- Walking over pred_begin/pred_end is an expensive operation.
- PHINodes contain a value for each predecessor anyway.
- While it may look like we used to save a few iterations with the set,
  be aware that getIncomingValueForBlock does a linear search on
  the values of the phi node.
- Another -5% on ARMDisassembler.cpp (Release build). This was the last
  entry in the profile that was obviously wasting time.

llvm-svn: 145937

llvm-svn: 145934

llvm-svn: 145929

Fix a bunch of SSE/AVX patterns to use v2i64/v4i64 loads since all other integer vector loads are promoted to those.

llvm-svn: 145927

llvm-svn: 145926

FIXME: Restore more other arch-dependent MachO tests. (eg. r126401 and r133856)
llvm-svn: 145925

Clean up some of the shuffle decoding code for UNPCK instructions. Add instruction commenting for AVX/AVX2 forms for integer UNPCKs.

llvm-svn: 145924

Same as r145922, just for ARM mode.

llvm-svn: 145923

Fix the alias to encode 'mul r5, r6' as if it were 'mul r5, r6, r5' so we
match gas.

rdar://10532439

llvm-svn: 145922

Merge isSHUFPMask and isCommutedSHUFPMask into single function that can do both. Do the same for the 256-bit version. Use loops to reduce size of isVSHUFPYMask. Fix test cases that were incorrectly passing due to isCommutedSHUFPMask not checking for the vector being 128-bit. This caused some 256-bit shuffles to be incorrectly commuted.

llvm-svn: 145921

Using encoding T1 for offset of #0 and encoding T2 for #-0.

rdar://10532413

llvm-svn: 145919

MC/MachO assumes x86.

llvm-svn: 145916

recent discussions. Poison can't make every value that depends on
it act in maximally undefined ways, because the optimizer may still
hoist code following the usual rules for undef. Make Poison invoke
its full undefined behavior only when it reaches an instruction with
externally visible side effects.

llvm-svn: 145913

llvm-svn: 145912

llvm-svn: 145911

llvm-svn: 145910

llvm-svn: 145908

purpose, and to avoid ambiguity with other uses of the word "trap"
in LangRef.

llvm-svn: 145907

It's always good to prune early, but formulae that are unsatisfactory
in their own right need to be removed before running any other pruning
heuristics. We easily avoid generating such formulae, but we need them
as an intermediate basis for forming other good formulae.

llvm-svn: 145906

llvm-svn: 145903

The new register allocator is much more able to split back up ranges too constrained by register classes.

Fixes <rdar://problem/10466609>

llvm-svn: 145899

llvm-svn: 145898

llvm-svn: 145897

llvm-svn: 145896

llvm-svn: 145895

llvm-svn: 145894

llvm-svn: 145893

rdar://10528060

llvm-svn: 145891

Previously, all ARM::CONSTPOOL_ENTRY instructions had a hardwired
alignment of 4 bytes emitted by ARMAsmPrinter.  Now the same alignment
is set on the basic block.

This is in preparation of supporting ARM constant pool islands with
different alignments.

llvm-svn: 145890

This was actually a bit of a mess. TLI.setPrefLoopAlignment was clearly
documented as taking log2(bytes) units, but the x86 target would still
set a preferred loop alignment of '16'.

CodePlacementOpt passed this number on to the basic block, and
AsmPrinter interpreted it as bytes.

Now both MachineFunction and MachineBasicBlock use logarithmic
alignments.

Obviously, MachineConstantPool still measures alignments in bytes, so we
can emulate the thrill of using as.

llvm-svn: 145889

value over that much.

llvm-svn: 145888

rdar://10069056

llvm-svn: 145885

llvm-svn: 145883

llvm-svn: 145882

Whether a fixup needs relaxation for the associated instruction is a
target-specific function, as the FIXME indicated. Create a hook for that
and use it.

llvm-svn: 145881

llvm-svn: 145880

don't do this now, but add a test case to prevent this from happening in the
future.
Additional test for rdar://9892684

llvm-svn: 145879

llvm-svn: 145878

llvm-svn: 145877

llvm-svn: 145876