Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > comp.os.linux.hardware > #1534

Re: WD Passport commands

Show all headers | View raw

On Saturday, November 17th, 2012, at 13:25:58h -0500, Haines Brown wrote:
  
> I finally concluded I was unnecessarily beating my head against the wall
> with this Passport (it took 10 hours to reformat it NTFS!)

10 hours sounds excessive -- was this done from GNU/Linux or Windoze
and did you definitely have the disk plugged in via a USB 2.0
socket rather than an old USB 1.0 socket?

Very occasionally plugging in a USB 2.0 disk into a USB 2.0
socket on the PC does not work quite right and the speed is
dropped down to the USB 1.0 level.

The easy way to check is to run dmesg immediately after plugging
in the device and checking to say that dmesg says "new high speed USB device"
and then after a few minute to check again with dmesg to see that there
are no messages indicating some sort of problem on the USB link and the
dreaded message along the lines of dropping speed down.

> Interesting, reformating changes the device UUID.

The correct term is no formatting.  Formatting is a low leve
operation on the disk, and what you are doing with mkfs or
mknfs is *MaKeing* a file system on the specified partion /dev/sdc1
or whatever.

By default, when you invoke mkfs, it generates a new UUID for
the file system on the specified partition.  If you ever juggle
around your disks changing the root / and/or possibly separate
file system for /usr/share and you are using Grub 2, you have
to worry about the UUID, because the Grub 2 configuration file
contains these specific UUIDs.

> I still have a problem I didn't have in earlier kernels, which is that
> when user writes a file to this device, the file ends being owned by
> root, not by user. 

Because you have forced it to be by the use of the uid (and for group, gid)
parameters on the mount command (either command line or in /etc/fstab).

> But I'd just a soon a file written by user keep user's
> ownership.

So do no specify the uid for the mount command.  Consider using
the umask and gmask though if you want to force the files to
be readable or non-readable to others, independent of the 
particular user's umask setting.

> Years ago I think I simply changed owner of the dev file or
> something like that to give user free access to a device, but surely
> that's not right.

It is not a case of "right" but what is appropriate for your particular
machine and usage pattern.  If there is only ever a single user, then
the flexible way is to either use the GiveDevices / TakeDevices script
method related to the display manager login or to look at PolicyKit
for setting permissions per user login.

openSuSE used to use its own system wide script based method (via cron if I
recall correctly) of ensuring the default permissions on important files
and devices, I do not know if they still have that, or if they have moved
on to Policy Kit for everything.

Have a look at  <ttps://wiki.archlinux.org/index.php/PolicyKit>
and in particular 5.2 "Mounting USB Drives"

I suspect that Policy Kit is too much for the type of system which
you are running since its main audience is multi user systems, or
maybe you do in fact have lots of different research workers logging
in to your machines?  ;)

There is also at the desktop level (as opposed to desktop independent Policy Kit
which applies whether the login is a terminal session or X session) the package
called Console Kit which works via dbus.  I think it should be possible to
configure Console Kit to set up permissions on devices and related USB drives 
via its configuration files and an appropriate script.

But Console Kit is going away, so it is best not to pursue that option.
It will be replaced by a feature of systemd, and the mere mention of
that package (good in theory, not so good in implementation) may elicit
some vituperative remarks from others on "why systemd is a bad thing".

Of course for a single default user system you can always hard code the
device ownership and permissions for USB mount points via udev.

So the usual problem on a GNU/Linux system is not a case of being
unable to make something happen, but first choosing the most appropriate
way for the particular machine and the way in which it is used way.

Back to comp.os.linux.hardware | Previous | Next — Previous in thread | Next in thread | Find similar | Unroll thread

Thread

WD Passport commands Haines Brown <haines@histomat.net> - 2012-11-05 23:08 -0500
  Re: WD Passport commands Richard Kettlewell <rjk@greenend.org.uk> - 2012-11-06 08:50 +0000
  Re: WD Passport commands Haines Brown <haines@histomat.net> - 2012-11-16 15:31 -0500
    Re: WD Passport commands Richard Kettlewell <rjk@greenend.org.uk> - 2012-11-16 21:32 +0000
      Re: WD Passport commands Haines Brown <haines@histomat.net> - 2012-11-17 05:57 -0500
        Re: WD Passport commands J G Miller <miller@yoyo.ORG> - 2012-11-17 13:44 +0000
          Re: WD Passport commands Aragorn <stryder@telenet.be.invalid> - 2012-11-17 19:19 +0100
          Re: WD Passport commands Haines Brown <haines@histomat.net> - 2012-11-17 13:25 -0500
            Re: WD Passport commands Richard Kettlewell <rjk@greenend.org.uk> - 2012-11-17 20:30 +0000
              Re: WD Passport commands Haines Brown <haines@histomat.net> - 2012-11-20 05:03 -0500
                Re: WD Passport commands Richard Kettlewell <rjk@greenend.org.uk> - 2012-11-20 10:25 +0000
                Re: WD Passport commands Haines Brown <haines@histomat.net> - 2012-11-20 11:27 -0500
            Re: WD Passport commands J G Miller <miller@yoyo.ORG> - 2012-11-18 18:16 +0000
            Re: WD Passport commands Bill Marcum <bill@nowhere.invalid> - 2012-11-19 03:03 -0500

csiph-web