oota-llvm.git
9 years agoYAML: Assign a value returned by the default constructor to the value in an optional...
Alex Lorenz [Wed, 17 Jun 2015 23:26:01 +0000 (23:26 +0000)]
YAML: Assign a value returned by the default constructor to the value in an optional mapping.

This commit ensures that a value that's passed into YAML's IO mapOptional method
is going to be assigned a value returned by the default constructor for that
value's type when the appropriate key is not present in the YAML mapping.

Reviewers: Duncan P. N. Exon Smith

Differential Revision: http://reviews.llvm.org/D10492

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@239972 91177308-0d34-0410-b5e6-96231b3b80d8

9 years ago[X86][SSE] Improved support for vector i16 to float conversions.
Simon Pilgrim [Wed, 17 Jun 2015 22:43:34 +0000 (22:43 +0000)]
[X86][SSE] Improved support for vector i16 to float conversions.

Added explicit sign extension for v4i16/v8i16 to v4i32/v8i32 before conversion to floats. Matches existing support for v4i8/v8i8.

Follow up to D10433

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@239966 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoAdd NVPTXLowerAlloca pass to convert alloca'ed memory to local address
Jingyue Wu [Wed, 17 Jun 2015 22:31:02 +0000 (22:31 +0000)]
Add NVPTXLowerAlloca pass to convert alloca'ed memory to local address

Summary:
This is done by first adding two additional instructions to convert the
alloca returned address to local and convert it back to generic. Then
replace all uses of alloca instruction with the converted generic
address. Then we can rely NVPTXFavorNonGenericAddrSpace pass to combine
the generic addresscast and the corresponding Load, Store, Bitcast, GEP
Instruction together.

Patched by Xuetian Weng (xweng@google.com).

Test Plan: test/CodeGen/NVPTX/lower-alloca.ll

Reviewers: jholewinski, jingyue

Reviewed By: jingyue

Subscribers: meheff, broune, eliben, jholewinski, llvm-commits

Differential Revision: http://reviews.llvm.org/D10483

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@239964 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoDevirtualize and pack MCFragment to reduce memory usage.
Pete Cooper [Wed, 17 Jun 2015 22:01:28 +0000 (22:01 +0000)]
Devirtualize and pack MCFragment to reduce memory usage.

MCFragment didn't really need vtables.  The majority of virtual methods were just getters and setters.

This removes the vtables and uses dispatch on the kind to do things like delete which needs to
get the appropriate class.

This reduces memory on the verify use list order test case by about 2MB out of 800MB.

Reviewed by Rafael EspĂ­ndola

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@239952 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoRe-land "[X86] Cache variables that only depend on the subtarget"
Reid Kleckner [Wed, 17 Jun 2015 21:50:02 +0000 (21:50 +0000)]
Re-land "[X86] Cache variables that only depend on the subtarget"

Re-instates r239949 without accidentally flipping the sense of UseLEA.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@239950 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoRevert "[X86] Cache variables that only depend on the subtarget"
Reid Kleckner [Wed, 17 Jun 2015 21:35:02 +0000 (21:35 +0000)]
Revert "[X86] Cache variables that only depend on the subtarget"

This reverts commit r239948, tests seem to be failing.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@239949 91177308-0d34-0410-b5e6-96231b3b80d8

9 years ago[X86] Cache variables that only depend on the subtarget
Reid Kleckner [Wed, 17 Jun 2015 21:31:17 +0000 (21:31 +0000)]
[X86] Cache variables that only depend on the subtarget

There is a one-to-one relationship between X86Subtarget and
X86FrameLowering, but every frame lowering method would previously pull
the subtarget off the MachineFunction and query some subtarget
properties.

Over time, these locals began to grow in complexity and it became
important to keep their names and meaning in sync across all of the
frame lowering methods, leading to duplication. We can eliminate that
duplication by computing them once in the constructor.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@239948 91177308-0d34-0410-b5e6-96231b3b80d8

9 years ago[docs] Fix "WARNING: Title underline too short."
David Majnemer [Wed, 17 Jun 2015 21:21:16 +0000 (21:21 +0000)]
[docs] Fix "WARNING: Title underline too short."

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@239947 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoAdd missing include.
Benjamin Kramer [Wed, 17 Jun 2015 21:08:22 +0000 (21:08 +0000)]
Add missing include.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@239946 91177308-0d34-0410-b5e6-96231b3b80d8

9 years ago[Bitcode] Replace hand-coded little endian handling with Endian.h functions.
Benjamin Kramer [Wed, 17 Jun 2015 20:55:30 +0000 (20:55 +0000)]
[Bitcode] Replace hand-coded little endian handling with Endian.h functions.

No functional change intended.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@239944 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoAMDGPU: Change unreachable into reported error
Matt Arsenault [Wed, 17 Jun 2015 20:55:25 +0000 (20:55 +0000)]
AMDGPU: Change unreachable into reported error

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@239943 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoremove unnecessary casts; NFC
Sanjay Patel [Wed, 17 Jun 2015 20:54:46 +0000 (20:54 +0000)]
remove unnecessary casts; NFC

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@239942 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoMove the personality function from LandingPadInst to Function
David Majnemer [Wed, 17 Jun 2015 20:52:32 +0000 (20:52 +0000)]
Move the personality function from LandingPadInst to Function

The personality routine currently lives in the LandingPadInst.

This isn't desirable because:
- All LandingPadInsts in the same function must have the same
  personality routine.  This means that each LandingPadInst beyond the
  first has an operand which produces no additional information.

- There is ongoing work to introduce EH IR constructs other than
  LandingPadInst.  Moving the personality routine off of any one
  particular Instruction and onto the parent function seems a lot better
  than have N different places a personality function can sneak onto an
  exceptional function.

Differential Revision: http://reviews.llvm.org/D10429

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@239940 91177308-0d34-0410-b5e6-96231b3b80d8

9 years ago[CodeGenPrepare] Generalize inserted set from truncs to any inst.
Ahmed Bougacha [Wed, 17 Jun 2015 20:44:32 +0000 (20:44 +0000)]
[CodeGenPrepare] Generalize inserted set from truncs to any inst.

It's been used before to avoid infinite loops caused by separate CGP
optimizations undoing one another.  We found one more such issue
caused by r238054.  To avoid it, generalize the "InsertedTruncs"
set to any inst, and use it to avoid touching those again.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@239938 91177308-0d34-0410-b5e6-96231b3b80d8

9 years ago[Hexagon] Adding a number of other tests for min/max instructions and loading i1s.
Colin LeMahieu [Wed, 17 Jun 2015 20:29:33 +0000 (20:29 +0000)]
[Hexagon] Adding a number of other tests for min/max instructions and loading i1s.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@239935 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoMove IsUsedInReloc from MCSymbolELF to MCSymbol.
Rafael Espindola [Wed, 17 Jun 2015 20:08:20 +0000 (20:08 +0000)]
Move IsUsedInReloc from MCSymbolELF to MCSymbol.

There is a free bit is MCSymbol and MachO needs the same information.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@239933 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoLowerBitSets: Do not assign names to aliases of unnamed bitset element objects.
Peter Collingbourne [Wed, 17 Jun 2015 18:31:02 +0000 (18:31 +0000)]
LowerBitSets: Do not assign names to aliases of unnamed bitset element objects.

The restriction on unnamed aliases was removed in r239921. Mostly reverts
r239590, but we keep the test.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@239923 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoAllow aliases to be unnamed.
Rafael Espindola [Wed, 17 Jun 2015 17:53:31 +0000 (17:53 +0000)]
Allow aliases to be unnamed.

If globals can be unnamed, there is no reason for aliases to be different.

The restriction was there since the original implementation in r36435. I
can only guess it was there because of the old bison parser for the old
alias syntax.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@239921 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoUse a range loop. NFC.
Rafael Espindola [Wed, 17 Jun 2015 17:33:37 +0000 (17:33 +0000)]
Use a range loop. NFC.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@239919 91177308-0d34-0410-b5e6-96231b3b80d8

9 years ago[Hexagon] Adding some compare tests, fixing existing XFAILed tests, and removing...
Colin LeMahieu [Wed, 17 Jun 2015 17:19:05 +0000 (17:19 +0000)]
[Hexagon] Adding some compare tests, fixing existing XFAILed tests, and removing mcpu=hexagonv4 since that's the minimum version anyway.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@239917 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agofix typos in comments; NFC
Sanjay Patel [Wed, 17 Jun 2015 16:34:48 +0000 (16:34 +0000)]
fix typos in comments; NFC

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@239916 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoAdd documentation for new backedge mass propagation in irregular loops.
Diego Novillo [Wed, 17 Jun 2015 16:28:22 +0000 (16:28 +0000)]
Add documentation for new backedge mass propagation in irregular loops.

Tweak test cases and rename headerIndexFor -> getHeaderIndex.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@239915 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoUse named temporaries for directional labels.
Rafael Espindola [Wed, 17 Jun 2015 16:26:47 +0000 (16:26 +0000)]
Use named temporaries for directional labels.

Directional labels can show up in symbol tables (and we have a llvm-mc test for
that). Given that, we need to make sure they are named.

With that out of the way, use setUseNamesOnTempLabels in llvm-mc so that it
too benefits from the memory saving.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@239914 91177308-0d34-0410-b5e6-96231b3b80d8

9 years ago[ArchiveWriter] Use EndianStream. No functional change intended.
Benjamin Kramer [Wed, 17 Jun 2015 16:02:56 +0000 (16:02 +0000)]
[ArchiveWriter] Use EndianStream. No functional change intended.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@239913 91177308-0d34-0410-b5e6-96231b3b80d8

9 years ago[MC/Dwarf] Encode DW_CFA_advance_loc in target endianess.
Benjamin Kramer [Wed, 17 Jun 2015 15:14:35 +0000 (15:14 +0000)]
[MC/Dwarf] Encode DW_CFA_advance_loc in target endianess.

This matches GNU as output.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@239911 91177308-0d34-0410-b5e6-96231b3b80d8

9 years ago[mips] [IAS] Add support for expanding LASym with a source register operand.
Toma Tabacu [Wed, 17 Jun 2015 14:31:51 +0000 (14:31 +0000)]
[mips] [IAS] Add support for expanding LASym with a source register operand.

Reviewers: dsanders

Reviewed By: dsanders

Subscribers: llvm-commits

Differential Revision: http://reviews.llvm.org/D9348

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@239910 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoTweak wording of alignment static_assert messages.
James Y Knight [Wed, 17 Jun 2015 13:53:12 +0000 (13:53 +0000)]
Tweak wording of alignment static_assert messages.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@239907 91177308-0d34-0410-b5e6-96231b3b80d8

9 years ago[mips] [IAS] Add support for the B{L,G}{T,E}(U) branch pseudo-instructions.
Toma Tabacu [Wed, 17 Jun 2015 13:20:24 +0000 (13:20 +0000)]
[mips] [IAS] Add support for the B{L,G}{T,E}(U) branch pseudo-instructions.

Summary:
This does not include support for the immediate variants of these pseudo-instructions.
Fixes llvm.org/PR20968.

Reviewers: dsanders

Reviewed By: dsanders

Subscribers: seanbruno, emaste, llvm-commits

Differential Revision: http://reviews.llvm.org/D8537

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@239905 91177308-0d34-0410-b5e6-96231b3b80d8

9 years ago[mips] [IAS] Fix LA with relative label operands.
Toma Tabacu [Wed, 17 Jun 2015 12:30:37 +0000 (12:30 +0000)]
[mips] [IAS] Fix LA with relative label operands.

Summary:
Call MCSymbolRefExpr::create() with a MCSymbol* argument, not with a StringRef
of the Symbol's name, in order to avoid creating invalid temporary symbols for
relative labels (e.g. {$,.L}tmp00, {$,.L}tmp10 etc.).

Reviewers: dsanders

Reviewed By: dsanders

Subscribers: llvm-commits

Differential Revision: http://reviews.llvm.org/D10498

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@239901 91177308-0d34-0410-b5e6-96231b3b80d8

9 years ago[mips] [IAS] Add test for SW with relative label operands. NFC.
Toma Tabacu [Wed, 17 Jun 2015 11:46:37 +0000 (11:46 +0000)]
[mips] [IAS] Add test for SW with relative label operands. NFC.

Reviewers: dsanders

Reviewed By: dsanders

Subscribers: llvm-commits

Differential Revision: http://reviews.llvm.org/D10497

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@239899 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoRemove empty directories left after r239657
Alexander Kornienko [Wed, 17 Jun 2015 11:31:18 +0000 (11:31 +0000)]
Remove empty directories left after r239657

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@239898 91177308-0d34-0410-b5e6-96231b3b80d8

9 years ago[mips] [IAS] Fix LW with relative label operands.
Toma Tabacu [Wed, 17 Jun 2015 10:43:45 +0000 (10:43 +0000)]
[mips] [IAS] Fix LW with relative label operands.

Summary:
Previously, MCSymbolRefExpr::create() was called with a StringRef of the symbol
name, which it would then search for in the Symbols StringMap (from MCContext).

However, relative labels (which are temporary symbols) are apparently not stored
in the Symbols StringMap, so we end up creating a new {$,.L}tmp symbol
({$,.L}tmp00, {$,.L}tmp10 etc.) each time we create an MCSymbolRefExpr by
passing in the symbol name as a StringRef.

Fortunately, there is a version of MCSymbolRefExpr::create() which takes an
MCSymbol* and we already have an MCSymbol* at that point, so we can just pass
that in instead of the StringRef.

I also removed the local StringRef calls to MCSymbolRefExpr::create() from
expandMemInst(), as those cases can be handled by evaluateRelocExpr() anyway.

Reviewers: dsanders

Reviewed By: dsanders

Subscribers: llvm-commits

Differential Revision: http://reviews.llvm.org/D9938

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@239897 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoAVX-512: cvtusi2ss/d intrinsics.
Igor Breger [Wed, 17 Jun 2015 07:23:57 +0000 (07:23 +0000)]
AVX-512: cvtusi2ss/d intrinsics.
Change builtin function name and signature ( add third parameter - rounding mode ).
Added tests for intrinsics.

Differential Revision: http://reviews.llvm.org/D10473

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@239888 91177308-0d34-0410-b5e6-96231b3b80d8

9 years ago[PM/AA] Suffix lots of member variables that directly use enumeration
Chandler Carruth [Wed, 17 Jun 2015 07:21:41 +0000 (07:21 +0000)]
[PM/AA] Suffix lots of member variables that directly use enumeration
names for counts with the word 'Count' to make them less ambiguous.

This will be an actual error if we use unscoped enums for any of these,
and generally this seems much clearer to read.

Also, use clang-format to normalize the formatting of this code which
seems to have been needlessly odd.

No functionality changed here.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@239887 91177308-0d34-0410-b5e6-96231b3b80d8

9 years ago[PM/AA] Remove the UnknownSize static member from AliasAnalysis.
Chandler Carruth [Wed, 17 Jun 2015 07:21:38 +0000 (07:21 +0000)]
[PM/AA] Remove the UnknownSize static member from AliasAnalysis.

This is now living in MemoryLocation, which is what it pertains to. It
is also an enum there rather than a static data member which is left
never defined.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@239886 91177308-0d34-0410-b5e6-96231b3b80d8

9 years ago[PM/AA] Remove the Location typedef from the AliasAnalysis class now
Chandler Carruth [Wed, 17 Jun 2015 07:18:54 +0000 (07:18 +0000)]
[PM/AA] Remove the Location typedef from the AliasAnalysis class now
that it is its own entity in the form of MemoryLocation, and update all
the callers.

This is an entirely mechanical change. References to "Location" within
AA subclases become "MemoryLocation", and elsewhere
"AliasAnalysis::Location" becomes "MemoryLocation". Hope that helps
out-of-tree folks update.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@239885 91177308-0d34-0410-b5e6-96231b3b80d8

9 years ago[PM/AA] Split the location computation out of getArgLocation so the
Chandler Carruth [Wed, 17 Jun 2015 07:12:40 +0000 (07:12 +0000)]
[PM/AA] Split the location computation out of getArgLocation so the
virtual interface on AliasAnalysis only deals with ModRef information.

This interface was both computing memory locations by using TLI and
other tricks to estimate the size of memory referenced by an operand,
and computing ModRef information through similar investigations. This
change narrows the scope of the virtual interface on AliasAnalysis
slightly.

Note that all of this code could live in BasicAA, and be done with
a single investigation of the argument, if it weren't for the fact that
the generic code in AliasAnalysis::getModRefBehavior for a callsite
calls into the virtual aspect of (now) getArgModRefInfo. But this
patch's arrangement seems a not terrible way to go for now.

The other interesting wrinkle is how we could reasonably extend LLVM
with support for custom memory location sizes and mod/ref behavior for
library routines. After discussions with Hal on the review, the
conclusion is that this would be best done by fleshing out the much
desired support for extensions to TLI, and support these types of
queries in that interface where we would likely be doing other library
API recognition and analysis.

Differential Revision: http://reviews.llvm.org/D10259

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@239884 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoRevert "AArch64: Use CMP;CCMP sequences for and/or/setcc trees."
Matthias Braun [Wed, 17 Jun 2015 04:02:32 +0000 (04:02 +0000)]
Revert "AArch64: Use CMP;CCMP sequences for and/or/setcc trees."

The patch triggers a miscompile on SPEC 2006 403.gcc with the (ref)
200.i and scilab.i inputs. I opened PR23866 to track analysis of this.

This reverts commit r238793.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@239880 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoTry to fix the MSVC build.
Rafael Espindola [Wed, 17 Jun 2015 03:13:26 +0000 (03:13 +0000)]
Try to fix the MSVC build.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@239877 91177308-0d34-0410-b5e6-96231b3b80d8

9 years ago[Hexagon] Adding MC ELF streamer and updating addend relocation test which shows...
Colin LeMahieu [Wed, 17 Jun 2015 03:06:16 +0000 (03:06 +0000)]
[Hexagon] Adding MC ELF streamer and updating addend relocation test which shows correct ELF symbol.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@239876 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoFix alignment issues in LLVM.
James Y Knight [Wed, 17 Jun 2015 01:21:20 +0000 (01:21 +0000)]
Fix alignment issues in LLVM.

Adds static_asserts to ensure alignment of concatenated objects is
correct, and fixes them where they are not.

Also changes the definition of AlignOf to use constexpr, except on
MSVC, to avoid enum comparison warnings from GCC.

(There's not too much of this in llvm itself, most of the fun is in
clang).

This seems to make LLVM actually work without Bus Error on 32bit
sparc.

Differential Revision: http://reviews.llvm.org/D10271

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@239872 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoHandle forward referenced function when streaming bitcode.
Rafael Espindola [Wed, 17 Jun 2015 01:15:47 +0000 (01:15 +0000)]
Handle forward referenced function when streaming bitcode.

Without this the included unit test would assert in

  assert(BasicBlockFwdRefs.empty() && "Unresolved blockaddress fwd references");

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@239871 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoHandle MaterializeAll in getLazyBitcodeModuleImpl. NFC.
Rafael Espindola [Wed, 17 Jun 2015 00:40:56 +0000 (00:40 +0000)]
Handle MaterializeAll in getLazyBitcodeModuleImpl. NFC.

This just handles both cases in the same place.

Extracted from a patch by Karl Schimpf.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@239870 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoUse std::unique_ptr to manage the DataStreamer in bitcode parsing.
Rafael Espindola [Tue, 16 Jun 2015 23:29:49 +0000 (23:29 +0000)]
Use std::unique_ptr to manage the DataStreamer in bitcode parsing.

We were already deleting it, this just makes it explicit.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@239867 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoRename and improve emitSectionOffset.
Rafael Espindola [Tue, 16 Jun 2015 23:22:02 +0000 (23:22 +0000)]
Rename and improve emitSectionOffset.

Different object formats represent references from dwarf in different ways.

ELF uses a relocation to the referenced point (except for .dwo) and
COFF/MachO use the offset of the referenced point inside its section.

This patch renames emitSectionOffset because

* It doesn't produce an offset on ELF.
* It changes behavior depending on how DWARF is represented, so adding
dwarf to its name is probably a good thing.

The patch also adds an option to force the use of offsets.That avoids
funny looking code like

  if (!UseOffsets)
    Asm->emitSectionOffset....

It was correct, but read as if the ! was inverted.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@239866 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoRefactor RecurrenceInstDesc
Tyler Nowicki [Tue, 16 Jun 2015 22:59:45 +0000 (22:59 +0000)]
Refactor RecurrenceInstDesc

Moved RecurrenceInstDesc into RecurrenceDescriptor to simplify the namespaces.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@239862 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoAdd some tests based on PR21711
Sanjay Patel [Tue, 16 Jun 2015 22:37:50 +0000 (22:37 +0000)]
Add some tests based on PR21711

These were originally added in r227242,
but that patch was reverted because it
caused a failure on AArch64.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@239860 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoReturn a unique_ptr from getLazyBitcodeModule and parseBitcodeFile. NFC.
Rafael Espindola [Tue, 16 Jun 2015 22:27:55 +0000 (22:27 +0000)]
Return a unique_ptr from getLazyBitcodeModule and parseBitcodeFile. NFC.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@239858 91177308-0d34-0410-b5e6-96231b3b80d8

9 years ago[llvm-readobj] Print MIPS .reginfo section content
Simon Atanasyan [Tue, 16 Jun 2015 21:47:43 +0000 (21:47 +0000)]
[llvm-readobj] Print MIPS .reginfo section content

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@239856 91177308-0d34-0410-b5e6-96231b3b80d8

9 years ago[X86][SSE] Vectorize v2i32 to v2f64 conversions
Simon Pilgrim [Tue, 16 Jun 2015 21:40:28 +0000 (21:40 +0000)]
[X86][SSE] Vectorize v2i32 to v2f64 conversions

This patch enables support for the conversion of v2i32 to v2f64 to use the CVTDQ2PD xmm instruction and stay on the SSE unit instead of scalarizing, sign extending to i64 and using CVTSI2SDQ scalar conversions.

Differential Revision: http://reviews.llvm.org/D10433

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@239855 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agorename variables; NFC
Sanjay Patel [Tue, 16 Jun 2015 20:47:19 +0000 (20:47 +0000)]
rename variables; NFC

...because I see 'StoreBW' and read it as 'store bandwidth'

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@239850 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoReapply 239795 - [InstCombine] Propagate non-null facts to call parameters
Philip Reames [Tue, 16 Jun 2015 20:24:25 +0000 (20:24 +0000)]
Reapply 239795 - [InstCombine] Propagate non-null facts to call parameters

The original change broke clang side tests.  I will be submitting those momentarily.  This change includes post commit feedback on the original change from from Pete Cooper.

Original Submission comments:
If a parameter to a function is known non-null, use the existing parameter attributes to record that fact at the call site. This has no optimization benefit by itself - that I know of - but is an enabling change for http://reviews.llvm.org/D9129.

Differential Revision: http://reviews.llvm.org/D9132

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@239849 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoextract some code into a helper function for MergeConsecutiveStores(); NFCI
Sanjay Patel [Tue, 16 Jun 2015 20:05:00 +0000 (20:05 +0000)]
extract some code into a helper function for MergeConsecutiveStores(); NFCI

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@239847 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoImprove handling of end of file in the bitcode reader.
Rafael Espindola [Tue, 16 Jun 2015 20:03:39 +0000 (20:03 +0000)]
Improve handling of end of file in the bitcode reader.

Before this patch the bitcode reader would read a module from a file
that contained in order:

* Any number of non MODULE_BLOCK sub blocks.
* One MODULE_BLOCK
* Any number of non MODULE_BLOCK sub blocks.
* 4 '\n' characters to handle OS X's ranlib.

Since we support lazy reading of modules, any information that is relevant
for the module has to be in the MODULE_BLOCK or before it. We don't gain
anything from checking what is after.

This patch then changes the reader to stop once the MODULE_BLOCK has been
successfully parsed.

This avoids the ugly special case for .bc files in an archive and makes it
easier to embed bitcode files.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@239845 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoFix PR 23525 - Separate header mass propagation in irregular loops.
Diego Novillo [Tue, 16 Jun 2015 19:10:58 +0000 (19:10 +0000)]
Fix PR 23525 - Separate header mass propagation in irregular loops.

Summary:
When propagating mass through irregular loops, the mass flowing through
each loop header may not be equal. This was causing wrong frequencies
to be computed for irregular loop headers.

Fixed by keeping track of masses flowing through each of the headers in
an irregular loop. To do this, we now keep track of per-header backedge
weights. After the loop mass is distributed through the loop, the
backedge weights are used to re-distribute the loop mass to the loop
headers.

Since each backedge will have a mass proportional to the different
branch weights, the loop headers will end up with a more approximate
weight distribution (as opposed to the current distribution that assumes
that every loop header is the same).

Reviewers: dexonsmith

Subscribers: llvm-commits

Differential Revision: http://reviews.llvm.org/D10348

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@239843 91177308-0d34-0410-b5e6-96231b3b80d8

9 years ago[Statepoints] Test only change. Check that statepoint lowering didn't generate more...
Igor Laevsky [Tue, 16 Jun 2015 19:07:05 +0000 (19:07 +0000)]
[Statepoints] Test only change. Check that statepoint lowering didn't generate more than expected amount of spills.
See http://reviews.llvm.org/D10402 for related discussion.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@239842 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoVirtRegMap: Add undef flag when reading undefined subregisters.
Matthias Braun [Tue, 16 Jun 2015 18:22:28 +0000 (18:22 +0000)]
VirtRegMap: Add undef flag when reading undefined subregisters.

While completely undefined registers are easy to catch and get their
<undef> flag early in ProcessImplicitDefs/RegisterCoalescer reading from
a partially defined register where just the subreg happens to be
undefined is harder to catch so we only add the undef flag in the
virtual register rewriting step.

No testcase as I cannot reproduce the problem on any of the in-tree targets at
the moment.

This fixes rdar://21387089

Differential Revision: http://reviews.llvm.org/D10470

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@239838 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoTargetRegisterInfo: Make the concept of imprecise lane masks explicit
Matthias Braun [Tue, 16 Jun 2015 18:22:26 +0000 (18:22 +0000)]
TargetRegisterInfo: Make the concept of imprecise lane masks explicit

LaneMasks as given by getSubRegIndexLaneMask() have a limited number of
of bits, so for targets with more than 31 disjunct subregister there may
be cases where:

   getSubReg(Reg,A) does not overlap getSubReg(Reg,B)

but we still have

   (getSubRegIndexLaneMask(A) & getSubRegIndexLaneMask(B)) != 0.

I had hoped to keep this an implementation detail of the tablegen but as
my next commit shows we can avoid unnecessary imp-defs operands if we
know that the lane masks in use are precise.

This is in preparation to http://reviews.llvm.org/D10470.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@239837 91177308-0d34-0410-b5e6-96231b3b80d8

9 years ago[X86] Rename some frame lowering variables
Reid Kleckner [Tue, 16 Jun 2015 18:08:57 +0000 (18:08 +0000)]
[X86] Rename some frame lowering variables

Old names, new names, and what they really mean:

- IsWin64 -> IsWin64CC: This is true on non-Windows x86_64 platforms
  when the ms_abi calling convention is used.
- IsWinEH -> IsWin64Prologue: True when the target is Win64, regardless
  of calling convention. Changes the prologue to obey the constraints of
  the Win64 unwinder.
- NeedsWinEH -> NeedsWinCFI: We're using the win64 prologue *and* the we
  want .xdata unwind tables. Analogous to NeedsDwarfCFI.

NFC

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@239836 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoRename Reduction variables/structures to Recurrence.
Tyler Nowicki [Tue, 16 Jun 2015 18:07:34 +0000 (18:07 +0000)]
Rename Reduction variables/structures to Recurrence.

A reduction is a special kind of recurrence. In the loop vectorizer we currently
identify basic reductions. Future patches will extend this to identifying basic
recurrences.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@239835 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoHave MachOObjectFile::isValidArch() accept armv7
Frederic Riss [Tue, 16 Jun 2015 17:37:03 +0000 (17:37 +0000)]
Have MachOObjectFile::isValidArch() accept armv7

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@239833 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoMIR Parser: Report an error when a machine function doesn't have a corresponding...
Alex Lorenz [Tue, 16 Jun 2015 17:06:29 +0000 (17:06 +0000)]
MIR Parser: Report an error when a machine function doesn't have a corresponding function.

This commit reports an error when a machine function from a MIR file that contains
LLVM IR can't find a function with the same name in the loaded LLVM IR module.

Reviewers: Duncan P. N. Exon Smith

Differential Revision: http://reviews.llvm.org/D10468

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@239831 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoAdd a test for padded bitcode files.
Rafael Espindola [Tue, 16 Jun 2015 16:36:15 +0000 (16:36 +0000)]
Add a test for padded bitcode files.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@239829 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agopropagate IR-level fast-math-flags to DAG nodes, disabled by default
Sanjay Patel [Tue, 16 Jun 2015 16:25:43 +0000 (16:25 +0000)]
propagate IR-level fast-math-flags to DAG nodes, disabled by default

This is an updated version of the patch that was checked in at:
http://reviews.llvm.org/rL237046

but subsequently reverted because it exposed a bug in the DAG Combiner:
http://reviews.llvm.org/D9893

This time, there's an enablement flag ("EnableFMFInDAG") around the code in
SelectionDAGBuilder where we copy the set of FP optimization flags from IR
instructions to DAG nodes. So, in theory, there should be no functional change
from this patch as-is, but it will allow testing with the added functionality
to proceed via "-enable-fmf-dag" passed to llc.

This patch adds the minimum plumbing necessary to use IR-level
fast-math-flags (FMF) in the backend without actually using
them for anything yet. This is a follow-on to:
http://reviews.llvm.org/rL235997

Differential Revision: http://reviews.llvm.org/D10403

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@239828 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoProperly handle the mftb instruction.
Kit Barton [Tue, 16 Jun 2015 16:01:15 +0000 (16:01 +0000)]
Properly handle the mftb instruction.

The mftb instruction was incorrectly marked as deprecated in the PPC
Backend. Instead, it should not be treated as deprecated, but rather be
implemented using the mfspr instruction. A similar patch was put into GCC last
year. Details can be found at:

https://sourceware.org/ml/binutils/2014-11/msg00383.html.
This change will replace instances of the mftb instruction with the mfspr
instruction for all CPUs except 601 and pwr3. This will also be the default
behaviour.

Additional details can be found in:

https://llvm.org/bugs/show_bug.cgi?id=23680

Phabricator review: http://reviews.llvm.org/D10419

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@239827 91177308-0d34-0410-b5e6-96231b3b80d8

9 years ago[Hexagon] Alphabetical ordering of functions, NFC.
Colin LeMahieu [Tue, 16 Jun 2015 15:59:53 +0000 (15:59 +0000)]
[Hexagon] Alphabetical ordering of functions, NFC.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@239826 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoRevert "Revert "Fix merges of non-zero vector stores""
Matt Arsenault [Tue, 16 Jun 2015 15:51:48 +0000 (15:51 +0000)]
Revert "Revert "Fix merges of non-zero vector stores""

Reapply r239539. Don't assume the collected number of
stores is the same vector size. Just take the first N
stores to fill the vector.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@239825 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoClean up redundant copies of Triple objects. NFC
Daniel Sanders [Tue, 16 Jun 2015 15:44:21 +0000 (15:44 +0000)]
Clean up redundant copies of Triple objects. NFC

Summary:

Reviewers: rengolin

Reviewed By: rengolin

Subscribers: llvm-commits, rengolin, jholewinski

Differential Revision: http://reviews.llvm.org/D10382

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@239823 91177308-0d34-0410-b5e6-96231b3b80d8

9 years ago[InstSimplify] Allow folding of fdiv X, X with just NaNs ignored
Benjamin Kramer [Tue, 16 Jun 2015 14:57:29 +0000 (14:57 +0000)]
[InstSimplify] Allow folding of fdiv X, X with just NaNs ignored

Any combination of +-inf/+-inf is NaN so it's already ignored with
nnan and we can skip checking for ninf. Also rephrase logic in comments
a bit.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@239821 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoRepair cmake libatomic check.
James Y Knight [Tue, 16 Jun 2015 14:00:01 +0000 (14:00 +0000)]
Repair cmake libatomic check.

The cmake check for whether libatomic could be used had been
unconditionally setting the result to false. Which was somewhat
fortunate, because the prerequisite check for whether it was *needed*
was always claiming it was, even if it was not.

However, this made platforms where libatomic is actually necessary
fail to link.

Differential Revision: http://reviews.llvm.org/D10453

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@239819 91177308-0d34-0410-b5e6-96231b3b80d8

9 years ago[mips][ias] Expand on r238751 to cover as many relocs as possible.
Daniel Sanders [Tue, 16 Jun 2015 13:46:26 +0000 (13:46 +0000)]
[mips][ias] Expand on r238751 to cover as many relocs as possible.

Summary:
Relocs that can be converted from absolute to PC-relative now do so if IsPCRel
is true. Relocs that require PC-relative now call llvm_unreachable() if IsPCRel
is false and similarly those that require absolute assert that IsPCRel is false.

Note that while it looks like some relocs (e.g. R_MIPS_26) can be converted into
the MIPS32r6/MIPS64r6 relocs (R_MIPS_PC*_S2), it isn't actually valid to do so.

Placeholders have been left in the testcase for unsupported relocs and relocs
that cannot be generated at the moment.

Reviewers: vkalintiris

Reviewed By: vkalintiris

Subscribers: llvm-commits, rafael

Differential Revision: http://reviews.llvm.org/D10184

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@239817 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoReplace string GNU Triples with llvm::Triple in TargetMachine::getTargetTriple()...
Daniel Sanders [Tue, 16 Jun 2015 13:15:50 +0000 (13:15 +0000)]
Replace string GNU Triples with llvm::Triple in TargetMachine::getTargetTriple(). NFC.

Summary:
This continues the patch series to eliminate StringRef forms of GNU triples
from the internals of LLVM that began in r239036.

Reviewers: rengolin

Reviewed By: rengolin

Subscribers: llvm-commits, rengolin

Differential Revision: http://reviews.llvm.org/D10381

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@239815 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoSilence an MSVC warning about not all control paths returning a value; NFC.
Aaron Ballman [Tue, 16 Jun 2015 13:14:59 +0000 (13:14 +0000)]
Silence an MSVC warning about not all control paths returning a value; NFC.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@239814 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoRecommit r239721: Replace string GNU Triples with llvm::Triple in InitMCObjectFileInf...
Daniel Sanders [Tue, 16 Jun 2015 12:18:07 +0000 (12:18 +0000)]
Recommit r239721: Replace string GNU Triples with llvm::Triple in InitMCObjectFileInfo. NFC.

Summary:
This affects other tools so the previous C++ API has been retained as a
deprecated function for the moment. Clang has been updated with a trivial
patch (not covered by the pre-commit review) to avoid breaking -Werror builds.
Other in-tree tools will be fixed with similar patches.

This continues the patch series to eliminate StringRef forms of GNU triples
from the internals of LLVM that began in r239036.

The first time this was committed it accidentally fixed an inconsistency in
triples in llvm-mc and this caused a failure. This inconsistency was fixed in
r239808.

Reviewers: rengolin

Reviewed By: rengolin

Subscribers: llvm-commits, rengolin

Differential Revision: http://reviews.llvm.org/D10366

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@239812 91177308-0d34-0410-b5e6-96231b3b80d8

9 years ago[mips] [IAS] Refactor symbol-address loading code into a helper function. NFC.
Toma Tabacu [Tue, 16 Jun 2015 12:16:24 +0000 (12:16 +0000)]
[mips] [IAS] Refactor symbol-address loading code into a helper function. NFC.

Reviewers: dsanders

Reviewed By: dsanders

Subscribers: llvm-commits

Differential Revision: http://reviews.llvm.org/D9523

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@239811 91177308-0d34-0410-b5e6-96231b3b80d8

9 years ago[llvm-mc] The object form of the GNU triple should be the same as the string form.
Daniel Sanders [Tue, 16 Jun 2015 09:57:38 +0000 (09:57 +0000)]
[llvm-mc] The object form of the GNU triple should be the same as the string form.

Summary:
GetTarget() may modify TripleName without also updating TheTriple.
This can lead to situations where the MCObjectStreamer has a different triple
to the rest of LLVM.

This inconsistency caused sparc-little-endian.s to pass on Windows because most
of LLVM had sparcel-pc-win32 while MCObjectStreamer had "". I believe the same
kind of thing was also true of Darwin.

Reviewers: rengolin

Reviewed By: rengolin

Subscribers: llvm-commits, rengolin, rafael

Differential Revision: http://reviews.llvm.org/D10450

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@239808 91177308-0d34-0410-b5e6-96231b3b80d8

9 years ago[MachineSink] Address post-commit review comments
Arnaud A. de Grandmaison [Tue, 16 Jun 2015 08:57:21 +0000 (08:57 +0000)]
[MachineSink] Address post-commit review comments

The successors cache is now a local variable, making it more visible that it
is only valid for the MBB being processed.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@239807 91177308-0d34-0410-b5e6-96231b3b80d8

9 years ago[AVX512] add integer min/max intrinsics support.
Asaf Badouh [Tue, 16 Jun 2015 08:39:27 +0000 (08:39 +0000)]
[AVX512] add integer min/max intrinsics support.

review:
http://reviews.llvm.org/D10439

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@239806 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoDisable llvm/test/CodeGen/MIR/machine-function.mir on x86 msc18 for now. Investigating.
NAKAMURA Takumi [Tue, 16 Jun 2015 06:57:35 +0000 (06:57 +0000)]
Disable llvm/test/CodeGen/MIR/machine-function.mir on x86 msc18 for now. Investigating.

The emission was as below;

  ---
  name:            foo
  alignment:       31428584
  exposesReturnsTwice: true
  hasInlineAsm:    false
  ...
  ---
  name:            bar
  alignment:       1701667182
  exposesReturnsTwice: false
  hasInlineAsm:    false
  ...
  ---
  name:            func
  alignment:       8
  exposesReturnsTwice: false
  hasInlineAsm:    false
  ...
  ---
  name:            func2
  alignment:       16
  exposesReturnsTwice: true
  hasInlineAsm:    true
  ...

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@239805 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agollvm/unittests/Support/Path.cpp: Use <windows.h> instead of <Windows.h>.
NAKAMURA Takumi [Tue, 16 Jun 2015 06:46:16 +0000 (06:46 +0000)]
llvm/unittests/Support/Path.cpp: Use <windows.h> instead of <Windows.h>.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@239804 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoX86: optimized i64 vector multiply with constant
Elena Demikhovsky [Tue, 16 Jun 2015 06:07:24 +0000 (06:07 +0000)]
X86: optimized i64 vector multiply with constant

When we multiply two 64-bit vectors, we extract lower and upper part and use the PMULUDQ instruction.
When one of the operands is a constant, the upper part may be zero, we know this at compile time.
Example: %a = mul <4 x i64> %b, <4 x i64> < i64 5, i64 5, i64 5, i64 5>.
I'm checking the value of the upper part and prevent redundant "multiply", "shift" and "add" operations.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@239802 91177308-0d34-0410-b5e6-96231b3b80d8

9 years ago[TableGen] Remove unused method declaration. NFC
Craig Topper [Tue, 16 Jun 2015 05:44:06 +0000 (05:44 +0000)]
[TableGen] Remove unused method declaration. NFC

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@239801 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoRevert 239795
Philip Reames [Tue, 16 Jun 2015 01:20:53 +0000 (01:20 +0000)]
Revert 239795

I forgot to update some clang test cases.  I'll fix and resubmit tomorrow.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@239800 91177308-0d34-0410-b5e6-96231b3b80d8

9 years ago[AArch64] Generalize extract-high DUP extension to MOVI/MVNI.
Ahmed Bougacha [Tue, 16 Jun 2015 01:18:14 +0000 (01:18 +0000)]
[AArch64] Generalize extract-high DUP extension to MOVI/MVNI.

These are really immediate DUPs, and suffer from the same problem
with long instructions with a high/2 variant (e.g. smull).

By extending a MOVI (or DUP, before this patch), we can avoid an ext
on the other operand of the long instruction, e.g. turning:
    ext.16b v0, v0, v0, #8
    movi.4h v1, #0x53
    smull.4s  v0, v0, v1
into:
    movi.8h v1, #0x53
    smull2.4s  v0, v0, v1

While there, add a now-necessary combine to fold (VT NVCAST (VT x)).

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@239799 91177308-0d34-0410-b5e6-96231b3b80d8

9 years ago[AArch64] Robustize neon-2velem-high test. NFC.
Ahmed Bougacha [Tue, 16 Jun 2015 01:05:39 +0000 (01:05 +0000)]
[AArch64] Robustize neon-2velem-high test. NFC.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@239798 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoMove logic from JumpThreading into LazyValue info to simplify caller.
Philip Reames [Tue, 16 Jun 2015 00:49:59 +0000 (00:49 +0000)]
Move logic from JumpThreading into LazyValue info to simplify caller.

This change is hopefully NFC. The only tricky part is that I changed the context instruction being used to the branch rather than the comparison. I believe both to be correct, but the branch is strictly more powerful. With the moved code, using the branch instruction is required for the basic block comparison test to return the same result. The previous code was able to directly access both the branch and the comparison where the revised code is not.

Differential Revision: http://reviews.llvm.org/D9652

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@239797 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agomodules: Add explicit dependency on intrinsics_gen
Duncan P. N. Exon Smith [Tue, 16 Jun 2015 00:44:12 +0000 (00:44 +0000)]
modules: Add explicit dependency on intrinsics_gen

`LLVM_ENABLE_MODULES` builds sometimes fail because `Intrinsics.td`
needs to regenerate `Instrinsics.h` before anyone can include anything
from the LLVM_IR module.  Represent the dependency explicitly to prevent
that.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@239796 91177308-0d34-0410-b5e6-96231b3b80d8

9 years ago[InstCombine] Propagate non-null facts to call parameters
Philip Reames [Tue, 16 Jun 2015 00:43:54 +0000 (00:43 +0000)]
[InstCombine] Propagate non-null facts to call parameters

If a parameter to a function is known non-null, use the existing parameter attributes to record that fact at the call site. This has no optimization benefit by itself - that I know of - but is an enabling change for http://reviews.llvm.org/D9129.

Differential Revision: http://reviews.llvm.org/D9132

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@239795 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agomodules: Mark CodeGen/DIEValues.def as a textual inclusion
Duncan P. N. Exon Smith [Tue, 16 Jun 2015 00:43:52 +0000 (00:43 +0000)]
modules: Mark CodeGen/DIEValues.def as a textual inclusion

Mark CodeGen/DIEValues.def as a textual inclusion to fix the
`LLVM_ENABLE_MODULES` build.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@239794 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agomodules: Move ProfileKinds to an anonymous namespace
Duncan P. N. Exon Smith [Tue, 16 Jun 2015 00:43:04 +0000 (00:43 +0000)]
modules: Move ProfileKinds to an anonymous namespace

Fix a build failure with `LLVM_ENABLE_MODULES` due to
`ProfileData::instr` conflicting with a function `instr()` in
`<curses.h>`.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@239793 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoMIR Serialization: Print and parse simple machine function attributes.
Alex Lorenz [Tue, 16 Jun 2015 00:10:47 +0000 (00:10 +0000)]
MIR Serialization: Print and parse simple machine function attributes.

This commit serializes the simple, scalar attributes from the
'MachineFunction' class.

Reviewers: Duncan P. N. Exon Smith

Differential Revision: http://reviews.llvm.org/D10449

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@239790 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoMIR Serialization: move the MIR printer out of the MIR printing pass.
Alex Lorenz [Mon, 15 Jun 2015 23:52:35 +0000 (23:52 +0000)]
MIR Serialization: move the MIR printer out of the MIR printing pass.

This commit decouples the MIR printer and the MIR printing pass so
that it will be possible to move the MIR printer into a separate
machine IR library later on.

Reviewers: Duncan P. N. Exon Smith

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@239788 91177308-0d34-0410-b5e6-96231b3b80d8

9 years ago[X86] Try to shorten dwarf CFI emission
Reid Kleckner [Mon, 15 Jun 2015 23:45:08 +0000 (23:45 +0000)]
[X86] Try to shorten dwarf CFI emission

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@239786 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoDebug Info IR: Switch DIObjCProperty to use DITypeRef.
Adrian Prantl [Mon, 15 Jun 2015 23:18:03 +0000 (23:18 +0000)]
Debug Info IR: Switch DIObjCProperty to use DITypeRef.
This is a prerequisite for turning on ODR type uniquing for ObjC++.

rdar://problem/21377883

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@239780 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoMIR Serialization: Create dummy functions when the MIR file doesn't have LLVM IR.
Alex Lorenz [Mon, 15 Jun 2015 23:07:38 +0000 (23:07 +0000)]
MIR Serialization: Create dummy functions when the MIR file doesn't have LLVM IR.

This commit creates a dummy LLVM IR function with one basic block and an unreachable
instruction for each parsed machine function when the MIR file doesn't have LLVM IR.
This change is required as the machine function analysis pass creates machine
functions only for the functions that are defined in the current LLVM module.

Reviewers: Duncan P. N. Exon Smith

Differential Revision: http://reviews.llvm.org/D10135

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@239778 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoMIR Serialization: Report an error when machine functions have the same name.
Alex Lorenz [Mon, 15 Jun 2015 22:23:23 +0000 (22:23 +0000)]
MIR Serialization: Report an error when machine functions have the same name.

This commit reports an error when the MIR parser encounters a machine
function with the name that is the same as the name of a different
machine function.

Reviewers: Duncan P. N. Exon Smith

Differential Revision: http://reviews.llvm.org/D10130

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@239774 91177308-0d34-0410-b5e6-96231b3b80d8

9 years agoAdd safestack attribute to LLVMAttribute enum and Go bindings. Correct
Peter Collingbourne [Mon, 15 Jun 2015 22:16:51 +0000 (22:16 +0000)]
Add safestack attribute to LLVMAttribute enum and Go bindings. Correct
constants in commented-out part of LLVMAttribute enum. Add tests that verify
that the safestack attribute is only allowed as a function attribute.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@239772 91177308-0d34-0410-b5e6-96231b3b80d8

9 years ago[Hexagon] Using readobj rather than objdump.
Colin LeMahieu [Mon, 15 Jun 2015 21:57:41 +0000 (21:57 +0000)]
[Hexagon] Using readobj rather than objdump.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@239770 91177308-0d34-0410-b5e6-96231b3b80d8

9 years ago[Hexagon] PC-relative offsets are relative to packet start rather than the offset...
Colin LeMahieu [Mon, 15 Jun 2015 21:52:13 +0000 (21:52 +0000)]
[Hexagon] PC-relative offsets are relative to packet start rather than the offset of the relocation.  Set relocation addend and check it's correct in the ELF.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@239769 91177308-0d34-0410-b5e6-96231b3b80d8

9 years ago[X86][SSE] Added tests for vector i8/i16 to f32/f64 conversions
Simon Pilgrim [Mon, 15 Jun 2015 21:49:31 +0000 (21:49 +0000)]
[X86][SSE] Added tests for vector i8/i16 to f32/f64 conversions

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@239767 91177308-0d34-0410-b5e6-96231b3b80d8