Sam McCall [Wed, 10 Feb 2021 21:20:23 +0000 (22:20 +0100)]
[clangd] Remove redundant -fno-delayed-template-parsing in tests. NFCI
We now (since a while) turn this off centrally in ParsedAST and CodeComplete.
Sam McCall [Thu, 28 Jan 2021 00:16:47 +0000 (01:16 +0100)]
[clangd] Remove support for pre-standard semanticHighlighting notification
This is obsoleted by the standard semanticTokens request family.
As well as the protocol details, this allows us to remove a bunch of plumbing
around pushing highlights to clients.
This should not land until the new protocol has feature parity, see D77702.
Differential Revision: https://reviews.llvm.org/D95576
Jacques Pienaar [Wed, 10 Feb 2021 20:44:18 +0000 (12:44 -0800)]
Make gCrashRecoveryEnabled thread local
If context is enabled/disabled and queried concurrently then this
results in a data race/TSAN failure with RunSafely (where boolean
variable was not locked).
There doesn't seem to be a reasonable way to enable threads that enable
and disable recovery in parallel (without also keeping
gCrashRecoveryEnabled's lock held during Fn execution which seems
undesirable). This makes enable checking if enabled thread local and
consistent with other thread local usage of crash context here.
Differential Revision: https://reviews.llvm.org/D93907
Hongtao Yu [Mon, 8 Feb 2021 06:49:20 +0000 (22:49 -0800)]
[CSSPGO] Unblock optimizations with pseudo probe instrumentation.
The IR/MIR pseudo probe intrinsics don't get materialized into real machine instructions and therefore they don't incur runtime cost directly. However, they come with indirect cost by blocking certain optimizations. Some of the blocking are intentional (such as blocking code merge) for better counts quality while the others are accidental. This change unblocks perf-critical optimizations that do not affect counts quality. They include:
1. IR InstCombine, sinking load operation to shorten lifetimes.
2. MIR LiveRangeShrink, similar to #1
3. MIR TwoAddressInstructionPass, i.e, opeq transform
4. MIR function argument copy elision
5. IR stack protection. (though not perf-critical but nice to have).
Reviewed By: wmi
Differential Revision: https://reviews.llvm.org/D95982
Ilya Tokar [Wed, 10 Feb 2021 18:05:19 +0000 (13:05 -0500)]
[libc++] Use builtins in more math.h functions.
Not using builtins doesn't always imply worse code,
but for e. g. isinf, this is 30%+ faster.
Before:
name time/op
BM_isinf 2.14ns ± 2%
After:
name time/op
BM_isinf 1.33ns ± 2%
Reviewed By: #libc, ldionne
Differential Revision: https://reviews.llvm.org/D88854
Adrian Prantl [Wed, 10 Feb 2021 20:14:45 +0000 (12:14 -0800)]
Add missing nullptr check.
salvageDebugInfoImpl() may fail and return a nullptr.
Philip Reames [Wed, 10 Feb 2021 20:06:46 +0000 (12:06 -0800)]
[SCEV] Add a missing AssumptionCache parameter
The AssumptionCache mechanism is used to feed assumes into known bits computations. Most places in SCEV passed it in, but one place appears to have been missed.
Spotted via inspection, don't have a test case which actually exercises this, but it seemed like an obvious fixit.
Sanjay Patel [Wed, 10 Feb 2021 19:57:31 +0000 (14:57 -0500)]
[InstCombine] fold lshr(mul X, SplatC), C2
This is a special-case multiply that replicates bits of
the source operand. We need this fold to avoid regression
if we make canonicalization to `mul` more aggressive for
shl+or patterns.
I did not see a way to make Alive generalize the bit width
condition for even-number-of-bits only, but an example of
the proof is:
Name: i32
Pre: isPowerOf2(C1 - 1) && log2(C1) == C2 && (C2 * 2 == width(C2))
%m = mul nuw i32 %x, C1
%t = lshr i32 %m, C2
=>
%t = and i32 %x, C1 - 2
Name: i14
%m = mul nuw i14 %x, 129
%t = lshr i14 %m, 7
=>
%t = and i14 %x, 127
https://rise4fun.com/Alive/e52
Sanjay Patel [Wed, 10 Feb 2021 19:40:08 +0000 (14:40 -0500)]
[InstCombine] add tests for lshr with mul; NFC
Mehdi Amini [Wed, 10 Feb 2021 20:01:20 +0000 (20:01 +0000)]
Fix StridedMemRefType operator[] SFINAE to allow correctly selecting the `int64_t` overload for non-container operands
Pavel Labath [Wed, 10 Feb 2021 19:45:30 +0000 (20:45 +0100)]
[lldb/test] Move and improve TestPlatformProcessConnect.py
Although it is located under tools/lldb-server, this test is very
different that other lldb-server tests. The most important distinction
is that it does not test lldb-server directly, but rather interacts with
it through the lldb client. It also tests the relevant client
functionality (the platform connect command, which is even admitted in
the test name). The fact that this test is structured as a lldb-server
test means it cannot access most of the goodies available to the
"normal" lldb tests (the runCmd function, which it reimplements; the
run_break_set_by_symbol utility function; etc.).
This patch makes it a full-fledged lldb this, and rewrites the relevant
bits to make use of the standard features. I also move the test into the
"commands" subtree to better reflect its new status.
Nawrin Sultana [Fri, 30 Oct 2020 18:04:25 +0000 (13:04 -0500)]
[OpenMP] Add lower and upper bound in num_teams clause
This patch adds lower-bound and upper-bound to num_teams clause
according to OpenMP 5.1 specification. The initial number of teams
created is implementation defined, but it will be greater than or
equal to lower-bound and less than or equal to upper-bound. If
num_teams clause is not specified, the number of teams created is
implementation defined, but it will be greater or equal to 1.
Differential Revision: https://reviews.llvm.org/D95820
Jing Pu [Wed, 10 Feb 2021 19:58:05 +0000 (11:58 -0800)]
Change type constraint of the "index" in "shape.split_at" to Shape_SizeOrIndexType
Make the type contraint consistent with other shape dialect operations.
Reviewed By: jpienaar
Differential Revision: https://reviews.llvm.org/D96377
Aart Bik [Wed, 10 Feb 2021 00:22:22 +0000 (16:22 -0800)]
[mlir][sparse] sparse tensor storage implementation
This revision connects the generated sparse code with an actual
sparse storage scheme, which can be initialized from a test file.
Lacking a first-class citizen SparseTensor type (with buffer),
the storage is hidden behind an opaque pointer with some "glue"
to bring the pointer back to tensor land. Rather than generating
sparse setup code for each different annotated tensor (viz. the
"pack" methods in TACO), a single "one-size-fits-all" implementation
has been added to the runtime support library. Many details and
abstractions need to be refined in the future, but this revision
allows full end-to-end integration testing and performance
benchmarking (with on one end, an annotated Lingalg
op and, on the other end, a JIT/AOT executable).
Reviewed By: nicolasvasilache, bixia
Differential Revision: https://reviews.llvm.org/D95847
Christopher Di Bella [Mon, 8 Feb 2021 05:08:25 +0000 (05:08 +0000)]
[libcxx] adds concepts `std::invocable` and `std::regular_invocable`
Implements parts of:
- P0898R3 Standard Library Concepts
- P1754 Rename concepts to standard_case for C++20, while we still can
Differential Revision: https://reviews.llvm.org/D96235
Christopher Di Bella [Thu, 4 Feb 2021 03:41:44 +0000 (03:41 +0000)]
[libcxx] adds concept std::derived_from
Implements parts of:
- P0898R3 Standard Library Concepts
- P1754 Rename concepts to standard_case for C++20, while we still can
Reviewed By: ldionne, #libc
Differential Revision: https://reviews.llvm.org/D74292
Michael Kruse [Wed, 10 Feb 2021 19:21:27 +0000 (13:21 -0600)]
[Polly] Added dedicated test for working -O3 pipeline.
Test the NewPM as well as the legacy PM.
Michael Kruse [Wed, 10 Feb 2021 19:04:58 +0000 (13:04 -0600)]
[Polly] Do not use -O3 pipeline for single pass test.
Jameson Nash [Wed, 10 Feb 2021 18:31:42 +0000 (13:31 -0500)]
Renovate CMake files in the `llvm-exegesis` tool.
This attempts to move all tools over to using `add_llvm_library` for
better consistency. After doing this, I noticed it ended up as nearly a
reimplementation of https://reviews.llvm.org/rL342148, which later got
reverted in r342336 (
b09a8c9bd9b819741b38071a7ccd95042ef2643a).
With ccache and ninja on a large core machine (40), I haven't run into
build errors, so I'm hopeful it's better now, though it doesn't seem to
be any different / new.
Reviewed By: stephenneuendorffer
Differential Revision: https://reviews.llvm.org/D90970
Arthur Eubanks [Fri, 5 Feb 2021 03:34:09 +0000 (19:34 -0800)]
[opt][NewPM] Add a --print-passes flag to print all available passes
It seems nicer to list passes given a flag rather than displaying all
passes in opt --help.
This is awkwardly structured because a PassBuilder is required, but
reusing the PassBuilder in runPassPipeline() doesn't work because we
read the input IR before getting to runPassPipeline(). So printing the
list of passes needs to happen before reading the input IR. If we remove
the legacy PM code in main() and move everything from NewPMDriver.cpp
into opt.cpp, we can create the PassBuilder before reading IR and check
if we should print the list of passes and exit. But until then this hack
seems fine.
Compared to the legacy PM, the new PM passes are lacking descriptions.
We'll need to figure out a way to add descriptions if we think this is
important.
Also, this only works for passes specified in PassRegistry.def. If we
want to print other custom registered passes, we'll need a different
mechanism.
Reviewed By: asbirlea
Differential Revision: https://reviews.llvm.org/D96101
Craig Topper [Wed, 10 Feb 2021 19:20:07 +0000 (11:20 -0800)]
[RISCV] Remove superfluous semicolon. NFC
Christopher Di Bella [Thu, 21 Jan 2021 02:33:04 +0000 (02:33 +0000)]
[libcxx] adds [concepts.arithmetic]
Implements parts of:
* P0898R3 Standard Library Concepts
* P1754 Rename concepts to standard_case for C++20, while we still can
Differential Revision: https://reviews.llvm.org/D88131
Nick Desaulniers [Wed, 10 Feb 2021 18:45:12 +0000 (10:45 -0800)]
[Thumb2] support `movs pc, lr` alias for `subs pc, lr, #0`/`eret`
This is used by the Linux kernel built with CONFIG_THUMB2_KERNEL.
Because different operands are not permitted to `movs`, the diagnostics now provide multiple suggestions along the lines of using a non-pc destination operand or lr source operand.
Forked from D95586.
Signed-off-by: Nick Desaulniers <ndesaulniers@google.com>
Reviewed By: DavidSpickett
Differential Revision: https://reviews.llvm.org/D96304
Mehdi Amini [Wed, 3 Feb 2021 21:45:00 +0000 (21:45 +0000)]
Add convenience C++ helper to manipulate ranked strided memref
Reland
11f32a41c21 that was reverted in
e49967fbd90 after fixing the build.
Differential Revision: https://reviews.llvm.org/D96192
Arthur Eubanks [Fri, 5 Feb 2021 03:34:09 +0000 (19:34 -0800)]
Specify that some flags are legacy PM-specific
Reviewed By: asbirlea
Differential Revision: https://reviews.llvm.org/D96100
Louis Dionne [Wed, 10 Feb 2021 18:30:38 +0000 (13:30 -0500)]
[libc++] Remove c++98 Lit features in the test suite
We don't populate a Lit feature named c++98 since
31cbe0f240f.
Erik Pilkington [Wed, 10 Feb 2021 16:06:23 +0000 (11:06 -0500)]
[clang] Add support for attribute 'swift_async_error'
This attribute specifies how an error is represented for a swift async method.
rdar://
71941280
Differential revision: https://reviews.llvm.org/D96175
Craig Topper [Wed, 10 Feb 2021 18:15:57 +0000 (10:15 -0800)]
[RISCV] Add support for matching .vf forms of fadd/fsub/fmul/fdiv/fma for fixed vectors.
fma+neg will come in a different patch since I haven't done it for .vv
yet either.
Differential Revision: https://reviews.llvm.org/D96375
Tom Stellard [Wed, 10 Feb 2021 15:36:49 +0000 (07:36 -0800)]
[CMake] Remove some dead code in llvm_install_library_symlink()
Reviewed By: smeenai
Differential Revision: https://reviews.llvm.org/D95666
Mehdi Amini [Wed, 10 Feb 2021 18:09:38 +0000 (18:09 +0000)]
Revert "Add convenience C++ helper to manipulate ranked strided memref"
This reverts commit
11f32a41c2144aeec80d1dce8cc6908fa91794a3.
The build is broken because this commit conflits with the refactoring of
the DialectRegistry APIs in the context. It'll reland shortly after
fixing the API usage.
Craig Topper [Wed, 10 Feb 2021 17:40:28 +0000 (09:40 -0800)]
[RISCV] Add support for selecting vrgather.vx/vi for fixed vector splat shuffles.
The test cases extract a fixed element from a vector and splat it
into a vector. This gets DAG combined into a splat shuffle.
I've used some very wide vectors in the test to make sure we have
at least a couple tests where the element doesn't fit into the
uimm5 immediate of vrgather.vi so we fall back to vrgather.vx.
Reviewed By: frasercrmck
Differential Revision: https://reviews.llvm.org/D96186
Christopher Di Bella [Mon, 8 Feb 2021 03:01:27 +0000 (03:01 +0000)]
[libcxx] adds concept `std::copy_constructible`
Implements parts of:
- P0898R3 Standard Library Concepts
- P1754 Rename concepts to standard_case for C++20, while we still can
Depends on D96230
Differential Revision: https://reviews.llvm.org/D96232
Fangrui Song [Wed, 10 Feb 2021 17:47:10 +0000 (09:47 -0800)]
DebugInfo/Symbolize: Retrieve filename from the preceding STT_FILE for .symtab symbolization
The ELF spec says:
> STT_FILE: Conventionally, the symbol's name gives the name of the source file associated with the object file. A file symbol has STB_LOCAL binding, its section index is SHN_ABS, and it precedes the other STB_LOCAL symbols for the file, if it is present.
For a local symbol, the preceding STT_FILE symbol is almost always in the same
file[1]. GNU addr2line uses this heuristic to retrieve the filename associated
with a local symbol (e.g. internal linkage functions in C/C++).
GNU addr2line can assign STT_FILE filename to a non-local symbol, too, but the trick
only works if no regular symbol precede STT_FILE. This patch does not implement this corner case
(not useful for most executables which have more than one files).
In case of filename mismatch between .debug_line & .symtab, arbitrarily make .debug_line win.
[1]: LLD does not synthesize STT_FILE symbols
(https://bugs.llvm.org/show_bug.cgi?id=48023 see also
https://sourceware.org/bugzilla/show_bug.cgi?id=26822). An assembly file
without `.file` directives can cause mis-attribution. This is an edge case.
Differential Revision: https://reviews.llvm.org/D95927
Fangrui Song [Wed, 10 Feb 2021 17:44:13 +0000 (09:44 -0800)]
[llvm-cfi-verify] Set UseSymbolTable to false
parseSectionContents expects to skip regions not described by DWARF. With my
pending DebugInfo/Symbolize change, the filename can be recovered and there
will be more IndirectInstructions entries.
Mehdi Amini [Wed, 3 Feb 2021 21:45:00 +0000 (21:45 +0000)]
Add convenience C++ helper to manipulate ranked strided memref
Differential Revision: https://reviews.llvm.org/D96192
Christopher Di Bella [Mon, 8 Feb 2021 02:00:42 +0000 (02:00 +0000)]
[libcxx] adds concept `std::move_constructible`
Implements parts of:
- P0898R3 Standard Library Concepts
- P1754 Rename concepts to standard_case for C++20, while we still can
Depends on D77961
Differential Revision: https://reviews.llvm.org/D96230
Fangrui Song [Wed, 10 Feb 2021 17:19:05 +0000 (09:19 -0800)]
[Polly] Fix -Wunused-lambda-capture
Fangrui Song [Wed, 10 Feb 2021 17:17:13 +0000 (09:17 -0800)]
[Polly] Fix -DPOLLY_ENABLE_GPGPU_CODEGEN=off build after
222d380d2f57cc71bb613b5c01ecf17cd1f61fa2
Mitch Phillips [Wed, 10 Feb 2021 16:38:34 +0000 (08:38 -0800)]
[GWP-ASan] Add back some headers removed by IWYU.
These headers are required for Android.
Reviewed By: eugenis
Differential Revision: https://reviews.llvm.org/D96374
Nicolas Vasilache [Wed, 10 Feb 2021 15:12:05 +0000 (15:12 +0000)]
[mlir][Linalg] Fix pad hoisting.
This revision fixes the indexing logic into the packed tensor that result from hoisting padding. Previously, the index was incorrectly set to the loop induction variable when in fact we need to compute the iteration count (i.e. `(iv - lb).ceilDiv(step)`).
Differential Revision: https://reviews.llvm.org/D96417
Tom Weaver [Wed, 10 Feb 2021 16:40:07 +0000 (16:40 +0000)]
Revert "Revert "[clang][driver] Only warn once about invalid library values""
This reverts commit
a743702a1f4880e4492196b1ce9a9a63e0b4c075.
Test was fixed in
c6a1b16db7dc7a0af8951b39f29ddbe639a98a3b
Tom Weaver [Wed, 10 Feb 2021 16:37:34 +0000 (16:37 +0000)]
Revert "[clang][driver] Only warn once about invalid library values"
This reverts commit
a6439b52088b1d58d8e7aa9891c9011648710593.
Caused buildbot failure http://lab.llvm.org:8014/#/builders/125/builds/125
Colin Finck [Wed, 10 Feb 2021 16:00:04 +0000 (11:00 -0500)]
[libc++] Fix copy-paste mistake in __threading_support
Differential Revision: https://reviews.llvm.org/D96115
Jeremy Morse [Wed, 10 Feb 2021 15:40:47 +0000 (15:40 +0000)]
Reland [DWARF] Location-less inlined variables should not have DW_TAG_variable
Originally landed in
ddc2f1e3fb4 and reverted in
d32deaab4d because of
a Generic test objecting. That was fixed up in
013613964fd9. Original
landing commit message follows:
[DWARF] Location-less inlined variables should not have DW_TAG_variable
Discussed in this thread:
https://lists.llvm.org/pipermail/llvm-dev/2021-January/148139.html
DwarfDebug::collectEntityInfo accidentally distinguishes between variable
locations that never have a location specified, and variable locations that
have an empty location specified. The latter leads to the creation of an
empty variable referring to the abstract origin.
Fix this by seeking a non-empty location before producing a concrete
entity, to guarantee a DW_AT_location will be produced. Other loops in
collectEntityInfo and endFunctionImpl take care of examining the
retainedNodes collection and ensuring optimised-out variables are created.
Differential Revision: https://reviews.llvm.org/D95617
Paul Robinson [Tue, 9 Feb 2021 22:41:52 +0000 (14:41 -0800)]
Avoid conflicts between debug-info and pseudo-probe profiling
After D93264, using both -fdebug-info-for-profiling and
-fpseudo-probe-for-profiling will cause the compiler to crash.
Diagnose these conflicting options in the driver.
Also, the existing CodeGen test was using the driver when it should be
running cc1.
Differential Revision: https://reviews.llvm.org/D96354
Jay Foad [Wed, 10 Feb 2021 14:37:29 +0000 (14:37 +0000)]
[AMDGPU] Add another test case for combining DS reads
Jay Foad [Tue, 9 Feb 2021 17:19:47 +0000 (17:19 +0000)]
[AMDGPU] Fix comments in SILoadStoreOptimizer::offsetsCanBeCombined
Nico Weber [Wed, 10 Feb 2021 14:45:04 +0000 (09:45 -0500)]
clang: try to fix Driver/undefined-libs.cpp on non-linux
Luís Marques [Wed, 10 Feb 2021 14:24:15 +0000 (14:24 +0000)]
[DAGCombiner] Don't fold FCOPYSIGN vector sign operand casts
Avoid doing the following combine for vector types:
```
copysign(x, fp_extend(y)) -> copysign(x, y)
copysign(x, fp_round(y)) -> copysign(x, y)
```
That combine seemed to impede the selection of vector instruction and cause
a mess in some circumstances.
Differential Revision: https://reviews.llvm.org/D96037
Nicolas Vasilache [Wed, 10 Feb 2021 13:15:23 +0000 (13:15 +0000)]
[mlir][Linalg] Add a vectorization pattern for linalg::PadTensorOp
The new pattern is exercised from the TestLinalgTransforms pass.
Differential Revision: https://reviews.llvm.org/D96410
Timm Bäder [Wed, 10 Feb 2021 13:34:04 +0000 (14:34 +0100)]
[clang][NFC] Fix undefined-libs tests
Not all platforms accept -stdlib or -rtlib. Instead of complaining about
the wrong argument to these options, clang complains about the option
itself being present.
Pass an appropriate -target to the clang invocations.
Nico Weber [Wed, 10 Feb 2021 13:59:07 +0000 (08:59 -0500)]
[gn build] (manually) port
e89fcbfad6a3
Daniel Cederman [Wed, 10 Feb 2021 13:50:38 +0000 (14:50 +0100)]
[Sparc] Support relocatable expressions in the assembler
Allow assembler expressions to start with an identifier. This allows for expressions such as
```
b symbol + 4
```
and
```
mov symEnd - symStart, %g1
```
The patch builds upon https://reviews.llvm.org/D47136.
Reviewed By: joerg
Differential Revision: https://reviews.llvm.org/D47458
Andrzej Warzynski [Wed, 10 Feb 2021 12:10:07 +0000 (12:10 +0000)]
[flang][driver] Add missing dependency in unit tests (nfc)
The following patch revealed a missing dependency in the CMake script
for Flang driver unit tests:
* https://reviews.llvm.org/D96032
The following buildbots are failing ("BUILD_SHARED_LIBS" is set to ON):
* http://lab.llvm.org:8011/#/builders/134/builds/1840
* http://lab.llvm.org:8011/#/builders/66/builds/1785
* http://lab.llvm.org:8011/#/builders/33/builds/2436
This patch adds the missing dependency.
From what I can see, FortranSemantics and FortranParser are only
indirect dependencies of FlangFrontendTests and IIUC shouldn't be
required here. This is something that we should revisit at some point.
In this patch I focus on fixing the build.
Jan Svoboda [Wed, 10 Feb 2021 12:54:08 +0000 (13:54 +0100)]
[clang][cli] Declare local variable for marshalling macros
Some Windows build bots report `FileSystemOpts` and `MigratorOpts` as undeclared. This fix renames the parameter and declares a local variable with the original name.
Jan Svoboda [Wed, 10 Feb 2021 11:47:06 +0000 (12:47 +0100)]
[clang][cli] Extract FileSystem and Migrator options parsing/generation
This patch splits out the last two option groups (`Filesystem` and `Migrator`) into their own `Parse`/`Generate` functions.
This effectively removes the need for `parseSimpleArgs` and marshalling block in `CompilerInvocation::generateCC1CommandLine`.
The two new `Parse`/`Generate` functions are not part of the round-trip, because they contain no custom code and the very next patch starts round-tripping the whole `CompilerInvocation`.
Reviewed By: dexonsmith
Differential Revision: https://reviews.llvm.org/D96278
Jan Svoboda [Wed, 10 Feb 2021 11:24:19 +0000 (12:24 +0100)]
[clang][cli] Generate and round-trip Diagnostic options
This patch implements generation of remaining diagnostic options and tests it by performing parse-generate-parse round trip.
Reviewed By: dexonsmith
Differential Revision: https://reviews.llvm.org/D96274
Jan Svoboda [Wed, 10 Feb 2021 10:47:33 +0000 (11:47 +0100)]
[clang][cli] Generate and round-trip DependencyOutput options
This patch implements generation of remaining dependency output options and tests it by performing parse-generate-parse round trip.
Reviewed By: dexonsmith
Differential Revision: https://reviews.llvm.org/D96273
Alex Zinenko [Wed, 10 Feb 2021 09:11:50 +0000 (10:11 +0100)]
[mlir] avoid exposing mutable DialectRegistry from MLIRContext
MLIRContext allows its users to access directly to the DialectRegistry it
contains. While sometimes useful for registering additional dialects on an
already existing context, this breaks the encapsulation by essentially giving
raw accesses to a part of the context's internal state. Remove this mutable
access and instead provide a method to append a given DialectRegistry to the
one already contained in the context. Also provide a shortcut mechanism to
construct a context from an already existing registry, which seems to be a
common use case in the wild. Keep read-only access to the registry contained in
the context in case it needs to be copied or used for constructing another
context.
With this change, DialectRegistry is no longer concerned with loading the
dialects and deciding whether to invoke delayed interface registration. Loading
is concentrated in the MLIRContext, and the functionality of the registry
better reflects its name.
Depends On D96137
Reviewed By: mehdi_amini
Differential Revision: https://reviews.llvm.org/D96331
Alex Zinenko [Wed, 10 Feb 2021 09:11:40 +0000 (10:11 +0100)]
[mlir] enable delayed registration of dialect interfaces
This introduces a mechanism to register interfaces for a dialect without making
the dialect itself depend on the interface. The registration request happens on
DialectRegistry and, if the dialect has not been loaded yet, the actual
registration is delayed until the dialect is loaded. It requires
DialectRegistry to become aware of the context that contains it and the context
to expose methods for querying if a dialect is loaded.
This mechanism will enable a simple extension mechanism for dialects that can
have interfaces defined outside of the dialect code. It is particularly helpful
for, e.g., translation to LLVM IR where we don't want the dialect itself to
depend on LLVM IR libraries.
Reviewed By: mehdi_amini
Differential Revision: https://reviews.llvm.org/D96137
Fraser Cormack [Tue, 9 Feb 2021 12:09:10 +0000 (12:09 +0000)]
[RISCV] Add support for selecting vid.v from build_vector
This patch optimizes a build_vector "index sequence" and lowers it to
the existing custom RISCVISD::VID node. This pattern is common in
autovectorized code.
The custom node was updated to allow it to be used by both scalable and
fixed-length vectors, thus avoiding pattern duplication.
Reviewed By: craig.topper
Differential Revision: https://reviews.llvm.org/D96332
Jeremy Morse [Wed, 10 Feb 2021 10:27:00 +0000 (10:27 +0000)]
Reapply [DebugInfo] Re-engineer a test to be stricter, add XFails
Was
e05c10380ce, reverted in
d7d0b17de77, see D95617 for details. I've
added "arm64" to the XFail list (as well as aarch64), will follow up on
the mailing list about whether there's anything else to be done.
Jan Svoboda [Wed, 10 Feb 2021 10:33:24 +0000 (11:33 +0100)]
[clang][cli] Generate and round-trip Target options
This patch implements generation of remaining target options and tests it by performing parse-generate-parse round trip.
Reviewed By: dexonsmith
Differential Revision: https://reviews.llvm.org/D96158
Jan Svoboda [Wed, 10 Feb 2021 10:16:10 +0000 (11:16 +0100)]
[clang][cli] Generate and round-trip PreprocessorOutput options
This patch implements generation of remaining preprocessor output options and tests it by performing parse-generate-parse round trip.
Reviewed By: dexonsmith
Differential Revision: https://reviews.llvm.org/D96156
Simon Pilgrim [Wed, 10 Feb 2021 10:26:30 +0000 (10:26 +0000)]
Revert rGe1172959226689a "[X86][AVX] canonicalizeLaneShuffleWithRepeatedOps - merge VPERMILPD ops with different low/high masks."
Revert this while I investigate a downstream breakage report.
Faris Rehman [Wed, 10 Feb 2021 09:24:45 +0000 (09:24 +0000)]
[flang][driver] Add support for -fopenmp and -fopenacc
Add support for the following options:
* -fopenmp
* -fopenacc
Update OpenMP and OpenACC semantics tests to use the new driver if it is built, otherwise use f18.
OpenMP tests that include `use omp_lib` or run `test_symbols.sh` have not been updated as they require options `-intrinsic-module-directory` and `-funparse-with-symbols` which are currently not implemented in the new driver.
Similarly OpenACC tests that run `test_symbols.sh` have not been updated.
This patch also moves semanticsContext to CompilerInvocation and creates it in CompilerInvocation#setSemanticsOpts so that the semantics context can use Fortran::parser::Options#features.
Summary of changes:
- Move semanticsContext to CompilerInvocation.h
- Update OpenMP and OpenACC semantics tests that do not rely on `-intrinsic-module-directory` and `-funparse-with-symbols` to use %flang
Differential Revision: https://reviews.llvm.org/D96032
Sven van Haastregt [Wed, 10 Feb 2021 09:44:52 +0000 (09:44 +0000)]
[OpenCL] Add cl_khr_subgroup_clustered_reduce to TableGen BIFs
Add the builtin functions brought by the
cl_khr_subgroup_clustered_reduce extension to
`-fdeclare-opencl-builtins`.
Sven van Haastregt [Wed, 10 Feb 2021 09:44:39 +0000 (09:44 +0000)]
[OpenCL] Add cl_khr_subgroup_non_uniform_arithmetic to TableGen BIFs
Add the builtin functions brought by the
cl_khr_subgroup_non_uniform_arithmetic extension to
`-fdeclare-opencl-builtins`.
Differential Revision: https://reviews.llvm.org/D95951
Sander de Smalen [Wed, 10 Feb 2021 08:52:10 +0000 (08:52 +0000)]
[LoopVectorize] NFC: Change computeFeasibleMaxVF to operate on ElementCount.
This patch is NFC and changes occurrences of `unsigned MaxVectorSize`
to work on type ElementCount.
This patch is a preparatory patch with the ultimate goal of making
`computeMaxVF()` return both a max fixed VF and a max scalable VF,
so that `selectVectorizationFactor()` can pick the most cost-effective
vectorization factor.
Reviewed By: kmclaughlin
Differential Revision: https://reviews.llvm.org/D96018
Sander de Smalen [Wed, 10 Feb 2021 08:27:03 +0000 (08:27 +0000)]
[ValueTypes] Add MVT for nxv1bf16.
Reviewed By: frasercrmck
Differential Revision: https://reviews.llvm.org/D96249
Sam Parker [Wed, 10 Feb 2021 08:22:45 +0000 (08:22 +0000)]
[WebAssembly] Enable loop unrolling
Enable partial and runtime unrolling with a threshold of 30, which
was derived from a large number of kernels running on node and
wasmtime for amd64 and aarch64.
Unrolling is enabled by default at -O2 and -O3 and is disabled at
-Oz and -Os. Compiling with -Os is recommended if the wasm binary
size is the most important factor.
Differential Revision: https://reviews.llvm.org/D95125
Jessica Paquette [Wed, 10 Feb 2021 02:04:44 +0000 (18:04 -0800)]
[AArch64][GlobalISel] Fold selects fed by G_PTR_ADD
Similar to the case for G_ADD.
There was a function in CTMark/pairlocalalign which was missing this case,
causing GlobalISel to emit a add + csel when a csinc is all that is necessary.
https://godbolt.org/z/ax69E9
Minor code size improvements on CTMark at -Os.
Differential Revision: https://reviews.llvm.org/D96390
Artem Dergachev [Wed, 10 Feb 2021 07:21:20 +0000 (23:21 -0800)]
Revert "[analyzer] RetainCountChecker: Add a suppression for OSSymbols."
This reverts commit
3500cc8d891bb3825bb3275affe6db8b12f2f695.
This old commit was made over a completely false premise. OSSymbols
aren't different from other OSObjects and we shouldn't treat them
differently for the purposes of static analysis.
Tres Popp [Mon, 1 Feb 2021 08:49:54 +0000 (09:49 +0100)]
[mlir][shape] Generalize broadcast to a variadic number of shapes
Previously broadcast was a binary op. Now it can support more inputs.
This has been changed in such a way that for now, this is an NFC for
all broadcast operations that were previously legal.
Differential Revision: https://reviews.llvm.org/D95777
Kazu Hirata [Wed, 10 Feb 2021 06:14:30 +0000 (22:14 -0800)]
[SelectionDAG] Use range-based for loops (NFC)
Kazu Hirata [Wed, 10 Feb 2021 06:14:28 +0000 (22:14 -0800)]
[TableGen] Drop unnecessary const from return types (NFC)
Kazu Hirata [Wed, 10 Feb 2021 06:14:26 +0000 (22:14 -0800)]
[TableGen] Use ListSeparator (NFC)
Michael Kruse [Wed, 10 Feb 2021 05:20:09 +0000 (23:20 -0600)]
[Polly] Make the NewPM pass pipeline more similar to the legacy's.
Even though it has some oddities, both pipelines should be as similar as
possible. Also use report_fatal_error instead of assertions to ensure a
proper failure in release builds for unsupported options.
This finalizes the patch serious to make Polly run in the default
configuration when using the NewPM by default.
Michael Kruse [Wed, 10 Feb 2021 05:08:15 +0000 (23:08 -0600)]
[Polly] Make NewPM's IslAstAnalysis more similar to the legacy IslAstInfoWrapperPass.
In particular, print the ast with -debug-only=polly-ast, print a
per-scop header with print<polly-ast> and force-add the analysis with
-polly-code-generation=ast.
Michael Kruse [Wed, 10 Feb 2021 04:58:54 +0000 (22:58 -0600)]
[Polly] Improve Simplify pass PM integration.
1. LegacyPM: Rename SimplifyLegacyPass to SimplifyWrapperPass.
2. LegacyPM: Complete create/init functions in LinkAllPasses.h
3. NewPM: Only invalidate non-Scop passes if changed.
4. NewPM: Add to default pass pipeline.
5. NewPM: Print -analyze header for each print<polly-simplify>
Michael Kruse [Wed, 10 Feb 2021 04:44:05 +0000 (22:44 -0600)]
[Polly] Port IslScheduleOptimizer to the NewPM.
Michael Kruse [Wed, 10 Feb 2021 04:30:42 +0000 (22:30 -0600)]
[Polly] Add TargetTransformInfo to ScopPass standard analysis results.
TargetTransformInfo is required by IslScheduleOptimizer, as ScopPass.
Unfortunately it is not possible to get arbitrary larger-unit analyses
in for as ScopPass. Loop passes also already use TargetTransformInfo as
LoopStandardAnalysisResults, hence wei might expect it to be available
to Scop passes as well.
Michael Kruse [Wed, 10 Feb 2021 03:57:13 +0000 (21:57 -0600)]
[Polly] Register pass-instrumentation for NewPM's Scop level.
The pass-instrumentation pass is implicitly execute by the NewPM
whenever a new analysis runs. Not registering it will cause the crash
whenever a scop pass requests an analysis.
For instance this is the case for the IstAstAnalysis requesting the
DependenceAnalsis result.
Michael Kruse [Wed, 10 Feb 2021 03:53:14 +0000 (21:53 -0600)]
[Polly] Port PruneUnprofitable to the NewPM.
Michael Kruse [Wed, 10 Feb 2021 03:32:29 +0000 (21:32 -0600)]
[Polly] Port DeLICM to the NewPM.
Michael Kruse [Wed, 10 Feb 2021 03:19:47 +0000 (21:19 -0600)]
[Polly] Port ForwardOpTree to the NewPM.
Ta-Wei Tu [Wed, 10 Feb 2021 05:52:50 +0000 (13:52 +0800)]
Fix deprecated usage of `mallinfo`
glibc deprecates `mallinfo` in the latest version of 2.33. This patch replaces the usage of `mallinfo` with the new `mallinfo2` when it's available.
Reviewed By: lattner
Differential Revision: https://reviews.llvm.org/D96359
Max Kazantsev [Wed, 10 Feb 2021 05:28:08 +0000 (12:28 +0700)]
[Test] Two more tests on usub
They are analogous to the existing tests, but use different starting offset
which can be important for some transforms.
Timm Bäder [Tue, 2 Feb 2021 17:25:36 +0000 (18:25 +0100)]
[clang][driver] Only warn once about invalid library values
Since ToolChain::GetCXXStdlibType() is a simple getter that might emit
the "invalid library name in argument" warning, it can conceivably be
called several times while initializing the build pipeline.
Before this patch, a simple 'clang++ -stdlib=foo ./test.cpp' would print
the warning twice, -rt=lib=foo would print 6 times.
Change this and always only print the warning once. Keep the rest of the
semantics of the functions.
Differential Revision: https://reviews.llvm.org/D95915
Uday Bondhugula [Wed, 10 Feb 2021 05:12:20 +0000 (10:42 +0530)]
[MLIR] NFC Fix vector transforms build warnings
Fix build warnings from VectorTransforms.cpp.
Uday Bondhugula [Tue, 9 Feb 2021 12:36:12 +0000 (18:06 +0530)]
[MLIR] Update affine.for unroll utility for iter_args support
Update affine.for loop unroll utility for iteration arguments support.
Fix promoteIfSingleIteration as well.
Fixes PR49084: https://bugs.llvm.org/show_bug.cgi?id=49084
Differential Revision: https://reviews.llvm.org/D96383
Christopher Di Bella [Thu, 4 Feb 2021 03:19:26 +0000 (03:19 +0000)]
[libcxx] adds concept `std::convertible_to`
Implements parts of:
- P0898R3 Standard Library Concepts
- P1754 Rename concepts to standard_case for C++20, while we still can
Differential Revision: https://reviews.llvm.org/D77961
Yang Fan [Wed, 10 Feb 2021 03:15:39 +0000 (11:15 +0800)]
[clang][cli] Fix gcc "enumeral and non-enumeral type in conditional expression" warning (NFC)
Todd Lipcon [Wed, 10 Feb 2021 01:59:31 +0000 (12:59 +1100)]
Fix JSON formatting when converting to trace event format
Reviewed By: dberris
Differential Revision: https://reviews.llvm.org/D96384
Todd Lipcon [Wed, 10 Feb 2021 01:54:00 +0000 (12:54 +1100)]
Fix xray fdr mode to allow multiple flushes
Reviewed By: dberris
Differential Revision: https://reviews.llvm.org/D96382
Jim Ingham [Wed, 10 Feb 2021 01:48:04 +0000 (17:48 -0800)]
Use internal_dict everywhere we refer to the python session dict in docs.
Andrew Pritchard [Wed, 10 Feb 2021 00:57:40 +0000 (00:57 +0000)]
Add LLVMIR Dialect counterparts of @llvm.maximum and @llvm.minimum.
These are similar to maxnum and minnum, but they're defined to treat -0
as less than +0. This behavior can't be expressed using float
comparisons and selects, since comparisons are defined to treat
different-signed zeros as equal. So, the only way to communicate this
behavior into LLVM IR without defining target-specific intrinsics is to
add the corresponding ops.
Reviewed By: mehdi_amini
Differential Revision: https://reviews.llvm.org/D96373
Andrew Pritchard [Wed, 10 Feb 2021 00:48:08 +0000 (00:48 +0000)]
Fix side-effect detection in LLVMIRIntrinsicGen.
Previously it reported an op had side-effects iff it declared that it
didn't have any side-effects. This had the undesirable result that
canonicalization would always delete any intrinsic calls that did memory
stores and returned void.
Reviewed By: ftynse, mehdi_amini
Differential Revision: https://reviews.llvm.org/D96369
Tyker [Wed, 10 Feb 2021 00:32:00 +0000 (01:32 +0100)]
Revert "[InstCombine] convert assumes to operand bundles"
This reverts commit
5eb2e994f9b3a5aff0a156d0a1f7e6121342cc11.
Michael Kruse [Wed, 10 Feb 2021 00:05:43 +0000 (18:05 -0600)]
[Polly] Remove use of -O3 in regression test.
In addition to that regression tests should not test the intire pass
pipeline (unless they are testing the pipeline itself), the Polly-ACC
currently does not support the new pass manager. If enabled by default,
such tests will therefore fail.
Use the -polly-gpu-runtime and -polly-gpu-arch options also as default
values for the PPCGCodeGeneration pass. This requires to move the option
to be moved from the pipeline-building Register passes to the
PPCGCodeGeneration implementation.
Fixes the spir-typesize.ll buildbot fail.