following it. However, the EmitGlobalConstant method wasn't emitting a body for
the constant. The assembler doesn't like that. Before, we were generating this:

  .zerofill __DATA, __common, __cmd, 1, 3

This fix puts us back to that semantic.

llvm-svn: 95336

llvm-svn: 95335

category implementation, which showed up during (attempted) typo
correction. Fixes <rdar://problem/7605289>.

llvm-svn: 95334

short-circuited conditions to AND/OR expressions, and those expressions
are often converted back to a short-circuited form in code gen.  The
original source order may have been optimized to take advantage of the
expected values, and if we reassociate them, we change the order and
subvert that optimization.  Radar 7497329.

llvm-svn: 95333

of a C++ record.  Exposed a lot of problems where various routines were
silently doing The Wrong Thing (or The Acceptable Thing in The Wrong Order)
when presented with a non-definition.  Also cuts down on memory usage.

llvm-svn: 95330

llvm-svn: 95324

Instruction selection for X86 now can choose an instruction
sequence that will fit any address of any symbol, no matter
the pointer width. X86-64 uses a mov+call-via-reg sequence
for this.

llvm-svn: 95323

llvm-svn: 95322

This makes the inliner about as agressive as it was before my changes to the
inliner cost calculations. These levels give the same performance and slightly
smaller code than before.

llvm-svn: 95320

ProcessImplicitDefs would only mark one operand per instruction with <undef>.
This fixed PR6086.

llvm-svn: 95319

llvm-svn: 95318

ton of potential crashes of the same kind. The fundamental problem is
that type creation was following a dangerous pattern when using its
FoldingSets:
  1) Use FindNodeOrInsertPos to see if the type is available
  2) If not, and we aren't looking at a canonical type, build the
  canonical type
  3) Build and insert the new node into the FoldingSet

The problem here is that building the canonical type can, in very rare
circumstances, force the hash table inside the FoldingSet to
reallocate. That invalidates the insertion position we computed in
step 1, and in step 3 we end up inserting the new node into the wrong
place. BOOM!

I've audited all of ASTContext, fixing this problem everywhere I found
it. The vast majority of wrong code was C++-specific (and *ahem*
written by me), so I also audited other major folding sets in the C++
code (e.g., template specializations), but found no other instances of
this problem.

llvm-svn: 95315

With this fix, and the other fixes committed today a make check-all with a clang-built LLVM now gives:

Expected Passes    : 6933
Expected Failures  : 46
Unsupported Tests  : 40
Unexpected Failures: 27

which means that we pass 99.96% of all tests :) The resulting 27 tests are all LLVMC tests and seem to be because of differences in the clang and gcc drivers.

llvm-svn: 95313

llvm-svn: 95312

Fix a bug where we would not mark temporaries as conditional when emitting a conditional operator as an lvalue.

llvm-svn: 95311

llvm-svn: 95310

template parameter, perform array/function decay (if needed), take the
address of the argument (if needed), perform qualification conversions
(if needed), and remove any top-level cv-qualifiers from the resulting
expression. Fixes PR6226.

llvm-svn: 95309

Rename StartConditionalBranch/FinishConditionalBranch to BeginConditionalBranch/EndConditionalBranch.

llvm-svn: 95308

Fix another pointer-to-member function miscompile, this time when trying to call a virtual member function.

llvm-svn: 95307

llvm-svn: 95306

llvm-svn: 95305

llvm-svn: 95303

This allows to show the explicit files that need to be built/linked to get an
LLVM component.

llvm-svn: 95300

llvm-svn: 95299

direct bit manipulation.  This is is less error prone, and fixes a bug
in the handling of the LeadingZeroes flag as pointed out by Cristian
Draghici.

llvm-svn: 95298

llvm-svn: 95297

after r95292 

llvm-svn: 95296

llvm-svn: 95295

failure.

llvm-svn: 95294

64-bit sparc codegen.  Patch by Nathan Keynes!

llvm-svn: 95293

"Attached patch removes the extra NUL bytes from the output and changes
test/Archive/MacOSX.toc from a binary to a text file (removes
svn:mime-type=application/octet-stream and adds svn:eol-style=native).  I can't
figure out how to get SVN to include the new contents of the file in the patch
so I'm attaching it separately."

Patch by James Abbatiello!

llvm-svn: 95292

llvm-svn: 95291

llvm-svn: 95290

llvm-svn: 95287

llvm-svn: 95286

llvm-svn: 95284

Fix bugs where we would compute out of bounds as in bounds, and where
we couldn't know that the linker could override the size of an array.

Add a few new testcases, change existing testcase to use a private
global array instead of extern.

llvm-svn: 95283

It's too risky to eliminate sext / zext of call results for tail call optimization even if the caller / callee attributes completely match. The callee may have been bitcast'ed (or otherwise lied about what it's doing).

llvm-svn: 95282

than int. This will make it more convenient for LSR, which does
a lot of things with int64_t offsets.

llvm-svn: 95281

Indirect tail call has to go through a call preserved register since it's after callee register pops. X86 isel lowering is using EAX / R11 and it was somehow adding that to function live out. That prevented the real function return register from being added to the function live out list and bad things happen.
This fixes 483.xalancbmk (with tail call opt).

llvm-svn: 95280