From mboxrd@z Thu Jan 1 00:00:00 1970 From: Arnd Bergmann Subject: Re: [GIT PULL 1/3] ARM: tegra: core SoC code changes for 3.18 Date: Thu, 25 Sep 2014 19:55:46 +0200 Message-ID: <5666206.hBBQzilaqm@wuerfel> References: <1411062692-22338-1-git-send-email-swarren@wwwdotorg.org> <201409251801.18466.arnd@arndb.de> <5424458E.7020603@wwwdotorg.org> Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7Bit Return-path: In-Reply-To: <5424458E.7020603-3lzwWm7+Weoh9ZMKESR00Q@public.gmane.org> Sender: linux-tegra-owner-u79uwXL29TY76Z2rM5mHXA@public.gmane.org To: linux-arm-kernel-IAPFreCvJWM7uuMidbF8XUB+6BGkLq7r@public.gmane.org Cc: Stephen Warren , linux-tegra-u79uwXL29TY76Z2rM5mHXA@public.gmane.org, arm-DgEjT+Ai2ygdnm+yROfE0A@public.gmane.org List-Id: linux-tegra@vger.kernel.org On Thursday 25 September 2014 10:40:46 Stephen Warren wrote: > > I don't recall the exact commits, but I ended up needing to base one of > the branches on -rc2. So, I then based all the branches on the same > commit. When I create Tegra's for-next, I start with that baseline > branch and merge in each Tegra topic branch. That way, the merges only > bring in exactly what's in that branch, and it's easy to validate the > final result in gitk/similar. I see. For us it's actually more helpful if every single topic branch is on the oldest possible state. > I suppose it'd work fine if I start with > the newest base of any branch and then merged the topic branches in; > that would at least restrict the merges to only the contents of the > branch, although it'd create a far more difficult to validate history, > since the different topic branches would be based in different places. Yes, I fear there is no perfect solution to this. Someone also suggested switching around the parent commits during a merge that updates one of our branches to a newer base. That would solve the backmerge problem but in turn break 'git log --first-parent'. Arnd