[GIT PULL] Renesas ARM Based SoC DT Timers Updates for v3.17

Olof Johansson olof at lixom.net
Wed Jul 23 17:00:21 PDT 2014


Ah, sorry. Got lost in the other pull requests.

On Wed, Jul 23, 2014 at 4:54 PM, Simon Horman <horms at verge.net.au> wrote:
> Hi Olof,
>
> I'd value your feedback on this if you have a moment.
>
> On Sun, Jul 20, 2014 at 10:51:25PM +0900, Simon Horman wrote:
>> On Fri, Jul 18, 2014 at 10:27:58PM -0700, Olof Johansson wrote:
>> > On Thu, Jul 17, 2014 at 09:40:20AM +0900, Simon Horman wrote:
>> > > Hi Olof, Hi Kevin, Hi Arnd,
>> > >
>> > > Please consider these Renesas ARM based SoC DT Timers updates for v3.17.
>> > >
>> > > This pull request is based on a merge of the following to provide
>> > > all dependencies and try to eliminate conflicts. It turns out the changes
>> > > in this pull requests are a nexus for dependencies due to modifying DT,
>> > > SoC, board and recently moved header files as well as requiring driver
>> > > changes.
>> > >
>> > > * The clockevents/renesas-timers-dt branch of Daniel Lezcano's tree.
>> > >   He has indicated that this branch has stable commit ids and will
>> > >   be included in v3.17. Olof and arm at kernel.org were CCed on the
>> > >   thread where he, Laurent Pinchart and I discussed the use of that branch.
>> > >
>> > >   The clockevents/renesas-timers-dt's branch is in turn based on v3.16-rc3.
>> > >
>> > > * "Third Round of Renesas ARM Based SoC DT Updates for v3.17",
>> > >   tagged as renesas-dt3-for-v3.17, which I have sent a pull request for.
>> > >
>> > > * "Renesas ARM Based SoC Clock Updates for v3.17",
>> > >   tagged as renesas-clock-for-v3.17, which you have merged
>> > >   into next/soc
>> > >
>> > > * "Second Round of Renesas ARM Based SoC soc-cleanup Updates for v3.17",
>> > >   tagged as renesas-soc-cleanup2-for-v3.17, which you have merged
>> > >   into next/cleanup.
>> > >
>> > > * "Third Round of Renesas ARM Based SoC r8a7779 Multiplatform Updates for
>> > >   v3.17", tagged as renesas-r8a7779-multiplatform3-for-v3.17, which
>> > >   you have merged into next/soc
>> > >
>> > > * "Renesas ARM Based SoC Boards Updates for v3.17",
>> > >   tagged as renesas-boards-for-v3.17, which you have merged
>> > >   into next/boards
>> > >
>> > > * "Third Round of Renesas ARM Based SoC Updates for v3.17",
>> > >   tagged as renesas-soc3-for-v3.17, which you have merged
>> > >   into next/soc
>> > >
>> > >
>> > > The following changes since commit 5c174afd407acc7a90701900b279578151bc007f:
>> > >
>> > >   Merge branch 'clockevents/renesas-timers-dt' of git://git.linaro.org/people/daniel.lezcano/linux into dt-timers-for-v3.17.base (2014-07-15 16:31:45 +0900)
>> > >
>> > > are available in the git repository at:
>> > >
>> > >
>> > >   git://git.kernel.org/pub/scm/linux/kernel/git/horms/renesas.git tags/renesas-dt-timers-for-v3.17
>> > >
>> > > for you to fetch changes up to 9394af4314554d15762585a3464cefaa2e6d0420:
>> > >
>> > >   ARM: shmobile: genmai-reference: Enable MTU2 in device tree (2014-07-15 21:26:42 +0900)
>> > >
>> > > ----------------------------------------------------------------
>> > > Renesas ARM Based SoC DT Timers Updates for v3.17
>> > >
>> > > * Enable timers using DT when booting boards without Legacy-C code
>> > >
>> > > ----------------------------------------------------------------
>> > > Laurent Pinchart (8):
>> > >       ARM: shmobile: r8a7790: Add CMT devices to DT
>> > >       ARM: shmobile: r8a7791: Add CMT devices to DT
>> > >       ARM: shmobile: r8a7779: Add TMU devices to DT
>> > >       ARM: shmobile: lager-reference: Enable CMT0 in device tree
>> > >       ARM: shmobile: koelsch-reference: Enable CMT0 in device tree
>> > >       ARM: shmobile: marzen-reference: Enable TMU0 in device tree
>> > >       ARM: shmobile: r7s72100: Add MTU2 device to DT
>> > >       ARM: shmobile: genmai-reference: Enable MTU2 in device tree
>> >
>> > Ok, this branch definitely contains a lot more than this. For dependent
>> > external branches such as clocksource, we still prefer to see a pull request so
>> > that we can merge in the dependency and get a clean diffstat when we do the
>> > merge of your branch, otherwise it gets awkward to compare that what we're
>> > getting is what you thought you sent (which is one of the things we check on
>> > merges).
>> >
>> > Please regenerate this pull request as appropriate.
>>
>> Hi Olof,
>>
>> FWIW, I believe that that the diffstat between
>> 5c174afd407acc7a90701900b279578151bc007f and
>> 9394af4314554d15762585a3464cefaa2e6d0420 is what was included in the
>> pull-request. But I guess that all the merged-in branches are hampering
>> your verification process.
>>
>> Would it help if things were arranged as follows?
>>
>> 1. Use the clocksource branch as a base and then;
>> 2. Merge in each of my branches (the ones listed above) and then;
>> 3. Add the patches on top

So there's nothing wrong per se with the way you arranged it, even
though it is more convenient for us from a review perspective to get
merges of branches at the tips of previous branch heads (i.e.
tags/merge requests). It just makes it easier to spot "Oh, that side
of the merge is from this branch that we've already reviewed", etc.

For generating pull requests for these complex merges, what we do when
sending stuff upstream is that we generate a dummy merge and
"manually" (through a script) generate the diffstat and shortlog from
that. Git can get confused about what is actually the merge-base
otherwise, which is what happened in this case for you.


-Olof



More information about the linux-arm-kernel mailing list