[GIT PULL 7/7] Samsung SoC for v3.14

[Olof Johansson]

On Tue, Dec 24, 2013 at 10:35:22AM +0900, Kukjin Kim wrote:
> Olof Johansson wrote:
> > 
> > On Mon, Dec 23, 2013 at 10:36:18AM +0100, Lukasz Majewski wrote:
> > > Hi Tomasz,
> > >
> > > > Hi Olof,
> > > >
> > > > On Sunday 22 of December 2013 14:11:41 Olof Johansson wrote:
> > > > > On Mon, Dec 16, 2013 at 05:23:01AM +0900, Kukjin Kim wrote:
> > > > > > The following changes since commit
> > > > > > 6ce4eac1f600b34f2f7f58f9cd8f0503d79e42ae:
> > > > > >
> > > > > >   Linux 3.13-rc1 (2013-11-22 11:30:55 -0800)
> > > > > >
> > > > > > are available in the git repository at:
> > > > > >
> > > > >
> > > > > > git://git.kernel.org/pub/scm/linux/kernel/git/kgene/linux-
> > samsung.git
> > > > > > tags/samsung-soc
> > > > > >
> > > > > > for you to fetch changes up to
> > > > > > 538cfbb4c40ab59688236484138133b8e3e89220:
> > > > > >
> > > > > >   ARM: dts: Add initial support for Arndale Octa board (2013-12-16
> > > > > > 05:05:43 +0900)
> > > > >
> > > > > Hi,
> > > > >
> > > > > I'm not going to pull this.
> > > > >
> > > > > I've asked you for about 6 months now (ever since Arnd went on
> > > > > leave) to please finish his multiplatform work for Exynos. Nothing
> > > > > has happened.
> > > > >
> > > > > I'm not going to merge any more platform or SoC enablement code
> > > > > for Samsung platforms until Exynos has been converted over to
> > > > > multiplatform. Sorry.
> > > >
> > > > Well, it's hard to disagree with you on this. Keeping adding new
> > > > things constantly definitely does not make us closer to multiplatform
> > > > support, as it only increases codebase of code that needs to be
> > > > cleaned up and made multiplatform aware.
> > > >
> 
> Well, I don't know why this will hurt multiplatform, even you requested that
> long time ago. I thought I could finish that by end of this year but it will
> be done in Jan. Of course, I'm not sure it could be merged in 3.14, because
> it's up to arm-soc tree, it could be sent to before merge window though.
> 
> Samsung SoC patches in this pull-request already got review in mailing-list
> and it's ready to merge in the mainline. I think, multiplatform support
> issue should be handled separately, it's up to you and arm-soc maintainers
> though.

Well, multiplatform support _has_ been "handled separately" for 6 months
already and as a result not much has happened.

Btw, this isn't personal -- we're not handling Exynos differently from
any other vendor out there. For example, we have made agreements with the
Renesas developers that we will not merge any more non-multiplatform SoC
support from them, and they are now working on migrating theirs over too. But
Exynos is one of the last remaining large platforms that have not yet been
moved over, and it's time to make it a high priority.

> > > > However it's not that we're not doing anything towards mutliplatform
> > > > support. We're slowly getting there. A bit too slowly, but I'm afraid
> > > > this is due to the fact that we have definitely too little manpower
> > > > working on mainline support of Samsung SoCs.
> > > >
> > > > As for good news, I already have a series cleaning up PM/sleep support
> > > > and making it multiplatform friendly and I'm going to post it
> > > > tomorrow. The bad news is that it depends on other series and I'm not
> > > > sure if we can get it merged in this release. I'll try to do whatever
> > > > possible to merge things as soon as possible, though.
> > > >
> > > > From smaller things, we still need to sort out Exynos cpufreq driver
> > > > that has dependencies on headers in plat-samsung/ and mach-exynos/.
> > > > Unfortunately I don't have so much time to work on all the things at
> > > > the same time, so I hope that someone else could pick this task up.
> > > > Marek, Lukasz, what do you think?
> > >
> > > Regarding the cpufreq for Exynos, we do need to clean things up.
> > >
> > > It shall be possible to reuse generic cpufreq-cpu0.c and
> > > arm_big_little.c code instead of several copy pasted
> > > exynos[4|5]xxx-cpufreq.c ones.
> > >
> > > This change will require some "virtual" clocks implementation for atomic
> > > clocks dividers change.
> > >
> > > I've already committed myself to fix this code. When all goes smooth, I
> > > shall deliver some RFC code after new year.
> > 
> > Great!
> > 
> > Sounds like if you can start with this after new year that 3.15 will be
> > a good target for Exynos multiplatform. It's way overdue, and I would
> > be very happy to see it completed.
> > 
> And Thomas P Abraham is working on multiplatform for exynos stuff and I
> think he could finish in Jan. So I think Samsung folks need to talk
> internally to avoid duplication efforts, let me talk to Marek in SRPC.

Sounds good.

Do you have a list of things that still need resolution, besides cpufreq and
resurrecting Arnd's old patches to move things over?


-Olof