[Dwarf-Discuss] Representing the location of smaller values in vector registers

Andreas Arnez arnez at linux.vnet.ibm.com
Wed Mar 9 03:35:06 PST 2016


On Fri, Mar 04 2016, Cary Coutant wrote:

> It's up to the ABI to define how smaller values are placed within a register.

Right, that seems to be the common understanding.  Maybe the confusion
arose because the standard does not explicitly make this statement.  Or
did I miss it?

Also, I wonder whether everybody is on the same page as to what this
really means.  Consider the following questions:

* What if the object is too large for the given register?  For instance,
  can the ABI map the object to a register pair?

* Is DW_OP_regx(n) followed by DW_OP_piece(size of object type) always
  equivalent to omitting the DW_OP_piece operation?

Answers are welcome, since I'm currently trying to fix various issues
with GDB's handling of composite location descriptions.  More questions
about DWARF pieces came up as well:

  https://sourceware.org/ml/gdb/2016-01/msg00013.html

-- 
Andreas Arnez



More information about the Dwarf-Discuss mailing list