The AttributeWithIndex class exposed the interior structure of the AttributeSet
class. That was gross. Remove it and all of the code that relied upon it.

llvm-svn: 173722

in fact, resolve undef uses.

llvm-svn: 173721

llvm-svn: 173720

llvm-svn: 173718

llvm-svn: 173717

llvm-svn: 173716

llvm-svn: 173715

llvm-svn: 173714

error. Jordan is right.

llvm-svn: 173713

llvm-svn: 173712

Remove the use of the 't' length modifier to avoid a gcc warning. Based
on usage, 32 bits of precision is good enough for printing a stack
offset for a stack trace.

't' length modifier isn't in C++03 but it *is* in C++11. Added a FIXME
to reintroduce once LLVM makes the switch to C++11.

Reviewer: gribozavr
llvm-svn: 173711

llvm-svn: 173709

<rdar://problem/13098104>.

llvm-svn: 173708

-fno-modules-global-index -cc1 option to allow one to disable the
index for performance testing purposes, but with a 10% win in
-fsyntax-only time, there is no reason a user would do this.

llvm-svn: 173707

The common code in the post-RA scheduler to break anti-dependencies on the
critical path contained a flaw.  In the reported case, an anti-dependency
between the overlapping registers %X4 and %R4 exists:

	%X29<def> = OR8 %X4, %X4
	%R4<def>, %X3<def,dead,tied3> = LBZU 1, %X3<kill,tied1>

The unpatched code breaks the dependency by replacing %R4 and its uses
with %R3, the first register on the available list.  However, %R3 and
%X3 overlap, so this creates two overlapping definitions on the same
instruction.

The fix is straightforward, preventing selection of a register that
overlaps any other defined register on the same instruction.

The test case is reduced from the bug report, and verifies that we no
longer produce "lbzu 3, 1(3)" when breaking this anti-dependency.

llvm-svn: 173706

pool in the global module index is not worthwhile. Update comments to
limit the scope of the global module index to identifiers.

llvm-svn: 173705

llvm-svn: 173703

llvm-svn: 173702

Fixes a crash. Thanks, Richard.

llvm-svn: 173701

llvm-svn: 173700

llvm-svn: 173698

llvm-svn: 173697

llvm-svn: 173696

Before:
    for (aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa aaaaaaaaaaa = aaaaaaaaaaaaaaa
             .aaaaaaaaaaaaaaaa;
         aaaaaaaaaaa != aaaaaaaaaaaaaaaaaaa; ++aaaaaaaaaaa) {}

After:
    for (aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa aaaaaaaaaaa =
             aaaaaaaaaaaaaaa.aaaaaaaaaaaaaaaa;
         aaaaaaaaaaa != aaaaaaaaaaaaaaaaaaa; ++aaaaaaaaaaa) {}

llvm-svn: 173695

llvm-svn: 173694

llvm-svn: 173693

ModuleManager::visit() by keeping a free list of the two data
structures used to store state (a preallocated stack and a visitation
number vector). Improves -fsyntax-only performance for my modules test
case by 2.8%. Modules has pulled ahead by almost 10% with the global
module index.

llvm-svn: 173692

llvm-svn: 173691

This would be against the style guide:
http://google-styleguide.googlecode.com/svn/trunk/cppguide.xml?showone=Function_Declarations_and_Definitions#Function_Declarations_and_Definitions

Not sure what to do as a last resort if the function signature does not
fit onto a single line in Google style ..

llvm-svn: 173690

better job with them. Shaves off 0.7% of -fparse-only time for a
modules test case. Sure makes you wonder...

llvm-svn: 173689

Before:
TypeSpecDecl *TypeSpecDecl::Create(ASTContext &C, DeclContext *DC,
                                   SourceLocation L, IdentifierInfo *II,
                                   Type *T) {}

After:
TypeSpecDecl *
TypeSpecDecl::Create(ASTContext &C, DeclContext *DC, SourceLocation L,
                     IdentifierInfo *II, Type *T) {}

This fixes llvm.org/PR14717.

llvm-svn: 173688

llvm-svn: 173687

llvm-svn: 173686

Before:
Constructor
    : LooooooooooooooooooooongVariable(
        LooooooooooooooooooooongValue) {}

After:
Constructor
    : LooooooooooooooooooooongVariable(
          LooooooooooooooooooooongValue) {}

llvm-svn: 173685

Before (in good cases):
for (auto aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa
     : aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa) {}
for (auto aaaaaaaaaaaaaaaaaaaa : aaaaaaaaaaaaaaaaaaaa(aaaaaaaaaaaaaaaaaaaa,
aaaa)) {}

After:
for (auto aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa :
     aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa) {}
for (auto aaaaaaaaaaaaaaaaaaaa :
     aaaaaaaaaaaaaaaaaaaa(aaaaaaaaaaaaaaaaaaaa, aaaa)) {}

llvm-svn: 173684

Before we did not really systematically format those. Now, we format the
different cases as:

- 1 Line:  a ? b : c;
- 2 Lines: short ? loooooooooong
                 : loooooooooong
- 2 Lines: loooooooooooooooong
               ? short : short
- 3 Lines: loooooooooooooooong
               ? loooooooooooooong
               : loooooooooooooong

Not sure whether "?" and ":" should go on the new line, but it seems to
be the most consistent approach.

llvm-svn: 173683

It is way too slow. Change the default option value to 0.
Always do exact shadow propagation for unsigned ICmp with constants, it is
cheap (under 1% cpu time) and required for correctness.

llvm-svn: 173682

llvm-svn: 173681

1. Use a hanging ident for function calls nested in binary expressions.
   E.g.:
   int aaaaa = aaaaaaaaa && aaaaaaaaaa(
                                aaaaaaaaaa);

2. Slightly improve heuristic for builder type expressions and reduce
   penalty for breaking before "." and "->" in those.

3. Remove  mostly obsolete metric of decreasing indent level. This
   fixes: llvm.org/PR14931.

Changes #1 and #2 were necessary to keep tests passing after #3.

llvm-svn: 173680

Broken tests.

llvm-svn: 173679