The SectionMemoryManager now supports (and requires) applying section-specific page permissions. Clients using this memory manager must call either MCJIT::finalizeObject() or SectionMemoryManager::applyPermissions() before executing JITed code.
See r168718 for changes from the previous implementation.
llvm-svn: 168721
The default for 64-bit PowerPC is small code model, in which TOC entries
must be addressable using a 16-bit offset from the TOC pointer. Additionally,
only TOC entries are addressed via the TOC pointer.
With medium code model, TOC entries and data sections can all be addressed
via the TOC pointer using a 32-bit offset. Cooperation with the linker
allows 16-bit offsets to be used when these are sufficient, reducing the
number of extra instructions that need to be executed. Medium code model
also does not generate explicit TOC entries in ".section toc" for variables
that are wholly internal to the compilation unit.
Consider a load of an external 4-byte integer. With small code model, the
compiler generates:
ld 3, .LC1@toc(2)
lwz 4, 0(3)
.section .toc,"aw",@progbits
.LC1:
.tc ei[TC],ei
With medium model, it instead generates:
addis 3, 2, .LC1@toc@ha
ld 3, .LC1@toc@l(3)
lwz 4, 0(3)
.section .toc,"aw",@progbits
.LC1:
.tc ei[TC],ei
Here .LC1@toc@ha is a relocation requesting the upper 16 bits of the
32-bit offset of ei's TOC entry from the TOC base pointer. Similarly,
.LC1@toc@l is a relocation requesting the lower 16 bits. Note that if
the linker determines that ei's TOC entry is within a 16-bit offset of
the TOC base pointer, it will replace the "addis" with a "nop", and
replace the "ld" with the identical "ld" instruction from the small
code model example.
Consider next a load of a function-scope static integer. For small code
model, the compiler generates:
ld 3, .LC1@toc(2)
lwz 4, 0(3)
.section .toc,"aw",@progbits
.LC1:
.tc test_fn_static.si[TC],test_fn_static.si
.type test_fn_static.si,@object
.local test_fn_static.si
.comm test_fn_static.si,4,4
For medium code model, the compiler generates:
addis 3, 2, test_fn_static.si@toc@ha
addi 3, 3, test_fn_static.si@toc@l
lwz 4, 0(3)
.type test_fn_static.si,@object
.local test_fn_static.si
.comm test_fn_static.si,4,4
Again, the linker may replace the "addis" with a "nop", calculating only
a 16-bit offset when this is sufficient.
Note that it would be more efficient for the compiler to generate:
addis 3, 2, test_fn_static.si@toc@ha
lwz 4, test_fn_static.si@toc@l(3)
The current patch does not perform this optimization yet. This will be
addressed as a peephole optimization in a later patch.
For the moment, the default code model for 64-bit PowerPC will remain the
small code model. We plan to eventually change the default to medium code
model, which matches current upstream GCC behavior. Note that the different
code models are ABI-compatible, so code compiled with different models will
be linked and execute correctly.
I've tested the regression suite and the application/benchmark test suite in
two ways: Once with the patch as submitted here, and once with additional
logic to force medium code model as the default. The tests all compile
cleanly, with one exception. The mandel-2 application test fails due to an
unrelated ABI compatibility with passing complex numbers. It just so happens
that small code model was incredibly lucky, in that temporary values in
floating-point registers held the expected values needed by the external
library routine that was called incorrectly. My current thought is to correct
the ABI problems with _Complex before making medium code model the default,
to avoid introducing this "regression."
Here are a few comments on how the patch works, since the selection code
can be difficult to follow:
The existing logic for small code model defines three pseudo-instructions:
LDtoc for most uses, LDtocJTI for jump table addresses, and LDtocCPT for
constant pool addresses. These are expanded by SelectCodeCommon(). The
pseudo-instruction approach doesn't work for medium code model, because
we need to generate two instructions when we match the same pattern.
Instead, new logic in PPCDAGToDAGISel::Select() intercepts the TOC_ENTRY
node for medium code model, and generates an ADDIStocHA followed by either
a LDtocL or an ADDItocL. These new node types correspond naturally to
the sequences described above.
The addis/ld sequence is generated for the following cases:
* Jump table addresses
* Function addresses
* External global variables
* Tentative definitions of global variables (common linkage)
The addis/addi sequence is generated for the following cases:
* Constant pool entries
* File-scope static global variables
* Function-scope static variables
Expanding to the two-instruction sequences at select time exposes the
instructions to subsequent optimization, particularly scheduling.
The rest of the processing occurs at assembly time, in
PPCAsmPrinter::EmitInstruction. Each of the instructions is converted to
a "real" PowerPC instruction. When a TOC entry needs to be created, this
is done here in the same manner as for the existing LDtoc, LDtocJTI, and
LDtocCPT pseudo-instructions (I factored out a new routine to handle this).
I had originally thought that if a TOC entry was needed for LDtocL or
ADDItocL, it would already have been generated for the previous ADDIStocHA.
However, at higher optimization levels, the ADDIStocHA may appear in a
different block, which may be assembled textually following the block
containing the LDtocL or ADDItocL. So it is necessary to include the
possibility of creating a new TOC entry for those two instructions.
Note that for LDtocL, we generate a new form of LD called LDrs. This
allows specifying the @toc@l relocation for the offset field of the LD
instruction (i.e., the offset is replaced by a SymbolLo relocation).
When the peephole optimization described above is added, we will need
to do similar things for all immediate-form load and store operations.
The seven "mcm-n.ll" test cases are kept separate because otherwise the
intermingling of various TOC entries and so forth makes the tests fragile
and hard to understand.
The above assumes use of an external assembler. For use of the
integrated assembler, new relocations are added and used by
PPCELFObjectWriter. Testing is done with "mcm-obj.ll", which tests for
proper generation of the various relocations for the same sequences
tested with the external assembler.
llvm-svn: 168708
Added in first optimization using fast-math flags to serve as an example for following optimizations. SimplifyInstruction will now try to optimize an fmul observing its FastMathFlags to see if it can fold multiply by zero when 'nnan' and 'nsz' flags are set.
llvm-svn: 168648
Added in bitcode enum for the serializing of fast-math flags. Added in the reading/writing of fast-math flags from the OptimizationFlags record for BinaryOps.
llvm-svn: 168646
Add in getter/setter methods for Instructions, allowing them to be the interface to FPMathOperator similarly to now NUS/NSW is handled.
llvm-svn: 168642
Created FastMathFlags convenience struct for the getting and setting of fast-math flags en masse. Added SubclassOptionalData bitfields and corresponding getters/setters to FPMathOperator for the various fast-math flags.
llvm-svn: 168641
- Widespread trailing space removal
- A dash of OCD spacing to block align enums
- joined a line that probably needed 80 cols a while back
llvm-svn: 168566
to support it. Original patch with the parsing and plumbing by the PaX team and
Roman Divacky. I added the bits in MCDwarf.cpp and the test.
llvm-svn: 168565
Necessary to give disassembler users (like darwin's otool) a possibility to
dlopen libLTO and still initialize the required LLVM bits. This used to go
through libMCDisassembler but that's a gross layering violation, the MC layer
can't pull in functions from the targets. Adding a function to libLTO is a bit
of a hack but not worse than exposing other disassembler bits from libLTO.
Fixes PR14362.
llvm-svn: 168545
Give MCCFIInstruction a single, private constructor and add helper static
methods that create each type of cfi instruction. This is is preparation
for changing its representation. The representation with a pair
MachineLocations older than MC and has been abused quiet a bit to support
more cfi instructions.
llvm-svn: 168532
I discovered a few more missing functions while migrating optimizations
from the simplify-libcalls pass to the instcombine (I already added some
in r167659).
llvm-svn: 168501
so that I can (someday) call SE->getSCEV without complaint.
No semantic change intended.
Patch from Preston Briggs <preston.briggs@gmail.com>.
llvm-svn: 168391
When code deletes the context, the AttributeImpls that the AttrListPtr points to
are now invalid. Therefore, instead of keeping a separate managed static for the
AttrListPtrs that's reference counted, move it into the LLVMContext and delete
it when deleting the AttributeImpls.
llvm-svn: 168354
The rationale is to get YAML filenames in diagnostics from
yaml::Stream::printError -- currently the filename is hard-coded as
"YAML" because there's no buffer information available.
Patch by Kim Gräsman!
llvm-svn: 168341
This patch moves the isInlineViable function from the InlineAlways pass into
the InlineCostAnalyzer and then changes the InlineCost computation to use that
simple check for always-inline functions. All the special-case checks for
AlwaysInline in the CallAnalyzer can then go away.
llvm-svn: 168300
positive.
In this particular case, R6 was being spilled by the register scavenger when it
was in fact dead. The isUsed function reported R6 as used because the R6_R7
alias was reserved (due to the fact that we've reserved R7 as the FP). The
solution is to only check if the original register (i.e., R6) isReserved and
not the aliases. The aliases are only checked to make sure they're available.
The test case is derived from one of the nightly tester benchmarks and is rather
intractable and difficult to reproduce, so I haven't included it.
rdar://12592448
llvm-svn: 168054
For global variables that get the same value stored into them
everywhere, GlobalOpt will replace them with a constant. The problem is
that a thread-local GlobalVariable looks like one value (the address of
the TLS var), but is different between threads.
This patch introduces Constant::isThreadDependent() which returns true
for thread-local variables and constants which depend on them (e.g. a GEP
into a thread-local array), and teaches GlobalOpt not to track such
values.
llvm-svn: 168037
This seems like redundant leftovers from r142288 - exposing
TargetData::parseSpecifier to LLParser - which got reverted. Removes
redunant td != NULL checks in parseSpecifier, and simplifies the
interface to parseSpecifier and init.
llvm-svn: 167924
Previously in a vector of pointers, the pointer couldn't be any pointer type,
it had to be a pointer to an integer or floating point type. This is a hassle
for dragonegg because the GCC vectorizer happily produces vectors of pointers
where the pointer is a pointer to a struct or whatever. Vector getelementptr
was restricted to just one index, but now that vectors of pointers can have
any pointer type it is more natural to allow arbitrary vector getelementptrs.
There is however the issue of struct GEPs, where if each lane chose different
struct fields then from that point on each lane will be working down into
unrelated types. This seems like too much pain for too little gain, so when
you have a vector struct index all the elements are required to be the same.
llvm-svn: 167828
This allows me to begin enabling (or backing out) misched by default
for one subtarget at a time. To run misched we typically want to:
- Disable SelectionDAG scheduling (use the source order scheduler)
- Enable more aggressive coalescing (until we decide to always run the coalescer this way)
- Enable MachineScheduler pass itself.
Disabling PostRA sched may follow for some subtargets.
llvm-svn: 167826
This patch migrates the math library call simplifications from the
simplify-libcalls pass into the instcombine library call simplifier.
I have typically migrated just one simplifier at a time, but the math
simplifiers are interdependent because:
1. CosOpt, PowOpt, and Exp2Opt all depend on UnaryDoubleFPOpt.
2. CosOpt, PowOpt, Exp2Opt, and UnaryDoubleFPOpt all depend on
the option -enable-double-float-shrink.
These two factors made migrating each of these simplifiers individually
more of a pain than it would be worth. So, I migrated them all together.
llvm-svn: 167815
If we have a type 'int a[1]' and a type 'int b[0]', the generated DWARF is the
same for both of them because we use the 'upper_bound' attribute. Instead use
the 'count' attrbute, which gives the correct number of elements in the array.
<rdar://problem/12566646>
llvm-svn: 167806
getNumContainedPasses() used to compute the size of the vector on demand. It is
called repeated in loops (such as runOnFunction()) and it can be updated while
inside the loop.
llvm-svn: 167759
Uses the infrastructure from r167742 to support clustering instructure
that the target processor can "fuse". e.g. cmp+jmp.
Next step: target hook implementations with test cases, and enable.
llvm-svn: 167744
This infrastructure is generally useful for any target that wants to
strongly prefer two instructions to be adjacent after scheduling.
A following checkin will add target-specific hooks with unit
tests. Then this feature will be enabled by default with misched.
llvm-svn: 167742
This adds support for weak DAG edges to the general scheduling
infrastructure in preparation for MachineScheduler support for
heuristics based on weak edges.
llvm-svn: 167738
In some cases the library call simplifier may need to replace instructions
other than the library call being simplified. In those cases it may be
necessary for clients of the simplifier to override how the replacements
are actually done. As such, a new overrideable method for replacing
instructions was added to LibCallSimplifier.
A new subclass of LibCallSimplifier is also defined which overrides
the instruction replacement method. This is because the instruction
combiner defines its own replacement method which updates the worklist
when instructions are replaced.
llvm-svn: 167681
In the process of migrating optimizations from the simplify-libcalls pass
to the instcombine pass I noticed that a few functions are missing from
the target library information. These functions need to be available for
querying in the instcombine library call simplifiers. More functions will
probably be added in the future as more simplifiers are migrated to
instcombine.
llvm-svn: 167659
- Add RTM code generation support throught 3 X86 intrinsics:
xbegin()/xend() to start/end a transaction region, and xabort() to abort a
tranaction region
llvm-svn: 167573
misched is disabled by default. With -enable-misched, these heuristics
balance the schedule to simultaneously avoid saturating processor
resources, expose ILP, and minimize register pressure. I've been
analyzing the performance of these heuristics on everything in the
llvm test suite in addition to a few other benchmarks. I would like
each heuristic check to be verified by a unit test, but I'm still
trying to figure out the best way to do that. The heuristics are still
in considerable flux, but as they are refined we should be rigorous
about unit testing the improvements.
llvm-svn: 167527
This patch adds the interface to expose events from MCJIT when an object is emitted or freed and implements the MCJIT functionality to send those events. The IntelJITEventListener implementation is left empty for now. It will be fleshed out in a future patch.
llvm-svn: 167475
Expose the processor resources defined by the machine model to the
scheduler and other clients through the TargetSchedule interface.
Normalize each resource count with respect to other kinds of
resources. This allows scheduling heuristics to balance resources
against other kinds of resources and latency.
llvm-svn: 167444
Prior to this patch RuntimeDyld attempted to re-apply relocations every time reassignSectionAddress was called (via MCJIT::mapSectionAddress). In addition to being inefficient and redundant, this led to a problem when a section was temporarily moved too far away from another section with a relative relocation referencing the section being moved. To fix this, I'm adding a new method (finalizeObject) which the client can call to indicate that it is finished rearranging section addresses so the relocations can safely be applied.
llvm-svn: 167400
is that the unit test doesn't have IntTy equal to APInt, instead it uses a class
derived from APInt. When, as in these lines, an IntTy& reference is returned
but is assigned to an APInt&, the compiler destroys the temporary the IntTy& was
referring to, leaving the APInt& referring to garbage. This causes the unittest
to fail systematically on my machine; it can also be caught by running the test
under valgrind.
llvm-svn: 167356
the inttoptr instruction. The conceptual model here is that
'getAddressSpace' refers to the address space of this instruction's
type. It just happens that for GEPs, that is always the same as the
pointer operand's address space. We want both names so that access
patterns can be consistent between different instruction types.
llvm-svn: 167229
compute the address space in the one place it was used.
Also write the getPointerAddressSpace member in terms of the
getPointerOperandType member.
llvm-svn: 167226
'@brief' doxygen markup to the now standard '\brief' markup form, in
conformance with the coding standards. This will let me continue to
write new comments in this form without making things inconsistent.
llvm-svn: 167225
politely and document this feature.
This simple API extension then allows us to write all of the
Instructions' address space query methods much more simply. No
functionality change intended here.
llvm-svn: 167223
r165941: Resubmit the changes to llvm core to update the functions to
support different pointer sizes on a per address space basis.
Despite this commit log, this change primarily changed stuff outside of
VMCore, and those changes do not carry any tests for correctness (or
even plausibility), and we have consistently found questionable or flat
out incorrect cases in these changes. Most of them are probably correct,
but we need to devise a system that makes it more clear when we have
handled the address space concerns correctly, and ideally each pass that
gets updated would receive an accompanying test case that exercises that
pass specificaly w.r.t. alternate address spaces.
However, from this commit, I have retained the new C API entry points.
Those were an orthogonal change that probably should have been split
apart, but they seem entirely good.
In several places the changes were very obvious cleanups with no actual
multiple address space code added; these I have not reverted when
I spotted them.
In a few other places there were merge conflicts due to a cleaner
solution being implemented later, often not using address spaces at all.
In those cases, I've preserved the new code which isn't address space
dependent.
This is part of my ongoing effort to clean out the partial address space
code which carries high risk and low test coverage, and not likely to be
finished before the 3.2 release looms closer. Duncan and I would both
like to see the above issues addressed before we return to these
changes.
llvm-svn: 167222
getIntPtrType support for multiple address spaces via a pointer type,
and also introduced a crasher bug in the constant folder reported in
PR14233.
These commits also contained several problems that should really be
addressed before they are re-committed. I have avoided reverting various
cleanups to the DataLayout APIs that are reasonable to have moving
forward in order to reduce the amount of churn, and minimize the number
of commits that were reverted. I've also manually updated merge
conflicts and manually arranged for the getIntPtrType function to stay
in DataLayout and to be defined in a plausible way after this revert.
Thanks to Duncan for working through this exact strategy with me, and
Nick Lewycky for tracking down the really annoying crasher this
triggered. (Test case to follow in its own commit.)
After discussing with Duncan extensively, and based on a note from
Micah, I'm going to continue to back out some more of the more
problematic patches in this series in order to ensure we go into the
LLVM 3.2 branch with a reasonable story here. I'll send a note to
llvmdev explaining what's going on and why.
Summary of reverted revisions:
r166634: Fix a compiler warning with an unused variable.
r166607: Add some cleanup to the DataLayout changes requested by
Chandler.
r166596: Revert "Back out r166591, not sure why this made it through
since I cancelled the command. Bleh, sorry about this!
r166591: Delete a directory that wasn't supposed to be checked in yet.
r166578: Add in support for getIntPtrType to get the pointer type based
on the address space.
llvm-svn: 167221
Explicitly allow composition of null sub-register indices, and handle
that common case in an inlinable stub.
Use a compressed table implementation instead of the previous nested
switches which generated pretty bad code.
llvm-svn: 167190
the MachineInstr MayLoad/MayLoad flags are based on the tablegen implementation.
For inline assembly, however, we need to compute these based on the constraints.
Revert r166929 as this is no longer needed, but leave the test case in place.
rdar://12033048 and PR13504
llvm-svn: 167040
When the switch-to-lookup tables transform landed in SimplifyCFG, it
was pointed out that this could be inappropriate for some targets.
Since there was no way at the time for the pass to know anything about
the target, an awkward reverse-transform was added in CodeGenPrepare
that turned lookup tables back into switches for some targets.
This patch uses the new TargetTransformInfo to determine if a
switch should be transformed, and removes
CodeGenPrepare::ConvertLoadToSwitch.
llvm-svn: 167011
checks to avoid performing compile-time arithmetic on PPCDoubleDouble.
Now that APFloat supports arithmetic on PPCDoubleDouble, those checks
are no longer needed, and we can treat the type like any other.
llvm-svn: 166958
wrapper returns a vector of integers when passed a vector of pointers) by having
getIntPtrType itself return a vector of integers in this case. Outside of this
wrapper, I didn't find anywhere in the codebase that was relying on the old
behaviour for vectors of pointers, so give this a whirl through the buildbots.
llvm-svn: 166939
We may need to change the way profile counter values are stored, but
saturation is the wrong thing to do. Just remove it for now.
Patch by Alastair Murray!
llvm-svn: 166938
Add getCostXXX calls for different families of opcodes, such as casts, arithmetic, cmp, etc.
Port the LoopVectorizer to the new API.
The LoopVectorizer now finds instructions which will remain uniform after vectorization. It uses this information when calculating the cost of these instructions.
llvm-svn: 166836
APInt::shl generated llvm.trap to guard against shifts greater than bit-width.
This was already checked with an assert, and there was a special case for
shifts equal to bit-width. Modify this check to catch shifts greater than or
equal to bit-width, so llvm.trap isn't generated.
Patch contributed by JF Bastien
llvm-svn: 166803
Enabled with -verify-scev. This could be extended significantly but hopefully
catches the common cases now. Note that it's not enabled by default in any
configuration because the way it tries to distinguish SCEVs is still fragile and
may produce false positives. Also the test-suite isn't clean yet, one example
is that it fails if a pass drops an NSW bit but it's still present in SCEV's
cached. Cleaning up all those cases will take some time.
llvm-svn: 166786
As discussed on IRC, add VectorTargetTransform::getNumberOfParts
to provide a stable interface to the vector legalization splitting factor.
llvm-svn: 166751
include/llvm/MC/MCTargetAsmParser.h:46:8: error: 'llvm::ParseInstructionInfo' has a field 'llvm::ParseInstructionInfo::AsmRewrites' whose type uses the anonymous namespace [-Werror]
llvm-svn: 166729
Most places can use PrintFatalError as the unwinding mechanism was not
used for anything other than printing the error. The single exception
was CodeGenDAGPatterns.cpp, where intermediate errors during type
resolution were ignored to simplify incremental platform development.
This use is replaced by an error flag in TreePattern and bailout earlier
in various places if it is set.
llvm-svn: 166712
Relationship maps are represented as InstrMapping records which are parsed by
TableGen and the information is used to construct mapping tables to represent
appropriate relations between instructions. These tables are emitted into
XXXGenInstrInfo.inc file along with the functions to query them.
Patch by Jyotsna Verma <jverma@codeaurora.org>.
llvm-svn: 166685
This patch adds initial PPC64 TOC MC object creation using the small mcmodel
(a single 64K TOC) adding the some TOC relocations (R_PPC64_TOC,
R_PPC64_TOC16, and R_PPC64_TOC16DS).
The addition of 'undefinedExplicitRelSym' hook on 'MCELFObjectTargetWriter'
is meant to avoid the creation of an unreferenced ".TOC." symbol (used in
the .odp creation) as well to set the R_PPC64_TOC relocation target as the
temporary ".TOC." symbol. On PPC64 ABI, the R_PPC64_TOC relocation should
not point to any symbol.
llvm-svn: 166677
a bunch of errors for all the Operator subclasses such as:
include/llvm/Operator.h:76:7: error: deleted function 'virtual llvm::OverflowingBinaryOperator::~OverflowingBinaryOperator()'
include/llvm/Operator.h:43:3: error: overriding non-deleted function 'virtual llvm::Operator::~Operator()'
include/llvm/Operator.h:76:7: error: 'virtual llvm::OverflowingBinaryOperator::~OverflowingBinaryOperator()' is implicitly deleted because the default definition would be ill-formed:
include/llvm/Operator.h:43:3: error: 'virtual llvm::Operator::~Operator()' is private
include/llvm/Operator.h:76:7: error: within this context
llvm-svn: 166611
called. Provide an (asserting) definition of Operator's private destructor.
Remove destructors from all classes derived from Operator. We don't need them
for safety, because their implicit definitions would be ill-formed (they'd call
Operator's private destructor), and we don't need them to avoid emitting
vtables, because we don't do anything with Operator subclasses which would
trigger vtable instantiation.
The Operator hierarchy is still a complete disaster with regard to undefined
behavior, but this at least allows LLVM to link when using Clang's
-fcatch-undefined-behavior with a new vptr-based type checking mechanism.
llvm-svn: 166530
Per the October 12, 2012 Proposal for annotated disassembly output sent out by
Jim Grosbach this set of changes implements this for X86 and arm. The llvm-mc
tool now has a -mdis option to produced the marked up disassembly and a couple
of small example test cases have been added.
rdar://11764962
llvm-svn: 166445
Patch by Quentin Colombet <qcolombet@apple.com>
Original description:
"""
The attached patch is the first step to have a better control on Oz related optimizations.
The Oz optimization level focuses on code size, thus I propose to add an attribute called ForceSizeOpt.
"""
llvm-svn: 166422
a memory operand. Retain this information and then add the sizing directives
to the IR. This allows the backend to do proper instruction selection.
llvm-svn: 166316
which is supposed to consistently raise SIGTRAP across all systems. In contrast,
__builtin_trap() behave differently on different systems. e.g. it raises SIGTRAP on ARM, and
SIGILL on X86. The purpose of __builtin_debugtrap() is to consistently provide "trap"
functionality, in the mean time preserve the compatibility with on gcc on __builtin_trap().
The X86 backend is already able to handle debugtrap(). This patch is to:
1) make front-end recognize "__builtin_debugtrap()" (emboddied in the one-line change to Clang).
2) In DAG legalization phase, by default, "debugtrap" will be replaced with "trap", which
make the __builtin_debugtrap() "available" to all existing ports without the hassle of
changing their code.
3) If trap-function is specified (via -trap-func=xyz to llc), both __builtin_debugtrap() and
__builtin_trap() will be expanded into the function call of the specified trap function.
This behavior may need change in the future.
The provided testing-case is to make sure 2) and 3) are working for ARM port, and we
already have a testing case for x86.
llvm-svn: 166300
Andrew Trick