[Date Prev][Date Next] [Thread Prev][Thread Next]
[Thread Index]
[Date Index]
[Author Index]
Re: submount?
- From: Alan Cox <alan redhat com>
- To: David Zeuthen <david fubar dk>
- Cc: Development discussions related to Fedora Core <fedora-devel-list redhat com>, alan redhat com, arjanv redhat com
- Subject: Re: submount?
- Date: Mon, 31 May 2004 14:30:00 -0400
On Mon, May 31, 2004 at 12:02:07PM +0200, David Zeuthen wrote:
> I've just put some code in HAL to detect whether the eject button is
> pressed (using MMC-2) so ideally HAL would relay this to gnome-vfs which
> would emit a PreUnmount signal such that apps closes their files and
> then it can do the unmount(1), eject(1) dance. Thing is, I don't know
> how many drives correctly implement this MMC-2 bit, any insight?
My Dad has an HP drive that does. Its the only one I've ever seen
> Maybe it would be smarter to get HAL to use some of the volumagic code
> [1], since I guess a similar technique can be used for partitions on a
> USB drive to give safe removal when the drive is not in use. Would this
> be a good idea? Or are there kernel changes on the horizon to solve all
> of these problems?
Volumagic takes a very different approach to the entire problem. Basically
it looks like this
/mnt/cdrom <---NFS--->volumagic daemon <--mount on demand--> CD iso fs
It uses the unfsd code which has path persistant handles for a given volume so
works even with the CD being mounted and unmounted by the nfs/volumagic daemon.
I wrote it as a hack after Al Viro reckoned it could be done that way just
to evaluate the idea. Really the same logic needs to be more kernel side so
that streaming file I/O bypasses the mount daemon logic.
[Date Prev][Date Next] [Thread Prev][Thread Next]
[Thread Index]
[Date Index]
[Author Index]