llvm-svn: 193958

llvm-svn: 193957

Instead of doing a RPO traversal of the whole function remember the blocks
containing gathers (typically <= 2) and scan them in dominator-first order.

The actual CSE is still quadratic, but I'm not confident that adding a
scoped hash table here is worth it as we're only looking at the generated
instructions and not arbitrary code.

llvm-svn: 193956

This reverts commit r193356, it caused PR17781.

A reduced test case covering this regression has been added to the test suite.

llvm-svn: 193955

llvm-svn: 193954

llvm-svn: 193953

OCaml's type unit is not compatible with C's type void.

llvm-svn: 193952

llvm-svn: 193951

llvm-svn: 193950

Llvm_target tests did not check for return values. This actually
caused them to miss a bug.

llvm-svn: 193949

llvm-svn: 193948

llvm-svn: 193947

llvm-svn: 193946

llvm-svn: 193945

These changes had accumulated internally at Apple for no good reason.

llvm-svn: 193944

This adds an SimplifyLibCalls case which converts the special __sinpi and
__cospi (float & double variants) into a __sincospi_stret where appropriate to
remove duplicated work.

Patch by Tim Northover

llvm-svn: 193943

rdar://12856873
Patch by Evan Cheng, with a fix for rdar://13209539 by Tilmann Scheller

llvm-svn: 193942

llvm-svn: 193941

Patch by Pete Cooper.

llvm-svn: 193940

llvm-svn: 193939

r102230 added an 'echo' making this a no-op.

Also fixes FAIL on native Windows with no shell/GnuWin32.

llvm-svn: 193938

When LLVM is embedded in a larger application, it's not OK for LLVM to intercept crashes.  LLVM already has 
the ability to disable this functionality.  This patch exposes it via the C API.

llvm-svn: 193937

llvm-svn: 193936

rdar://13973577

Patch by Fariborz Jahanian.

llvm-svn: 193935

llvm-svn: 193934

llvm-svn: 193933

llvm-svn: 193932

This is a fixed version of 193928 which keeps these uses in sync.

llvm-svn: 193931

Investigating a bot failure.

This reverts commit r193928.

llvm-svn: 193929

llvm-svn: 193928

llvm-svn: 193927

Doing this with a hash map doesn't change behavior and avoids calling
isIdenticalTo O(n^2) times. This should probably eventually move into a utility
class shared with EarlyCSE and the limited CSE in the SLPVectorizer.

llvm-svn: 193926

For for-nodes that are translated to a set of vector lanes, we already know the
overall number of iterations. Calculating the upper bound is consequently not
necessary. This change removes the code for upper bound calculation, which was
probably copy/pasted from the code generation for the normal for-loop.

This issue was found by Sylvestre's scan-build server.

llvm-svn: 193925

llvm-svn: 193924

Flexible array members inherently index off of the end of their parent
type.

We shouldn't allow this type to be used as a base, virtual or otherwise,
because indexing off the end may find us inside of another base or the
derived types members.

llvm-svn: 193923

llvm-svn: 193922

llvm-svn: 193921

Flexible array members only work out if they are the last field of a
record, however virtual bases would give us many situations where the
flexible array member would overlap with the virtual base fields.

It is unlikely in the extreme that this behavior was intended by the
user so raise a diagnostic instead of accepting.  This is will not
reject conforming code because flexible array members are an extension
in C++ mode.

llvm-svn: 193920

The determination of which diagnostics would be issued for certain
anonymous unions started to get a little ridiculous.  Clean this up by
inverting the condition-tree's logic from dialect -> issue to
issue -> diagnostic.

As part of this cleanup, move ext_c99_flexible_array_member from
DiagnosticParseKinds.td to DiagnosticSemaKinds.td because it's driven by
Sema, not Parse.

Also, the liberty was taken to edit ext_c99_flexible_array_member to
match other, similar, diagnostics.

llvm-svn: 193919

The declaration of a flexible array member was correctly diagnosed as an
extension in C89 mode but not in C++.

llvm-svn: 193918