Add an API on ValueObject to generate a 'synthetic child' of base class type. Note that in this commit, the term synthetic child is not meant to refer to data formatters, but to the programmatically-generated children stored inside a ValueObject itself

llvm-svn: 216483

Depending on llvm-config allows for a few nice things (auto-detecting
LLVM source tree, version numbers, etc), but it's makes bootstrapping a
pain. Keep the llvm-config features around, but also fall back to being
able to configure based on -DLLVM_PATH=path/to/llvm.

Reviewers: jroelofs, loladiro

Reviewed By: loladiro

Subscribers: cfe-commits

Differential Revision: http://reviews.llvm.org/D5016

llvm-svn: 216482

The existing matcher has lots of AT&T assembly dialect assumptions baked
into it.  In particular, the hack for resolving the size of a memory
operand by appending the four most common suffixes doesn't work at all.
The Intel assembly dialect mnemonic table has ambiguous entries, so we
need to try matching multiple times with different operand sizes, since
that's the only way to choose different instruction variants.

This makes us more compatible with gas's implementation of Intel
assembly syntax.  MSVC assumes you want byte-sized operations for the
instructions that we reject as ambiguous.

Reviewed By: grosbach

Differential Revision: http://reviews.llvm.org/D4747

llvm-svn: 216481

[libclang] Fixup the cursor spelling range for C++ destructors, operator overloads, and conversion functions.

Patch by Craig Tenenbaum!

llvm-svn: 216480

llvm-svn: 216479

llvm-svn: 216478

llvm-svn: 216477

llvm-svn: 216476

PR 20642.

llvm-svn: 216475

lowering of the intrinsics.
Prior to this commit, most of the copy-related intrinsics could be optimized
away. The situation is still not ideal as there are several possibilities to
lower a given intrinsic. Currently, we match LLVM behavior.

llvm-svn: 216474

And likewise for qProcessInfo on Linux, but ensures cputype/cpusubtype is not defined.  The
Apple case is the more important one, since we take a slightly different path to initialize
ProcessGDBRemote-related remote host/process info if triple is present.

Related to http://llvm.org/bugs/show_bug.cgi?id=20755.

llvm-svn: 216473

flag.

llvm-svn: 216472

proper flag.

llvm-svn: 216471

This change addresses this bug:
http://llvm.org/bugs/show_bug.cgi?id=20755

This change:
* Modifies llgs to send triple instead of cputype and cpusubtype when not on Apple platforms in qProcessInfo.
* Modifies lldb's GDBRemoteCommunicationClient to handle the triple returned from qProcessInfo if given.
  When given, it will prefer to use triple over cputype and cpusubtype.
* Adds gdb-remote protocol tests to verify that cputype and cpusubtype are specified on darwin, and that triple is specified on Linux.

llvm-svn: 216470

feature is c11 about nested struct declarations must have
struct-declarator-list. Without this change, code
which was meant for c99 breaks. rdar://18125536

llvm-svn: 216469

Thanks to David Blaikie for the suggestion.

llvm-svn: 216468

llvm-svn: 216467

llvm-svn: 216466

I'm about to add some more qProcessInfo tests so I wanted to first pull them out
of the monolithic TestLldbGdbServer test case class.

Related to http://llvm.org/bugs/show_bug.cgi?id=20755

llvm-svn: 216465

The memory management in BugPoint is fairly convoluted, so this just unwraps
one layer by changing the return type of functions that always return
owned Modules.

llvm-svn: 216464

llvm-svn: 216463

we can also fix the original header guard.

We still allow an _ at the end of a header guard since it's so common, but
remove it now when the #endif comment is changed.

llvm-svn: 216462

Just call parseBitcodeFile instead of getLazyBitcodeModule followed by
materializeAllPermanently.

llvm-svn: 216461

We had two functions for finding the temp or cache directory. Each had a
different set of smarts about OS specific APIs.

With this patch system_temp_directory becomes the only way to do it.

llvm-svn: 216460

llvm-svn: 216459

llvm-svn: 216458

Some versions of MSVC do not support initializer list construction of vectors, so this fixes a broken build from r216385.

llvm-svn: 216457

This test was updated in r216397, but was failing on Windows due to mixed path separators as well as case sensitivity of the "no" in "no such file or directory." Rather than revert this file back to its original form, I've made some incredibly ugly regexes so that it will pass everywhere.

Note, the path this test reports a failure on (for my Windows setup) is: E:\llvm\llvm\tools\clang\test\Frontend/doesnotexist/somename

llvm-svn: 216456

It seems on Darwin the illegal round-trip ::iterator -> MachineInstr* -> ::iterator breaks execution horribly when the iterator is not a real MachineInstr, like ::end().

llvm-svn: 216455

llvm-svn: 216454

This completes all ACLE hint intrinsics.

llvm-svn: 216453

llvm-svn: 216452

llvm-svn: 216451

This warning makes it a bit hard to define strlen and wcslen interceptors in an elegant yet portable way

llvm-svn: 216450

Before:
  int a[5];
  a[3] += 42;

After:
  int a[ 5 ];
  a[ 3 ] += 42;

Fixes LLVM bug #17887 (http://llvm.org/bugs/show_bug.cgi?id=17887).

Patch by Marek Kurdej, thank you!

llvm-svn: 216449

llvm-svn: 216448

With this patch, "check-asan" passes all the tests with both MT and MD ASan RTL if you set COMPILER_RT_BUILD_SHARED_ASAN to ON
(PR20214)

llvm-svn: 216447

Insert the LDREX/STREX instruction sequence specified in ARM ACLE 2.0,
as SWP instruction is deprecated since ARMv6.

llvm-svn: 216446

Before:
  std::vector<int> v = {
      1, 0 /* comment */
  };

After:
  std::vector<int> v = {1, 0 /* comment */};

llvm-svn: 216445

(X >> Z) & (Y >> Z)  -> (X&Y) >> Z  for all shifts.
(X >> Z) | (Y >> Z)  -> (X|Y) >> Z  for all shifts.
(X >> Z) ^ (Y >> Z)  -> (X^Y) >> Z  for all shifts.

These patterns were previously handled separately in visitAnd()/visitOr()/visitXor().

Differential Revision: http://reviews.llvm.org/D4951

llvm-svn: 216443