The need to re-detect usb-storage devices (on resume and on-the-fly plugging)

[Michel Salim]

2007/3/23, David Zeuthen <david at fubar.dk>:
> On Fri, 2007-03-23 at 14:22 -0400, David Zeuthen wrote:
> > >  Right
> > > now, the laptop would resume with the device node still there (e.g.
> > > sdb, sdb1), and the device still "mounted" but not accessible, and a
> > > new device node is created (sdc, sdc1).
> >
> > That sounds like another bug; it works for me fwiw.
>
> Actually, I must admit, this doesn't work for me - I just tried. Upon
> resume (from from disk and from RAM), the kernel emits uevents to remove
> the devices, then adds them again. I don't see other device nodes being
> used though - my guess is that you have open files on the drive when you
> tested this? Hmm.. I can only guess that you didn't mount the partitions
> through HAL (using gnome-mount or GNOME) because if you did, we would
> have lazy unmounted the file system when we saw the device node going
> away...
>
Yup; I see extra nodes being created only I manually mounted the
removable partition in the first place (e.g. when I forgot to plug the
device at boot)

> Anyway, I submit that this is a bug with the kernel... although I'm sure
> some people would argue this is how it works (since bus enum and sending
> out deltas is *hard*) and that it's not a bug. Anyway, I filed this bug
>
>  https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=233697
>
Will get Cc:ed on it. Thanks!

-- 
Michel