Stop folding constant adds into GEP when the type size doesn't match.
Otherwise, the adds' operands are effectively being promoted, changing the
conditions of an overflow.  Results are different when:

    sext(a) + sext(b) != sext(a + b)

Problem originally found on x86-64, but also fixed issues with ARM and PPC,
which used similar code.

<rdar://problem/15292280>

Patch by Duncan Exon Smith!

llvm-svn: 194840

llvm-svn: 194838

Even if we don't support a flag, we should be able to parse it
to provide a better error message than the current default
"error: no such file or directory: '/foo'" (which we should probably
also tweak, btw).

This also tries to clean up the test file a bit.

llvm-svn: 194837

This fixes a crash with GNOME settings manager.

llvm-svn: 194836

llvm-svn: 194835

Up until now we were expecting that when libc++ is installed alongside
clang the headers would be in lib/, which was true if the configure
build was used and false if the cmake build was.

We've now corrected the configure build to install in include/, and
with this change we'll be able to find the correct headers with both
build systems.

llvm-svn: 194834

When using the configure build system, the libc++ headers were being
installed in lib/, whereas cmake installs them in include/. Since
include/ makes more sense for headers, we'll make both systems install
headers there.

llvm-svn: 194833

Patch by James Lyon!

llvm-svn: 194832

Patch by Dale Martin!

llvm-svn: 194831

ivar when property belongs to a super class and
currnt class happens to have a method with same name as
property. // rdar//15473432

llvm-svn: 194830

llvm-svn: 194829

llvm-svn: 194828

Instead of storing the vtable offset directly in the function pointer and
doing a branch to check for virtualness at each call site, the MS ABI
generates a thunk for calling the function at a specific vtable offset,
and puts that in the function pointer.

This patch adds support for emitting such thunks. However, it doesn't support
pointers to virtual member functions that are variadic, have an incomplete
aggregate return type or parameter, or are overriding a function in a virtual
base class.

Differential Revision: http://llvm-reviews.chandlerc.com/D2104

llvm-svn: 194827

Now that FileCheck supports multiple check prefixes, we don't need to keep the
little and big endian versions of this test separate anymore. Merge them back
together.

llvm-svn: 194826

easier to use freshly-built clang with freshly-built libc++.

Basically, this makes it possible to run clang with libc++ without
having to install it, even if you don't have any version of libc++
installed in /usr/

llvm-svn: 194825

llvm-svn: 194824

llvm-svn: 194823

llvm-svn: 194822

llvm-svn: 194821

Add a test.

llvm-svn: 194820

[mips][msa] lowerMSABitClear() should use SelectionDAG::getNOT() instead of using a long-winded equivalent.

Now that getConstant(-1, MVT::v2i64) works correctly on MIPS32 we can use
SelectionDAG::getNOT() to produce the bitmask.

llvm-svn: 194819

llvm-svn: 194818

Summary:
Currently with clang:
$ clang -O20 foo.c
error: invalid value '20' in '-O20'

With the patch:
$ clang -O20 foo.c
warning: optimization level '-O20' is unsupported; using '-O3' instead.
1 warning generated.

This matches the gcc behavior (with a warning added)

Pass all tests:
Testing: 0 .. 10.. 20.. 30.. 40.. 50.. 60.. 70.. 80.. 90..
Testing Time: 94.14s
  Expected Passes    : 6721
  Expected Failures  : 20
  Unsupported Tests  : 17

(which was not the case of http://llvm-reviews.chandlerc.com/D2125)

Reviewers: chandlerc, rafael, rengolin, hfinkel

Reviewed By: rengolin

CC: cfe-commits

Differential Revision: http://llvm-reviews.chandlerc.com/D2152

llvm-svn: 194817

llvm-svn: 194816

Thanks to Kal Conley for the patch

llvm-svn: 194815

By adding a default config.excludes pattern we can avoid individual
suppressions in subdirectories.

This matches LLVM's lit.cfg which also excludes a few other common non-test
filenames for consistency.

llvm-svn: 194814

llvm-svn: 194813

llvm-svn: 194812

Summary:
When getConstant() is called for an expanded vector type, it is split into
multiple scalar constants which are then combined using appropriate build_vector
and bitcast operations.

In addition to the usual big/little endian differences, the case where the
element-order of the vector does not have the same endianness as the elements
themselves is also accounted for.  For example, for v4i32 on big-endian MIPS,
the byte-order of the vector is <3210,7654,BA98,FEDC>. For little-endian, it is
<0123,4567,89AB,CDEF>.
Handling this case turns out to be a nop since getConstant() returns a splatted
vector (so reversing the element order doesn't change the value)

This fixes a number of cases in MIPS MSA where calling getConstant() during
operation legalization introduces illegal types (e.g. to legalize v2i64 UNDEF
into a v2i64 BUILD_VECTOR of illegal i64 zeros). It should also handle bigger
differences between illegal and legal types such as legalizing v2i64 into v8i16.

lowerMSASplatImm() in the MIPS backend no longer needs to avoid calling
getConstant() so this function has been updated in the same patch.

For the sake of transparency, the steps I've taken since the review are:
* Added 'virtual' to isVectorEltOrderLittleEndian() as requested. This revealed
  that the MIPS tests were falsely passing because a polymorphic function was
  not actually polymorphic in the reviewed patch.
* Fixed the tests that were now failing. This involved deleting the code to
  handle the MIPS MSA element-order (which was previously doing an byte-order
  swap instead of an element-order swap). This left
  isVectorEltOrderLittleEndian() unused and it was deleted.
* Fixed build failures caused by rebasing beyond r194467-r194472. These build
  failures involved the bset, bneg, and bclr instructions added in these commits
  using lowerMSASplatImm() in a way that was no longer valid after this patch.
  Some of these were fixed by calling SelectionDAG::getConstant() instead,
  others were fixed by a new function getBuildVectorSplat() that provided the
  removed functionality of lowerMSASplatImm() in a more sensible way.

Reviewers: bkramer

Reviewed By: bkramer

CC: llvm-commits

Differential Revision: http://llvm-reviews.chandlerc.com/D1973

llvm-svn: 194811

Using a special machine node is cleaner than an InlineAsm node, and fixes an assertion failure in InstrEmitter

llvm-svn: 194810

llvm-svn: 194809

llvm-svn: 194808

In response to post-commit feedback by Jordan, this is required to
make sure path-sensitive checks work correctly.

llvm-svn: 194807

Summary:
This patch (correctly) breaks some MSA tests by exposing the cases when
SelectionDAG::getConstant() produces illegal types. These have been temporarily
marked XFAIL and the XFAIL flag will be removed when
SelectionDAG::getConstant() is fixed.

There are three categories of failure:
* Immediate instructions are not selected in one endian mode.
* Immediates used in ldi.[bhwd] must be different according to endianness.
  (this only affects cases where the 'wrong' ldi is used to load the correct
   bitpattern. E.g. (bitcast:v2i64 (build_vector:v4i32 ...)))
* Non-immediate instructions that rely on immediates affected by the
  previous two categories as part of their match pattern.
  For example, the bset match pattern is the vector equivalent of
  'ws | (1 << wt)'.

One test needed correcting to expect different output depending on whether big
or little endian was in use. This test was
test/CodeGen/Mips/msa/basic_operations.ll and experiences the second category
of failure shown above. The little endian version of this test is named
basic_operations_little.ll and will be merged back into basic_operations.ll in
a follow up commit now that FileCheck supports multiple check prefixes.

Reviewers: bkramer, jacksprat, dsanders

Reviewed By: dsanders

CC: llvm-commits

Differential Revision: http://llvm-reviews.chandlerc.com/D1972

llvm-svn: 194806

llvm-svn: 194805

llvm-svn: 194804

this allows gdb to unwind through our hacky call

llvm-svn: 194803

location, update this reference to reflect that.

llvm-svn: 194802

externally to simplify our integration of GoogleTest into LLVM. Also,
build the single source file gtest-all.cc instead of the individual
source files as we don't expect these to change and thus gain nothing
from increased incrementality in compiles.

This makes our standard build of googletest exactly like upstream's
recommended build and the sanitizer's build. It also simplifies the
steps of importing a new version should we ever want one.

llvm-svn: 194801

llvm-svn: 194800