condition doesn't return a result, instead
of blindly trying to use that result.

<rdar://problem/14009519>

llvm-svn: 182875

llvm-svn: 182874

llvm-svn: 182873

This allows rematerialization during register coalescing to handle
more cases involving operations like SUBREG_TO_REG which might need to
be rematerialized using sub-register indices.

For example, code like:
    v1(GPR64):sub_32 = MOVZ something
    v2(GPR64) = COPY v1(GPR64)
should be convertable to:
    v2(GPR64):sub_32 = MOVZ something

but previously we just gave up in places like this

llvm-svn: 182872

[libclang] For "@import .." code-completion results, associate a CXCursor_ModuleImportDecl cursor instead of CXCursor_NotImplemented.

llvm-svn: 182871

While we can't yet emit vbtables, this allows us to find virtual bases
of objects constructed in other TUs.

This make iostream hello world work, since basic_ostream virtually
inherits from basic_ios.

Differential Revision: http://llvm-reviews.chandlerc.com/D795

llvm-svn: 182870

llvm-svn: 182869

llvm-svn: 182868

Since the testing case uses ref_addr, which requires version 3+ to work,
we will solve the dwarf version issue first.

This patch also causes failures in one of the bots. I will update the patch
accordingly in my next attempt.

rdar://13926659

llvm-svn: 182867

Peter pointed out that C::f() is not a key function.  C's key function
is actually C::~C().

llvm-svn: 182866

MSVC's class data is always comdat, so clang's should always be
linkonce_odr in LLVM IR.

Reviewers: pcc

Differential Revision: http://llvm-reviews.chandlerc.com/D838

llvm-svn: 182865

newFrontendActionFactory() took a pointer to a callback to call when a source
file was done being processed by an action. This revision updates the callback
to include an ante-processing callback as well.

Callback-providing class renamed and callback functions themselves renamed.
Functions are no longer pure-virtual so users aren't forced to implement both
callbacks if one isn't needed.

llvm-svn: 182864

Tidy up three places where the register class for ARM and Thumb wasn't
restrictive enough:
 - No PC dest for reg-reg add/orr/sub.
 - No PC dest for shifts.
 - No PC or SP for Thumb2 reg-imm add.

I encountered this while combining FastISel with
-verify-machineinstrs. These instructions defined registers whose
classes weren't restrictive enough, and the uses failed
verification. They're also undefined in the ISA, or would produce code
that FastISel wouldn't want. This doesn't fix the register class
narrowing issue (where uses should restrict definitions), and isn't
thorough, but it's a small step in the right direction.

llvm-svn: 182863

llvm-svn: 182862

Now that the TokenAnnotator does not require stack space anymore,
reconstructing the lines has become the limiting factor. This patch
fixes that problem, allowing large files with multiple megabytes of
single unwrapped lines to be formatted.

llvm-svn: 182861

- The original test now passes on Linux with clang because a breakpoint is hit prior to evaluation of text_list, which improves text coverage.
- The new test fails because 4 steps are requested, and only two occur prior to evaluation of text_list.
--- Note that the loss of every second "next" command can be reproduced using lldb manually with this script.

llvm-svn: 182860

Gets rid of AnnotatedToken, putting everything into FormatToken.
FormatTokens are created once, and only referenced by pointer.  This
enables multiple future features, like having tokens shared between
multiple UnwrappedLines (while there's still work to do to fully enable
that).

llvm-svn: 182859

llvm-svn: 182858

llvm-svn: 182857

llvm-svn: 182856

If an identifier is on its own line and it is all upper case, it is highly
likely that this is a macro that is meant to stand on a line by itself.

Before:
class A : public QObject {
  Q_OBJECT A() {}
};

Ater:
class A : public QObject {
  Q_OBJECT

  A() {}
};

llvm-svn: 182855

This CL enables thread support in LSan when used on top of ASan.

llvm-svn: 182854

Instead of using arch_prctl(ARCH_GET_FS), read the address from the
tread descriptor itself. This lets us avoid sandboxing issues. Also,
GetThreadStackAndTls() can now be implemented on i386.

llvm-svn: 182853

llvm-svn: 182852

Fixes the Go build.

Differential Revision: http://llvm-reviews.chandlerc.com/D877

llvm-svn: 182851

llvm-svn: 182850

With option enabled (e.g. in Google-style):
template <typename T>
void f() {}

With option disabled:
template <typename T> void f() {}

Enabling this for Google-style and Chromium-style, not sure which other
styles would prefer that.

llvm-svn: 182849

llvm-svn: 182848

llvm-svn: 182847

This patch adds support for the CIJ and CGIJ instructions.

llvm-svn: 182846

llvm-svn: 182845

llvm-svn: 182844

llvm-svn: 182843

llvm-svn: 182842

definitions from test.

llvm-svn: 182841

[sanitizer] Use raw check in internal allocator to prevent self-deadlocks. Try to fix Windows build.

llvm-svn: 182840

llvm-svn: 182839

llvm-svn: 182838

llvm-svn: 182837

llvm-svn: 182836