Hi Greg, On 13/10/2021 18:48, Greg Kroah-Hartman wrote: > On Wed, Oct 13, 2021 at 04:47:55PM +0200, Matthieu Baerts wrote: >> Hi Greg, Sasha, >> >> First, thank you for your great job maintaining the stable versions! >> >> In our work related to MPTCP, we were wondering if we should/can add the >> "Fixes" tag for trivial/stable fixes. >> >> It is certainly easier to explain that with an example: we have a small >> patch [1] to stop exposing a function that is only used from one .c file >> and declared there too. So the signature is removed from the .h file and >> the 'static' keyword is added in the .c file. It should have been like >> that since the introduction of this function. >> >> We don't know if we can/should add the "Fixes" tag for such cases: the >> "mistake" has been introduced by one specific commit so we could add the >> "Fixes" tag but we also know patches with such tags are certainly going >> to be automatically backported. The patch is not really fixing a bug, >> more a "cleaning". Does it make sense to backport these patches then? >> >> On one hand, we might think it would be interesting to backport it to >> reduce the differences with the last version: if the idea is to backport >> simple fixes to ease future and maybe more complex backports later. But >> on the other hand, it is more work for you to backport it: if the idea >> is to backport only actual bug-fix patches. So what is the preferred policy? >> >> We didn't find anything in the doc on "when not to add the 'Fixes' tag" >> but we know the Stable Kernel Rules doc mentions to avoid trivial fixes. >> Maybe this patch is not "trivial", it is not really a bug-fix either but >> that's not the real question here, more: does this rule -- and other >> ones from Stable Kernel doc -- apply to the "Fixes" tag as well? > > Please use the Fixes: tag whenever you want to. Having it there does > NOT mean the patch is automatically backported to stable releases, we > look at them all and choose if they are valid or not. Wow, that's a lot of patches to read! :) > There are lots of tiny "Fixes:" commits that we do not backport for > obvious reasons that they do not fit the stable kernel requirements. > > If you know a patch should be backported to a stable tree, then put the > cc: stable on it, as documented. Again, "Fixes:" is no guarantee that > a commit will be backported at all. Thank you for your reply, that's very clear! Cheers, Matt -- Tessares | Belgium | Hybrid Access Solutions www.tessares.net
