llvm-svn: 100058

* Set the "DestA" and "DestB" according to how they're understood by the
  method. I.e., if one or both of them should point to the "fall through" block,
  then point to the fall through block.

* Improve the loop that removes superfluous edges to be more understandable.

llvm-svn: 100056

Debug info can now properly represent functions inside classes inside other functions.  Partial fix for Radar 7424645.

llvm-svn: 100048

PHIs.  The previous algorithm was unable to reliably detect when existing
PHIs in a cycle can be reused.  I'm still working on reducing a testcase.
Radar 7711900.

llvm-svn: 100047

generate wrong code pretty much anywhere AFAICT.
A case that hits the bug reproducibly is impossible,
but the situation was like this:
Addr = ...
Store -> Addr
Addr2 = GEP , 0, 0
Store -> Addr2
Handling the first store, the code changed replaced Addr
with a sunkaddr and deleted Addr, but not its table
entry.  Code in OptimizedBlock replaced Addr2 with a
bitcast; if that happened to reuse the memory of Addr,
the old table entry was erroneously found when handling
the second store.

llvm-svn: 100044

llvm-svn: 100042

llvm-svn: 100038

llvm-svn: 100037

POD-like anyway, so we don't even care about calling their d'tors (DIEBlock
being the exception).

~6% less mallocs and ~1% compile time improvement on clang -O0 -g oggenc.c

llvm-svn: 100035

llvm-svn: 100033

llvm-svn: 100031

llvm-svn: 100016

llvm-svn: 100011

llvm-svn: 100010

Patch from Gianluigi Tiesi!

llvm-svn: 100003

llvm-svn: 99999

llvm-svn: 99998

consistent with other similar ones.

llvm-svn: 99997

llvm-svn: 99995

llvm-svn: 99993

llvm-svn: 99992

llvm-svn: 99989

llvm-svn: 99986

llvm-svn: 99984

instructions.  In addition to being a convenience,
they are faster than the old apis, particularly when
not going from an MDKindID like people should be
doing.

llvm-svn: 99982

llvm-svn: 99975

llvm-svn: 99974

the storage of !dbg metadata kinds in the instruction themselves.
The on-the-side hash table works great for metadata that not-all
instructions get, or for metadata that only exists when optimizing.
But when compile-time is everything, it isn't great.

I'm not super thrilled with the fact that this plops a TrackingVH in
Instruction, because it grows it by 3 words.  I'm investigating 
alternatives, but this should be a step in the right direction in any
case.

llvm-svn: 99957

llvm-svn: 99954

Since it is just a pxor in disguise, we should probably expand it to a full
polymorphic triple.

llvm-svn: 99953

SSEDomainFix will collapse to the domain with the lower number when it has a
choice. The SSEPackedSingle domain often has smaller instructions, so prefer
that.

llvm-svn: 99952

llvm-svn: 99948

llvm-svn: 99938

Not running 'make check-all' before committing is a bad idea.

llvm-svn: 99933

llvm-svn: 99931

e.g., llvm.memcpy.i32(i8*, i8*, i32, i32) -> llvm.memcpy.p0i8.p0i8.i32(i8*, i8*, i32, i32, i1)
A update of langref will occur in a subsequent checkin.

llvm-svn: 99928

llvm-svn: 99927

only a single type of object to be allocated. Use it to make VNInfo destruction
typesafe.

llvm-svn: 99919

llvm-svn: 99918

llvm-svn: 99917