llvm-svn: 106439

"llvm.dbg.lv.~A" is not a valid name.

llvm-svn: 106438

Move part of SSE 1 & 2 compare, shuffle and unpack instructions closely. Preparing them for refactoring and to the addition of their AVX forms

llvm-svn: 106437

Add AVX regular (non-aliased ones) and,or,xor,andn packed instructions. They are already tested in the MC framework, no test needed

llvm-svn: 106436

handling, plus correct R9 handling in that mode.

llvm-svn: 106434

Based on a patch by Patrick Marlier!

llvm-svn: 106433

llvm-svn: 106430

llvm-svn: 106428

llvm-svn: 106427

particular sequence causes its definition to not be generated in the object file.
(fixes radar 8071804).

llvm-svn: 106424

llvm-svn: 106423

Split the code for materializing a value out of
SelectionDAGBuilder::getValue into a helper function, so that it can
be used in other ways. Add a new getNonRegisterValue function which
uses it, for use in code which doesn't want a CopyFromReg even
when FuncMap.ValueMap already has an entry for it.

llvm-svn: 106422

llvm-svn: 106421

This allows the fast regiser allocator to remove redundant 
register moves.
Update a set of tests that depend on the register allocator
to be linear scan. 

llvm-svn: 106420

llvm-svn: 106419

an unitialized variable.

valgrind_errors -= 1;

llvm-svn: 106418

dependent on the last byte of the buffer, which could be unitialized.

llvm-svn: 106417

This also fixes a bug where we were trying to copy m_string into itself
via a format string. The pointer was invalidated by m_string.resize and
lldb (sometimes) crashed inside vsnprintf.

llvm-svn: 106416

llvm-svn: 106415

the LocalValueMap, to make it more flexible when fast-isel isn't
proceding straight top-down.

llvm-svn: 106414

I was trying to do some refactoring on the copyRegToReg, but this is realyl a work in progress and not generally useful yet.

llvm-svn: 106413

llvm-svn: 106412

used to choke llc with the attached test.
 

llvm-svn: 106411

const_cast.

llvm-svn: 106410

llvm-svn: 106409

llvm-svn: 106408

llvm-svn: 106407

Add support for returning multiple vectors via sret, which is how the ARM target expects the intrinsics to work.

llvm-svn: 106406

llvm-svn: 106405

llvm-svn: 106404

llvm-svn: 106403

Adds analyzer support for idempotent and tautological binary operations such as "a*0" and "a+0". This is not very powerful, but does make the analyzer look a little smarter than it actually is.

llvm-svn: 106402

Casting to void* or any other pointer-to-sizeless type (e.g. function pointers) causes a divide-by-zero error. Simple fix: check if the pointee type size is 0 and bail out early if it is.

llvm-svn: 106401

llvm-svn: 106399

in refactoring.

llvm-svn: 106398

llvm-svn: 106397

use sharing map. The reconcileNewOffset logic already forces a
separate use if the kinds differ, so incorporating the kind in the
key means we can track more sharing opportunities.

More sharing means fewer total uses to track, which means smaller
problem sizes, which means the conservative throttles don't kick
in as often.

llvm-svn: 106396

llvm-svn: 106395

llvm-svn: 106394

llvm-svn: 106393