The standard function epilog includes a .size directive, but ppc64 uses
an alternate local symbol to tag the actual start of each function.

Until recently, binutils accepted the .size directive as:
 .size	test1, .Ltmp0-test1
however, using this directive with recent binutils will result in the error:
 .size expression for XXX does not evaluate to a constant
so we must use the label which actually tags the start of the function.

llvm-svn: 151200

llvm-svn: 151189

llvm-svn: 151186

Patch by Joe Groff!

llvm-svn: 151183

llvm-svn: 151178

llvm-svn: 151169

llvm-svn: 151167

llvm-svn: 151166

Add some data structures to represent for loops.  These will be
referenced during object processing to do any needed iteration and
instantiation.

Add foreach keyword support to the lexer.

Add a mode to indicate that we're parsing a foreach loop.  This allows
the value parser to early-out when processing the foreach value list.

Add a routine to parse foreach iteration declarations.  This is
separate from ParseDeclaration because the type of the named value
(the iterator) doesn't match the type of the initializer value (the
value list).  It also needs to add two values to the foreach record:
the iterator and the value list.

Add parsing support for foreach.

Add the code to process foreach loops and create defs based
on iterator values.

Allow foreach loops to be matched at the top level.

When parsing an IDValue check if it is a foreach loop iterator for one
of the active loops.  If so, return a VarInit for it.

Add Emacs keyword support for foreach.

Add VIM keyword support for foreach.

Add tests to check foreach operation.

Add TableGen documentation for foreach.

Support foreach with multiple objects.

Support non-braced foreach body with one object.

Do not require types for the foreach declaration.  Assume the iterator
type from the iteration list element type.

llvm-svn: 151164

llvm-svn: 151163

llvm-svn: 151160

chip in r139383, and the PSP components of the triple are really
annoying to parse. Let's leave this chapter behind. There is no reason
to expect LLVM to see a PSP-related triple these days, and so no
reasonable motivation to support them.

It might be reasonable to prune a few of the older MIPS triple forms in
general, but as those at least cause no burden on parsing (they aren't
both a chip and an OS!), I'm happy to leave them in for now.

llvm-svn: 151156

Worth another 45k (1%) off of a large C++ testcase.

rdar://10909458

llvm-svn: 151144

llvm-svn: 151143

llvm-svn: 151142

Declare register classes as const. Fix a couple pointers to register classes that weren't already const.

llvm-svn: 151138

The vast majority of virtual register definitions don't need an entry
in the DAG builder's VRegDefs set.

llvm-svn: 151136

Affect on SD scheduling and postRA scheduling:
Printing the DAG will display the nodes in top-down topological order.
This matches the order within the MBB and makes my life much easier in general.

Affect on misched:
We don't need to track virtual register uses at all. This is awesome.
I also intend to rely on the SUnit ID as a topo-sort index. So if A < B then we cannot have an edge B -> A.

llvm-svn: 151135

Make all pointers to TargetRegisterClass const since they are all pointers to static data that should not be modified.

llvm-svn: 151134

llvm-svn: 151127

the dominance once the dominates method is fixed and why we can use the builder's
insertion point.
Fixes pr12048.

llvm-svn: 151125

llvm-svn: 151123

llvm-svn: 151115

llvm-svn: 151114

llvm-svn: 151113

This makes RAFast 4% faster, and it gets rid of the dodgy DenseMap
iteration.

This also revealed that RAFast would sometimes dereference DenseMap
iterators after erasing other elements from the map. That does seem to
work in the current DenseMap implementation, but SparseSet doesn't allow
it.

llvm-svn: 151111

For objects that can be identified by small unsigned keys, SparseSet
provides constant time clear() and fast deterministic iteration. Insert,
erase, and find operations are typically faster than hash tables.

SparseSet is useful for keeping information about physical registers,
virtual registers, or numbered basic blocks.

llvm-svn: 151110

llvm-svn: 151107

This test case was way too strict, matching the entire assembly output.
Every non-trivial change to the ppc backend  or -O0 pipeline required
the test to be updated.

It should be replaced with a test of the specific vaarg feature.

llvm-svn: 151105

Even if a call instruction has %SP<imp-def> operands, it doesn't change
the value of the stack pointer.

llvm-svn: 151104

bundles. This method takes a bundle start and an MI being bundled, and makes
the intervals for the MI's operands appear to start/end on the bundle start.

Also fixes some minor cosmetic issues (whitespace, naming convention) in the
HMEditor code.

llvm-svn: 151099

rdar://10493979

llvm-svn: 151098

Part of rdar://10493979 where it reduces by about .5% (10k)

llvm-svn: 151097

they'll be simple enough to simulate, and to reduce the chance we'll encounter
equal but different simple pointer constants.

This removes the symptoms from PR11352 but is not a full fix. A proper fix would
either require a guarantee that two constant objects we simulate are folded
when equal, or a different way of handling equal pointers (ie., trying a
constantexpr icmp on them to see whether we know they're equal or non-equal or
unsure).

llvm-svn: 151093

llvm-svn: 151083

llvm-svn: 151079

llvm-svn: 151056

This transformation is not safe in some pathological cases (signed icmp of pointers should be an
extremely rare thing, but it's valid IR!). Add an explanatory comment.

Kudos to Duncan for pointing out this edge case (and not giving up explaining it until I finally got it).

llvm-svn: 151055

and into StringRef.cpp, which is where the other StringRef stuff is.

llvm-svn: 151054

llvm-svn: 151053