'ischaincompatible' when a pattern has more than one input chain.  Need
to do some commenting and cleanup now that I understand how this works.

llvm-svn: 96443

llvm-svn: 96440

It's not clear why this is really required, but it was explicitly
added in r48808 with no real explanation or rdar #.

llvm-svn: 96438

llvm-svn: 96437

variables.  Use the fancy OpNo variable instead of i,
which has the right index including chains.

llvm-svn: 96436

This pass is supposed to be run on the linked .bc module.
It traveses the module call graph twice. Once starting from the main function
and marking each reached function as "ML". Again, starting from the ISR
and cloning any reachable function that was marked as "ML". After cloning
the function, it remaps all the call sites in IL functions to call the
cloned functions. 

Currently only marking is being done.

llvm-svn: 96435

llvm-svn: 96434

llvm-svn: 96433

llvm-svn: 96432

llvm-svn: 96429

have overflowed.

llvm-svn: 96428

CheckComplexPattern function.  Though it is logically const,
I don't have the fortitude to clean up all the targets now,
and it not being const doesn't block anything.

llvm-svn: 96426

llvm-svn: 96422

64 bits, fixing a variety of problems.

llvm-svn: 96421

for evaluating complex patterns.  Some cleanup has to happen before
this can be used though.

llvm-svn: 96419

indentation.  No functional changes.

llvm-svn: 96418

like llvm-gcc-x86_64-darwin10-selfhost

llvm-svn: 96417

llvm-svn: 96416

llvm-svn: 96413

record all their results.

llvm-svn: 96412

does the same thing as getComplexPatternInfo.

llvm-svn: 96411

llvm-svn: 96410

llvm-svn: 96409

into a roundss intrinsic, producing a cyclic dag.  The root cause
of this is badness handling ComplexPattern nodes in the old dagisel
that I noticed through inspection.  Eliminate a copy of the of the
code that handled ComplexPatterns by making EmitChildMatchCode call
into EmitMatchCode.

llvm-svn: 96408

new lack of R2.

llvm-svn: 96407

llvm-svn: 96404

build failures due to my fix for pr6111.

llvm-svn: 96402

llvm-svn: 96401

llvm-svn: 96399

llvm-svn: 96395

llvm-svn: 96393

llvm-svn: 96391

If there exists a use of a build_vector that's the bitwise complement of the mask,
then transform the node to
(and (xor x, (build_vector -1,-1,-1,-1)), (build_vector ~c1,~c2,~c3,~c4)).

Since this transformation is only useful when 1) the given build_vector will
become a load from constpool, and 2) (and (xor x -1), y) matches to a single
instruction, I decided this is appropriate as a x86 specific transformation.
rdar://7323335

llvm-svn: 96389

llvm-svn: 96388

llvm-svn: 96387

Add support for emitting non-temporal stores for DAGs marked
non-temporal.  Fix from r96241 for botched encoding of MOVNTDQ.

Add documentation for !nontemporal metadata.

Add a simpler movnt testcase.

llvm-svn: 96386

llvm-svn: 96385

to have the predicate on the pattern itself instead. Support for the new
ISel. Remove definitions of CarryDefIsUnused and CarryDefIsUsed since they are
no longer used anywhere.

llvm-svn: 96384

llvm-svn: 96383

llvm-svn: 96382