OSDN Git Service
Emil Engler [Thu, 7 Jan 2021 02:28:54 +0000 (18:28 -0800)]
[doc] Place sha256 in lld/README.md into backticks
Reviewed By: smeenai
Differential Revision: https://reviews.llvm.org/D93984
Timm Bäder [Tue, 12 Jan 2021 18:15:21 +0000 (13:15 -0500)]
Add -ansi option to CompileOnly group
-ansi is documented as being the "same as -std=c89", but there are
differences when passing it to a link.
Adding -ansi to said group makes sense since it's supposed to be an
alias for -std=c89 and resolves this inconsistency.
Cullen Rhodes [Tue, 12 Jan 2021 17:48:52 +0000 (17:48 +0000)]
[SVE][NFC] Regenerate a few CodeGen tests
Regenerated using llvm/utils/update_llc_test_checks.py as part of
D94504, committing separately to reduce the diff for D94504.
Simon Pilgrim [Tue, 12 Jan 2021 18:01:41 +0000 (18:01 +0000)]
[AMDGPU] Regenerate umax crash test
Akira Hatanaka [Tue, 12 Jan 2021 17:56:06 +0000 (09:56 -0800)]
Fix typo in diagnostic message
rdar://
66684531
Simon Pilgrim [Tue, 12 Jan 2021 17:24:34 +0000 (17:24 +0000)]
[X86] Regenerate sdiv_fix_sat.ll + udiv_fix_sat.ll tests
Adding missing libcall PLT qualifiers
Rahul Joshi [Thu, 7 Jan 2021 00:32:59 +0000 (16:32 -0800)]
[MLIR] Disallow `sym_visibility`, `sym_name` and `type` attributes in the parsed attribute dictionary.
Differential Revision: https://reviews.llvm.org/D94200
Jinsong Ji [Tue, 12 Jan 2021 15:56:58 +0000 (15:56 +0000)]
[PowerPC][NFCI] PassSubtarget to ASMWriter
Subtarget feature bits are needed to change instprinter's behavior based
on feature bits.
Most of the other popular targets were updated back in 2015,
in https://reviews.llvm.org/rGb46d0234a6969
we should update it too.
Reviewed By: sfertile
Differential Revision: https://reviews.llvm.org/D94449
Lei Zhang [Tue, 12 Jan 2021 16:11:45 +0000 (11:11 -0500)]
[mlir][spirv] NFC: split deserialization into multiple source files
This avoids large source files and gives a better structure. It also
allows leveraging compilation parallelism.
Reviewed By: mravishankar
Differential Revision: https://reviews.llvm.org/D94360
Marek Kurdej [Tue, 12 Jan 2021 16:06:58 +0000 (17:06 +0100)]
[libc++] [C++2b] [P1048] Add is_scoped_enum and is_scoped_enum_v.
* https://wg21.link/p1048
Reviewed By: ldionne, #libc
Differential Revision: https://reviews.llvm.org/D94409
Vladislav Vinogradov [Tue, 12 Jan 2021 16:06:06 +0000 (17:06 +0100)]
[mlir] Fix for LIT tests
Add `MLIR_SPIRV_CPU_RUNNER_ENABLED` to `llvm_canonicalize_cmake_booleans`.
Reviewed By: mehdi_amini
Differential Revision: https://reviews.llvm.org/D94407
Vladislav Vinogradov [Tue, 12 Jan 2021 16:02:56 +0000 (17:02 +0100)]
[mlir][CAPI] Fix inline function declaration
Add `static` keyword, otherwise build fail with linker error for some cases.
Reviewed By: ftynse
Differential Revision: https://reviews.llvm.org/D94496
Lei Zhang [Mon, 11 Jan 2021 13:50:00 +0000 (08:50 -0500)]
Reland "[mlir][linalg] Support parsing attributes in named op spec"
With this, now we can specify a list of attributes on named ops
generated from the spec. The format is defined as
```
attr-id ::= bare-id (`?`)?
attr-typedef ::= type (`[` `]`)?
attr-def ::= attr-id `:` attr-typedef
tc-attr-def ::= `attr` `(` attr-def-list `)`
tc-def ::= `def` bare-id
`(`tensor-def-list`)` `->` `(` tensor-def-list`)`
(tc-attr-def)?
```
For example,
```
ods_def<SomeCppOp>
def some_op(...) -> (...)
attr(
f32_attr: f32,
i32_attr: i32,
array_attr : f32[],
optional_attr? : f32
)
```
where `?` means optional attribute and `[]` means array type.
Reviewed By: hanchung, nicolasvasilache
Differential Revision: https://reviews.llvm.org/D94240
Nemanja Ivanovic [Tue, 12 Jan 2021 15:46:11 +0000 (09:46 -0600)]
[PowerPC] Add support for embedded devices with EFPU2
PowerPC cores like e200z759n3 [1] using an efpu2 only support single precision
hardware floating point instructions. The single precision instructions efs*
and evfs* are identical to the spe float instructions while efd* and evfd*
instructions trigger a not implemented exception.
This patch introduces a new command line option -mefpu2 which leads to
single-hardware / double-software code generation.
[1] Core reference:
https://www.nxp.com/files-static/32bit/doc/ref_manual/e200z759CRM.pdf
Differential revision: https://reviews.llvm.org/D92935
Bjorn Pettersson [Tue, 12 Jan 2021 15:28:16 +0000 (16:28 +0100)]
[SLP] Add test case showing a bug when dealing with padded types
We shouldn't vectorize stores of non-packed types (i.e. types that
has padding between consecutive variables in a scalar layout,
but being packed in a vector layout).
The problem was detected as a miscompile in a downstream test case.
This is a pre-commit of a test case for the fix in D94446.
Lei Zhang [Mon, 11 Jan 2021 14:58:31 +0000 (09:58 -0500)]
[mlir][spirv] NFC: place ops in the proper file for their categories
This commit moves dangling ops in the main ops.td file to the proper
file matching their categories. This makes ops.td as purely including
all category files.
Differential Revision: https://reviews.llvm.org/D94413
Kazushi (Jam) Marukawa [Tue, 12 Jan 2021 12:36:55 +0000 (21:36 +0900)]
[VE] Update VELIntrinsic tests
Update comment and style of regression tests for VELIntrinsic
Reviewed By: simoll
Differential Revision: https://reviews.llvm.org/D94490
Bevin Hansson [Tue, 12 Jan 2021 14:40:36 +0000 (15:40 +0100)]
[X86] Improved lowering for saturating float to int.
Adapted from D54696 by @nikic.
This patch improves lowering of saturating float to
int conversions, FP_TO_[SU]INT_SAT, for X86.
Reviewed By: craig.topper
Differential Revision: https://reviews.llvm.org/D86079
Valentin Clement [Tue, 12 Jan 2021 14:42:25 +0000 (09:42 -0500)]
[mlir][openacc] Use TableGen information for default enum
Use TableGen and information in ACC.td for the Default enum in the OpenACC dialect.
This patch generalize what was done for OpenMP for directives.
Follow up patch after D93576
Reviewed By: kiranchandramohan
Differential Revision: https://reviews.llvm.org/D93710
Paul C. Anagnostopoulos [Mon, 11 Jan 2021 14:46:27 +0000 (09:46 -0500)]
[TableGen] Improve error message for semicolon after braced body.
Add a test for this message.
Differential Revision: https://reviews.llvm.org/D94412
Nicolas Vasilache [Tue, 12 Jan 2021 14:01:59 +0000 (14:01 +0000)]
[mlir][Linalg] NFC - Refactor fusion APIs
This revision uniformizes fusion APIs to allow passing OpOperand, OpResult and adds a finer level of control fusion.
Differential Revision: https://reviews.llvm.org/D94493
Simon Pilgrim [Tue, 12 Jan 2021 14:07:53 +0000 (14:07 +0000)]
[X86][SSE] getFauxShuffleMask - handle PACKSS(SRAI(),SRAI()) shuffle patterns.
We can't easily treat ASHR a faux shuffle, but if it was just feeding a PACKSS then it was likely being used as sign-extension for a truncation, so just peek through and adjust the mask accordingly.
Simon Pilgrim [Tue, 12 Jan 2021 13:43:56 +0000 (13:43 +0000)]
[X86][SSE] combineSubToSubus - add v16i32 handling on pre-AVX512BW targets.
v16i32 -> v16i16/v8i16 truncation is now good enough using PACKSS/PACKUS + shuffle combining that its no longer necessary to early-out on pre-AVX512BW targets.
This was noticed while looking at completing PR40111 and moving combineSubToSubus to DAGCombine entirely.
Bevin Hansson [Mon, 11 Jan 2021 21:46:42 +0000 (22:46 +0100)]
[Fixed Point] Add codegen for conversion between fixed-point and floating point.
The patch adds the required methods to FixedPointBuilder
for converting between fixed-point and floating point,
and uses them from Clang.
This depends on D54749.
Reviewed By: leonardchan
Differential Revision: https://reviews.llvm.org/D86632
Simon Pilgrim [Tue, 12 Jan 2021 11:50:09 +0000 (11:50 +0000)]
[X86][SSE] combineSubToSubus - remove SSE2 early-out.
SSE2 truncation codegen has improved over the past few years (mainly due to better shuffle lowering/combining and computeKnownBits) - its no longer necessary to early-out from v8i32/v8i64 truncations.
This was noticed while looking at completing PR40111 and moving combineSubToSubus to DAGCombine entirely.
Fraser Cormack [Fri, 8 Jan 2021 17:14:08 +0000 (17:14 +0000)]
[RISCV] Improve scalable-vector shift tests (NFC)
All i8/i16 and several i32 tests were testing immediate shift amounts
which exceeded the bits in the vector elements, creating poison values.
Amend the tests to test well-behaved shift amounts.
Christian Sigg [Thu, 7 Jan 2021 08:41:36 +0000 (09:41 +0100)]
Change the LLVM_ATTRIBUTE_DEPRECATED macro to use C++14 attribute.
C++14 attributes are superior because they can be applied to functions with inline definition and the syntax is cleaner.
I intend to convert all uses and then remove the macro.
One issue that might hold back switching uses to C++14 attributes is that
clang-format does not put long attributes on separate lines and formatted code will look like:
```
template <typename T>
[[deprecated("blah blah")]] void
foooooooooooooooooooooooooooo() {
...
}
```
Putting long attributes on a separate line would be prettier.
See https://stackoverflow.com/questions/
45740466/clang-format-setting-to-control-c-attributes
AttributeMacros probably won't help because it can't match the custom message.
https://clang.llvm.org/docs/ClangFormatStyleOptions.html
Reviewed By: rriddle, MaskRay
Differential Revision: https://reviews.llvm.org/D94219
Nico Weber [Tue, 12 Jan 2021 11:30:32 +0000 (06:30 -0500)]
Revert "[Test] Add failing test for PR48725"
This reverts commit
e8287cb2b2923af9da72fd953e2ec5495c33861a.
Test unexpectedly passes on mac, see comment 2 on PR48725.
Georgii Rymar [Tue, 22 Dec 2020 14:36:16 +0000 (17:36 +0300)]
[obj2yaml] - Don't crash when an object has an empty symbol table.
Currently we crash when we have an object with SHT_SYMTAB/SHT_DYNSYM sections
of size 0.
With this patch instead of the crash we start to dump them properly.
Differential revision: https://reviews.llvm.org/D93697
Georgii Rymar [Mon, 28 Dec 2020 09:20:51 +0000 (12:20 +0300)]
[obj2yaml,yaml2obj] - Fix issues with creating/dumping group sections.
We have the following issues related to group sections:
1) yaml2obj is unable to set the custom `sh_entsize` value, because the `EntSize`
key is currently ignored.
2) obj2yaml is unable to dump the group section which `sh_entsize != 4`.
3) obj2yaml always dumps the "EntSize" for group sections, though
usually we are trying to omit dumping default values when dumping keys.
I.e. we should not print the "EntSize" key when `sh_entsize` == 4.
This patch fixes (1),(3) and adds the test case to document the behavior of (2).
Differential revision: https://reviews.llvm.org/D93854
Jay Foad [Wed, 26 Aug 2020 13:08:14 +0000 (14:08 +0100)]
[AMDGPU][GlobalISel] Remove some duplicate RUN lines
Differential Revision: https://reviews.llvm.org/D86618
Jay Foad [Fri, 8 Jan 2021 13:40:29 +0000 (13:40 +0000)]
[SlotIndexes] Fix and simplify basic block splitting
Remove the InsertionPoint argument from SlotIndexes::insertMBBInMaps
because it was confusing: what does it mean to insert a new block
between two instructions, in the middle of an existing block?
Instead, support the case that MachineBasicBlock::splitAt really needs,
where the new block contains some instructions that are already in the
maps because they have been moved there from the tail of the previous
block.
In all other use cases the new block is empty.
Based on work by Carl Ritson!
Differential Revision: https://reviews.llvm.org/D94311
Mikhail Maltsev [Tue, 12 Jan 2021 10:22:35 +0000 (10:22 +0000)]
[clang][AST] Get rid of an alignment hack in DeclObjC.h [NFCI]
This code currently uses a union object to increase the
alignment of the type ObjCTypeParamList. The original intent of this
trick was to be able to use the expression `this + 1` to access the
beginning of a tail-allocated array of `ObjCTypeParamDecl *` pointers.
The code has since been refactored and uses `llvm::TrailingObjects` to
manage the tail-allocated array. This template takes care of
alignment, so the hack is no longer necessary.
This patch removes the union so that the `SourceRange` class can be
used directly instead of being re-implemented with raw representations
of source locations.
Reviewed By: aprantl
Differential Revision: https://reviews.llvm.org/D94224
Georgii Rymar [Tue, 12 Jan 2021 10:17:23 +0000 (13:17 +0300)]
[llvm-readobj] - One more attempt to fix BB.
Add `this->` for `W`, which is the member of `ObjDumper`
An example of error:
readobj/ELFDumper.cpp:738:13: error: use of undeclared identifier 'W'
assert(&W.getOStream() == &llvm::fouts());
Sourabh Singh Tomar [Tue, 12 Jan 2021 10:13:33 +0000 (15:43 +0530)]
[mlir][openmp][NFCI] Rename `continuationIP` to `continuationBlock`
Argument is a `Block` not a `point`.
Georgii Rymar [Tue, 12 Jan 2021 10:09:49 +0000 (13:09 +0300)]
[llvm-readobj] - An attempt to fix BB.
This adds the `template` keyword for 'getAsArrayRef' calls.
An example of error:
/b/1/openmp-gcc-x86_64-linux-debian/llvm.src/llvm/tools/llvm-readobj/ELFDumper.cpp:4491:50: error: use 'template' keyword to treat 'getAsArrayRef' as a dependent template name
for (const Elf_Rel &Rel : this->DynRelRegion.getAsArrayRef<Elf_Rel>())
Georgii Rymar [Tue, 12 Jan 2021 10:01:15 +0000 (13:01 +0300)]
[llvm-readobj] - Add 'override' to fix build bots.
This should fix bots after landing D93900.
An example of error is:
/home/worker/2.0.1/lldb-x86_64-debian/llvm-project/llvm/tools/llvm-readobj/ELFDumper.cpp:883:8: warning: 'printSectionMapping' overrides a member function but is not marked 'override' [-Winconsistent-missing-override]
void printSectionMapping() {}
Georgii Rymar [Tue, 29 Dec 2020 13:03:37 +0000 (16:03 +0300)]
[llvm-readef/obj] - Change the design structure of ELF dumper. NFCI.
This is a refactoring for design of stuff in `ELFDumper.cpp`.
The current design of ELF dumper is far from ideal.
Currently most overridden functions (inherited from `ObjDumper`) in `ELFDumper` just forward to
the functions of `ELFDumperStyle` (which can be either `GNUStyle` or `LLVMStyle`).
A concrete implementation may be in any of `ELFDumper`/`DumperStyle`/`GNUStyle`/`LLVMStyle`.
This patch reorganizes the classes by introducing `GNUStyleELFDumper`/`LLVMStyleELFDumper`
which inherit from `ELFDumper`. The implementations are moved:
`DumperStyle` -> `ELFDumper`
`GNUStyle` -> `GNUStyleELFDumper`
`LLVMStyle` -> `LLVMStyleELFDumper`
With that we can avoid having a lot of redirection calls and helper methods.
The number of code lines changes from 7142 to 6922 (reduced by ~3%) and the
code overall looks cleaner.
Differential revision: https://reviews.llvm.org/D93900
Heejin Ahn [Tue, 29 Dec 2020 03:48:44 +0000 (19:48 -0800)]
[WebAssembly] Remove more unnecessary brs in CFGStackify
After placing markers, we removed some unnecessary branches, but it only
handled the simplest case. This makes more unnecessary branches to be
removed.
Reviewed By: dschuff, tlively
Differential Revision: https://reviews.llvm.org/D94047
Max Kazantsev [Tue, 12 Jan 2021 09:04:12 +0000 (16:04 +0700)]
[Test] Add failing test for PR48725
Alex Zinenko [Mon, 11 Jan 2021 12:58:05 +0000 (13:58 +0100)]
[mlir] use built-in vector types instead of LLVM dialect types when possible
Continue the convergence between LLVM dialect and built-in types by using the
built-in vector type whenever possible, that is for fixed vectors of built-in
integers and built-in floats. LLVM dialect vector type is still in use for
pointers, less frequent floating point types that do not have a built-in
equivalent, and scalable vectors. However, the top-level `LLVMVectorType` class
has been removed in favor of free functions capable of inspecting both built-in
and LLVM dialect vector types: `LLVM::getVectorElementType`,
`LLVM::getNumVectorElements` and `LLVM::getFixedVectorType`. Additional work is
necessary to design an implemented the extensions to built-in types so as to
remove the `LLVMFixedVectorType` entirely.
Note that the default output format for the built-in vectors does not have
whitespace around the `x` separator, e.g., `vector<4xf32>` as opposed to the
LLVM dialect vector type format that does, e.g., `!llvm.vec<4 x fp128>`. This
required changing the FileCheck patterns in several tests.
Reviewed By: mehdi_amini, silvas
Differential Revision: https://reviews.llvm.org/D94405
Jan Svoboda [Tue, 12 Jan 2021 08:09:06 +0000 (09:09 +0100)]
[clang][cli] Remove -f[no-]trapping-math from -cc1 command line
This patch removes the -f[no-]trapping-math flags from the -cc1 command line. These flags are ignored in the command line parser and their semantics is fully handled by -ffp-exception-mode.
This patch does not remove -f[no-]trapping-math from the driver command line. The driver flags are being used and do affect compilation.
Reviewed By: dexonsmith, SjoerdMeijer
Differential Revision: https://reviews.llvm.org/D93395
Sebastian Neubauer [Mon, 11 Jan 2021 13:20:36 +0000 (14:20 +0100)]
[AMDGPU] Fix failing assert with scratch ST mode
In ST mode, flat scratch instructions have neither an sgpr nor a vgpr
for the address. This lead to an assertion when inserting hard clauses.
Differential Revision: https://reviews.llvm.org/D94406
Sander de Smalen [Tue, 12 Jan 2021 08:32:21 +0000 (08:32 +0000)]
[LiveDebugValues] Fix comparison operator in VarLocBasedImpl
The issue was introduced in commit rG84a1120943a651184bae507fed5d648fee381ae4
and would cause a VarLoc's StackOffset to be compared with its own, instead of
the StackOffset from the other VarLoc. This patch fixes that.
Stephan Bergmann [Wed, 2 Sep 2020 14:04:56 +0000 (16:04 +0200)]
Adapt CastExpr::getSubExprAsWritten to ConstantExpr
Differential Revision: https://reviews.llvm.org/D87030
Heejin Ahn [Mon, 28 Dec 2020 13:23:48 +0000 (05:23 -0800)]
[WebAssembly] Misc. refactoring in CFGStackify (NFC)
Updating `ScopeTops` is something we frequently do in CFGStackify, so
this factors it out as a function. This also makes a few utility
functions templated so that they are not dependent on input vector
types and simplifies function parameters.
Reviewed By: tlively
Differential Revision: https://reviews.llvm.org/D94046
Fangrui Song [Tue, 12 Jan 2021 08:07:28 +0000 (00:07 -0800)]
[ELF] Drop .rel[a].debug_gnu_pub{names,types} for --gdb-index --emit-relocs
Fixes PR48693: --emit-relocs keeps relocation sections. --gdb-index drops
.debug_gnu_pubnames and .debug_gnu_pubtypes but not their relocation sections.
This can cause a null pointer dereference in `getOutputSectionName`.
Also delete debug-gnu-pubnames.s which is covered by gdb-index.s
Reviewed By: grimar
Differential Revision: https://reviews.llvm.org/D94354
Craig Topper [Tue, 12 Jan 2021 07:38:49 +0000 (23:38 -0800)]
[DAGCombiner] Replace static helper function isConstantFPBuildVectorOrConstantFP with the identical version in SelectionDAG. NFC
Petr Hosek [Wed, 16 Dec 2020 18:26:42 +0000 (10:26 -0800)]
[CMake] Split the target side of runtimes build
Previously, llvm/runtimes/CMakeLists.txt played two different roles:
1. host side which could used to set up the build of runtimes for
different targets in the right order;
2. target side to build the runtimes for the specified target.
This change splits llvm/runtimes/CMakeLists.txt and moves the target
side to runtimes/CMakeLists laying down the foundation for the "A vision
for building the runtimes" proposal. From the user perspective, there
shouldn't be any visible difference at the moment.
Differential Revision: https://reviews.llvm.org/D93408
Craig Topper [Tue, 12 Jan 2021 07:18:57 +0000 (23:18 -0800)]
[SelectionDAG] Make isConstantIntBuildVectorOrConstantInt and isConstantFPBuildVectorOrConstantFP methods const.
Jason Molenda [Tue, 12 Jan 2021 06:17:10 +0000 (22:17 -0800)]
Add unconditional logging to debugserver for launch/attach processes
Debugging app launch/attach failures can be difficult because of
all of the messages logged to the console on a darwin system;
emitting specific messages around critical API calls can make it
easier to narrow the search for the console messages related to
the failure.
<rdar://problem/
67220442>
Differential revision: https://reviews.llvm.org/D94357
Craig Topper [Tue, 12 Jan 2021 03:58:35 +0000 (19:58 -0800)]
[CodeGen] Try to make the print of memory operand alignment a little more user friendly.
Memory operands store a base alignment that does not factor in
the effect of the offset on the alignment.
Previously the printing code only printed the base alignment if
it was different than the size. If there is an offset, the reader
would need to figure out the effective alignment themselves. This
has confused me before and someone else was recently confused on
IRC.
This patch prints the possibly offset adjusted alignment if it is
different than the size. And prints the base alignment if it is
different than the alignment. The MIR parser has been updated to
read basealign in addition to align.
Reviewed By: arsenm
Differential Revision: https://reviews.llvm.org/D94344
Kazu Hirata [Tue, 12 Jan 2021 02:48:09 +0000 (18:48 -0800)]
[llvm] Simplify string comparisons (NFC)
Identified with readability-string-compare.
Kazu Hirata [Tue, 12 Jan 2021 02:48:07 +0000 (18:48 -0800)]
[llvm] Use *Set::contains (NFC)
Kazu Hirata [Tue, 12 Jan 2021 02:48:05 +0000 (18:48 -0800)]
[llvm] Use llvm::find_if (NFC)
Shilei Tian [Tue, 12 Jan 2021 02:33:25 +0000 (21:33 -0500)]
[OpenMP] Move memory manager to plugin and make it a common interface
The lifetime of `libomptarget` and its opened plugins are not aligned
and it's hard for `libomptarget` to determine when the plugins are destroyed.
As a result, some issues (see D94256 for details) occur on some platforms.
Actually, if we take target memory as target resources, same as other resources,
such as CUDA streams, in each plugin, then the memory manager should also be in
the plugin. Also considering some platforms may want to opt out the feature, it
makes sense to move the memory manager to plugin, make it a common interface, and
let plguin developers determine whether they need it. This is what this patch does.
CUDA plugin is taken as example to show how to integrate it. In this way, we can
also get a bonus that different thresholds can be set for different platforms.
Reviewed By: jdoerfert, JonChesterfield
Differential Revision: https://reviews.llvm.org/D94379
Quentin Colombet [Tue, 12 Jan 2021 00:31:40 +0000 (16:31 -0800)]
[NFC][LICM] Minor improvements to debug output
Added a utility function in Value class to print block name and use
block labels for unnamed blocks.
Changed LICM to call this function in its debug output.
Patch by Xiaoqing Wu <xiaoqing_wu@apple.com>
Differential Revision: https://reviews.llvm.org/D93577
Heejin Ahn [Sun, 27 Dec 2020 11:02:22 +0000 (03:02 -0800)]
[WebAssembly] Ensure terminate pads are a single BB
This ensures every single terminate pad is a single BB in the form of:
```
%exn = catch $__cpp_exception
call @__clang_call_terminate(%exn)
unreachable
```
This is a preparation for HandleEHTerminatePads pass, which will be
added in a later CL and will run after CFGStackify. That pass duplicates
terminate pads with a `catch_all` instruction, and duplicating it
becomes simpler if we can ensure every terminate pad is a single BB.
Reviewed By: dschuff, tlively
Differential Revision: https://reviews.llvm.org/D94045
Kazushi (Jam) Marukawa [Fri, 4 Dec 2020 23:41:46 +0000 (08:41 +0900)]
[VE][compiler-rt] Add VE as a target of crt
SX Aurora VE is an experimental target. We upstreamed many part of
ported llvm and clang. In order to continue this move, we need to
support libraries next, then we need to show the ability of llvm for
VE through test cases. As a first step for that, we need to use
crt in compiler-rt. VE has it's own crt but they are a part of
proprietary compiler. So, we want to use crt in compiler-rt as an
alternative.
This patch enables VE as a candidate of crt in compiler-rt.
Reviewed By: phosek, compnerd
Differential Revision: https://reviews.llvm.org/D92748
Lang Hames [Mon, 11 Jan 2021 09:00:45 +0000 (20:00 +1100)]
[JITLink] Add a new PostAllocationPasses list.
Passes in the new PostAllocationPasses list will run immediately after memory
allocation and address assignment for defined symbols, and before
JITLinkContext::notifyResolved is called. These passes can set up state
associated with the addresses of defined symbols before any query for these
addresses completes.
Jonas Devlieghere [Tue, 12 Jan 2021 00:56:02 +0000 (16:56 -0800)]
[MC] Make getEHFrameSection const like every other getter (NFC)
Rob Suderman [Fri, 8 Jan 2021 22:03:12 +0000 (14:03 -0800)]
[MLIR][Linalg] Refactor transforms to use linalg::getDynOperands helper
getDynOperands behavior is commonly used in a number of passes. Refactored to
use a helper function and avoid code reuse.
Differential Revision: https://reviews.llvm.org/D94340
Tom Stellard [Tue, 12 Jan 2021 00:20:08 +0000 (16:20 -0800)]
[github] Move repo lockdown config into llvm-project repo
We were storing this in the .github repo and moving this into llvm-project
will allow us to delete the .github repo.
Hubert Tong [Mon, 11 Jan 2021 23:36:28 +0000 (18:36 -0500)]
[clang] Fix message text for `-Wpointer-sign` to account for plain char
The `-Wpointer-sign` warning text is inappropriate for describing the
incompatible pointer conversion between plain `char` and explicitly
`signed`/`unsigned` `char` (whichever plain `char` has the same range
as) and vice versa.
Specifically, in part, it reads "converts between pointers to integer
types with different sign". This patch changes that portion to read
instead as "converts between pointers to integer types where one is of
the unique plain 'char' type and the other is not" when one of the types
is plain `char`.
C17 subclause 6.5.16.1 indicates that the conversions resulting in
`-Wpointer-sign` warnings in assignment-like contexts are constraint
violations. This means that strict conformance requires a diagnostic for
the case where the message text is wrong before this patch. The lack of
an even more specialized warning group is consistent with GCC.
Reviewed By: aaron.ballman
Differential Revision: https://reviews.llvm.org/D93999
Hubert Tong [Mon, 11 Jan 2021 23:35:54 +0000 (18:35 -0500)]
NFC: Pre-commit test: -Wpointer-sign with plain char to [un]signed char
Add tests with bad message text for `-Wpointer-sign` and run them with
both signed and unsigned versions of plain `char`.
Evandro Menezes [Thu, 31 Dec 2020 03:51:41 +0000 (21:51 -0600)]
[RISCV] Define the vfclass RVV intrinsics
Define the `vfclass` IR intrinsics for the respective V instructions.
Authored-by: Roger Ferrer Ibanez <rofirrim@gmail.com>
Co-Authored-by: Evandro Menezes <evandro.menezes@sifive.com>
Differential Revision: https://reviews.llvm.org/D94356
Duncan P. N. Exon Smith [Wed, 23 Dec 2020 21:17:45 +0000 (13:17 -0800)]
ADT: Fix pointer comparison UB in SmallVector
The standard requires comparisons of pointers to unrelated storage to
use `std::less`. Split out some helpers that do that and update all the
code that was comparing using `<` and friends (mostly assertions).
Differential Revision: https://reviews.llvm.org/D93777
Roman Lebedev [Mon, 11 Jan 2021 21:36:18 +0000 (00:36 +0300)]
[SimplifyCFGPass] iterativelySimplifyCFG(): support lazy DomTreeUpdater
This boils down to how we deal with early-increment iterator
over function's basic blocks: not only we need to early-increment,
after that we also need to skip all the blocks
that are scheduled for removal, as per DomTreeUpdater.
Roman Lebedev [Mon, 11 Jan 2021 21:35:15 +0000 (00:35 +0300)]
[SimplifyCFGPass] mergeEmptyReturnBlocks(): skip blocks scheduled for removal as per DomTreeUpdater
Thus supporting lazy DomTreeUpdater mode,
where the domtree updates (and thus block removals)
aren't applied immediately, but are delayed
until last possible moment.
Roman Lebedev [Mon, 11 Jan 2021 21:29:55 +0000 (00:29 +0300)]
[NFCI][Utils/Local] removeUnreachableBlocks(): cleanup support for lazy DomTreeUpdater
When DomTreeUpdater is in lazy update mode, the blocks
that were scheduled to be removed, won't be removed
until the updates are flushed, e.g. by asking
DomTreeUpdater for a up-to-date DomTree.
From the function's current code, it is pretty evident
that the support for the lazy mode is an afterthought,
see e.g. how we roll-back NumRemoved statistic..
So instead of considering all the unreachable blocks
as the blocks-to-be-removed, simply additionally skip
all the blocks that are already scheduled to be removed
Roman Lebedev [Mon, 11 Jan 2021 21:21:25 +0000 (00:21 +0300)]
[SimplifyCFG] FoldValueComparisonIntoPredecessors(): don't insert a DomTree edge if it already exists
When we are adding edges to the terminator and potentially turning it
into a switch (if it wasn't already), it is possible that the
case we're adding will share it's destination with one of the
preexisting cases, in which case there is no domtree edge to add.
Indeed, this change does not have a test coverage change.
This failure has been exposed in an existing test coverage
by a follow-up patch that switches to lazy domtreeupdater mode,
and removes domtree verification from
SimplifyCFGOpt::simplifyOnce()/SimplifyCFGOpt::run(),
IOW it does not appear feasible to add dedicated test coverage here.
Roman Lebedev [Mon, 11 Jan 2021 21:18:23 +0000 (00:18 +0300)]
[SimplifyCFG] SimplifyBranchOnICmpChain(): don't insert a DomTree edge that already exists
BB was already always branching to EdgeBB, there is no edge to add.
Indeed, this change does not have a test coverage change.
This failure has been exposed in an existing test coverage
by a follow-up patch that switches to lazy domtreeupdater mode,
and removes domtree verification from
SimplifyCFGOpt::simplifyOnce()/SimplifyCFGOpt::run(),
IOW it does not appear feasible to add dedicated test coverage here.
Roman Lebedev [Mon, 11 Jan 2021 13:01:57 +0000 (16:01 +0300)]
[SimplifyCFG] SwitchToLookupTable(): don't insert a DomTree edge that already exists
SI is the terminator of BB, so the edge we are adding obviously already existed.
Indeed, this change does not have a test coverage change.
This failure has been exposed in an existing test coverage
by a follow-up patch that switches to lazy domtreeupdater mode,
and removes domtree verification from
SimplifyCFGOpt::simplifyOnce()/SimplifyCFGOpt::run(),
IOW it does not appear feasible to add dedicated test coverage here.
Craig Topper [Mon, 11 Jan 2021 23:03:29 +0000 (15:03 -0800)]
[RISCV] Use vmv.v.i vd, 0 instead of vmv.v.x vd, x0 for llvm.riscv.vfmv.v.f with 0.0
This matches what we use for integer 0. It's also consistent with
the scalar 'mv' pseudo that uses addi rather than add with x0.
Richard Uhler [Mon, 11 Jan 2021 22:57:05 +0000 (22:57 +0000)]
Update syntax for tensor and memref types to match parser.
Based on the comments in lib/Parser/TypeParser.cpp on the
parseMemRefType and parseTensorType functions.
Reviewed By: mehdi_amini
Differential Revision: https://reviews.llvm.org/D94262
Nathan Chancellor [Mon, 11 Jan 2021 22:39:01 +0000 (14:39 -0800)]
clang: Always pass PowerPC endian information to GNU as
When building a 64-bit big endian PowerPC Linux kernel with a 64-bit
little endian PowerPC target, the 32-bit vDSO errors:
```
$ make ARCH=powerpc CC=clang CROSS_COMPILE=powerpc64le-linux-gnu- \
pseries_defconfig arch/powerpc/kernel/vdso32/
ld.lld: error: arch/powerpc/kernel/vdso32/sigtramp.o is incompatible with elf32-powerpc
ld.lld: error: arch/powerpc/kernel/vdso32/gettimeofday.o is incompatible with elf32-powerpc
ld.lld: error: arch/powerpc/kernel/vdso32/datapage.o is incompatible with elf32-powerpc
ld.lld: error: arch/powerpc/kernel/vdso32/cacheflush.o is incompatible with elf32-powerpc
ld.lld: error: arch/powerpc/kernel/vdso32/note.o is incompatible with elf32-powerpc
ld.lld: error: arch/powerpc/kernel/vdso32/getcpu.o is incompatible with elf32-powerpc
ld.lld: error: arch/powerpc/kernel/vdso32/vgettimeofday.o is incompatible with elf32-powerpc
...
```
This happens because the endian information is missing from the call to
the assembler, even though it was explicitly passed to clang. See the
below example.
```
$ echo | clang --target=powerpc64le-linux-gnu \
--prefix=/usr/bin/powerpc64le-linux-gnu- \
-no-integrated-as -m32 -mbig-endian -### -x c -c -
".../clang-12" "-cc1" "-triple" "powerpc-unknown-linux-gnu" ...
...
"/usr/bin/powerpc64le-linux-gnu-as" "-a32" "-mppc" "-many" "-o" "-.o" "/tmp/--e69e28.s"
```
clang sets the right target with -m32 and -mbig-endian but -mbig-endian
does not make it to the assembler, resulting in a 32-bit little endian
binary. This differs from the little endian targets, which always pass
-mlittle-endian.
```
$ echo | clang --target=powerpc64-linux-gnu \
--prefix=/usr/bin/powerpc64-linux-gnu- \
-no-integrated-as -m32 -mlittle-endian -### -x c -c -
".../clang-12" "-cc1" "-triple" "powerpcle-unknown-linux-gnu" ...
...
"/usr/bin/powerpc64-linux-gnu-as" "-a32" "-mppc" "-mlittle-endian" "-many" "-o" "-.o" "/tmp/--405dbd.s"
```
Do the same thing for the big endian targets so that there is no more
error. This matches GCC's behavior, where -mbig and -mlittle are always
passed along to GNU as.
```
$ echo | powerpc64-linux-gcc -### -x c -c -
...
.../powerpc64-linux/bin/as -a64 -mpower4 -many -mbig -o -.o /tmp/ccVn7NAm.s
...
$ echo | powerpc64le-linux-gcc -### -x c -c -
...
.../powerpc64le-linux/bin/as -a64 -mpower8 -many -mlittle -o -.o /tmp/ccPN9ato.s
...
```
Reviewed By: nickdesaulniers, MaskRay
Differential Revision: https://reviews.llvm.org/D94442
Fraser Cormack [Fri, 8 Jan 2021 11:47:12 +0000 (11:47 +0000)]
[RISCV] Add scalable vector vselect ISel patterns
Reviewed By: craig.topper
Differential Revision: https://reviews.llvm.org/D94294
Shilei Tian [Mon, 11 Jan 2021 22:34:17 +0000 (17:34 -0500)]
[OpenMP] Take elf_common.c as a interface library
For now `elf_common.c` is taken as a common part included into
different plugin implementations directly via
`#include "../../common/elf_common.c"`, which is not a best practice. Since it
is simple enough such that we don't need to create a real library for it, we just
take it as a interface library so that other targets can link it directly. Another
advantage of this method is, we don't need to add the folder into header search
path which can potentially pollute the search path.
VE and AMD platforms have not been tested because I don't have target machines.
Reviewed By: JonChesterfield
Differential Revision: https://reviews.llvm.org/D94443
Hongtao Yu [Sat, 9 Jan 2021 06:57:03 +0000 (22:57 -0800)]
Rename debug linkage name with -funique-internal-linkage-names
Functions that are renamed under -funique-internal-linkage-names have their debug linkage name updated as well.
Reviewed By: dblaikie
Differential Revision: https://reviews.llvm.org/D93747
Fangrui Song [Mon, 11 Jan 2021 21:39:03 +0000 (13:39 -0800)]
CGDebugInfo: Delete unused parameters
Nikita Popov [Mon, 11 Jan 2021 21:32:59 +0000 (22:32 +0100)]
[SCCP] Fix misclassified conditions in test (NFC)
Nikita Popov [Mon, 11 Jan 2021 19:42:47 +0000 (20:42 +0100)]
[PredicateInfo] Add test for one unknown condition in and/or (NFC)
Test the case where one part of and/or is an icmp, while the other
one is an arbitrary value.
Aart Bik [Mon, 11 Jan 2021 19:34:10 +0000 (11:34 -0800)]
[mlir][vector] verify memref of vector memory ops
This ensures the memref base + indices expression is well-formed
Reviewed By: ThomasRaoux, ftynse
Differential Revision: https://reviews.llvm.org/D94441
Fraser Cormack [Mon, 11 Jan 2021 14:02:35 +0000 (14:02 +0000)]
[RISCV] Add scalable vector fadd/fsub/fmul/fdiv ISel patterns
Original patch by @rogfer01.
This patch adds ISel patterns for the above operations to the
corresponding vector/vector and vector/scalar RVV instructions, as well
as extra patterns to match operand-swapped scalar/vector vfrsub and
vfrdiv.
Authored-by: Roger Ferrer Ibanez <rofirrim@gmail.com>
Co-Authored-by: Fraser Cormack <fraser@codeplay.com>
Reviewed By: craig.topper
Differential Revision: https://reviews.llvm.org/D94408
Fangrui Song [Mon, 11 Jan 2021 21:25:20 +0000 (13:25 -0800)]
Add an assert to CGDebugInfo::getTypeOrNull
Richard Smith [Mon, 11 Jan 2021 21:12:43 +0000 (13:12 -0800)]
[c++20] Don't consider string literal operator templates for numeric
literals.
A literal interpretation of the standard wording allows this, but it was
never intended that string literal operator templates would be used for
anything other than user-defined string literals.
Siva Chandra Reddy [Mon, 11 Jan 2021 20:35:34 +0000 (12:35 -0800)]
[libc][NFC] Make __support/common.h an in tree header.
It was previously a generated header. It can easily converted to a
generated header if required in future.
Reviewed By: michaelrj
Differential Revision: https://reviews.llvm.org/D94445
Bjorn Pettersson [Mon, 11 Jan 2021 15:19:01 +0000 (16:19 +0100)]
[GlobalISel] Map extractelt to G_EXTRACT_VECTOR_ELT
Before this patch there was generic mapping from vector_extract
to G_EXTRACT_VECTOR_ELT added in SelectionDAGCompat.td. That
mapping is now replaced by a mapping from extractelt instead.
The reasoning is that vector_extract is marked as deprecated,
so it is assumed that a majority of targets will use extractelt
and not vector_extract (and that the long term solution for all
targets would be to use extractelt).
Targets like AArch64 that still use vector_extract can add an
additional mapping from the deprecated vector_extract as target
specific tablegen definitions. Such a mapping is added for AArch64
in this patch to avoid breaking tests.
When adding the extractelt => G_EXTRACT_VECTOR_ELT mapping we
triggered some new code paths in GlobalISelEmitter, ending up in
an assert when trying to import a pattern containing EXTRACT_SUBREG
for ARM. Therefore this patch also adds a "failedImport" warning
for that situation (instead of hitting the assert).
Differential Revision: https://reviews.llvm.org/D93416
Sanjay Patel [Mon, 11 Jan 2021 20:36:22 +0000 (15:36 -0500)]
[InstCombine] reduce icmp(ashr X, C1), C2 to sign-bit test
This is a more basic pattern that we should handle before trying to solve:
https://llvm.org/PR48640
There might be a better way to think about this because the pre-condition
that I came up with (number of sign bits in the compare constant) misses a
potential transform for each of ugt and ult as commented on in the test file.
Tried to model this is in Alive:
https://rise4fun.com/Alive/juX1
...but I couldn't get the ComputeNumSignBits() pre-condition to work as
expected, so replaced with leading 0/1 preconditions instead.
Name: ugt
Pre: countLeadingZeros(C2) <= C1 && countLeadingOnes(C2) <= C1
%a = ashr %x, C1
%r = icmp ugt i8 %a, C2
=>
%r = icmp slt i8 %x, 0
Name: ult
Pre: countLeadingZeros(C2) <= C1 && countLeadingOnes(C2) <= C1
%a = ashr %x, C1
%r = icmp ult i4 %a, C2
=>
%r = icmp sgt i4 %x, -1
Also approximated in Alive2:
https://alive2.llvm.org/ce/z/u5hCcz
https://alive2.llvm.org/ce/z/__szVL
Differential Revision: https://reviews.llvm.org/D94014
Mehdi Amini [Mon, 11 Jan 2021 20:42:10 +0000 (20:42 +0000)]
Revert "[mlir][linalg] Support parsing attributes in named op spec"
This reverts commit
df86f15f0c53c395dac5a14aba08745bc12b9b9b.
The gcc-5 build was broken by this change:
mlir/tools/mlir-linalg-ods-gen/mlir-linalg-ods-gen.cpp:1275:77: required from here
/usr/include/c++/5/ext/new_allocator.h:120:4: error: no matching function for call to 'std::pair<const std::__cxx11::basic_string<char>, {anonymous}::TCParser::RegisteredAttr>::pair(llvm::StringRef&, {anonymous}::TCParser::RegisteredAttr'
Stella Laurenzo [Sun, 10 Jan 2021 03:01:39 +0000 (19:01 -0800)]
[mlir][CAPI] Introduce standard source layout for mlir-c dialect registration.
* Registers a small set of sample dialects.
* NFC with respect to existing C-API symbols but some headers have been moved down a level to the Dialect/ sub-directory.
* Adds an additional entry point per dialect that is needed for dynamic discovery/loading.
* See discussion: https://llvm.discourse.group/t/dialects-and-the-c-api/2306/16
Differential Revision: https://reviews.llvm.org/D94370
Stella Laurenzo [Sun, 10 Jan 2021 01:14:47 +0000 (17:14 -0800)]
Enable python bindings for tensor, shape and linalg dialects.
* We've got significant missing features in order to use most of these effectively (i.e. custom builders, region-based builders).
* We presently also lack a mechanism for actually registering these dialects but they can be use with contexts that allow unregistered dialects for further prototyping.
Differential Revision: https://reviews.llvm.org/D94368
Thomas Raoux [Mon, 11 Jan 2021 18:35:30 +0000 (10:35 -0800)]
[mlir][vector] Add side-effect information to different load/store ops
Differential Revision: https://reviews.llvm.org/D94434
Mircea Trofin [Mon, 11 Jan 2021 17:34:45 +0000 (09:34 -0800)]
[NFC] Disallow unused prefixes under llvm/test/CodeGen
This patch finishes addressing unused prefixes under CodeGen: 2
remaining tests fixed, and then undo-ing the lit.local.cfg changes under
various subdirs and moving the policy under CodeGen.
Differential Revision: https://reviews.llvm.org/D94430
Abhina Sreeskantharajan [Mon, 11 Jan 2021 20:13:40 +0000 (15:13 -0500)]
[tools] Mark output of tools as text if it is really text
This is a continuation of https://reviews.llvm.org/D67696. The following tools also need to set the OF_Text flag correctly.
- llvm-profdata
- llvm-link
Reviewed By: rnk
Differential Revision: https://reviews.llvm.org/D94313
Nathan James [Mon, 11 Jan 2021 20:12:53 +0000 (20:12 +0000)]
[ADT] Add makeIntrusiveRefCnt helper function
Works like std::make_unique but for IntrusiveRefCntPtr objects.
See https://lists.llvm.org/pipermail/llvm-dev/2021-January/147729.html
Reviewed By: dblaikie, MaskRay
Differential Revision: https://reviews.llvm.org/D94440
River Riddle [Mon, 11 Jan 2021 19:55:09 +0000 (11:55 -0800)]
[mlir][IR][NFC] Move the definitions of Complex/Function/Integer/Opaque/TupleType to ODS
The type tablegen backend now has enough support to represent these types well enough, so we can now move them to be declaratively defined.
Differential Revision: https://reviews.llvm.org/D94275
River Riddle [Mon, 11 Jan 2021 19:55:00 +0000 (11:55 -0800)]
[mlir][TypeDefGen] Add support for adding builders when generating a TypeDef
This allows for specifying additional get/getChecked methods that should be generated on the type, and acts similarly to how OpBuilders work. TypeBuilders have two additional components though:
* InferredContextParam
- Bit indicating that the context parameter of a get method is inferred from one of the builder parameters
* checkedBody
- A code block representing the body of the equivalent getChecked method.
Differential Revision: https://reviews.llvm.org/D94274
River Riddle [Mon, 11 Jan 2021 19:54:51 +0000 (11:54 -0800)]
[mlir][ODS] Add a C++ abstraction for OpBuilders
This removes the need for OpDefinitionsGen to use raw tablegen API, and will also
simplify adding builders to TypeDefs as well.
Differential Revision: https://reviews.llvm.org/D94273
Tony Tye [Sat, 9 Jan 2021 10:20:42 +0000 (10:20 +0000)]
[NFC][AMDGPU] Clarify memory model support for volatile
Reorder the AMDGPUUage description of the memory model code sequences
for volatile so clear that it applies independent of the nontemporal
setting.
Differential Revision: https://reviews.llvm.org/D94358