oota-llvm.git
12 years agoRevert "Fix a quadratic algorithm in MachineBranchProbabilityInfo."
Jakob Stoklund Olesen [Wed, 8 Aug 2012 01:10:31 +0000 (01:10 +0000)]
Revert "Fix a quadratic algorithm in MachineBranchProbabilityInfo."

It caused an assertion failure when compiling consumer-typeset.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@161463 91177308-0d34-0410-b5e6-96231b3b80d8

12 years agoX86: enable CSE between CMP and SUB
Manman Ren [Wed, 8 Aug 2012 00:51:41 +0000 (00:51 +0000)]
X86: enable CSE between CMP and SUB

We perform the following:
1> Use SUB instead of CMP for i8,i16,i32 and i64 in ISel lowering.
2> Modify MachineCSE to correctly handle implicit defs.
3> Convert SUB back to CMP if possible at peephole.

Removed pattern matching of (a>b) ? (a-b):0 and like, since they are handled
by peephole now.

rdar://11873276

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@161462 91177308-0d34-0410-b5e6-96231b3b80d8

12 years agoDon't scan physreg use-def chains looking for a PIC base.
Jakob Stoklund Olesen [Wed, 8 Aug 2012 00:40:47 +0000 (00:40 +0000)]
Don't scan physreg use-def chains looking for a PIC base.

We can't rematerialize a PIC base after register allocation anyway, and
scanning physreg use-def chains is very expensive in a function with
many calls.

<rdar://problem/12047515>

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@161461 91177308-0d34-0410-b5e6-96231b3b80d8

12 years agoFix a quadratic algorithm in MachineBranchProbabilityInfo.
Jakob Stoklund Olesen [Wed, 8 Aug 2012 00:20:37 +0000 (00:20 +0000)]
Fix a quadratic algorithm in MachineBranchProbabilityInfo.

The getSumForBlock function was quadratic in the number of successors
because getSuccWeight would perform a linear search for an already known
iterator.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@161460 91177308-0d34-0410-b5e6-96231b3b80d8

12 years agoAvoid recomputing the unique exit blocks and their insert points when doing
Dan Gohman [Wed, 8 Aug 2012 00:00:26 +0000 (00:00 +0000)]
Avoid recomputing the unique exit blocks and their insert points when doing
multiple scalar promotions on a single loop. This also has the effect of
preserving the order of stores sunk out of loops, which is aesthetically
pleasing, and it happens to fix the testcase in PR13542, though it doesn't
fix the underlying problem.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@161459 91177308-0d34-0410-b5e6-96231b3b80d8

12 years agoSkip tied operand pairs that already have the same register.
Jakob Stoklund Olesen [Tue, 7 Aug 2012 22:47:06 +0000 (22:47 +0000)]
Skip tied operand pairs that already have the same register.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@161454 91177308-0d34-0410-b5e6-96231b3b80d8

12 years agoAdd SelectionDAG::getTargetIndex.
Jakob Stoklund Olesen [Tue, 7 Aug 2012 22:37:05 +0000 (22:37 +0000)]
Add SelectionDAG::getTargetIndex.

This adds support for TargetIndex operands during isel. The meaning of
these (index, offset, flags) operands is entirely defined by the target.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@161453 91177308-0d34-0410-b5e6-96231b3b80d8

12 years agoFix a serious typo in InstCombine's optimization of comparisons.
Bob Wilson [Tue, 7 Aug 2012 22:35:16 +0000 (22:35 +0000)]
Fix a serious typo in InstCombine's optimization of comparisons.

An unsigned value converted to floating-point will always be greater than
a negative constant.  Unfortunately InstCombine reversed the check so that
unsigned values were being optimized to always be greater than all positive
floating-point constants.  <rdar://problem/12029145>

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@161452 91177308-0d34-0410-b5e6-96231b3b80d8

12 years agoX86 cmp lowering is looking past truncate on the condition node. It should only
Evan Cheng [Tue, 7 Aug 2012 22:21:00 +0000 (22:21 +0000)]
X86 cmp lowering is looking past truncate on the condition node. It should only
do so when the high bits are known zero. This caused a subtle miscompilation.

rdar://12027825

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@161451 91177308-0d34-0410-b5e6-96231b3b80d8

12 years agoFor non-Darwin platforms, we want to generate stack protectors only for
Bill Wendling [Tue, 7 Aug 2012 20:59:05 +0000 (20:59 +0000)]
For non-Darwin platforms, we want to generate stack protectors only for
character arrays. This is in line with what GCC does.
<rdar://problem/10529227>

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@161446 91177308-0d34-0410-b5e6-96231b3b80d8

12 years agoAdd a new kind of MachineOperand: MO_TargetIndex.
Jakob Stoklund Olesen [Tue, 7 Aug 2012 18:56:39 +0000 (18:56 +0000)]
Add a new kind of MachineOperand: MO_TargetIndex.

A target index operand looks a lot like a constant pool reference, but
it is completely target-defined. It contains the 8-bit TargetFlags, a
32-bit index, and a 64-bit offset. It is preserved by all code generator
passes.

TargetIndex operands can be used to carry target-specific information in
cases where immediate operands won't suffice.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@161441 91177308-0d34-0410-b5e6-96231b3b80d8

12 years agoEnable lazy compilation in MCJIT
Andrew Kaylor [Tue, 7 Aug 2012 18:33:00 +0000 (18:33 +0000)]
Enable lazy compilation in MCJIT

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@161438 91177308-0d34-0410-b5e6-96231b3b80d8

12 years agoFix a couple of typos.
Jakob Stoklund Olesen [Tue, 7 Aug 2012 18:32:57 +0000 (18:32 +0000)]
Fix a couple of typos.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@161437 91177308-0d34-0410-b5e6-96231b3b80d8

12 years agoRemove empty directory.
Eli Friedman [Tue, 7 Aug 2012 18:06:01 +0000 (18:06 +0000)]
Remove empty directory.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@161434 91177308-0d34-0410-b5e6-96231b3b80d8

12 years agoAdd trace accessor methods, implement primitive if-conversion heuristic.
Jakob Stoklund Olesen [Tue, 7 Aug 2012 18:02:19 +0000 (18:02 +0000)]
Add trace accessor methods, implement primitive if-conversion heuristic.

Compare the critical paths of the two traces through an if-conversion
candidate. If the difference is larger than the branch brediction
penalty, reject the if-conversion. If would never pay.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@161433 91177308-0d34-0410-b5e6-96231b3b80d8

12 years agoTidy up a bit.
Jim Grosbach [Tue, 7 Aug 2012 17:53:14 +0000 (17:53 +0000)]
Tidy up a bit.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@161430 91177308-0d34-0410-b5e6-96231b3b80d8

12 years agoThe dominance computation already has logic for computing if an edge dominates
Rafael Espindola [Tue, 7 Aug 2012 17:30:46 +0000 (17:30 +0000)]
The dominance computation already has logic for computing if an edge dominates
a use or a BB, but it is inline in the handling of the invoke instruction.

This patch refactors it so that it can be used in other cases. For example, in

define i32 @f(i32 %x) {
bb0:
  %cmp = icmp eq i32 %x, 0
  br i1 %cmp, label %bb2, label %bb1
bb1:
  br label %bb2
bb2:
  %cond = phi i32 [ %x, %bb0 ], [ 0, %bb1 ]
  %foo = add i32 %cond, %x
  ret i32 %foo
}

GVN should be able to replace %x with 0 in any use that is dominated by the
true edge out of bb0. In the above example the only such use is the one in
the phi.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@161429 91177308-0d34-0410-b5e6-96231b3b80d8

12 years agoAdd a comment about mftb vs. mfspr on PPC.
Hal Finkel [Tue, 7 Aug 2012 17:04:20 +0000 (17:04 +0000)]
Add a comment about mftb vs. mfspr on PPC.

Thanks to Alex Rosenberg for the suggestion.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@161428 91177308-0d34-0410-b5e6-96231b3b80d8

12 years agoFix the representation of debug line table in DebugInfo LLVM library,
Alexey Samsonov [Tue, 7 Aug 2012 11:46:57 +0000 (11:46 +0000)]
Fix the representation of debug line table in DebugInfo LLVM library,
and "instruction address -> file/line" lookup.

Instead of plain collection of rows, debug line table for compilation unit is now
treated as the number of row ranges, describing sequences (series of contiguous machine
instructions). The sequences are not always listed in the order of increasing
address, so previously used std::lower_bound() sometimes produced wrong results.
Now the instruction address lookup consists of two stages: finding the correct
sequence, and searching for address in range of rows for this sequence.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@161414 91177308-0d34-0410-b5e6-96231b3b80d8

12 years agoPR13095: Give an inline cost bonus to functions using byval arguments.
Benjamin Kramer [Tue, 7 Aug 2012 11:13:19 +0000 (11:13 +0000)]
PR13095: Give an inline cost bonus to functions using byval arguments.

We give a bonus for every argument because the argument setup is not needed
anymore when the function is inlined. With this patch we interpret byval
arguments as a compact representation of many arguments. The byval argument
setup is implemented in the backend as an inline memcpy, so to model the
cost as accurately as possible we take the number of pointer-sized elements
in the byval argument and give a bonus of 2 instructions for every one of
those. The bonus is capped at 8 elements, which is the number of stores
at which the x86 backend switches from an expanded inline memcpy to a real
memcpy. It would be better to use the real memcpy threshold from the backend,
but it's not available via TargetData.

This change brings the performance of c-ray in line with gcc 4.7. The included
test case tries to reproduce the c-ray problem to catch regressions for this
benchmark early, its performance is dominated by the inline decision of a
specific call.

This only has a small impact on most code, more on x86 and arm than on x86_64
due to the way the ABI works. When building LLVM for x86 it gives a small
inline cost boost to virtually any function using StringRef or STL allocators,
but only a 0.01% increase in overall binary size. The size of gcc compiled by
clang actually shrunk by a couple bytes with this patch applied, but not
significantly.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@161413 91177308-0d34-0410-b5e6-96231b3b80d8

12 years agoFix PR13412, a nasty miscompile due to the interleaved
Chandler Carruth [Tue, 7 Aug 2012 10:59:59 +0000 (10:59 +0000)]
Fix PR13412, a nasty miscompile due to the interleaved
instsimplify+inline strategy.

The crux of the problem is that instsimplify was reasonably relying on
an invariant that is true within any single function, but is no longer
true mid-inline the way we use it. This invariant is that an argument
pointer != a local (alloca) pointer.

The fix is really light weight though, and allows instsimplify to be
resiliant to these situations: when checking the relation ships to
function arguments, ensure that the argumets come from the same
function. If they come from different functions, then none of these
assumptions hold. All credit to Benjamin Kramer for coming up with this
clever solution to the problem.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@161410 91177308-0d34-0410-b5e6-96231b3b80d8

12 years agoAdd a much more conservative strategy for aligning branch targets.
Chandler Carruth [Tue, 7 Aug 2012 09:45:24 +0000 (09:45 +0000)]
Add a much more conservative strategy for aligning branch targets.
Previously, MBP essentially aligned every branch target it could. This
bloats code quite a bit, especially non-looping code which has no real
reason to prefer aligned branch targets so heavily.

As Andy said in review, it's still a bit odd to do this without a real
cost model, but this at least has much more plausible heuristics.

Fixes PR13265.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@161409 91177308-0d34-0410-b5e6-96231b3b80d8

12 years agoMachineCSE: Update the heuristics for isProfitableToCSE.
Manman Ren [Tue, 7 Aug 2012 06:16:46 +0000 (06:16 +0000)]
MachineCSE: Update the heuristics for isProfitableToCSE.

If the result of a common subexpression is used at all uses of the candidate
expression, CSE should not increase the live range of the common subexpression.

rdar://11393714 and rdar://11819721

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@161396 91177308-0d34-0410-b5e6-96231b3b80d8

12 years agoRevert r161371. Removing the 'const' before Type is a "good thing".
Bill Wendling [Tue, 7 Aug 2012 05:51:59 +0000 (05:51 +0000)]
Revert r161371. Removing the 'const' before Type is a "good thing".

--- Reverse-merging r161371 into '.':
U    include/llvm/Target/TargetData.h
U    lib/Target/TargetData.cpp

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@161394 91177308-0d34-0410-b5e6-96231b3b80d8

12 years agoThe define for 64 bit sign extension neglected to
Jack Carter [Tue, 7 Aug 2012 00:35:22 +0000 (00:35 +0000)]
The define for 64 bit sign extension neglected to
initialize fields of the class that it used.

The result was nonsense code.

Before:
0000000000000000 <foo>:
   0:    00441100     0x441100
   4:    03e00008     jr    ra
   8:    00000000     nop

After:
0000000000000000 <foo>:
   0:    00041000     sll    v0,a0,0x0
   4:    03e00008     jr    ra
   8:    00000000     nop

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@161377 91177308-0d34-0410-b5e6-96231b3b80d8

12 years agoConstify the Type parameter to some methods (which are const anyway).
Bill Wendling [Tue, 7 Aug 2012 00:26:35 +0000 (00:26 +0000)]
Constify the Type parameter to some methods (which are const anyway).

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@161371 91177308-0d34-0410-b5e6-96231b3b80d8

12 years agoAllow x86 subtargets to use the GenericModel defined in X86Schedule.td.
Andrew Trick [Tue, 7 Aug 2012 00:25:30 +0000 (00:25 +0000)]
Allow x86 subtargets to use the GenericModel defined in X86Schedule.td.

This allows codegen passes to query properties like
InstrItins->SchedModel->IssueWidth. It also ensure's that
computeOperandLatency returns the X86 defaults for loads and "high
latency ops". This should have no significant impact on existing
schedulers because X86 defaults happen to be the same as global
defaults.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@161370 91177308-0d34-0410-b5e6-96231b3b80d8

12 years agoMips relocation R_MIPS_64 relocates a 64 bit double word.
Jack Carter [Tue, 7 Aug 2012 00:01:14 +0000 (00:01 +0000)]
Mips relocation R_MIPS_64 relocates a 64 bit double word.

I hit this in a very large program (spirit.cpp), but
have not figured out how to make a small make check
test for it.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@161366 91177308-0d34-0410-b5e6-96231b3b80d8

12 years agoThe Mips64InstrInfo.td definitions DynAlloc64 LEA_ADDiu64
Jack Carter [Mon, 6 Aug 2012 23:29:06 +0000 (23:29 +0000)]
The Mips64InstrInfo.td definitions DynAlloc64 LEA_ADDiu64
were using a class defined for 32 bit instructions and
thus the instruction was for addiu instead of daddiu.

This was corrected by adding the instruction opcode as a
field in the  base class to be filled in by the defs.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@161359 91177308-0d34-0410-b5e6-96231b3b80d8

12 years agoReduce indentation by early exiting.
Bill Wendling [Mon, 6 Aug 2012 22:52:45 +0000 (22:52 +0000)]
Reduce indentation by early exiting.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@161356 91177308-0d34-0410-b5e6-96231b3b80d8

12 years agoFix typo.
Jakob Stoklund Olesen [Mon, 6 Aug 2012 22:34:51 +0000 (22:34 +0000)]
Fix typo.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@161354 91177308-0d34-0410-b5e6-96231b3b80d8

12 years agoAdd a way to grab the target options from the LTO command line.
Bill Wendling [Mon, 6 Aug 2012 21:34:54 +0000 (21:34 +0000)]
Add a way to grab the target options from the LTO command line.

When the command line target options were removed from the LLVM libraries, LTO
lost its ability to specify things like `-disable-fp-elim'. Add this back by
adding the command line variables to the `lto' project.
<rdar://problem/12038729>

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@161353 91177308-0d34-0410-b5e6-96231b3b80d8

12 years agoMips relocations R_MIPS_HIGHER and R_MIPS_HIGHEST.
Jack Carter [Mon, 6 Aug 2012 21:26:03 +0000 (21:26 +0000)]
Mips relocations R_MIPS_HIGHER and R_MIPS_HIGHEST.

These 2 relocations gain access to the
highest and the second highest 16 bits
of a 64 bit object.

R_MIPS_HIGHER %higher(A+S)
The %higher(x) function is [ (((long long) x + 0x80008000LL) >> 32) & 0xffff ].

R_MIPS_HIGHEST %highest(A+S)
The %highest(x) function is [ (((long long) x + 0x800080008000LL) >> 48) & 0xffff ].

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@161348 91177308-0d34-0410-b5e6-96231b3b80d8

12 years agoMFTB on PPC64 should really be encoded using MFSPR.
Hal Finkel [Mon, 6 Aug 2012 21:21:44 +0000 (21:21 +0000)]
MFTB on PPC64 should really be encoded using MFSPR.

The MFTB instruction itself is being phased out, and its functionality
is provided by MFSPR. According to the ISA docs, using MFSPR works on all known
chips except for the 601 (which did not have a timebase register anyway)
and the POWER3.

Thanks to Adhemerval Zanella for pointing this out!

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@161346 91177308-0d34-0410-b5e6-96231b3b80d8

12 years agoAdd support for the OpenBSD for Bitrig.
Eric Christopher [Mon, 6 Aug 2012 20:52:18 +0000 (20:52 +0000)]
Add support for the OpenBSD for Bitrig.

Patch by David Hill.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@161344 91177308-0d34-0410-b5e6-96231b3b80d8

12 years agoFix MIPS DSP Rev1 intrinsics memory properties.
Simon Atanasyan [Mon, 6 Aug 2012 19:47:47 +0000 (19:47 +0000)]
Fix MIPS DSP Rev1 intrinsics memory properties.
The patch reviewed by Akira Hatanaka.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@161332 91177308-0d34-0410-b5e6-96231b3b80d8

12 years agoPut up warning signs around MO::getNextOperandForReg().
Jakob Stoklund Olesen [Mon, 6 Aug 2012 18:48:43 +0000 (18:48 +0000)]
Put up warning signs around MO::getNextOperandForReg().

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@161329 91177308-0d34-0410-b5e6-96231b3b80d8

12 years agoRemove empty overrides of processFunctionBeforeFrameFinalized().
Roman Divacky [Mon, 6 Aug 2012 18:14:18 +0000 (18:14 +0000)]
Remove empty overrides of processFunctionBeforeFrameFinalized().

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@161328 91177308-0d34-0410-b5e6-96231b3b80d8

12 years agoImplement proper handling for pcmpistri/pcmpestri intrinsics. Requires custom handlin...
Craig Topper [Mon, 6 Aug 2012 06:22:36 +0000 (06:22 +0000)]
Implement proper handling for pcmpistri/pcmpestri intrinsics. Requires custom handling in DAGISelToDAG due to limitations in TableGen's implicit def handling. Fixes PR11305.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@161318 91177308-0d34-0410-b5e6-96231b3b80d8

12 years agoUpdate test to check for r161305
Craig Topper [Sun, 5 Aug 2012 09:06:28 +0000 (09:06 +0000)]
Update test to check for r161305

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@161307 91177308-0d34-0410-b5e6-96231b3b80d8

12 years agoRemove custom inserter for MWAIT. It doesn't do anything that couldn't be represented...
Craig Topper [Sun, 5 Aug 2012 00:36:57 +0000 (00:36 +0000)]
Remove custom inserter for MWAIT. It doesn't do anything that couldn't be represented in a pattern.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@161306 91177308-0d34-0410-b5e6-96231b3b80d8

12 years agoUse a COPY node instead of an explicit MOVA opcode in the custom insterter for pcmpes...
Craig Topper [Sun, 5 Aug 2012 00:17:48 +0000 (00:17 +0000)]
Use a COPY node instead of an explicit MOVA opcode in the custom insterter for pcmpestrm/pcmpistrm. Allows the register allocator to handle it better and prevent wasted identity moves.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@161305 91177308-0d34-0410-b5e6-96231b3b80d8

12 years agoAdd readcyclecounter lowering on PPC64.
Hal Finkel [Sat, 4 Aug 2012 14:10:46 +0000 (14:10 +0000)]
Add readcyclecounter lowering on PPC64.

On PPC64, this can be done with a simple TableGen pattern.
To enable this, I've added the (otherwise missing) readcyclecounter
SDNode definition to TargetSelectionDAG.td.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@161302 91177308-0d34-0410-b5e6-96231b3b80d8

12 years agoSkip impdef regs during eabi save/restore list emission to workaround PR11902
Anton Korobeynikov [Sat, 4 Aug 2012 13:25:58 +0000 (13:25 +0000)]
Skip impdef regs during eabi save/restore list emission to workaround PR11902

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@161301 91177308-0d34-0410-b5e6-96231b3b80d8

12 years agoRecognize vst1.64 / vld1.64 with 3 and 4 regs as load from / store to stack stuff
Anton Korobeynikov [Sat, 4 Aug 2012 13:22:14 +0000 (13:22 +0000)]
Recognize vst1.64 / vld1.64 with 3 and 4 regs as load from / store to stack stuff
(this corresponds by spilling/reloading regs in DTriple / DQuad reg classes).
No testcase, found by inspection.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@161300 91177308-0d34-0410-b5e6-96231b3b80d8

12 years agoAdd stack spill / reload instructions for DTriple and DQuad register classes, which
Anton Korobeynikov [Sat, 4 Aug 2012 13:16:12 +0000 (13:16 +0000)]
Add stack spill / reload instructions for DTriple and DQuad register classes, which
were missed for no reason. This fixes PR13377

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@161299 91177308-0d34-0410-b5e6-96231b3b80d8

12 years agoRemove extraneous ';'.
Bill Wendling [Sat, 4 Aug 2012 10:31:40 +0000 (10:31 +0000)]
Remove extraneous ';'.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@161298 91177308-0d34-0410-b5e6-96231b3b80d8

12 years agoUpdate cmake build.
Benjamin Kramer [Sat, 4 Aug 2012 09:52:39 +0000 (09:52 +0000)]
Update cmake build.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@161297 91177308-0d34-0410-b5e6-96231b3b80d8

12 years agoPostpone the deletion of the old name in StructType::setName to allow using a slice...
Benjamin Kramer [Sat, 4 Aug 2012 09:47:02 +0000 (09:47 +0000)]
Postpone the deletion of the old name in StructType::setName to allow using a slice of the old name.

Fixes PR13522. Add a rudimentary unit test to exercise the behavior.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@161296 91177308-0d34-0410-b5e6-96231b3b80d8

12 years ago[CMake] add_lit_target: Remove comments about add_dependencies. It is not a bug in...
NAKAMURA Takumi [Sat, 4 Aug 2012 02:14:44 +0000 (02:14 +0000)]
[CMake] add_lit_target: Remove comments about add_dependencies. It is not a bug in cmake that add_custom_target(DEPENDS) would not accept targets but file-level dependencies.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@161295 91177308-0d34-0410-b5e6-96231b3b80d8

12 years agollc: Try to suppress failures since r161262 .
NAKAMURA Takumi [Sat, 4 Aug 2012 01:29:37 +0000 (01:29 +0000)]
llc: Try to suppress failures since r161262 .

FIXME: Fix several tests on i686-win32 due to lacking of many libraries.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@161292 91177308-0d34-0410-b5e6-96231b3b80d8

12 years agoDelete a dead variable.
Jakob Stoklund Olesen [Sat, 4 Aug 2012 00:04:03 +0000 (00:04 +0000)]
Delete a dead variable.

TwoAddressInstructionPass doesn't remat any more.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@161285 91177308-0d34-0410-b5e6-96231b3b80d8

12 years agoTwoAddressInstructionPass refactoring: Extract another method.
Jakob Stoklund Olesen [Fri, 3 Aug 2012 23:57:58 +0000 (23:57 +0000)]
TwoAddressInstructionPass refactoring: Extract another method.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@161284 91177308-0d34-0410-b5e6-96231b3b80d8

12 years agoRefactor and check "onlyReadsMemory" before optimizing builtins.
Bob Wilson [Fri, 3 Aug 2012 23:29:17 +0000 (23:29 +0000)]
Refactor and check "onlyReadsMemory" before optimizing builtins.

This patch is mostly just refactoring a bunch of copy-and-pasted code, but
it also adds a check that the call instructions are readnone or readonly.
That check was already present for sin, cos, sqrt, log2, and exp2 calls, but
it was missing for the rest of the builtins being handled in this code.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@161282 91177308-0d34-0410-b5e6-96231b3b80d8

12 years agoTwoAddressInstructionPass refactoring: Extract a method.
Jakob Stoklund Olesen [Fri, 3 Aug 2012 23:25:45 +0000 (23:25 +0000)]
TwoAddressInstructionPass refactoring: Extract a method.

No functional change intended, except replacing a DenseMap with a
SmallDenseMap which should behave identically.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@161281 91177308-0d34-0410-b5e6-96231b3b80d8

12 years agoBegin adding support for updating LiveIntervals in TwoAddressInstructionPass.
Jakob Stoklund Olesen [Fri, 3 Aug 2012 22:58:34 +0000 (22:58 +0000)]
Begin adding support for updating LiveIntervals in TwoAddressInstructionPass.

This is far from complete, and only changes behavior when the
-early-live-intervals flag is passed to llc.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@161273 91177308-0d34-0410-b5e6-96231b3b80d8

12 years ago1. Redo mips16 instructions to avoid multiple opcodes for same instruction.
Akira Hatanaka [Fri, 3 Aug 2012 22:57:02 +0000 (22:57 +0000)]
1. Redo mips16 instructions to avoid multiple opcodes for same instruction.
Change these to patterns.
2. Add another 16 instructions.

Patch by Reed Kotler.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@161272 91177308-0d34-0410-b5e6-96231b3b80d8

12 years agoAdd an experimental -early-live-intervals option.
Jakob Stoklund Olesen [Fri, 3 Aug 2012 22:12:54 +0000 (22:12 +0000)]
Add an experimental -early-live-intervals option.

This option runs LiveIntervals before TwoAddressInstructionPass which
will eventually learn to exploit and update the analysis.

Eventually, LiveIntervals will run before PHIElimination, and we can get
rid of LiveVariables.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@161270 91177308-0d34-0410-b5e6-96231b3b80d8

12 years agoDelete merged physreg copies in joinReservedPhysReg().
Jakob Stoklund Olesen [Fri, 3 Aug 2012 22:12:51 +0000 (22:12 +0000)]
Delete merged physreg copies in joinReservedPhysReg().

Previously, the identity copy would survive through register allocation
before it was removed by the rewriter.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@161269 91177308-0d34-0410-b5e6-96231b3b80d8

12 years agoTry to reduce the compile time impact of r161232.
Bob Wilson [Fri, 3 Aug 2012 21:26:24 +0000 (21:26 +0000)]
Try to reduce the compile time impact of r161232.

The previous change caused fast isel to not attempt handling any calls to
builtin functions.  That included things like "printf" and caused some
noticable regressions in compile time.  I wanted to avoid having fast isel
keep a separate list of functions that had to be kept in sync with what the
code in SelectionDAGBuilder.cpp was handling.  I've resolved that here by
moving the list into TargetLibraryInfo.  This is somewhat redundant in
SelectionDAGBuilder but it will ensure that we keep things consistent.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@161263 91177308-0d34-0410-b5e6-96231b3b80d8

12 years agoFix memcmp code-gen to honor -fno-builtin.
Bob Wilson [Fri, 3 Aug 2012 21:26:18 +0000 (21:26 +0000)]
Fix memcmp code-gen to honor -fno-builtin.

I noticed that SelectionDAGBuilder::visitCall was missing a check for memcmp
in TargetLibraryInfo, so that it would use custom code for memcmp calls even
with -fno-builtin.  I also had to add a new -disable-simplify-libcalls option
to llc so that I could write a test for this.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@161262 91177308-0d34-0410-b5e6-96231b3b80d8

12 years agoCompletely eliminate VNInfo flags.
Jakob Stoklund Olesen [Fri, 3 Aug 2012 20:59:32 +0000 (20:59 +0000)]
Completely eliminate VNInfo flags.

The 'unused' state of a value number can be represented as an invalid
def SlotIndex. This also exposed code that shouldn't have been looking
at unused value VNInfos.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@161258 91177308-0d34-0410-b5e6-96231b3b80d8

12 years agoFix a couple of loops that were processing unused value numbers.
Jakob Stoklund Olesen [Fri, 3 Aug 2012 20:59:29 +0000 (20:59 +0000)]
Fix a couple of loops that were processing unused value numbers.

Unused VNInfos should be left alone. Their def SlotIndex doesn't point
to anything.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@161257 91177308-0d34-0410-b5e6-96231b3b80d8

12 years agoSilence unused variable warning in -asserts build
Matt Beaumont-Gay [Fri, 3 Aug 2012 20:54:11 +0000 (20:54 +0000)]
Silence unused variable warning in -asserts build

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@161256 91177308-0d34-0410-b5e6-96231b3b80d8

12 years agoEliminate the VNInfo::hasPHIKill() flag.
Jakob Stoklund Olesen [Fri, 3 Aug 2012 20:19:44 +0000 (20:19 +0000)]
Eliminate the VNInfo::hasPHIKill() flag.

The only real user of the flag was removeCopyByCommutingDef(), and it
has been switched to LiveIntervals::hasPHIKill().

All the code changed by this patch was only concerned with computing and
propagating the flag.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@161255 91177308-0d34-0410-b5e6-96231b3b80d8

12 years agoMake the hasPHIKills flag a computed property.
Jakob Stoklund Olesen [Fri, 3 Aug 2012 20:10:24 +0000 (20:10 +0000)]
Make the hasPHIKills flag a computed property.

The VNInfo::HAS_PHI_KILL is only half supported. We precompute it in
LiveIntervalAnalysis, but it isn't properly updated by live range
splitting and functions like shrinkToUses().

It is only used in one place: RegisterCoalescer::removeCopyByCommutingDef().

This patch changes that function to use a new LiveIntervals::hasPHIKill()
function that computes the flag for a given value number.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@161254 91177308-0d34-0410-b5e6-96231b3b80d8

12 years agoAdd a configure flag for enabling -Werror on the command line
Eric Christopher [Fri, 3 Aug 2012 19:58:20 +0000 (19:58 +0000)]
Add a configure flag for enabling -Werror on the command line
while building as requested by Lang.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@161253 91177308-0d34-0410-b5e6-96231b3b80d8

12 years agoAdd support for detecting libxml for Dmitri's work. He'll
Eric Christopher [Fri, 3 Aug 2012 19:47:19 +0000 (19:47 +0000)]
Add support for detecting libxml for Dmitri's work. He'll
commit code in clang that uses this shortly.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@161252 91177308-0d34-0410-b5e6-96231b3b80d8

12 years agoAdd a configure option to pass -std=c++11 on the command line.
Eric Christopher [Fri, 3 Aug 2012 19:47:14 +0000 (19:47 +0000)]
Add a configure option to pass -std=c++11 on the command line.

rdar://11366674

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@161251 91177308-0d34-0410-b5e6-96231b3b80d8

12 years agoRemove a dead prototype.
Jakob Stoklund Olesen [Fri, 3 Aug 2012 18:34:32 +0000 (18:34 +0000)]
Remove a dead prototype.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@161250 91177308-0d34-0410-b5e6-96231b3b80d8

12 years agoRevert previous patch here, we should instead configure in specific
Eric Christopher [Fri, 3 Aug 2012 17:45:31 +0000 (17:45 +0000)]
Revert previous patch here, we should instead configure in specific
packages for particular uses.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@161246 91177308-0d34-0410-b5e6-96231b3b80d8

12 years agoDelete dead function.
Jakob Stoklund Olesen [Fri, 3 Aug 2012 15:21:21 +0000 (15:21 +0000)]
Delete dead function.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@161242 91177308-0d34-0410-b5e6-96231b3b80d8

12 years agoDon't delete dead code in TwoAddressInstructionPass.
Jakob Stoklund Olesen [Fri, 3 Aug 2012 15:11:57 +0000 (15:11 +0000)]
Don't delete dead code in TwoAddressInstructionPass.

This functionality was added before we started running
DeadMachineInstructionElim on all targets. It serves no purpose now.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@161241 91177308-0d34-0410-b5e6-96231b3b80d8

12 years agoallow 'make CPPFLAGS=<something>' work again
Gabor Greif [Fri, 3 Aug 2012 13:31:24 +0000 (13:31 +0000)]
allow 'make CPPFLAGS=<something>' work again

this makes this hack a bit more bearable
for poor souls who need to pass custom
preprocessor flags to the build process

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@161240 91177308-0d34-0410-b5e6-96231b3b80d8

12 years agoAdd a BUILD_FLAGS variable so that autoconf checks have a place
Eric Christopher [Fri, 3 Aug 2012 05:18:00 +0000 (05:18 +0000)]
Add a BUILD_FLAGS variable so that autoconf checks have a place
to store additional flag options since too many things can
and do override CPPFLAGS. Also, this is exported, unlike CPPFLAGS
so it can be actually used elsewhere. This should enable us
to remove the AC_SUBSTs in the intel checks, but I have no way
of testing it.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@161233 91177308-0d34-0410-b5e6-96231b3b80d8

12 years agoFall back to selection DAG isel for calls to builtin functions.
Bob Wilson [Fri, 3 Aug 2012 04:06:28 +0000 (04:06 +0000)]
Fall back to selection DAG isel for calls to builtin functions.

Fast isel doesn't currently have support for translating builtin function
calls to target instructions.  For embedded environments where the library
functions are not available, this is a matter of correctness and not
just optimization.  Most of this patch is just arranging to make the
TargetLibraryInfo available in fast isel.  <rdar://problem/12008746>

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@161232 91177308-0d34-0410-b5e6-96231b3b80d8

12 years agoAdd new getLibFunc method to TargetLibraryInfo.
Bob Wilson [Fri, 3 Aug 2012 04:06:22 +0000 (04:06 +0000)]
Add new getLibFunc method to TargetLibraryInfo.

This just provides a way to look up a LibFunc::Func enum value for a
function name.  Alphabetize the enums and function names so we can use a
binary search.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@161231 91177308-0d34-0410-b5e6-96231b3b80d8

12 years ago[arm-fast-isel] Add support for shl, lshr, and ashr.
Jush Lu [Fri, 3 Aug 2012 02:37:48 +0000 (02:37 +0000)]
[arm-fast-isel] Add support for shl, lshr, and ashr.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@161230 91177308-0d34-0410-b5e6-96231b3b80d8

12 years ago[CMake] Add yaml2obj to check-llvm.
NAKAMURA Takumi [Fri, 3 Aug 2012 00:45:32 +0000 (00:45 +0000)]
[CMake] Add yaml2obj to check-llvm.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@161229 91177308-0d34-0410-b5e6-96231b3b80d8

12 years agoMove the "findUsedStructTypes" functionality outside of the Module class.
Bill Wendling [Fri, 3 Aug 2012 00:30:35 +0000 (00:30 +0000)]
Move the "findUsedStructTypes" functionality outside of the Module class.

The "findUsedStructTypes" method is very expensive to run. It needs to be
optimized so that LTO can run faster. Splitting this method out of the Module
class will help this occur. For instance, it can keep a list of seen objects so
that it doesn't process them over and over again.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@161228 91177308-0d34-0410-b5e6-96231b3b80d8

12 years agoAdd support for the ARM GHC calling convention, this patch was in 3.0,
Eric Christopher [Fri, 3 Aug 2012 00:05:53 +0000 (00:05 +0000)]
Add support for the ARM GHC calling convention, this patch was in 3.0,
but somehow managed to be dropped later.

Patch by Karel Gardas.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@161226 91177308-0d34-0410-b5e6-96231b3b80d8

12 years agoRemove redundant '== true' after a comparison.
Richard Trieu [Thu, 2 Aug 2012 23:22:39 +0000 (23:22 +0000)]
Remove redundant '== true' after a comparison.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@161223 91177308-0d34-0410-b5e6-96231b3b80d8

12 years agoARM: Tidy up. Remove unused template parameters.
Jim Grosbach [Thu, 2 Aug 2012 22:08:27 +0000 (22:08 +0000)]
ARM: Tidy up. Remove unused template parameters.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@161222 91177308-0d34-0410-b5e6-96231b3b80d8

12 years agoARM: More InstAlias refactors to use #NAME#.
Jim Grosbach [Thu, 2 Aug 2012 21:59:52 +0000 (21:59 +0000)]
ARM: More InstAlias refactors to use #NAME#.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@161220 91177308-0d34-0410-b5e6-96231b3b80d8

12 years agoMove test yaml files under Inputs until they are converted to be the actual
Matt Beaumont-Gay [Thu, 2 Aug 2012 21:52:49 +0000 (21:52 +0000)]
Move test yaml files under Inputs until they are converted to be the actual
test files.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@161219 91177308-0d34-0410-b5e6-96231b3b80d8

12 years agoARM: Refactor instaliases using TableGen support for #NAME#.
Jim Grosbach [Thu, 2 Aug 2012 21:50:41 +0000 (21:50 +0000)]
ARM: Refactor instaliases using TableGen support for #NAME#.

Now that TableGen supports references to NAME w/o it being explicitly
referenced in the definition's own name, use that to simplify
assembly InstAlias definitions in multiclasses.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@161218 91177308-0d34-0410-b5e6-96231b3b80d8

12 years agoX86 Peephole: fold loads to the source register operand if possible.
Manman Ren [Thu, 2 Aug 2012 19:37:32 +0000 (19:37 +0000)]
X86 Peephole: fold loads to the source register operand if possible.

Add more comments and use early returns to reduce nesting in isLoadFoldable.
Also disable folding for V_SET0 to avoid introducing a const pool entry and
a const pool load.

rdar://10554090 and rdar://11873276

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@161207 91177308-0d34-0410-b5e6-96231b3b80d8

12 years ago[yaml2obj] Fix build. Apparently I've gotten too familiar with C++11.
Michael J. Spencer [Thu, 2 Aug 2012 19:36:30 +0000 (19:36 +0000)]
[yaml2obj] Fix build. Apparently I've gotten too familiar with C++11.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@161206 91177308-0d34-0410-b5e6-96231b3b80d8

12 years agoAdd yaml2obj. A utility to convert YAML to binaries.
Michael J. Spencer [Thu, 2 Aug 2012 19:16:56 +0000 (19:16 +0000)]
Add yaml2obj. A utility to convert YAML to binaries.

yaml2obj takes a textual description of an object file in YAML format
and outputs the binary equivalent. This greatly simplifies writing
tests that take binary object files as input.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@161205 91177308-0d34-0410-b5e6-96231b3b80d8

12 years agoTableGen: Allow use of #NAME# outside of 'def' names.
Jim Grosbach [Thu, 2 Aug 2012 18:46:42 +0000 (18:46 +0000)]
TableGen: Allow use of #NAME# outside of 'def' names.

Previously, def NAME values were only populated, and references to NAME
resolved, when NAME was referenced in the 'def' entry of the multiclass
sub-entry. e.g.,
multiclass foo<...> {
  def prefix_#NAME : ...
}

It's useful, however, to be able to reference NAME even when the default
def name is used. For example, when a multiclass has 'def : Pat<...>'
or 'def : InstAlias<...>' entries which refer to earlier instruction
definitions in the same multiclass. e.g.,
multiclass myMulti<RegisterClass rc> {
  def _r : myI<(outs rc:$d), (ins rc:$r), "r $d, $r", []>;

  def : InstAlias<\"wilma $r\", (!cast<Instruction>(NAME#\"_r\") rc:$r, rc:$r)>;
}

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@161198 91177308-0d34-0410-b5e6-96231b3b80d8

12 years agoCompute the critical path length through a trace.
Jakob Stoklund Olesen [Thu, 2 Aug 2012 18:45:54 +0000 (18:45 +0000)]
Compute the critical path length through a trace.

Whenever both instruction depths and instruction heights are known in a
block, it is possible to compute the length of the critical path as
max(depth+height) over the instructions in the block.

The stored live-in lists make it possible to accurately compute the
length of a critical path that bypasses the current (small) block.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@161197 91177308-0d34-0410-b5e6-96231b3b80d8

12 years agoMove the code that creates instances of MipsInstrInfo and MipsFrameLowering out
Akira Hatanaka [Thu, 2 Aug 2012 18:21:47 +0000 (18:21 +0000)]
Move the code that creates instances of MipsInstrInfo and MipsFrameLowering out
of MipsTargetMachine.cpp.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@161191 91177308-0d34-0410-b5e6-96231b3b80d8

12 years agoSet transient stack alignment in constructor of MipsFrameLowering and re-enable
Akira Hatanaka [Thu, 2 Aug 2012 18:15:13 +0000 (18:15 +0000)]
Set transient stack alignment in constructor of MipsFrameLowering and re-enable
test o32_cc_vararg.ll.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@161189 91177308-0d34-0410-b5e6-96231b3b80d8

12 years agoVerify regunit intervals along with virtreg intervals.
Jakob Stoklund Olesen [Thu, 2 Aug 2012 16:36:50 +0000 (16:36 +0000)]
Verify regunit intervals along with virtreg intervals.

Don't cause regunit intervals to be computed just to verify them. Only
check the already cached intervals.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@161183 91177308-0d34-0410-b5e6-96231b3b80d8

12 years agoAvoid creating dangling physreg live ranges during DCE.
Jakob Stoklund Olesen [Thu, 2 Aug 2012 16:36:47 +0000 (16:36 +0000)]
Avoid creating dangling physreg live ranges during DCE.

LiveRangeEdit::eliminateDeadDefs() can delete a dead instruction that
reads unreserved physregs. This would leave the corresponding regunit
live interval dangling because we don't have shrinkToUses() for physical
registers.

Fix this problem by turning the instruction into a KILL instead of
deleting it. This happens in a landing pad in
test/CodeGen/X86/2012-05-19-CoalescerCrash.ll:

  %vreg27<def,dead> = COPY %EDX<kill>; GR32:%vreg27

becomes:

  KILL %EDX<kill>

An upcoming fix to the machine verifier will catch problems like this by
verifying regunit live intervals.

This fixes PR13498. I am not including the test case from the PR since
we already have one exposing the problem once the verifier is fixed.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@161182 91177308-0d34-0410-b5e6-96231b3b80d8

12 years agoAdd TargetRegisterInfo::hasRegUnit().
Jakob Stoklund Olesen [Thu, 2 Aug 2012 14:45:53 +0000 (14:45 +0000)]
Add TargetRegisterInfo::hasRegUnit().

This trivial helper function tests if a register contains a register
unit. It is similar to regsOverlap(), but with asymmetric arguments.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@161180 91177308-0d34-0410-b5e6-96231b3b80d8

12 years agoinclude/llvm/Analysis/RegionInfo.h: Appease msvc.
NAKAMURA Takumi [Thu, 2 Aug 2012 14:44:01 +0000 (14:44 +0000)]
include/llvm/Analysis/RegionInfo.h: Appease msvc.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@161179 91177308-0d34-0410-b5e6-96231b3b80d8

12 years agoAdd report() functions that take a LiveInterval argument.
Jakob Stoklund Olesen [Thu, 2 Aug 2012 14:31:49 +0000 (14:31 +0000)]
Add report() functions that take a LiveInterval argument.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@161178 91177308-0d34-0410-b5e6-96231b3b80d8

12 years agoImplement the block_iterator of Region based on df_iterator.
Hongbin Zheng [Thu, 2 Aug 2012 14:20:02 +0000 (14:20 +0000)]
Implement the block_iterator of Region based on df_iterator.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@161177 91177308-0d34-0410-b5e6-96231b3b80d8

12 years agoJIT::runFunction(): add a fast path for functions with a single argument that is...
Nuno Lopes [Thu, 2 Aug 2012 12:09:32 +0000 (12:09 +0000)]
JIT::runFunction(): add a fast path for functions with a single argument that is a pointer.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@161171 91177308-0d34-0410-b5e6-96231b3b80d8