Bug#78011: storeio (in)activation (was: Re: CDROM lock)

[Marcus Brinkmann]

Package: hurd
Severity: wishlist

On Wed, Oct 18, 2000 at 01:52:25AM -0400, Roland McGrath wrote:
> In oskit-mach, device_close is a no-op and releasing the last send right to
> the port returned by device_open is what closes the device.  Then there is
> nothing extra to worry about, even in case of untimely death of translator.
> So there.

This is not good enough, IMHO. Alexey Dejneka sent in a patch for gnumach
that does the same, and it works just fine. But after doing:

settrans -ag /cdrom

you have to do

settrans -ag /dev/hd3

(or whatever your CD ROM device is) as well, or the CD ROM remains busy.
I consider this to be a bit inconvenient, and also repetitive. In fact, we
don't handle removeable media well currently (CD ROM, ZIP).
I think that storeio should release such ressources it doesn't need anymore.

Thanks,
Marcus

I wrote (just to log it in the BTS):
> The problem is that the responsibility is not entirely clear to me.
> isofs does use storeio on /dev/hd3, so for CD changes to work, clearing the
> active isofs translator should free the device. This means, a close on an
> open_hook in storeio should set the INACTIVE flag in the store if it is the
> last one. Also, the right thing has to be done when the translator should go
> away, and this even if nosync is true.

-- 
`Rhubarb is no Egyptian god.' Debian http://www.debian.org brinkmd@debian.org
Marcus Brinkmann              GNU    http://www.gnu.org    marcus@gnu.org
Marcus.Brinkmann@ruhr-uni-bochum.de
http://www.marcus-brinkmann.de