The cmake script is failing to copy cxxabi.h to the right place because it was generating to destination path like so /include// and dstdir can legally be blank from my interpretation of the script, and this would then generate a path like libcxx/include// which is illegal.

llvm-svn: 178579

- make an overload of Launch() that takes an init list of const char* if all you need to tweak in the launch info are the command-line arguments
- make Run() return an int that you can use as an exit-code
- make dynamic values work properly when recursing in FetchVariables()
- make the po output more obvious in verbose mode

llvm-svn: 178578

Fixing a bug where LLDB was not handling correctly CFStrings that have an explicit length but no NULL terminator
The data formatter was showing garbled data as part of the summary
The fix is to explicitly figure out the explicit length if we need to (bitfields tell us when that is the case) and use that as a size delimiter

llvm-svn: 178577

implementation of std::is_polymorphic does this:

template <class _Tp> struct __is_polymorphic1 : public _Tp {};

... and that g++ rejects this if _Tp has an inaccessible virtual destructor
(because __is_polymorphic1<_Tp> would have a deleted virtual destructor
overriding _Tp's non-deleted destructor). Clang was failing to reject this;
I've fixed that in r178563, but that causes libc++'s corresponding test
case to fail with both clang and gcc when using the fallback
implementation. The fallback code also incorrectly rejects final types.

The attached patch fixes the fallback implementation of is_polymorphic; we
now use dynamic_cast's detection of polymorphic class types rather than
trying to determine if adding a virtual function makes the type larger:

  enable_if<sizeof((_Tp*)dynamic_cast<const volatile
void*>(declval<_Tp*>())) != 0, ...>

Two things of note here:
* the (_Tp*) cast is necessary to work around bugs in Clang and g++ where
we otherwise don't instantiate the dynamic_cast (filed as PR15656)
* the 'const volatile' is here to treat is_polymorphic<cv T> as true for a
polymorphic class type T -- my reading of the standard suggests this is
incorrect, but it matches our builtin __is_polymorphic and gcc

llvm-svn: 178576

For variable declarations initialized with new expressions, use 'auto' for the
type specifier.

The 'auto' replacement happens only when the type of the VarDecl exactly
matches the type of the initializer and the VarDecl is *not* CV-qualified. The
only case that is currently handled is if the pointer type of the VarDecl is
itself CV qualified.

Some improvements need to be made to Clang's TypeLoc information in order for
other CV qualifier cases to be successfully handled. See the new test suite
new_cv_failing.cpp for examples of usages that could be handled with such an
improvement.

Function pointers are, for now, not transformed until the identifier info can
be extracted.

Reviewer: klimek
llvm-svn: 178575

Don't crash when there is no register context for a thread with kernel debugging. The kernel debugging uses the OperatingSystemPlugin that may behave badly when trying to get thread state, so be prepared to have invalid register contexts in threads.

llvm-svn: 178574

when we changed "rb" to "rbreak".

<rdar://problem/13552724>

llvm-svn: 178573

llvm-svn: 178572

llvm-svn: 178569

llvm-svn: 178568

llvm-svn: 178567

qualifiers.

This patch only adds support for parsing these identifiers in the
X86AsmParser.  The front-end interface isn't capable of looking up
these identifiers at this point in time.  The end result is the
compiler now errors during object file emission, rather than at
parse time.  Test case coming shortly.
Part of rdar://13499009 and PR13340

llvm-svn: 178566

llvm-svn: 178565

Add utilities to create struct nodes in TBAA type DAG and to create path-aware
tags. The format of struct nodes in TBAA type DAG: a unique name, a list of
fields with field offsets and field types. The format of path-aware tags:
a base type in TBAA type DAG, an access type and an offset relative to the base
type.

llvm-svn: 178564

overriding a non-deleted virtual function. The existing check for this doesn't
catch this case, because it fires before we mark the method as deleted.

llvm-svn: 178563

about 'isa' ivar being explicitely accessed
when base is a user class object reference.
// rdar://13503456

llvm-svn: 178562

llvm-svn: 178561

When doing a partword atomic operation, a lwarx was being paired with
a stdcx. instead of a stwcx. when compiling for a 64-bit target.  The
target has nothing to do with it in this case; we always need a stwcx.

Thanks to Kai Nacke for reporting the problem.

llvm-svn: 178559

This should fix the PPC buildbots.

llvm-svn: 178558

Updated tests and docs.

llvm-svn: 178556

llvm-svn: 178554

The new instruction scheduling models provide information about the
number of cycles consumed on each processor resource. This makes it
possible to estimate ILP more accurately than simply counting
instructions / issue width.

The functions getResourceDepth() and getResourceLength() now identify
the limiting processor resource, and return a cycle count based on that.

This gives more precise resource information, particularly in traces
that use one resource a lot more than others.

llvm-svn: 178553

llvm-svn: 178552

The IHM_ prefix was a fairly gross abbreviation to try to hit three
characters for uniqueness.

llvm-svn: 178551

llvm-svn: 178550

llvm-svn: 178549

Summary:
This makes it possible to share code between lib/AST/MicrosoftCXXABI.cpp
and lib/CodeGen/MicrosoftCXXABI.cpp.  No functionality change.

Also adds comments about the layout of the member pointer structs as I
currently understand them.

Reviewers: rjmccall

CC: timurrrr, cfe-commits

Differential Revision: http://llvm-reviews.chandlerc.com/D590

llvm-svn: 178548

This is helps on architectures where i8,i16 are not legal but we have byte, and
short loads/stores. Allowing us to merge copies like the one below on ARM.

copy(char *a, char *b, int n) {
 do {
   int t0 = a[0];
   int t1 = a[1];
   b[0] = t0;
   b[1] = t1;

radar://13536387

llvm-svn: 178546

llvm-svn: 178545

llvm-svn: 178544

case when [[clang::fallthrough]]; is used in a method of a local class.

llvm-svn: 178543

Basically we have always special-cased the top-level statement of an
unwrapped line (the one with ParenLevel == 0) and that lead to several
inconsistencies. All added tests were formatted in a strange way, for
example:

Before:
aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa
    .aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa
    .aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa
    .aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa();
if (aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa
        .aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa
            .aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa
                .aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa()) {
}

After:
aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa
    .aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa
    .aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa
    .aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa();
if (aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa
        .aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa
        .aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa
        .aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa()) {
}

llvm-svn: 178542

call memory indirect (32 and 64 bit).

llvm-svn: 178541

Fixes PR15642.

llvm-svn: 178540

llvm-svn: 178539

llvm-svn: 178538

Summary: Store comments in ScopedLineState

Reviewers: klimek, djasper

Reviewed By: klimek

CC: cfe-commits

Differential Revision: http://llvm-reviews.chandlerc.com/D609

llvm-svn: 178537

llvm-svn: 178536

WTFAnnotateBenignRaceSized and AnnotateBenignRaceSized.

llvm-svn: 178534

The iterator could be invalidated when it's recursively deleting a whole bunch
of constant expressions in a constant initializer.

Note: This was only reproducible if `opt' was run on a `.bc' file. If `opt' was
run on a `.ll' file, it wouldn't crash. This is why the test first pushes the
`.ll' file through `llvm-as' before feeding it to `opt'.

PR15440

llvm-svn: 178531