* Loop invariant code does not dominate the loop header, but rather
    the end of the loop preheader.

  * The base for a reduced GEP isn't a constant unless all of its
    operands (preceding the induction variable) are constant.

  * Allow induction variable elimination for the simple case after all.

Also made changes recommended by Chris for properly deleting
instructions.

llvm-svn: 20383

llvm-svn: 20382

llvm-svn: 20381

llvm-svn: 20380

llvm-svn: 20379

Justin Wick!

llvm-svn: 20378

llvm-svn: 20377

llvm-svn: 20376

llvm-svn: 20375

llvm-svn: 20374

llvm-svn: 20373

llvm-svn: 20372

llvm-svn: 20371

ensure that -L paths don't contain both bytecode and native libraries.
This patch contributed by Adam Treat.

llvm-svn: 20370

llvm-svn: 20369

llvm-svn: 20368

for reasons explained in the comments, I also deactivated this code as it needs
more thought.

llvm-svn: 20367

llvm-svn: 20360

llvm-svn: 20358

llvm-svn: 20357

llvm-svn: 20356

llvm-svn: 20355

llvm-svn: 20354

llvm-svn: 20353

llvm-svn: 20352

llvm-svn: 20351

llvm-svn: 20350

llvm-svn: 20349

llvm-svn: 20345

llvm-svn: 20343

llvm-svn: 20342

llvm-svn: 20341

llvm-svn: 20340

llvm-svn: 20339

bytecode file members or not.
Patch Contributed By Adam Treat

llvm-svn: 20338

If we fold three constants together (c1+c2+c3), make sure to keep
LHSC updated, instead of reusing (in this case), the 1 instead of the
partial sum.

llvm-svn: 20337

llvm-svn: 20336

a ternary commutative expr.  Remove FIXME that does not need to be fixed
(can't happen).

llvm-svn: 20335

llvm-svn: 20334

make BB labels be exported for debuging, add fp negation optimization, further pecimise the FP instructions

llvm-svn: 20332