llvm-svn: 20654

PHASE 1: write instruction selector
PHASE 2: ???
PHASE 3: profit!

llvm-svn: 20652

llvm-svn: 20651

Bill Wendling!!

llvm-svn: 20649

llvm-svn: 20623

llvm-svn: 20599

llvm-svn: 20598

using Function::arg_{iterator|begin|end}.  Likewise Module::g* -> Module::global_*.

This patch is contributed by Gabor Greif, thanks!

llvm-svn: 20597

llvm-svn: 20589

llvm-svn: 20579

llvm-svn: 20577

llvm-svn: 20563

llvm-svn: 20542

llvm-svn: 20535

llvm-svn: 20523

to cygwin)

llvm-svn: 20520

struct types.  This fixes Regression/CodeGen/CBackend/2005-03-08-RecursiveTypeCrash.ll,
a crash on Java output that Alkis reported.

llvm-svn: 20519

llvm-svn: 20515

llvm-svn: 20471

llvm-svn: 20455

llvm-svn: 20453

llvm-svn: 20442

fix FCMOVxx typo, set rem and div to hardcode target reg to be the same as the one the assembler uese, update ISel to put values in regs used by assembler for rem and div

llvm-svn: 20434

llvm-svn: 20425

llvm-svn: 20424

double test(double l1_X) {
  return (-l1_X);
}

instead of like this:

double test(double l1_X) {
  return (-0x0p+0 - l1_X);
}

llvm-svn: 20423

llvm-svn: 20422

  ltmp_0_7 = malloc(((unsigned )(&(((signed char (*)[784])/*NULL*/0)[1u]))));

Instead, just emit the literal constant, like this:

  ltmp_0_7 = malloc(784u);

This works around a bug in ICC 8.1 compiling the CBE generated code.  :-(

llvm-svn: 20415

llvm-svn: 20410

llvm-svn: 20408

llvm-svn: 20407

Loop Strength Reduction.

llvm-svn: 20399

llvm-svn: 20380

Justin Wick!

llvm-svn: 20378

llvm-svn: 20372

llvm-svn: 20343

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

llvm-svn: 20332

llvm-svn: 20306

Only print out machine instructions before modulo scheduling if we are actually doing modulo scheduling! :)

llvm-svn: 20292

llvm-svn: 20290