llvm-6502/test/CodeGen/Thumb
Evan Cheng 9fe2009956 Sorry, several patches in one.
TargetInstrInfo:
Change produceSameValue() to take MachineRegisterInfo as an optional argument.
When in SSA form, targets can use it to make more aggressive equality analysis.

Machine LICM:
1. Eliminate isLoadFromConstantMemory, use MI.isInvariantLoad instead.
2. Fix a bug which prevent CSE of instructions which are not re-materializable.
3. Use improved form of produceSameValue.

ARM:
1. Teach ARM produceSameValue to look pass some PIC labels.
2. Look for operands from different loads of different constant pool entries
   which have same values.
3. Re-implement PIC GA materialization using movw + movt. Combine the pair with
   a "add pc" or "ldr [pc]" to form pseudo instructions. This makes it possible
   to re-materialize the instruction, allow machine LICM to hoist the set of
   instructions out of the loop and make it possible to CSE them. It's a bit
   hacky, but it significantly improve code quality.
4. Some minor bug fixes as well.

With the fixes, using movw + movt to materialize GAs significantly outperform the
load from constantpool method. 186.crafty and 255.vortex improved > 20%, 254.gap
and 176.gcc ~10%.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@123905 91177308-0d34-0410-b5e6-96231b3b80d8
2011-01-20 08:34:58 +00:00
..
2007-01-31-RegInfoAssert.ll
2007-02-02-JoinIntervalsCrash.ll
2007-03-06-AddR7.ll
2007-05-05-InvalidPushPop.ll Correct bogus module triple specifications. 2010-08-30 10:48:29 +00:00
2009-06-18-ThumbCommuteMul.ll
2009-07-19-SPDecBug.ll
2009-07-20-TwoAddrBug.ll
2009-07-27-PEIAssert.ll
2009-08-12-ConstIslandAssert.ll
2009-08-12-RegInfoAssert.ll
2009-08-20-ISelBug.ll
2009-12-17-pre-regalloc-taildup.ll
2010-01-15-local-alloc-spill-physical.ll
2010-04-07-DbgValueOtherTargets.ll If dbg_declare() or dbg_value() is not lowered by isel then emit DEBUG message instead of creating DBG_VALUE for undefined value in reg0. 2010-12-06 22:39:26 +00:00
2010-06-18-SibCallCrash.ll
2010-07-01-FuncAlign.ll ARM function alignments were off by a power of two. svn 83242 changed 2010-07-01 22:26:26 +00:00
2010-07-15-debugOrdering.ll Try again to disable critical edge splitting in CodeGenPrepare. 2010-09-30 20:51:52 +00:00
2011-EpilogueBug.ll Spill R4 if it's going to be used to restore SP from FP. 2011-01-16 05:14:33 +00:00
asmprinter-bug.ll
barrier.ll Do not use MEMBARRIER_MCR for any Thumb code. 2010-11-09 22:50:44 +00:00
dg.exp
dyn-stackalloc.ll Fix epilogue codegen to avoid leaving the stack pointer in an invalid 2010-11-22 18:12:04 +00:00
fpconv.ll
fpow.ll
frame_thumb.ll
iabs.ll
inlineasm-imm-thumb.ll
ispositive.ll
large-stack.ll Thumb1 had two patterns for the same load-from-constant-pool instruction. 2010-12-15 23:52:36 +00:00
ldr_ext.ll
ldr_frame.ll
long_shift.ll
long-setcc.ll
long.ll These tests are looking for library function names that 2010-11-17 21:57:32 +00:00
mul.ll
pop.ll
push.ll Propagate the AlignStack bit in InlineAsm's to the 2010-07-02 20:16:09 +00:00
select.ll These tests are looking for library function names that 2010-11-17 21:57:32 +00:00
stack-frame.ll
thumb-imm.ll
trap.ll
tst_teq.ll
unord.ll
vargs.ll Update test for 112609 2010-08-31 17:58:47 +00:00