llvm-svn: 176302

This was causing the folding set to fail to fold attributes, because it was
being calculated in one spot without an empty values string but here with an
empty values string.

llvm-svn: 176301

The instcombine recognized pattern looks like:
a = b * c
d = a +/- Cst
or
a = b * c
d = Cst +/- a

When creating the new operands for fadd or fsub instruction following the related fmul, the first operand was created with the second original operand (M0 was created with C1) and the second with the first (M1 with Opnd0).

The fix consists in creating the new operands with the appropriate original operand, i.e., M0 with Opnd0 and M1 with C1.

llvm-svn: 176300

llvm-svn: 176298

our bitwise compare is equal to the field we're looking for.

Noticed on inspection.

llvm-svn: 176296

Some buildbots use a shell that doesn't know 'exit'. So instead of using XFAIL
for UseAuto/iterator.cpp (since this problem doesn't exist for cmake) leave the
'grep' command and turn off the others.

llvm-svn: 176295

llvm-svn: 176294

    
The way 'make check-all' causes configuration files to be generated results in
the cpp11-migrate/Makefile not being triggered to generated any files. These
files are necessary for UseAuto/iterator.cpp to run. Turning off this test
until the problem is fixed properly.

llvm-svn: 176293

llvm-svn: 176292

at a time.  Enforce this for now (we should relax the requirement when we have a little more time.)

<rdar://problem/13314462>

llvm-svn: 176291

llvm-svn: 176289

llvm-svn: 176288

We make the cost for calling libm functions extremely high as emitting the
calls is expensive and causes spills (on x86) so performance suffers. We still
vectorize important calls like ceilf and friends on SSE4.1. and fabs.

Differential Revision: http://llvm-reviews.chandlerc.com/D466

llvm-svn: 176287

calls and declarations.

LLVM has a default CC determined by the target triple.  This is
not always the actual default CC for the ABI we've been asked to
target, and so we sometimes find ourselves annotating all user
functions with an explicit calling convention.  Since these
calling conventions usually agree for the simple set of argument
types passed to most runtime functions, using the LLVM-default CC
in principle has no effect.  However, the LLVM optimizer goes
into histrionics if it sees this kind of formal CC mismatch,
since it has no concept of CC compatibility.  Therefore, if this
module happens to define the "runtime" function, or got LTO'ed
with such a definition, we can miscompile;  so it's quite
important to get this right.

Defining runtime functions locally is quite common in embedded
applications.

llvm-svn: 176286

llvm-svn: 176285

llvm-svn: 176283

Fix the getCompilerVersion() function to do the right thing with clang 5.0 which now says "LLVM version" instead of "clang version".

llvm-svn: 176282

llvm-svn: 176281

llvm-svn: 176280

llvm-svn: 176279

llvm-svn: 176278

llvm-svn: 176277

llvm-svn: 176276

llvm-svn: 176275

Also removed the use of llvm::raw_string_ostream as it wasn't needed. 

Also fixed a crasher that could occur when the following line returned a C string tied to a local variable:

    return OS.str().c_str();

I am guessing "static std::string buf;" was supposed to get assigned to "OS.str()" and then have "buf.c_str()" returned. 

Modified the non-apple version code to cache its value and not recompute the version every time.

llvm-svn: 176274

other per-instruction statistics.

llvm-svn: 176273

llvm-svn: 176272

have their own code completion comments. Use those in 
their properties in this case. 
// rdar://12791315

llvm-svn: 176271

llvm-svn: 176270

llvm-svn: 176269

- make LLDB version number match Clang (and the Debian package)
- use the same revision detection magic that Clang uses to report SVN/Git revisions
- update test case as per above

Example output:

  $ lldb -v
  lldb version 3.3 (https://dmalea@llvm.org/svn/llvm-project/lldb/trunk revision 176211 clang revision 176208 llvm revision 176208)
ss

This line, and those below, will be ignored--

M    source/lldb.cpp
M    test/help/TestHelp.py
M    source/Makefile

llvm-svn: 176268

Patch by Ariel Bernal.

llvm-svn: 176267

The new Use-Auto transform replaces the type specifier for variable
declarations with the special C++11 'auto' type specifier. For now, the
replacement is done only for variables that are iterators of any of the
std containers and only if the type used is one of those explicitly
allowed by the standard (i.e. not an implementation-specific type).

Reviewers: gribozavr, silvas, klimek
llvm-svn: 176266

llvm-svn: 176264

This fixes llvm.org/PR15379.

Before:
const uint8_t aaaaaaaaaaaaaaaaaaaaaa[0] = { 0x00, 0x00, 0x00, 0x00, 0x00,
                                            0x00,                  // comment
                                            0x00, 0x00, 0x00, 0x00, 0x00,
                                            0x00,                  // comment
                                            0x00, 0x00, 0x00, 0x00 // comment
};

After:
const uint8_t aaaaaaaaaaaaaaaaaaaaaa[0] = {
  0x00, 0x00, 0x00, 0x00, 0x00, 0x00, // comment
  0x00, 0x00, 0x00, 0x00, 0x00, 0x00, // comment
  0x00, 0x00, 0x00, 0x00              // comment
};

llvm-svn: 176262

The work done by the post-encoder (setting architecturally unused bits to 0 as
required) can be done by the existing operand that covers the "#0.0". This
removes at least one use of the discouraged PostEncoderMethod uses.

llvm-svn: 176261

Before:
void aaaaaaaaaaaaaaaaaa() __attribute__(
    (aaaaaaaaaaaaaaaaaaaaaaaaa, aaaaaaaaaaaaaaaaaaaaaaaaa,
     aaaaaaaaaaaaaaaaaaaaaaaaa));

After:
void aaaaaaaaaaaaaaaaaa()
    __attribute__((aaaaaaaaaaaaaaaaaaaaaaaaa, aaaaaaaaaaaaaaaaaaaaaaaaa,
                   aaaaaaaaaaaaaaaaaaaaaaaaa));

llvm-svn: 176260

If an otherwise weak var is actually defined in this unit, it can't be
undefined at runtime so we can use normal global variable sequences (ADRP/ADD)
to access it.

llvm-svn: 176259

llvm-svn: 176258

- Consistency with opt (which supports the same option with the same meaning and
  description).
- Debugging gold plugin-based linking without optimizations getting in the way.
- Debugging programs linked with the gold plugin while preserving the original
  debug info.
- Fine-grained control over LTO passes using the gold plugin in combination with
  opt (or clang/dragonegg).

Patch by Cristiano Giuffrida!

llvm-svn: 176257