llvm-svn: 70522

Add a comment to refer to the section of the programmer's manual that explains what the pointer tagging in Use is for.

llvm-svn: 70521

-Replace DebugLocTuple's Source ID with CompileUnit's GlobalVariable*
-Remove DwarfWriter::getOrCreateSourceID
-Make necessary changes for the above (fix callsites, etc.)

llvm-svn: 70520

Don't include memory allocated for global variables during relocations resolution.

llvm-svn: 70517

comment a bit.

llvm-svn: 70515

llvm-svn: 70513

classes.

This is implemented as a function rather than a method on TargetRegisterClass
because it is symmetric in its arguments.

llvm-svn: 70512

A subclass is allowed to have a larger spill size than the superclass, and the
spill alignment must be a multiple of the superclass alignment. This causes
the following new subclass relations:

=== Alpha ===
F4RC -> F8RC

=== PPC ===
F4RC -> F8RC

=== SPU ===
R8C -> R16C -> R32C/R32FP -> R64C/R64FP -> GPRC/VECREG

=== X86 ===
FR32  -> FR64  -> VR128
RFP32 -> RFP64 -> RFP80

These subclass relations are consistent with the behaviour of -join-cross-class-copies.

llvm-svn: 70511

llvm-svn: 70510

compute an upper-bound value for the trip count, in addition to
the actual trip count. Use this to allow getZeroExtendExpr and
getSignExtendExpr to fold casts in more cases.

This may eventually morph into a more general value-range
analysis capability; there are certainly plenty of places where
more complete value-range information would allow more folding.

llvm-svn: 70509

Add a smarter heuristic to determine when to coalesce a virtual register with a physical one. More specifically, it avoid tying a virtual register in the loop with a physical register defined / used outside the loop. When it determines it's not profitable, it will use the physical register as the allocation preference instead.

This is *not* turned on by default. Testing indicates this is just as likely to pessimize code. The main issue seems to be allocation preference doesn't work effectively. That will change once I've taught register allocator "swapping".

llvm-svn: 70503

so that it doesn't shadow the instance variable of the same name.
Make the parameter names in method declarations match the definitions.

llvm-svn: 70502

llvm-svn: 70499

of "class", so that it matches the subsequent definition.

llvm-svn: 70498

Make dataflow iteration possible on Value*, not only on User*: df_ext_iterator<Value*, SmallPtrSet<const Value*, 16> >

llvm-svn: 70496

in getSCEVAtScope.

llvm-svn: 70495

llvm-svn: 70491

run when assembling.

Wire this up to the gold plugin. You can now pass --plugin-opt gcc=/foo/bar/gcc
and it will run that gcc instead of looking for it on the path.

llvm-svn: 70490

class.

llvm-svn: 70488

llvm-svn: 70466

a SCEVable type, not just integer types.

llvm-svn: 70463

llvm-svn: 70461

llvm-svn: 70459

memory operands otherwise the writebacks get lost when the inline asm 
doesn't otherwise have side effects.  This fixes rdar://6839427, though
clang really shouldn't generate these anymore.

llvm-svn: 70455

llvm-svn: 70454

llvm-svn: 70453

llvm-svn: 70449

llvm-svn: 70445

which better identifies what the optimization is doing. And is more flexible for
future uses.

llvm-svn: 70440

llvm-svn: 70437

llvm-svn: 70427

llvm-svn: 70425

llvm-svn: 70422

(sext i8 {-128,+,1} to i64) to i64 {-128,+,1}, where the iteration
crosses from negative to positive, but is still safe if the trip
count is within range.

llvm-svn: 70421

llvm-svn: 70416

llvm-svn: 70411

llvm-svn: 70410

llvm-svn: 70408

print sext, zext, and trunc, instead of signextend, zeroextend,
and truncate, respectively, for consistency with the main IR.

llvm-svn: 70405

llvm-svn: 70391