[PATCH]AArch64 Neon Post-index Load/Store multiple N-element instructions.

[Hao Liu]

Hi Tim,

 

This llvm patch implements post-index load/store multiple N-element
instructions. As these instructions share the same ACLE functions with
normal load/store multiple N-element instructions (my last patch), there is
no patch for Clang.

 

The implementation of these instructions are according to ARM backend, which
replace normal load/store nodes in ISelLowing with post-index  nodes. 

There is a little difference with fixed offset in ARM backend. If the
increment is an immediate and equal to the length of the vector list, the
instruction is like:

           ld1, {v0.8b, v1,8b}, [x0], #16

The '#16' is similar to '!' of load/store post instructions in ARM backend.

 

The patch is also added to link: http://llvm-reviews.chandlerc.com/D2091.

Please review it.

 

Thanks,

-Hao   
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20131102/f2fcd829/attachment.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: ldst-post-llvm.patch
Type: application/octet-stream
Size: 74685 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20131102/f2fcd829/attachment.obj>