emit less than complete types on purpose on occasion and so
our caches aren't useful for this kind of lazy emitting.

llvm-svn: 150856

llvm-svn: 150855

llvm-svn: 150838

We now generate temporary arrays to back std::initializer_list objects
initialized with braces. The initializer_list is then made to point at
the array. We support both ptr+size and start+end forms, although
the latter is untested.

Array lifetime is correct for temporary std::initializer_lists (e.g.
call arguments) and local variables. It is untested for new expressions
and member initializers.

Things left to do:
Massively increase the amount of testing. I need to write tests for
start+end init lists, temporary objects created as a side effect of
initializing init list objects, new expressions, member initialization,
creation of temporary objects (e.g. std::vector) for initializer lists,
and probably more.
Get lifetime "right" for member initializers and new expressions. Not
that either are very useful.
Implement list-initialization of array new expressions.

llvm-svn: 150803

variable ends, if the variable has a trivial destructor and no mutable
subobjects then emit an llvm.invariant.start call for it. globalopt knows to
make the variable const when evaluating this.

llvm-svn: 150798

Fixes lots of gdb testsuite failures.

llvm-svn: 150797

as constants.

Refactor and simplify all the separate checks for whether a type can be
emitted as a constant.

llvm-svn: 150793

 1) It has a const-qualified type, and
 2) It has no mutable members, and
 3) It has no dynamic initialization, and
 4) It has trivial destruction.
Remove the unnecessary requirement that the type be POD. This allows us to
mark all constexpr objects with no mutable members as 'constant'.

llvm-svn: 150792

optional argument passed through the variadic ellipsis)
potentially affects how we need to lower it.  Propagate
this information down to the various getFunctionInfo(...)
overloads on CodeGenTypes.  Furthermore, rename those
overloads to clarify their distinct purposes, and make
sure we're calling the right one in the right place.
This has a nice side-effect of making it easier to construct
a function type, since the 'variadic' bit is no longer
separable.

This shouldn't really change anything for our existing
platforms, with one minor exception --- we should now call
variadic ObjC methods with the ... in the "right place"
(see the test case), which I guess matters for anyone
running GNUStep on MIPS.  Mostly it's just a substantial
clean-up.

llvm-svn: 150788

conversion to function pointer. Rather than having IRgen synthesize
the body of this function, we instead introduce a static member
function "__invoke" with the same signature as the lambda's
operator() in the AST. Sema then generates a body for the conversion
to function pointer which simply returns the address of __invoke. This
approach makes it easier to evaluate a call to the conversion function
as a constant, makes the linkage of the __invoke function follow the
normal rules for member functions, and may make life easier down the
road if we ever want to constexpr'ify some of lambdas.

Note that IR generation is responsible for filling in the body of
__invoke (Sema just adds a dummy body), because the body can't
generally be expressed in C++.

Eli, please review!

llvm-svn: 150783

    "Add a completed/incomplete type difference. This allows us to have
    partial types for contexts and forward decls while allowing us to
    complete types later on for debug purposes.

    This piggy-backs on the metadata replacement and rauw changes
    for temporary nodes and takes advantage of the incremental
    support I added in earlier. This allows us to, if we decide,
    to limit adding methods and variables to structures in order
    to limit the amount of debug information output into a .o file.

    The caching is a bit complicated though so any thoughts on
    untangling that are welcome."

with a fix:

 - Remove all RAUW during type construction by adding stub versions
   of types that we later complete.

and some TODOs:

 - Add an RAUW cache for forward declared types so that we can replace
   them at the end of compilation.
 - Remove the code that updates on completed types because we no
   longer need to have that happen. We emit incomplete types on
   purpose and only want to know when we want to complete them.

llvm-svn: 150752

llvm-svn: 150738

Revert "Revert "Make CXXNewExpr contain only a single initialier, and not hold the used constructor itself.""

This reintroduces commit r150682 with a fix for the Bullet benchmark crash.

llvm-svn: 150685

Revert "Make CXXNewExpr contain only a single initialier, and not hold the used constructor itself."
It leads to a compiler crash in the Bullet benchmark.

This reverts commit r12014.

llvm-svn: 150684

Holding the constructor directly makes no sense when list-initialized arrays come into play. The constructor is now held in a CXXConstructExpr, if construction is what is done. The new design can also distinguish properly between list-initialization and direct-initialization, as well as implicit default-initialization constructors and explicit value-initialization constructors. Finally, doing it this way removes redundance from the AST because CXXNewExpr doesn't try to handle both the allocation and the initialization responsibilities.

This breaks the static analysis of new expressions. I've filed PR12014 to track this.

llvm-svn: 150682

llvm-svn: 150660

llvm-svn: 150649

The garbage collection metadata needs to be merged "intelligently", when two or
more modules are linked together, and not merely appended. (Appending creates a
section which is too large.) The module flags metadata method is the way to do
this.
<rdar://problem/8198537>

llvm-svn: 150648

-fno-objc-arc-exceptions. This will allow the optimizer to perform
optimizations which are only safe under that flag.

This is a part of rdar://10803830.

llvm-svn: 150644

as unscanned. // rdar://10832643

llvm-svn: 150639

This reverts commit 9a68d4584afcd0853b930bd80235b58736e785b4.

llvm-svn: 150637

partial types for contexts and forward decls while allowing us to
complete types later on for debug purposes.

This piggy-backs on the metadata replacement and rauw changes
for temporary nodes and takes advantage of the incremental
support I added in earlier. This allows us to, if we decide,
to limit adding methods and variables to structures in order
to limit the amount of debug information output into a .o file.

The caching is a bit complicated though so any thoughts on
untangling that are welcome.

llvm-svn: 150631

is general goodness because representations of member pointers are
not always equivalent across member pointer types on all ABIs
(even though this isn't really standard-endorsed).

Take advantage of the new information to teach IR-generation how
to do these reinterprets in constant initializers.  Make sure this
works when intermingled with hierarchy conversions (although
this is not part of our motivating use case).  Doing this in the
constant-evaluator would probably have been better, but that would
require a *lot* of extra structure in the representation of
constant member pointers:  you'd really have to track an arbitrary
chain of hierarchy conversions and reinterpretations in order to
get this right.  Ultimately, this seems less complex.  I also
wasn't quite sure how to extend the constant evaluator to handle
foldings that we don't actually want to treat as extended
constant expressions.

llvm-svn: 150551

llvm-svn: 150475

llvm-svn: 150459

llvm-svn: 150452

Refactor out new function EmitInitializerForField from EmitMemberInitializer.  The new function will be used to initialize the fields of lambda expressions.

llvm-svn: 150451

llvm-svn: 150435

constructor, and that constructor is used to initialize an object of static
storage duration such that all members and bases are initialized by constant
expressions, constant initialization is performed. In this case, the object
can still have a non-trivial destructor, and if it does, we must emit a dynamic
initializer which performs no initialization and instead simply registers that
destructor.

llvm-svn: 150419

llvm-svn: 150390

while reworking how we handle wanting to emit only parts of structures.

Fixes PR11970.

llvm-svn: 150388

instead of having a special-purpose function.

- ActOnCXXDirectInitializer, which was mostly duplication of
  AddInitializerToDecl (leading e.g. to PR10620, which Eli fixed a few days
  ago), is dropped completely.
- MultiInitializer, which was an ugly hack I added, is dropped again.
- We now have the infrastructure in place to distinguish between
  int x = {1};
  int x({1});
  int x{1};
-- VarDecl now has getInitStyle(), which indicates which of the above was used.
-- CXXConstructExpr now has a flag to indicate that it represents list-
   initialization, although this is not yet used.
- InstantiateInitializer was renamed to SubstInitializer and simplified.
- ActOnParenOrParenListExpr has been replaced by ActOnParenListExpr, which
  always produces a ParenListExpr. Placed that so far failed to convert that
  back to a ParenExpr containing comma operators have been fixed. I'm pretty
  sure I could have made a crashing test case before this.

The end result is a (I hope) considerably cleaner design of initializers.
More importantly, the fact that I can now distinguish between the various
initialization kinds means that I can get the tricky generalized initializer
test cases Johannes Schaub supplied to work. (This is not yet done.)

This commit passed self-host, with the resulting compiler passing the tests. I
hope it doesn't break more complicated code. It's a pretty big change, but one
that I feel is necessary.

llvm-svn: 150318

Basic support for referring to captured variables from lambdas.  Some simple examples seem to work.  Tests coming up soon.

llvm-svn: 150293

This changes function prolog in such a way as to avoid out-of-bounds
stack store in the case when coerce-to type has a larger storage size
than the real argument type.

Fixes PR11905.

llvm-svn: 150238

llvm-svn: 150210

This looks like it had a small, but measurable performance improvement on -O0
compile time for our ObjC tests.

llvm-svn: 150208

Double fields of by-value class objects should be passed in floating point
registers.

llvm-svn: 150200

class objects follow the same rules as those returning struct objects.

llvm-svn: 150196

difference in the qual type. This is a workaround for the fact that
the type isn't artificial but the this decl is, however, we don't
have any way of representing it in the current metadata. For now,
however, just don't cache the full type.

Fixes rdar://10831526 and probably a couple of others.

llvm-svn: 150159

llvm-svn: 150147