On Thursday 22 September 2011, Pawel Moll wrote:

Executive summary:

* I have second map_desc with pfn for RS1 memory map, but using the same
virtual address as the legacy one. The legacy one is used if root of the
tree is compatible with "arm,vexpress-legacy".

* The devices I need to use in v2m.c have aliases in DTS so I can find
their offsets in the flat tree (the *_find_node_by_alias() function is
rather generic and could be moved to drivers/of/fdt.c if you think it
would be useful for others).

* There are no more users of MMIO_P2V in v2m.c, next thing I will do is
the same in core tile; then the macro can be killed. Once this happens
the virtual address currently taken from __MMIO_P2V(V2M_PA_CS7) will be
replaced by some kind of "#define V2M_PERIPH_BASE 0xf8000000".

* Once Nico's changes regarding static maps are in, the manual pointer
operations in v2m_dt_map_io can be replaced with neat ioremap()-s.

All feedback appreciated, cheers!

Look ok to me. Just a comment on the submission:

Even when you post something for review instead of inclusion, please add
a Signed-off-by: line and a diffstat.

	Arnd