NAK: [PATCH 0/3][linux-firmware][SRU][G] Support Killer 500s (QCA6390) WLAN/BT

[Seth Forshee]

On Tue, Dec 15, 2020 at 11:53:20AM +0800, You-Sheng Yang wrote:
> On 12/15/20 12:25 AM, Seth Forshee wrote:
> > On Fri, Dec 04, 2020 at 11:11:44PM +0800, You-Sheng Yang wrote:
> >> BugLink: https://bugs.launchpad.net/bugs/1879633
> >>
> >> [Impact]
> >>
> >> Qualcomm QCA6390 series not recognized due to the lack of ath11k driver.
> >>
> >> $ lspci
> >> 0000:55:00.0 Unassigned class [ff00]: Qualcomm Device [17cb:1101]
> >>  Subsystem: Bigfoot Networks, Inc. Device [1a56:a501]
> >>
> >> [Fix]
> >>
> >> 25 additional fixes in ath tree
> >> (https://git.kernel.org/pub/scm/linux/kernel/git/kvalo/ath.git), tag
> >> ath11k-qca6390-bringup-202011301608 to resolve issues on TGL platforms,
> >> and firmware updates from mainline are required.
> >>
> >> [Test Case]
> >>
> >> 1. Install firmwares & boot from patched kernel,
> >> 2. ath11k should be now up and running. Check with
> >>    `sudo dmesg | grep ath11k`
> >>
> >> [Where problems could occur]
> >>
> >> There has been a lot patches landed in kernel since 5.7 devel cycle. The
> >> last bits here has merge base with mainline at v5.10-rc4, and since then
> >> only patches for ath11k itself has been committed. The SRU here installs
> >> new firmware blobs for ath11k itself, so it shouldn't introduce any
> >> regression.
> > 
> > First, this is not true. As the diffstat below clearly shows,
> > qca/htbtfw20.tlv and qca/htnv20.bin are existing firmware which are
> > updated by these patches. So at minimum you need to address this in the
> > regression potential.
> 
> The existing firmware blobs for AX500 bluetooth have some defects in the
> init sequence and may result in inability to reconnect a previously
> paired device over reboots/airplane mode. It was not fully tested and
> was at most for a initial taste.

Thanks. This information should be included in the SRU justification.

> > Second, some of the files added by this update aren't currently used in
> > the groovy kernel, and I don't see patches on the list which would add
> > their use. Did I overlook the corresponding kernel patches?
> 
> Its backport to OEM 5.6 tree is still under testing as it would require
> 380 patches in which 220 belongs to ath11k itself and many more for
> qrtr, mhi bus, ath11k bluetooth, and certainly some to the wireless
> stack. Regression tests are being deployed to both ath11k and non-ath11k
> platforms to them, so I'm hold the patches to 5.8 kernel (which probably
> cannot pass due to the quantity), and see if the plan is to use DKMS
> instead or any other possible path.
> 
> So, yes, there is no 5.8 patch for ath11k proposed currently.

I think we should hold off on the new firmware files until we have some
consumer of these files on its way into groovy. If you want to send just
the qca/* updates in the meantime, we can go ahead and get those in.

Thanks,
Seth