platform/upstream/llvm.git
5 years ago[llvm-exegesis] Teach llvm-exegesis to handle instructions with multiple tied variables.
Clement Courbet [Tue, 26 Feb 2019 10:54:45 +0000 (10:54 +0000)]
[llvm-exegesis] Teach llvm-exegesis to handle instructions with multiple tied variables.

Reviewers: gchatelet

Subscribers: tschuett, llvm-commits

Tags: #llvm

Differential Revision: https://reviews.llvm.org/D58285

llvm-svn: 354862

5 years agoCodeGen: Explicitly initialize structure padding in the -ftrivial-auto-var-init mode
Alexander Potapenko [Tue, 26 Feb 2019 10:46:21 +0000 (10:46 +0000)]
CodeGen: Explicitly initialize structure padding in the -ftrivial-auto-var-init mode

When generating initializers for local structures in the
-ftrivial-auto-var-init mode, explicitly wipe the padding bytes with
either 0x00 or 0xAA.

This will allow us to automatically handle the padding when splitting
the initialization stores (see https://reviews.llvm.org/D57898).

Reviewed at https://reviews.llvm.org/D58188

llvm-svn: 354861

5 years ago[llvm-objcopy] Add --set-start, --change-start and --adjust-start
Eugene Leviant [Tue, 26 Feb 2019 09:24:22 +0000 (09:24 +0000)]
[llvm-objcopy] Add --set-start, --change-start and --adjust-start

Differential revision: https://reviews.llvm.org/D58173

llvm-svn: 354854

5 years ago[compiler-rt] disable asan bcmp tests on android.
Clement Courbet [Tue, 26 Feb 2019 09:16:51 +0000 (09:16 +0000)]
[compiler-rt] disable asan bcmp tests on android.

Android does not have bcmp.

sanitizer-x86_64-linux-android: run instrumented asan tests [arm/aosp_marlin-userdebug/PI] -  stdio

llvm-svn: 354853

5 years ago[compiler-rt] Fix test broken by r354851.
Clement Courbet [Tue, 26 Feb 2019 08:17:49 +0000 (08:17 +0000)]
[compiler-rt] Fix test broken by r354851.

error: CHECK: expected string not found in input
// CHECK: Uninitialized bytes in __interceptor_memcmp at offset 3

llvm-svn: 354852

5 years ago[compiler-rt] Intercept the bcmp() function.
Clement Courbet [Tue, 26 Feb 2019 07:43:01 +0000 (07:43 +0000)]
[compiler-rt] Intercept the bcmp() function.

Summary:
I have not introduced a separate hook for `bcmp()` as I don't think there
should be any reason for a sanitizer to treat it differently from `memcmp()`.

This is only enabled when building on POSIX with GNU extensions.

Context: this is to avoid losing coverage when emitting `bcmp() == 0` instead
of `memcmp() == 0` in llvm, see https://reviews.llvm.org/D56593.

Reviewers: mgorny, krytarowski, vitalybuka, dvyukov

Subscribers: kubamracek, dberris, delcypher, jdoerfert, #sanitizers, llvm-commits, jyknight

Tags: #llvm, #sanitizers

Differential Revision: https://reviews.llvm.org/D58379

llvm-svn: 354851

5 years ago[ThinLTO] Use defined node and edge order when dumping DOT file
Eugene Leviant [Tue, 26 Feb 2019 07:38:21 +0000 (07:38 +0000)]
[ThinLTO] Use defined node and edge order when dumping DOT file

Differential revision: https://reviews.llvm.org/D58631

llvm-svn: 354850

5 years agoRevert "Improve "llvm-nm -f sysv" output for Elf files"
Vlad Tsyrklevich [Tue, 26 Feb 2019 07:04:56 +0000 (07:04 +0000)]
Revert "Improve "llvm-nm -f sysv" output for Elf files"

This reverts commit r354833, it was causing ASan test failures on
sanitizer-x86_64-linux-fast.

llvm-svn: 354849

5 years ago[libc++] Rename _NOALIAS macro to _LIBCPP_NOALIAS
Louis Dionne [Tue, 26 Feb 2019 06:34:42 +0000 (06:34 +0000)]
[libc++] Rename _NOALIAS macro to _LIBCPP_NOALIAS

Summary:
For consistency, libc++ macros always start with _LIBCPP. This should
have no functionality change.

Reviewers: EricWF, mclow.lists

Subscribers: christof, jkorous, dexonsmith, libcxx-commits

Differential Revision: https://reviews.llvm.org/D58558

llvm-svn: 354848

5 years ago[NFC] Add to contributor list.
Chen Zheng [Tue, 26 Feb 2019 05:46:45 +0000 (05:46 +0000)]
[NFC] Add to contributor list.

llvm-svn: 354847

5 years ago[WebAssembly] Properly align fp128 arguments in outgoing varargs arguments
Dan Gohman [Tue, 26 Feb 2019 05:20:19 +0000 (05:20 +0000)]
[WebAssembly] Properly align fp128 arguments in outgoing varargs arguments

For outgoing varargs arguments, it's necessary to check the OrigAlign field
of the corresponding OutputArg entry to determine argument alignment, rather
than just computing an alignment from the argument value type. This is
because types like fp128 are split into multiple argument values, with
narrower types that don't reflect the ABI alignment of the full fp128.

This fixes the printf("printfL: %4.*Lf\n", 2, lval); testcase.

Differential Revision: https://reviews.llvm.org/D58656

llvm-svn: 354846

5 years ago[ARM] Be super conservative about atomics
Philip Reames [Tue, 26 Feb 2019 04:30:33 +0000 (04:30 +0000)]
[ARM] Be super conservative about atomics

As requested during review of D57601 <https://reviews.llvm.org/D57601> https://reviews.llvm.org/D57601, be equally conservative for atomic MMOs as for volatile MMOs in all in tree backends. At the moment, all atomic MMOs are also volatile, but I'm about to change that.

Differential Revision: https://reviews.llvm.org/D58490

Note: D58498 landed in several pieces as individual backends were approved.  This is the last chunk.
llvm-svn: 354845

5 years ago[WebAssembly] Fix a bug deleting instruction in a ranged for loop
Heejin Ahn [Tue, 26 Feb 2019 04:08:49 +0000 (04:08 +0000)]
[WebAssembly] Fix a bug deleting instruction in a ranged for loop

Summary: We shouldn't delete elements while iterating a ranged for loop.

Reviewers: dschuff

Subscribers: sbc100, jgravelle-google, sunfish, llvm-commits

Tags: #llvm

Differential Revision: https://reviews.llvm.org/D58519

llvm-svn: 354844

5 years ago[CGDebugInfo] Set NonTrivial DIFlag to a c++ record if it's not trivial
Aaron Smith [Tue, 26 Feb 2019 03:49:05 +0000 (03:49 +0000)]
[CGDebugInfo] Set NonTrivial DIFlag to a c++ record if it's not trivial

This goes with https://reviews.llvm.org/D44406

llvm-svn: 354843

5 years ago[WebAssembly] Improve readability of EH tests
Heejin Ahn [Tue, 26 Feb 2019 03:29:59 +0000 (03:29 +0000)]
[WebAssembly] Improve readability of EH tests

Summary:
- Indent check lines to easily figure out try-catch-end structure
- Add the original C++ code the tests were genereated from
- Add a few more lines to make the structure more readable
- Rename a couple function / structures
- Add label and branch annotations to cfg-stackify-eh.ll
- Temporarily delete check lines for `test1` in `cfg-stackify-eh.ll`
  because it will be updated in a later CL soon and there's no point of
  making it look better here

Reviewers: dschuff

Subscribers: sunfish, sbc100, jgravelle-google, llvm-commits

Tags: #llvm

Differential Revision: https://reviews.llvm.org/D58562

llvm-svn: 354842

5 years ago[CodeView] Emit HasConstructorOrDestructor class option for non-trivial constructors
Aaron Smith [Tue, 26 Feb 2019 03:23:56 +0000 (03:23 +0000)]
[CodeView] Emit HasConstructorOrDestructor class option for non-trivial constructors

Reviewers: zturner, rnk, llvm-commits, aleksandr.urakov

Reviewed By: zturner, rnk

Subscribers: jdoerfert, majnemer, asmith

Tags: #llvm

Differential Revision: https://reviews.llvm.org/D44406

llvm-svn: 354841

5 years ago[llvm-cov] Fix llvm-cov on Windows and un-XFAIL test
Reid Kleckner [Tue, 26 Feb 2019 02:30:00 +0000 (02:30 +0000)]
[llvm-cov] Fix llvm-cov on Windows and un-XFAIL test

Summary:
The llvm-cov tool needs to be able to find coverage names in the
executable, so the .lprfn and .lcovmap sections cannot be merged into
.rdata.

Also, the linker merges .lprfn$M into .lprfn, so llvm-cov needs to
handle that when looking up sections. It has to support running on both
relocatable object files and linked PE files.

Lastly, when loading .lprfn from a PE file, llvm-cov needs to skip the
leading zero byte added by the profile runtime.

Reviewers: vsk

Subscribers: hiraditya, #sanitizers, llvm-commits

Tags: #sanitizers, #llvm

Differential Revision: https://reviews.llvm.org/D58661

llvm-svn: 354840

5 years agoRevert r354832 "[ASTImporter] Add support for importing ChooseExpr AST nodes."
Reid Kleckner [Tue, 26 Feb 2019 02:22:22 +0000 (02:22 +0000)]
Revert r354832 "[ASTImporter] Add support for importing ChooseExpr AST nodes."

Test does not pass on Windows

llvm-svn: 354839

5 years ago[MS] Fix for Bug 8446, template instantiation without a 'typename' keyword
Reid Kleckner [Tue, 26 Feb 2019 02:22:17 +0000 (02:22 +0000)]
[MS] Fix for Bug 8446, template instantiation without a 'typename' keyword

Patch by Zahira Ammarguellat!

Differential Revision: https://reviews.llvm.org/D41950

llvm-svn: 354838

5 years ago[X86] Fix bug in x86_intrcc with arg copy elision
Reid Kleckner [Tue, 26 Feb 2019 02:11:25 +0000 (02:11 +0000)]
[X86] Fix bug in x86_intrcc with arg copy elision

Summary:
Use a custom calling convention handler for interrupts instead of fixing
up the locations in LowerMemArgument. This way, the offsets are correct
when constructed and we don't need to account for them in as many
places.

Depends on D56883

Replaces D56275

Reviewers: craig.topper, phil-opp

Subscribers: hiraditya, llvm-commits

Differential Revision: https://reviews.llvm.org/D56944

llvm-svn: 354837

5 years ago[winasan] Unpoison stack memory when threads exit (redux)
David Major [Tue, 26 Feb 2019 01:35:48 +0000 (01:35 +0000)]
[winasan] Unpoison stack memory when threads exit (redux)

This is a second attempt at r342652 using a TLS callback instead of an
interceptor.

In long-running builds we've seen some ASan complaints during thread creation
that we suspect are due to leftover poisoning from previous threads whose
stacks occupied that memory. This patch adds a callback that unpoisons the
stack memory when a thread exits.

Differential Revision: https://reviews.llvm.org/D58641

llvm-svn: 354836

5 years ago[sanitizer] Re-disable a few tests on android.
Evgeniy Stepanov [Tue, 26 Feb 2019 00:22:22 +0000 (00:22 +0000)]
[sanitizer] Re-disable a few tests on android.

Tests were accidentally enabled r354829.

llvm-svn: 354834

5 years agoImprove "llvm-nm -f sysv" output for Elf files
Sunil Srivastava [Tue, 26 Feb 2019 00:19:39 +0000 (00:19 +0000)]
Improve "llvm-nm -f sysv" output for Elf files

Specifically, compute and Print Type and Section columns.

Differential Revision: https://reviews.llvm.org/D58263

llvm-svn: 354833

5 years ago[ASTImporter] Add support for importing ChooseExpr AST nodes.
Tom Roeder [Mon, 25 Feb 2019 23:24:58 +0000 (23:24 +0000)]
[ASTImporter] Add support for importing ChooseExpr AST nodes.

Summary:
This allows ASTs to be merged when they contain ChooseExpr (the GNU
__builtin_choose_expr construction). This is needed, for example, for
cross-CTU analysis of C code that makes use of __builtin_choose_expr.

The node is already supported in the AST, but it didn't have a matcher
in ASTMatchers. So, this change adds the matcher and adds support to
ASTImporter.

Reviewers: shafik, a_sidorin, martong, aaron.ballman

Subscribers: aaron.ballman, rnkovacs, jdoerfert, cfe-commits

Tags: #clang

Differential Revision: https://reviews.llvm.org/D58292

llvm-svn: 354832

5 years ago[NFC] Reorder some mis-ordered tests
JF Bastien [Mon, 25 Feb 2019 23:09:34 +0000 (23:09 +0000)]
[NFC] Reorder some mis-ordered tests

I somehow had misaligned some of the tests when I originally wrote this. Re-order them properly.

llvm-svn: 354831

5 years ago[AMDGPU] Added target to mir test. NFC.
Stanislav Mekhanoshin [Mon, 25 Feb 2019 22:59:55 +0000 (22:59 +0000)]
[AMDGPU] Added target to mir test. NFC.

Test was used without -mcpu, although tested instructions
not available on all ASICs.

llvm-svn: 354830

5 years ago[sanitizer] Remove "-android" from test_arch.
Evgeniy Stepanov [Mon, 25 Feb 2019 22:32:30 +0000 (22:32 +0000)]
[sanitizer] Remove "-android" from test_arch.

Summary:
ASan and Scudo tests are adding "-android" to test arch.
There are no tests that depend on it as far as I can see.
If necessary, do this instead:
  REQUIRES: aarch64-target-arch && android

Reviewers: pcc, vitalybuka

Subscribers: srhines, kubamracek, mgorny, javed.absar, kristof.beyls, cryptoad, #sanitizers, llvm-commits

Tags: #sanitizers, #llvm

Differential Revision: https://reviews.llvm.org/D58532

llvm-svn: 354829

5 years agoRegBankSelect: Handle slightly more complex value mappings
Matt Arsenault [Mon, 25 Feb 2019 22:24:13 +0000 (22:24 +0000)]
RegBankSelect: Handle slightly more complex value mappings

Try to use concat_vectors. Also remove unnecessary assert on
pointers. Fixes asserting for <4 x s16> operations and 64-bit pointers
for AMDGPU.

llvm-svn: 354828

5 years agoReapply "Make static counters in ASTContext non-static." with fixes.
Alexander Kornienko [Mon, 25 Feb 2019 22:22:09 +0000 (22:22 +0000)]
Reapply "Make static counters in ASTContext non-static." with fixes.

This reverts commit e50038e4dc53caee1acc811362ac0b15e00ef5eb.

llvm-svn: 354827

5 years ago[CodeGenObjC] Fix a nullptr dyn_cast
Erik Pilkington [Mon, 25 Feb 2019 21:35:14 +0000 (21:35 +0000)]
[CodeGenObjC] Fix a nullptr dyn_cast

ObjCMessageExpr::getInstanceReceiver returns nullptr if the receiver
is 'super'. Make this check more strict, since we don't care about
messages to super here.

rdar://48247290

llvm-svn: 354826

5 years agoAMDGPU/GlobalISel: Fix bit ops for non-power-of-2 sizes
Matt Arsenault [Mon, 25 Feb 2019 21:32:48 +0000 (21:32 +0000)]
AMDGPU/GlobalISel: Fix bit ops for non-power-of-2 sizes

llvm-svn: 354825

5 years ago[libclang] Expose warn_unused and warn_unused_result attributes.
Emilio Cobos Alvarez [Mon, 25 Feb 2019 21:24:52 +0000 (21:24 +0000)]
[libclang] Expose warn_unused and warn_unused_result attributes.

This is helpful to properly detect them, and fixing issues like
https://github.com/rust-lang/rust-bindgen/issues/1518.

Differential Revision: https://reviews.llvm.org/D58570

llvm-svn: 354824

5 years ago[libclang] Fix a trivial error introduced in D57946.
Emilio Cobos Alvarez [Mon, 25 Feb 2019 21:15:34 +0000 (21:15 +0000)]
[libclang] Fix a trivial error introduced in D57946.

The value for CXCursor_ConvergentAttr is not 420. I'm not really sure how easy
it is to test this, and I'm not familiar with the python bindings, just noticed
the error while looking at D57946 to write D58570.

Differential Revision: https://reviews.llvm.org/D58571

llvm-svn: 354823

5 years agoRevert "[Support] Make raw_string_ostream unbuffered"
Roman Lebedev [Mon, 25 Feb 2019 21:11:19 +0000 (21:11 +0000)]
Revert "[Support] Make raw_string_ostream unbuffered"

Shame on me, did not run all the tests, bots are angry.

This reverts commit r354819.

llvm-svn: 354822

5 years ago[LangRef] *.overflow intrinsics now support vectors
Simon Pilgrim [Mon, 25 Feb 2019 21:05:09 +0000 (21:05 +0000)]
[LangRef] *.overflow intrinsics now support vectors

We have all the necessary legalization, expansion and unrolling support required for the *.overflow intrinsics with vector types, so update the docs to make that clear.

Note: vectorization is not in place yet (the non-homogenous return types aren't well supported) so we still must explicitly use the vectors intrinsics and not reply on slp/loop.

Differential Revision: https://reviews.llvm.org/D58618

llvm-svn: 354821

5 years ago[Support] Make raw_string_ostream unbuffered
Roman Lebedev [Mon, 25 Feb 2019 20:51:49 +0000 (20:51 +0000)]
[Support] Make raw_string_ostream unbuffered

Summary:
In D58580 i have noted that `llvm::to_string()` is a memory hog.
It uses `raw_string_ostream`, and since it was buffered,
every `raw_string_ostream` had a cost of `BUFSIZ` bytes
(which is `8192` at least here). So every `llvm::to_string()`
call, even to just print an `int`, costed `8192` bytes.

In D58580, getting rid of that buffering //had// significant
performance and memory consumption improvements for `llvm-xray convert`.

Similarly, in D58580 @rnk pointed out that the `raw_svector_ostream`
is already unbuffered, and `write_unsigned_impl` and friends
do internal buffering. So it should be ok performance-wise to just
make the `raw_string_ostream` itself unbuffered.

Here, i don't have any perf measurements.
Another letdown is that i'm leaving a loose-end - not deleting the
`flush()` method. I  don't expect that cleanup to be anything more
than just fixing every new compiler error, but i'm presently unable
to do that. Will look into that later.

Reviewers: rnk, zturner

Reviewed By: rnk

Subscribers: kristina, jdoerfert, llvm-commits, rnk

Tags: #llvm

Differential Revision: https://reviews.llvm.org/D58643

llvm-svn: 354819

5 years agoAMDGPU/GlobalISel: Clamp max implicit_def elements
Matt Arsenault [Mon, 25 Feb 2019 20:46:06 +0000 (20:46 +0000)]
AMDGPU/GlobalISel: Clamp max implicit_def elements

llvm-svn: 354818

5 years ago[OpenMP 5.0] Parsing/sema support for from clause with mapper modifier.
Michael Kruse [Mon, 25 Feb 2019 20:34:15 +0000 (20:34 +0000)]
[OpenMP 5.0] Parsing/sema support for from clause with mapper modifier.

This patch implements the parsing and sema support for the OpenMP
'from'-clause with potential user-defined mappers attached.
User-defined mappers are a new feature in OpenMP 5.0. A 'from'-clause
can have an explicit or implicit associated mapper, which instructs the
compiler to generate and use customized mapping functions. An example is
shown below:

    struct S { int len; int *d; };
    #pragma omp declare mapper(id: struct S s) map(s, s.d[0:s.len])
    struct S ss;
    #pragma omp target update from(mapper(id): ss) // use the mapper with name 'id' to map ss from device

Contributed-by: Lingda Li <lildmh@gmail.com>
Differential Revision: https://reviews.llvm.org/D58638

llvm-svn: 354817

5 years agoRegisterScavenger: Allow fail without spill
Matt Arsenault [Mon, 25 Feb 2019 20:29:04 +0000 (20:29 +0000)]
RegisterScavenger: Allow fail without spill

AMDGPU wants to use this in some contexts where
the spilling is either impossible, or a worse alternative
to doing something else.

llvm-svn: 354816

5 years agoAMDGPU: Remove IntrReadMem from memtime/memrealtime intrinsics
Matt Arsenault [Mon, 25 Feb 2019 20:16:11 +0000 (20:16 +0000)]
AMDGPU: Remove IntrReadMem from memtime/memrealtime intrinsics

EarlyCSE with MemorySSA was able to use this to merge multiple calls
with no intervening store.

llvm-svn: 354814

5 years agoGlobalISel: Make legalizer/regbankselect clear NoPHIs property
Matt Arsenault [Mon, 25 Feb 2019 20:00:25 +0000 (20:00 +0000)]
GlobalISel: Make legalizer/regbankselect clear NoPHIs property

If no phi existed in the original MIR and these introduced one, the
verifier would fail.

llvm-svn: 354813

5 years agoRevert "Make static counters in ASTContext non-static."
Vlad Tsyrklevich [Mon, 25 Feb 2019 19:53:13 +0000 (19:53 +0000)]
Revert "Make static counters in ASTContext non-static."

This reverts commit r354795, I suspect it is causing test failures
on MSan sanitizer bots.

llvm-svn: 354812

5 years ago[X86] Improve detection of unneeded shift amount masking to also handle the case...
Craig Topper [Mon, 25 Feb 2019 19:42:47 +0000 (19:42 +0000)]
[X86] Improve detection of unneeded shift amount masking to also handle the case that the LHS has known zeroes in it

If the LHS has known zeros, the RHS immediate will have had bits removed. So call computeKnownBits to get the known zeroes so we can handle this case.

Differential Revision: https://reviews.llvm.org/D58475

llvm-svn: 354811

5 years agoFix a sign compare warning breaking the -Werror build.
Andrea Di Biagio [Mon, 25 Feb 2019 19:33:58 +0000 (19:33 +0000)]
Fix a sign compare warning breaking the -Werror build.

The warning was introduced at r354793.

llvm-svn: 354810

5 years agoAMDGPU: Correct definitions for bitset instructions
Matt Arsenault [Mon, 25 Feb 2019 19:24:46 +0000 (19:24 +0000)]
AMDGPU: Correct definitions for bitset instructions

These really read and write the result register, so these need a tied
input.

llvm-svn: 354809

5 years ago[Mips] Fix missing masking in fast-isel of br (PR40325)
Nikita Popov [Mon, 25 Feb 2019 18:54:17 +0000 (18:54 +0000)]
[Mips] Fix missing masking in fast-isel of br (PR40325)

Fixes https://bugs.llvm.org/show_bug.cgi?id=40325 by zero extending
(and x, 1) the condition before branching on it.

To avoid regressing trivial cases, I'm combining emission of cmp+br
sequences for the single-use + same block case (similar to what we
do in x86). icmpbr1.ll still regresses due to the cross-bb usage
of the condition.

Differential Revision: https://reviews.llvm.org/D58576

llvm-svn: 354808

5 years ago[AArch64][GlobalISel] Refactor selectBuildVector to use MachineIRBuilder. NFC.
Amara Emerson [Mon, 25 Feb 2019 18:52:54 +0000 (18:52 +0000)]
[AArch64][GlobalISel] Refactor selectBuildVector to use MachineIRBuilder. NFC.

This is a preparatory change as I want to use emitScalarToVector() elsewhere,
and in general we want to transition to MIRBuilder instead of using BuildMI
directly.

Differential Revision: https://reviews.llvm.org/D58528

llvm-svn: 354807

5 years ago[analyzer] Fix infinite recursion in printing macros
Kristof Umann [Mon, 25 Feb 2019 18:49:42 +0000 (18:49 +0000)]
[analyzer] Fix infinite recursion in printing macros

#define f(y) x
#define x f(x)
int main() { x; }

This example results a compilation error since "x" in the first line was not
defined earlier. However, the macro expression printer goes to an infinite
recursion on this example.

Patch by Tibor Brunner!

Differential Revision: https://reviews.llvm.org/D57892

llvm-svn: 354806

5 years agoLWG3101 - span's Container constructors need another constraint. Reviewed as https...
Marshall Clow [Mon, 25 Feb 2019 18:32:57 +0000 (18:32 +0000)]
LWG3101 - span's Container constructors need another constraint. Reviewed as https://reviews.llvm.org/D57058.

llvm-svn: 354805

5 years ago[lldb-mi] Return source line number in proper format
Tatyana Krasnukha [Mon, 25 Feb 2019 18:32:46 +0000 (18:32 +0000)]
[lldb-mi] Return source line number in proper format

Line number is a decimal number and is printed as such, however for some
reason it was prefixed with '0x', thus turning printed value invalid.

Patch by Anton Kolesov <Anton.Kolesov@synopsys.com>

llvm-svn: 354804

5 years ago[lldb-mi] Fix conversion warning for 64-bit build
Tatyana Krasnukha [Mon, 25 Feb 2019 18:23:44 +0000 (18:23 +0000)]
[lldb-mi] Fix conversion warning for 64-bit build

llvm-svn: 354803

5 years agoCommit LWG3144 - span does not have a const_pointer typedef. Reviewed as D57039.
Marshall Clow [Mon, 25 Feb 2019 17:58:03 +0000 (17:58 +0000)]
Commit LWG3144 - span does not have a const_pointer typedef. Reviewed as D57039.

llvm-svn: 354802

5 years agoFirst part of P1024: Usability Enhancements for std::span. Remove operator() for...
Marshall Clow [Mon, 25 Feb 2019 17:54:08 +0000 (17:54 +0000)]
First part of P1024: Usability Enhancements for std::span. Remove operator() for indexing, and add 'front' and 'back' calls.

llvm-svn: 354801

5 years ago[Lanai] Be super conservative about atomics
Philip Reames [Mon, 25 Feb 2019 17:36:10 +0000 (17:36 +0000)]
[Lanai] Be super conservative about atomics

As requested during review of D57601 <https://reviews.llvm.org/D57601>, be equally conservative for atomic MMOs as for volatile MMOs in all in tree backends. At the moment, all atomic MMOs are also volatile, but I'm about to change that.

Reviewed as part of https://reviews.llvm.org/D58490, with other backends still pending review.

llvm-svn: 354800

5 years ago[lldb-mi] Check raw pointers before passing them to std::string ctor/assignment
Tatyana Krasnukha [Mon, 25 Feb 2019 16:40:11 +0000 (16:40 +0000)]
[lldb-mi] Check raw pointers before passing them to std::string ctor/assignment

Differential Revision: https://reviews.llvm.org/D55653

llvm-svn: 354798

5 years ago[SelectionDAG] Add demanded elts variants to isConstOrConstSplat helpers. NFCI.
Simon Pilgrim [Mon, 25 Feb 2019 16:31:58 +0000 (16:31 +0000)]
[SelectionDAG] Add demanded elts variants to isConstOrConstSplat helpers. NFCI.

These helpers extend the existing isConstOrConstSplat helper checks to support DemandedElts masks as well.

We already had a local version of this in SelectionDAG that computeKnownBits/ComputeNumSignBits made use of, but this adds the functionality directly to the BuildVectorSDNode node and extends isConstOrConstSplat etc. to use that.

This will allow us to reuse the functionality in SimplifyDemandedVectorElts/SimplifyDemandedBits.

Differential Revision: https://reviews.llvm.org/D58503

llvm-svn: 354797

5 years agoUpdate status page with papers/issues adopted in Kona
Marshall Clow [Mon, 25 Feb 2019 16:12:00 +0000 (16:12 +0000)]
Update status page with papers/issues adopted in Kona

llvm-svn: 354796

5 years agoMake static counters in ASTContext non-static.
Alexander Kornienko [Mon, 25 Feb 2019 16:08:46 +0000 (16:08 +0000)]
Make static counters in ASTContext non-static.

Summary:
Fixes a data race and makes it possible to run clang-based tools in
multithreaded environment with TSan.

Reviewers: ilya-biryukov, riccibruno

Reviewed By: riccibruno

Subscribers: riccibruno, jfb, cfe-commits

Tags: #clang

Differential Revision: https://reviews.llvm.org/D58612

llvm-svn: 354795

5 years ago[DAGCombine] Add undef shuffle elt support to partitionShuffleOfConcats
Simon Pilgrim [Mon, 25 Feb 2019 16:02:01 +0000 (16:02 +0000)]
[DAGCombine] Add undef shuffle elt support to partitionShuffleOfConcats

Support undef shuffle mask indices in the shuffle(concat_vectors, concat_vectors) -> concat_vectors fold

Differential Revision: https://reviews.llvm.org/D58585

llvm-svn: 354793

5 years ago[clangd] Drop documentation in static index if symbols are not indexed for completion.
Haojian Wu [Mon, 25 Feb 2019 16:00:00 +0000 (16:00 +0000)]
[clangd] Drop documentation in static index if symbols are not indexed for completion.

Summary:
This is a further optimization of r350803, we drop docs in static index for
symbols not being indexed for completion, while keeping the docs in dynamic
index (we rely on dynamic index to get docs for class members).

Reviewers: ilya-biryukov

Reviewed By: ilya-biryukov

Subscribers: ioeric, MaskRay, jkorous, arphaman, kadircet, cfe-commits

Differential Revision: https://reviews.llvm.org/D56539

llvm-svn: 354792

5 years ago[ARM] Add some more missing T1 opcodes for the peephole optimisier
David Green [Mon, 25 Feb 2019 15:50:54 +0000 (15:50 +0000)]
[ARM] Add some more missing T1 opcodes for the peephole optimisier

This adds a few extra Thumb1 opcodes to improve the peephole opimisers
ability to remove redundant cmp instructions. tADC and tSBC require
a small fixup to prevent MOVS being moved past the instruction, giving
the wrong flags.

Differential Revision: https://reviews.llvm.org/D58281

llvm-svn: 354791

5 years ago[Vectorizer] Add vectorization support for fixed smul/umul intrinsics
Simon Pilgrim [Mon, 25 Feb 2019 15:42:02 +0000 (15:42 +0000)]
[Vectorizer] Add vectorization support for fixed smul/umul intrinsics

This requires a couple of tweaks to existing vectorization functions as they were assuming that only the second call argument (ctlz/cttz/powi) could ever be the 'always scalar' argument, but for smul.fix + umul.fix its the third argument.

Differential Revision: https://reviews.llvm.org/D58616

llvm-svn: 354790

5 years ago[AArch64] Add support for Cortex-A76 and Cortex-A76AE
Luke Cheeseman [Mon, 25 Feb 2019 15:11:31 +0000 (15:11 +0000)]
[AArch64] Add support for Cortex-A76 and Cortex-A76AE

- Add LLVM backend support for Cortex-A76 and Cortex-A76AE
- Documentation can be found at
  https://developer.arm.com/products/processors/cortex-a/cortex-a76

Differential Revision: https://reviews.llvm.org/D57764

llvm-svn: 354789

5 years ago[AArch64] Add support for Cortex-A76 and Cortex-A76AE
Luke Cheeseman [Mon, 25 Feb 2019 15:08:27 +0000 (15:08 +0000)]
[AArch64] Add support for Cortex-A76 and Cortex-A76AE

- Add LLVM backend support for Cortex-A76 and Cortex-A76AE
- Documentation can be found at
  https://developer.arm.com/products/processors/cortex-a/cortex-a76

llvm-svn: 354788

5 years ago[llvm-objcopy] Add --add-symbol
Eugene Leviant [Mon, 25 Feb 2019 14:12:41 +0000 (14:12 +0000)]
[llvm-objcopy] Add --add-symbol

Differential revision: https://reviews.llvm.org/D58234

llvm-svn: 354787

5 years agoMoved clangd docs to a separate directory in preparation to restructure them into...
Dmitri Gribenko [Mon, 25 Feb 2019 13:43:48 +0000 (13:43 +0000)]
Moved clangd docs to a separate directory in preparation to restructure them into multiple files

Reviewers: ilya-biryukov

Subscribers: ioeric, MaskRay, jkorous, arphaman, kadircet, jdoerfert, cfe-commits

Tags: #clang

Differential Revision: https://reviews.llvm.org/D58607

llvm-svn: 354786

5 years agoFixed typos in tests: s/CHEKC/CHECK/
Dmitri Gribenko [Mon, 25 Feb 2019 13:41:59 +0000 (13:41 +0000)]
Fixed typos in tests: s/CHEKC/CHECK/

Reviewers: ilya-biryukov

Subscribers: nemanjai, javed.absar, jsji, cfe-commits, llvm-commits

Tags: #clang, #llvm

Differential Revision: https://reviews.llvm.org/D58611

llvm-svn: 354785

5 years ago[TTI] Add generic cost model for smul/umul overflow intrinsics
Simon Pilgrim [Mon, 25 Feb 2019 13:30:23 +0000 (13:30 +0000)]
[TTI] Add generic cost model for smul/umul overflow intrinsics

Based off smul/umul fixed costs and the implementation in TargetLowering::expandMULO.

llvm-svn: 354784

5 years ago[SLPVectorizer][X86] Add fixed smul/umul tests
Simon Pilgrim [Mon, 25 Feb 2019 13:26:30 +0000 (13:26 +0000)]
[SLPVectorizer][X86] Add fixed smul/umul tests

Baseline tests - fixed mul intrinsics aren't flagged as vectorizable yet

llvm-svn: 354783

5 years ago[llvm-objdump] Add `Version References` dumper
Xing GUO [Mon, 25 Feb 2019 13:13:19 +0000 (13:13 +0000)]
[llvm-objdump] Add `Version References` dumper

Summary: Add symbol version dumper for [#30241](https://bugs.llvm.org/show_bug.cgi?id=30241)

Reviewers: jhenderson, MaskRay, kristina, emaste, grimar

Reviewed By: jhenderson, grimar

Subscribers: grimar, rupprecht, jakehehrlich, llvm-commits

Tags: #llvm

Differential Revision: https://reviews.llvm.org/D54697

llvm-svn: 354782

5 years agoFixed typos in tests: s/CEHCK/CHECK/
Dmitri Gribenko [Mon, 25 Feb 2019 13:12:33 +0000 (13:12 +0000)]
Fixed typos in tests: s/CEHCK/CHECK/

Reviewers: ilya-biryukov

Subscribers: sanjoy, sdardis, javed.absar, jrtc27, atanasyan, llvm-commits

Tags: #llvm

Differential Revision: https://reviews.llvm.org/D58608

llvm-svn: 354781

5 years ago[clang-tidy] misc-string-integer-assignment: ignore toupper/tolower
Clement Courbet [Mon, 25 Feb 2019 13:09:02 +0000 (13:09 +0000)]
[clang-tidy] misc-string-integer-assignment: ignore toupper/tolower

Summary: Tis represents ~20% of false positives. See PR27723.

Reviewers: xazax.hun, alexfh

Subscribers: rnkovacs, cfe-commits

Tags: #clang

Differential Revision: https://reviews.llvm.org/D58604

llvm-svn: 354780

5 years agoUpdated the documentation build instructions for the current CMake build system
Dmitri Gribenko [Mon, 25 Feb 2019 13:03:44 +0000 (13:03 +0000)]
Updated the documentation build instructions for the current CMake build system

Reviewers: ilya-biryukov

Subscribers: cfe-commits

Tags: #clang

Differential Revision: https://reviews.llvm.org/D58603

llvm-svn: 354779

5 years agoFixed grammar in index.rst
Dmitri Gribenko [Mon, 25 Feb 2019 12:49:27 +0000 (12:49 +0000)]
Fixed grammar in index.rst

Subscribers: arphaman, cfe-commits

Tags: #clang

Differential Revision: https://reviews.llvm.org/D58601

llvm-svn: 354778

5 years agoRemoved an unhelpful comment in index.rst
Dmitri Gribenko [Mon, 25 Feb 2019 12:48:52 +0000 (12:48 +0000)]
Removed an unhelpful comment in index.rst

Reviewers: ilya-biryukov

Subscribers: arphaman, jdoerfert, cfe-commits

Tags: #clang

Differential Revision: https://reviews.llvm.org/D58602

llvm-svn: 354777

5 years agoTest commit (remove a blank space)
Ganesh Gopalasubramanian [Mon, 25 Feb 2019 12:27:49 +0000 (12:27 +0000)]
Test commit (remove a blank space)

Change-Id: I69175571d3b1defeb85e96fdd87db5c3ccadcb63
llvm-svn: 354775

5 years ago[TTI] Add generic cost model for fixed point smul/umul
Simon Pilgrim [Mon, 25 Feb 2019 11:59:23 +0000 (11:59 +0000)]
[TTI] Add generic cost model for fixed point smul/umul

Based on an IR equivalent of target lowering's generic expansion - target specific costs will typically be lower (IR doesn't have a good mull/mulh equivalent) but we need a baseline.

Differential Revision: https://reviews.llvm.org/D57925

llvm-svn: 354774

5 years ago[SYCL] Add clang front-end option to enable SYCL device compilation flow.
Alexey Bader [Mon, 25 Feb 2019 11:48:48 +0000 (11:48 +0000)]
[SYCL] Add clang front-end option to enable SYCL device compilation flow.

Patch by Mariya Podchishchaeva <mariya.podchishchaeva@intel.com>

llvm-svn: 354773

5 years ago[mips] Reduce number of tools invocations in the test. NFC
Simon Atanasyan [Mon, 25 Feb 2019 11:30:33 +0000 (11:30 +0000)]
[mips] Reduce number of tools invocations in the test. NFC

llvm-svn: 354772

5 years ago[X86] Merge ISD::ADD/SUB nodes into X86ISD::ADD/SUB equivalents (PR40483)
Simon Pilgrim [Mon, 25 Feb 2019 11:19:37 +0000 (11:19 +0000)]
[X86] Merge ISD::ADD/SUB nodes into X86ISD::ADD/SUB equivalents (PR40483)

Avoid ADD/SUB instruction duplication by reusing the X86ISD::ADD/SUB results.

Includes ADD commutation - I tried to include NEG+SUB SUB commutation as well but this causes regressions as we don't have good combine coverage to simplify X86ISD::SUB.

Differential Revision: https://reviews.llvm.org/D58597

llvm-svn: 354771

5 years ago[yaml2obj]Re-allow dynamic sections to have raw content
James Henderson [Mon, 25 Feb 2019 11:02:24 +0000 (11:02 +0000)]
[yaml2obj]Re-allow dynamic sections to have raw content

Recently, support was added to yaml2obj to allow dynamic sections to
have a list of entries, to make it easier to write tests with dynamic
sections. However, this change also removed the ability to provide
custom contents to the dynamic section, making it hard to test
malformed contents (e.g. because the section is not a valid size to
contain an array of entries). This change reinstates this. An error is
emitted if raw content and dynamic entries are both specified.

Reviewed by: grimar, ruiu

Differential Review: https://reviews.llvm.org/D58543

llvm-svn: 354770

5 years ago[ELF][ARM] Accept and ignore -p and -no-pipleline-knowledge
Peter Smith [Mon, 25 Feb 2019 10:48:31 +0000 (10:48 +0000)]
[ELF][ARM] Accept and ignore -p and -no-pipleline-knowledge

The linux kernel uses an old flag -p/-no-pipeline-knowledge that is
accepted by bfd and gold but ignored by modern versions of them. The
original option is very old and is pre-ABI, it sometimes comes up in
code-bases that had support for pre ABI toolchains. The Linux kernel uses
it in 3 places in the ARM specific section.

Differential Revision: https://reviews.llvm.org/D58540

llvm-svn: 354769

5 years ago[ARM] Make fullfp16 instructions not conditionalisable.
Simon Tatham [Mon, 25 Feb 2019 10:39:53 +0000 (10:39 +0000)]
[ARM] Make fullfp16 instructions not conditionalisable.

More or less all the instructions defined in the v8.2a full-fp16
extension are defined as UNPREDICTABLE if you put them in an IT block
(Thumb) or use with any condition other than AL (ARM). LLVM didn't
know that, and was happy to conditionalise them.

In order to force these instructions to count as not predicable, I had
to make a small Tablegen change. The code generation back end mostly
decides if an instruction was predicable by looking for something it
can identify as a predicate operand; there's an isPredicable bit flag
that overrides that check in the positive direction, but nothing that
overrides it in the negative direction.

(I considered the alternative approach of actually removing the
predicate operand from those instructions, but thought that it would
be more painful overall for instructions differing only in data type
to have different shapes of operand list. This way, the only code that
has to notice the difference is the if-converter.)

So I've added an isUnpredicable bit alongside isPredicable, and set
that bit on the right subset of FP16 instructions, and also on the
VSEL, VMAXNM/VMINNM and VRINT[ANPM] families which should be
unpredicable for all data types.

I've included a couple of representative regression tests, both of
which previously caused an fp16 instruction to be conditionalised in
ARM state and (with -arm-no-restrict-it) to be put in an IT block in
Thumb.

Reviewers: SjoerdMeijer, t.p.northover, efriedma

Reviewed By: efriedma

Subscribers: jdoerfert, javed.absar, kristof.beyls, hiraditya, llvm-commits

Tags: #llvm

Differential Revision: https://reviews.llvm.org/D57823

llvm-svn: 354768

5 years ago[llvm-exegesis] Split Epsilon param into two (PR40787)
Roman Lebedev [Mon, 25 Feb 2019 09:36:12 +0000 (09:36 +0000)]
[llvm-exegesis] Split Epsilon param into two (PR40787)

Summary:
This eps param is used for two distinct things:
* initial point clusterization
* checking clusters against the llvm values

What if one wants to only look at highly different clusters, without changing
the clustering itself? In particular, this helps to weed out noisy measurements
(since the clusterization epsilon is still small, so there is a better chance
that noisy measurements from the same opcode will go into different clusters)

By splitting it into two params it is now possible.

This is nearly-free performance-wise:
Old:
```
$ perf stat -r 25 ./bin/llvm-exegesis -mode=analysis -benchmarks-file=/home/lebedevri/PileDriver-Sched/benchmarks-latency-1.yaml -analysis-inconsistencies-output-file=/tmp/clusters-old.html
no exegesis target for x86_64-unknown-linux-gnu, using default
Parsed 10099 benchmark points
Printing sched class consistency analysis results to file '/tmp/clusters-old.html'
...
 Performance counter stats for './bin/llvm-exegesis -mode=analysis -benchmarks-file=/home/lebedevri/PileDriver-Sched/benchmarks-latency-1.yaml -analysis-inconsistencies-output-file=/tmp/clusters-old.html' (25 runs):

            390.01 msec task-clock                #    0.998 CPUs utilized            ( +-  0.25% )
                12      context-switches          #   31.735 M/sec                    ( +- 27.38% )
                 0      cpu-migrations            #    0.000 K/sec
              4745      page-faults               # 12183.732 M/sec                   ( +-  0.54% )
        1562711900      cycles                    # 4012303.327 GHz                   ( +-  0.24% )  (82.90%)
         185567822      stalled-cycles-frontend   #   11.87% frontend cycles idle     ( +-  0.52% )  (83.30%)
         392106234      stalled-cycles-backend    #   25.09% backend cycles idle      ( +-  1.31% )  (33.79%)
        1839236666      instructions              #    1.18  insn per cycle
                                                  #    0.21  stalled cycles per insn  ( +-  0.15% )  (50.37%)
         407035764      branches                  # 1045074878.710 M/sec              ( +-  0.12% )  (66.80%)
          10896459      branch-misses             #    2.68% of all branches          ( +-  0.17% )  (83.20%)

          0.390629 +- 0.000972 seconds time elapsed  ( +-  0.25% )
```
```
$ perf stat -r 9 ./bin/llvm-exegesis -mode=analysis -benchmarks-file=/home/lebedevri/PileDriver-Sched/benchmarks-latency.yml -analysis-inconsistencies-output-file=/tmp/clusters-old.html
no exegesis target for x86_64-unknown-linux-gnu, using default
Parsed 50572 benchmark points
Printing sched class consistency analysis results to file '/tmp/clusters-old.html'
...
 Performance counter stats for './bin/llvm-exegesis -mode=analysis -benchmarks-file=/home/lebedevri/PileDriver-Sched/benchmarks-latency.yml -analysis-inconsistencies-output-file=/tmp/clusters-old.html' (9 runs):

           6803.36 msec task-clock                #    0.999 CPUs utilized            ( +-  0.96% )
               262      context-switches          #   38.546 M/sec                    ( +- 23.06% )
                 0      cpu-migrations            #    0.065 M/sec                    ( +- 76.03% )
             13287      page-faults               # 1953.206 M/sec                    ( +-  0.32% )
       27252537904      cycles                    # 4006024.257 GHz                   ( +-  0.95% )  (83.31%)
        1496314935      stalled-cycles-frontend   #    5.49% frontend cycles idle     ( +-  0.97% )  (83.32%)
       16128404524      stalled-cycles-backend    #   59.18% backend cycles idle      ( +-  0.30% )  (33.37%)
       17611143370      instructions              #    0.65  insn per cycle
                                                  #    0.92  stalled cycles per insn  ( +-  0.05% )  (50.04%)
        3894906599      branches                  # 572537147.437 M/sec               ( +-  0.03% )  (66.69%)
         116314514      branch-misses             #    2.99% of all branches          ( +-  0.20% )  (83.35%)

            6.8118 +- 0.0689 seconds time elapsed  ( +-  1.01%)
```
New:
```
$ perf stat -r 25 ./bin/llvm-exegesis -mode=analysis -benchmarks-file=/home/lebedevri/PileDriver-Sched/benchmarks-latency-1.yaml -analysis-inconsistencies-output-file=/tmp/clusters-new.html
no exegesis target for x86_64-unknown-linux-gnu, using default
Parsed 10099 benchmark points
Printing sched class consistency analysis results to file '/tmp/clusters-new.html'
...
 Performance counter stats for './bin/llvm-exegesis -mode=analysis -benchmarks-file=/home/lebedevri/PileDriver-Sched/benchmarks-latency-1.yaml -analysis-inconsistencies-output-file=/tmp/clusters-new.html' (25 runs):

            400.14 msec task-clock                #    0.998 CPUs utilized            ( +-  0.66% )
                12      context-switches          #   29.429 M/sec                    ( +- 25.95% )
                 0      cpu-migrations            #    0.100 M/sec                    ( +-100.00% )
              4714      page-faults               # 11796.496 M/sec                   ( +-  0.55% )
        1603131306      cycles                    # 4011840.105 GHz                   ( +-  0.66% )  (82.85%)
         199538509      stalled-cycles-frontend   #   12.45% frontend cycles idle     ( +-  2.40% )  (83.10%)
         402249109      stalled-cycles-backend    #   25.09% backend cycles idle      ( +-  1.19% )  (34.05%)
        1847783963      instructions              #    1.15  insn per cycle
                                                  #    0.22  stalled cycles per insn  ( +-  0.18% )  (50.64%)
         407162722      branches                  # 1018925730.631 M/sec              ( +-  0.12% )  (67.02%)
          10932779      branch-misses             #    2.69% of all branches          ( +-  0.51% )  (83.28%)

           0.40077 +- 0.00267 seconds time elapsed  ( +-  0.67% )

lebedevri@pini-pini:/build/llvm-build-Clang-release$ perf stat -r 9 ./bin/llvm-exegesis -mode=analysis -benchmarks-file=/home/lebedevri/PileDriver-Sched/benchmarks-latency.yml -analysis-inconsistencies-output-file=/tmp/clusters-new.html
no exegesis target for x86_64-unknown-linux-gnu, using default
Parsed 50572 benchmark points
Printing sched class consistency analysis results to file '/tmp/clusters-new.html'
...
 Performance counter stats for './bin/llvm-exegesis -mode=analysis -benchmarks-file=/home/lebedevri/PileDriver-Sched/benchmarks-latency.yml -analysis-inconsistencies-output-file=/tmp/clusters-new.html' (9 runs):

           6947.79 msec task-clock                #    1.000 CPUs utilized            ( +-  0.90% )
               217      context-switches          #   31.236 M/sec                    ( +- 36.16% )
                 1      cpu-migrations            #    0.096 M/sec                    ( +- 50.00% )
             13258      page-faults               # 1908.389 M/sec                    ( +-  0.34% )
       27830796523      cycles                    # 4006032.286 GHz                   ( +-  0.89% )  (83.30%)
        1504554006      stalled-cycles-frontend   #    5.41% frontend cycles idle     ( +-  2.10% )  (83.32%)
       16716574843      stalled-cycles-backend    #   60.07% backend cycles idle      ( +-  0.65% )  (33.38%)
       17755545931      instructions              #    0.64  insn per cycle
                                                  #    0.94  stalled cycles per insn  ( +-  0.09% )  (50.04%)
        3897255686      branches                  # 560980426.597 M/sec               ( +-  0.06% )  (66.70%)
         117045395      branch-misses             #    3.00% of all branches          ( +-  0.47% )  (83.34%)

            6.9507 +- 0.0627 seconds time elapsed  ( +-  0.90% )
```

I.e. it's +2.6% slowdown for one whole sweep, or +2% for 5 whole sweeps.
Within noise i'd say.

Should help with [[ https://bugs.llvm.org/show_bug.cgi?id=40787 | PR40787 ]].

Reviewers: courbet, gchatelet

Reviewed By: courbet

Subscribers: tschuett, RKSimon, llvm-commits

Tags: #llvm

Differential Revision: https://reviews.llvm.org/D58476

llvm-svn: 354767

5 years agoFinish revert of r354706
Pavel Labath [Mon, 25 Feb 2019 09:30:41 +0000 (09:30 +0000)]
Finish revert of r354706

The revert in r354711 wasn't complete. Finish the job.

llvm-svn: 354766

5 years ago[clangd] Add thread priority lowering for MacOS as well
Kadir Cetinkaya [Mon, 25 Feb 2019 09:19:26 +0000 (09:19 +0000)]
[clangd] Add thread priority lowering for MacOS as well

Reviewers: ilya-biryukov

Subscribers: ioeric, MaskRay, jkorous, arphaman, cfe-commits

Tags: #clang

Differential Revision: https://reviews.llvm.org/D58492

llvm-svn: 354765

5 years ago[XRay][tools] Revert "Use Support/JSON.h in llvm-xray convert"
Roman Lebedev [Mon, 25 Feb 2019 07:39:07 +0000 (07:39 +0000)]
[XRay][tools] Revert "Use Support/JSON.h in llvm-xray convert"

Summary:
This reverts D50129 / rL338834: [XRay][tools] Use Support/JSON.h in llvm-xray convert

Abstractions are great.
Readable code is great.
JSON support library is a *good* idea.

However unfortunately, there is an internal detail that one needs
to be aware of in `llvm::json::Object` - it uses `llvm::DenseMap`.
So for **every** `llvm::json::Object`, even if you only store a single `int`
entry there, you pay the whole price of `llvm::DenseMap`.

Unfortunately, it matters for `llvm-xray`.

I was trying to analyse the `llvm-exegesis` analysis mode performance,
and for that i wanted to view the LLVM X-Ray log visualization in Chrome
trace viewer. And the `llvm-xray convert` is sluggish, and sometimes
even ended up being killed by OOM.

`xray-log.llvm-exegesis.lwZ0sT` was acquired from `llvm-exegesis`
(compiled with ` -fxray-instruction-threshold=128`)
analysis mode over `-benchmarks-file` with 10099 points (one full
latency measurement set), with normal runtime of 0.387s.

Timings:
Old: (copied from D58580)
```
$ perf stat -r 5 ./bin/llvm-xray convert -sort -symbolize -instr_map=./bin/llvm-exegesis -output-format=trace_event -output=/tmp/trace.yml xray-log.llvm-exegesis.lwZ0sT

 Performance counter stats for './bin/llvm-xray convert -sort -symbolize -instr_map=./bin/llvm-exegesis -output-format=trace_event -output=/tmp/trace.yml xray-log.llvm-exegesis.lwZ0sT' (5 runs):

          21346.24 msec task-clock                #    1.000 CPUs utilized            ( +-  0.28% )
               314      context-switches          #   14.701 M/sec                    ( +- 59.13% )
                 1      cpu-migrations            #    0.037 M/sec                    ( +-100.00% )
           2181354      page-faults               # 102191.251 M/sec                  ( +-  0.02% )
       85477442102      cycles                    # 4004415.019 GHz                   ( +-  0.28% )  (83.33%)
       14526427066      stalled-cycles-frontend   #   16.99% frontend cycles idle     ( +-  0.70% )  (83.33%)
       32371533721      stalled-cycles-backend    #   37.87% backend cycles idle      ( +-  0.27% )  (33.34%)
       67896890228      instructions              #    0.79  insn per cycle
                                                  #    0.48  stalled cycles per insn  ( +-  0.03% )  (50.00%)
       14592654840      branches                  # 683631198.653 M/sec               ( +-  0.02% )  (66.67%)
         212207534      branch-misses             #    1.45% of all branches          ( +-  0.94% )  (83.34%)

           21.3502 +- 0.0585 seconds time elapsed  ( +-  0.27% )
```
New:
```
$ perf stat -r 9 ./bin/llvm-xray convert -sort -symbolize -instr_map=./bin/llvm-exegesis -output-format=trace_event -output=/tmp/trace.yml xray-log.llvm-exegesis.lwZ0sT

 Performance counter stats for './bin/llvm-xray convert -sort -symbolize -instr_map=./bin/llvm-exegesis -output-format=trace_event -output=/tmp/trace.yml xray-log.llvm-exegesis.lwZ0sT' (9 runs):

           7178.38 msec task-clock                #    1.000 CPUs utilized            ( +-  0.26% )
               182      context-switches          #   25.402 M/sec                    ( +- 28.84% )
                 0      cpu-migrations            #    0.046 M/sec                    ( +- 70.71% )
             33701      page-faults               # 4694.994 M/sec                    ( +-  0.88% )
       28761053971      cycles                    # 4006833.933 GHz                   ( +-  0.26% )  (83.32%)
        2028297997      stalled-cycles-frontend   #    7.05% frontend cycles idle     ( +-  1.61% )  (83.32%)
       10773154901      stalled-cycles-backend    #   37.46% backend cycles idle      ( +-  0.38% )  (33.36%)
       36199132874      instructions              #    1.26  insn per cycle
                                                  #    0.30  stalled cycles per insn  ( +-  0.03% )  (50.02%)
        6434504227      branches                  # 896420204.421 M/sec               ( +-  0.03% )  (66.68%)
          73355176      branch-misses             #    1.14% of all branches          ( +-  1.46% )  (83.33%)

            7.1807 +- 0.0190 seconds time elapsed  ( +-  0.26% )
```

So using `llvm::json` nearly triples run-time on that test case.
(+3x is times, not percent.)

Memory:
Old:
```
total runtime: 39.88s.
bytes allocated in total (ignoring deallocations): 79.07GB (1.98GB/s)
calls to allocation functions: 33267816 (834135/s)
temporary memory allocations: 5832298 (146235/s)
peak heap memory consumption: 9.21GB
peak RSS (including heaptrack overhead): 147.98GB
total memory leaked: 1.09MB
```
New:
```
total runtime: 17.42s.
bytes allocated in total (ignoring deallocations): 5.12GB (293.86MB/s)
calls to allocation functions: 21382982 (1227284/s)
temporary memory allocations: 232858 (13364/s)
peak heap memory consumption: 350.69MB
peak RSS (including heaptrack overhead): 2.55GB
total memory leaked: 79.95KB
```
Diff:
```
total runtime: -22.46s.
bytes allocated in total (ignoring deallocations): -73.95GB (3.29GB/s)
calls to allocation functions: -11884834 (529155/s)
temporary memory allocations: -5599440 (249307/s)
peak heap memory consumption: -8.86GB
peak RSS (including heaptrack overhead): 0B
total memory leaked: -1.01MB
```
So using `llvm::json` increases *peak* memory consumption on *this* testcase ~+27x.
And total allocation count +15x. Both of these numbers are times, *not* percent.

And note that memory usage is clearly unbound with `llvm::json`, it directly depends
on the length of the log, so peak memory consumption is always increasing.
This isn't so with the dumb code, there is no accumulating memory consumption,
peak memory consumption is fixed. Naturally, that means it will handle *much*
larger logs without OOM'ing.

Readability is good, but the price is simply unacceptable here.
Too bad none of this analysis was done as part of the development/review D50129 itself.

Reviewers: dberris, kpw, sammccall

Reviewed By: dberris

Subscribers: riccibruno, hans, courbet, jdoerfert, llvm-commits

Tags: #llvm

Differential Revision: https://reviews.llvm.org/D58584

llvm-svn: 354764

5 years ago[SelectionDAG] Add a OPC_CheckChild2CondCode to SelectionDAGISel to remove a MoveChil...
Craig Topper [Mon, 25 Feb 2019 03:11:44 +0000 (03:11 +0000)]
[SelectionDAG] Add a OPC_CheckChild2CondCode to SelectionDAGISel to remove a MoveChild and MoveParent pair.

OPC_CheckCondCode is always used as operand 2 of a setcc. And its always surrounded by a MoveChild2 and a MoveParent. By having a dedicated opcode for this case we can reduce the number of bytes needed for this pattern from 4 bytes to 2.

This saves ~3000 bytes in the X86 table.

llvm-svn: 354763

5 years ago[PowerPC] [PowerPC] Enhance the fast selection of fptoi & fptrunc instruction and...
Kang Zhang [Mon, 25 Feb 2019 02:46:16 +0000 (02:46 +0000)]
[PowerPC] [PowerPC] Enhance the fast selection of fptoi & fptrunc instruction and clean up related asserts

Summary:
Fast selection of llvm fptoi & fptrunc instructions is not handled well about
VSX instruction support.
We'd use VSX float convert integer instruction instead of non-vsx float convert
integer instruction if the operand register class is VSSRC or VSFRC because i32
and i64 are mapped to VSSRC and VSFRC correspondingly if VSX feature is
openeded.
For float trunc instruction, we do this silimar work like float convert integer
instruction to try to use VSX instruction.

Reviewed By: jsji

Differential Revision: https://reviews.llvm.org/D58430

llvm-svn: 354762

5 years ago[clangd] Enhance macro hover to see full definition
Marc-Andre Laperle [Sun, 24 Feb 2019 23:47:03 +0000 (23:47 +0000)]
[clangd] Enhance macro hover to see full definition

Summary: Signed-off-by: Marc-Andre Laperle <malaperle@gmail.com>

Reviewers: simark, ilya-biryukov, sammccall, ioeric, hokein

Reviewed By: ilya-biryukov

Subscribers: ilya-biryukov, ioeric, MaskRay, jkorous, arphaman, kadircet, cfe-commits

Tags: #clang

Differential Revision: https://reviews.llvm.org/D55250

llvm-svn: 354761

5 years ago[InstCombine] Add tests for PR40846; NFC
Nikita Popov [Sun, 24 Feb 2019 21:55:37 +0000 (21:55 +0000)]
[InstCombine] Add tests for PR40846; NFC

The icmps are the same as the overflow result of the intrinsic.

llvm-svn: 354760

5 years ago[InstCombine] Move with.overflow tests to separate file; NFC
Nikita Popov [Sun, 24 Feb 2019 21:55:31 +0000 (21:55 +0000)]
[InstCombine] Move with.overflow tests to separate file; NFC

And regenerate checks. I had to rename some variables, because
update_test_checks can't deal with the same variable names used
in lower and upper case. I've also dropped the result type aliases,
as just using the type directly gives a cleaner result.

llvm-svn: 354759

5 years ago[X86] Add PR40483 test cases
Simon Pilgrim [Sun, 24 Feb 2019 21:13:29 +0000 (21:13 +0000)]
[X86] Add PR40483 test cases

Demonstrate failure to merge ISD::ADD(x,y)/X86ISD::ADD(x,y) + ISD::SUB(x,y)/X86ISD::SUB(x,y) equivalent ops

llvm-svn: 354758

5 years ago[X86] Combine zext(packus(x),packus(y)) -> concat(x,y) (PR39637)
Simon Pilgrim [Sun, 24 Feb 2019 19:57:52 +0000 (19:57 +0000)]
[X86] Combine zext(packus(x),packus(y)) -> concat(x,y) (PR39637)

Its proving tricky to combine shuffles across multiple vector sizes, so for now I'm adding this more specific combine - the pattern is common enough to be worth it as a first step.

llvm-svn: 354757

5 years ago[X86] Fix tls variable lowering issue with large code model
Craig Topper [Sun, 24 Feb 2019 19:33:37 +0000 (19:33 +0000)]
[X86] Fix tls variable lowering issue with large code model

Summary:
The problem here is the lowering for tls variable. Below is the DAG for the code.
SelectionDAG has 11 nodes:

t0: ch = EntryToken
      t8: i64,ch = load<(load 8 from `i8 addrspace(257)* null`, addrspace 257)> t0, Constant:i64<0>, undef:i64
        t10: i64 = X86ISD::WrapperRIP TargetGlobalTLSAddress:i64<i32* @x> 0 [TF=10]
      t11: i64,ch = load<(load 8 from got)> t0, t10, undef:i64
    t12: i64 = add t8, t11
  t4: i32,ch = load<(dereferenceable load 4 from @x)> t0, t12, undef:i64
t6: ch = CopyToReg t0, Register:i32 %0, t4
And when mcmodel is large, below instruction can NOT be folded.

  t10: i64 = X86ISD::WrapperRIP TargetGlobalTLSAddress:i64<i32* @x> 0 [TF=10]
t11: i64,ch = load<(load 8 from got)> t0, t10, undef:i64
So "t11: i64,ch = load<(load 8 from got)> t0, t10, undef:i64" is lowered to " Morphed node: t11: i64,ch = MOV64rm<Mem:(load 8 from got)> t10, TargetConstant:i8<1>, Register:i64 $noreg, TargetConstant:i32<0>, Register:i32 $noreg, t0"

When llvm start to lower "t10: i64 = X86ISD::WrapperRIP TargetGlobalTLSAddress:i64<i32* @x> 0 [TF=10]", it fails.

The patch is to fold the load and X86ISD::WrapperRIP.

Fixes PR26906

Patch by LuoYuanke

Reviewers: craig.topper, rnk, annita.zhang, wxiao3

Reviewed By: rnk

Subscribers: llvm-commits

Tags: #llvm

Differential Revision: https://reviews.llvm.org/D58336

llvm-svn: 354756

5 years ago[X86][SSE] Use pblendw for v4i32/v2i64 during isel.
Craig Topper [Sun, 24 Feb 2019 19:23:41 +0000 (19:23 +0000)]
[X86][SSE] Use pblendw for v4i32/v2i64 during isel.

Summary:

Previously we used BLENDPS/BLENDPD but that puts the blend in the FP domain. Under optsize, the two address instruction pass can cause blendps/blendpd to commute to blendps/blendpd. But we probably shouldn't do that if the original type was a integer. So use pblendw instead.

Reviewers: spatel, RKSimon

Reviewed By: RKSimon

Subscribers: jdoerfert, llvm-commits

Tags: #llvm

Differential Revision: https://reviews.llvm.org/D58574

llvm-svn: 354755

5 years ago[X86] Correct some ADC/SBB with immediate scheduler data for Broadwell and Skylake.
Craig Topper [Sun, 24 Feb 2019 19:23:39 +0000 (19:23 +0000)]
[X86] Correct some ADC/SBB with immediate scheduler data for Broadwell and Skylake.

Summary:
The AX/EAX/RAX with immediate forms are 2 uops just like the AL with immediate.

The modrm form with r8 and immediate is a single uop just like r16/r32/r64 with immediate.

Reviewers: RKSimon, andreadb

Reviewed By: RKSimon

Subscribers: gbedwell, llvm-commits

Tags: #llvm

Differential Revision: https://reviews.llvm.org/D58581

llvm-svn: 354754

5 years ago[LegalizeTypes][AArch64][X86] Make type legalization of vector (S/U)ADD/SUB/MULO...
Craig Topper [Sun, 24 Feb 2019 19:23:36 +0000 (19:23 +0000)]
[LegalizeTypes][AArch64][X86] Make type legalization of vector (S/U)ADD/SUB/MULO follow getSetCCResultType for the overflow bits. Make UnrollVectorOverflowOp properly convert from scalar boolean contents to vector boolean contents

Summary:
When promoting the over flow vector for these ops we should use the target's desired setcc result type. This way a v8i32 result type will use a v8i32 overflow vector instead of a v8i16 overflow vector. A v8i16 overflow vector will cause LegalizeDAG/LegalizeVectorOps to have to use v8i32 and truncate to v8i16 in its expansion. By doing this in type legalization instead, we get the truncate into the DAG earlier and give DAG combine more of a chance to optimize it.

We also have to fix unrolling to use the scalar setcc result type for the scalarized operation, and convert it to the required vector element type after the scalar operation. We have to observe the vector boolean contents when doing this conversion. The previous code was just taking the scalar result and putting it in the vector. But for X86 and AArch64 that would have only put a the boolean value in bit 0 of the element and left all other bits in the element 0. We need to ensure all bits in the element are the same. I'm using a select with constants here because that's what setcc unrolling in LegalizeVectorOps used.

Reviewers: spatel, RKSimon, nikic

Reviewed By: nikic

Subscribers: javed.absar, kristof.beyls, dmgreen, llvm-commits

Tags: #llvm

Differential Revision: https://reviews.llvm.org/D58567

llvm-svn: 354753

5 years agoFix accidentally used hard tabs. NFC
Kristina Brooks [Sun, 24 Feb 2019 18:06:10 +0000 (18:06 +0000)]
Fix accidentally used hard tabs. NFC

Big sorry. This undoes the indentation mess I made
in r354751.

llvm-svn: 354752

5 years agoWrap code for builtin_assume_aligned at 80 col.NFC
Kristina Brooks [Sun, 24 Feb 2019 17:57:33 +0000 (17:57 +0000)]
Wrap code for builtin_assume_aligned at 80 col.NFC

Minor style fix to avoid going over 80 cols in handling
of case for Builtin::BI__builtin_assume_aligned. NFC.

llvm-svn: 354751