As 2.6.12 draws near (hopefully), I'd like again to solicit people's
opinion about inclusion of FUSE into mainline in the next cycle.
I'm asking now, and not when 2.6.12 is already released, because last
time there was a big rush of reviews and complaints, and by the time
things quieted down it was a bit too late. Thanks to everybody
involved BTW :)
So, if anybody still got a problem with the current version (as in -mm
or released as 2.3-rc1 on SF.net), please speak up now.
Christoph?
Thanks,
Miklos
On Thu, May 26, 2005 at 01:46:27PM +0200, Miklos Szeredi wrote:
> As 2.6.12 draws near (hopefully), I'd like again to solicit people's
> opinion about inclusion of FUSE into mainline in the next cycle.
>
> I'm asking now, and not when 2.6.12 is already released, because last
> time there was a big rush of reviews and complaints, and by the time
> things quieted down it was a bit too late. Thanks to everybody
> involved BTW :)
>
> So, if anybody still got a problem with the current version (as in -mm
> or released as 2.3-rc1 on SF.net), please speak up now.
FUSE_ALLOW_OTHER and FUSE_DEFAULT_PERMISSIONS are still there, and off by
default.
So the same NACK as last time.
Remove them and it's fine to go in.
> > So, if anybody still got a problem with the current version (as in -mm
> > or released as 2.3-rc1 on SF.net), please speak up now.
>
> FUSE_ALLOW_OTHER and FUSE_DEFAULT_PERMISSIONS are still there, and off by
> default.
Yes. The difference between last time lays in the details of the
implementation, which is now well documented in
Documentation/filesystems/fuse.txt, and to which everybody has agreed
to. Or am I mistaken? Do you have any specific objection to the
security measures layed out in there?
Thanks,
Miklos
On Sat, May 28, 2005 at 12:44:54PM +0200, Miklos Szeredi wrote:
> > > So, if anybody still got a problem with the current version (as in -mm
> > > or released as 2.3-rc1 on SF.net), please speak up now.
> >
> > FUSE_ALLOW_OTHER and FUSE_DEFAULT_PERMISSIONS are still there, and off by
> > default.
>
> Yes. The difference between last time lays in the details of the
> implementation, which is now well documented in
> Documentation/filesystems/fuse.txt, and to which everybody has agreed
> to. Or am I mistaken? Do you have any specific objection to the
> security measures layed out in there?
Just because it's documented it's not any better. It's still the horrible
hack it was at the beginning and no amount of discussion or documentation
will change that.
> Just because it's documented it's not any better.
Agreed.
> It's still the horrible hack it was at the beginning and no amount
> of discussion or documentation will change that.
Well, the discussion did help. It helped me understand better why
this is necessary, and not only necessary, but basically the _only_
way to solve this problem. And the documentation is there so that
others can understand too.
Did you actually read it?
Miklos