values in a map that can be passed to consumers. Add a testcase that
ensures this works for llvm-dwarfdump.

llvm-svn: 167558

<rdar://problem/12153915>  (partial fix)

Remove an assert and place an error message instead so we don't crash when we run into a type tag that we don't recognize. We will now emit a warning so that hopefully we can get a bug report that has example code that shows what we are missing.

Also fixed a case when trying to unique one type to another where we would confuse concrete instances of methods with their definitions and end up not correctly registering the types.

llvm-svn: 167557

is available.

llvm-svn: 167552

llvm-svn: 167551

the base class.  If the base class deduction succeeds, use those results.  If
it fails, keep using the results from the derived class template deduction.

This prevents an assertion later where the type of deduction failure doesn't
match up with the template deduction info.

llvm-svn: 167550

captured block variable layout meta-data. No intended change
in functionality.

llvm-svn: 167549

Patch by Sasa Stankovic.

llvm-svn: 167548

llvm-svn: 167547

llvm-svn: 167546

llvm-svn: 167545

llvm-svn: 167544

llvm-svn: 167543

directive on a new line.

llvm-svn: 167542

This also changes the default architecture emitted for a generic CPU to "v7".

llvm-svn: 167540

module in place. <rdar://problem/10138913>

llvm-svn: 167539

Patch by Philip Craig.

llvm-svn: 167538

Patch by Sean McBride!

llvm-svn: 167537

llvm-svn: 167536

llvm-svn: 167535

a bug in the inliner still causes the wrong thing to happen at -O2 and above
(PR14116).

llvm-svn: 167534

llvm-svn: 167533

llvm-svn: 167532

llvm-svn: 167530

[asan] fix bug 14277 (asan needs to fail with fata error if an __asan interface function is being redefined. Before this fix asan asserts)

llvm-svn: 167529

misched is disabled by default. With -enable-misched, these heuristics
balance the schedule to simultaneously avoid saturating processor
resources, expose ILP, and minimize register pressure. I've been
analyzing the performance of these heuristics on everything in the
llvm test suite in addition to a few other benchmarks. I would like
each heuristic check to be verified by a unit test, but I'm still
trying to figure out the best way to do that. The heuristics are still
in considerable flux, but as they are refined we should be rigorous
about unit testing the improvements.

llvm-svn: 167527

instructions without relying on liveintervals.

llvm-svn: 167526

llvm-svn: 167525

updating an abstract DIE or not. If we are, then we use that. Its children will
be added on later, as well as the object pointer attribute. Otherwise, this
function may be called with a concrete DIE twice and adding the children and
object pointer attribute to it twice.
<rdar://problem/12401423&12600340>

llvm-svn: 167524

No functionality change (doc comments only).

llvm-svn: 167523

No functionality change; this checker is only used for documentation.

llvm-svn: 167522

llvm-svn: 167521

x/9i

we actually calculated the size of 9 instructions.  Then we multiplied it by the count again 
to get the total amount we should fetch, so we thought 9 x86_64 instructions took over 1K
to fetch...

<rdar://problem/12649027>

llvm-svn: 167520

llvm-svn: 167518

llvm-svn: 167517

Also fixed a bug where the Objective-C runtime
would not properly report that it found a class if
(a) it had to build the ObjCInterfaceDecl for the
class and (b) logging was enabled.

<rdar://problem/12641180>

llvm-svn: 167516

llvm-svn: 167515

instantiate it if it can be instantiated and implicitly define it if it can be
implicitly defined. This matches g++'s approach. Remove some cases from
SemaOverload which were marking functions as referenced when just planning how
overload resolution would proceed; such cases are not actually references.

llvm-svn: 167514

llvm-svn: 167511

llvm-svn: 167510

This is preparation for adding Decl dumping.

Patch by Philip Craig.

llvm-svn: 167509