On 20 April 2016 at 10:20, Lennart Poettering <lenn...@poettering.net> wrote:
> On Wed, 20.04.16 08:00, James Hogarth (james.hoga...@gmail.com) wrote: > > > Regardless of how the hibernate generator might potentially be changed to > > attempt to locate a valid swap partition with hibernate data, or how > > difficult that may be given the nature of running in an initrd and as a > > generator, we still need to follow the kernel requirements for > > suspend-to-disk. > > > > The documented way there is resume= so I feel that's the best way to > direct > > the Fedora bug as it stands. > > > > If the anaconda guys disagree then I think the best course of action to > > avoid unexpected data loss is to have upower not default to HybridSleep > but > > go straight to shutdown. > > > > I'll update the bug comments later today with the results of this thread. > > I am still open to add searching for the resume swap partition to the > generator, but only where this may be done reliably, i.e. GPT, but not > MBR. And we should either define a new GPT partition type UUID for > "hibernation swap", or reuse the swap GPT partition type UUID but then > set a flag in it which marks it for use in resume= usage. > > Given that Windows mandates EFI in modern laptops (and thus GPT), I am > pretty sure you can cover large majority of systems with that. But > again, it won't be a 100% solution, since there are of course people > who still use MBR. > > > It seems like it'd be a good idea with the GPT type (though there would still be an issue of someone dualbooting between linux distros of course). There is a large Fedora user base without EFI of course, or with legacy boot enabled even if they have an EFI capable laptop so in terms of the bug itself we'll need a solution that will help them too. James
_______________________________________________ systemd-devel mailing list systemd-devel@lists.freedesktop.org https://lists.freedesktop.org/mailman/listinfo/systemd-devel