Jamie Schmeiser [Mon, 5 Dec 2022 19:11:15 +0000 (14:11 -0500)]
Reland "A new hidden option exec-on-ir-change=exe that calls exe each time IR changes"
Summary:
This relands commit
dff0e8b4ff13af311512c369d059f1e095e83a60. The test is now
guarded with a lit.local.cfg that ensures /bin/cat is available. Also, the
code has been updated to match changes made to relevant code.
Author: Jamie Schmeiser <schmeise@ca.ibm.com>
Reviewed By:aeubanks (Arthur Eubanks)
Differential Revision: https://reviews.llvm.org/D110776
LLVM GN Syncbot [Mon, 5 Dec 2022 18:56:16 +0000 (18:56 +0000)]
[gn build] Port
5ecd36329508
Jonas Paulsson [Mon, 5 Dec 2022 14:14:40 +0000 (08:14 -0600)]
Reapply "[CodeGen] Add new pass for late cleanup of redundant definitions."
This reverts commit
122efef8ee9be57055d204d52c38700fe933c033.
- Patch fixed to not reuse definitions from predecessors in EH landing pads.
- Late review suggestions (by MaskRay) have been addressed.
- M68k/pipeline.ll test updated.
- Init captures added in processBlock() to avoid capturing structured bindings.
- RISCV has this disabled for now.
Original commit message:
A new pass MachineLateInstrsCleanup is added to be run after PEI.
This is a simple pass that removes redundant and identical instructions
whenever found by scanning the MF once while keeping track of register
definitions in a map. These instructions are typically immediate loads
resulting from rematerialization, and address loads emitted by target in
eliminateFrameInde().
This is enabled by default, but a target could easily disable it by means of
'disablePass(&MachineLateInstrsCleanupID);'.
This late cleanup is naturally not "optimal" in removing instructions as it
is done by looking at phys-regs, but still quite effective. It would be
desirable to improve other parts of CodeGen and avoid these redundant
instructions in the first place, but there are no ideas for this yet.
Differential Revision: https://reviews.llvm.org/D123394
Reviewed By: RKSimon, foad, craig.topper, arsenm, asb
Peter Klausler [Fri, 18 Nov 2022 00:30:49 +0000 (16:30 -0800)]
[flang] Catch attempts to copy pointers in allocatables in PURE
In a pure context, a pointer acquired from an INTENT(IN) dummy argument
may not be copied. Catch the case in which the pointer is a component
of an allocatable component at some depth of nesting.
(This patch adds a new component iterator kind that is a variant of
a potential subobject component iterator; it visits all potential
subobject components, plus pointers, into which it does not descend.)
Differential Revision: https://reviews.llvm.org/D139161
Michael Buch [Mon, 5 Dec 2022 18:12:56 +0000 (18:12 +0000)]
[lldb][Test] TestRerunAndExprDylib.py on Linux: dlclose solib to force destruction module
Previously we didn't properly trigger the destructor of
the `lldb_private::Module` backing `libfoo.so`. So the newly
rebuilt version wouldn't actually be loaded on a program re-run.
The test expects the fresh module to be loaded.
Roman Lebedev [Mon, 5 Dec 2022 18:10:54 +0000 (21:10 +0300)]
[NFC] Port all runlines for SimplifyCFG pass tests to -passes syntax
Matt Arsenault [Fri, 25 Nov 2022 21:52:01 +0000 (16:52 -0500)]
AMDGPU: Convert a test to generated checks
Also switch to not using kernels to test, since it shouldn't matter
for these addressing modes. This should reduce some of the diffs
between subtargets, but it's still not enough to share the same
checks.
The test could use some additional modernization. Probably should also
round this out with the new maximium offsets.
Peter Klausler [Wed, 16 Nov 2022 00:23:13 +0000 (16:23 -0800)]
[flang] Fold the RHS of assignment
In order to emit overflow warnings from assignment statements whose
right-hand sides are constants that undergo conversions, run the
right-hand sides of assignments through constant folding after the
conversions have been made explicit in expression analysis.
Differential Revision: https://reviews.llvm.org/D139151
Valentin Clement [Mon, 5 Dec 2022 17:28:13 +0000 (18:28 +0100)]
[flang] Fix pointer association with remap on polymorphic entities
Runtime is expecting a 1d array. This patch fixes the generation
of the array holding the bounds to be passed to the runtime function call.
Reviewed By: jeanPerier, PeteSteinfeld
Differential Revision: https://reviews.llvm.org/D139324
Valentin Clement [Mon, 5 Dec 2022 17:26:16 +0000 (18:26 +0100)]
[flang] Retrieve rank before updating the pointer
The code is iterating on the rank of the pointer to set the bounds.
If the rank is retrieved after the `pointer = target` it does not
reflect the actual rank of the pointer.
This could happen in code like the following:
```
type t1
integer :: a
end type
type(t), pointer :: p(:)
class(t), pointer :: q(:,:)
q(0:1,-2:2) => p(10:1:-1)
```
Reviewed By: klausler
Differential Revision: https://reviews.llvm.org/D139327
Jamie Schmeiser [Mon, 5 Dec 2022 17:10:53 +0000 (12:10 -0500)]
Expand loop peeling phi computation to handle binary ops and casts
Summary:
Expand the capabilities of the code for computing how many peels are
needed to make phis determined. A cast gets the peel count for the
value being casted while a binary op gets the maximum of the operands.
Respond to review comments: remove redundant asserts.
Author: Jamie Schmeiser <schmeise@ca.ibm.com>
Reviewed By:mkazantsev (Max Kazantsev),syzaara (Zaara Syeda)
Differential Revision: https://reviews.llvm.org/D138719
Navdeep Katel [Mon, 5 Dec 2022 10:44:56 +0000 (16:14 +0530)]
Support `transpose` mode for `gpu.subgroup` WMMA ops
Add support for loading, computing, and storing `gpu.subgroup` WMMA ops
in transpose mode as well. Update the GPU to NVVM lowerings to support
`transpose` mode and update integration tests as well.
Reviewed By: ThomasRaoux
Differential Revision: https://reviews.llvm.org/D139021
Paul Walker [Mon, 21 Nov 2022 16:56:55 +0000 (16:56 +0000)]
[InstCombine] Increase test coverage of vector.reverse ready for follow on work.
Matt Arsenault [Mon, 5 Dec 2022 16:45:17 +0000 (11:45 -0500)]
AMDGPU: Bulk update memory legalizer tests to use opaque pointers
Michael Buch [Fri, 25 Nov 2022 14:45:09 +0000 (14:45 +0000)]
Reland "[lldb][Target] Flush the scratch TypeSystem when owning lldb_private::Module gets unloaded"
This relands commit `
71f3cac7895ad516ec25438f803ed3c9916c215a`
Fixes LLDB Linux bots and improves TypeSystem flushing for shared libraries.
Differential Revision: https://reviews.llvm.org/D138724
Matt Arsenault [Mon, 4 Jun 2018 10:12:39 +0000 (13:12 +0300)]
DAG: ComputeNumSignBits from load range metadata
The cases where the result type doesn't match the range type
are inadequately tested, but I'm not sure how to write such a
test. During the pre-legalize combine, any obviously optimizable
code gets handled so it's harder to test legalized extloads.
Matt Arsenault [Sat, 12 Nov 2022 18:55:48 +0000 (10:55 -0800)]
AMDGPU: Add baseline tests for known sign bits of load range metadata
Philip Reames [Mon, 5 Dec 2022 16:49:42 +0000 (08:49 -0800)]
[SDAG] Allow scalable vectors in ComputeKnownBits (try 2)
This was previously reverted due to a hang on a Hexagon bot. This turned out to be a bug in the Hexagon backend around how splat_vectors are legalized (which they're using for fixed length vectors!). I adjusted this patch to remove the implicit truncate support. This hides the hexagon bug for now, and unblocks the rest of the change.
Original commit message:
This is the SelectionDAG equivalent of D136470, and is thus an alternate patch to D128159.
The basic idea here is that we track a single lane for scalable vectors which corresponds to an unknown number of lanes at runtime. This is enough for us to perform lane wise reasoning on many arithmetic operations.
This patch also includes an implementation for SPLAT_VECTOR as without it, the lane wise reasoning has no base case. The original patch which inspired this (D128159), also included STEP_VECTOR. I plan to do that as a separate patch.
Differential Revision: https://reviews.llvm.org/D137140
Florian Hahn [Mon, 5 Dec 2022 16:44:19 +0000 (16:44 +0000)]
[ConstraintElim] Queue facts and checks directly.
This allows interleaving facts and checks in a single block. In
particular this enables using facts from assumes for conditions in the
same block that come after the assume.
This could be extended to only try to simplify checks at the point where
a condition is used.
Reviewed By: nikic
Differential Revision: https://reviews.llvm.org/D138452
Matt Arsenault [Sat, 26 Nov 2022 22:23:01 +0000 (17:23 -0500)]
PGOProfile: Only keep opaque pointer tests
The typed pointer tests were already converted, these were identical
now.
Matt Arsenault [Sat, 26 Nov 2022 20:43:31 +0000 (15:43 -0500)]
PredicateInfo: Convert test to opaque pointers
Drops the "has predicate info" for the first block.
Paul Walker [Mon, 21 Nov 2022 17:35:49 +0000 (17:35 +0000)]
[NFC][PatternMatch] Add helper for m_Intrinsic<Intrinsic::experimental_vector_reverse>.
Florian Hahn [Mon, 5 Dec 2022 16:32:45 +0000 (16:32 +0000)]
[ConstraintElim] Add test subtracting SIGNED_MIN.
chenglin.bi [Mon, 5 Dec 2022 16:31:58 +0000 (00:31 +0800)]
[Instcombine] Precommit tests for some or canonicalization; NFC
~((A & B) ^ (A | ?)) -> (A & B) | ~(A | ?)
~(A & B) ^ (A | ?) -> (A & B) | ~(A | ?)
Timm Bäder [Mon, 5 Dec 2022 16:30:24 +0000 (17:30 +0100)]
[clang][Interp][NFC] Move to std::optional
chenglin.bi [Mon, 5 Dec 2022 16:23:28 +0000 (00:23 +0800)]
[Instcombine] Precommit tests for D139080; NFC
Test pattern for
(C & X) | ~(C | Y) -> C ? X : ~Y
Samuel Parker [Mon, 5 Dec 2022 16:01:36 +0000 (16:01 +0000)]
[NFC][WebAssembly] Add codegen tests
Matt Arsenault [Sun, 27 Nov 2022 18:47:09 +0000 (13:47 -0500)]
ExecutionEngine: Convert tests to opaque pointers
chenglin.bi [Mon, 5 Dec 2022 16:00:54 +0000 (00:00 +0800)]
[InstCombine] Precommit tests for D139253; NFC
Nico Weber [Fri, 2 Dec 2022 14:29:18 +0000 (09:29 -0500)]
[lld/mac] Fix --start-lib/--end-lib with split thinlto inputs
Fixes #59162. The test has a comment explaining what's going on.
See also Symbol::extract() in lld/ELF/Symbols.cpp.
The included test sadly also passes if I pass just
bd448f01a62,
while doing that isn't enough to make my bigger repro case work
(if I port just that, something else asserts later on, but with
this fix here everything's fine in my bigger repro).
Differential Revision: https://reviews.llvm.org/D139199
Michał Górny [Mon, 5 Dec 2022 15:47:17 +0000 (16:47 +0100)]
Revert "[test] Fix CodeGen/M68k/pipeline.ll after D123394 MachineLateInstrsCleanupPass"
This reverts commit
f55880e830e150d98e5340cdc3c4c41867a5514d.
The original change was reverted.
Sjoerd Meijer [Mon, 5 Dec 2022 15:36:24 +0000 (15:36 +0000)]
[LoopFlatten] Precommit test for GH issue #59339. NFC.
Sjoerd Meijer [Mon, 5 Dec 2022 15:34:23 +0000 (15:34 +0000)]
[LoopFlatten] Add some LLVM_DEBUG messages. NFC.
Peter Klausler [Fri, 2 Dec 2022 15:19:49 +0000 (07:19 -0800)]
[flang] Map symbols in expressions when copying interface symbols
Given a MODULE SUBROUTINE or MODULE FUNCTION interface followed
later by a corresponding separate module subprogram definition in a
MODULE PROCEDURE, the copies of the interface's dummy argument and
function result symbols that populate the initial scope of that
MODULE PROCEDURE need to have any symbol references in their types
or bounds adjusted to point to their new counterparts.
Differential Revision: https://reviews.llvm.org/D139200
Samira Bazuzi [Mon, 5 Dec 2022 15:30:29 +0000 (07:30 -0800)]
[flang][NFC] Add a no-arg constructor for `Verbatim`.
In C++20, types that declare or delete any constructors are no longer aggregates, breaking compilation of many existing uses of aggregate initialization.
Although `Verbatim` declares itself to not have a no-arg default constructor, this is circumvented in `basic-parsers.h` which returns a `RESULT{}` a.k.a. `Verbatim{}`. Adding the no-arg constructor while still deleting the copy/assignment constructors maintains the current state and also supports eventually building this in c++20 mode.
Fix suggested in https://discourse.llvm.org/t/build-failure-when-attempting-to-build-flang-with-c-20/66953.
Reviewed By: klausler
Differential Revision: https://reviews.llvm.org/D139228
Florian Hahn [Mon, 5 Dec 2022 15:25:07 +0000 (15:25 +0000)]
[ConstraintElim] Add test with implicit/explicit GEP index truncate.
The implicit index truncate example is currently being miscompiled.
Sjoerd Meijer [Mon, 5 Dec 2022 14:52:40 +0000 (14:52 +0000)]
Revert "Recommit "[LoopFlatten] Enable it by default""
This reverts commit
3ea6a9a469fde168c527b1c34c09f6d684ec86af because of the
reported miscompilation in: https://github.com/llvm/llvm-project/issues/59339
bipmis [Mon, 5 Dec 2022 15:08:31 +0000 (15:08 +0000)]
[AAch64] Optimize muls with operands having enough sign bits.
Muls with 64bit operands where each of the operand is having more than 32 sign bits, we can generate a single smull instruction on a 32bit operand.
Differential Revision: https://reviews.llvm.org/D138817
Matt Arsenault [Sun, 4 Dec 2022 00:40:06 +0000 (19:40 -0500)]
ValueTracking: Teach canCreateUndefOrPoison about more intrinsics
I tried to test the fallthrough to noundef callsite return attribute
case, but it seems that folds out as-is.
Emmmer [Sat, 3 Dec 2022 13:49:50 +0000 (21:49 +0800)]
[LLDB][RISCV] Add RV64F instruction support for EmulateInstructionRISCV
Reviewed By: DavidSpickett
Differential Revision: https://reviews.llvm.org/D139294
Matt Arsenault [Sun, 4 Dec 2022 00:37:57 +0000 (19:37 -0500)]
ValueTracking: Add some baseline intrinsic with freeze tests
LLVM GN Syncbot [Mon, 5 Dec 2022 14:27:33 +0000 (14:27 +0000)]
[gn build] Port
dbe8c2c316c4
Dmitry Vyukov [Mon, 5 Dec 2022 14:12:37 +0000 (15:12 +0100)]
compiler-rt/test/metadata: restrict tests to x86_64
aarch64 builder broke with:
-- Supported architectures for crt: aarch64
CMake Error at compiler-rt/cmake/config-ix.cmake:244 (message):
Unsupported architecture: x86_64
Call Stack (most recent call first):
compiler-rt/cmake/config-ix.cmake:280 (get_target_flags_for_arch)
compiler-rt/test/metadata/CMakeLists.txt:7 (get_test_cc_for_arch)
-- Configuring incomplete, errors occurred!
https://lab.llvm.org/buildbot/#/builders/223/builds/11423
Restrict the tests to x86_64 only.
Reviewed By: melver
Differential Revision: https://reviews.llvm.org/D139323
Roman Lebedev [Mon, 5 Dec 2022 14:21:29 +0000 (17:21 +0300)]
[NFC] Fix test/CodeGen/PowerPC/O0-pipeline.ll
Matt Arsenault [Sat, 3 Dec 2022 19:25:43 +0000 (14:25 -0500)]
clang: Convert builtins test to opaque pointers
Joseph Huber [Mon, 5 Dec 2022 13:48:30 +0000 (07:48 -0600)]
[LinkerWrapper] Fix use-after-free when using JIT
Summary:
This portion of the code was unused until now and wasn't properly
updated after we moved to using `StringRef`.
Dmitry Vyukov [Mon, 17 Oct 2022 13:13:56 +0000 (15:13 +0200)]
Use-after-return sanitizer binary metadata
Currently per-function metadata consists of:
(start-pc, size, features)
This adds a new UAR feature and if it's set an additional element:
(start-pc, size, features, stack-args-size)
Reviewed By: melver
Differential Revision: https://reviews.llvm.org/D136078
Matt Arsenault [Sun, 4 Dec 2022 03:49:47 +0000 (22:49 -0500)]
ValueTracking: Teach CannotBeOrderedLessThanZero about trivial ops
Handle canonicalize and arithmetic.fence
Matt Arsenault [Sun, 4 Dec 2022 03:28:57 +0000 (22:28 -0500)]
ValueTracking: Teach isKnownNeverNaN about arithmetic_fence
Matt Arsenault [Sun, 4 Dec 2022 03:25:34 +0000 (22:25 -0500)]
ValueTracking: Teach isKnownNeverInfinity about arithmetic.fence
Matt Arsenault [Sun, 4 Dec 2022 15:54:44 +0000 (10:54 -0500)]
Intrinsics: Fix not speculating llvm.fptrunc.round
Move its definition to be with the other FP rounding intrinsics.
Vladislav Dzhidzhoev [Mon, 20 Jun 2022 12:49:08 +0000 (15:49 +0300)]
[GlobalISel][DebugInfo] Propagate debug location for localized constants
After IRTranslator pass, constants are deduplicated and translated into instructions at entry block, having debug locations lost.
Localization of constants may cause emission of extra zero lines in debug_line section, like here https://godbolt.org/z/ecvsxxfKn. In this example, constant gets placed as
a first instruction in entry block, and despite it has no debug location, AsmPrinter emits zero line for it.
If a localized constant has the only user, we can assume that it has the same debug location as its user, since they are placed consequently.
Differential Revision: https://reviews.llvm.org/D128192
Juan Manuel MARTINEZ CAAMAÑO [Mon, 5 Dec 2022 12:30:51 +0000 (06:30 -0600)]
[NFC][Clang][Driver][AMDGPU] Avoid temporary copies of std::string by using Twine and StringRef
Reviewed By: tra
Differential Revision: https://reviews.llvm.org/D139023
Dmitry Preobrazhensky [Mon, 5 Dec 2022 13:03:17 +0000 (16:03 +0300)]
[AMDGPU][MC] Correct handling of mandatory literals
Differential Revision: https://reviews.llvm.org/D138661
Haojian Wu [Mon, 5 Dec 2022 13:18:33 +0000 (14:18 +0100)]
An iwyu cleanup in flang/Frontend/TargetOptions.h
Sander de Smalen [Mon, 5 Dec 2022 12:06:10 +0000 (12:06 +0000)]
[AArch64][SME] Avoid going through memory for streaming-compatible splats
Reviewed By: david-arm, paulwalker-arm
Differential Revision: https://reviews.llvm.org/D139111
Florian Hahn [Mon, 5 Dec 2022 12:19:09 +0000 (12:19 +0000)]
[ConstraintElim] Update comment after recent changes.
Update the comment to reflect the current code.
Guray Ozen [Fri, 2 Dec 2022 15:34:58 +0000 (16:34 +0100)]
[mlir] Fix infinite loop in collapse
Incrementing `counter` variable is inside the if statement. If the code does not enter there, the while loop will iterate infinitely. This revision moves the codes outside of if statement.
Reviewed By: mravishankar
Differential Revision: https://reviews.llvm.org/D139005
luxufan [Mon, 5 Dec 2022 11:38:25 +0000 (19:38 +0800)]
[SCCP] Propagate equality of a not-constant
The equality state of a not-constant can be used to do constant
propagation. For example,
```
define i32 @equal_not_constant(ptr noundef %p, ptr noundef %q) {
entry:
%cmp = icmp ne ptr %p, null
br i1 %cmp, label %if.then, label %if.end
if.then: ; preds = %entry
%cmp.then = icmp eq ptr %p, %q
br i1 %cmp.then, label %if.then1, label %if.end
if.then1: ; preds = %if.then
%cmp.then1 = icmp ne ptr %q, null
call void @use(i1 %cmp.then1)
br label %if.end
if.end:
ret i32 0
}
```
In this case, we can fold `%cmp.then1` as `true`.
Reviewed By: nikic
Differential Revision: https://reviews.llvm.org/D139289
Petar Avramovic [Mon, 5 Dec 2022 10:49:20 +0000 (11:49 +0100)]
AMDGPU/MC: Simplify AsmParser for VOP3P
Remove check that instruction has src_modifiers because
isRegOrImmWithInputMods essentially includes this check.
isMod is no longer used after the refactor. Also, was unreachable case
inside for loop. clamp and omod operands are handled after the loop.
Differential Revision: https://reviews.llvm.org/D139194
David Spickett [Mon, 5 Dec 2022 11:33:29 +0000 (11:33 +0000)]
[LLVM][Runtimes] Surround unknown runtime with ""
So that when you do something like:
-DLLVM_ENABLE_RUNTIMES=;compiler-rt;libcxx;libcxxabi;libunwind
You get:
Runtime "" is not a supported runtime. Supported runtimes are:
Instead of:
Runtime is not a supported runtime. Supported runtimes are:
Which implies that "Runtime" is the name of the runtime you
tried to build, but in fact it was "".
Haojian Wu [Mon, 5 Dec 2022 11:18:58 +0000 (12:18 +0100)]
[include-cleaner] Add two command-line flags to the include-cleaner
category.
Tomas Matheson [Mon, 5 Dec 2022 11:09:03 +0000 (11:09 +0000)]
Revert "[AArch64] Improve TargetParser API"
This reverts commit
e83f1502f1be7a2a3b9a33f5a73867767e78ba6b.
Did not build with C++20 and caused problems with dynamic libs.
Nikita Popov [Wed, 9 Nov 2022 11:09:03 +0000 (12:09 +0100)]
[SCCP] Add support for with.overflow intrinsics
This adds SCCP support for extractvalues of with.overflow.
We compute both the range of the result value and determine when
the overflow value is always false.
Differential Revision: https://reviews.llvm.org/D137713
Michael Buch [Mon, 5 Dec 2022 10:47:08 +0000 (10:47 +0000)]
[lldb][Test] TestModulesCompileError.py: fix expected error string
Max Kazantsev [Mon, 5 Dec 2022 09:35:18 +0000 (16:35 +0700)]
[NFC] Rename variable MaxBECount -> ConstantMaxBECount
Just to distinguish it from symbolic max which we plan to compute
here as well.
Simon Pilgrim [Mon, 5 Dec 2022 10:39:49 +0000 (10:39 +0000)]
Fix MSVC "not all control paths return a value" warning. NFC.
Adrian Kuegel [Mon, 5 Dec 2022 10:22:20 +0000 (11:22 +0100)]
[mlir][Tensor] Apply ClangTidy performance finding (NFC)
Tiehu Zhang [Mon, 5 Dec 2022 07:50:30 +0000 (15:50 +0800)]
[AArch64][SVE2] Add patterns for eor3
Add patterns for:
eor x, (eor y, z) -> eor3 x, y, z
Reviewed By: dmgreen, sdesmalen
Differential Revision: https://reviews.llvm.org/D138793
John McIver [Mon, 5 Dec 2022 10:13:18 +0000 (10:13 +0000)]
[NFC][clang] Strengthen checks in matrix-type-operators.c
* Add tbaa attribute checks
* Add end-of-line check to load instructions
John McIver [Mon, 5 Dec 2022 10:12:08 +0000 (10:12 +0000)]
[NFC][clang] Strengthen checks in matrix-type-operators.cpp
* Add tbaa attribute checks
* Add end-of-line check to load instructions
Valentin Clement [Mon, 5 Dec 2022 09:32:08 +0000 (10:32 +0100)]
[flang][NFC] Remove implemented TODOs
Max Kazantsev [Mon, 5 Dec 2022 08:56:23 +0000 (15:56 +0700)]
[Test] Add 2 more simplified test with missing opts
These are simplified versions of one existing test, but we cannot deal
with them either.
Valentin Clement [Mon, 5 Dec 2022 08:35:50 +0000 (09:35 +0100)]
[flang] Pointer assignment with remapping involcing polymorphic entities
Lower pointer assignment with remapping involving polymorphic entities
to runtime call to PointerAssociateRemapping.
For the time being all pointer assignment involcing polymorphic entities are
done with the runtime call. When lhs is not unlimited polymorphic
we might be able to do it inlined as well.
Reviewed By: jeanPerier, PeteSteinfeld
Differential Revision: https://reviews.llvm.org/D139198
Nikita Popov [Mon, 14 Nov 2022 15:13:51 +0000 (16:13 +0100)]
[LAA] Use cross-iteration alias analysis
LAA analyzes cross-iteration memory dependencies, as such AA should
not make assumptions about equality of values inside the loop, as
they may come from different iterations.
Fix this by exposing the MayBeCrossIteration AA flag and enabling
it for LAA.
Differential Revision: https://reviews.llvm.org/D137958
Matthias Springer [Mon, 5 Dec 2022 08:16:05 +0000 (09:16 +0100)]
[mlir][tensor] Fold rank-reducing insert_slice with inverse collapse_shape
Differential Revision: https://reviews.llvm.org/D139221
Matthias Springer [Mon, 5 Dec 2022 08:15:52 +0000 (09:15 +0100)]
[mlir][tensor] Fold rank-reducing extract_slice with inverse expand_shape
Differential Revision: https://reviews.llvm.org/D139220
Fangrui Song [Mon, 5 Dec 2022 08:09:23 +0000 (08:09 +0000)]
[LegacyPM] Remove dead declarations
Jean Perier [Mon, 5 Dec 2022 08:05:37 +0000 (09:05 +0100)]
[flang] Lower function return to HLFIR
The only special thing that is needed is to update the bridge symbol
lookup to deal with the HLFIR symbol lookup (symbols are mapped to
fir::FortranVariableInterface operations, not Fortran::Lower::SymbolBox).
Differential Revision: https://reviews.llvm.org/D139201
Fangrui Song [Mon, 5 Dec 2022 07:54:11 +0000 (07:54 +0000)]
[PowerPC] Remove unused MacOSX check
PPC Darwin support was removed long ago.
Fangrui Song [Mon, 5 Dec 2022 07:51:17 +0000 (07:51 +0000)]
[LegacyPM] Remove AnnotationRemarksLegacyPass
Following recent changes to remove non-core features of the legacy
PM/optimization pipeline. This is a new pass (2020-11) to generate remarks from
!annotation metadata (e.g. -ftrivial-auto-var-init=).
Nikita Popov [Tue, 8 Nov 2022 11:26:34 +0000 (12:26 +0100)]
[PowerPC] Use default attributes for intrinsics
This switches a large subset of PowerPC intrinsics to use default
attributes (nosync, nofree, nocallback and willreturn). In
particular the presence of willreturn is important to avoid
optimization regression in the future.
This patch primarily covers readnone/readonly intrinsics.
Differential Revision: https://reviews.llvm.org/D137629
Nikita Popov [Wed, 5 Oct 2022 15:13:25 +0000 (17:13 +0200)]
[llvm-c] Remove C API functions that are incompatible with opaque pointers
These deprecated functions are incompatible with opaque pointers,
and have replacements that accept an explicit type. Drop them now
as a final warning to consumers of the C API to migrate their code
(while LLVMGetElementType still exists as a temporary workaround).
Differential Revision: https://reviews.llvm.org/D135271
Vassil Vassilev [Mon, 5 Dec 2022 07:18:23 +0000 (07:18 +0000)]
[lldb] Fix the way we set up the lldb modules infrastructure.
D127284 introduced a new language option which is not benign from modules
perspective. Before this patch lldb would set up the compiler invocation and
later enable incremental processing. Post-D127284 this does not work because
the option causes a module hash mismatch for implicit modules.
In addition, D127284 enables parsing statements on the global scope if
incremental processing is on and thus `syntax_error_for_lldb_to_find` was
rightfully not recognized as a declaration and is considered a statement
which produces a slightly different diagnostic.
Thanks to Michael Buch for the help in understanding this issue. This patch
should appease the lldb bots.
More discussion available at: https://reviews.llvm.org/D127284
Nikita Popov [Mon, 5 Dec 2022 07:37:08 +0000 (08:37 +0100)]
[Kaleidoscope] Fix build after std::optional changes
Adrian Kuegel [Mon, 5 Dec 2022 07:18:00 +0000 (08:18 +0100)]
[mlir][Linalg] Apply ClangTidy fixes (NFC)
Nikita Popov [Mon, 14 Nov 2022 14:08:44 +0000 (15:08 +0100)]
[AST] Make AliasSetTracker work on BatchAA
D138014 restricted AST to work on immutable IR. This means it is
also safe to use a single BatchAA instance for the entire AST
lifetime, instead of only batching parts of individual queries.
The primary motivation for this is not compile-time, but rather
having a central place to control cross-iteration AA, which will
be used by D137958.
Differential Revision: https://reviews.llvm.org/D137955
Fangrui Song [Mon, 5 Dec 2022 07:07:19 +0000 (07:07 +0000)]
Transforms/IPO: llvm::Optional => std::optional
Vitaly Buka [Mon, 5 Dec 2022 06:29:15 +0000 (22:29 -0800)]
[test][msan] Update for noundef on retval
Fangrui Song [Mon, 5 Dec 2022 06:31:11 +0000 (06:31 +0000)]
Remove unused #include "llvm/ADT/Optional.h"
Fangrui Song [Mon, 5 Dec 2022 06:26:26 +0000 (06:26 +0000)]
[iwyu] Add Hashing.h
Kazu Hirata [Mon, 5 Dec 2022 05:43:10 +0000 (21:43 -0800)]
Forward-declare raw_ostream (NFC)
This patch adds forward declarations of raw_ostream to those header
files that are relying on the forward declaration of raw_ostream in
llvm/include/llvm/ADT/Optional.h.
I'm planning to move operator<< for Optional<T> and std::optional<T>
from Optional.h to llvm/include/llvm/Support/raw_ostream.h. Once I do
so, we no longer need to forward-declare raw_ostream in Optional.h.
Fangrui Song [Mon, 5 Dec 2022 05:08:37 +0000 (21:08 -0800)]
[ELF] Produce DT_RISCV_VARIANT_CC
https://github.com/riscv/riscv-elf-psabi-doc/pull/190 introduced STO_RISCV_VARIANT_CC.
The linker should:
* Copy the STO_RISCV_VARIANT_CC bit to .symtab/.dynsym: already fulfilled after
82ed93ea0552c8f82df05859ee93e70b71c4e65d
* Produce DT_RISCV_VARIANT_CC if at least one R_RISCV_JUMP_SLOT relocation
references a symbol with the STO_RISCV_VARIANT_CC bit. Done by this patch.
Reviewed By: kito-cheng
Differential Revision: https://reviews.llvm.org/D107951
esmeyi [Mon, 5 Dec 2022 04:57:33 +0000 (23:57 -0500)]
[NFC][XCOFF] format `XCOFFObjectWriter.cpp`
Fangrui Song [Mon, 5 Dec 2022 04:53:28 +0000 (04:53 +0000)]
[iwyu] Fix files which would break without #include Optional.h
Kazu Hirata [Mon, 5 Dec 2022 04:31:05 +0000 (20:31 -0800)]
[clang] Use std::nullopt instead of None in comments (NFC)
This is part of an effort to migrate from llvm::Optional to
std::optional:
https://discourse.llvm.org/t/deprecating-llvm-optional-x-hasvalue-getvalue-getvalueor/63716
Fangrui Song [Mon, 5 Dec 2022 04:30:54 +0000 (04:30 +0000)]
CheckedArithmetic: llvm::Optional => std::optional
Fangrui Song [Mon, 5 Dec 2022 04:21:07 +0000 (04:21 +0000)]
Remove unused #include "llvm/ADT/Optional.h"
Yeting Kuo [Sat, 19 Nov 2022 14:46:24 +0000 (22:46 +0800)]
[RISCV] Support .variant_cc directive for the assembler.
The patch is split from D103435. The patch supported a new directive .variant_cc
that annotates function with STO_RISCV_VARIANT_CC. Symbols marked with
STO_RISCV_VARIANT_CC do not use standard calling conversion or use parameter not
passed in GPR/FPR.
Related: https://github.com/riscv/riscv-elf-psabi-doc/pull/190
Initial authored by: HsiangKai
Reviewed By: MaskRay
Differential Revision: https://reviews.llvm.org/D138352
Fangrui Song [Mon, 5 Dec 2022 04:13:11 +0000 (04:13 +0000)]
[IR] llvm::Optional => std::optional
Many llvm/IR/* files have been migrated by other contributors.
This migrates most remaining files.