b43/SSB: SPROM revision problem

b43/SSB: SPROM revision problem

[Rafał Miłecki]

I've problem with initializing N radio (2055) on my Broadcom card.

In case of my card simple workaround [1] for radio initialization is
needed. If I do not use workaround radio simply does not start. I
dumped MMIO operations of "wl" and it uses workaround for my card as
well.

The problem is that workaround condition is SPROM-based. It looks like this:
if (sprom->revision < 4)
	workaround = (some_specific_card);
else
	workaround = ((sprom->boardflags_hi & B43_BFH_NOPA) == 0);

So now goes the tricky part. My SSB has following SPROM:
[ 6208.080097] ssb: SPROM offset is 0x1000
[ 6208.089815] ssb: SPROM revision 5 detected.

However in drivers/ssb/pci.c there are some translations, including
one for my card. My SPROM revision gets translated to 4. The result
is:
sprom->revision: 4
sprom->boardflags_hi: 0x1

With that revision and boardflags workaround is not used and my radio
does not init. There are two possibilities:
1) Specs are wrong
2) We do SPROM translation incorrectly

Any hints about that?


[1] http://bcm-v4.sipsolutions.net/802.11/Radio/2055/Init#Post_Init

-- 
Rafa?

b43/SSB: SPROM revision problem

[Rafał Miłecki]

W dniu 22 pa?dziernika 2010 13:23 u?ytkownik Rafa? Mi?ecki
<zajec5@gmail.com> napisa?:
> I've problem with initializing N radio (2055) on my Broadcom card.
>
> In case of my card simple workaround [1] for radio initialization is
> needed. If I do not use workaround radio simply does not start. I
> dumped MMIO operations of "wl" and it uses workaround for my card as
> well.
>
> The problem is that workaround condition is SPROM-based. It looks like this:
> if (sprom->revision < 4)
> ? ? ? ?workaround = (some_specific_card);
> else
> ? ? ? ?workaround = ((sprom->boardflags_hi & B43_BFH_NOPA) == 0);
>
> So now goes the tricky part. My SSB has following SPROM:
> [ 6208.080097] ssb: SPROM offset is 0x1000
> [ 6208.089815] ssb: SPROM revision 5 detected.
>
> However in drivers/ssb/pci.c there are some translations, including
> one for my card. My SPROM revision gets translated to 4. The result
> is:
> sprom->revision: 4
> sprom->boardflags_hi: 0x1
>
> With that revision and boardflags workaround is not used and my radio
> does not init. There are two possibilities:
> 1) Specs are wrong
> 2) We do SPROM translation incorrectly
>
> Any hints about that?
>
>
> [1] http://bcm-v4.sipsolutions.net/802.11/Radio/2055/Init#Post_Init

Ahhh, I should check for boardflags2! Expect patches soon ;)

-- 
Rafa?

b43/SSB: SPROM revision problem

[Rafał Miłecki]

W dniu 22 pa?dziernika 2010 14:41 u?ytkownik Rafa? Mi?ecki
<zajec5@gmail.com> napisa?:
> W dniu 22 pa?dziernika 2010 13:23 u?ytkownik Rafa? Mi?ecki
> <zajec5@gmail.com> napisa?:
>> I've problem with initializing N radio (2055) on my Broadcom card.
>>
>> In case of my card simple workaround [1] for radio initialization is
>> needed. If I do not use workaround radio simply does not start. I
>> dumped MMIO operations of "wl" and it uses workaround for my card as
>> well.
>>
>> The problem is that workaround condition is SPROM-based. It looks like this:
>> if (sprom->revision < 4)
>> ? ? ? ?workaround = (some_specific_card);
>> else
>> ? ? ? ?workaround = ((sprom->boardflags_hi & B43_BFH_NOPA) == 0);
>>
>> So now goes the tricky part. My SSB has following SPROM:
>> [ 6208.080097] ssb: SPROM offset is 0x1000
>> [ 6208.089815] ssb: SPROM revision 5 detected.
>>
>> However in drivers/ssb/pci.c there are some translations, including
>> one for my card. My SPROM revision gets translated to 4. The result
>> is:
>> sprom->revision: 4
>> sprom->boardflags_hi: 0x1
>>
>> With that revision and boardflags workaround is not used and my radio
>> does not init. There are two possibilities:
>> 1) Specs are wrong
>> 2) We do SPROM translation incorrectly
>>
>> Any hints about that?
>>
>>
>> [1] http://bcm-v4.sipsolutions.net/802.11/Radio/2055/Init#Post_Init
>
> Ahhh, I should check for boardflags2! Expect patches soon ;)

Larry: could you check this and fix specs if I am right?

The way I understand this is following:
1) Workaround registers are:
a) 0x77: Core 1 RX Baseband Regulator
b) 0xa6: Core 2 RX Baseband Regulator
that really looks like regulatory-related
2) 0x00010000 BFL_NOPA has no PA
doesn't look like regulatory related
3) 0x00000001 BFL2_RXBB_INT_REG_DI external RX BB regulator present
looks like sth-related, so I guess we need to check for boardflags2

-- 
Rafa?