Stanislav Gatev [Tue, 4 Jan 2022 13:47:14 +0000 (13:47 +0000)]
[clang][dataflow] Add transfer functions for assignment
This is part of the implementation of the dataflow analysis framework.
See "[RFC] A dataflow analysis framework for Clang AST" on cfe-dev.
Differential Revision: https://reviews.llvm.org/D116596
Alex Xu (Hello71) [Mon, 10 Jan 2022 17:42:56 +0000 (09:42 -0800)]
Revert "[CodeGen] Mark fma as const for Android"
This code is intended to give a special exception for platforms which set errno in some math functions but not fma. This does not apply to Android, which does not set errno in any math functions (https://cs.android.com/android/platform/superproject/+/master:bionic/libc/include/math.h;drc=master;l=59). The correct implementation for Android is to set -fno-math-errno by default, which was done in https://reviews.llvm.org/D51068. Therefore, this special exception is no longer needed for Android. Deleting it slightly reduces code complexity, clang executable size, compile time, and test time.
This reverts
fbfba29d74748b8fffd2f7cc654fb4c3659c8f67.
Reviewed By: pirama
Differential Revision: https://reviews.llvm.org/D116755
Matt Arsenault [Sun, 9 Jan 2022 17:32:15 +0000 (12:32 -0500)]
AMDGPU/GlobalISel: Explicitly track d16 for image legalization
We were trying to guess at the original IR type for image intrinsics
after legalization to figure out if they were d16, but this didn't
work. Explicitly track if this is a d16 operation or not in the
opcode, as is done for the buffer intrinsics.
The OpenCL library is using f32 image writes with a dmask of 15 for
some reason, and this was incorrectly switching them to use d16. Fixes
image failures in the OpenCL conformance test. The equivalent dmask
for loads doesn't even select in either selector.
Alexander Shaposhnikov [Mon, 10 Jan 2022 18:53:31 +0000 (18:53 +0000)]
[cmake] Enable users to specify archive creation commands
This diff enables users to override CMAKE_C_ARCHIVE_CREATE & CMAKE_CXX_ARCHIVE_CREATE
(currently set in HandleLLVMOptions.cmake).
For example, one can specify
cmake -DCMAKE_C_ARCHIVE_CREATE="<CMAKE_AR> TDqc <TARGET> <LINK_FLAGS> <OBJECTS>" \
-DCMAKE_CXX_ARCHIVE_CREATE="<CMAKE_AR> TDqc <TARGET> <LINK_FLAGS> <OBJECTS>" ...
to make the build create thin archives instead of regular ones.
For a clean run `ninja lld` using thin archives seems to reduce the size
of the build directory from ~14GB to ~8GB
Differential revision: https://reviews.llvm.org/D116850
Matt Arsenault [Sun, 9 Jan 2022 15:25:36 +0000 (10:25 -0500)]
GlobalISel: Pass DebugLoc to getFunctionLiveInPhysReg
Fixes crash in assertion about dropping debug info.
Logan Smith [Fri, 7 Jan 2022 18:22:56 +0000 (10:22 -0800)]
[clang-tidy] Fix RenamerClangTidyChecks suggesting invalid macro identifiers
This behavior was fixed for regular identifiers in
9f3edc323a88c1a179a0a5a9dc9a87a2964c0d48, but the same fix was not applied to
macro fixits. This addresses https://github.com/llvm/llvm-project/issues/52895.
Differential Revision: https://reviews.llvm.org/D116824
Matt Arsenault [Tue, 28 Dec 2021 15:22:13 +0000 (10:22 -0500)]
AMDGPU: Avoid null check during addrspacecast lowering
If we know the source is a valid object, we do not need to insert a
null check. This misses a lot of opportunities from
metadata/attributes not tracked in codegen.
Ivan Butygin [Mon, 10 Jan 2022 14:52:41 +0000 (17:52 +0300)]
[mlir][NFC] Fully spell mlir typenames in BaseOpWithOffsetSizesAndStrides
Differential Revision: https://reviews.llvm.org/D116951
Stella Stamenova [Mon, 10 Jan 2022 18:21:12 +0000 (10:21 -0800)]
[lldb] Disable several lldb tests that are flaky on Windows
This tests have recently become flaky (flakier?) causing occasional failures in the windows lldb buildbot
Jean Perier [Mon, 10 Jan 2022 18:09:45 +0000 (19:09 +0100)]
[flang] Do not lose call in shape inquiry on function reference
Currently, something like `print *, size(foo(n,m))` was rewritten
to `print *, size(foo_result_symbol)` when foo result is a non constant
shape array. This cannot be processed by lowering or reprocessed by a
Fortran compiler since the syntax is wrong (`foo_result_symbol` is
unknown on the caller side) and the arguments are lost when they might
be required to compute the result shape.
It is not possible (and probably not desired) to make GetShape fail in
general in such case since returning nullopt seems only expected for
scalars or assumed rank (see GetRank usage in lib/Semantics/check-call.cpp),
and returning a vector with nullopt extent may trigger some checks to
believe they are facing an assumed size (like here in intrinsic argument
checks: https://github.com/llvm/llvm-project/blob/
196204c72c68a577c72af95d70f18e3550939a5e/flang/lib/Evaluate/intrinsics.cpp#L1530).
Hence, I went for a solution that limits the rewrite change to folding
(where the original expression is returned if the shape depends on a non
constant shape from a call).
I added a non default option to GetShapeHelper that prevents the rewrite
of shape inquiry on calls to descriptor inquiries. At first I wanted to
avoid touching GetShapeHelper, but it would require to re-implement all
its logic to determine if the shape comes from a function call or not
(the expression could be `size(1+foo(n,m))`). So added an alternate
entry point to GetShapeHelper seemed the cleanest solution to me.
Differential Revision: https://reviews.llvm.org/D116933
Fangrui Song [Mon, 10 Jan 2022 18:03:21 +0000 (10:03 -0800)]
[ELF] Support mixed TLSDESC and TLS GD
We only support both TLSDESC and TLS GD for x86 so this is an x86-specific
problem. If both are used, only one R_X86_64_TLSDESC is produced and TLS GD
accesses will incorrectly reference R_X86_64_TLSDESC. Fix this by introducing
SymbolAux::tlsDescIdx.
Reviewed By: ikudrin
Differential Revision: https://reviews.llvm.org/D116900
Roman Lebedev [Mon, 10 Jan 2022 17:49:41 +0000 (20:49 +0300)]
[SCEV] Sequential/in-order `UMin` expression
As discussed in https://github.com/llvm/llvm-project/issues/53020 / https://reviews.llvm.org/D116692,
SCEV is forbidden from reasoning about 'backedge taken count'
if the branch condition is a poison-safe logical operation,
which is conservatively correct, but is severely limiting.
Instead, we should have a way to express those
poison blocking properties in SCEV expressions.
The proposed semantics is:
```
Sequential/in-order min/max SCEV expressions are non-commutative variants
of commutative min/max SCEV expressions. If none of their operands
are poison, then they are functionally equivalent, otherwise,
if the operand that represents the saturation point* of given expression,
comes before the first poison operand, then the whole expression is not poison,
but is said saturation point.
```
* saturation point - the maximal/minimal possible integer value for the given type
The lowering is straight-forward:
```
compare each operand to the saturation point,
perform sequential in-order logical-or (poison-safe!) ordered reduction
over those checks, and if reduction returned true then return
saturation point else return the naive min/max reduction over the operands
```
https://alive2.llvm.org/ce/z/Q7jxvH (2 ops)
https://alive2.llvm.org/ce/z/QCRrhk (3 ops)
Note that we don't need to check the last operand: https://alive2.llvm.org/ce/z/abvHQS
Note that this is not commutative: https://alive2.llvm.org/ce/z/FK9e97
That allows us to handle the patterns in question.
Reviewed By: nikic, reames
Differential Revision: https://reviews.llvm.org/D116766
Arthur O'Dwyer [Thu, 30 Dec 2021 01:45:08 +0000 (20:45 -0500)]
[libc++] Properly handle specializations of std::is_placeholder.
Before this patch, the user needed to specialize both of
`is_placeholder<MyType>` and `is_placeholder<const MyType>`.
After this patch, only the former is needed (although the
latter is harmless if provided).
The new tests don't actually fail unless return type deduction
is used, which is a C++14 feature. Specializing `is_placeholder`
is still allowed in C++11, though.
Fixes #51095.
Differential Revision: https://reviews.llvm.org/D116388
Philip Reames [Mon, 10 Jan 2022 17:32:23 +0000 (09:32 -0800)]
Minor style tweaks following fb93659
David Goldman [Mon, 6 Dec 2021 21:33:29 +0000 (16:33 -0500)]
[clang][HeaderSearch] Support framework includes in suggestPath...
Clang will now search through the framework includes to identify
the framework include path to a file, and then suggest a framework
style include spelling for the file.
Differential Revision: https://reviews.llvm.org/D115183
Kazu Hirata [Mon, 10 Jan 2022 17:21:59 +0000 (09:21 -0800)]
Revert "[clang] Remove redundant member initialization (NFC)"
This reverts commit
80e2c587498a7b2bf14dde47a33a058da6e88a9a.
The original patch causes a lot of warnings on gcc like:
llvm-project/clang/include/clang/Basic/Diagnostic.h:1329:3: warning:
base class ‘class clang::StreamingDiagnostic’ should be explicitly
initialized in the copy constructor [-Wextra]
Bryce Wilson [Mon, 10 Jan 2022 17:08:55 +0000 (09:08 -0800)]
[instcombine] Add align return attributes for operator new(..., align_val)
(Split from original patch to separate non-NFC part and add coverage. I typoed when adding the new test, so this change includes the typo fix to let libfunc recongize the signature. Didn't figure it was worth another separate commit.)
Differential Revision: https://reviews.llvm.org/D116851 (part 2 of 2)
Bryce Wilson [Mon, 10 Jan 2022 16:58:44 +0000 (08:58 -0800)]
[MemoryBuiltins] Add field for alignment argument [NFC]
There are a few places where the alignment argument for AlignedAllocLike functions was previously hardcoded. This patch adds an getAllocAlignment function and a change to the MemoryBuiltin table to allow alignment arguments to be found generically.
This will shortly allow alignment inference on operator new's with align_val params and an extension to Attributor's HeapToStack. The former will follow shortly - I split Bryce's patch for purpose of having the large change be NFC. The later will be reviewed separately.
Differential Revision: https://reviews.llvm.org/D116851 (part 1 of 2)
Craig Topper [Mon, 10 Jan 2022 16:57:38 +0000 (08:57 -0800)]
[RISCV] Use FP ABI on some of the FP tests to reduce the number of CHECK lines. NFC
These tests are interested in the FP instructions being used, not
the conversions needed to pass the arguments/returns in GPRs.
Reviewed By: asb
Differential Revision: https://reviews.llvm.org/D116869
Simon Pilgrim [Mon, 10 Jan 2022 17:03:43 +0000 (17:03 +0000)]
[ConstantFolding] Clean up Intrinsics::abs undef handling
Match cttz/ctlz handling by assuming C1 == 0 if C1 != 1 - I've added an assertion as well.
Fixes static analyzer nullptr dereference warnings.
Nico Weber [Mon, 10 Jan 2022 17:01:19 +0000 (12:01 -0500)]
Revert "[clangd] Enable expand-auto for decltype(auto)."
This reverts commit
37ec65e1d705f56fe5551de1dfcbac1e071588a2.
Its prerequisite
55d96ac3dc56bdebea854952a724c2a50d96ce19 wsa
reverted in
c2293bc17dd09d552c5fdd13ff2b7c5738c5a10a. c2293bc's
patch description claimed that it reverted 37ec65 as well,
but it apparently didn't.
See https://reviews.llvm.org/D116921#3231802
William S. Moses [Fri, 7 Jan 2022 22:26:38 +0000 (17:26 -0500)]
[MLIR] Generalize select to arithmetic canonicalization
Given a select whose result is an i1, we can eliminate the conditional in the select completely by adding a few arithmetic operations.
Reviewed By: ftynse
Differential Revision: https://reviews.llvm.org/D116839
Kazu Hirata [Mon, 10 Jan 2022 16:49:46 +0000 (08:49 -0800)]
[AMDGPU] Fix an unused variable warning (NFC)
This patch fixes:
llvm/lib/Target/AMDGPU/AMDGPULegalizerInfo.cpp:2245:12: error:
unused variable 'Ins' [-Werror,-Wunused-variable]
Philip Reames [Mon, 10 Jan 2022 16:38:40 +0000 (08:38 -0800)]
Add test coverage for D116851
Philip Reames [Mon, 10 Jan 2022 16:25:20 +0000 (08:25 -0800)]
[instcombine] Infer alignment for aligned_alloc with potentially zero size
This change removes a previous restriction where we had to prove the allocation performed by aligned_alloc was non-zero in size before using the align parameter to annotate the result. I believe this was conservatism around the C11 specification of this routine which allowed UB when size was not a multiple of alignment, but if so, it was a partial one at best. (ex: align 32, size 16 was equally UB, but not restricted) The spec has since been clarified to require nullptr return, not UB.
A nullptr - the documented return for this function on failure for all cases after UB mentioned above was removed - is trivially aligned for any power of two. This isn't totally new behavior even for this transform, we'd previously annotate potentially failing allocs (e.g. huge sizes) meaning we were putting align on potentially null pointers anyways. This change simpy does the same for all failure modes.
Arthur O'Dwyer [Mon, 10 Jan 2022 16:26:29 +0000 (11:26 -0500)]
[libc++] libcxx/utils: s/preambule/preamble/g. NFC.
Simon Pilgrim [Mon, 10 Jan 2022 16:28:04 +0000 (16:28 +0000)]
[SemaDecl] Use castAs<> instead of getAs<> to avoid dereference of nullptr
This will assert the cast is correct instead of returning nullptr
James Y Knight [Mon, 10 Jan 2022 16:24:20 +0000 (16:24 +0000)]
Jeremy Morse [Mon, 10 Jan 2022 12:11:15 +0000 (12:11 +0000)]
[Dexter] Allow DexUnreachable in supplementary .dex files
DexUnreachable is a useful tool for specifying that lines shouldn't be
stepped on. Right now they have to be placed in the source file; lets allow
them to be placed instead in a detached .dex file, by adding on_line and
line-range keyword arguments to the command.
Differential Revision: https://reviews.llvm.org/D115449
Roman Lebedev [Mon, 10 Jan 2022 16:17:32 +0000 (19:17 +0300)]
[NFC] clang-format the whole ScalarEvolutionExpressions.h
This file has completely wrong formatting,
and modifying it leads to having to fight around that. every time.
This is a pure reformatting, there are *NO* other changes here.
Johannes Doerfert [Mon, 10 Jan 2022 15:53:12 +0000 (09:53 -0600)]
[Attributor][FIX] Ensure "IsExact" is false for non-exact accesses
If we look at potentially interfering accesses we need to ensure the
"IsExact" flag is set appropriately. Accesses that have an "unknown"
size or offset cannot be exact matches and we missed to flag that.
Error and test reported by Serguei N. Dmitriev.
Simon Pilgrim [Mon, 10 Jan 2022 15:34:24 +0000 (15:34 +0000)]
[PGOInstrumentation] populateEHOperandBundle - earlyout if !isa<CallBase>
All paths (that actually do anything) require a successful dyn_cast<CallBase> - so just earlyout if the cast fails
Fixes static analyzer nullptr deference warning
Simon Pilgrim [Mon, 10 Jan 2022 15:23:49 +0000 (15:23 +0000)]
[LowerExpectIntrinsic] Use cast<> instead of dyn_cast<> to avoid dereference of nullptr. NFC
Pavel Labath [Mon, 3 Jan 2022 15:49:58 +0000 (16:49 +0100)]
[lldb/platform-gdb] Clear cached protocol state upon disconnection
Previously we would persist the flags indicating whether the remote side
supports a particular feature across reconnects, which is obviously not
a good idea.
I implement the clearing by nuking (its the only way to be sure :) the
entire GDBRemoteCommunication object in the disconnect operation and
creating a new one upon connection. This allows us to maintain a nice
invariant that the GDBRemoteCommunication object (which is now a
pointer) exists only if it is connected. The downside to that is that a
lot of functions now needs to check the validity of the pointer instead
of blindly accessing the object.
The process communication does not suffer from the same issue because we
always destroy the entire Process object for a relaunch.
Differential Revision: https://reviews.llvm.org/D116539
Nicolas Vasilache [Sun, 9 Jan 2022 21:09:24 +0000 (16:09 -0500)]
[mlir][Bufferize] Fix incorrect bufferization of rank-reducing tensor ops.
This revision fixes SubviewOp, InsertSliceOp, ExtractSliceOp construction during bufferization
where not all offset/size/stride operands were properly specified.
A test that exhibited problematic behaviors related to incorrect memref casts is introduced.
Init tensor optimization is disabled in teh testing func bufferize pass.
Differential Revision: https://reviews.llvm.org/D116899
Nikita Popov [Mon, 10 Jan 2022 15:12:29 +0000 (16:12 +0100)]
[Thumb2] Regenerate test checks (NFC)
Yitzhak Mandelbaum [Fri, 7 Jan 2022 19:39:12 +0000 (19:39 +0000)]
[clang][dataflow] Change `transfer` function to update lattice element in place.
Currently, the transfer function returns a new lattice element, which forces an
unnecessary copy on processing each CFG statement.
Differential Revision: https://reviews.llvm.org/D116834
Matthias Springer [Mon, 10 Jan 2022 14:36:46 +0000 (23:36 +0900)]
[mlir][linalg][bufferize][NFC] Pass missing BufferizationState objs as const ref
These should have been updated as part of D116742.
Differential Revision: https://reviews.llvm.org/D116937
Nicolas Vasilache [Mon, 10 Jan 2022 14:03:23 +0000 (09:03 -0500)]
[mlir][Linalg] Disable init_tensor elimination by default
init_tensor elimination is arguably a pre-optimization that should be separated from comprehensive bufferization.
In any case it is still experimental and easily results in wrong IR with violated SSA def-use orderings.
Isolate the optimization behind a flag, separate the test cases and add a test case that would results in wrong IR.
Differential Revision: https://reviews.llvm.org/D116936
Haojian Wu [Mon, 10 Jan 2022 14:09:17 +0000 (15:09 +0100)]
Revert "[AST] Add RParen loc for decltype AutoTypeloc."
This breaks a clang-tidy check, needs to investigate and fix. Revert
them to bring the buildbot back.
This reverts commit
55d96ac3dc56bdebea854952a724c2a50d96ce19 and
37ec65e1d705f56fe5551de1dfcbac1e071588a2
David Sherwood [Wed, 3 Nov 2021 17:09:34 +0000 (17:09 +0000)]
[LoopVectorize] Make VPWidenCanonicalIVRecipe::execute work for scalable vectors
The code in VPWidenCanonicalIVRecipe::execute only worked for fixed-width
vectors due to the way we generate the values per lane. This patch changes
the code to use a combination of vector splats and step vectors to get
the same result. This then works for both fixed-width and scalable vectors.
Tests that exercise this code path for scalable vectors have been added here:
Transforms/LoopVectorize/AArch64/sve-tail-folding.ll
Differential Revision: https://reviews.llvm.org/D113180
Nuno Lopes [Mon, 10 Jan 2022 13:11:44 +0000 (13:11 +0000)]
[SROA] Switch replacement of dead/UB/unreachable ops from undef to poison
SROA has 3 data-structures where it stores sets of instructions that should
be deleted:
- DeadUsers -> instructions that are UB or have no users
- DeadOperands -> instructions that are UB or operands of useless phis
- DeadInsts -> "dead" instructions, including loads of uninitialized memory
with users
The first 2 sets can be RAUW with poison instead of undef. No brainer as UB
can be replaced with poison, and for instructions with no users RAUW is a
NOP.
The 3rd case cannot be currently replaced with poison because the set mixes
the loads of uninit memory. I leave that alone for now.
Another case where we can use poison is in the construction of vectors from
multiple loads. The base vector for the first insertelement is now poison as
it doesn't matter as it is fully overwritten by inserts.
Differential Revision: https://reviews.llvm.org/D116887
Nikita Popov [Mon, 10 Jan 2022 13:58:38 +0000 (14:58 +0100)]
[CodeGen] Avoid some pointer element type accesses
Possibly this is sufficient to fix PR53089.
Louis Dionne [Wed, 15 Dec 2021 16:35:09 +0000 (11:35 -0500)]
[libc++] Refactor the test for join_view's default constructor
In particular, this removes the need for adding a ad-hoc `operator==`
to forward_iterator.
Differential Revision: https://reviews.llvm.org/D116614
Serge Guelton [Mon, 3 Jan 2022 18:32:19 +0000 (13:32 -0500)]
Use a sorted array instead of a map to store AttrBuilder string attributes
Using and std::map<SmallString, SmallString> for target dependent attributes is
inefficient: it makes its constructor slightly heavier, and involves extra
allocation for each new string attribute. Storing the attribute key/value as
strings implies extra allocation/copy step.
Use a sorted vector instead. Given the low number of attributes generally
involved, this is cheaper, as showcased by
https://llvm-compile-time-tracker.com/compare.php?from=
5de322295f4ade692dc4f1823ae4450ad3c48af2&to=
05bc480bf641a9e3b466619af43a2d123ee3f71d&stat=instructions
Differential Revision: https://reviews.llvm.org/D116599
Nikita Popov [Mon, 10 Jan 2022 09:40:33 +0000 (10:40 +0100)]
[TypeFinder] Support opaque pointers
We need to explicitly visit a number of types, as these are no
longer reachable through the pointer type if opaque pointers are
enabled. This is similar to ValueEnumerator changes that have
been done previously.
Alex Xu (Hello71) [Mon, 10 Jan 2022 13:44:44 +0000 (08:44 -0500)]
set __NO_MATH_ERRNO__ if -fno-math-errno
This causes modern glibc to unset math_errhandling MATH_ERRNO. gcc 12
also sets some other macros, but most of them are associated with
flags ignored by clang, so without library examples, it is difficult to
determine whether they should be set. I think setting this one macro is
OK for now.
Louis Dionne [Mon, 10 Jan 2022 13:37:54 +0000 (08:37 -0500)]
[libc++] Add missing `return 0` to main() functions in the tests
Matthias Springer [Mon, 10 Jan 2022 13:31:49 +0000 (22:31 +0900)]
[mlir][linalg][bufferize][NFC] Update comments in BufferizableOpInterface
Differential Revision: https://reviews.llvm.org/D116932
Louis Dionne [Mon, 10 Jan 2022 13:33:14 +0000 (08:33 -0500)]
[libc++] Fix link to bug tracker
Simon Pilgrim [Mon, 10 Jan 2022 12:40:02 +0000 (12:40 +0000)]
[CostModel][X86] Update ROTL/ROTR vXi8/vXi16 costs on AVX512BW targets
Refresh based off recent improvements to codegen and the helper script from D103695
Louis Dionne [Sun, 9 Jan 2022 19:40:10 +0000 (14:40 -0500)]
[libc++] Fix the documentation and re-enable documentation CI
The documentation CI job is very cheap, so we can afford to keep it
around even with reduced capacity. This commit fixes the documentation
(which had an invalid reference in it) and re-enables that CI step.
Differential Revision: https://reviews.llvm.org/D116897
Matthias Springer [Mon, 10 Jan 2022 12:40:05 +0000 (21:40 +0900)]
[mlir][linalg][bufferize][NFC] Clean up bufferization entry point
Differential Revision: https://reviews.llvm.org/D116892
Haojian Wu [Mon, 10 Jan 2022 09:10:41 +0000 (10:10 +0100)]
[clangd] Enable expand-auto for decltype(auto).
Based on https://reviews.llvm.org/D116919.
Fixes https://github.com/clangd/clangd/issues/121
Differential Revision: https://reviews.llvm.org/D116921
Martin Storsjö [Mon, 3 Jan 2022 13:51:37 +0000 (15:51 +0200)]
[llvm-objcopy] Implement the PE-COFF specific --subsystem option
This implements the parsing of the highly PE-COFF specific option
in ConfigManager.cpp, setting Optional<> values in COFFConfig, which
then are used in COFFObjcopy.
This should fix https://github.com/mstorsjo/llvm-mingw/issues/239.
Differential Revision: https://reviews.llvm.org/D116556
Tim Northover [Tue, 16 Nov 2021 12:26:16 +0000 (12:26 +0000)]
AArch64: don't claim to preserve registers used by prologue code
Petar Avramovic [Mon, 10 Jan 2022 12:14:28 +0000 (13:14 +0100)]
AMDGPU/GlobalISel: Rework legalization for extract/insert vector elt
Use G_MERGE_VALUES and G_UNMERGE_VALUES on vector elements instead of
G_EXTRACT and G_INSERT when doing custom legalization for
G_EXTRACT_VECTOR_ELT and G_INSERT_VECTOR_ELT.
With this approach legalization artifact combiner gets direct access
to all vector elements.
Differential Revision: https://reviews.llvm.org/D116115
Florian Hahn [Mon, 10 Jan 2022 12:09:13 +0000 (12:09 +0000)]
[SROA] Reduce the number of times a IRBuilder is constructed (NFC).
This patch reduces the number of times IRBuilders need to be constructed
in SROA.cpp by passing existing ones by reference to the appropriate
places.
Haojian Wu [Fri, 7 Jan 2022 23:12:18 +0000 (00:12 +0100)]
[AST] Add RParen loc for decltype AutoTypeloc.
Differential Revision: https://reviews.llvm.org/D116919
Haojian Wu [Fri, 7 Jan 2022 21:55:58 +0000 (22:55 +0100)]
[AST] Use recovery-expr to preserve incomplete-type-member-access expression.
Fixes https://github.com/clangd/clangd/issues/502
Differential Revision: https://reviews.llvm.org/D116922
Alex Zinenko [Mon, 10 Jan 2022 11:40:38 +0000 (12:40 +0100)]
[mlir] address post-commit review for D116759
Florian Hahn [Mon, 10 Jan 2022 11:31:26 +0000 (11:31 +0000)]
[SCEVExpander] Only create trunc when needed.
9345ab3a4550 updated generateOverflowCheck to skip creating checks that
always evaluate to false. This in turn means that we only need to
create TruncTripCount if it is actually used.
Sink the TruncTripCount creating into ComputeEndCheck, so it is only
created when there's an actual check.
Jeremy Morse [Mon, 10 Jan 2022 11:22:51 +0000 (11:22 +0000)]
[Dexter] Allow tests to specify command line options
This patch adds a "DexCommandLine" command, allowing dexter tests to
specify what command line options the test should be started with. I've
also plumbed it through into the debuggers.
This eases the matter of pointing Dexter at larger tests, or controlling
different paths through a single binary from a Dexter test.
Differential Revision: https://reviews.llvm.org/D115330
Mubashar Ahmad [Wed, 5 Jan 2022 16:53:59 +0000 (16:53 +0000)]
[Clang][AArch64][ARM] PMUv3.4 Option Added
An option has been added to Clang to enable or disable
the PMU v3.4 architecture extension.
Differential Revision: https://reviews.llvm.org/D116748
Sam McCall [Fri, 10 Dec 2021 03:17:50 +0000 (04:17 +0100)]
[clangd] Include fixer for missing functions in C
A function call `unresolved()` in C will generate an implicit declaration
of the missing function and warn `ext_implicit_function_decl` or so.
(Compared to in C++ where we get `err_undeclared_var_use`).
We want to try to resolve these names.
Unfortunately typo correction is disabled in sema for performance
reasons unless this warning is promoted to error.
(We need typo correction for include-fixer.)
It's not clear to me where a switch to force this correction on should
go, include-fixer is kind of a hack. So hack more by telling sema we're
promoting them to error.
Fixes https://github.com/clangd/clangd/issues/937
Differential Revision: https://reviews.llvm.org/D115490
Alex Zinenko [Mon, 10 Jan 2022 10:39:06 +0000 (11:39 +0100)]
[mlir] Don't inline calls from dead SCCs
During iterative inlining of the functions in a multi-step call chain, the
inliner could add the same call operation several times to the worklist, which
led to use-after-free when this op was considered more than once.
Closes #52887.
Reviewed By: wsmoses
Differential Revision: https://reviews.llvm.org/D116820
David Green [Mon, 10 Jan 2022 11:03:49 +0000 (11:03 +0000)]
Revert "[LoopFlatten] Move it to a LoopPassManager"
This commit caused performance regressions due to differences in the
expected code during loop flattening. Reverting it until the fix is
ready, which hopefully wont take too long.
This reverts commit
86825fc2fb363b807569327880c05e4b0b5393ec.
Stephan Herhut [Fri, 7 Jan 2022 08:53:14 +0000 (09:53 +0100)]
[mlir][memref] Tighten verification of memref.reinterpret_cast
We allow the omission of a map in memref.reinterpret_cast under the assumption,
that the cast might cast to an identity layout. This change adds verification
that the static knowledge that is present in the reinterpret_cast supports
this assumption.
Differential Revision: https://reviews.llvm.org/D116601
David Sherwood [Tue, 26 Oct 2021 08:55:30 +0000 (09:55 +0100)]
[LoopVectorize] Add support for tail folding using scalable vectors
This patch fixes up an issue with InnerLoopVectorizer::getOrCreateVectorTripCount
whereby we weren't correctly generating the runtime trip count
for scalable vectors when tail-folding.
It also removes some asserts in the tail-folding path for cases when
the VF is not scalable.
In this patch I have only permitted tail-folding to be enabled
explicitly for scalable vectors when the user has specified one
of the following flags:
-prefer-predicate-over-epilogue=predicate-dont-vectorize
-prefer-predicate-over-epilogue=predicate-else-scalar-epilogue
For now it's best not to enable tail-folding with scalable vectors for
low trip counts or when optimising for code size, since there has been
no analysis on whether this is worth it.
Various tests have been added here:
Transforms/LoopVectorize/AArch64/sve-tail-folding.ll
Transforms/LoopVectorize/AArch64/sve-tail-folding-forced.ll
The tests cannot be target independent because they require masked
load/store support, i.e. TTI.isLegalMaskedLoad and TTI.isLegalMaskedStore
need to return true.
Differential Revision: https://reviews.llvm.org/D113003
Cullen Rhodes [Mon, 10 Jan 2022 10:21:52 +0000 (10:21 +0000)]
[AArch64][SVE] Fold predicate into compare
Codegen of added testcase before this patch:
ptrue p0.s
cmpgt p1.s, p0/z, z0.s, z1.s
cmpge p2.s, p0/z, z2.s, z1.s
and p0.b, p0/z, p1.b, p2.b
ret
Patterns originally authored by Will Lovett.
Reviewed By: david-arm
Differential Revision: https://reviews.llvm.org/D116749
Shraiysh Vaishay [Mon, 10 Jan 2022 09:56:20 +0000 (15:26 +0530)]
[mlir][OpenMP] Change the syntax of omp.atomic.read op
This patch changes the syntax of omp.atomic.read to take the address of
destination, instead of having the value in a result. This will allow
using omp.atomic.read operation within an omp.atomic.capture operation
thus making its implementation less complex.
Reviewed By: peixin
Differential Revision: https://reviews.llvm.org/D116396
David Sherwood [Fri, 7 Jan 2022 14:01:29 +0000 (14:01 +0000)]
[NFC] Add tests for splats of illegal integer vector types
Differential Revision: https://reviews.llvm.org/D116810
Sam McCall [Thu, 6 Jan 2022 01:01:13 +0000 (02:01 +0100)]
[clangd] Support configuration of inlay hints.
The idea is that the feature will always be advertised at the LSP level, but
depending on config we'll return partial or no responses.
We try to avoid doing the analysis for hints we're not going to return.
Examples of syntax:
```
InlayHints:
Enabled: No
---
InlayHints:
ParameterNames: No
---
InlayHints:
ParameterNames: Yes
DeducedTypes: Yes
```
Differential Revision: https://reviews.llvm.org/D116713
Sam McCall [Thu, 11 Nov 2021 14:27:10 +0000 (15:27 +0100)]
[Parse] Use empty RecoveryExpr when if/while/do/switch conditions fail to parse
This allows the body to be parsed.
An special-case that would replace a missing if condition with OpaqueValueExpr
was removed as it's now redundant (unless recovery-expr is disabled).
For loops are not handled at this point, as the parsing is more complicated.
Differential Revision: https://reviews.llvm.org/D113752
Florian Hahn [Mon, 10 Jan 2022 08:39:12 +0000 (08:39 +0000)]
[SCEVExpander] Only create multiplication if needed.
9345ab3a4550 updated generateOverflowCheck to skip creating checks that
always evaluate to false. This in turn means that we only need to
compute |Step| * Trip count if the result of the multiplication is
actually used.
Sink the multiplication into ComputeEndCheck, so it is only created
when there's an actual check.
LLVM GN Syncbot [Mon, 10 Jan 2022 08:36:39 +0000 (08:36 +0000)]
[gn build] Port
c0fdc748871f
Haojian Wu [Thu, 6 Jan 2022 19:46:33 +0000 (20:46 +0100)]
[AST] Add more source information for DecltypeTypeLoc.
Adds the paren source location, and removes the hack in clangd.
Differential Revision: https://reviews.llvm.org/D116793
mydeveloperday [Mon, 10 Jan 2022 08:28:42 +0000 (08:28 +0000)]
[clang-format] Ensure we can correctly parse lambda in the template argument list
https://github.com/llvm/llvm-project/issues/46505
The presence of a lambda in an argument template list ignored the [] as a lambda at all, this caused the contents of the <> to be incorrectly analyzed.
```
struct Y : X < [] {
return 0;
} > {};
```
Fixes: #46505
Reviewed By: curdeius
Differential Revision: https://reviews.llvm.org/D116806
Lang Hames [Mon, 10 Jan 2022 08:24:38 +0000 (19:24 +1100)]
[ORC][JITLink] Add dependence on OrcShared to JITLink.
JITLink depends on OrcShared as of
c0fdc748871. This should fix the build
failure at https://lab.llvm.org/buildbot#builders/61/builds/19796.
Vincent Lee [Mon, 10 Jan 2022 08:07:42 +0000 (00:07 -0800)]
[lld-macho] Fix shadowed variable
This fixes a windows build failure from D115416.
Nikita Popov [Fri, 7 Jan 2022 08:57:53 +0000 (09:57 +0100)]
[MemoryBuiltins] Remove isNoAliasFn() in favor of isNoAliasCall()
We currently have two similar implementations of this concept:
isNoAliasCall() only checks for the noalias return attribute.
isNoAliasFn() also checks for allocation functions.
We should switch to only checking the attribute. SLC is responsible
for inferring the noalias return attribute for non-new allocation
functions (with a missing case fixed in
https://github.com/llvm/llvm-project/commit/
348bc76e3548c52dbcd442590ca0a7f5b09b7534).
For new, clang is responsible for setting the attribute,
if -fno-assume-sane-operator-new is not passed.
Differential Revision: https://reviews.llvm.org/D116800
Lang Hames [Mon, 10 Jan 2022 08:08:59 +0000 (19:08 +1100)]
[ORC] Add helper functions for running finalize / dealloc actions.
runFinalizeActions takes an AllocActions vector and attempts to run its finalize
actions. If any finalize action fails then all paired dealloc actions up to the
failing pair are run, and the error(s) returned. If all finalize actions succeed
then a vector containing the dealloc actions is returned.
runDeallocActions takes a vector<WrapperFunctionCall> containing dealloc action
calls and runs them all, returning any error(s).
These helpers are intended to simplify the implementation of
JITLinkMemoryManager::InFlightAlloc::finalize and
JITLinkMemoryManager::deallocate overrides by taking care of execution (and
potential roll-back) of allocation actions.
Craig Topper [Mon, 10 Jan 2022 07:23:45 +0000 (23:23 -0800)]
[RISCV] Generalize (srl (and X, 0xffff), C) -> (srli (slli X, (XLen-16), (XLen-16) + C) optimization.
This can be generalized to (srl (and X, C2), C) ->
(srli (slli X, (XLen-C3), (XLen-C3) + C). Where C2 is a mask with
C3 trailing ones.
This can avoid constant materialization for C2. This is beneficial
even when C2 can be selected to ANDI because the SLLI can become
C.SLLI, but C.ANDI cannot cover all the immediates of ANDI.
This also enables CSE in some cases of i8 sdiv by constant codegen.
Stephen Neuendorffer [Mon, 10 Jan 2022 06:16:58 +0000 (22:16 -0800)]
Fix exported MLIR_TABLEGEN_EXE
LLVM_OPTIMIZED_TABLEGEN results in MLIR_TABLEGEN_EXE pointing to
an absolute path in the build directory. This doesn't work
when exporting to an install directory. This patch fixes the exported
information for an install directory to refer to the installed
mlir-tblgen. (Note that this is probably a debug version if
LLVM_OPTIMIZED_TABLEGEN is set)
Pavel Kosov [Mon, 10 Jan 2022 06:57:13 +0000 (09:57 +0300)]
[SchedModels][CortexA55] Fix scheduling of FP loads
Patch fixes scheduling of FP load instructions with pre/post increment adding WriteAdr for address operand.
Reviewed By: dmgreen
Differential Revision: https://reviews.llvm.org/D116361
OS Laboratory. Huawei Russian Research Institute. Saint-Petersburg
Jonas Devlieghere [Mon, 10 Jan 2022 06:54:08 +0000 (22:54 -0800)]
[lldb] Remove LLDB_RECORD_RESULT macro
Dave Lee [Mon, 10 Jan 2022 06:34:47 +0000 (22:34 -0800)]
[lldb] Skip TestTargetXMLArch if no support for x86 target
If LLVM is configured without X86 as one of its TARGETS_TO_BUILD, then lldb
will crash when using X86 disassembler (which it does while running `image
show-unwind`).
Zi Xuan Wu [Thu, 6 Jan 2022 09:21:49 +0000 (17:21 +0800)]
[CSKY] Lower leaf DAG node such as global symbol, frame address and jumptable, etc.
Lower global symbols such as call/external symbol.
Lower other leaf DAG node such as frame address/block address/jumptable/vastart.
Normally some leaf symbols need reside in constant pool as ABI prefers, and are addressed by
lrw or jsri instructions.
Every symbol in constant pool is lowered with one entry in target constant pool. The
entry has different type corresponding to different leaf node such as blockaddress,
jumptable, or global value.
Siva Chandra Reddy [Mon, 10 Jan 2022 05:51:30 +0000 (05:51 +0000)]
[libc] Re-enable thrd_test.
Other threads related tests have been using the API that is tested here
without any problems. The main reason for disabling the test has also been
fixed many months ago.
Dave Lee [Mon, 10 Jan 2022 06:01:02 +0000 (22:01 -0800)]
[lldb] Skip TestVSCode_coreFile if no x86 target support
Jonas Devlieghere [Sat, 8 Jan 2022 00:26:40 +0000 (16:26 -0800)]
[lldb] Remove reproducer instrumentation
This patch removes most of the reproducer instrumentation. It keeps
around the LLDB_RECORD_* macros for logging. See [1] for more details.
[1] https://lists.llvm.org/pipermail/lldb-dev/2021-September/017045.html
Differential revision: https://reviews.llvm.org/D116847
Dave Lee [Sun, 9 Jan 2022 23:01:53 +0000 (15:01 -0800)]
[lldb] Guard libstdc++ specific 'frame var' test
While working on D116788 (properly error out of `frame var`), this libstdc++
specific `frame var` invocation was found in the tests. This test is in the
generic directory, but has this one case that requires libstdc++. The fix here
is to put the one `expect()` inside of a condition that checks for libstdc++.
Differential Revision: https://reviews.llvm.org/D116901
Dave Lee [Mon, 10 Jan 2022 04:10:38 +0000 (20:10 -0800)]
[lldb] Check for arm64 in TestDisassembleRawData
This test checks for `aarch64` but the lit config could also contain `arm64`.
This change adds `arm64` to make the test pass in all cases.
Differential Revision: https://reviews.llvm.org/D116912
Dave Lee [Mon, 10 Jan 2022 05:29:39 +0000 (21:29 -0800)]
[lldb] Require x86 support for dwo-relative-path test
Craig Topper [Mon, 10 Jan 2022 03:55:13 +0000 (19:55 -0800)]
[RISCV] Isel (sra (sext_inreg X, i16), C) -> (srai (slli X, (XLen-16), (XLen-16) + C).
Similar for (sra (sext_inreg X, i8), C).
With Zbb, sext_inreg of i8 and i16 are legal for sext.b and sext.h.
This transform makes the Zbb codegen the same as without Zbb. The
shifts are more compressible. This also exposes an opportunity for
CSE with another slli in the i16 sdiv by constant codegen.
Alexander Shaposhnikov [Mon, 10 Jan 2022 05:20:37 +0000 (05:20 +0000)]
[lld][ELF] Support adrp+ldr GOT optimization for AArch64
This diff adds first bits to support relocation relaxations for AArch64
discussed on https://github.com/ARM-software/abi-aa/pull/106.
In particular, the case of
adrp x0, :got: symbol
ldr x0, [x0, :got_lo12: symbol]
is handled.
Test plan: make check-all
Differential revision: https://reviews.llvm.org/D112063
jacquesguan [Thu, 22 Jul 2021 03:40:31 +0000 (11:40 +0800)]
[RISCV] Disable EEW=64 for index values when XLEN=32.
Disable EEW=64 for vector index load/store when XLEN=32.
Differential Revision: https://reviews.llvm.org/D106518
jinge90 [Mon, 10 Jan 2022 01:23:28 +0000 (09:23 +0800)]
Control-flow Enforcement Technology (CET), published by Intel, introduces
indirect branch tracking(IBT) feature aiming to ensure the target address
of an indirect jump/call is not tampered.
When IBT is enabled, each function or target of any indirect jump/call will start
with an 'endbr32/64' instruction otherwise the program will crash during execution.
To build an application with CET enabled. we need to ensure:
1. build the source code with "-fcf-protection=full"
2. all the libraries linked with .o files must be CET enabled too
This patch aims to enable CET for compiler-rt builtins library, we add an option
"COMPILER_RT_ENABLE_CET" whose default value is OFF to enable CET for compiler-rt
in building time and when this option is "ON", "-fcf-protection=full" is added to
BUILTINS_CFLAG and the "endbr32/64" will be placed in the beginning of each assembly
function. We also enabled CET for crtbegin, crtend object files in this patch.
Reviewed by: MaskRay, compnerd, manojgupta, efriedma
Differential Revision: https://reviews.llvm.org/D109811
Signed-off-by: jinge90 <ge.jin@intel.com>
Jim Lin [Mon, 10 Jan 2022 01:43:20 +0000 (09:43 +0800)]
[Builtins] Add missing the macro 'y' description in comments
New type str 'y' added from D76077. But missed its description in
comments.
Reviewed By: stuij
Differential Revision: https://reviews.llvm.org/D116509
Craig Topper [Mon, 10 Jan 2022 02:15:49 +0000 (18:15 -0800)]
[RISCV] Fold (srl (and X, 0xffff), C)->(srli (slli X, (XLen-16), (XLen-16) + C) even with Zbb/Zbp.
We can use zext.h with Zbb, but srli/slli may offer more opportunities
for compression.