C is a constant which can be sign-extended from 8 bits without value loss,
and op is one of: add, sub, imul, and, or, xor.

This allows the JIT to emit the one byte version of the constant instead of
the two or 4 byte version.  Because these instructions are very common, this
can save a LOT of code space.  For example, I sampled two benchmarks, 176.gcc
and 254.gap.

BM        Old     New    Reduction
176.gcc 2673621 2548962  4.89%
254.gap  498261  475104  4.87%

Note that while the percentage is not spectacular, this did eliminate
124.6 _KILOBYTES_ of codespace from gcc.  Not bad.

Note that this doesn't effect the llc version at all, because the assembler
already does this optimization.

llvm-svn: 9284

llvm-svn: 9283

llvm-svn: 9282

llvm-svn: 9281

llvm-svn: 9280

* Added description of `mem2reg'

llvm-svn: 9279

* Implement R1 = R2 * C where R1 and R2 are 32 or 16 bits. This avoids an
  extra copy into a register, reducing register pressure.

llvm-svn: 9278

llvm-svn: 9274

debugging with the JIT

llvm-svn: 9273

llvm-svn: 9272

llvm-svn: 9270

Change the Opcode enum for PHI nodes from "Instruction::PHINode" to "Instruction::PHI" to be more consistent with the other instructions.

llvm-svn: 9269

Change the Opcode enum for PHI nodes from "Instruction::PHINode" to "Instruction::PHI" to be more consistent with the other instructions.

This allows removal of a special case from the instvisitor

llvm-svn: 9268

  getelementptr code path for use by other code paths (like malloc and alloca).
* Optimize comparisons with zero
* Generate neg, not, inc, and dec instructions, when possible.

This gives some code size wins, which might translate into performance.  We'll
see tommorow in the nightly tester.

llvm-svn: 9267

llvm-svn: 9266

llvm-svn: 9265

Mention that they should really read the gcc manual.

llvm-svn: 9264

llvm-svn: 9263

llvm-svn: 9262

llvm-svn: 9261

llvm-svn: 9260

llvm-svn: 9259

llvm-svn: 9258

Refer to the various "modules" of bugpoint as "debuggers", not
 "debugging modes" or "modules". I think this is more evocative of
 how they typically act, and it's shorter.
Add a note about how a linker bug can crash bugpoint early.

llvm-svn: 9257

llvm-svn: 9256

Add Design Philosophy as a separate subsection.  Clarify its last sentence.
Give Automatic Mode Selection a uniquely-named anchor.
Always call the program the "test program", instead of the "initial program",
 the "LLVM program", the "test case", the "resultant module", etc.
Try to explain the assumptions a little more, instead of just describing the
 process.

llvm-svn: 9255

Many minor edits.
Rewrite some of the options section for grammatical parallelism,
 clarity, and brevity.

llvm-svn: 9254

Add printouts about failed building and testing.

llvm-svn: 9253

This fixes PR#44.

llvm-svn: 9252

llvm-svn: 9251

llvm-svn: 9250

This also fixes miscompilation of 176.gcc.

llvm-svn: 9249

llvm-svn: 9248

llvm-svn: 9247

llvm-svn: 9246

all, making it pointless for use with the code generator debugger.  With it,
it works like a charm.

llvm-svn: 9245

llvm-svn: 9244

llvm-svn: 9243

llvm-svn: 9242

llvm-svn: 9241