While this generally helped x86-64, there was some large regressions
for i386.

llvm-svn: 136571

rules to the new explicitly listed TableGen rules. Somehow I missed this
in my original sweep.

llvm-svn: 136567

appropriate sentinel at the end of the argument list. Also, put the
sentinel warnings under -Wsentinel. Fixes <rdar://problem/8764236>.

llvm-svn: 136566

sub-library for the targets depended on the core target CodeGen library.
This completely undermined the careful work to separate the those
libraries, especially the MC-layer ones. This surfaced as circular
dependencies when the libraries were built as shared libraries where
CMake doesn't allow cycles.

This should fix PR10537. I'll watch the bots to see if there is fallout
on other platforms.

llvm-svn: 136565

statement. Fixes <rdar://problem/9229438>.

llvm-svn: 136564

variadic bit (", ..." or ", args...") into the prior placeholder, like
we do with functions and methods. Fixes <rdar://problem/9740808>.

llvm-svn: 136563

(such as in, inout, byref, and oneway) to code completion
results. Fixes <rdar://problem/8844158>.

llvm-svn: 136562

llvm-svn: 136561

libstdc++ hack has reverted these type traits to keywords. Icky, but
fixes <rdar://problem/9836262>.

llvm-svn: 136560

already-defined and forward-declared results. Already-defined results
are fine because they could be the start of a category. Fixes
<rdar://problem/9811691>.

llvm-svn: 136559

special diagnostic for ARC ownership-qualified types. We wouldn't want
to expose Objective-C programmers to the term "POD", would we? Fixes
<rdar://problem/9772982>.

llvm-svn: 136558

a fallback, if normal header search fails. Another attempt at
<rdar://problem/9824020>.

llvm-svn: 136557

r136339, r136341, r136369, r136387, r136392, r136396, r136429, r136430, r136444,
r136445, r136446, r136253 pending review.

llvm-svn: 136556

completes the support in the LLDB expression parser
for incomplete types.  Clang now imports types
lazily, and we complete those types as necessary.

Changes include:

- ClangASTSource now supports three APIs which it
  passes to ClangExpressionDeclMap.  CompleteType
  completes a TagDecl or an ObjCInterfaceDecl when
  needed; FindExternalVisibleDecls finds named
  entities that are visible in the expression's
  scope; and FindExternalLexicalDecls performs a
  (potentially restricted) search for entities
  inside a lexical scope like a namespace.  These
  changes mean that entities in namespaces should
  work normally.

- The SymbolFileDWARF code for searching a context
  for a specific name is now more general, and can
  search arbitrary contexts.

- We are continuing to adapt our calls into LLVM
  from interfaces that take start and end iterators
  when accepting multiple items to interfaces that
  use ArrayRef.

- I have cleaned up some code, especially our use
  of namespaces.

This change is neutral for our testsuite and greatly
improves correctness for large programs (like Clang)
with complicated type systems.  It should also lay
the groundwork for improving the expression parser's
performance as we are lazier and lazier about
providing type information.

llvm-svn: 136555

we have a nil NSString *. Also added blank lines between functions in the
CFString.py files.

llvm-svn: 136554

The test driver now takes an option "+b" which enables to run just the benchmarks tests.
By default, tests decorated with the @benchmarks_test decorator do not get run.

Add an example benchmarks test directory which contains nothing for the time being,
just to demonstrate the @benchmarks_test concept.

For example,

$ ./dotest.py -v benchmarks

...

----------------------------------------------------------------------
Collected 2 tests

1: test_with_gdb (TestRepeatedExprs.RepeatedExprssCase)
   Test repeated expressions with gdb. ... skipped 'benchmarks tests'
2: test_with_lldb (TestRepeatedExprs.RepeatedExprssCase)
   Test repeated expressions with lldb. ... skipped 'benchmarks tests'

----------------------------------------------------------------------
Ran 2 tests in 0.047s

OK (skipped=2)
$ ./dotest.py -v +b benchmarks

...

----------------------------------------------------------------------
Collected 2 tests

1: test_with_gdb (TestRepeatedExprs.RepeatedExprssCase)
   Test repeated expressions with gdb. ... running test_with_gdb
benchmarks result for test_with_gdb
ok
2: test_with_lldb (TestRepeatedExprs.RepeatedExprssCase)
   Test repeated expressions with lldb. ... running test_with_lldb
benchmarks result for test_with_lldb
ok

----------------------------------------------------------------------
Ran 2 tests in 0.270s

OK

Also mark some Python API tests which are missing the @python_api_test decorator.

llvm-svn: 136553

llvm-svn: 136552

llvm-svn: 136551

Fixed again. // rdar://9846759

llvm-svn: 136550

for copying a captured block object. // rdar://9846759

llvm-svn: 136549

The ARM target depends on CPSR liveness being tracked after register
allocation.

llvm-svn: 136548

implementation of previous().

llvm-svn: 136547

the hash_table. I think this is the correct solution to PR10507, but I'm
not sure since this is a little bit cargo-culted. Howard, please review.

llvm-svn: 136546

more than just a bit broken. This one should compile and run without
infinite loops.

llvm-svn: 136545

changes in the new GetMinimumLanguages() ; robustness improvements in the CFStringSynthProvider object ; made a CFString_SummaryProvider function you can use if all you care about is the summary string for your NSString objects

llvm-svn: 136544

globally scoped constructs. Also, round-trip these dependencies through
the LLVMConfig.cmake.in file thata is used by CMake-based clients of
"installed" (or built) LLVM trees.

llvm-svn: 136543

together to get it working with code, and is neither optimal
(erase(Iterator, Iterator) calculates the previous iterator twice,
rather than calculating the previous iterator of the first one, then
advancing it until the second is found) nor complete (splice() was not
implemented). Most of the implementation is borrowed from forward_list
via using-declarations.

llvm-svn: 136542

This includes registers like EFLAGS and ST0-ST7. We don't check for
liveness issues in the verifier and scavenger because registers will
never be allocated from these classes.

While in SSA form, we do care about the liveness of unallocatable
unreserved registers. Liveness of EFLAGS and ST0 neds to be correct for
MachineDCE and MachineSinking.

llvm-svn: 136541

avoids bugs where, when the allocator's size_type was smaller than int,
the multiplication or division would cause integral promotions and, with
two different integer types as arguments, deduction of the template
arguments would fail.

llvm-svn: 136540

are two motivations for this.

First, this allows users who are specializing __gnu_ext::hash to
continue doing so without changing their code.

Second, SGI specifies hash overloads for char* and const char* that
perform a hash of the string, not of the pointer.

In order to support this, the hashing code for string is factored out.

llvm-svn: 136539

the type name 'data_type', which is specified by the SGI spec as being
the correct type name for the mapped type. The second is an overload of
insert found in standard containers, taking an iterator as a 'hint'
(which we ignore in the standard containers as well). libstdc++'s
implementation includes these overloads, and they are needed to make
insert_iterator work (which I suspect is the real motivation for
including them in the standard containers).

The motivation for including these overloads of insert and leaving the
mapped_type typedef is to make it easier for clients to migrate to the
standard containers.

llvm-svn: 136538

imported a forward declaration, but later the full definition of the
same entity becomes available. When this happens, import the definition.

llvm-svn: 136537

llvm-svn: 136536

llvm-svn: 136535

llvm-svn: 136534

to find out the tests which failed/errored and need re-running.  The dotest.py test driver
script is modified to allow specifying multiple -f testclass.testmethod in the command line
to accommodate the redo functionality.

An example,

 $ ./redo.py -n 2011-07-29-11_50_14
adding filterspec: TargetAPITestCase.test_find_global_variables_with_dwarf
adding filterspec: DisasmAPITestCase.test_with_dsym
Running ./dotest.py -v  -f TargetAPITestCase.test_find_global_variables_with_dwarf -f DisasmAPITestCase.test_with_dsym

...

----------------------------------------------------------------------
Collected 2 tests

1: test_with_dsym (TestDisasmAPI.DisasmAPITestCase)
   Exercise getting SBAddress objects, disassembly, and SBAddress APIs. ... ok
2: test_find_global_variables_with_dwarf (TestTargetAPI.TargetAPITestCase)
   Exercise SBTarget.FindGlobalVariables() API. ... ok

----------------------------------------------------------------------
Ran 2 tests in 15.328s

OK

llvm-svn: 136533

This flag is true from isel to register allocation when the machine
function is required to be in SSA form.  The TwoAddressInstructionPass
and PHIElimination passes clear the flag.

The SSA flag wil be used by the machine code verifier to check for SSA
form, and eventually an assertion can enforce it in +Asserts builds.
This will catch the common target error of creating machine code with
multiple defs of a virtual register.

llvm-svn: 136532

Remove const qualifiers from Init references, per Chris' request.

llvm-svn: 136531

llvm-svn: 136530

llvm-svn: 136529