ath10k-firmware: QCA6174 hw3.2 target 0x05030000 chip_id 0x00340aff sub 168c:3370 board file

[Hans de Goede]

Hi,

On 1/21/22 15:16, Kalle Valo wrote:
> Hans de Goede <hans at hansg.org> writes:
> 
>> Here is a board-file for the QCA6174 PCI 168c:003e subsys: 168c:3370 wifi
>> card found on Microsoft Surface Go devices.
>>
>> I'm submitting this per the docs in:
>> https://wireless.wiki.kernel.org/en/users/drivers/ath10k/boardfiles
>> I hope those are still up to date.
>>
>> Currently the kernel gives the following message when loading on this card:
>> "failed to fetch board data for
>> bus=pci,vendor=168c,device=003e,subsystem-vendor=168c,subsystem-device=3370
>> from ath10k/QCA6174/hw3.0/board-2.bin"
>>
>> And then errors out with a bunch of -110 (-ETIMEOUT) errors.
>>
>> Replacing the fallback board.bin file with the attached .bin file fixes
>> these errors and results in working wifi (I tested at both 2.4G and 5G).
>>
>> There are various internet forum posts which point to this board file:
>> https://web.archive.org/web/20201111213909/http://www.killernetworking.com/support/K1535_Debian/board.bin
>>
>> But the killer 1535 wifi card uses completely different subsys-ids
>> of 1a56:1535, so even though that happens to work I do not believe
>> that is the correct file.
>>
>> So instead, inspired by this report:
>> https://bugzilla.redhat.com/show_bug.cgi?id=1651779
>>
>> I started looking at the driverstore directory of the original windows
>> install on the device I found "netathr10x.inf" there which has this line:
>>
>> eDesc.6320_3% = ATHR_DEV_OS63_988x_TX8.ndi,
>> PCI\VEN_168C&DEV_003E&SUBSYS_3370168C&REV_32; WiFi Only SKU
>>
>> Note the "WiFi Only SKU" remark is there to differentiate this from
>> the 168c:3371 subsys one which also has LTE (marked as "WiFi/LTE SKU")
>>
>> This led me to the "ATHR_DEV_OS63_988x_TX8.ndi" section of the .inf file:
>>
>> [ATHR_DEV_OS63_988x_TX8.ndi.NTamd64]
>> ...
>> CopyFiles       = atheros.CopyFiles, atheros_FW_3X.CopyFiles, atheros_BD_TX8.CopyFiles
>> ...
>>
>> And "atheros_BD_TX8.CopyFiles" is:
>>
>> [atheros_BD_TX8.CopyFiles]
>> eeprom_ar6320_3p0_TX8_clpc.bin,,,2
>>
>> The "windows\system32\driverstore\netathr10x" dir has a ton of
>> eeprom files (let me know if you want a copy of the whole dir)
>> including the mentioned one. Where as "windows\system32\drivers"
>> only has the one eeprom_ar6320_3p0_TX8_clpc.bin eeprom... file,
>> further indicating that that is the one actually being used.
>>
>> So this is the file which I've tested with and which is attached to this email as:
>> "bus=pci,vendor=168c,device=003e,subsystem-vendor=168c,subsystem-device=3370.bin"
>>
>> md5sum:
>> bc52aa5640b27fa50f9d4d144f81e169
>> bus=pci,vendor=168c,device=003e,subsystem-vendor=168c,subsystem-device=3370.bin
> 
> I assume this is the same version as you submitted directly to
> linux-firmware[1],

Correct, note that my upstream linux-firmware submission contains
2 new board files. I've also submitted this to the ath10k-firmware repo here:

https://github.com/kvalo/ath10k-firmware/pull/11

Regards,

Hans