passes to a create-pass function instead of a direct constructor call.

llvm-svn: 171622

LLVM ignores this data for now - patch for that to follow.

llvm-svn: 171605

and add stack alignment information.

llvm-svn: 171588

It is somewhat hard to test linkage, so I decided to try to add an assert. This
already found some interesting cases where there were different.

llvm-svn: 171585

In my last patch initialize the destination to null (with a simple store) before doing a storeStrong to it.
// rdar://12530881

llvm-svn: 171572

Some of this is still pretty rough (note the load of FIXMEs), but it is
strictly an improvement and fixes various bugs that were related to
macro processing but are also imporant in non-macro use cases.

Specific fixes:
- correctly puts espaced newlines at the end of the line
- fixes counting of white space before a token when escaped newlines are
  present
- fixes parsing of "trailing" tokens when eof() is hit
- puts macro parsing orthogonal to parsing other structure
- general support for parsing of macro definitions

Due to the fix to format trailing tokens, this change also includes a
bunch of fixes to the c-index tests.

llvm-svn: 171556

__strong __block variables, perform objc_storeStrong on
source and destination instead of direct move. This 
is done with -O0 and to improve some analysis.
// rdar://12530881

llvm-svn: 171555

llvm-svn: 171545

Thanks for dgregor for noticing it.

llvm-svn: 171532

Before: InvalidRegions[ &R] = 0;
After:  InvalidRegions[&R] = 0;

This fixes llvm.org/PR14793

llvm-svn: 171522

llvm-svn: 171521

llvm-svn: 171516

llvm-svn: 171505

Add __has_feature support to detect if clang supports the explicit "atomic" keyword for ObjC properties.

Fixes <rdar://problem/12953378>.

llvm-svn: 171504

NSErrorChecker: remove quoting the parameter name in the diagnostic until we actually include it's name.

This is a possible regression of moving to using ImplicitNullDerefEvent.
Fixing this for real (including the parameter name) requires more
plumbing in ImplicitNullDerefEvent.  This is just a stop gap fix.

llvm-svn: 171502

llvm-svn: 171501

llvm-svn: 171497

This prevents code like:

namespace {
class Foo {
  Foo(
  };
}  // comment

from causing segfaults (see llvm.org/PR14774).

llvm-svn: 171495

llvm-svn: 171494

The line information was changed when emitting debug information for all the
DeclRefExprs and we should change it back to get ready for PopClenupBlocks
called from FinishFunction.

rdar://11562117

llvm-svn: 171493

to match those foung in objc.h an avoid spurious warnings. 
// rdar://12489098

llvm-svn: 171492

rdar://11074996

llvm-svn: 171485

variable, thus emitting the "switch case is in protected scope" error.

rdar://12952016

llvm-svn: 171484

llvm-svn: 171483

rdar://12952025

llvm-svn: 171482

llvm-svn: 171478

field number was 0 anyhow.

llvm-svn: 171472

To do so we have to wait until we know that the type of a variable has been
deduced. Sema::FinalizeDeclaration is the first callback that is used for
decl with or without initializers.

llvm-svn: 171458

llvm-svn: 171457

  c = Block_copy(b);
  Block_release(c);
 ---->
  c = [b copy];
  <removed>

rdar://9408211

llvm-svn: 171454

llvm-svn: 171439

static local variable from the parent scope.  PR14773.

llvm-svn: 171433

This better reflects when callback is called and what the checkers
are relying on. (Both names meant the same pre-IPA.)

llvm-svn: 171432

Patch by Will Dietz:

Minor touchup so the values of Offset/ID reflect their intention.
Previously, the sum (Offset+ID) was correct, but Offset/ID
individually were wrong.

Caught by investigating unsigned overflow reported by -fsanitize=integer.

llvm-svn: 171421

Patch by Krzysztof Parzyszek!

llvm-svn: 171415

llvm-svn: 171402

Fixes:
- incorrect handling of multiple consecutive preprocessor directives
- crash when trying to right align the escpaed newline for a line that
  is longer than the column limit
- using only ColumnLimit-1 columns when layouting with escaped newlines
  inside preprocessor directives

llvm-svn: 171401

This fixes llvm.org/PR14786.

We will need to split there as a last resort, but that should be done
consistently independent of whether the type is a template type or not.

Before:
template <typename T>
aaaaaaaaaaaaaaaaaaa(aaaaaaaaaaaaaaaaaa, aaaaaaaaaaaaaaaaaaaaaaaaaa<T>
                    ::aaaaaaaaaa, aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa);

After:
template <typename T>
aaaaaaaaaaaaaaaaaaa(aaaaaaaaaaaaaaaaaa,
                    aaaaaaaaaaaaaaaaaaaaaaaaaa<T>::aaaaaaaaaa,
                    aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa);

llvm-svn: 171400

This fixes llvm.org/PR14687.
Also fixes segfault for lines starting with * or &.

Before:
a *~b;
*a = 1;  // <- this segfaulted

After:
a * ~b;
*a = 1;  // no segfault :-)

llvm-svn: 171396

This is the first step towards handling preprocessor directives. This
patch only fixes the most pressing issue, namely correctly escaping
newlines for tokens within a sequence of a preprocessor directive.

The next step will be to fix incorrect format decisions on #define
directives.

llvm-svn: 171393