews.llvm.org/D144677

20 months ago[flang] Fix CONTIGUOUS attribute for construct entities
Peter Klausler [Sat, 25 Feb 2023 14:46:32 +0000 (06:46 -0800)]
[flang] Fix CONTIGUOUS attribute for construct entities

Currently, the selector of a construct entity (e.g., ASSOCIATE(x=>a(1:20)))
is inheriting the CONTIGUOUS attribute from its associated variable
even if it has subscripts that make it noncontiguous (a(1:20:2)).
Add construct entities to the dynamic contiguity predicate instead.

Differential Revision: https://reviews.llvm.org/D145114

20 months ago[ASAN] Simplify masked load/store test [nfc]
Philip Reames [Thu, 2 Mar 2023 22:00:21 +0000 (14:00 -0800)]
[ASAN] Simplify masked load/store test [nfc]

Only use of the globals was to load a pointer, we can just pass in a pointer instead.

20 months ago[libc][obvious] Fix errno includes in unistd
Michael Jones [Thu, 2 Mar 2023 22:03:03 +0000 (14:03 -0800)]
[libc][obvious] Fix errno includes in unistd

Minor typo swapped quotes for angle brackets. This patch moves
everything to quotes.

Differential Revision: https://reviews.llvm.org/D145188

20 months ago[Linux] Add kernel.yama.ptrace_scope note when ptrace fails to attach.
Jordan Rupprecht [Thu, 2 Mar 2023 21:51:30 +0000 (13:51 -0800)]
[Linux] Add kernel.yama.ptrace_scope note when ptrace fails to attach.

A common reason for LLDB failing to attach to an already-running process on Linux is the Yama security module: https://www.kernel.org/doc/Documentation/security/Yama.txt. This patch adds an explaination and suggested fix when it detects that case happening.

This was previously proposed in D106226, but hasn't been updated in a while. The last request was to put the check in a target-specific location, which is the approach this patch takes. I believe Yama only exists on Linux, so it's put in that package.

This has no end-to-end test because I'm not sure how to set `ptrace_scope` in a test environment -- if there are suggestions on how to do that, I'd be happy to add it. (Also, setting it to `3` is comically irreversible). I tested this locally.

Reviewed By: DavidSpickett, labath

Differential Revision: https://reviews.llvm.org/D144904

20 months ago[clang][deps] NFC: Simplify worker loop
Jan Svoboda [Wed, 1 Mar 2023 19:51:20 +0000 (11:51 -0800)]
[clang][deps] NFC: Simplify worker loop

This patch simplifies the loop inside each worker by extracting index retrieval into a lambda function.

Reviewed By: benlangmuir

Differential Revision: https://reviews.llvm.org/D145101

20 months ago[flang] Move check for statement function in BLOCK construct
Peter Klausler [Sat, 25 Feb 2023 01:36:15 +0000 (17:36 -0800)]
[flang] Move check for statement function in BLOCK construct

A BLOCK construct may not contain a statement function definition;
but it may of course contain an assignment statement with an array
element on its left-hand side that looks like a statement function
definition.  These misparsed statement functions are converted
into assignment statements during semantics once it is clear what
they are.  Move the C1107 check for a statement function definition
in a block construct into declaration checking, which is where it
probably should have been in the first place anyway.

Differential Revision: https://reviews.llvm.org/D145112

20 months ago[ADT] Clean up `enumerate` implementation. NFC.
Jakub Kuderski [Thu, 2 Mar 2023 21:37:56 +0000 (16:37 -0500)]
[ADT] Clean up `enumerate` implementation. NFC.

*  Remove unnecessary member functions.
*  Fix code sample.

This is in preparation for landing future changes in https://reviews.llvm.org/D144503.

Reviewed By: zero9178

Differential Revision: https://reviews.llvm.org/D145025

20 months agoRevert "[Support] Implement findModulesAndOffsets on Apple 64-bit platforms"
Luís Marques [Thu, 2 Mar 2023 21:42:23 +0000 (21:42 +0000)]
Revert "[Support] Implement findModulesAndOffsets on Apple 64-bit platforms"

This reverts commit b8b8aa6f06458212193c4202291c9f68364b2025.
It broke AIX.

20 months ago[clang][deps] Preserve input ordering in the full output
Jan Svoboda [Wed, 1 Mar 2023 19:31:49 +0000 (11:31 -0800)]
[clang][deps] Preserve input ordering in the full output

This patch makes sure the ordering of TUs in the input CDB is preserved in the full output. Previously, the results were pushed into a global vector, potentially out-of-order and then sorted by the input file name. This is non-deterministic when the CDB contains multiple entries with the same input file. This patch fixes that by pre-allocating the output vector and writing directly to the position corresponding to the current input. This also eliminates one critical section.

Reviewed By: benlangmuir

Differential Revision: https://reviews.llvm.org/D145098

20 months ago[libc] disable integration tests when using gcc
Michael Jones [Thu, 2 Mar 2023 00:27:02 +0000 (16:27 -0800)]
[libc] disable integration tests when using gcc

The integration tests were failing to build under GCC due to missing
compile options for using LLVM's libc++ and compiler-rt.

This should unblock https://github.com/llvm/llvm-project/issues/60467

Reviewed By: sivachandra, lntue

Differential Revision: https://reviews.llvm.org/D145128

20 months agoAdd SBValue::GetValueAsAddress API for removing non-addressing metadata
Jason Molenda [Thu, 2 Mar 2023 21:28:34 +0000 (13:28 -0800)]
Add SBValue::GetValueAsAddress API for removing non-addressing metadata

On target where metadata is stored in bits that aren't used for
virtual addressing -- AArch64 Top Byte Ignore and pointer authentication
are two examples -- an SBValue object representing a pointer will
return the address with metadata for SBValue::GetValueAsUnsigned.
Users may want to get the virtual address without the metadata;
this new method gives them a way to do this.

Differential Revision: https://reviews.llvm.org/D142792

20 months ago[mlir] Use std::optional instead of llvm::Optional (NFC)
Kazu Hirata [Thu, 2 Mar 2023 21:19:42 +0000 (13:19 -0800)]
[mlir] Use std::optional instead of llvm::Optional (NFC)

20 months ago[mlir][sparse] convenience util to test for "sparse" op
Aart Bik [Thu, 2 Mar 2023 20:06:29 +0000 (12:06 -0800)]
[mlir][sparse] convenience util to test for "sparse" op

Reviewed By: cota

Differential Revision: https://reviews.llvm.org/D145176

20 months ago[Docs] update MemorySSA for opaque ptrs
Nick Desaulniers [Thu, 2 Mar 2023 20:58:53 +0000 (12:58 -0800)]
[Docs] update MemorySSA for opaque ptrs

Reviewed By: asbirlea, nikic

Differential Revision: https://reviews.llvm.org/D145130

20 months ago[lldb/Interpreter] Fix build failures in ScriptInterpreterLua
Med Ismail Bennani [Thu, 2 Mar 2023 20:46:34 +0000 (12:46 -0800)]
[lldb/Interpreter] Fix build failures in ScriptInterpreterLua

This patch should fix the build failures in the Lua ScriptedInterpreter
introduced by 9a9fce1fed6d.

Signed-off-by: Med Ismail Bennani <medismail.bennani@gmail.com>
20 months agoRevert "[lldb/Interpreter] Fix build failures in ScriptInterpreterLua"
Med Ismail Bennani [Thu, 2 Mar 2023 20:46:18 +0000 (12:46 -0800)]
Revert "[lldb/Interpreter] Fix build failures in ScriptInterpreterLua"

This reverts commit 6de18eb050a66544cc38210024860366b84faf35.

20 months ago[lldb/Interpreter] Fix build failures in ScriptInterpreterLua
Med Ismail Bennani [Thu, 2 Mar 2023 20:43:15 +0000 (12:43 -0800)]
[lldb/Interpreter] Fix build failures in ScriptInterpreterLua

This patch should fix the build failures in the Lua ScriptedInterpreter
introduced by 9a9fce1fed6d.

Signed-off-by: Med Ismail Bennani <medismail.bennani@gmail.com>
20 months agoRevert "Revert "Revert "[scudo] Only prepare PageMap entry for partial region"""
Chia-hung Duan [Thu, 2 Mar 2023 20:37:55 +0000 (20:37 +0000)]
Revert "Revert "Revert "[scudo] Only prepare PageMap entry for partial region"""

ScudoReleaseTest.ReleaseFreeMemoryToOSAndroid failed on Fuchsia

This reverts commit c6ef6bbd8d964028ee6c2f03441604d7a7ba5375.

20 months ago[libc++][chrono] XFAIL hh_mm_ss formatter test on FreeBSD
Ed Maste [Wed, 1 Mar 2023 22:09:38 +0000 (17:09 -0500)]
[libc++][chrono] XFAIL hh_mm_ss formatter test on FreeBSD

This test did not exist when I added the previous set of XFAILs (and the
libc++ FreeBSD CI runner is not yet active).

Reviewed by: Mordante

Differential Revision: https://reviews.llvm.org/D145120

20 months agoRevert "[Clang] Refactor "Designators" into a unified implementation [NFC]"
Bill Wendling [Thu, 2 Mar 2023 20:08:07 +0000 (12:08 -0800)]
Revert "[Clang] Refactor "Designators" into a unified implementation [NFC]"

This reverts commit 3c07db5f58e9852f35202f0fffed50fc7506f37b.

This caused https://github.com/llvm/llvm-project/issues/61118. Reverting
to ensure this is a pure NFC change.

20 months ago[SelectionDAG][AArch64] Constant fold in SelectionDAG::getVScale if VScaleMin==VScaleMax.
Craig Topper [Thu, 2 Mar 2023 20:02:37 +0000 (12:02 -0800)]
[SelectionDAG][AArch64] Constant fold in SelectionDAG::getVScale if VScaleMin==VScaleMax.

Reviewed By: paulwalker-arm

Differential Revision: https://reviews.llvm.org/D145113

20 months ago[clang-format] Add -j to clang-format-diff to speed up formatting
Sean Maher [Thu, 2 Mar 2023 19:53:57 +0000 (11:53 -0800)]
[clang-format] Add -j to clang-format-diff to speed up formatting

This patch changes the implementation of clang-format-diff.py to
start up many clang-format processes in parallel in order to speed
up clang-format-diff.py by several orders of magnitude on large
patches.

Differential Revision: https://reviews.llvm.org/D141230

20 months ago[C2x] Add tests for WG14 N3035 and update the C status page
Aaron Ballman [Thu, 2 Mar 2023 19:35:03 +0000 (14:35 -0500)]
[C2x] Add tests for WG14 N3035 and update the C status page

The paper was making minor corrections to the standard that Clang had
already implemented. This adds (most of) the test coverage for the
paper and claims support. In my testing, we supported this in Clang 15.

20 months ago[PowerPC] Recognize long CPU name for -mtune in Clang
Nemanja Ivanovic [Thu, 2 Mar 2023 18:54:13 +0000 (13:54 -0500)]
[PowerPC] Recognize long CPU name for -mtune in Clang

There are two ways of specifying a CPU on PowerPC:
power<N> and pwr<N>. Clang/LLVM traditionally
supports the latter and Clang replaces the former
with the latter when passing it to the back end for
the -mcpu= option. However, when the -mtune= option
was introduced, this replacement was not implemented for it.

This leaves us in an inconsistent state of accepting
both forms for -mcpu= and and only the latter for
-mtune=. Furthermore, it leaves us incompatible with
GCC which only understands the power<N> version for
both options.

This patch just adds the same handling for the long
names for -mtune= as already exists for -mcpu=.

Differential revision: https://reviews.llvm.org/D144967

20 months ago[test][ASAN] Add test coverage for (fixed) vector load and stores
Philip Reames [Thu, 2 Mar 2023 19:10:57 +0000 (11:10 -0800)]
[test][ASAN] Add test coverage for (fixed) vector load and stores

20 months ago[libc][NFC] Move all developer docs into a sub-directory docs/dev.
Siva Chandra Reddy [Thu, 2 Mar 2023 18:58:04 +0000 (10:58 -0800)]
[libc][NFC] Move all developer docs into a sub-directory docs/dev.

20 months ago[SLP][NFC]Add a test with reused scalars in 3 tree nodes with different
Alexey Bataev [Thu, 2 Mar 2023 18:44:53 +0000 (10:44 -0800)]
[SLP][NFC]Add a test with reused scalars in 3 tree nodes with different
VF, NFC.

20 months ago[ASAN] Rename TypeSize to TypeStoreSize [mostly NFC]
Philip Reames [Thu, 2 Mar 2023 18:39:06 +0000 (10:39 -0800)]
[ASAN] Rename TypeSize to TypeStoreSize [mostly NFC]

This is a mechanical prep change for scalable vector support.  All it does is move the point of TypeSize to unsigned (i.e. the unsafe cast) closer to point of use.

20 months ago[flang][runtime] Handle explicit-length character padding & truncation in Assign()
Peter Klausler [Wed, 22 Feb 2023 18:19:25 +0000 (10:19 -0800)]
[flang][runtime] Handle explicit-length character padding & truncation in Assign()

When the left-hand side of an allocatable assignment has an explicit character length,
rather than a deferred length that might imply reallocation, handle any discrepancy
in lengths via truncation or blank padding.

Differential Revision: https://reviews.llvm.org/D145111