[Dwarf-Discuss] Register kinds

Relph, Richard Richard.Relph at amd.com
Fri Oct 29 15:27:14 PDT 2010


When establishing a mapping from DWARF on to a new machine architecture, one has to establish a mechanism for referring to registers.
If the architecture supports more than one kind of register, what's the recommended approach for 'naming' registers of a kind other than the most common?
For example, if an architecture defines an abstract machine that has an arbitrary number of 'normal' registers, but also includes support for some special purpose registers (say, a kind of register designed to optimize access to memory, or vector registers, or even FP registers), how can DWARF compactly refer to registers of these non-normal kinds?
The obvious approach is to establish an arbitrary ceiling on the number of normal registers and then encode the register kind in bits above that (or some variation of encoding register kind and register number in a single 'number'). But I wanted to see if there is some other approach that has been successfully tried.

Thanks,
Richard
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.dwarfstd.org/private.cgi/dwarf-discuss-dwarfstd.org/attachments/20101029/b1e2e033/attachment-0001.htm>


More information about the Dwarf-Discuss mailing list