Hi, On Fri, Oct 13, 2017 at 01:53:13PM +0200, Alexander Dahl wrote: > I'm currently investigating a minor annoying issue. > > For our BSP we set PTXCONF_PLATFORM to something like 'at91foo'. Recently we > added some kernel patches (we didn't have those before when using a pure > vanilla kernel, this is no option anymore) and we put those to $ > (PTXDIST_WORKSPACE)/patches (where our other patches are) and created a file > 'linux-4.9.47/series.at91foo' in the subdirectory matching the currently used > kernel version. This works so far, build is successful, image is created and > runs on target. > > To those patches: we track our patches in a separate Git repository and use > it > as a Git submodule in more than one BSP, not below the > PTXDIST_PLATFORMCONFIGDIR (where the platform specific linux patches maybe > should go), but below PTXDIST_WORKSPACE, because it's a shared patches folder > anyway. Maybe also not the best decision, but it's like that currently. > > The linux kernel patch stack was created with 'git ptx-patches', but on a > clean build I deactivated PTXCONF_SETUP_PATCHIN_GIT so patches are applied > without git, this saves quite some time on extract stage. > > Now after a `ptxdist clean` and a `ptxdist prepare kernel` a new 'series' > file > appears in $(PTXDIST_WORKSPACE)/patches/linux-4.9.47 and it's only named > 'series' and put besides the already present 'series.at91foo' which is > tracked > in Git. In the patches folder Git lists this as untracked file. In the parent > folder, which is PTXDIST_WORKSPACE aka the BSP using the patches folder as a > Git submodule and tracked in a different Git repo, on `git status` I see this: > > modified: patches (untracked content) > > And with `git describe --tags --dirty` this: > > v2017.03.0-275-gd8e18cb-dirty > > We use this somewhere in our firmware. Now on a clean build, we must abort in > between, remove this additional series file, and proceed the build to get a > clean version string without '-dirty' added. :-( > > From my intuition I would say ptxdist should not create this additional > series > file, or delete it again, if it's just temporary. I tried to find the spot in > ptxdist sources, but up to now with no luck. > > Is this a bug in ptxdist? Should we organize our BSPs and patches somehow > different to not run into this? Are there any other possibilities or > suggestions? This behaviour is problematic for fully automatic builds. :-/
The only place ptxdist creates a series file, is, when it cannot find one (This is in scripts/lib/ptxd_make_world_patchin.sh line 282). In that case, you should see a warning, when it is created. Maybe PTXCONF_KERNEL_SERIES is not set correctly? Michael -- Pengutronix e.K. | | Industrial Linux Solutions | http://www.pengutronix.de/ | Peiner Str. 6-8, 31137 Hildesheim, Germany | Phone: +49-5121-206917-0 | Amtsgericht Hildesheim, HRA 2686 | Fax: +49-5121-206917-5555 | _______________________________________________ ptxdist mailing list [email protected]
