Re: [PATCH 10/10] brcmfmac: Add support for multiple PCIE devices in nvram.

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

On 04/21/15 12:16, Rafał Miłecki wrote:
> On 20 April 2015 at 22:16, Arend van Spriel<arend@broadcom.com>  wrote:
>> On 04/20/15 20:49, Rafał Miłecki wrote:
>>>
>>> On 20 April 2015 at 19:12, Arend van Spriel<arend@broadcom.com>   wrote:
>>>>
>>>> On 04/20/15 13:26, Rafał Miłecki wrote:
>>>>>
>>>>>
>>>>> On 17 April 2015 at 10:50, Arend van Spriel<arend@broadcom.com>    wrote:
>>>>>>
>>>>>> Another option is to add the parsing stuff in that nvram code and have
>>>>>> an
>>>>>> api to get the appropriate portion based on pcie domain and bus number
>>>>>> as
>>>>>> used in brcmf_fw_get_firmwares_pcie(). However, I would prefer to have
>>>>>> this
>>>>>> in the driver and not in arch specific code as there may be other
>>>>>> platforms
>>>>>> like our set-top boxes needing this.
>>>>>
>>>>>
>>>>>
>>>>> This is some plan for the future I was lacking from the beginning. It
>>>>> makes things more clear now, thanks.
>>>>
>>>>
>>>> You're welcome. Do you want to see this clarification in the commit
>>>> message?
>>>
>>>
>>> I don't really need that, I'm leaving it up to you :)
>>>
>>> The last remaining question from me is if  about this NVRAM validation
>>> function (if it exists).
>>
>>
>> Ok. I can try to answer that one. The nvram parsing code in firmware.c
>> intends to do just that. So apart from stripping comments and whitespace it
>> validates each line and gives off a warning if a wrongly formatted entry is
>> found.
>
> OK, thanks. So I guess this patch is OK after all?

Until proven otherwise. I tested it with a number of format violations, 
but there could be more creative people out there coming up with a 
pattern uncovering some issue.

Regards,
Arend