On Sat, Feb 15, 2014 at 02:23:23PM +0100, Arnd Bergmann wrote:
On Saturday 15 February 2014 11:20:08 Andrew Lunn wrote:
quoted
Instantiate the L2 cache from DT. Indicate in DT where the cache
control register is and if write through should be made.
Signed-off-by: Andrew Lunn <andrew@lunn.ch>
cc: devicetree at vger.kernel.org
I guess this answers part of my question for patch 5, but I also
wonder if the run-time setting is correct now with the hardcoded
#ifdef in arch/arm/mm/proc-feroceon.S checkign for the
Kconfig option. Presumably the code should match whatever is
set in the cache control register.
Humm, yes, good point.
None of the _defconfig's ever turn on
CACHE_FEROCEON_L2_WRITETHROUGH. I also did a quick google and could
not find any usage of it.
So i see two options:
1) Remove the wr-override from the DT binding and use
CACHE_FEROCEON_L2_WRITETHROUGH.
2) Remove CACHE_FEROCEON_L2_WRITETHROUGH and make proc-feroceon.S do
the right thing at runtime.
I suspect i will go for 1), it is simpler.
Thanks
Andrew