platform/upstream/llvm.git
6 years ago[OpenMP] Use LIBOMPTARGET_DEVICE_RTL_DEBUG env var to control debug messages on the...
Guansong Zhang [Fri, 4 May 2018 19:29:28 +0000 (19:29 +0000)]
[OpenMP] Use LIBOMPTARGET_DEVICE_RTL_DEBUG env var to control debug messages on the device side

Summary:
Enable the device side debug messages at compile time, use env var to control at runtime.

To achieve this, an environment data block is passed to the device lib when it is loaded.

By default, the message is off, to enable it, a user need to set LIBOMPDEVICE_DEBUG=1.

Reviewers: grokos

Reviewed By: grokos

Subscribers: openmp-commits

Tags: #openmp

Differential Revision: https://reviews.llvm.org/D46210

llvm-svn: 331550

6 years ago[MachineLICM] Debug intrinsics shouldn't affect hoist decisions
Geoff Berry [Fri, 4 May 2018 19:25:09 +0000 (19:25 +0000)]
[MachineLICM] Debug intrinsics shouldn't affect hoist decisions

Summary:
When checking if an instruction stores to a given frame index, check
that the instruction can write to memory before looking at the memory
operands list to avoid e.g. DBG_VALUE instructions that reference a
frame index preventing a load from that index from being hoisted.

Reviewers: dblaikie, MatzeB, qcolombet, reames, javed.absar

Subscribers: mcrosier, llvm-commits

Differential Revision: https://reviews.llvm.org/D46284

llvm-svn: 331549

6 years ago[ObjCARC] Account for catchswitch in bitcast insertion
Shoaib Meenai [Fri, 4 May 2018 19:03:11 +0000 (19:03 +0000)]
[ObjCARC] Account for catchswitch in bitcast insertion

A catchswitch is both a pad and a terminator, meaning it must be the
only non-phi instruction in its basic block. When we're inserting a
bitcast in the incoming basic block for a phi, if that incoming block is
a catchswitch, we should go up the dominator tree to find a valid
insertion point rather than attempting to insert before the catchswitch
(which would result in invalid IR).

Differential Revision: https://reviews.llvm.org/D46412

llvm-svn: 331548

6 years agoFast Math Flag mapping into SDNode
Michael Berg [Fri, 4 May 2018 18:48:20 +0000 (18:48 +0000)]
Fast Math Flag mapping into SDNode

Summary: Adding support for Fast flags in the SDNode to leverage fast math sub flag usage.

Reviewers: spatel, arsenm, jbhateja, hfinkel, escha, qcolombet, echristo, wristow, javed.absar

Reviewed By: spatel

Subscribers: llvm-commits, rampitec, nhaehnle, tstellar, FarhanaAleen, nemanjai, javed.absar, jbhateja, hfinkel, wdng

Differential Revision: https://reviews.llvm.org/D45710

llvm-svn: 331547

6 years ago[X86] Add WriteEMMS scheduler class
Simon Pilgrim [Fri, 4 May 2018 18:16:13 +0000 (18:16 +0000)]
[X86] Add WriteEMMS scheduler class

Filled in the missing values from Btver2 SoG or Agner

llvm-svn: 331546

6 years agoAdd support for ObjC property name to be a single acronym.
Yan Zhang [Fri, 4 May 2018 18:14:08 +0000 (18:14 +0000)]
Add support for ObjC property name to be a single acronym.

Summary:
This change will support cases like:

```
@property(assign, nonatomic) int ID;
```

Reviewers: benhamilton, hokein

Reviewed By: benhamilton

Subscribers: klimek, cfe-commits

Tags: #clang-tools-extra

Differential Revision: https://reviews.llvm.org/D46374

llvm-svn: 331545

6 years ago[clang-format] Refactor #include insertion/deletion functionality into a class.
Eric Liu [Fri, 4 May 2018 17:55:13 +0000 (17:55 +0000)]
[clang-format] Refactor #include insertion/deletion functionality into a class.

Summary:
The class will be moved into libToolingCore as followup.

The new behaviors in this patch:
- New #include is inserted in the right position in a #include block to
preserver sorted #includes. This is best effort - only works when the
block is already sorted.
- When inserting multiple #includes to the end of a file which doesn't
end with a "\n" character, a "\n" will be prepended to each #include.
This is a special and rare case that was previously handled. This is now
relaxed to avoid complexity as it's rare in practice.

Reviewers: ilya-biryukov

Reviewed By: ilya-biryukov

Subscribers: klimek, cfe-commits, djasper

Differential Revision: https://reviews.llvm.org/D46180

llvm-svn: 331544

6 years ago[X86] Finish splitting WriteVecShift and WriteVecIMul to remove InstRW overrides.
Simon Pilgrim [Fri, 4 May 2018 17:47:46 +0000 (17:47 +0000)]
[X86] Finish splitting WriteVecShift and WriteVecIMul to remove InstRW overrides.

llvm-svn: 331543

6 years ago[LoopIdiomRecognize] Don't create an IRBuilder just to call getTrue/getFalse.
Craig Topper [Fri, 4 May 2018 17:39:08 +0000 (17:39 +0000)]
[LoopIdiomRecognize] Don't create an IRBuilder just to call getTrue/getFalse.

We can call the methods in ConstantInt directly. We just need a context.

llvm-svn: 331542

6 years ago[AArch64] Add missing testcase for r331522
Adhemerval Zanella [Fri, 4 May 2018 17:21:26 +0000 (17:21 +0000)]
[AArch64] Add missing testcase for r331522

llvm-svn: 331541

6 years ago[llvm-mca] Add descriptive names for the TimelineView report characters. NFC.
Matt Davis [Fri, 4 May 2018 17:19:40 +0000 (17:19 +0000)]
[llvm-mca] Add descriptive names for the TimelineView report characters. NFC.

Summary:
This change makes the TimelineView source simpler to read and easier to modify in the future.
This patch introduces a class of static chars used as the display values in the TimelineView report, this change just eliminates a few magic characters.

Reviewers: andreadb, courbet, RKSimon

Reviewed By: andreadb

Subscribers: tschuett, gbedwell, llvm-commits

Differential Revision: https://reviews.llvm.org/D46409

llvm-svn: 331540

6 years ago[clang-doc] Attaching a name to reference data
Julie Hockett [Fri, 4 May 2018 17:02:13 +0000 (17:02 +0000)]
[clang-doc] Attaching a name to reference data

This adds the name of the referenced decl, in addition to its USR, to
the saved data, so that the backend can look at an info in isolation and
still be able to construct a human-readable name for it.

Differential Revision: https://reviews.llvm.org/D46281

llvm-svn: 331539

6 years agoObject: The default alignment of a section without alignment flags is 16.
Peter Collingbourne [Fri, 4 May 2018 16:45:57 +0000 (16:45 +0000)]
Object: The default alignment of a section without alignment flags is 16.

Differential Revision: https://reviews.llvm.org/D46420

llvm-svn: 331538

6 years agoobj2yaml: Correctly round-trip default alignment.
Peter Collingbourne [Fri, 4 May 2018 16:28:41 +0000 (16:28 +0000)]
obj2yaml: Correctly round-trip default alignment.

Previously we were emitting the "cooked" alignment, which made it hard
to distinguish between that and the default alignment.

Differential Revision: https://reviews.llvm.org/D46418

llvm-svn: 331537

6 years ago[NFC]Convert Class to use member initialization instead of inline.
Erich Keane [Fri, 4 May 2018 16:19:53 +0000 (16:19 +0000)]
[NFC]Convert Class to use member initialization instead of inline.

llvm-svn: 331536

6 years agoDwarfCompileUnit: Fix another assertion failure on malformed input
Adrian Prantl [Fri, 4 May 2018 16:10:43 +0000 (16:10 +0000)]
DwarfCompileUnit: Fix another assertion failure on malformed input
that is not rejected by the Verifier.

Thanks to Björn Pettersson for providing a reproducer!

llvm-svn: 331535

6 years ago[PPC64] Remove support for ELF V1 ABI in LLD - buildbot fix
Zaara Syeda [Fri, 4 May 2018 16:04:04 +0000 (16:04 +0000)]
[PPC64] Remove support for ELF V1 ABI in LLD - buildbot fix

Fix buildbot error, failure to build with msvc due to error C2446
Use switch instead of ternary operator.

Differential Revision: https://reviews.llvm.org/D46316

llvm-svn: 331534

6 years ago[clang-cl] Print /showIncludes to stderr, if used in combination with /E, /EP or /P
Erich Keane [Fri, 4 May 2018 15:58:31 +0000 (15:58 +0000)]
[clang-cl] Print /showIncludes to stderr, if used in combination with /E, /EP or /P

This replicates 'cl.exe' behavior and allows for both preprocessor output and
dependency information to be extraced with a single compiler invocation.

This is especially useful for compiler caching with tools like Mozilla's sccache.

See: https://github.com/mozilla/sccache/issues/246

Patch By: fxb

Differential Revision: https://reviews.llvm.org/D46394

llvm-svn: 331533

6 years ago[llvm-exegesis] Fix pfm counter names for BDW.
Clement Courbet [Fri, 4 May 2018 15:26:12 +0000 (15:26 +0000)]
[llvm-exegesis] Fix pfm counter names for BDW.

Summary: They are not consistent with other microarchitectures.

Reviewers: gchatelet

Subscribers: tschuett, llvm-commits

Differential Revision: https://reviews.llvm.org/D46434

llvm-svn: 331532

6 years ago[X86] Cleanup SchedWriteFMA classes and use X86SchedWriteWidths directly.
Simon Pilgrim [Fri, 4 May 2018 15:20:18 +0000 (15:20 +0000)]
[X86] Cleanup SchedWriteFMA classes and use X86SchedWriteWidths directly.

Rename scalar and XMM versions, this is to match/simplify an upcoming change to split MUL/DIV/SQRT scalar/xmm/ymm/zmm classes.

llvm-svn: 331531

6 years ago[bindings/go] Add Go bindings for volatile loads/stores
David Bolvansky [Fri, 4 May 2018 15:14:00 +0000 (15:14 +0000)]
[bindings/go] Add Go bindings for volatile loads/stores

Reviewers: whitequark

Reviewed By: whitequark

Subscribers: llvm-commits

Differential Revision: https://reviews.llvm.org/D46235

llvm-svn: 331530

6 years ago[PPC64] Remove support for ELF V1 ABI in LLD
Zaara Syeda [Fri, 4 May 2018 15:09:49 +0000 (15:09 +0000)]
[PPC64] Remove support for ELF V1 ABI in LLD

The current support for V1 ABI in LLD is incomplete.
This patch removes V1 ABI support and changes the default behavior to V2 ABI,
issuing an error when using the V1 ABI. It also updates the testcases to V2
and removes any V1 specific tests.

Differential Revision: https://reviews.llvm.org/D46316

llvm-svn: 331529

6 years ago[Hexagon] Remove leftover debugging code after r331527
Krzysztof Parzyszek [Fri, 4 May 2018 15:07:30 +0000 (15:07 +0000)]
[Hexagon] Remove leftover debugging code after r331527

llvm-svn: 331528

6 years ago[Hexagon] Handle non-immediate constants in HexagonSplitDouble
Krzysztof Parzyszek [Fri, 4 May 2018 15:04:48 +0000 (15:04 +0000)]
[Hexagon] Handle non-immediate constants in HexagonSplitDouble

llvm-svn: 331527

6 years ago[mips] Correct the predicates of sign extension instructions
Simon Dardis [Fri, 4 May 2018 15:00:54 +0000 (15:00 +0000)]
[mips] Correct the predicates of sign extension instructions

And eliminatw the duplication of those instructions for microMIPS32r6.

Reviewers: smaksimovic, abeserminji, atanasyan

Differential Revision: https://reviews.llvm.org/D46117

llvm-svn: 331526

6 years ago[X86] Add WriteVecMOVMSKY scheduler class
Simon Pilgrim [Fri, 4 May 2018 14:54:33 +0000 (14:54 +0000)]
[X86] Add WriteVecMOVMSKY scheduler class

llvm-svn: 331525

6 years agoComment fix.
David Major [Fri, 4 May 2018 14:41:49 +0000 (14:41 +0000)]
Comment fix.

Test commit!

llvm-svn: 331523

6 years ago[AArch64] Custom Lower MULLH{S,U} for v16i8, v8i16, and v4i32
Adhemerval Zanella [Fri, 4 May 2018 14:33:55 +0000 (14:33 +0000)]
[AArch64] Custom Lower MULLH{S,U} for v16i8, v8i16, and v4i32

This patch adds a custom lowering for ISD::MULH{S,U} used on divide by
constant optimization (DAGCombiner::BuildSDIV and DAGCombiner::BuildUDIV).

New patterns for smull and umull are added, so AArch64ISD::{S,U}MULL
can be correctly lowered to smull2 and umull2.

Reviewed By: SjoerdMeijer

Differential Revision: https://reviews.llvm.org/D46009

llvm-svn: 331522

6 years ago[ELF][AArch64] Add aarch64_elf64_le_vec emulation
Peter Smith [Fri, 4 May 2018 14:28:29 +0000 (14:28 +0000)]
[ELF][AArch64] Add aarch64_elf64_le_vec emulation

Android AOSP has started specifying -m aarch64_elf64_le_vec as supported
by gold and BFD. This is a simple change to add the emulation so that LLD
doesn't immediately error when used as a linker in an AOSP build.

Differential Revision: https://reviews.llvm.org/D46429

llvm-svn: 331521

6 years agoRemove explicit cfg-temporary-dtors=true
Alexander Kornienko [Fri, 4 May 2018 14:13:14 +0000 (14:13 +0000)]
Remove explicit cfg-temporary-dtors=true

Summary:
Remove explicit -analyzer-config cfg-temporary-dtors=true in analyzer tests,
since this option defaults to true since r326461.

Reviewers: NoQ

Reviewed By: NoQ

Subscribers: cfe-commits

Differential Revision: https://reviews.llvm.org/D46393

llvm-svn: 331520

6 years ago[Coroutines] Catch exceptions in await_resume
Brian Gesiak [Fri, 4 May 2018 14:02:37 +0000 (14:02 +0000)]
[Coroutines] Catch exceptions in await_resume

Summary:
http://wg21.link/P0664r2 section "Evolution/Core Issues 24" describes a
proposed change to Coroutines TS that would have any exceptions thrown
after the initial suspend point of a coroutine be caught by the handler
specified by the promise type's 'unhandled_exception' member function.
This commit provides a sample implementation of the specified behavior.

Test Plan: `check-clang`

Reviewers: GorNishanov, EricWF

Reviewed By: GorNishanov

Subscribers: cfe-commits, lewissbaker, eric_niebler

Differential Revision: https://reviews.llvm.org/D45860

llvm-svn: 331519

6 years ago[Hexagon] Skip reserved physical registers when updating liveness
Krzysztof Parzyszek [Fri, 4 May 2018 13:59:05 +0000 (13:59 +0000)]
[Hexagon] Skip reserved physical registers when updating liveness

llvm-svn: 331518

6 years ago[llvm-mca] use colors for warnings and notes generated by InstrBuilder.
Andrea Di Biagio [Fri, 4 May 2018 13:52:12 +0000 (13:52 +0000)]
[llvm-mca] use colors for warnings and notes generated by InstrBuilder.

llvm-svn: 331517

6 years ago[llvm-mca] remove unused argument from method InstrBuilder::createInstrDescImpl.
Andrea Di Biagio [Fri, 4 May 2018 13:10:10 +0000 (13:10 +0000)]
[llvm-mca] remove unused argument from method InstrBuilder::createInstrDescImpl.

We don't need to pass the instruction index to the method that constructs new
instruction descriptors.

No functional change intended.

llvm-svn: 331516

6 years ago[X86] Add SchedWriteFRnd fp rounding scheduler classes
Simon Pilgrim [Fri, 4 May 2018 12:59:24 +0000 (12:59 +0000)]
[X86] Add SchedWriteFRnd fp rounding scheduler classes

Split off from SchedWriteFAdd for fp rounding/bit-manipulation instructions.

Fixes an issue on btver2 which only had the ymm version using the JSTC pipe instead of JFPA.

llvm-svn: 331515

6 years ago[X86] Add test case for PR30290s failing behaviour
Jeremy Morse [Fri, 4 May 2018 10:05:10 +0000 (10:05 +0000)]
[X86] Add test case for PR30290s failing behaviour

Following the advice in review D45022, this currently tests for the broken llc
output where an instruction is mis-scheduled. This test is committed in advance
to improve the eventual fixing patch in D45022, making the bad behaviour that
that patch fixes clearer.

llvm-svn: 331514

6 years ago[ELF][AArch64] Add REQUIRES aarch64 to test [NFC]
Peter Smith [Fri, 4 May 2018 09:10:18 +0000 (09:10 +0000)]
[ELF][AArch64] Add REQUIRES aarch64 to test [NFC]

Forgot to add REQUIRES aarch64 to the test I recently added for D46255.

llvm-svn: 331513

6 years agoWord wrap a test-file comment to 80 columns
Jeremy Morse [Fri, 4 May 2018 08:58:06 +0000 (08:58 +0000)]
Word wrap a test-file comment to 80 columns

This is a test commit to check whether my account works.

llvm-svn: 331512

6 years ago[ELF][AArch64] Implement the AArch64 TLSLD_LDST_LO12 family of relocs
Peter Smith [Fri, 4 May 2018 08:53:34 +0000 (08:53 +0000)]
[ELF][AArch64] Implement the AArch64 TLSLD_LDST_LO12 family of relocs

Implement the following relocations for AArch64:
R_AARCH64_TLSLE_LDST8_TPREL_LO12_NC
R_AARCH64_TLSLE_LDST16_TPREL_LO12_NC
R_AARCH64_TLSLE_LDST32_TPREL_LO12_NC
R_AARCH64_TLSLE_LDST64_TPREL_LO12_NC
R_AARCH64_TLSLE_LDST128_TPREL_LO12_NC

These are specified in ELF for the 64-bit Arm Architecture.

Fixes pr36727

Differential Revision: https://reviews.llvm.org/D46255

llvm-svn: 331511

6 years ago[SelectionDAG] Refactor code by adding RegsForValue::getRegsAndSizes(). NFCI
Bjorn Pettersson [Fri, 4 May 2018 08:50:48 +0000 (08:50 +0000)]
[SelectionDAG] Refactor code by adding RegsForValue::getRegsAndSizes(). NFCI

Summary:
Added a helper method in RegsForValue to get a list with
all the <RegNumber, RegSize> pairs that we want to iterate
over in SelectionDAGBuilder::EmitFuncArgumentDbgValue and
in SelectionDAGBuilder::visitIntrinsicCall.

Reviewers: vsk

Reviewed By: vsk

Subscribers: llvm-commits

Differential Revision: https://reviews.llvm.org/D46360

llvm-svn: 331510

6 years ago[RegUsageInfoCollector] Bugfix for handling of register aliases.
Jonas Paulsson [Fri, 4 May 2018 07:50:05 +0000 (07:50 +0000)]
[RegUsageInfoCollector]  Bugfix for handling of register aliases.

Don't assume the alias of a defined reg is always already in the set.

As the test case in https://bugs.llvm.org/show_bug.cgi?id=36587 discovered,
it is wrong to assume that all the aliases of the defined register in the
*current function* is already present in the UsedPhysRegsMask.

This patch changes this so that any definition in the current function of a
phys-reg always results in all its aliases inserted into the set of defined
registers.

Review: Quentin Colombet
https://reviews.llvm.org/D45157

llvm-svn: 331509

6 years ago[IRCE] Fix misuse of dyn_cast which leads to UB
Max Kazantsev [Fri, 4 May 2018 07:34:35 +0000 (07:34 +0000)]
[IRCE] Fix misuse of dyn_cast which leads to UB

llvm-svn: 331508

6 years ago[XRay][compiler-rt] Support string-based config for Basic mode.
Dean Michael Berris [Fri, 4 May 2018 06:27:53 +0000 (06:27 +0000)]
[XRay][compiler-rt] Support string-based config for Basic mode.

Summary:
This addresses http://llvm.org/PR36790.

This change allows the XRay Basic Mode implementation to use the
string-based initialization routine provided through
`__xray_log_init_mode(...)`. In the process, we've also deprecated some
flags defined for the `XRAY_OPTIONS` environment variable.

We then introduce another environment variable that can control the XRay
Basic Mode implementation through `XRAY_BASIC_OPTIONS`.

We also rename files from `xray_inmemory_log` to `xray_basic_logging` to
be more in line with the mode implementation.

Depends on D46174.

Reviewers: echristo, kpw, pelikan, eizan

Reviewed By: kpw

Subscribers: mgorny, llvm-commits

Differential Revision: https://reviews.llvm.org/D46246

llvm-svn: 331507

6 years ago[XRay][compiler-rt] Support string-based config for FDR mode
Dean Michael Berris [Fri, 4 May 2018 06:13:35 +0000 (06:13 +0000)]
[XRay][compiler-rt] Support string-based config for FDR mode

Summary:
In this chage we add support for the string-based configuration
mechanism for configuring FDR mode.

We deprecate most of the `xray_fdr_log_*` flags that are set with the
`XRAY_OPTIONS` environment variable. Instead we make the FDR
implementation take defaults from the `XRAY_FDR_OPTIONS` environment
variable, and use the flags defined in `xray_fdr_flags.{h,cc,inc}` for
the options we support.

This change addresses http://llvm.org/PR36790.

Depends on D46173.

Reviewers: eizan, pelikan, kpw, echristo

Subscribers: llvm-commits, mgorny

Differential Revision: https://reviews.llvm.org/D46174

llvm-svn: 331506

6 years ago[COFF] Implement the remaining ARM64 relocations
Martin Storsjo [Fri, 4 May 2018 06:06:27 +0000 (06:06 +0000)]
[COFF] Implement the remaining ARM64 relocations

Now only IMAGE_REL_ARM64_ABSOLUTE and IMAGE_REL_ARM64_TOKEN
are unhandled.

Also add range checks for the existing BRANCH26 relocation.

Differential Revision: https://reviews.llvm.org/D46354

llvm-svn: 331505

6 years ago[Driver] Don't warn about unused inputs in config files
Martin Storsjo [Fri, 4 May 2018 06:05:58 +0000 (06:05 +0000)]
[Driver] Don't warn about unused inputs in config files

This avoids warnings about unused linker parameters, just like
other flags are ignored if they're from config files.

Differential Revision: https://reviews.llvm.org/D46286

llvm-svn: 331504

6 years ago[XRay][compiler-rt+docs] Introduce __xray_log_init_mode(...).
Dean Michael Berris [Fri, 4 May 2018 06:01:12 +0000 (06:01 +0000)]
[XRay][compiler-rt+docs] Introduce __xray_log_init_mode(...).

Summary:
This addresses http://llvm.org/PR36790.

The change Deprecates a number of functions and types in
`include/xray/xray_log_interface.h` to recommend using string-based
configuration of XRay through the __xray_log_init_mode(...) function. In
particular, this deprecates the following:

- `__xray_set_log_impl(...)` -- users should instead use the
`__xray_log_register_mode(...)` and `__xray_log_select_mode(...)` APIs.

- `__xray_log_init(...)` -- users should instead use the
`__xray_log_init_mode(...)` function, which also requires using the
`__xray_log_register_mode(...)` and `__xray_log_select_mode(...)`
functionality.

- `__xray::FDRLoggingOptions` -- in following patches, we'll be
migrating the FDR logging implementations (and tests) to use the
string-based configuration. In later stages we'll remove the
`__xray::FDRLoggingOptions` type, and ask users to migrate to using the
string-based configuration mechanism instead.

- `__xray::BasicLoggingOptions` -- same as `__xray::FDRLoggingOptions`,
we'll be removing this type later and instead rely exclusively on the
string-based configuration API.

We also update the documentation to reflect the new advice and remove
some of the deprecated notes.

Reviewers: eizan, kpw, echristo, pelikan

Reviewed By: kpw

Subscribers: llvm-commits

Differential Revision: https://reviews.llvm.org/D46173

llvm-svn: 331503

6 years ago[MachineCSE] Rewrite a loop checking if a block is in a set of blocks without using...
Michael Zolotukhin [Fri, 4 May 2018 01:40:05 +0000 (01:40 +0000)]
[MachineCSE] Rewrite a loop checking if a block is in a set of blocks without using a set. NFC.

Summary:
Using a set is unnecessary here an in some cases (see e.g. PR37277)
takes significant amount of time to just insert values into it. In this
particular case all we need is just to check if we find the block we are
looking for or not.

Reviewers: davide

Subscribers: hiraditya, llvm-commits

Differential Revision: https://reviews.llvm.org/D46411

llvm-svn: 331502

6 years agoAdd children and child[N] properties to SBValue.i.
Jim Ingham [Fri, 4 May 2018 01:31:47 +0000 (01:31 +0000)]
Add children and child[N] properties to SBValue.i.

Also fixed some bad formatting in SBValue.i.

llvm-svn: 331501

6 years ago[LoopIdiomRecognize] Replace more unchecked dyn_casts with cast.
Craig Topper [Fri, 4 May 2018 01:04:28 +0000 (01:04 +0000)]
[LoopIdiomRecognize] Replace more unchecked dyn_casts with cast.

Two of these are immediately dereferenced on the next line. The other two are passed immediately to the IRBuilder constructor which can't handle a nullptr.

llvm-svn: 331500

6 years ago[LoopIdiomRecognize] Use a regular array instead of a SmallVector and explicit ArrayRef.
Craig Topper [Fri, 4 May 2018 01:04:26 +0000 (01:04 +0000)]
[LoopIdiomRecognize] Use a regular array instead of a SmallVector and explicit ArrayRef.

llvm-svn: 331499

6 years ago[LoopIdiomRecognize] Turn two uncheck dyn_casts into regular casts.
Craig Topper [Fri, 4 May 2018 01:04:24 +0000 (01:04 +0000)]
[LoopIdiomRecognize] Turn two uncheck dyn_casts into regular casts.

These are casts on users of a PHINode to Instruction. I think since PHINode is an Instruction any users would also be Instructions. At least a cast will give us an assertion if its wrong.

llvm-svn: 331498

6 years agoThe on-ios-device command line lldb has an optimization where
Jason Molenda [Fri, 4 May 2018 00:59:37 +0000 (00:59 +0000)]
The on-ios-device command line lldb has an optimization where
when it and the inferior process both have the same shared cache
(a conglomeration of all libraries at the same fixed address for
all processes), lldb will read data out of its own memory to speed
things up.  The shared cache has a UUID, so lldb currently checks
that the UUID of its own shared cache matches that of the inferior.

This change adds one refinement to that -- it checks that the UUID
is the same and that the base address of the shared cache is the
same.  And only uses its local shared cache if they are both identical.

This involved using a different style of SPI with dyld to get lldb's
shared cache load address, but it's not especially difficult.

One unattractive part of the change is that I'm using the real
underlying types of task_t and kern_return_t instead of picking
them up from mach/mach.h.  The defines that get picked up there (a
lot from machine.h but others too) conflict with llvm/Support/MachO.h
even when I have mach.h included before our SafeMachO.h which
undefines most of the defines before including llvm/Support/MachO.h.
I'll need to augment the #undefs in SafeMachO.h to get this to
compile cleanly, but that'll be another day.

<rdar://problem/39868238>

llvm-svn: 331497

6 years ago[analyzer] NFC: Remove unused parameteer of StoreManager::CastRetrievedVal().
Artem Dergachev [Fri, 4 May 2018 00:53:41 +0000 (00:53 +0000)]
[analyzer] NFC: Remove unused parameteer of StoreManager::CastRetrievedVal().

llvm-svn: 331496

6 years ago[LoopIdiomRecognize] Add a test case to show incorrect transformation of an infinite...
Craig Topper [Thu, 3 May 2018 23:50:29 +0000 (23:50 +0000)]
[LoopIdiomRecognize] Add a test case to show incorrect transformation of an infinite loop with side effets into a countable loop using ctlz.

We currently recognize this idiom where x is signed and thus the shift in an ashr.

int cnt = 0;
while (x) {
  x >>= 1; // arithmetic shift right
  ++cnt;
}

and turn it into (bitwidth - ctlz(x)). And if there is anything else in the loop we will create a new loop that runs that many times.

If x is initially negative, the shift result will never be 0 and thus the loop is infinite. If you put something with side effects in the loop, that side effect will now only happen bitwidth times instead of an infinite number of times.

So this transform is only safe for logical shift right (which we don't currently recognize) or if we can prove that x cannot be negative before the loop.

llvm-svn: 331493

6 years agoDWARFExpression: Convert file addresses to load addresses early on.
Adrian Prantl [Thu, 3 May 2018 23:32:47 +0000 (23:32 +0000)]
DWARFExpression: Convert file addresses to load addresses early on.

This is a change that only affects Swift and is NFC for the language
plugins on llvm.org. In Swift, we can have global variables with a
location such as DW_OP_addr <addr> DW_OP_deref. The DWARF expression
evaluator doesn't know how to apply a DW_OP_deref to a file address,
but at the very end we convert the file address into a load address.

This patch moves the file->load address conversion to right after the
result of the DW_OP_addr is pushed onto the stack so that a subsequent
DW_OP_deref (and potentially other operations) can be interpreted.

rdar://problem/39767528

Differential revision: https://reviews.llvm.org/D46362

llvm-svn: 331492

6 years agoRevert "Follow-up to r331378. Update tests to allow to use C atomics in C++."
Volodymyr Sapsai [Thu, 3 May 2018 23:06:37 +0000 (23:06 +0000)]
Revert "Follow-up to r331378. Update tests to allow to use C atomics in C++."

It reverts commit r331484 because it caused test failures

    ThreadSanitizer-x86_64 :: Darwin/gcd-groups-destructor.mm
    ThreadSanitizer-x86_64 :: Darwin/libcxx-shared-ptr-stress.mm
    ThreadSanitizer-x86_64 :: Darwin/xpc-race.mm

Foundation.h transitively includes <atomic>, so we have a case of benign mixing
<stdatomic.h> and <atomic>.

llvm-svn: 331491

6 years agoAMDGPU: Make getSubRegFromChannel a static member of AMDGPURegisterInfo
Tom Stellard [Thu, 3 May 2018 22:38:06 +0000 (22:38 +0000)]
AMDGPU: Make getSubRegFromChannel a static member of AMDGPURegisterInfo

Summary:
This makes is possible to have R600RegisterInfo and SIRegisterInfo
not inherit from AMDGPURegisterInfo.

Reviewers: arsenm, nhaehnle

Reviewed By: arsenm

Subscribers: kzhuravl, wdng, yaxunl, dstuttard, tpr, llvm-commits, t-tye

Differential Revision: https://reviews.llvm.org/D46280

llvm-svn: 331490

6 years ago[X86] Add WriteDPPD/WriteDPPS dot product scheduler classes
Simon Pilgrim [Thu, 3 May 2018 22:31:19 +0000 (22:31 +0000)]
[X86] Add WriteDPPD/WriteDPPS dot product scheduler classes

llvm-svn: 331489

6 years ago[X86][Znver1] Use SchedAlias to tag microcoded scheduler classes
Simon Pilgrim [Thu, 3 May 2018 22:12:23 +0000 (22:12 +0000)]
[X86][Znver1] Use SchedAlias to tag microcoded scheduler classes

Avoids extra entries in the class tables.

Found a typo that missed the MMX_PHSUBSW instruction.

llvm-svn: 331488

6 years agoFix include of config.h that was incorrectly changed in r331184
Justin Bogner [Thu, 3 May 2018 21:59:13 +0000 (21:59 +0000)]
Fix include of config.h that was incorrectly changed in r331184

The RWMutex implementation depends on config.h macros (specifically
HAVE_PTHREAD_H and HAVE_PTHREAD_RWLOCK_INIT), so we need to be
including it and not just llvm-config.h here or we fall back to a much
slower implementation.

llvm-svn: 331487

6 years ago[InstCombine] refine select-of-constants to bitwise ops
Sanjay Patel [Thu, 3 May 2018 21:58:44 +0000 (21:58 +0000)]
[InstCombine] refine select-of-constants to bitwise ops

Add logic for the special case when a cmp+select can clearly be
reduced to just a bitwise logic instruction, and remove an
over-reaching chunk of general purpose bit magic. The primary goal
is to remove cases where we are not improving the IR instruction
count when doing these select transforms, and in all cases here that
is true.

In the motivating 3-way compare tests, there are further improvements
because we can combine/propagate select values (not sure if that
belongs in instcombine, but it's there for now).

DAGCombiner has folds to turn some of these selects into bit magic,
so there should be no difference in the end result in those cases.
Not all constant combinations are handled there yet, however, so it
is possible that some targets will see more cmov/csel codegen with
this change in IR canonicalization.

Ideally, we'll go further to *not* turn selects into multiple
logic/math ops in instcombine, and we'll canonicalize to selects.
But we should make sure that this step does not result in regressions
first (and if it does, we should fix those in the backend).

The general direction for this change was discussed here:
http://lists.llvm.org/pipermail/llvm-dev/2016-September/105373.html
http://lists.llvm.org/pipermail/llvm-dev/2017-July/114885.html

Alive proofs for the new bit magic:
https://rise4fun.com/Alive/XG7

Differential Revision: https://reviews.llvm.org/D46086

llvm-svn: 331486

6 years agoGlobalISel: Use a callback to compute constrained reg class for unallocatble registers
Tom Stellard [Thu, 3 May 2018 21:44:16 +0000 (21:44 +0000)]
GlobalISel: Use a callback to compute constrained reg class for unallocatble registers

Summary:
constrainOperandRegClass() currently fails if it tries to constrain the
register class of an operand that is defeined with an unallocatable register
class.  This patch resolves this by adding a target callback to compute
register constriants in this case.

This is required by the AMDGPU because many of its instructions have source opreands
defined with the unallocatable register classe VS_32 which is a union of two allocatable
register classes VGPR_32 and SReg_32.

Reviewers: dsanders, aditya_nandakumar

Reviewed By: aditya_nandakumar

Subscribers: rovka, kristof.beyls, tpr, llvm-commits

Differential Revision: https://reviews.llvm.org/D45991

llvm-svn: 331485

6 years agoFollow-up to r331378. Update tests to allow to use C atomics in C++.
Volodymyr Sapsai [Thu, 3 May 2018 21:30:47 +0000 (21:30 +0000)]
Follow-up to r331378. Update tests to allow to use C atomics in C++.

Reviewers: kubamracek

Reviewed By: kubamracek

Subscribers: cfe-commits

Differential Revision: https://reviews.llvm.org/D46363

llvm-svn: 331484

6 years ago[X86] Make __builtin_ia32_directstore_u32 and __builtin_ia32_movdir64b 'nothrow'
Craig Topper [Thu, 3 May 2018 21:01:35 +0000 (21:01 +0000)]
[X86] Make __builtin_ia32_directstore_u32 and __builtin_ia32_movdir64b 'nothrow'

These builtins snuck in while I was in the middle of adding nothrow to the other builtins in my local clone and I guess I missed them.

llvm-svn: 331483

6 years ago[CodeGenFunction] Use the StringRef::split function that takes a char separator inste...
Craig Topper [Thu, 3 May 2018 21:01:33 +0000 (21:01 +0000)]
[CodeGenFunction] Use the StringRef::split function that takes a char separator instead of StringRef separator. NFC

The char separator version should be a little better optimized.

llvm-svn: 331482

6 years ago[ThinLTO] Add support for optimization remarks to thinBackend
Teresa Johnson [Thu, 3 May 2018 20:24:12 +0000 (20:24 +0000)]
[ThinLTO] Add support for optimization remarks to thinBackend

Summary:
Support was added to the regular LTO backend, but not thinBackend.
This patch adds that support.

Reviewers: pcc, davide

Subscribers: mehdi_amini, inglorion, llvm-commits

Differential Revision: https://reviews.llvm.org/D46376

llvm-svn: 331481

6 years agoRevert "DWARFExpression: Convert file addresses to load addresses early on."
Adrian Prantl [Thu, 3 May 2018 20:19:39 +0000 (20:19 +0000)]
Revert "DWARFExpression: Convert file addresses to load addresses early on."

This reverts commit 331462 while investigating bot breakage.

llvm-svn: 331480

6 years agoAdd back condition that was accidentally removed in r331462.
Adrian Prantl [Thu, 3 May 2018 20:13:58 +0000 (20:13 +0000)]
Add back condition that was accidentally removed in r331462.

This should make the bots much happier.

llvm-svn: 331479

6 years agoAdded ThinLTO test to verify that index files are not generated if thinlto-index...
Rumeet Dhindsa [Thu, 3 May 2018 20:13:38 +0000 (20:13 +0000)]
Added ThinLTO test to verify that index files are not generated if thinlto-index-only is not enabled.

Differential Revision: https://reviews.llvm.org/D46405

llvm-svn: 331478

6 years ago[sanitizer] Remove unused 32-bit allocator TransferBatch parameter
Kostya Kortchinsky [Thu, 3 May 2018 19:14:07 +0000 (19:14 +0000)]
[sanitizer] Remove unused 32-bit allocator TransferBatch parameter

Summary:
NFC. Remove an unused parameter in
`SizeClassAllocator32::TransferBatch::SetFromArray`, and thus get rid of the
compilation warning.

Reviewers: alekseyshl, vitalybuka

Reviewed By: vitalybuka

Subscribers: kubamracek, delcypher, #sanitizers, llvm-commits

Differential Revision: https://reviews.llvm.org/D46397

llvm-svn: 331477

6 years ago[PowerPC] add more FMF debug output; NFC
Sanjay Patel [Thu, 3 May 2018 18:49:35 +0000 (18:49 +0000)]
[PowerPC] add more FMF debug output; NFC

We can't see all of the problems currently unless
we look at debug output when the global 'unsafe' is
on. It's a mess. This is another attempt to make
sure that D45710 is not making changes unintentionally.

llvm-svn: 331476

6 years agoSimplify test clang-tidy-__clang_analyzer__macro.cpp
Zinovy Nis [Thu, 3 May 2018 18:31:39 +0000 (18:31 +0000)]
Simplify test clang-tidy-__clang_analyzer__macro.cpp

llvm-svn: 331475

6 years ago[clang-tidy] Define __clang_analyzer__ macro for clang-tidy for compatibility with...
Zinovy Nis [Thu, 3 May 2018 18:26:39 +0000 (18:26 +0000)]
[clang-tidy] Define __clang_analyzer__ macro for clang-tidy for compatibility with clang static analyzer

This macro is widely used in many well-known projects, ex. Chromium.
But it's not set for clang-tidy, so for ex. DCHECK in Chromium is not considered
as [[no-return]], and a lot of false-positive warnings about nullptr
dereferenced are emitted.

Differential Revision: https://reviews.llvm.org/D46325

llvm-svn: 331474

6 years ago[X86][AVX512] VPLZCNT instructions match SchedWriteVecIMul scheduling class not Sched...
Simon Pilgrim [Thu, 3 May 2018 18:22:49 +0000 (18:22 +0000)]
[X86][AVX512] VPLZCNT instructions match SchedWriteVecIMul scheduling class not SchedWriteVecALU.

llvm-svn: 331473

6 years ago[X86] Split WriteVecShift/WriteVarVecShift into MMX, XMM and YMM/ZMM scheduler classes
Simon Pilgrim [Thu, 3 May 2018 17:56:43 +0000 (17:56 +0000)]
[X86] Split WriteVecShift/WriteVarVecShift into MMX, XMM and YMM/ZMM scheduler classes

This took a bit of extra work as on Intel targets the old (V)PSLLDrr/(V)PSLLDrm style instructions act differently - I ended up creating WriteVecShiftImm classes for XMM/YMM/ZMM vector shift by immediate and retaining WriteVecShift as the default (used only by MMX) plus WriteVecShiftX/WriteVecShiftY. X86SchedWriteWidths hides most of this thank goodness.

llvm-svn: 331472

6 years ago[PowerPC] add tests for FMF propagation; NFC
Sanjay Patel [Thu, 3 May 2018 17:41:37 +0000 (17:41 +0000)]
[PowerPC] add tests for FMF propagation; NFC

I'm choosing PPC out of convenience because it does
all of the transforms of interest in these tests by
default. There are multiple FMF problems shown in the
current checks. D45710 is proposing to fix part of
that.

llvm-svn: 331471

6 years ago[ELF][MIPS] Check that a section has a valid reference to a file in the isMipsPIC...
Simon Atanasyan [Thu, 3 May 2018 17:33:10 +0000 (17:33 +0000)]
[ELF][MIPS] Check that a section has a valid reference to a file in the isMipsPIC routine

llvm-svn: 331470

6 years ago[OPENMP] Fix test typos: CHECK-DAG-N -> CHECK-N-DAG
Joel E. Denny [Thu, 3 May 2018 17:22:04 +0000 (17:22 +0000)]
[OPENMP] Fix test typos: CHECK-DAG-N -> CHECK-N-DAG

Reviewed by: ABataev

Differential Revision: https://reviews.llvm.org/D46370

llvm-svn: 331469

6 years agoRevert r331466: [OPENMP] Fix test typos: CHECK-DAG-N -> CHECK-N-DAG"
Joel E. Denny [Thu, 3 May 2018 17:22:01 +0000 (17:22 +0000)]
Revert r331466: [OPENMP] Fix test typos: CHECK-DAG-N -> CHECK-N-DAG"

Sorry, forgot to add commit log attributes.

llvm-svn: 331468

6 years ago[WebAssembly] Add --stack-first option which places the shadow stack at start of...
Sam Clegg [Thu, 3 May 2018 17:21:53 +0000 (17:21 +0000)]
[WebAssembly] Add --stack-first option which places the shadow stack at start of linear memory

Fixes https://bugs.llvm.org/show_bug.cgi?id=37181

Differential Revision: https://reviews.llvm.org/D46141

llvm-svn: 331467

6 years ago[OPENMP] Fix test typos: CHECK-DAG-N -> CHECK-N-DAG
Joel E. Denny [Thu, 3 May 2018 17:15:44 +0000 (17:15 +0000)]
[OPENMP] Fix test typos: CHECK-DAG-N -> CHECK-N-DAG

llvm-svn: 331466

6 years ago[DebugInfo] Correction for an assert in DIExpression::createFragmentExpression
Bjorn Pettersson [Thu, 3 May 2018 17:04:21 +0000 (17:04 +0000)]
[DebugInfo] Correction for an assert in DIExpression::createFragmentExpression

Summary:
When we create a fragment expression, and there already is an
old fragment expression, we assert that the new fragment is
within the range for the old fragment.

If for example the old fragment expression says that we
describe bit 10-16 of a variable (Offset=10, Size=6),
and we now want to create a new fragment expression only
describing bit 3-6 of the original value, then the resulting
fragment expression should have Offset=13, Size=3.

The assert is supposed to catch if the resulting fragment
expression is outside the range for the old fragment. However,
it used to verify that the Offset+Size of the new fragment was
smaller or equal than Offset+Size for the old fragment. What
we really want to check is that Offset+Size of the new fragment
is smaller than the Size of the old fragment.

Reviewers: aprantl, vsk

Reviewed By: aprantl

Subscribers: davide, llvm-commits, JDevlieghere

Differential Revision: https://reviews.llvm.org/D46391

llvm-svn: 331465

6 years agoReapply "[SelectionDAG] Selection of DBG_VALUE using a PHI node result (pt 2)"
Bjorn Pettersson [Thu, 3 May 2018 17:04:16 +0000 (17:04 +0000)]
Reapply "[SelectionDAG] Selection of DBG_VALUE using a PHI node result (pt 2)"

Summary:
This reverts SVN r331441 (reapplies r331337), together with a fix
in to handle an already existing fragment expression in the
dbg.value that must be fragmented due to a split PHI node.

This should solve the problem seen in PR37321, which was the
reason for the revert of r331337.

The situation in PR37321 is that we have a PHI node like this

   %u.sroa = phi i80 [ %u.sroa.x, %if.x ],
                     [ %u.sroa.y, %if.y ],
                     [ %u.sroa.z, %if.z ]

and a dbg.value like this

  call void @llvm.dbg.value(metadata i80 %u.sroa,
                            metadata !13,
                            metadata !DIExpression(DW_OP_LLVM_fragment, 0, 80))

The phi node is split into three 32-bit PHI nodes

  %30:gr32 = PHI %11:gr32, %bb.4, %14:gr32, %bb.5, %27:gr32, %bb.8
  %31:gr32 = PHI %12:gr32, %bb.4, %15:gr32, %bb.5, %28:gr32, %bb.8
  %32:gr32 = PHI %13:gr32, %bb.4, %16:gr32, %bb.5, %29:gr32, %bb.8

but since the original value only is 80 bits we need to adjust the size
of the last fragment expression, and with this patch we get

  DBG_VALUE debug-use %30:gr32, debug-use $noreg, !"u", !DIExpression(DW_OP_LLVM_fragment, 0, 32)
  DBG_VALUE debug-use %31:gr32, debug-use $noreg, !"u", !DIExpression(DW_OP_LLVM_fragment, 32, 32)
  DBG_VALUE debug-use %32:gr32, debug-use $noreg, !"u", !DIExpression(DW_OP_LLVM_fragment, 64, 16)

Reviewers: vsk, aprantl, mstorsjo

Reviewed By: aprantl

Subscribers: llvm-commits

Differential Revision: https://reviews.llvm.org/D46384

llvm-svn: 331464

6 years ago[CMake] Unify and relayer testing
Jonas Devlieghere [Thu, 3 May 2018 16:54:10 +0000 (16:54 +0000)]
[CMake] Unify and relayer testing

This patch restructures part of LLDB's testing configuration:

1. I moved the test dependencies up the chain so every dotest dependency
   becomes a lit dependency as well. It wouldn't make sense for dotest to
   have other dependencies when it's being run by lit. Lit on the other
   hand can still specify extra dependencies.

2. I replaced as much generator expressions with variables as possible.
   This is consistent with the rest of LLVM and doesn't break generators
   that support multiple targets (MSVC, Xcode). This wasn't a problem
   before, but now we need to expand the dotest arguments in the lit
   configuration and there's only one test suite even with multiple
   targets.

3. I moved lldb-dotest into it's own directory under utils since there's
   no need anymore for it to located under `test/`.

Differential revision: https://reviews.llvm.org/D46334

llvm-svn: 331463

6 years agoDWARFExpression: Convert file addresses to load addresses early on.
Adrian Prantl [Thu, 3 May 2018 16:51:37 +0000 (16:51 +0000)]
DWARFExpression: Convert file addresses to load addresses early on.

This is a change that only affects Swift and is NFC for the language
plugins on llvm.org. In Swift, we can have global variables with a
location such as DW_OP_addr <addr> DW_OP_deref. The DWARF expression
evaluator doesn't know how to apply a DW_OP_deref to a file address,
but at the very end we convert the file address into a load address.

This patch moves the file->load address conversion to right after the
result of the DW_OP_addr is pushed onto the stack so that a subsequent
DW_OP_deref (and potentially other operations) can be interpreted.

rdar://problem/39767528

Differential revision: https://reviews.llvm.org/D46362

llvm-svn: 331462

6 years agoAdded trailing periods.
Alexander Kornienko [Thu, 3 May 2018 16:01:49 +0000 (16:01 +0000)]
Added trailing periods.

llvm-svn: 331461

6 years agoAdd a trailing period in release notes.
Alexander Kornienko [Thu, 3 May 2018 15:59:39 +0000 (15:59 +0000)]
Add a trailing period in release notes.

llvm-svn: 331460

6 years agoAllow writing calling convention attributes on function types.
Aaron Ballman [Thu, 3 May 2018 15:33:50 +0000 (15:33 +0000)]
Allow writing calling convention attributes on function types.

Calling convention attributes notionally appertain to the function type -- they modify the mangling of the function, change the behavior of assignment operations, etc. This commit allows the calling convention attributes to be written in the type position as well as the declaration position.

llvm-svn: 331459

6 years agoRemove the timed_out out-argument from Predicate::WaitForValueEqualTo
Pavel Labath [Thu, 3 May 2018 15:33:41 +0000 (15:33 +0000)]
Remove the timed_out out-argument from Predicate::WaitForValueEqualTo

The function can only return in one of two ways: the Predicate value is
successfully set within the allotted time, or it isn't (the wait times
out). These states can be represented in the return value, and the extra
arg adds no value.

llvm-svn: 331458

6 years ago[clangd] Incorporate #occurrences in scoring code complete results.
Sam McCall [Thu, 3 May 2018 14:53:02 +0000 (14:53 +0000)]
[clangd] Incorporate #occurrences in scoring code complete results.

Summary: needs tests

Reviewers: ilya-biryukov

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

Differential Revision: https://reviews.llvm.org/D46183

llvm-svn: 331457

6 years ago[clang-tidy] Remove AnalyzeTemporaryDtors option.
Alexander Kornienko [Thu, 3 May 2018 14:40:37 +0000 (14:40 +0000)]
[clang-tidy] Remove AnalyzeTemporaryDtors option.

Remove the `AnalyzeTemporaryDtors` option, since the corresponding
`cfg-temporary-dtors` option of the Static Analyzer defaults to `true` since
r326461.

llvm-svn: 331456

6 years agouse LLVM's standard CMakeLists.txt layout for llvm-xray
Nico Weber [Thu, 3 May 2018 14:25:57 +0000 (14:25 +0000)]
use LLVM's standard CMakeLists.txt layout for llvm-xray

llvm-svn: 331455

6 years ago[CodeGen][X86][NFC] Copy two selectcc tests from AArch64.
Roman Lebedev [Thu, 3 May 2018 13:33:07 +0000 (13:33 +0000)]
[CodeGen][X86][NFC] Copy two selectcc tests from AArch64.

These tests are for DAGCombiner::foldSelectCCToShiftAnd().
Right now, they were only tested for AArch64,
but given the upcoming X86 changes to the hasAndNot(),
the test coverage needs to be added.

These tests originated from D27489 / rL289738

llvm-svn: 331454

6 years ago[X86] Split WriteVecALU/WritePHAdd into XMM and YMM/ZMM scheduler classes
Simon Pilgrim [Thu, 3 May 2018 13:27:10 +0000 (13:27 +0000)]
[X86] Split WriteVecALU/WritePHAdd into XMM and YMM/ZMM scheduler classes

llvm-svn: 331453

6 years ago[LLD][AArch64] Simplify relocations sharing same encoding [NFC]
Peter Smith [Thu, 3 May 2018 12:59:52 +0000 (12:59 +0000)]
[LLD][AArch64] Simplify relocations sharing same encoding [NFC]

The code to encode the result in relocateOne for the relocations:
R_AARCH64_LD64_GOT_LO12_NC
R_AARCH64_TLSIE_LD64_GOTTPREL_LO12_NC
R_AARCH64_TLSDESC_LD64_LO12

is equivalent to that for R_AARCH64_LDST64_ABS_LO12_NC. This is described
in the ABI as "Set the LD/ST immediate field bits [11:3] of X. No overflow
check; check that X&7 =0.

Differential Revision: https://reviews.llvm.org/D46247

llvm-svn: 331452

6 years agoARM: don't try to over-align large vectors as arguments.
Tim Northover [Thu, 3 May 2018 12:54:25 +0000 (12:54 +0000)]
ARM: don't try to over-align large vectors as arguments.

By default LLVM thinks very large vectors get aligned to their size when
passed across functions. Unfortunately no-one told the ARM backend so it
doesn't trigger stack realignment and so accesses can cause the usual
misalignment issues (e.g. a data abort).

This changes the ABI alignment to the stack alignment, which in practice
(and as a bonus) also coincides with the alignment "natural" vectors get.

llvm-svn: 331451

6 years agoFix test failure for missing _LIBCPP_ENABLE_CXX17_REMOVED_UNEXPECTED_FUNCTIONS
Nico Weber [Thu, 3 May 2018 12:44:27 +0000 (12:44 +0000)]
Fix test failure for missing _LIBCPP_ENABLE_CXX17_REMOVED_UNEXPECTED_FUNCTIONS

This is a follow-up change to r331150. The CL moved the macro from individual
file to build file, but the macro is missed in a test config file.

https://reviews.llvm.org/D46385
Patch from Taiju Tsuiki <tzik@chromium.org>!

llvm-svn: 331450

6 years agoperform DSE through launder.invariant.group
Piotr Padlewski [Thu, 3 May 2018 11:03:53 +0000 (11:03 +0000)]
perform DSE through launder.invariant.group

Summary:
Alias Analysis knows that llvm.launder.invariant.group
returns pointer that mustalias argument, but this information
wasn't used, therefor we didn't DSE through launder.invariant.group

Reviewers: chandlerc, dberlin, bogner, hfinkel, efriedma

Reviewed By: dberlin

Subscribers: amharc, llvm-commits, nlewycky, rsmith

Differential Revision: https://reviews.llvm.org/D31581

llvm-svn: 331449

6 years agoRename invariant.group.barrier to launder.invariant.group
Piotr Padlewski [Thu, 3 May 2018 11:03:01 +0000 (11:03 +0000)]
Rename invariant.group.barrier to launder.invariant.group

Summary:
This is one of the initial commit of "RFC: Devirtualization v2" proposal:
https://docs.google.com/document/d/16GVtCpzK8sIHNc2qZz6RN8amICNBtvjWUod2SujZVEo/edit?usp=sharing

Reviewers: rsmith, amharc, kuhar, sanjoy

Subscribers: arsenm, nhaehnle, javed.absar, hiraditya, llvm-commits

Differential Revision: https://reviews.llvm.org/D45111

llvm-svn: 331448