|
From: Ø. H. <oyv...@zy...> - 2008-02-18 19:50:03
|
On Feb 18, 2008 7:12 PM, Dominic Rath <Dom...@gm...> wrote: > On Monday 18 February 2008 09:08:44 Øyvind Harboe wrote: > > I'd like to see working_area deal with RAM moving after the MMU has > > been enabled. > > > > How about adding an optional argument for a physical address? > > > > If the MMU is enabled the virtual address is used, otherwise the > > phsyical address is used. Default physical address is the same > > as the virtual address. > > > > Note that the virtual address does not need to point to the physical > > memory area used. > > > > Any better ideas? Objections? > > I suppose the "clean" approach would be to make OpenOCD's target layer aware > of virtual memory. Currently, MMU handling is implemented on a per-target > level, where it is inaccessible to the code deciding where the working_area > goes. So no objections or other ideas to the change in syntax & functionality of working_area from the users point of view? If there is a consensus on the new syntax/semantics of working_area, I'll start tinkering with an implementation(target independent). -- Øyvind Harboe http://www.zylin.com - eCos ARM & FPGA developer kit |