[Linux-Xtensa] Re: Audio Driver - applied your git patch, I'm looking into buildingyourconfig on an LX200...

Piet Delaney pdelaney at tensilica.com
Fri Sep 17 02:11:55 PDT 2010


benn.huang wrote:
> Hi Truby:
> 
> 
> ÓÚ Fri, 17 Sep 2010 16:55:21 +0800
> "truby.zhuang" <truby.zhuang at chipsbank.com> дµÀ:
> 
> 
>>>> ----------------------------------------------------------------------------------
>> I fixed the backtrace:
>> ----------------------------------------------------------------------------------------
> Piet already haved fix this issue by using non sleep memory allocation
> printk as temporary solution.

This seems like a good point to take a break and not end up
coming into work to late.

Marc was wondering about were we stand with the current
code with the kernel setting CPENABLE on the first usage
and it being associate with the task. If you access the
port in an interrupt handler you have to set and restore
it your selves in the interrupt handler.In the current
model once your driver does any optional access checks
the user space is free to access the I/O port. Other processes
are not allowed to access it from user space without some
kernel code initiating the access. One nice part is that
this approach piggy backs on the existing coprocessor
code which has weeks of testing, including SMP support,
and is know to be rock solid. One limitation is that once
you own a coprocessor you are locked to the current processor.
If another processor is also using the coprocessor, then
when you no longer own the co-processor registers you  are
free to migrate. It's slightly limiting for a I/O port as
there are no active registers but it's simple and well tested.

I'd like to extend it to provide for the kernel also being
able to use co-processor registers. With that addition we
could make it possible that I/O ports would only be accessible
via the driver if so configured.

I don't think Transwitch is using the I/O ports or coprocessors,
if they are I believe Prasanna is on the linux-xtensa mailing list
and can let us know what he thinks/wants. Same is try for any other
readers of the list.


-piet



More information about the linux-xtensa mailing list