On Wed, Aug 09, 2006, EV wrote:
> 2006-08-09: Keith Bennett dixit:
> > [...]
> > That just leaves bind/unbind which is solved by the kernel
> > patch for ejecting the device. As Bob says, this is pretty
> > simple and unintrusive so there is a fair chance that it will
> > be accepted.
>
> Sorry, I have not followd the patch in detail. Can you please
> explain how eject is going to solve bind/unbind (both using shell
> commands and syscals)?
>
> Would it be possible to include full bind/unbind management in
> libkarma? I still see the problem of mount/umount, which would
> need root privilegies...
Okay, so you would still need to do system("mount"), etc.
The mount/umount calls are already suid.
I'm not totally sold on the idea of mount/umount from within libkarma
itself. I realise that this means that the application must be made
aware of the difference between USB and ethernet devices which is not
ideal, but the alternative is also pretty ugly.
Don't take this the wrong way but I *really* dislike the idea of invoking
a daemon for all this stuff. To me it seems an unnecessarily complex
solution to a fairly simple problem. I think I would sooner see those
system() calls in the library than add the dependency of a daemon. All
that it achieves is to move the problem elsewhere.
Of course, this is just my personal opinion. Others may feel differently.
For the benefit of others that may not be aware of the motivation
behind this, the issue is as follows.
Under Windows when using RMM on a USB detached device the RMM application
only grabs control of the karma when it needs to read or write data to
it. The rest of the time, it is left alone so you can leave it plugged
in whilst listening to music and only get interrupted when you are
copying music onto/off of the device.
This is quite convenient and it would be nice if riocp, etc. behaved
the same way. lkarmafs only needs to access the device itself at mount
time and when files are copied, read or deleted. If all you are doing is
browsing the directory structure then the device does not need to be
in storage mode. Again, it would be nice if lkarmafs only grabbed the
device when needed.
The problem is then that mounting/unmounting binding/unbinding the
device all need root privileges.
Keith.
-------------------------------------------------------------------------
Using Tomcat but need to do more? Need to support web services, security?
Get stuff done quickly with pre-integrated technology to make your job easier
Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo
http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642
_______________________________________________
linux-karma-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/linux-karma-devel
