Lennart Poettering
Lennart Poettering
Hmm, not sure I follow. If /efi/ is an automount then sysupdate should just trigger it and things should just work? or are you saying sysupdate for some reason doesn't...
I wonder if we might need something like 506c1bb594543f6fcb232048e76e60e2caddee86 somewhere in sysupdate
@mcassaniti do still have the precise .transfer file that triggered this?
(it might also make sense to add an explicit concept for setting fcaps to tmpfiles.d/)
We nowadays support a "~" modifier to deoce the 6th colum as base64. But we do not support it for setting xattrs so far. Would be trivial to add now,...
hmm, there are some other issues i think need addressing in that original PR, to keep track of that i filed #33413
So I think adding a knob for this might be OK. But: I think I agree with @prajnoha somewhat that the *correct* fix is to avoid these clashes, or to...
> One thing I keep wondering about now: the existing behaviour takes device properties into account (i.e. the link prio is configured by device), and if that doesn't help it...
> I'm afraid it doesn't make most sense, because it's not how it ever worked, and as proven again and again, changing it breaks loads of existing udev rules out...