On Fri, Feb 23, 2024 at 12:15 PM Konstantin Aladyshev <aladyshe...@gmail.com> wrote: > > Should I see the new git version in the linux-yocto devshell? > """ > bitbake linux-yocto -c devshell > """ > > I've tried to check it in the current poky master, and it gives me: > """ > $/data/poky/build/workspace/sources/linux-yocto# git --version > git version 2.30.2 > $/data/poky/build/workspace/sources/linux-yocto# which git > /data/poky/scripts/git > """ > So in this case host git is used. "bitbake kern-tools-native -c > devshell" gives the same result. > Can this be considered as a proof that git-native is not used? >
It is likely ASSUME_PROVIDED that is causing the issue, git-native is in there, which means it is going to use the host version and not build our -native recipe for it. We'll need some guidance from Richard on this, because the use of that flag in the git operations is in fact required for security reasons, so it isn't going to be optional. If that means we need a minimum git version or to use our own built one (and possibly the buildtools tarball, etc), then those are the hoops we'll have to jump through. When I pulled git-native out of ASSUME_PROVIDED locally, I ended up in a dependency loop. I can have a closer look at this again on Monday, but i'm out of time for it today. Bruce Bruce > Konstantin > > On Fri, Feb 23, 2024 at 4:11 PM Bruce Ashfield <bruce.ashfi...@gmail.com> > wrote: > > > > On Fri, Feb 23, 2024 at 6:43 AM Alexander Kanavin > > <alex.kana...@gmail.com> wrote: > > > > > > On Fri, 23 Feb 2024 at 12:32, Konstantin Aladyshev > > > <aladyshe...@gmail.com> wrote: > > > > Thanks for the response! What would be a proper way to fix this problem > > > > then? > > > > > > 1. show how to reproduce the issue in plain poky master. > > > 2. check why none of the tests expose the issue. You can for example > > > grep related keywords in meta/lib/oeqa/selftest/cases. I think there > > > are tests that check devtool kernel operations there. > > > 3. see if existing tests can be easily extended to cover it, so that > > > they fail without a fix. > > > 4. perhaps the fix would be to check git version first, and add the > > > argument only if the version is recent enough. > > > > Which unfortunately would still leave the problem that the change is > > intended to fix, so I wouldn't want to merge any patches to make it > > conditional in the kern-tools. > > > > Those changes were made to prevent git operations from running > > hooks from the host environment, for both security and consistency > > reasons. > > > > We added git-native to the kern-tools-native recipe to get a consistently > > new version available for any operations involving the tools. > > > > So I'd add another question .. why is this operation not using our > > natively built git ? Is there something going on with the path, environment > > or something else that we missed when adding that native dependency. > > > > Bruce > > > > > > > > Alex > > > > > > > > > > > > > > > -- > > - Thou shalt not follow the NULL pointer, for chaos and madness await > > thee at its end > > - "Use the force Harry" - Gandalf, Star Trek II -- - Thou shalt not follow the NULL pointer, for chaos and madness await thee at its end - "Use the force Harry" - Gandalf, Star Trek II
-=-=-=-=-=-=-=-=-=-=-=- Links: You receive all messages sent to this group. View/Reply Online (#62586): https://lists.yoctoproject.org/g/yocto/message/62586 Mute This Topic: https://lists.yoctoproject.org/mt/104509876/21656 Group Owner: yocto+ow...@lists.yoctoproject.org Unsubscribe: https://lists.yoctoproject.org/g/yocto/unsub [arch...@mail-archive.com] -=-=-=-=-=-=-=-=-=-=-=-
