o Parallel addition and subtraction, signed/unsigned
o Miscellaneous operations: QADD, QDADD, QSUB, QDSUB
o Unsigned sum of absolute differences [and accumulate]: USAD8, USADA8
o Signed/Unsigned saturate: SSAT, SSAT16, USAT, USAT16
o Signed multiply accumulate long (halfwords): SMLAL<x><y>
o Signed multiply accumulate/subtract [long] (dual): SMLAD[x], SMLALD[X], SMLSD[X], SMLSLD[X]
o Signed dual multiply add/subtract [long]: SMUAD[X], SMUSD[X]

llvm-svn: 97276

and SRS.

llvm-svn: 97164

llvm-svn: 97163

llvm-svn: 97159

llvm-svn: 96393

llvm-svn: 96388

to have the predicate on the pattern itself instead. Support for the new
ISel. Remove definitions of CarryDefIsUnused and CarryDefIsUsed since they are
no longer used anywhere.

llvm-svn: 96384

llvm-svn: 95603

llvm-svn: 95112

cannot be directly interchanged for comparisons against negated values.
Disable the CMN instructions for the time being.

llvm-svn: 94119

Fix r93758. Use isel patterns instead of c++ selection code to select rbit and make sure we pick different instructions for ARM vs. Thumb2.

llvm-svn: 93829

"On ARMv6T2 this turns cttz into rbit, clz instead of the 4 instruction
 sequence it is now."

llvm-svn: 93758

T2I_bin_ii12rs definition.

llvm-svn: 93006

llvm-svn: 92796

instructions.  Thumb does not have the restriction that t2 = t+1.

llvm-svn: 92785

llvm-svn: 91555

bytes of Inst to 0x0000 for the benefit of the Thumb decoder.

llvm-svn: 91496

llvm-svn: 91434

llvm-svn: 91329

llvm-svn: 91313

llvm-svn: 91310

than doing the same via constpool:
1. Load from constpool costs 3 cycles on A9, movt/movw pair - just 2.
2. Load from constpool might stall up to 300 cycles due to cache miss.
3. Movt/movw does not use load/store unit.
4. Less constpool entries => better compiler performance.

This is only enabled on ELF systems, since darwin does not have needed
relocations (yet).

llvm-svn: 89720

llvm-svn: 89718

fold immediate of a + Const into the user as a subtract if it can fit as a negated two-part immediate.

llvm-svn: 89694

llvm-svn: 89478

  load of a GV from constantpool and then add pc. It allows the code sequence to
  be rematerializable so it would be hoisted by machine licm.
- Add a late pass to break these pseudo instructions into a number of real
  instructions. Also move the code in Thumb2 IT pass that breaks up t2MOVi32imm
  to this pass. This is done before post regalloc scheduling to allow the
  scheduler to proper schedule these instructions. It also allow them to be
  if-converted and shrunk by later passes.

llvm-svn: 86304

Fix t2Int_eh_sjlj_setjmp. Immediate form of orr is a 32-bit instruction. So it should be 22 bytes instead of 20 bytes long.

llvm-svn: 85965

llvm-svn: 85824

llvm-svn: 85809

It's safe to remat t2LDRpci; Add PseudoSourceValue to load / store's to enable more machine licm. More changes coming.

llvm-svn: 85643

clang/test/CodeGen/indirect-goto.c runs! (unoptimized)

llvm-svn: 85577

bunch of associated comments, because it doesn't have anything to do
with DAGs or scheduling. This is another step in decoupling MachineInstr
emitting from scheduling.

llvm-svn: 85517

Change Thumb1 and Thumb2 instructions to separate opcode from operands with a tab instead of a space.

llvm-svn: 85184

immediate operands when they will fit into the using instruction.

llvm-svn: 84778

llvm-svn: 84751

llvm-svn: 84009

ld / st pairs, etc.

llvm-svn: 83197

Change ld/st multiples to explicitly model the writeback to base register. This fixes most of the -ldstopti-before-sched2 regressions.

llvm-svn: 83191

Patch by Sylvere Teissier.

llvm-svn: 83135

instruction. This makes it re-materializable.

Thumb2 will split it back out into two instructions so IT pass will generate the
right mask. Also, this expose opportunies to optimize the movw to a 16-bit move.

llvm-svn: 82982