getSubLoops/getTopLevelLoops methods, replacing them with iterator-based
accessors.

llvm-svn: 10714

llvm-svn: 10536

loop before hoisting any.

llvm-svn: 10534

AliasSetTracker as well.

llvm-svn: 10507

llvm-svn: 10458

Fix LICM/2003-12-11-SinkingToPHI.ll, and quite possibly all of the other known problems in the universe.

llvm-svn: 10409

Thanks for pointing this out John  :)

llvm-svn: 10387

llvm-svn: 10386

llvm-svn: 10371

llvm-svn: 10370

Make LICM itself a bit more efficient, and make the generated code more efficient too: don't insert a store in every exit block, because a particular block may be exited to more than once by a loop

llvm-svn: 10369

better job, but this is the majority of the work.  This implements
LICM/sink*.ll

llvm-svn: 10358

should make subsequent changes simpler.  This also allows us to hoist vaarg
and vanext instructions

llvm-svn: 10342

Code cleanups
Make LICM::SafeToHoist marginally more efficient

llvm-svn: 10341

llvm-svn: 9903

llvm-svn: 9768

Header files will be on the way.

llvm-svn: 9298

llvm-svn: 9073

llvm-svn: 9061

llvm-svn: 9027

llvm-svn: 8883

llvm-svn: 8469

llvm-svn: 8467

llvm-svn: 8431

llvm-svn: 8399

llvm-svn: 7614

llvm-svn: 7612

llvm-svn: 7492

llvm-svn: 5872

allows LICM to use access sizes to help alias analysis be more precise.

llvm-svn: 5693

llvm-svn: 5685

llvm-svn: 5675

llvm-svn: 5658

Convert to work with new AliasAnalysis interface by conservatively assuming all pointers are arbitrarily large accesses

llvm-svn: 5636

llvm-svn: 5613

    setPreservesCFG to be less confusing.

llvm-svn: 4255

    * Renamed StatisticReporter.h/cpp to Statistic.h/cpp
    * Broke constructor to take two const char * arguments instead of one, so
      that indendation can be taken care of automatically.
    * Sort the list by pass name when printing
    * Make sure to print all statistics as a group, instead of randomly when
      the statistics dtors are called.
    * Updated ProgrammersManual with new semantics.

llvm-svn: 4002

llvm-svn: 3980

rather than in random order.  This causes LICM to be DRAMATICALLY more
effective. For example, on bzip2.c, it is able to hoist 302 loads and
2380 total instructions, as opposed to 44/338 before.  This  obviously
makes other transformations much more powerful as well!

llvm-svn: 3978

 - Do not clone instructions then insert clone outside of loop.  Just move them.

llvm-svn: 3951