Commit Graph

395476 Commits

Author SHA1 Message Date
River Riddle 634b086f6e [vscode-mlir] Fix the package repo url.
It should be llvm/vscode-mlir not llvm-project/vscode-mlir.
2021-07-31 01:10:27 +00:00
Petr Hosek dcadd64986 [profile] Fix profile merging with binary IDs
This fixes support for merging profiles which broke as a consequence
of e50a38840d. The issue was missing
adjustment in merge logic to account for the binary IDs which are
now included in the raw profile just after header.

In addition, this change also:
* Includes the version in module signature that's used for merging
to avoid accidental attempts to merge incompatible profiles.
* Moves the binary IDs size field after version field in the header
as was suggested in the review.

Differential Revision: https://reviews.llvm.org/D107143
2021-07-30 17:38:53 -07:00
River Riddle 03b2d1a659 [mlir-vscode] Create a proper output channel for the MLIRContext
This allows for reusing the same output channel when the extension reloads after updating the server. Currently, whenever the extension restarts a new output channel is created (which can lead to a large number of seemingly dead output channels).
2021-07-30 23:53:50 +00:00
River Riddle 3e5906499f [vscode-mlir] Update package.json in preparation for publishing
Quite a few things were out-of-date, or just not
organized well. This revision updates the extension
name, repo, icon, and many other components in
preperation for publishing the extension to the
marketplace.
2021-07-30 23:53:50 +00:00
Siva Chandra Reddy 3d758e68a7 [libc] Move FPExceptMatcher out of TestHelpers.h
TestHelpers.h pulls few pieces from LLVM libc's unittest framework
which aren't available on platforms like Fuchsia which use their own
unittest framework. So, by moving FPExceptMatcher to a different file
we can exclude LLVM libc specific pieces in a cleaner way.

In a later pass, it might make more sense to rename TestHelpers.h also
to FPMatcher.h. That way, we can make macros like EXPECT_FP_EQ to be
equivalent to EXPECT_EQ on platforms like Fuchsia.

Reviewed By: michaelrj

Differential Revision: https://reviews.llvm.org/D107129
2021-07-30 23:44:29 +00:00
Vitaly Buka f08229f49e [sanitizer] Remove cpplint
As code diverge from Google style we need
to add more and more exceptions to suppress
conflicts with clang-format and clang-tidy.
As this point it does not provide a additional value.

Reviewed By: morehouse

Differential Revision: https://reviews.llvm.org/D107197
2021-07-30 15:21:04 -07:00
peter klausler 3338ef93b0 [flang] Produce proper "preprocessor output" for -E option
Rename the current -E option to "-E -Xflang -fno-reformat".

Add a new Parsing::EmitPreprocessedSource() routine to convert the
cooked character stream output of the prescanner back to something
more closely resembling output from a traditional preprocessor;
call this new routine when -E appears.

The new -E output is suitable for use as fixed form Fortran source to
compilation by (one hopes) any Fortran compiler.  If the original
top-level source file had been free form source, the output will be
suitable for use as free form source as well; otherwise there may be
diagnostics about missing spaces if they were indeed absent in the
original fixed form source.

Unless the -P option appears, #line directives are interspersed
with the output (but be advised, f18 will ignore these if presented
with them in a later compilation).

An effort has been made to preserve original alphabetic character case
and source indentation.

Add -P and -fno-reformat to the new drivers.

Tweak test options to avoid confusion with prior -E output; use
-fno-reformat where needed, but prefer to keep -E, sometimes
in concert with -P, on most, updating expected results accordingly.

Differential Revision: https://reviews.llvm.org/D106727
2021-07-30 15:13:56 -07:00
zoecarver 3df649e619 [libcxx][docs] Take locks on the last three views. 2021-07-30 15:08:01 -07:00
Lei Zhang 233a3a75ac [mlir][spirv] Add support for i8 serialization
Reviewed By: hanchung

Differential Revision: https://reviews.llvm.org/D107122
2021-07-30 18:06:11 -04:00
Lei Zhang 0065bd2ad5 [mlir][spirv] Fix loading bool with proper storage capabilities
If the source value to load is bool, and we have native storage
capability support for the source bitwidth, we still cannot directly
rewrite uses; we need to perform casting to bool first.

Reviewed By: hanchung

Differential Revision: https://reviews.llvm.org/D107119
2021-07-30 18:06:11 -04:00
Lei Zhang 9f5300c8be [mlir][spirv] Fix storing bool with proper storage capabilities
If the source value to store is bool, and we have native storage
capability support for the target bitwidth, we still cannot directly
store; we need to perform casting to match the target memref
element's bitwidth.

Reviewed By: hanchung

Differential Revision: https://reviews.llvm.org/D107114
2021-07-30 18:06:10 -04:00
Jon Chesterfield 567c8c7bfd [libomptarget][nfc] Only set cuda-path for nvptx tests
Remove --cuda-path=CUDA_TOOLKIT_ROOT_DIR-NOTFOUND
from the invocation of non-nvptx test cases. Better signal
to noise ratio on other architectures.

Reviewed By: tianshilei1992

Differential Revision: https://reviews.llvm.org/D107074
2021-07-30 23:01:09 +01:00
Shoaib Meenai b8f04a670f [builtins] Try to ensure single copy of emulated TLS state
Multiple copies of emulated TLS state means inconsistent results when
accessing the same thread-local variable from different shared objects
(https://github.com/android/ndk/issues/1551). Making `__emutls_get_address`
be a weak default visibility symbol should make the dynamic linker
ensure only a single copy gets used at runtime. This is best-effort, but
the more robust approach of putting emulated TLS into its own shared
object would (a) be a much bigger change, and (b) shared objects are
pretty heavyweight, and adding a new one to a space-constrained
environment isn't an easy sell. Given the expected rarity of direct
accesses to emulated TLS variables across different shared objects, the
best-effort approach should suffice.

Reviewed By: danalbert, rprichard

Differential Revision: https://reviews.llvm.org/D107127
2021-07-30 14:52:38 -07:00
Petr Hosek 6ea2f31f3d Revert "[profile] Fix profile merging with binary IDs"
This reverts commit 89d6eb6f8c, this
seemed to have break a few builders.
2021-07-30 14:32:52 -07:00
Florian Mayer b5b023638a Revert "[hwasan] Detect use after scope within function."
This reverts commit 84705ed913.
2021-07-30 22:32:04 +01:00
Brendon Cahoon c4c379d633 [LoopStrengthReduction] Fix pointer extend asserts
Additional asserts were added to ScalarEvolution to enforce
pointer/int type rules. An assert is triggered when the LSR pass
attempts to extend a pointer SCEV in GenerateTruncates.

This patch changes GenerateTruncates to exit early if the Formaula
contains a ScaledReg or BaseReg with a pointer type.

Differential Revision: https://reviews.llvm.org/D107185
2021-07-30 17:24:08 -04:00
Eric Leese 8dfd6cae9b [lldb] [DWARF-5] Be lazier about loading .dwo files
This change makes sure that DwarfUnit does not load a .dwo file until
necessary. I also take advantage of DWARF 5's guarantee that the first
support file is also the primary file to make it possible to create
a compile unit without loading the .dwo file.

Review By: jankratochvil, dblaikie

Differential Revision: https://reviews.llvm.org/D100299
2021-07-30 23:17:06 +02:00
Jon Chesterfield 7f97ddaf8a Revert "[OpenMP][AMDGCN] Initial math headers support"
Broke nvptx compilation on files including <complex>

This reverts commit 12da97ea10.
2021-07-30 22:07:00 +01:00
Eli Friedman 3f2828dc28 [polly] Fix up regression test config with current features.
Primarily, configure substitutions so we can copy-paste the "RUN" line
of failed tests without worrying about the paths.
2021-07-30 13:44:48 -07:00
Petr Hosek 89d6eb6f8c [profile] Fix profile merging with binary IDs
This fixes support for merging profiles which broke as a consequence
of e50a38840d. The issue was missing
adjustment in merge logic to account for the binary IDs which are
now included in the raw profile just after header.

In addition, this change also:
* Includes the version in module signature that's used for merging
to avoid accidental attempts to merge incompatible profiles.
* Moves the binary IDs size field after version field in the header
as was suggested in the review.

Differential Revision: https://reviews.llvm.org/D107143
2021-07-30 13:30:30 -07:00
Jay Foad 6e712fdf52 [AMDGPU] Autogenerate checks in kernel-args.ll
Differential Revision: https://reviews.llvm.org/D107052
2021-07-30 21:25:51 +01:00
Aart Bik 160399c7ce [mlir][sparse] move comments from cpp files into dialect doc
Reviewed By: rriddle

Differential Revision: https://reviews.llvm.org/D107191
2021-07-30 13:16:50 -07:00
Jonas Devlieghere 7ad854c41e [lldb] Fix remote macOS debugging on Apple Silicon
Update ARMGetSupportedArchitectureAtIndex to consider remote macOS
debugging. Currently, it defaults to an iOS triple when IsHost() returns
false. This fixes TestPlatformSDK.py on Apple Silicon.

Differential revision: https://reviews.llvm.org/D107179
2021-07-30 13:14:10 -07:00
Fanbo Meng bdf4c7b738 [z/OS]Remove overriding default attribute aligned value
Make DefaultAlignForAttributeAligned consistent with SystemZ.

Reviewed By: abhina.sreeskantharajan, anirudhp

Differential Revision: https://reviews.llvm.org/D107189
2021-07-30 15:51:40 -04:00
Nemanja Ivanovic 9019b55b60 [PowerPC] Fix byte ordering of ld/st with length on BE
The builtins vec_xl_len_r and vec_xst_len_r actually use the
wrong side of the vector on big endian Power9 systems. We never
spotted this before because there was no such thing as a big
endian distro that supported Power9. Now we have AIX and the
elements are in the wrong part of the vector. This just fixes
it so the elements are loaded to and stored from the right
side of the vector.
2021-07-30 14:37:24 -05:00
pooja2299 460d220872 [doc]Added examples for generic opcodes
Added examples to G_BR, G_BRCOND, G_BRJT, G_BRINDIRECT

Differential Revision: https://reviews.llvm.org/D104974
2021-07-31 00:44:32 +05:30
Rahman Lavaee 2256b359d7 Explain the symbols of basic block clusters with an example in the header comments.
This prevents from confusion with the ``labels`` option.

Reviewed By: snehasish

Differential Revision: https://reviews.llvm.org/D107128
2021-07-30 12:08:04 -07:00
Leonard Chan bb438f6cbf [clang][cache] Update Fuchsia-stage2.cmake to create hwasan multilibs
Differential Revision: https://reviews.llvm.org/D99364
2021-07-30 11:53:21 -07:00
Fangrui Song a1532ed275 [InstrProfiling] Make CountersPtr in __profd_ relative
Change `CountersPtr` in `__profd_` to a label difference, which is a link-time
constant. On ELF, when linking a shared object, this requires that `__profc_` is
either private or linkonce/linkonce_odr hidden. On COFF, we need D104564 so that
`.quad a-b` (64-bit label difference) can lower to a 32-bit PC-relative relocation.

```
# ELF: R_X86_64_PC64 (PC-relative)
.quad .L__profc_foo-.L__profd_foo

# Mach-O: a pair of 8-byte X86_64_RELOC_UNSIGNED and X86_64_RELOC_SUBTRACTOR
.quad l___profc_foo-l___profd_foo

# COFF: we actually use IMAGE_REL_AMD64_REL32/IMAGE_REL_ARM64_REL32 so
# the high 32-bit value is zero even if .L__profc_foo < .L__profd_foo
# As compensation, we truncate CountersDelta in the header so that
# __llvm_profile_merge_from_buffer and llvm-profdata reader keep working.
.quad .L__profc_foo-.L__profd_foo
```

(Note: link.exe sorts `.lprfc` before `.lprfd` even if the object writer
has `.lprfd` before `.lprfc`, so we cannot work around by reordering
`.lprfc` and `.lprfd`.)

With this change, a stage 2 (`-DLLVM_TARGETS_TO_BUILD=X86 -DLLVM_BUILD_INSTRUMENTED=IR`)
`ld -pie` linked clang is 1.74% smaller due to fewer R_X86_64_RELATIVE relocations.
```
% readelf -r pie | awk '$3~/R.*/{s[$3]++} END {for (k in s) print k, s[k]}'
R_X86_64_JUMP_SLO 331
R_X86_64_TPOFF64 2
R_X86_64_RELATIVE 476059  # was: 607712
R_X86_64_64 2616
R_X86_64_GLOB_DAT 31
```

The absolute function address (used by llvm-profdata to collect indirect call
targets) can be converted to relative as well, but is not done in this patch.

Differential Revision: https://reviews.llvm.org/D104556
2021-07-30 11:52:18 -07:00
David Green 69cdadddec [ARM] Distribute reductions based on ascending load offset
This distributes reductions based on the relative offset of loads, if
one is found from their operands. Given chains of reductions this will
then sort them in ascending load order, which in turn can help simple
prefetches latch on to increasing strides more easily.

Differential Revision: https://reviews.llvm.org/D106569
2021-07-30 19:50:07 +01:00
Anton Zabaznov 4e124ff256 [OpenCL] Replace test for pipe struct to test it with fixed triple
Reviewed By: Anastasia

Differential Revision: https://reviews.llvm.org/D107176
2021-07-30 21:49:20 +03:00
Sanjay Patel e5d9831cf2 [InstCombine] add more tests for vector cmp-bitcast; NFC
D107170
2021-07-30 14:48:11 -04:00
Sanjay Patel c6d147b0a5 [InstCombine] add more tests for vector-cmp bitcast; NFC 2021-07-30 14:48:11 -04:00
Leonard Chan 606ca48d79 [compiler-rt][Fuchsia] Support HWASan on Fuchsia
This allows for hwasan to be built targetting fuchsia.

Differential Revision: https://reviews.llvm.org/D103543
2021-07-30 11:43:03 -07:00
Leonard Chan b4c00b3882 [compiler-rt][hwasan] Remove __sanitizer allocation functions from hwasan interface
These functions should not be externally used. We also do not need them internally for Fuchsia.

Differential Revision: https://reviews.llvm.org/D99381
2021-07-30 11:37:19 -07:00
Mark de Wever d618a1cc5c [libc++] Improve LIBCXX_ENABLE_INCOMPLETE_FEATURES.
@tcanens pointed out the current behavior of the macro breaks the usage
pattern described in http://wg21.link/SD6
```
#  if __has_include(<optional>)
#    include <optional>
#    if __cpp_lib_optional >= 201606
#      define have_optional 1
#    endif
```

To support this usage pattern the hard errror is removed. Instead the
header includes nothing but the `<version>` header.

Reviewed By: ldionne, #libc

Differential Revision: https://reviews.llvm.org/D107134
2021-07-30 14:36:03 -04:00
Simon Pilgrim afc6b09dee [InstCombine] getMaskedTypeForICmpPair - remove dead code. NFCI.
Ok should be true at this point, so the early-out is dead - replace with an assert.
2021-07-30 19:23:05 +01:00
Simon Pilgrim 3c0b596ecc SelectionDAGDumper.cpp - remove nested if-else return chain. NFCI.
Match style and don't use an else after a return.
2021-07-30 19:23:05 +01:00
Simon Pilgrim 986841cca2 SelectionDAGDumper.cpp - printrWithDepthHelper - remove dead code. NFCI.
Fixes coverity warning - we have an early-out for unsigned depth == 0, so the depth < 1 early-out later on is dead code.
2021-07-30 19:23:04 +01:00
Aart Bik 1d77bb9e1b [mlir][sparse] template the memory resident coordinate scheme storage
Rationale:
External file formats always store the values as doubles, so this was
hard coded in the memory resident COO scheme used to pass data into the
final sparse storage scheme during setup. However, with alternative methods
on the horizon of setting up these temporary COO schemes, it is time to
properly template this data structure.

Reviewed By: gussmith23

Differential Revision: https://reviews.llvm.org/D107001
2021-07-30 11:21:05 -07:00
Jonas Devlieghere fe23f0c73a [lldb] Always codesign binaries on macOS
Always codesign binaries on macOS. Apple Silicon has stricter
codesigning requirements, for example requiring macCatalyst binaries to
be signed. Ad-hoc sign everything like we do for other Darwin platforms.
2021-07-30 11:19:40 -07:00
Jose M Monsalve Diaz 5424ceeda0 [OpenMP] Fixing llvm-omp-device-info compilation with runtimes
When using `-DLLVM_ENABLED_RUNTIMES` instead of `-DLLVM_ENABLED_PROJECTS`
the `llvm-omp-device-info` tool is not compiled or installed.
In general, no llvm tool would be build on runtimes, because the
-DLLVM_BUILD_TOOLS flag is removed by the way runtimes compilation calls
cmake again.

This patch is simple. Just forward the value of this flag to the
runtime cmake command.

I'm also removing an unnecessary comment in the compilation of the tool

Differential Revision: https://reviews.llvm.org/D107177
2021-07-30 13:09:08 -05:00
LLVM GN Syncbot c112483fab [gn build] Port 481ad59b9f 2021-07-30 17:57:30 +00:00
Walter Erquinigo c9308cc219 [nfc] improve a simple call
@jingham correctly pointed out that this call can be simplified. So
let's better do it.
2021-07-30 10:55:48 -07:00
zoecarver 481ad59b9f [libcxx][ranges] Add `std::ranges::single_view`.
Differential Revision: https://reviews.llvm.org/D106840
2021-07-30 10:53:20 -07:00
Paul Robinson 31d408f9f9 Trim failing test
Don't try to run the non-integrated assembler; just verify that the
invocations look like what we expect.  Do verify that the integrated
assembler handles warnings as expected.
2021-07-30 10:51:08 -07:00
Siva Chandra Reddy 381f9affab [libc][NFC] Add dummy errno target to satisfy mixed mode builds.
In mixed mode builds, we should not be including errno as part of
LLVM libc - errno from another library (or the system library) should be
used. But, other entrypoints which use errno list LLVM libc's errno as a
dep ta satisfy the full build mode. So, we add a dummy errno
implementation with empty files to make both mixed mode and full build
mode happy.
2021-07-30 17:42:48 +00:00
Matt Arsenault e46badd4e9 GlobalISel: Have lowerLoad scalarize unaligned vectors
This could be smarter by picking an ideal type, or at least splitting
the vector in half first. Also handles lower for non-power-of-2,
non-extending vector loads.

Currently this just avoids failing to legalize some odd vector AMDGPU
tests, but is a step towards removing the split logic from the
NarrowScalar logic.
2021-07-30 13:23:29 -04:00
Alexey Bataev 95e5d401ae [SLP]Improve splats vectorization.
Replace insertelement instructions for splats with just single
insertelement + broadcast shuffle. Also, try to merge these instructions
if they come from the same/shuffled gather node.

Differential Revision: https://reviews.llvm.org/D107104
2021-07-30 10:17:45 -07:00
Kerry McLaughlin 9d35594993 Reland "[LV] Use lookThroughAnd with logical reductions"
If a reduction Phi has a single user which `AND`s the Phi with a type mask,
`lookThroughAnd` will return the user of the Phi and the narrower type represented
by the mask. Currently this is only used for arithmetic reductions, whereas loops
containing logical reductions will create a reduction intrinsic using the widened
type, for example:

  for.body:
    %phi = phi i32 [ %and, %for.body ], [ 255, %entry ]
    %mask = and i32 %phi, 255
    %gep = getelementptr inbounds i8, i8* %ptr, i32 %iv
    %load = load i8, i8* %gep
    %ext = zext i8 %load to i32
    %and = and i32 %mask, %ext
    ...

^ this will generate an and reduction intrinsic such as the following:
    call i32 @llvm.vector.reduce.and.v8i32(<8 x i32>...)

The same example for an add instruction would create an intrinsic of type i8:
    call i8 @llvm.vector.reduce.add.v8i8(<8 x i8>...)

This patch changes AddReductionVar to call lookThroughAnd for other integer
reductions, allowing loops similar to the example above with reductions such
as and, or & xor to vectorize.

Reviewed By: david-arm, dmgreen

Differential Revision: https://reviews.llvm.org/D105632
2021-07-30 18:04:09 +01:00