On 2021-08-30 10:58:09, Bjorn Andersson wrote:
[...]

quoted

quoted

Afaik these devices all boots off a boot.img, which means that it's
unlikely that a new kernel is installed on a device with an older DT.
None of them would boot mainline off the DT that shipped with the
original product.

That was my understanding as well, DT overlays are a "new thing" afaik
and most devices (at least all Sony's that I'm working with) use an
appended DTB that's always in-sync with the kernel image.

I think that with the introduction of DT overlays the system becomes
more flexible and as such more susceptible for bugs caused by unexpected
DT versions.

Offtopic: We have some problems with this on newer Sony devices where
the BL indeed tries to overlay this DTBO on the DT, which is usually a
downstream DT not fitting on top of a mainline kernel+appended-DTB.  The
solution is to simply wipe DTBO, and afaik it should be fine to compile
the overlay bits directly inside the appended-DTB anyway.  Leads to
unsuspecting problems at times, but it is manageable.

I think in practice the real issues comes when the DTB is delivered
separately (i.e. not by boot.img) or inbetween two kernel releases where
the Qualcomm tree might not be in sync with the driver tree.

Dmitry sees this as a problem for msm8974 but I'm not familiar enough
with the board.  I take it this doesn't use appended DTBs then?

quoted

quoted

As such, if I pick this patch up as a fix for 5.15 you can respin the
other two patches and they can land in 5.16 and I would be surprised if
anyone will run into any issues with it.

I.e. I've applied this patch.

Sounds good, I'll leave this patch out from v2.  Should it have a Fixes:
tag to get backported too?

Sounds good, I added to this:

Fixes: 6969d1d9c615 ("ARM: dts: qcom-apq8064: Set 'cxo_board' as ref clock of the DSI PHY")

Did the same on the v2 respin since it seems like those patches were
incomplete without the driver change.

quoted

Since most review seems to be in I'll respin v2 shortly with the
addition of the "ref" clock to msm8974, that should probably get the
same treatment (added to 5.15 fixes) then we can land this patchset in
5.16 (maybe without .name= fallback if Dmitry is okay with that).

Sounds good.

Sent the msm8974 patch separately and re-spun a v2.  I haven't added the
.name="xo" fallback yet while awaiting Dmitry to see if that counts as
enough time for firmware to be delivered between kernel 5.15 and 5.16.

- Marijn