Change errs() to dbgs().

llvm-svn: 92545

Change errs() to dbgs().

llvm-svn: 92544

Change errs() to dbgs().

llvm-svn: 92542

Change errs() to dbgs().

llvm-svn: 92539

Change errs() to dbgs().

llvm-svn: 92535

llvm-svn: 92533

Change errs() to dbgs().

llvm-svn: 92532

Change errs() to dbgs().

llvm-svn: 92529

Change errs() to dbgs().

llvm-svn: 92528

Change errs() to dbgs().

llvm-svn: 92524

Change errs() to dbgs().

llvm-svn: 92520

Change errs() to dbgs().

llvm-svn: 92518

Change errs() to dbgs().

llvm-svn: 92516

Change errs() to dbgs().

llvm-svn: 92515

(or (x << c) | (y >> (64 - c))) ==> (shld64 x, y, c)

The isel patterns may not catch all the cases if general dag combine has reduced width of source operands.

llvm-svn: 92513

in an inline asm.

llvm-svn: 92512

llvm-svn: 92511

lets the test-elimination work in more conditional-move cases.

llvm-svn: 92508

rules as normal add, and, or, etc.

llvm-svn: 92507

This lets isel fold loads into them in more cases.

llvm-svn: 92506

llvm-svn: 92505

Change errs() to dbgs().

llvm-svn: 92502

Change errs() to dbgs().

llvm-svn: 92498

Change errs() to dbgs().

llvm-svn: 92496

when doing this transform if the GEP is not inbounds.  No testcase because
it is very difficult to trigger this: instcombine already canonicalizes
GEP indices to pointer size, so it relies specific permutations of the
instcombine worklist.

Thanks to Duncan for pointing this possible problem out.

llvm-svn: 92495

llvm-svn: 92494

llvm-svn: 92492

Change errs() to dbgs().

llvm-svn: 92490

llvm-svn: 92472

llvm-svn: 92471

llvm-svn: 92470

llvm-svn: 92469

file.  InstructionCombining.cpp is now down to a svelte 9300 lines :)

llvm-svn: 92468

a new InstCombineCompares.cpp file.

llvm-svn: 92467

llvm-svn: 92466

move the 'SimplifyDemandedFoo' methods out to their own file, cutting 1K lines out of instcombine.cpp

llvm-svn: 92465

among the instcombine library.

llvm-svn: 92463

llvm-svn: 92462

llvm-svn: 92461

llvm-svn: 92460