CrashRecovery: Clear the current context on the first crash, to avoid re-entering it if the cleanup code crashes.

llvm-svn: 111309

CrashRecovery: Add CrashRecoveryContext::GetCurrent(), so clients can find the active context from anywhere.

llvm-svn: 111308

llvm-svn: 111307

PR 7882.  Follows suggestion by Amaury Pouly, thanks.

llvm-svn: 111306

llvm-svn: 111291

  - Do not clobber al during variadic calls, this is AMD64 ABI-only feature
  - Emit wincall64, where necessary
Patch by Cameron Esfahani!

llvm-svn: 111289

Patch by Cameron Esfahani!

llvm-svn: 111288

PHI elimination shouldn't require machineloopinfo since it's used at -O0. Move the requirement to LiveIntervalAnalysis instead. Note this does not change the number of times machineloopinfo is computed.

llvm-svn: 111285

llvm-svn: 111281

llvm-svn: 111277

llvm-svn: 111274

llvm-svn: 111271

llvm-svn: 111266

Hello world builds & runs now on i386/ELF with -integrated-as.

llvm-svn: 111264

Modernize predicates a bit.

The Predicate_* methods are not used by TableGen any longer. They are only
emitted for the sake of legacy code.

llvm-svn: 111263

whether to allocate a virtual frame base register to resolve the frame
index reference in it. Implement a simple version for ARM to aid debugging.

In LocalStackSlotAllocation, scan the function for frame index references
to local frame indices and ask the target whether to allocate virtual
frame base registers for any it encounters. Purely infrastructural for
debug output. Next step is to actually allocate base registers, then add
intelligent re-use of them.

rdar://8277890

llvm-svn: 111262

llvm-svn: 111260

llvm-svn: 111259

Move the decision logic whether it's a good idea to split a critical edge to clients. Also fixed an erroneous check. An edge is only a back edge when the from and to blocks are in the same loop.

llvm-svn: 111256

loop, making the resulting loop significantly less ugly.  Also, zap
its trivial PHI nodes, since it's easy.

llvm-svn: 111255

llvm-svn: 111254

llvm-svn: 111252

printing "lsl #0".  This fixes the remaining parts of pr7792.  Make
corresponding changes for encoding/decoding these instructions.

llvm-svn: 111251

llvm-svn: 111250

what it does manually.

llvm-svn: 111248

llvm-svn: 111247

llvm-svn: 111241

against it. This affects Windows.

llvm-svn: 111240

cygwin when built with ENABLE_SHARED=1.  Patch by NAKAMURA Takumi!

llvm-svn: 111231

llvm-svn: 111226

Add an option to disable codegen prepare critical edge splitting. In theory, PHI elimination is already doing all (most?) of the splitting needed. But machine-licm and machine-sink seem to miss some important optimizations when splitting is disabled.

llvm-svn: 111224

llvm-svn: 111223

we're adding predicates and optional defs to the MachineInstrs.

llvm-svn: 111222

PHI elimination should not break back edge. It can cause some significant code placement issues. rdar://8263994

good:
LBB0_2:
  mov     r2, r0
  . . .
  mov     r1, r2
  bne     LBB0_2

bad:
LBB0_2:
  mov     r2, r0
  . . .
@ BB#3:
  mov     r1, r2
  b       LBB0_2

llvm-svn: 111221

llvm-svn: 111219

llvm-svn: 111214

llvm-svn: 111213

llvm-svn: 111210

llvm-svn: 111208

llvm-svn: 111206