Re: [PATCH] ath10k: support bus and device specific API 1 BDF selection

[Christian Lamparter]

On 14/10/2021 14:01, Robert Marko wrote:
> On Thu, 14 Oct 2021 at 13:54, Christian Lamparter <chunkeey@xxxxxxxxx> wrote:
> > On 10/10/2021 00:17, Robert Marko wrote:
> > > Some ath10k IPQ40xx devices like the MikroTik hAP ac2 and ac3 require the
> > > BDF-s to be extracted from the device storage instead of shipping packaged
> > > API 2 BDF-s.
> > >
> > > This is required as MikroTik has started shipping boards that require BDF-s
> > > to be updated, as otherwise their WLAN performance really suffers.
> > > This is however impossible as the devices that require this are release
> > > under the same revision and its not possible to differentiate them from
> > > devices using the older BDF-s.
> > >
> > > In OpenWrt we are extracting the calibration data during runtime and we are
> > > able to extract the BDF-s in the same manner, however we cannot package the
> > > BDF-s to API 2 format on the fly and can only use API 1 to provide BDF-s on
> > > the fly.
> > > This is an issue as the ath10k driver explicitly looks only for the
> > > board.bin file and not for something like board-bus-device.bin like it does
> > > for pre-cal data.
> > > Due to this we have no way of providing correct BDF-s on the fly, so lets
> > > extend the ath10k driver to first look for BDF-s in the
> > > board-bus-device.bin format, for example: board-ahb-a800000.wifi.bin
> > > If that fails, look for the default board file name as defined previously.
> > >
> > > Signed-off-by: Robert Marko <robimarko@xxxxxxxxx>
> > > ---
> >
> > As mentioned in Robert's OpenWrt Pull request:
> > https://github.com/openwrt/openwrt/pull/4679
> >
> > It looks like the data comes from an mtd-partition parser.
> > So the board data takes an extra detour through userspace
> > for this.
> >
> > Maybe it would be great, if that BDF (and likewise pre-cal)
> > files could be fetched via an nvmem-consumer there?
> > (Kalle: like the ath9k-nvmem patches)
>
> Christian, in this case, NVMEM wont work as this is not just read from
> an MTD device, it first needs to be parsed from the MikroTik TLV, and
> then decompressed as they use LZO with RLE to compress the caldata
> and BDF-s.

For more context here (it's unrelated to the patch):
There is more custom code than just the mtd splitter.
I do fear that this could be turning into a dreaded "separation between
mechanism vs policy"-proxy discussion with that in-kernel LZOR
decompressor/extractor and the way that the board-data then has be
rerouted through user-space back to ath10k.

---

As for the proposed feature: Yeah, back in 2017/2018-ish, I would have
really loved to have this "load separate board-1 based on device-location".
Instead the QCA4019's board-2.bin is now bigger than the device's
firmware itself. From what I can see, there are also more outstanding
board-2.bin merge requests too, though some those are updates.

Cheers,
Christian