OSDN Git Service

android-x86/external-llvm.git
7 years agoCleanup tests to not share a DISubprogram between multiple Functions.
Adrian Prantl [Thu, 4 May 2017 16:24:31 +0000 (16:24 +0000)]
Cleanup tests to not share a DISubprogram between multiple Functions.

rdar://problem/31926379

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@302166 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[test-release] Status update *before* long gzip
Renato Golin [Thu, 4 May 2017 16:21:30 +0000 (16:21 +0000)]
[test-release] Status update *before* long gzip

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@302165 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[globalisel][tablegen] Add several GINodeEquiv's for operators that do not require...
Daniel Sanders [Thu, 4 May 2017 14:24:50 +0000 (14:24 +0000)]
[globalisel][tablegen] Add several GINodeEquiv's for operators that do not require additional support.

Summary:
As of this patch, 350 out of 3938 rules are currently imported.

Depends on D32229

Reviewers: qcolombet, kristof.beyls, rovka, t.p.northover, ab, aditya_nandakumar

Reviewed By: ab

Subscribers: dberris, llvm-commits, igorb

Differential Revision: https://reviews.llvm.org/D32275

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@302154 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[DAGCombine] Transform (fadd A, (fmul B, -2.0)) -> (fsub A, (fadd B, B)).
Chad Rosier [Thu, 4 May 2017 14:14:44 +0000 (14:14 +0000)]
[DAGCombine] Transform (fadd A, (fmul B, -2.0)) -> (fsub A, (fadd B, B)).

Differential Revision: http://reviews.llvm.org/D32596

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@302153 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[X86][AVX512] Fix VPABSD file checks
Simon Pilgrim [Thu, 4 May 2017 13:42:57 +0000 (13:42 +0000)]
[X86][AVX512] Fix VPABSD file checks

Fix capitalization and string matching

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@302150 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[X86][SSE] Add i686 triple tests for partial vector and re-association
Simon Pilgrim [Thu, 4 May 2017 13:35:40 +0000 (13:35 +0000)]
[X86][SSE] Add i686 triple tests for partial vector and re-association

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@302149 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoRefactoring with range-based for, NFC
Krzysztof Parzyszek [Thu, 4 May 2017 13:35:17 +0000 (13:35 +0000)]
Refactoring with range-based for, NFC

Patch by Wei-Ren Chen.

Differential Revision: https://reviews.llvm.org/D32682

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@302148 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[SystemZ] Make copyPhysReg() add impl-use operands of super reg.
Jonas Paulsson [Thu, 4 May 2017 13:33:30 +0000 (13:33 +0000)]
[SystemZ]  Make copyPhysReg() add impl-use operands of super reg.

When a 128 bit COPY is lowered into two instructions, an impl-use operand of
the super-reg should be added to each new instruction in case one of the
sub-regs is undefined.

Review: Ulrich Weigand

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@302146 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[X86][SSE] Add i686 triple tests for PBLENDW commutation
Simon Pilgrim [Thu, 4 May 2017 13:08:09 +0000 (13:08 +0000)]
[X86][SSE] Add i686 triple tests for PBLENDW commutation

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@302145 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[X86][AVX1] Regenerate checks and add i686 triple tests for folded logical ops
Simon Pilgrim [Thu, 4 May 2017 13:00:30 +0000 (13:00 +0000)]
[X86][AVX1] Regenerate checks and add i686 triple tests for folded logical ops

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@302144 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoRevert "IR: Use pointers instead of GUIDs to represent edges in the module summary...
Eric Liu [Thu, 4 May 2017 11:49:39 +0000 (11:49 +0000)]
Revert "IR: Use pointers instead of GUIDs to represent edges in the module summary. NFCI."

This reverts commit r302108. This causes crash in clang bootstrap with LTO.

Contacted the auther in the original commit.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@302140 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[mips][XRay] Use the base version of emitXRayTable
Simon Dardis [Thu, 4 May 2017 11:03:50 +0000 (11:03 +0000)]
[mips][XRay] Use the base version of emitXRayTable

Follow up rL290858 by removing the MIPS specific version of XRayTable
emission in favour of the basic version.

This resolves a buildbot failure where the ELF sections were malformed
causing the linker to reject the object files with xray related sections.

Reviewers: dberris, slthakur

Differential Revision: https://reviews.llvm.org/D32808

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@302138 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[ArgPromotion] Fix a truncated variable
Martin Storsjo [Thu, 4 May 2017 10:54:35 +0000 (10:54 +0000)]
[ArgPromotion] Fix a truncated variable

This fixes a regression since SVN rev 273808 (which was supposed to
not change functionality).

The regression caused miscompilations (noted in the wild when targeting
AArch64) on platforms with 32 bit long.

Differential Revision: https://reviews.llvm.org/D32850

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@302137 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[LLVM][inline-asm][Altmacor] Altmacro string delimiter '<..>'
Michael Zuckerman [Thu, 4 May 2017 10:37:00 +0000 (10:37 +0000)]
[LLVM][inline-asm][Altmacor] Altmacro string delimiter '<..>'

In this patch, I introduce a new altmacro string delimiter.
This review is the second review in a series of four reviews.
(one for each altmacro feature: LOCAL, string delimiter, string '!' escape sign and absolute expression as a string '%' ).

In the alternate macro mode, you can delimit strings with matching angle brackets <..>
when using it as a part of calling macro arguments.

As described in the https://sourceware.org/binutils/docs-2.27/as/Altmacro.html
"<string>
You can delimit strings with matching angle brackets."

assumptions:

1. If an argument begins with '<' and ends with '>'. The argument is considered as a string.
2. Except adding new string mark '<..>', a regular macro behavior is expected.
3. The altmacro cannot affect the regular less/greater behavior.
4. If a comma is present inside an angle brackets it considered as a character and not as a separator.

Differential Revision: https://reviews.llvm.org/D32701

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@302135 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[X86][AVX-512] Allow EVEX encoded instruction selection when available for mul v8i32.
Igor Breger [Thu, 4 May 2017 07:34:58 +0000 (07:34 +0000)]
[X86][AVX-512] Allow EVEX encoded instruction selection when available for mul v8i32.

Differential Revision: https://reviews.llvm.org/D32679

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@302127 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[ARM] ACLE Chapter 9 intrinsics
Sam Parker [Thu, 4 May 2017 07:31:28 +0000 (07:31 +0000)]
[ARM] ACLE Chapter 9 intrinsics

Added the integer data processing intrinsics from ACLE v2.1 Chapter 9
but I have missed out the saturation_occurred intrinsics for now. For
the instructions that read and write the GE bits, a chain is included
and the only instruction that reads these flags (sel) is only
selectable via the implemented intrinsic.

Differential Revision: https://reviews.llvm.org/D32281

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@302126 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[X86] Disabling PLT in Regcall CC Functions
Oren Ben Simhon [Thu, 4 May 2017 07:22:49 +0000 (07:22 +0000)]
[X86] Disabling PLT in Regcall CC Functions

According to psABI, PLT stub clobbers XMM8-XMM15.
In Regcall calling convention those registers are used for passing parameters.
Thus we need to prevent lazy binding in Regcall.

Differential Revision: https://reviews.llvm.org/D32430

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@302124 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[AVX-512VL] Autogenerate checks. Add --show-mc-encoding to check instruction predicate.
Igor Breger [Thu, 4 May 2017 06:53:31 +0000 (06:53 +0000)]
[AVX-512VL] Autogenerate checks. Add --show-mc-encoding to check instruction predicate.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@302123 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[AVX] Fix vpcmpeqq predicate.
Igor Breger [Thu, 4 May 2017 06:24:52 +0000 (06:24 +0000)]
[AVX] Fix vpcmpeqq predicate.

Summary:
Fix vpcmpeqq predicate. AVX512 version of vpcmpeqq is not equivalent to AVX one.
Split from https://reviews.llvm.org/D32679

Reviewers: craig.topper, zvi, aymanmus

Reviewed By: craig.topper

Subscribers: llvm-commits

Differential Revision: https://reviews.llvm.org/D32786

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@302119 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoUse right function in LoopVectorize.
Jonas Paulsson [Thu, 4 May 2017 05:31:56 +0000 (05:31 +0000)]
Use right function in LoopVectorize.

-    unsigned AS = getMemInstAlignment(I);
+    unsigned AS = getMemInstAddressSpace(I);

Review: Hal Finkel

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@302114 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[XRay] Use wordsize-dependent alignment for index
Dean Michael Berris [Thu, 4 May 2017 04:55:46 +0000 (04:55 +0000)]
[XRay] Use wordsize-dependent alignment for index

This makes it simpler for the runtime to consistently handle the entries
in the function sled index in both 32 and 64 bit platforms where the
XRay runtime works.

Follow-up on D32693.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@302111 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[SelectionDAG] Improve known bits support for CTPOP.
Craig Topper [Thu, 4 May 2017 04:33:27 +0000 (04:33 +0000)]
[SelectionDAG] Improve known bits support for CTPOP.

This is based on the same concept from ValueTracking's version of computeKnownBits.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@302110 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[XRay] Create an Index of sleds per function
Dean Michael Berris [Thu, 4 May 2017 03:37:57 +0000 (03:37 +0000)]
[XRay] Create an Index of sleds per function

Summary:
This change adds a new section to the xray-instrumented binary that
stores an index into ranges of the instrumentation map, where sleds
associated with the same function can be accessed as an array. At
runtime, we can get access to this index by function ID offset allowing
for selective patching and unpatching by function ID.

Each entry in this new section (xray_fn_idx) will include two pointers
indicating the start and one past the end of the sleds associated with
the same function. These entries will be 16 bytes long on x86 and
aarch64. On arm, we align to 16 bytes anyway so the runtime has to take
that into consideration.

__{start,stop}_xray_fn_idx will be the symbols that the runtime will
look for when we implement the selective patching/unpatching by function
id APIs. Because XRay synthesizes the function id's in a monotonically
increasing manner at runtime now, implementations (and users) can use
this table to look up the sleds associated with a specific function.
This is useful in implementations that want to do things like:

  - Implement coverage mode for functions by patching everything
    pre-main, then as functions are encountered, the installed handler
    can unpatch the function that's been encountered after recording
    that it's been called.
  - Do "learning mode", so that the implementation can figure out some
    statistical information about function calls by function id for a
    time being, and then determine which functions are worth
    uninstrumenting at runtime.
  - Do "selective instrumentation" where an implementation can
    specifically instrument only certain function id's at runtime
    (either based on some external data, or through some other
    heuristics) instead of patching all the instrumented functions at
    runtime.

Reviewers: dblaikie, echristo, chandlerc, javed.absar

Subscribers: pelikan, aemerson, kpw, llvm-commits, rengolin

Differential Revision: https://reviews.llvm.org/D32693

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@302109 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoIR: Use pointers instead of GUIDs to represent edges in the module summary. NFCI.
Peter Collingbourne [Thu, 4 May 2017 03:36:16 +0000 (03:36 +0000)]
IR: Use pointers instead of GUIDs to represent edges in the module summary. NFCI.

When profiling a no-op incremental link of Chromium I found that the functions
computeImportForFunction and computeDeadSymbols were consuming roughly 10% of
the profile. The goal of this change is to improve the performance of those
functions by changing the map lookups that they were previously doing into
pointer dereferences.

This is achieved by changing the ValueInfo data structure to be a pointer to
an element of the global value map owned by ModuleSummaryIndex, and changing
reference lists in the GlobalValueSummary to hold ValueInfos instead of GUIDs.
This means that a ValueInfo will take a client directly to the summary list
for a given GUID.

Differential Revision: https://reviews.llvm.org/D32471

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@302108 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agollvm/IR/Function.h: Prune an obsolete @param in r302060. [-Wdocumentation]
NAKAMURA Takumi [Thu, 4 May 2017 02:01:18 +0000 (02:01 +0000)]
llvm/IR/Function.h: Prune an obsolete @param in r302060. [-Wdocumentation]

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@302106 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[XRay] Detect loops in functions being lowered
Dean Michael Berris [Thu, 4 May 2017 01:24:26 +0000 (01:24 +0000)]
[XRay] Detect loops in functions being lowered

Summary:
This is an implementation of the loop detection logic that XRay needs to
determine whether a function might take time at runtime. Without this
heuristic, XRay will tend to not instrument short functions that have
loops that might have runtime dependent on inputs or external values.

While this implementation doesn't do any further analysis than just
figuring out whether there is a loop in the MachineFunction being
code-gen'ed, we're paving the way for being able to perform more
sophisticated analysis of the function in the future (for example to
determine whether the trip count for the loop might be constant, and
make a decision on that instead). This enables us to cover more
functions with the default heuristics, and potentially identify ones
that have variable runtime latency just by looking for the presence of
loops.

Reviewers: chandlerc, rnk, pelikan

Subscribers: llvm-commits

Differential Revision: https://reviews.llvm.org/D32274

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@302103 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[SCEV] createAddRecFromPHI: Optimize for the most common case.
Michael Zolotukhin [Wed, 3 May 2017 23:53:38 +0000 (23:53 +0000)]
[SCEV] createAddRecFromPHI: Optimize for the most common case.

Summary:
The existing implementation creates a symbolic SCEV expression every
time we analyze a phi node and then has to remove it, when the analysis
is finished. This is very expensive, and in most of the cases it's also
unnecessary. According to the data I collected, ~60-70% of analyzed phi
nodes (measured on SPEC) have the following form:
  PN = phi(Start, OP(Self, Constant))
Handling such cases separately significantly speeds this up.

Reviewers: sanjoy, pete

Subscribers: llvm-commits

Differential Revision: https://reviews.llvm.org/D32663

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@302096 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agostrlen-1.ll: Fix test
Matthias Braun [Wed, 3 May 2017 23:32:51 +0000 (23:32 +0000)]
strlen-1.ll: Fix test

Change test for `strlen(x) == 0 --> *x == 0` to actually test the
pattern.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@302094 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[InstCombine][KnownBits] Use KnownBits better to detect nsw adds
Craig Topper [Wed, 3 May 2017 23:22:46 +0000 (23:22 +0000)]
[InstCombine][KnownBits] Use KnownBits better to detect nsw adds

Change checkRippleForAdd from a heuristic to a full check -
if it is provable that the add does not overflow return true, otherwise false.

Patch by Yoav Ben-Shalom

Differential Revision: https://reviews.llvm.org/D32686

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@302093 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoMark functions as not having CFI once we finalize an x86 stack frame
Reid Kleckner [Wed, 3 May 2017 23:13:42 +0000 (23:13 +0000)]
Mark functions as not having CFI once we finalize an x86 stack frame

We'll set it back to true in emitPrologue if it gets called. It doesn't
get called for naked functions.

Fixes PR32912

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@302092 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[KnownBits] Add methods for determining if KnownBits is a constant value
Craig Topper [Wed, 3 May 2017 23:12:29 +0000 (23:12 +0000)]
[KnownBits] Add methods for determining if KnownBits is a constant value

This patch adds isConstant and getConstant for determining if KnownBits represents a constant value and to retrieve the value. Use them to simplify code.

Differential Revision: https://reviews.llvm.org/D32785

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@302091 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[ValueTracking] Remove handling for BitWidth being 0 in ComputeSignBit and isKnownNon...
Craig Topper [Wed, 3 May 2017 22:25:19 +0000 (22:25 +0000)]
[ValueTracking] Remove handling for BitWidth being 0 in ComputeSignBit and isKnownNonZero.

I don't believe its possible to have non-zero values here since DataLayout became required. The APInt constructor inside of the KnownBits object will assert if this ever happens.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@302089 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[KnownBits] Add zext, sext, and trunc methods to KnownBits
Craig Topper [Wed, 3 May 2017 22:07:25 +0000 (22:07 +0000)]
[KnownBits] Add zext, sext, and trunc methods to KnownBits

This patch adds zext, sext, and trunc methods to KnownBits and uses them where possible.

Differential Revision: https://reviews.llvm.org/D32784

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@302088 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[TargetLowering] use isSubsetOf in SimplifyDemandedBits; NFCI
Sanjay Patel [Wed, 3 May 2017 21:55:34 +0000 (21:55 +0000)]
[TargetLowering] use isSubsetOf in SimplifyDemandedBits; NFCI

This is the DAG equivalent of https://reviews.llvm.org/D32255 ,
which will hopefully be committed again. The functionality
(preferring a 'not' op) is already here in the DAG, so this is
just intended to be a clean-up and performance improvement.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@302087 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoclang-format and restyle DWARFFormValue before working on it. NFC
Paul Robinson [Wed, 3 May 2017 21:53:21 +0000 (21:53 +0000)]
clang-format and restyle DWARFFormValue before working on it. NFC

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@302086 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoDebugInfo: elide type index entries for synthetic types
Saleem Abdulrasool [Wed, 3 May 2017 21:39:01 +0000 (21:39 +0000)]
DebugInfo: elide type index entries for synthetic types

Compiler emitted synthetic types may not have an associated DIFile
(translation unit).  In such a case, when generating CodeView debug type
information, we would attempt to compute an absolute filepath which
would result in a segfault due to a NULL DIFile*.  If there is no source
file associated with the type, elide the type index entry for the type
and record the type information.  This actually results in higher
fidelity debug information than clang/C2 as of this writing.

Resolves PR32668!

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@302085 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[AArch64] Fix variable name ambiguity in r302078.
Ahmed Bougacha [Wed, 3 May 2017 20:51:34 +0000 (20:51 +0000)]
[AArch64] Fix variable name ambiguity in r302078.

ArchKind is passed to the function, but it's also a type.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@302081 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[TailCallElim] Remove an unused argument. NFCI
Xin Tong [Wed, 3 May 2017 20:37:07 +0000 (20:37 +0000)]
[TailCallElim] Remove an unused argument. NFCI

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@302080 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[AArch64] Make the TargetParser add CPU exts provided by the arch.
Ahmed Bougacha [Wed, 3 May 2017 20:33:58 +0000 (20:33 +0000)]
[AArch64] Make the TargetParser add CPU exts provided by the arch.

Otherwise, each CPU has to manually specify the extensions it supports,
even though they have to be a superset of the base arch extensions.
And when there's redundant data there's stale data, so most of the CPUs
lie about the features they support (almost none lists AEK_FP).

Instead, do the saner thing: add the optional extensions on top of the
base extensions provided by the architecture.

The ARM TargetParser has the same behavior.

Differential Revision: https://reviews.llvm.org/D32780

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@302078 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[AArch64] armv8-A doesn't have CRC.
Ahmed Bougacha [Wed, 3 May 2017 20:33:52 +0000 (20:33 +0000)]
[AArch64] armv8-A doesn't have CRC.

That's only a required extension as of v8.1a.

Remove it from the "generic" CPU as well: it should only support the
base ISA (and binutils agrees).

Also unify the MC tests into crc.s and arm64-crc32.s

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@302077 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[libFuzzer] exit without running atexit handlers in libfuzzer's crash handler
Vitaly Buka [Wed, 3 May 2017 20:31:19 +0000 (20:31 +0000)]
[libFuzzer] exit without running atexit handlers in libfuzzer's crash handler

Summary:
It's not safe to assume that atexit handlers can be run once the app crashed.

Patch by Jochen Eisinger.

Reviewers: kcc, vitalybuka

Subscribers: llvm-commits

Differential Revision: https://reviews.llvm.org/D32640

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@302076 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[Hexagon] Use automatically-generated scheduling information for HVX
Krzysztof Parzyszek [Wed, 3 May 2017 20:10:36 +0000 (20:10 +0000)]
[Hexagon] Use automatically-generated scheduling information for HVX

Patch by Jyotsna Verma.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@302073 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoRemove unused private field.
Zachary Turner [Wed, 3 May 2017 19:42:06 +0000 (19:42 +0000)]
Remove unused private field.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@302069 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoAvoid warning of unused variable in release builds. NFC
Anna Thomas [Wed, 3 May 2017 19:25:04 +0000 (19:25 +0000)]
Avoid warning of unused variable in release builds. NFC

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@302068 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoFix typos in comment
Sanjoy Das [Wed, 3 May 2017 18:29:34 +0000 (18:29 +0000)]
Fix typos in comment

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@302063 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoBreak verification down into smaller functions to keep code clean.
Greg Clayton [Wed, 3 May 2017 18:25:46 +0000 (18:25 +0000)]
Break verification down into smaller functions to keep code clean.

Adrian requested that we break things down to make things clean in the DWARFVerifier. This patch breaks everything down into nice individual functions and cleans up the code quite a bit and prepares us for the next round of verifiers.

Differential Revision: https://reviews.llvm.org/D32812

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@302062 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoFix PPC64 warning for missing parantheses. NFC.
Anna Thomas [Wed, 3 May 2017 18:25:43 +0000 (18:25 +0000)]
Fix PPC64 warning for missing parantheses. NFC.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@302061 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[IR] Abstract away ArgNo+1 attribute indexing as much as possible
Reid Kleckner [Wed, 3 May 2017 18:17:31 +0000 (18:17 +0000)]
[IR] Abstract away ArgNo+1 attribute indexing as much as possible

Summary:
Do three things to help with that:
- Add AttributeList::FirstArgIndex, which is an enumerator currently set
  to 1. It allows us to change the indexing scheme with fewer changes.
- Add addParamAttr/removeParamAttr. This just shortens addAttribute call
  sites that would otherwise need to spell out FirstArgIndex.
- Remove some attribute-specific getters and setters from Function that
  take attribute list indices.  Most of these were only used from
  BuildLibCalls, and doesNotAlias was only used to test or set if the
  return value is malloc-like.

I'm happy to split the patch, but I think they are probably easier to
review when taken together.

This patch should be NFC, but it sets the stage to change the indexing
scheme to this, which is more convenient when indexing into an array:
  0: func attrs
  1: retattrs
  2...: arg attrs

Reviewers: chandlerc, pete, javed.absar

Subscribers: david2050, llvm-commits

Differential Revision: https://reviews.llvm.org/D32811

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@302060 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[CodeView] Remove constructor initialization of a removed field.
Davide Italiano [Wed, 3 May 2017 18:02:46 +0000 (18:02 +0000)]
[CodeView] Remove constructor initialization of a removed field.

I should've staged this with my last commit.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@302059 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[RuntimeLoopUnroller] Add assert that we dont unroll non-rotated loops
Anna Thomas [Wed, 3 May 2017 17:43:59 +0000 (17:43 +0000)]
[RuntimeLoopUnroller] Add assert that we dont unroll non-rotated loops

Summary:
Cloning basic blocks in the loop for runtime loop unroller depends on loop being
in rotated form (i.e. loop latch target is the exit block).
Assert that this is true, so that callers of runtime loop unroller pass in
canonical loops.
The single caller of this function has that check recently added:
https://reviews.llvm.org/rL301239

Reviewers: davide

Subscribers: llvm-commits

Differential Revision: https://reviews.llvm.org/D32801

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@302058 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoDelete dead function causing compilation failure.
Zachary Turner [Wed, 3 May 2017 17:38:49 +0000 (17:38 +0000)]
Delete dead function causing compilation failure.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@302057 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[CodeView] Remove unused private field. NFCI.
Davide Italiano [Wed, 3 May 2017 17:37:35 +0000 (17:37 +0000)]
[CodeView] Remove unused private field. NFCI.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@302056 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[bpf] add relocation support
Alexei Starovoitov [Wed, 3 May 2017 17:30:56 +0000 (17:30 +0000)]
[bpf] add relocation support

  . there should be no runtime relocation inside the bpf function.
  . relocation supported here mostly for debugging.
  . a test case is added.

Signed-off-by: Yonghong Song <yhs@fb.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@302055 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoRevert "CMake: Move sphinx detection into AddSphinxTarget.cmake"
Tom Stellard [Wed, 3 May 2017 17:22:23 +0000 (17:22 +0000)]
Revert "CMake: Move sphinx detection into AddSphinxTarget.cmake"

This reverts commit r302025.

clang and lld need to be updated too so they don't break with this patch.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@302054 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[CodeView] Use actual strings for dealing with checksums and lines.
Zachary Turner [Wed, 3 May 2017 17:11:40 +0000 (17:11 +0000)]
[CodeView] Use actual strings for dealing with checksums and lines.

The raw CodeView format references strings by "offsets", but it's
confusing what table the offset refers to.  In the case of line
number information, it's an offset into a buffer of records,
and an indirection is required to get another offset into a
different table to find the final string.  And in the case of
checksum information, there is no indirection, and the offset
refers directly to the location of the string in another buffer.

This would be less confusing if we always just referred to the
strings by their value, and have the library be smart enough
to correctly resolve the offsets on its own from the right
location.

This patch makes that possible.  When either reading or writing,
all the user deals with are strings, and the library does the
appropriate translations behind the scenes.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@302053 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[llvm-readobj] Update readobj to re-use parsing code.
Zachary Turner [Wed, 3 May 2017 17:11:11 +0000 (17:11 +0000)]
[llvm-readobj] Update readobj to re-use parsing code.

llvm-readobj hand rolls some CodeView parsing code for string
tables, so this patch updates it to re-use some of the newly
introduced parsing code in LLVMDebugInfoCodeView.

Differential Revision: https://reviews.llvm.org/D32772

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@302052 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoARM: add extra test for addrmode folding.
Tim Northover [Wed, 3 May 2017 16:54:30 +0000 (16:54 +0000)]
ARM: add extra test for addrmode folding.

I was worried we might replace a mul with a mul+shift even if there were later
uses. Turns out to be unfounded but I'd just as well add an actual test for it.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@302051 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[X86][LWP] Add stack folding mappings and tests for LWPINS/LWPVAL instructions
Simon Pilgrim [Wed, 3 May 2017 16:46:30 +0000 (16:46 +0000)]
[X86][LWP] Add stack folding mappings and tests for LWPINS/LWPVAL instructions

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@302049 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoSilence a 'enum and non-enum used in conditional' warning.
Simon Pilgrim [Wed, 3 May 2017 16:43:57 +0000 (16:43 +0000)]
Silence a 'enum and non-enum used in conditional' warning.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@302048 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[DAGCombine] (addcarry (add|uaddo X, Y), 0, Carry) -> (addcarry X, Y, Carry)
Amaury Sechet [Wed, 3 May 2017 16:28:10 +0000 (16:28 +0000)]
[DAGCombine] (addcarry (add|uaddo X, Y), 0, Carry) -> (addcarry X, Y, Carry)

Summary: Do the transform when the carry isn't used. It's a pattern exposed when legalizing large integers.

Reviewers: jyknight, nemanjai, mkuper, spatel, RKSimon, zvi, bkramer

Subscribers: llvm-commits

Differential Revision: https://reviews.llvm.org/D32755

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@302047 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoCreate DWARFVerifier.cpp and .h and move all DWARF verification code over into it.
Greg Clayton [Wed, 3 May 2017 16:02:29 +0000 (16:02 +0000)]
Create DWARFVerifier.cpp and .h and move all DWARF verification code over into it.

Adrian requested we create a DWARFVerifier.cpp file to contain all of the DWARF verification stuff. This change simply moves the functionality over into DWARFVerifier.h and DWARFVerifier.cpp, renames the DWARFVerifier methods to start with lower case, and switches DWARFContext.cpp over to using the new functionality.

Differential Revision: https://reviews.llvm.org/D32809

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@302044 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoResubmit r301986 and r301987 "Add codeview::StringTable"
Zachary Turner [Wed, 3 May 2017 15:58:37 +0000 (15:58 +0000)]
Resubmit r301986 and r301987 "Add codeview::StringTable"

This was reverted due to a "missing" file, but in reality
what happened was that I renamed a file, and then due to
a merge conflict both the old file and the new file got
added to the repository.  This led to an unused cpp file
being in the repo and not referenced by any CMakeLists.txt
but #including a .h file that wasn't in the repo.  In an
even more unfortunate coincidence, CMake didn't report the
unused cpp file because it was in a subdirectory of the
folder with the CMakeLists.txt, and not in the same directory
as any CMakeLists.txt.

The presence of the unused file was then breaking certain
tools that determine file lists by globbing rather than
by what's specified in CMakeLists.txt

In any case, the fix is to just remove the unused file from
the patch set.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@302042 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[X86][LWP] Add llvm support for LWP instructions (reapplied).
Simon Pilgrim [Wed, 3 May 2017 15:51:39 +0000 (15:51 +0000)]
[X86][LWP] Add llvm support for LWP instructions (reapplied).

This patch adds support for the the LightWeight Profiling (LWP) instructions which are available on all AMD Bulldozer class CPUs (bdver1 to bdver4).

Reapplied - this time without changing line endings of existing files.

Differential Revision: https://reviews.llvm.org/D32769

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@302041 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[APInt] Give the value union a name so we can remove assumptions on VAL being the...
Craig Topper [Wed, 3 May 2017 15:46:24 +0000 (15:46 +0000)]
[APInt] Give the value union a name so we can remove assumptions on VAL being the larger member

Currently several places assume the VAL member is always at least the same size as pVal. In particular for a memcpy in the move assignment operator. While this is a true assumption, it isn't good practice to assume this.

This patch gives the union a name so we can write the memcpy in terms of the union itself. This also adds a similar memcpy to the move constructor where we previously just copied using VAL directly.

This patch is mostly just a mechanical addition of the U in front of VAL and pVAL everywhere. But several constructors had to be modified since we can't directly initializer a field of named union from the initializer list.

Differential Revision: https://reviews.llvm.org/D30629

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@302040 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoVerify that no compile units share the same line table in "llvm-dwarfdump --verify"
Greg Clayton [Wed, 3 May 2017 15:45:31 +0000 (15:45 +0000)]
Verify that no compile units share the same line table in "llvm-dwarfdump --verify"

Check to make sure no compile units have the same DW_AT_stmt_list values. Report a verification error if they do.

Differential Revision: https://reviews.llvm.org/D32771

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@302039 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoRevert rL302028 due to accidental line ending changes.
Simon Pilgrim [Wed, 3 May 2017 15:42:29 +0000 (15:42 +0000)]
Revert rL302028 due to accidental line ending changes.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@302038 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[Hexagon] Handle S2_storerf_io in HexagonInstrInfo
Krzysztof Parzyszek [Wed, 3 May 2017 15:36:51 +0000 (15:36 +0000)]
[Hexagon] Handle S2_storerf_io in HexagonInstrInfo

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@302036 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[Hexagon] Misc fixes in HexagonInstrInfo, NFC
Krzysztof Parzyszek [Wed, 3 May 2017 15:34:52 +0000 (15:34 +0000)]
[Hexagon] Misc fixes in HexagonInstrInfo, NFC

Formatting changes + remove unused function.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@302035 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[Hexagon] Adjust latency between allocframe and the first store on stack
Krzysztof Parzyszek [Wed, 3 May 2017 15:33:09 +0000 (15:33 +0000)]
[Hexagon] Adjust latency between allocframe and the first store on stack

Allocframe and the following stores on the stack have a latency of 2 cycles
when not in the same packet. This happens because R29 is needed early by the
store instruction. Since one of such stores can be packetized along with
allocframe and use old value of R29, we can assign it 0 cycle latency
while leaving latency of other stores to the default value of 2 cycles.

Patch by Jyotsna Verma.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@302034 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[Hexagon] Handle J2_jumptpt and J2_jumpfpt in HexagonInstrInfo
Krzysztof Parzyszek [Wed, 3 May 2017 15:30:46 +0000 (15:30 +0000)]
[Hexagon] Handle J2_jumptpt and J2_jumpfpt in HexagonInstrInfo

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@302033 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[Hexagon] Implement undoing .cur instructions in packetizer
Krzysztof Parzyszek [Wed, 3 May 2017 15:28:56 +0000 (15:28 +0000)]
[Hexagon] Implement undoing .cur instructions in packetizer

The packetizer needs to convert .cur instruction to its regular form if
the use is not in the same packet as the .cur. The code in the packetizer
handles one type of .cur, which is the vector load case. This patch
updates the packetizer so that it can undo all the .cur instructions.
In the test case, the .cur is the 128B version, but there are also the
post-increment versions.

Patch by Brendon Cahoon.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@302032 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[Hexagon] Add memory operands to a rewritten load
Krzysztof Parzyszek [Wed, 3 May 2017 15:26:13 +0000 (15:26 +0000)]
[Hexagon] Add memory operands to a rewritten load

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@302030 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[Hexagon] Reset spill alignment when variable-sized objects are present
Krzysztof Parzyszek [Wed, 3 May 2017 15:23:53 +0000 (15:23 +0000)]
[Hexagon] Reset spill alignment when variable-sized objects are present

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@302029 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[X86][LWP] Add llvm support for LWP instructions.
Simon Pilgrim [Wed, 3 May 2017 15:18:34 +0000 (15:18 +0000)]
[X86][LWP] Add llvm support for LWP instructions.

This patch adds support for the the LightWeight Profiling (LWP) instructions which are available on all AMD Bulldozer class CPUs (bdver1 to bdver4).

Differential Revision: https://reviews.llvm.org/D32769

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@302028 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoCMake: Add LLVM_DYLIB_SYMBOL_VERSIONING option
Tom Stellard [Wed, 3 May 2017 14:43:44 +0000 (14:43 +0000)]
CMake: Add LLVM_DYLIB_SYMBOL_VERSIONING option

Summary:
When apps or other libraries link against a library with symbol
versions, the version string is recorded in the import table, and used
at runtime to resolve the symbol back to a library that provides that
version (vaguely like how two-level namespaces work in Mach-O).  ld's
--default-symver flag tags every exported symbol with a symbol version
string equal to the library's soname.  Using --default-symver means
multiple versions of libLLVM can coexist within the same process, at
least to the extent that they don't try to pass data between each
other's llvms.

As an example, imagine a language like Rust using llvm for CPU codegen,
binding to OpenGL, with Mesa as the OpenGL implementation using llvm for
R600 codegen.  With --default-symver Rust and Mesa will resolve their
llvm usage to the version each was linked against, which need not match.

(Other ELF platforms like BSD and Solaris might have similar semantics,
I've not checked.)

This is based on an autoconf version of this patch by Adam Jackson.

This new option can be used to add --default-symver to the linker flags
for libLLVM.so.

Reviewers: beanz

Reviewed By: beanz

Subscribers: mgorny, llvm-commits

Differential Revision: https://reviews.llvm.org/D30997

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@302026 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoCMake: Move sphinx detection into AddSphinxTarget.cmake
Tom Stellard [Wed, 3 May 2017 14:29:56 +0000 (14:29 +0000)]
CMake: Move sphinx detection into AddSphinxTarget.cmake

Reviewers: chandlerc, beanz, mgorny

Reviewed By: beanz

Subscribers: llvm-commits

Differential Revision: https://reviews.llvm.org/D31773

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@302025 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[X86][AVX512] remove unnecessary case. NFC
Guy Blank [Wed, 3 May 2017 13:34:05 +0000 (13:34 +0000)]
[X86][AVX512] remove unnecessary case. NFC

VFPCLASS is for vector types and not scalar, so it cannot get here.

Differential Revision: https://reviews.llvm.org/D32694

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@302023 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[SystemZ] Properly check number of operands in getCmpOpsType()
Jonas Paulsson [Wed, 3 May 2017 13:33:45 +0000 (13:33 +0000)]
[SystemZ]  Properly check number of operands in getCmpOpsType()

It is needed to check that the number of operands are 2 when
finding the case of a logic combination, e.g. 'and' of two compares.

Review: Ulrich Weigand

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@302022 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[X86] Support of no_caller_saved_registers attribute
Oren Ben Simhon [Wed, 3 May 2017 13:07:19 +0000 (13:07 +0000)]
[X86] Support of no_caller_saved_registers attribute

This patch implements the LLVM part for no_caller_saved_registers attribute as appears here: https://gcc.gnu.org/git/?p=gcc.git;a=commit;h=5ed3cc7b66af4758f7849ed6f65f4365be8223be.
In order to implement the attribute, we use the dynamic CSR mechanism to remove returned/passed arguments from the function regmask/CSR list.

Differential Revision: https://reviews.llvm.org/D31876

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@302020 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoSupport arbitrary address space pointers in masked gather/scatter intrinsics.
Elad Cohen [Wed, 3 May 2017 12:28:54 +0000 (12:28 +0000)]
Support arbitrary address space pointers in masked gather/scatter intrinsics.

Fixes PR31789 - When loop-vectorize tries to use these intrinsics for a
non-default address space pointer we fail with a "Calling a function with a
bad singature!" assertion. This patch solves this by adding the 'vector of
pointers' argument as an overloaded type which will determine the address
space.

Differential revision: https://reviews.llvm.org/D31490

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@302018 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[AVR] Reserve the Y register in all functions
Dylan McKay [Wed, 3 May 2017 11:56:01 +0000 (11:56 +0000)]
[AVR] Reserve the Y register in all functions

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@302017 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[Loop Deletion] Delete loops that are never executed
Anna Thomas [Wed, 3 May 2017 11:47:11 +0000 (11:47 +0000)]
[Loop Deletion] Delete loops that are never executed

Summary:
Currently, loop deletion deletes loop where the only values
that are used outside the loop are loop-invariant.
This patch adds logic to delete loops where the loop is proven to be
never executed (i.e. the only predecessor of the loop preheader has a
constant conditional branch as terminator, and the preheader is not the
taken target). This will remove loops that become dead after
loop-unswitching generates constant conditional branches.

The next steps are:
1. moving the loop deletion implementation to LoopUtils.
2. Add logic in loop-simplifyCFG which will support changing conditional
constant branches to unconditional branches. If loops become unreachable in this
process, they can be removed using `deleteDeadLoop` function.

Reviewers: chandlerc, efriedma, sanjoy, reames

Reviewed by: sanjoy

Subscribers: mzolotukhin, llvm-commits

Differential Revision: https://reviews.llvm.org/D32494

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@302015 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoRevert "[AVR] Enable the frame pointer for all functions"
Dylan McKay [Wed, 3 May 2017 11:36:42 +0000 (11:36 +0000)]
Revert "[AVR] Enable the frame pointer for all functions"

This reverts commit 358ad02d999e88853d2cfc954bd2f668308a51f7.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@302014 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[Triple] Add a "macos" OS type that acts as a synonym for "macosx"
Alex Lorenz [Wed, 3 May 2017 10:42:35 +0000 (10:42 +0000)]
[Triple] Add a "macos" OS type that acts as a synonym for "macosx"

The "macosx" OS type is still the canonical type. In the future "macos" will
become the canonical OS type (but we will still support "macosx").

rdar://27043820

Differential Revision: https://reviews.llvm.org/D32748

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@302011 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[X86] Refactored LowerINTRINSIC_W_CHAIN to use a switch statament. NFCI.
Simon Pilgrim [Wed, 3 May 2017 10:40:18 +0000 (10:40 +0000)]
[X86] Refactored LowerINTRINSIC_W_CHAIN to use a switch statament. NFCI.

Pre-commit as requested in D32769.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@302010 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoRevert r301986 (and subsequent r301987).
Daniel Jasper [Wed, 3 May 2017 07:29:25 +0000 (07:29 +0000)]
Revert r301986 (and subsequent r301987).

The patch is failing to add StringTableStreamBuilder.h, but that isn't
even discovered because the corresponding StringTableStreamBuilder.cpp
isn't added to any CMakeLists.txt file and thus never built. I think
this patch is just incomplete.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@302002 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoFix use after free in BinaryStream library.
Zachary Turner [Wed, 3 May 2017 05:34:00 +0000 (05:34 +0000)]
Fix use after free in BinaryStream library.

This was reported by the ASAN bot, and it turned out to be
a fairly fundamental problem with the design of VarStreamArray
and the way it passes context information to the extractor.

The fix was cumbersome, and I'm not entirely pleased with it,
so I plan to revisit this design in the future when I'm not
pressed to get the bots green again.  For now, this fixes
the issue by storing the context information by value instead
of by reference, and introduces some impossibly-confusing
template magic to make things "work".

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@301999 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoReplace hardcoded intrinsic list with speculatable attribute.
Matt Arsenault [Wed, 3 May 2017 02:26:10 +0000 (02:26 +0000)]
Replace hardcoded intrinsic list with speculatable attribute.

No change in which intrinsics should be speculated.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@301995 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoRevert r295861, "[ModuleSummaryAnalysis] Don't crash when referencing unnamed globals."
Peter Collingbourne [Wed, 3 May 2017 00:18:48 +0000 (00:18 +0000)]
Revert r295861, "[ModuleSummaryAnalysis] Don't crash when referencing unnamed globals."

We should always expect values to be named before running the module summary
analysis (see NameAnonGlobals pass), so it's fine if we crash in that case.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@301991 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[PowerPC, DAGCombiner] Fold a << (b % (sizeof(a) * 8)) back to a single instruction
Tim Shen [Wed, 3 May 2017 00:07:02 +0000 (00:07 +0000)]
[PowerPC, DAGCombiner] Fold a << (b % (sizeof(a) * 8)) back to a single instruction

Summary:
This is the corresponding llvm change to D28037 to ensure no performance
regression.

Reviewers: bogner, kbarton, hfinkel, iteratee, echristo

Subscribers: nemanjai, llvm-commits

Differential Revision: https://reviews.llvm.org/D28329

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@301990 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoFix type conversion error.
Zachary Turner [Tue, 2 May 2017 23:41:51 +0000 (23:41 +0000)]
Fix type conversion error.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@301987 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoMake codeview::StringTable.
Zachary Turner [Tue, 2 May 2017 23:36:17 +0000 (23:36 +0000)]
Make codeview::StringTable.

Previously we had knowledge of how to serialize and deserialize
a string table inside of DebugInfo/PDB, but the string table
that it serializes contains a piece that is actually considered
CodeView and can appear outside of a PDB.  We already have logic
in llvm-readobj and MCCodeView to read and write this format,
so it doesn't make sense to duplicate the logic in DebugInfoPDB
as well.

This patch makes codeview::StringTable (for writing) and
codeview::StringTableRef (for reading), updates DebugInfoPDB
to use these classes for its own writing, and updates llvm-readobj
to additionally use StringTableRef for reading.

It's a bit more difficult to get MCCodeView to use this for
writing, but it's a logical next step.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@301986 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoTypo in LangRef.rst. NFC
Xin Tong [Tue, 2 May 2017 23:24:12 +0000 (23:24 +0000)]
Typo in LangRef.rst. NFC

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@301985 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoAdd line table verification to lldb-dwarfdump --verify
Greg Clayton [Tue, 2 May 2017 22:48:52 +0000 (22:48 +0000)]
Add line table verification to lldb-dwarfdump --verify

This patch verifies the .debug_line:
- verify all addresses in a line table sequence have ascending addresses
- verify that all line table file indexes are valid

Unit tests added for both cases.

Differential Revision: https://reviews.llvm.org/D32765

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@301984 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoARM: avoid handing a deleted node back to TableGen during ISel.
Tim Northover [Tue, 2 May 2017 22:45:19 +0000 (22:45 +0000)]
ARM: avoid handing a deleted node back to TableGen during ISel.

When we replaced the multiplicand the destination node might already exist.
When that happens the original gets CSEd and deleted. However, it's actually
used as the offset so nonsense is produced.

Should fix PR32726.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@301983 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agoRe-land r301697 "[IR] Make add/remove Attributes use AttrBuilder instead of Attribute...
Reid Kleckner [Tue, 2 May 2017 22:07:37 +0000 (22:07 +0000)]
Re-land r301697 "[IR] Make add/remove Attributes use AttrBuilder instead of AttributeList"

This time, I fixed, built, and tested clang.

This reverts r301712.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@301981 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[AArch64] ILP32 Backend Relocation Support
Joel Jones [Tue, 2 May 2017 22:01:48 +0000 (22:01 +0000)]
[AArch64] ILP32 Backend Relocation Support

Remove "_NC" suffix and semantics from TLSDESC_LD{64,32}_LO12 and
  TLSDESC_ADD_LO12 relocations
Rearrange ordering in AArch64.def to follow relocation encoding
Fix name:
  R_AARCH64_P32_LD64_GOT_LO12_NC => R_AARCH64_P32_LD32_GOT_LO12_NC
Add support for several "TLS", "TLSGD", and "TLSLD" relocations for
  ILP32
Fix return values from isNonILP32reloc
Add implementations for
  R_AARCH64_ADR_PREL_PG_HI21_NC, R_AARCH64_P32_LD32_GOT_LO12_NC,
  R_AARCH64_P32_TLSIE_LD32_GOTTPREL_LO12_NC,
  R_AARCH64_P32_TLSDESC_LD32_LO12, R_AARCH64_LD64_GOT_LO12_NC,
  *TLSLD_LDST128_DTPREL_LO12, *TLSLD_LDST128_DTPREL_LO12_NC,
  *TLSLE_LDST128_TPREL_LO12, *TLSLE_LDST128_TPREL_LO12_NC
Modify error messages to give name of equivalent relocation in the
  ABI not being used, along with better checking for non-existent
  requested relocations.
Added assembler support for "pg_hi21_nc"
Relocation definitions added without implementations:
  R_AARCH64_P32_TLSDESC_ADR_PREL21, R_AARCH64_P32_TLSGD_ADR_PREL21,
  R_AARCH64_P32_TLSGD_ADD_LO12_NC, R_AARCH64_P32_TLSLD_ADR_PREL21,
  R_AARCH64_P32_TLSLD_ADR_PAGE21, R_AARCH64_P32_TLSLD_ADD_LO12_NC,
  R_AARCH64_P32_TLSLD_LD_PREL19, R_AARCH64_P32_TLSDESC_LD_PREL19,
  R_AARCH64_P32_TLSGD_ADR_PAGE21, R_AARCH64_P32_TLS_DTPREL,
  R_AARCH64_P32_TLS_DTPMOD, R_AARCH64_P32_TLS_TPREL,
  R_AARCH64_P32_TLSDESC
Fix encoding:
  R_AARCH64_P32_TLSDESC_ADR_PAGE21

Reviewers: Peter Smith

Patch by: Joel Jones (jjones@cavium.com)

Differential Revision: https://reviews.llvm.org/D32072

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@301980 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[DWARFv5] Parse new line-table header format.
Paul Robinson [Tue, 2 May 2017 21:40:47 +0000 (21:40 +0000)]
[DWARFv5] Parse new line-table header format.

The directory and file tables now have form-based content descriptors.
Parse these and extract the per-directory/file records based on the
descriptors.  For now we support only DW_FORM_string (inline) for the
path names; follow-up work will add support for indirect forms (i.e.,
DW_FORM_strp, strx<N>, and line_strp).

Differential Revision: http://reviews.llvm.org/D32713

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@301978 91177308-0d34-0410-b5e6-96231b3b80d8

7 years agorevert r301766: InstructionSimplify: Canonicalize shuffle operands. NFC-ish
Sanjay Patel [Tue, 2 May 2017 21:37:28 +0000 (21:37 +0000)]
revert r301766: InstructionSimplify: Canonicalize shuffle operands. NFC-ish

Turns out this wasn't NFC-ish at all because there's a bug processing shuffles
that change the size of their input vectors (that case always seems to trip us
up).

This should fix PR32872 while we investigate how it failed and reduce a testcase:
https://bugs.llvm.org/show_bug.cgi?id=32872

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@301977 91177308-0d34-0410-b5e6-96231b3b80d8

7 years ago[NewGVN] Fix typo and format comment. NFCI.
Davide Italiano [Tue, 2 May 2017 21:11:40 +0000 (21:11 +0000)]
[NewGVN] Fix typo and format comment. NFCI.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@301974 91177308-0d34-0410-b5e6-96231b3b80d8