Re: [GIT PULL 9/10] arm64: tegra: Device tree changes for v4.10-rc1

[Olof Johansson <olof-nZhT3qVonbNeoWH0uzbU5w@public.gmane.org>]

On Mon, Nov 21, 2016 at 09:22:10AM +0100, Thierry Reding wrote:
> On Fri, Nov 18, 2016 at 06:40:21PM -0800, Olof Johansson wrote:
> > On Fri, Nov 18, 2016 at 05:17:18PM +0100, Thierry Reding wrote:
> > > Hi ARM SoC maintainers,
> > > 
> > > The following changes since commit 1001354ca34179f3db924eb66672442a173147dc:
> > > 
> > >   Linux 4.9-rc1 (2016-10-15 12:17:50 -0700)
> > > 
> > > are available in the git repository at:
> > > 
> > >   git://git.kernel.org/pub/scm/linux/kernel/git/tegra/linux.git tags/tegra-for-4.10-arm64-dt
> > > 
> > > for you to fetch changes up to cc13b4fa4ac780cec6c21b64a39ab2950e95e8f6:
> > > 
> > >   arm64: tegra: Add NVIDIA P2771 board support (2016-11-18 14:35:53 +0100)
> > > 
> > > Thanks,
> > > Thierry
> > > 
> > > ----------------------------------------------------------------
> > > arm64: tegra: Device tree changes for v4.10-rc1
> > > 
> > > This adds initial support for Tegra186, the P3310 processor module as
> > > well as the P2771 development board. Not much is functional, but there
> > > is enough to boot to an initial ramdisk with debug serial output.
> > > 
> > > ----------------------------------------------------------------
> > > Dan Carpenter (1):
> > >       mailbox: tegra-hsp: Use after free in tegra_hsp_remove_doorbells()
> > > 
> > > Joseph Lo (6):
> > >       soc/tegra: Add Tegra186 support
> > >       dt-bindings: mailbox: Add Tegra HSP binding
> > >       dt-bindings: firmware: Add bindings for Tegra BPMP
> > >       arm64: tegra: Add Tegra186 support
> > >       arm64: tegra: Add NVIDIA P3310 processor module support
> > >       arm64: tegra: Add NVIDIA P2771 board support
> > > 
> > > Stephen Warren (2):
> > >       dt-bindings: Add power domains to Tegra BPMP firmware
> > >       dt-bindings: firmware: Allow child nodes inside the Tegra BPMP
> > > 
> > > Thierry Reding (12):
> > >       Merge branch 'for-4.10/soc' into for-4.10/mailbox
> > >       mailbox: Add Tegra HSP driver
> > >       Merge branch 'for-4.10/mailbox' into for-4.10/firmware
> > >       firmware: tegra: Add IVC library
> > >       firmware: tegra: Add BPMP support
> > >       Merge branch 'for-4.10/firmware' into for-4.10/arm64/dt
> > >       arm64: tegra: Add CPU nodes for Tegra186
> > >       arm64: tegra: Add serial ports on Tegra186
> > >       arm64: tegra: Add I2C controllers on Tegra186
> > >       arm64: tegra: Add SDHCI controllers on Tegra186
> > >       arm64: tegra: Add GPIO controllers on Tegra186
> > >       arm64: tegra: Enable PSCI on P3310
> > 
> > The drivers->dt dependency here is annoying. Any chance you can respin without
> > it?
> > 
> > We've been encouraging people to consider using numerical clock/gpio/reset
> > numbers on initial submission to avoid these dependencies on dt-bindings
> > includes, and then follow up with a move to the symbolic names between -rc1 and
> > -rc2. Mind doing the same here?
> 
> Yes, I can do that. Would it be acceptable to have a dt-bindings->dt
> dependency? Stephen's already done a good job of avoiding this kind of
> dependency by getting the bindings, and hence dt-bindings headers,
> merged ahead of Linux kernel support because he had already gotten the
> bindings reviewed and finalized during his work on U-Boot.
> 
> I've been told in the past that it's not necessary to strictly split DT
> bindings patches from driver patches, but I suppose if a dt-bindings->dt
> is acceptable, then splitting things up more strictly would actually be
> the preferable solution here because it also avoids the slight churn of
> converting to symbolic values later on.

Sorry, haven't been looking at this older for a while so this reply is
high-latency.

If you want to do a separate dt-bindings branch that includes the headers and
is present in both the dt and drivers branch, that'd work. If the changes are
trivial though, and just contains a few clocks, I think we should just have
them opencoded on the original merge, and then move over once the include files
have gone in. I'd be willing to merge such conversions between -rc1 and -rc2.

That's in particular the case when there's an external driver tree/maintainer,
since it avoids the three-way handshake, etc.


-Olof