[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: what is /.udev for ?

Once upon a time Ron Johnson said...
> On Wed, 2005-02-09 at 12:43 -0800, Ben Pfaff wrote:
> > Ron Johnson <ron.l.johnson@cox.net> writes:
> > 
> > > One thing I do know is that traditional apps like df (and anything
> > > that uses stat(), I guess) don't know about /.dev, and so return
> > > false information:
> > > [df output snipped]
> > It's not really false, it's just that /.dev is a subtree of / and
> > so shows the same information as / does.
> To me, reporting the same information 2 times means that one of
> them should not be there. [...]

The information is not the same. Some of it is, some of it isn't.
If you remove the line entirely, you lose information about the bind

It could be argued that df (presumable short for Disk Free) should
somehow identify situations where the same "device" appears multiple
times and show only the one set of usage data, but that is probably
almost impossible to identify.

> "So what?", you say.  Well, data should only be listed once, not
> twice.  gtkdiskfree sums up all total and free disk space, and
> having /.dev in there totally distorts the truth.

That's a problem with gtkdiskfree, not with having /.dev mounted.

The situation also occurs when you have a NFS server exporting a number
of disk trees from the one filesystem to a client that mounts those
multiple trees at different locations in its own filesystem. df (and
presumable gtkdiskfree) will display duplicate filesystem usage

This situation has been around since long before gtkdiskfree existed
(even before linux existed), so if it does not accomodate it, it is a
deficiency in gtkdiskfree, not NFS mounts. Likewise for bind mounts.

Reply to: