[PATCH 0/8] arm64: hi6220-hikey: Fix WiFi support

On 23 May 2017 at 14:00, Arnd Bergmann [off-list ref] wrote:

On Mon, May 22, 2017 at 10:40 AM, Ulf Hansson [off-list ref] wrote:

quoted

+ Arnd, Rob

On 8 May 2017 at 18:21, Ulf Hansson [off-list ref] wrote:

quoted

The Hikey board has a Wifi chip attached to the SDIO bus, which is managed by a
dwmmc controller. This is currently broken in that way that not even the SDIO
card is being detected.

To fix this we need to make sure the correct resources are described for the
dwmmc controller, so the mmc core can power on/off the SDIO/WiFi chip
correctly. This is the main issue today.

I have picked up one change from Daniel Lezcano's earlier submission [1] and
folded that into this series, as the Wifi support depends on it. As a matter of
fact I have even split Daniel's change into two halves, one for DT
documentation and one for arm64 hikey dts.

It's important that this series goes together as to conform with TI's HW spec
of the wl18xx Wifi chip, as to prevent it from being damaged. We can either go
via arm-soc, unless Wei Xu the Hikey SoC maintainer, prefers to ack the dts
changes, then I can pick the series via my mmc tree.

Moreover, I would suggest we send this as fixes for the 4.12 rcs to get the
Wifi working for Hikey again.

This series fixes my wifi issues on the Hikey board and currently all
but the hikey dts changes seems to be okay/acked.

Rob and me had some discussions regarding the new mmc DT binding,
however in the last response from Rob he indicates that he is not
objecting to adding the new binding.

So, unless I hear some objection from any of you, I am going to queue
this up for fixes via my mmc tree within a day or so.

Hi Ulf,

I looked at all the changes, and they seem reasonable to me. Please
add my

Acked-by: Arnd Bergmann <arnd@arndb.de>

Thanks Arnd!

and put them into a separate branch that you merge into your mmc
git, so we can merge the changes with the other updates in case
there are conflicts.

I do that!

However, these will be sent as fixes for 4.12 rcs. I guess you can
pull the changes from there later instead.

I see that the only code change is in patch 2. Are you able to put
that patch last in the series without introducing bisection problems?
That would be preferred, as it allows us to keep a pure 'DT' branch
if we end up merging your other seven patches in.

Unfortunate not. Because of one may then end up running an incorrect
power off sequence, potentially damaging the Wifi chip.

Kind regards
Uffe