llvm-svn: 37397

llvm-svn: 37394

llvm-svn: 37393

See test/CodeGen/X86/test-pic-jtbl.ll for a case where it works well;
shaves another 10K off our favorite benchmark.  I was hesitant about
this because of compile speed, but seems to do OK on a bootstrap.

llvm-svn: 37392

Avoid overwriting the APInt instance with 0 bytes which causes the bitwidth
to be set to 0 (illegal) producing a subsequent assert.

llvm-svn: 37391

llvm-svn: 37390

llvm-svn: 37389

llvm-svn: 37388

llvm-svn: 37387

llvm-svn: 37386

llvm-svn: 37385

llvm-svn: 37384

as a null i8 pointer not as a 0 i32.

llvm-svn: 37383

Ifcvt triangle: don't ifcvt 'true' BB if it has other predecessors; don't merge 'false' BB if it has other predecessors.

llvm-svn: 37382

llvm-svn: 37381

Remove a bogus check. Even terminators in a ifcvt need to be predicated. Unconditional branches can usually be converted to conditional ones.

llvm-svn: 37380

llvm-svn: 37379

llvm-svn: 37376

Do not remove empty landing pads (EH table needs to be updated)

llvm-svn: 37375

llvm-svn: 37374

smaller than the preferred alignment, but so that the target can actually
specify a minimum alignment if needed.  This fixes some objc protocol
failures Devang tracked down.

llvm-svn: 37373

llvm-svn: 37372

When a function has FP, the register scavenging spill slot offset already
was calculated.

llvm-svn: 37371

llvm-svn: 37368

llvm-svn: 37367

llvm-svn: 37366

llvm-svn: 37365

llvm-svn: 37362

llvm-svn: 37361

llvm-svn: 37360

http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20070528/050047.html

llvm-svn: 37359

half of PR1421, by not decimating structs with holes that are the source and
destination of a memcpy.

llvm-svn: 37358

llvm-svn: 37357

llvm-svn: 37356

llvm-svn: 37355

matter until my last change).  Reenable tail merging by default.

llvm-svn: 37354

llvm-svn: 37353

llvm-svn: 37352

For VFP2 fldm, fstm instructions, the condition code is printed after the address mode and size specifier. e.g. fstmiaseq, not fstmeqias.

llvm-svn: 37351

llvm-svn: 37350