Re: AP6275 / bcm43752 pcie on mainline brcmf

[Angus Ainslie <angus@akkea.ca>]

On 2022-11-17 10:12, Arend Van Spriel wrote:
> On November 17, 2022 5:06:25 PM Angus Ainslie <angus@akkea.ca> wrote:
> 
>> On 2022-11-17 04:01, Andy Green wrote:
>>> On 11/17/22 09:23, Arend van Spriel wrote:
>>>> On 11/16/2022 1:25 PM, Shengyu Qu wrote:
>>>>> Got a new version of downstream driver and firmware here:
>>>>> https://gitlab.com/rk3588_linux/linux/external/rkwifibt/-/tree/linux-5.10-gen-rkr3.4
>>>> 
>>>> Thanks for that. The firmware indeed seems to originate from 
>>>> Synaptics
>>>> and this module can be found on wifi-alliance product finder:
>>>> 
>>>> https://www.wi-fi.org/product-finder-results?sort_by=certified&sort_order=desc&keywords=ap6275
>>>> 
>>>> We have no known contacts there and they do not seem to have much
>>>> interest in open-source. In that gitlab repo I see they are 
>>>> providing
>>>> bcmdhd driver and I assume they consider that sufficient.
>>>> 
>>>> Will see what I can do. First have to think on how to approach this
>>>> failure to get most out of a debug effort. I do not have the module
>>>> over here so we would need to go back and forth. The fact that the
>>>> issue is seen on both SDIO and PCIe is somewhat encouraging.
>>> 
>>> Yes it seems that at some point with some firmware, SDIO has worked
>>> for, eg, Angus.  And, it should work well with bchdhd (but that does
>>> not seem to work for scan for me on v6.1-rc5 uplevel with the 
>>> firmware
>>> load changed to use kernel apis; but the firmware has no timeout on
>>> bcmdhd). So it doesn't feel a million miles away.
>> 
>> This is the firmware we have been using for SDIO. It was sent to us
>> directly by SparkLAN so not sure how it compares to other variants.
>> 
>> https://source.puri.sm/Librem5/firmware-brcm43752-nonfree/-/tree/master/firmware/brcm
> 
> It is a tad bit older:
> 
> $ strings brcmfmac43752-sdio.bin | grep FW-TAG
> <FW-TAG>4362a2-roml/config_sdio_ampak_release_he
> Version=18.35.387.23.7 (b328500@shgit) (wlan=r880297)
> Date=2020-07-13T10:34:00Z FWID=01-f50fdff0
> 

Currently it's broken for hotspot mode but I haven't had the chance to 
see if that is FW or something further up the stack.

Angus

> Thanks,
> Arend
> 
>> 
>> Angus
>> 
>>> I have the hw, and (paid) time to help with debugging this "pair of
>>> hands" style. I'm also a little familiar with the kernel code and can
>>> try jiggle things autonomously if you can prompt what is interesting.
>>> 
>>> -Andy
>>> 
>>>> Regards,
>>>> Arend