The recovery was failing due to a missing case in SkipUntil().

Also add back tests from r197553 that were reverted in the previous commit.

llvm-svn: 197598

These parser changes were redundant. The same or better recovery can be
achieved with a one-line fix to SkipUntil() due to land in the next commit.

This reverts commit r197553.

llvm-svn: 197597

(prompted by NAKAMURA Takumi)

llvm-svn: 197596

llvm-svn: 197595

This will cause llvm-cov to output branch counts instead of branch
probabilities. -b must be enabled.

Also updated tests.

llvm-svn: 197594

llvm-svn: 197593

llvm-svn: 197592

llvm-svn: 197591

This changes Clang standalone builds so that clang-tblgen lives in
clang/build/bin instead of llvm/build/bin, and so that with the Xcode
generator it's in clang/build/bin/Debug instead of llvm/build/bin/Debug/Debug/.
Yes, really.

llvm-svn: 197590

1. The arch directive now appears before the cpu directive
2. Long run lines were split across multiple lines
    
No functional changes.

llvm-svn: 197588

According to "Addenda to ABI for ARM architecture", Tag_FP_arch is the
new name for the equivalent Tag_VFP_arch.  This commit renames
Tag_VFP_arch to Tag_FP_arch.

llvm-svn: 197587

// rdar://15397430

llvm-svn: 197586

I am about to send a patch for review touching these and clang-formating first
makes the patch much easier to read.

llvm-svn: 197585

Implement the MSABI and SysVABI calling conventions for Objective-C method declarations. This appears to be an omission from r189644.

llvm-svn: 197584

I will introduce another flag, like --system-libs, later.

llvm-svn: 197583

Adding some comments about AST node requirements for attributes which create AST nodes but never actually make use of them. No functional changes.

llvm-svn: 197582

llvm-svn: 197581

llvm-svn: 197580

During testing I observed QEMU send "$T02thread:01;#04" upon connection,
before any command from LLDB.  This change from gclayton accepts (and
discards) a packet immediately after sending the initial ack, to flush
the GDB remote pipeline.

llvm-svn: 197579

This commit kills off custom type specifier and keyword handling of OpenCL C
data types.

Although the OpenCL spec describes them as keywords, we can handle them more
elegantly as predefined types. This should provide better error correction and
code completion as well as simplifying the implementation.

The primary intention is however to simplify the C/C++ parser and save some
packed bits on AST structures that had been extended in r170432 just for
OpenCL.

llvm-svn: 197578

A f64 inside a struct can be 32 bit aligned on darwin.

llvm-svn: 197577

llvm-svn: 197576

llvm-svn: 197575

This patch adds -f64:32:64 to 32 bit ppc darwin since a f64 inside a
structure are only 32 bit aligned.

The patch also drop -f128:64:128 from all ppc darwin, since f128 is
128 bit aligned.

llvm-svn: 197574

llvm-svn: 197573

Thanks for Iain Sandoe for testing this with the original gcc.

Clang was already getting this right.

llvm-svn: 197572

llvm-svn: 197571

llvm-svn: 197570

llvm-svn: 197569

This code is not robust enough and triggers when simply linking with
libdynamorio.so, without any code translation at all. Disabling it is safe
(i.e. we may unpoison too much memory and see false negatives, but never false
positives).

llvm-svn: 197568

The polly on-commit formatting checker notified me that I forgot to format
the code before submitting.

llvm-svn: 197560

   
No functional changes.

llvm-svn: 197559

We now report the following:

$ polly-clang -O3 -mllvm -polly -mllvm -polly-report test.c  -c \
  -gline-tables-only

note: Polly detected an optimizable loop region (scop) in function 'foo'
test.c:2: Start of scop
test.c:3: End of scop
note: Polly detected an optimizable loop region (scop) in function 'bar'
test.c:9: Start of scop
test.c:13: End of scop

llvm-svn: 197558

Before:
  aaaaaaaaaa(aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa(
      aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa)
                 .aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa);
  aaaaaaaaaa = aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa(
      aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa)
                   .aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa();

After:
  aaaaaaaaaa(aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa(
                 aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa)
                 .aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa);
  aaaaaaaaaa = aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa(
                   aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa)
                   .aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa();

Probably still not ideal, but should be a step into the right direction.

llvm-svn: 197557

llvm-svn: 197556

This should fix the ARM bots.

llvm-svn: 197555

Clang sets the float-abi target option manually, but no longer
annotates each function with its ABI. This can lead to confusing
mistmatch between "clang -emit-llvm | llc" and normal clang
invocations.

Besides which, gnueabihf actually *is* hard-float. Defaulting to soft
was just perverse.

llvm-svn: 197554

llvm-svn: 197553

Before:
  bool a = ([aaaaaaaa aaaaa] == aaaaaaaaaaaaaaaaa || [aaaaaaaa aaaaa] ==
                                                         aaaaaaaaaaaaaaaaaaaa);

After:
  bool a = ([aaaaaaaa aaaaa] == aaaaaaaaaaaaaaaaa ||
            [aaaaaaaa aaaaa] == aaaaaaaaaaaaaaaaaaaa);

This fixes llvm.org/PR18271.

llvm-svn: 197552

llvm-svn: 197551