Re: 2.6.25-rc6: CONFIG_USB_PERSIST forced on

[Pavel Machek]

On Fri 2008-06-06 10:39:19, Alan Stern wrote:
> On Fri, 6 Jun 2008, Pavel Machek wrote:
> 
> > Hi!
> > 
> > In 2.6.26-rc5, 
> > 
> > -config USB_PERSIST
> > -       bool "USB device persistence during system suspend (DANGEROUS)"
> > 
> > is now on. Given that it was previously marked "DANGEROUS", that seems
> > quite a change to me.
> 
> It is.  Basically I did it because Linus asked for it -- or at least,
> for something like it.  He felt that the danger of not having
> USB_PERSIST when you need it is worse than the danger of having it when
> you don't want it.

Well, not-usb-persists means we force-unplug disks during suspend,
after syncing them. Not _too_ bad.

If you unplug disk while hibernated, modify it, and plug it back,
youget _silent_ filesystem corruption. I call that bad.

> > Besides, it seems to have broken usblp hibernation support, and maybe
> > other devices that does not have reset_resume() present. (Big thanks
> > for Oliver for doing investigation).
> 
> URL?  It's hard to see how lack of reset-resume support would mess 
> anything up worse than it would be without USB_PERSIST.

usblp driver sets 'suspended' flag in .suspend(), and resets it in
.resume() -- except that with USB_PERSIST .resume() is not called :-(.

I have novell bugzilla bug number somewhere...

> > [Or is it that now USB_PERSIST is conditional on /sys fs setting, so
> > while setting it on individual device is dangerous, it is still N by
> > default? Changelog does not tell me...?]
> 
> USB_PERSIST is (and has always been) conditional on the power/persist
> sysfs attribute setting.  The difference is that now the setting starts
> out as On whereas before it started out as Off, and there no longer is
> a Kconfig option to remove USB_PERSIST entirely.

I consider the default to be dangerous here.

							Pavel

-- 
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/