llvm-svn: 120513

statements.

llvm-svn: 120512

GetCommandObject returns no matches in the match array when there is only one match.  That's odd, but I don't want to change that right now, just cope with it where I'm doing the command matching.

llvm-svn: 120511

llvm-svn: 120507

llvm-svn: 120506

    rdar://problem/8689814 test failure: test/macosx/universal (the i386 slice does not break?)

llvm-svn: 120505

llvm-svn: 120504

Possibly due to http://llvm.org/viewvc/llvm-project?rev=120484&view=rev.

llvm-svn: 120503

legalization time. Since at legalization time there is no mapping from
SDNode back to the corresponding LLVM instruction and the return
SDNode is target specific, this requires a target hook to check for
eligibility. Only x86 and ARM support this form of sibcall optimization
right now.
rdar://8707777

llvm-svn: 120501

llvm-svn: 120500

llvm-svn: 120499

llvm-svn: 120498

llvm-svn: 120497

llvm-svn: 120496

llvm-svn: 120495

llvm-svn: 120494

llvm-svn: 120493

llvm-svn: 120492

refer to opaque types, they must be held via PATypeHolders.  I'm
not sure why this hasn't blown up before.

llvm-svn: 120491

in favor of the widespread llvm style. Capitalize variables and add
newlines for visual parsing. Rename variables for readability.
And other cleanup.

llvm-svn: 120490

llvm-svn: 120489

llvm-svn: 120488

llvm-svn: 120487

literal declaration caused crash in CodeGen.
This patch fixes it. pr8707

llvm-svn: 120486

may-aliasing stores that partially overlap with different base
pointers.  This implements PR6043 and the non-variable part of
PR8657

llvm-svn: 120485

llvm-svn: 120484

Follow through references to catch returned stack addresses, local blocks, label addresses or references to temporaries, e.g:

const int& g2() {
  int s1;
  int &s2 = s1; // expected-note {{binding reference variable 's2' here}}
  return s2; // expected-warning {{reference to stack memory associated with local variable 's1' returned}}
}

llvm-svn: 120483

  t_addrmode_s4, but with a different scaling factor.

* Encode the Thumb1 load and store instructions. This involved a bit of
  refactoring (hi, Chris! :-). Some of the patterns became dead afterwards and
  were removed.

llvm-svn: 120482

Thumb2 encoding to share code with the ARM encoding, which gets use fixup support for free.
It also allows us to fold away at least one codegen-only pattern.

llvm-svn: 120481

llvm-svn: 120480

llvm-svn: 120479

llvm-svn: 120476

llvm-svn: 120474

copied by the copy constructor for ClangExpressionVariable.
This meant that a NULL m_register_info wouldn't be
copied, and instead the field was uninitialized, potentially
confusing the materializer.

llvm-svn: 120472

llvm-svn: 120471

llvm-svn: 120470

the code.  We now get accurate sizes on Loads, though it surely doesn't
matter in practice.

llvm-svn: 120469

1. if the underlying pointer passed in can be resolved
   to any argument or alloca, then we don't need to scan.
   Previously we would only avoid the scan if the alloca
   or byval was actually considered dead.
2. The dead store processing code is itself completely
   dead and didn't handle volatile stores right anyway,
   so delete it.  This allows simplifying the interface
   to RemoveAccessedObjects.

llvm-svn: 120467

while I'm in there.

llvm-svn: 120466

made sense to me.  We now have a set of dead stack objects, and
they become live when loaded.  Fix a theoretical problem where
we'd pass in the wrong pointer to the alias query.

llvm-svn: 120465