Re: L2 cache evbarm

[Frank Zerangue <frank.zerangue%gmail.com@localhost>]

Thank you.

> On Dec 17, 2014, at 2:55 PM, Matt Thomas <matt%3am-software.com@localhost> wrote:
> 
> 
>> On Dec 17, 2014, at 12:26 PM, Frank Zerangue <frank.zerangue%gmail.com@localhost> wrote:
>> 
>> I have a private port for an iMX31 based board which is itself based on evbarm. When I enable L2 cache in start.S the startup process freezes when I switch to the real L1 page table in initarm(). 
> 
> Don't do it that way.  Make sure the arml2cc config line has no flags or flags 0 and it will enable the cache when it gets attached if it's not already enabled.
> 
>> I wrote a custom flash boot program that loads the kernel with an initial ram disk. I enable L2 cache in the flash boot program and I see about a 15% to 20% improvement in load time; but when I start the kernel I do not enable L2 cache for the reason described above.
>> 
>> Can you give me some clues into the infrastructure that may need work to support L2 cache?
> 
> The above is all that needs to be done on cortex-a9.