Re: Suspend 2 merge: 49/51: Checksumming

[Pavel Machek <pavel@ucw.cz>]

Hi!

> > > If I recall correctly, someone replied that even a read only mount under
> > > one filesystem (XFS? Not sure), would replay the journal, so it wasn't a
> > > goer.
> > 
> > You could always special case the broken one until they fix it... :)
> 
> Mmm. I wonder how much code that would require us to add. I do like the
> idea of not interacting where the answer is obvious :>. I still think,
> however, that interacting when the answer isn't obvious is the right
> thing to do. Take for example the case where we find an image, but the
> device numbers look like they belong to 2.4 and we're a 2.6 kernel. We
> can't read the header (we can't be sure that this is the cause). The
> user - or their cat - might have selected the wrong boot image
> unintentionally. Why shouldn't we give them the opportunity to reboot
> and get the right one?

Well, kernel depending on user feedback has some interesting issues...
...like user not speaking english or user using speech output.
Thats why pushing "Shall I reboot?" etc prompts into userland
is good idea. (Distros probably will not get it right, either, but at least
they get a chance.)
				Pavel
-- 
64 bytes from 195.113.31.123: icmp_seq=28 ttl=51 time=448769.1 ms