should xfer_memory vector functions perform multiple transfers?

[Andrew Cagney]

"J.T. Conklin" wrote:
> 
> In many (most? (all?)) cases, the xfer_memory vector functions will
> perform multiple transfers if the size of the transfer is too large
> to do in one.  The high level logic in target_xfer_memory() will do
> the same.

(Ah, yes, that ...)

> Is there a rule about whether a too-large transfer should be split by
> xfer_memory vector functions or target_xfer_memory()?  If it can be
> done by vector functions, the value of target_xfer_memory_partial()
> is limited.  (target_xfer_memory_partial() is used (indirectly) in
> generic_load() so that GDB is not completely stalled during large
> transfers.

My strong preference is for the target_xfer() to just move as much data
as can be copied in a single atomic operation.

	Andrew