Re: [RFC] bcma: replace ssb_sprom structure with bcma_sprom structure

["Arend van Spriel" <arend@broadcom.com>]

On 03/07/2012 12:23 AM, Hauke Mehrtens wrote:
> On 03/06/2012 01:26 PM, Arend van Spriel wrote:
>> BCMA shared a structure with SSB in which sprom data was stored. This
>> could have been appropriate but with upcoming srom revisions that are
>> of no interest to SSB it is better to have own definitions with BCMA.
>>
>> This patch adds support for srom revisions 8 and 9.
>
> Will sprom version 10 contain many new or different attributes so the
> struct has to be extended a lot in comparison to the changes made from
> version 8 to 9?
>

For version 10 there are only a couple of new attributes, but version 11 
is a major overhaul. Initially, I intended to have the bcma_sprom 
structure contain a union of revision specific structs, but the flat 
ssb_sprom structure made me reconsider. Looking at the current size of 
bcma_sprom I am tempted to move back to the original union solution.

>>
>> Signed-off-by: Arend van Spriel<arend@broadcom.com>
>> ---
>> The BCMA structure is based on internal specifications. Therefor the structure
>> fields are not matching the SSB sprom structure. By dropping the use of the SSB
>> sprom structure in BCMA a fair amount of work is needed in b43 driver. If it
>> helps I can provide a mapping between bcma_sprom and ssb_sprom structure members.
> The bcm47xx architecture code will not compile with this patch either.
>
> A quick look into b43 showed that it uses the boardflags flags all other
> the code and the other stuff in most cases just in some initialization
> function of the specific phy so it should be manageable to use a
> different sprom strcut for bcma. Is there any other PHY than the N-PHY
> that is used with ssb and bcma?

I believe all AXI chips are having some flavour of N-PHY (or 11ac).

Gr. AvS