[systemd-devel] systemd shutdown vs ostree

[Lennart Poettering]

On Sun, 15.09.13 17:26, Koen Kooi (koen at dominion.thruhere.net) wrote:

> > On Sat, Sep 14, 2013 at 10:52 AM, Koen Kooi <koen at dominion.thruhere.net> wrote:
> >> Please keep in mind that pstore is x86 only. EFI isn't x86 only, but exceedingly rare outside of that arch.
> > 
> > What? Pstore itself isn't. It's a generic interface to persistent
> > platform storage. There are backends using EFI variables, NVRAM
> > (PowerPC) or the APEI ERST (x86). As a last resort there's also a
> > platform-independent RAM backend which may at least survive a simple
> > reboot.
> 
> OK, so x86 and ppc only. That will leaves out arm, mips and others.

I am pretty sure that it would be relatively easy to add support for an
addition ARM backend to pstore if such functionality is available on ARM
systems. The thing is simply that we should support pstore where it is
available, and get people to use that as the one and only API for these
kinds of things.

That all said, currently we cannot use pstore anyway, since there is no
way to tag the pstore data with the machine identity the data is
from. Which means that if you have a suse/Fedora dual-boot system there
is no way how we could make sure that the Suse instance doesn't end up
with Fedora's pstore logs in the journal and vice versa. 

To make pstore workable for us we'd need some scheme how we could echo
the machine ID into some virtual kernel file and then be sure that the
kernel adds that into the pstore data if something happens.

Lennart

-- 
Lennart Poettering - Red Hat, Inc.