This commit adds an x86-64 Pointer8 edge kind (8-bit pointer), and uses it to
implement support for the ELF::R_X86_64_8 relocation kind.

The assembly version of this testcase is shorter and easier to read.

We already had an x86_64::Pointer16 edge kind courtesy of Sunho's COFF/x86-64
work, it just needed to be wired up.

Tests in test/ExecutionEngine/JITLink/AArch64 are for aarch64 only, so we don't
need to repeat 'aarch64' / 'arm64' in the individual test names.

Commit 8ad75c10 made a similar change for the x86-64 tests.

Most of Orc and JITLink are movinng away from JITTargetAddress and
use ExecutorAddr instead.

Signed-off-by: Jun Zhang <jun@junz.org>

Differential Revision: https://reviews.llvm.org/D148434

The terminator of this op is special: it does not just yield a value,
but bufferizes to a memcpy. This requires special treatment to make sure
that deallocs are placed after the memcpy. (By default, deallocs are
placed right before the terminator.)

Differential Revision: https://reviews.llvm.org/D148408

This fixes a TODO in the first version.

Reviewed By: Peiming

Differential Revision: https://reviews.llvm.org/D148406

There are some typos in the previous patch e0117a3e

 that woulc break unit test and CI.
These typos are fixed now.

Signed-off-by: Peter Rong <PeterRong96@gmail.com>

Source and Sink are required when generating a new instruction.
(Term defined by previous author, in LLVM terms it's probably Use and User.)
Previously, only instructions in the same block is considered when taking source and sink.

In this patch, more source and sink types are considered.
For source, we have SrcFromInstInCurBlock, FunctionArgument, InstInDominator, SrcFromGlobalVariable, and NewConstOrStack.
For sink, we have SinkToInstInCurBlock, PointersInDominator, InstInDominatee, NewStore, and SinkToGlobalVariable.

A unit test to make sure source always dominates an instruction, and the instruction always dominates the sink is included.

Reviewed By: arsenm

Differential Revision: https://reviews.llvm.org/D139907

Fixes a crash in dump-config when checks emit warnings trying to read the config

Fixes https://github.com/llvm/llvm-project/issues/50876



Co-authored-by: Piotr Zegar <me@piotrzegar.pl>

Reviewed By: PiotrZSL

Differential Revision: https://reviews.llvm.org/D148436

This patch precommits a test for:

https://github.com/llvm/llvm-project/issues/61650

Tests in test/ExecutionEngine/JITLink/X86 were for x86-64 only (never i386) so
it makes sense to name the test directory x86-64, and drop x86-64 from the
individual test names.

Without this patch, std::bit_floor<uint32_t> in libstdc++ is compiled
as:

  %eq0 = icmp eq i32 %x, 0
  %lshr = lshr i32 %x, 1
  %ctlz = tail call i32 @llvm.ctlz.i32(i32 %lshr, i1 false)
  %sub = sub i32 32, %ctlz
  %shl = shl i32 1, %sub
  %sel = select i1 %eq0, i32 0, i32 %shl

With this patch:

  %eq0 = icmp eq i32 %x, 0
  %ctlz = call i32 @llvm.ctlz.i32(i32 %x, i1 false)
  %lshr = lshr i32 -2147483648, %1
  %sel = select i1 %eq0, i32 0, i32 %lshr

This patch recognizes the specific pattern emitted for std::bit_floor
in libstdc++.

https://alive2.llvm.org/ce/z/piMdFX

This patch fixes:

https://github.com/llvm/llvm-project/issues/61183

Differential Revision: https://reviews.llvm.org/D145890

This reverts commit 1ef4c3c8.

Two buildbots still haven't been updated.

The mapping from ELF relocation types to JITLink x86-64 edge kinds is 1-1, so
we don't need the intermediate step.

Some C++ DR tests are disrupting the usual ascending order, introducing confusion, and making people put new tests in the wrong place (myself included). This change affects 15 tests out of roughly 700.

Fix https://llvm.org/PR49498.
The check notices 2 sides of a conditional operator have types with a different constness and so tries to examine the implicit cast.
As one side is infinity, the float narrowing detection sees when its casted to a double(which it already was) it thinks the result is out of range.

I've fixed this by just disregarding expressions where the builtin type(without quals) match as no conversion would take place.

However this has opened a can of worms. Currenty `float a = std::numeric_limits<double>::infinity();` is marked as narrowing.
Whats more suspicious is `double a = std::numeric_limits<float>::infinity();` is also marked as narrowing.
It could be argued `double inf -> float inf` is narrowing, but `float inf -> double inf` definitely isnt.

Reviewed By: carlosgalvezp

Differential Revision: https://reviews.llvm.org/D98416

There is no type named sliceExpr, so it's likely a misspelling of
either slice_array or __slice_expr. Neither of which appear to
need the friend declaration.

This may indicate a unimplemented bit of `<valarray`, but ¯\_(ツ)_/¯
Nobody uses it anyway.

(Commit message provided by @EricWF.)

This reverts commit e13c43b2.

This reverts commit d5193e34.

Reverted at @EricWF's request so it can be relanded with a better commit
message.

This reverts commit 9ece8753.

Check detects cyclic #include dependencies between user-defined headers.

Reviewed By: njames93

Differential Revision: https://reviews.llvm.org/D144828

Summary:
This patch adds `fetch_or`. It is mainly useful for querying and set a
bit's status without using the more expensive `exchange` functions.

Also fix ordering in Release Notes.

Differential Revision: https://reviews.llvm.org/D148424

The class is never defined. This was discovered while validating modules
in libc++.

Reviewed By: #libc, philnik

Differential Revision: https://reviews.llvm.org/D148357

This reverts commit 92523a35.

Reland to see whether CIs are updated.

Per our policy we only support the last two releases.

Reviewed By: #libc, EricWF, philnik

Differential Revision: https://reviews.llvm.org/D148359

Using 'hasUnqualifiedDesugaredType' to skip all type aliases when
checking for handle.

Fixes #38779

Reviewed By: carlosgalvezp

Differential Revision: https://reviews.llvm.org/D148418

Added IgnoreTemplateInstantiations option to modernize-use-override
check. Allows to ignore virtual function overrides that are part of
template instantiations. This can be useful in cases where the use
of the "override" keyword is not appropriate or causes issues with
template specialization.

Fixes #38276.

Reviewed By: carlosgalvezp

Differential Revision: https://reviews.llvm.org/D147924

[[https://wg21.link/p1787 | P1787]]: CWG2331 is resolved by defining lookup from complete-class contexts and out-of-line member definitions.
Wording: The declaration set is the result of a single search in the scope of C for N from immediately after the class-specifier of C if P is in a complete-class context of C or from P otherwise. ([class.member.lookup]/4)

Reviewed By: #clang-language-wg, shafik

Differential Revision: https://reviews.llvm.org/D148260

[clang-tidy] Apply cppcoreguidelines-avoid-capture-default-when-capturin-this only to by-value capture default

Since Cpp Core Guidelines have accepted the change in the rules:
https://github.com/isocpp/CppCoreGuidelines/commit/3c90d590e138c3a1e4eb59234e410e00545326de

Also rename the check accordingly.

Differential Revision: https://reviews.llvm.org/D148340

Both had diverged in a few ways, so this brings them both back in sync.

Differential Revision: https://reviews.llvm.org/D148351

The optimization of (mul x, c) to (ADD (SLLI x, i0), (SLLI x, i1))
is only enabled for i32 multiplication on rv64, because of
the regression in i64 multiplication on rv32.

However we can change the condition to that the immediate 'c'
should only be used once, then the above regression can also be
avoided, and ohter chances of optimization can be enabled.

Reviewed By: craig.topper

Differential Revision: https://reviews.llvm.org/D147410

Support all floating point representations for an 0.5 floating const.
Added missing tests.

Depend on D147906.

Reviewed By: carlosgalvezp

Differential Revision: https://reviews.llvm.org/D147908

Using APFloat to compare floating numbers instead of float/double.

Fixes: #46424

Reviewed By: carlosgalvezp

Differential Revision: https://reviews.llvm.org/D147906