MSan instrumentation is driven by the original code. We take every
incoming instruction and emit another instruction (or ten) next to
it, operating on the shadow values (but sometimes on the real values,
too). Two programs in one, essentially. There can be any kinds of
redundancies in the second one, so we just run whatever is normally
run at -O2, and then exclude some passes that do not help much with
benchmarks.

llvm-svn: 174049

statements were only true for 65 CLs after they were introduced two years ago.

llvm-svn: 174043

constructor, retrieve our VTT parameter directly. Fixes PR14588 /
<rdar://problem/12867962>.

llvm-svn: 174042

have a direct mismatch between some component of the template and some
component of the argument. The diagnostic now says what the mismatch was, but
doesn't yet say which part of the template doesn't match.

llvm-svn: 174039

Objective-C message receiver, the user is as likely to want to write a
type name as any other declaration, so give types the same priority as
other declarations. Fixes <rdar://problem/12480600>.

llvm-svn: 174038

Allow the computation of the base priority for a declaration code completion result to consider the completion context

llvm-svn: 174037

in the one case where we've already factored out a reason code.

llvm-svn: 174036

filter the elements before emitting them into a PCH. No user-visible
functionality change, except that PCH files may be smaller?

llvm-svn: 174034

llvm-svn: 174033

llvm-svn: 174032

This is a hack to work around the fact that we don't track extents for our
default bindings:

  CGPoint p;
  p.x = 0.0;
  p.y = 0.0;
  rectParam.origin = p;
  use(rectParam.size); // warning: uninitialized value in rectParam.size.width

In this case, the default binding for 'p' gets copied into 'rectParam',
because the 'origin' field is at offset 0 within CGRect. From then on,
rectParam's old default binding (in this case a symbol) is lost.

This patch silences the warning by pretending that lazy bindings are never
made from uninitialized memory, but not only is that not true, the original
default binding is still getting overwritten (see FIXME test cases).
The long-term solution is tracked in <rdar://problem/12701038>

PR14765 and <rdar://problem/12875012>

llvm-svn: 174031

them the same if they are actually the same; having the same name isn't enough.

Similar to r174013, template template arguments were also mistakenly considered
the same when they had the same name but were in different namespaces.
In addition, when printing template template arguments, use the qualified name
if the regular name is the same.

llvm-svn: 174029

llvm-svn: 174025

Indents were given the color blue when outputting with color.
AST dumping now looks like this:

Node
|-Node
| `-Node
`-Node
  `-Node

Compared to the previous:

(Node
  (Node
    (Node))
  (Node
    (Node)))

llvm-svn: 174022

a workaround for this bug from the -Wundefined-internals warning.

llvm-svn: 174020

positives.

The includeSuffix was only set on the first iteration through the
function, resulting in invalid regions being produced by getLazyBinding
(ex: zoomRegion.y).

llvm-svn: 174016

Don't expect a character between the double-quotes and 'ld'.

llvm-svn: 174015

same if they are actually the same; having the same name isn't
enough. Fixes <rdar://problem/12931988>.

llvm-svn: 174013

Several places were still treating the Attribute object as respresenting
multiple attributes. Those places now use the AttributeSet to represent
multiple attributes.

llvm-svn: 174004

llvm-svn: 173996

  #define X X

for which there is no point warning, ever.

llvm-svn: 173991

Thanks, Sean.

llvm-svn: 173981

llvm-svn: 173980

file contents being autogenerated

llvm-svn: 173979

llvm-svn: 173978

Specifically, don't print snippets, caret diagnostics, or ranges for
lines over 4096 characters. We copy the line around a few times in our
diagnostics machinery, and we have to print a caret line that's just as
long. This uses a lot of memory just to create a poor user experience as
we print out a line much too long for anyone to read...or spend extra
energy trying to fit it to -fmessage-length.

<rdar://problem/13106850>

llvm-svn: 173976

This reverts commit r173952

llvm-svn: 173970

to printing the default case.  This is a fix for PR15023.

llvm-svn: 173965

llvm-svn: 173963

llvm-svn: 173961

This caused hangs as we processed the same invalid byte over and over.

<rdar://problem/13115651>

llvm-svn: 173959

Redefine the shallow mode to inline all functions for which we have a
definite definition (ipa=inlining). However, only inline functions that
are up to 4 basic blocks large and cut the max exploded nodes generated
per top level function in half.

This makes shallow faster and allows us to keep inlining small
functions. For example, we would keep inlining wrapper functions and
constructors/destructors.

With the new shallow, it takes 104s to analyze sqlite3, whereas
the deep mode is 658s and previous shallow is 209s.

llvm-svn: 173958

llvm-svn: 173957

llvm-svn: 173956

Thanks Jordan!

llvm-svn: 173955

llvm-svn: 173954

Just makes the code a little cleaner, and easier to reason about.

llvm-svn: 173953

for "#define X X".

This is a pattern that, for example, stdbool.h uses.
rdar://12435773

llvm-svn: 173952

This is faster for the analyzer to process than inlining the constructor
and performing a member-wise copy, and it also solves the problem of
warning when a partially-initialized POD struct is copied.

Before:
  CGPoint p;
  p.x = 0;
  CGPoint p2 = p; <-- assigned value is garbage or undefined

After:
  CGPoint p;
  p.x = 0;
  CGPoint p2 = p; // no-warning

This matches our behavior in C, where we don't see a field-by-field copy.

<rdar://problem/12305288>

llvm-svn: 173951

llvm-svn: 173950