On Thu, Apr 04, 2024 at 10:54:39PM -0400, Theodore Ts'o wrote: > On Thu, Apr 04, 2024 at 05:56:58PM +0200, Linux regression tracking (Thorsten Leemhuis) wrote: > > > > I know, as I wrote that (as you likely remember). ;-) But it seems it's > > not well known; and maybe making it explicit that this can be used to > > convey a "DoNotBackport" message is supported as well. > > > > Guess I'll prepare a patch to do that then and we'll see how it goes > > from there. > > Maybe something like "ManualBackportOnly"instead? The basic idea is > that it's not that the commit should *never* be backported, but only > with human intervention where someone has specifically requested the > backport, perhaps with qualification test. > > (For example, the XFS file system has an implicit ManualBackportOnly > on all commits, and the XFS stable maintainers are responsible for > backporting identifying commits with Fixes: tags, and running QA > before passing on a request to having them be backported.) For drivers/subsystems/files that no one wants to have backported at all UNLESS there is an explicit cc: stable tag, just email us at stable@vger and let us know and we will add you to the list of files that we ignore for this. We keep that list in the stable-queue repo for anyone to see if they are curious. That's what xfs and bcachefs and kvm and mm and other subsystems have done, that way they can tag things with "Fixes:" to their hearts content and we just ignore them. thanks, greg k-h
