Hi, Christian.
On Dec 14 2009, Cristian Ionescu-Idbohrn wrote:
> Bought today two identical usb-sticks and made some observations.
>
> 1. not related to the two usb-sticks being identical:
>
> * some env variables are already exported by udev:
>
> ID_BUS=usb
(...)
> scripts, so re-exporting:
>
> + export UM_FILESYSTEM=vfat
> same as ID_FS_TYPE=vfat
> + export UM_VENDOR=Ut165
> same as ID_VENDOR=Ut165
> + export 'UM_MODEL=USB Flash Disk'
> (almost) same as ID_MODEL=USB_Flash_Disk
>
> shouldn't be needed, and would simplify and improve performance.
Great. I wonder if ID_MODEL is guaranteed to contain no spaces in the
name. That would also make things more comfortable.
> 2. related to the two usb-sticks being identical:
>
> * they are in fact so identical, that the only differences lay in:
>
> ID_PATH=pci-0000:00:1d.7-usb-0:3.4:1.0-scsi-0:0:0:0
> ID_SERIAL=Ut165_USB_Flash_Disk_00000000000E7F-0:0
> ID_SERIAL_SHORT=00000000000E7F
I guess that the only thing that we could count on (and I'm not even
sure if we could really count on that) is the serial number.
> 3. the next thing is the mounted device is (for example) /dev/sdc instead
> of the expected /dev/sdc1:
>
> $ cat /proc/mounts
> /dev/sdc /media/usb0 vfat
> rw,nodev,noexec,noatime,nodiratime,gid=25,fmask=0117,dmask=0007,allow_utime=0020,codepage=cp437,iocharset=utf8,errors=remount-ro
> 0 0
(...)
I had not seen this before. I do have one stick that is mounted as
/dev/sdc, but it doesn't have a partition table.
> * fsck reports:
>
> # fsck.vfat /dev/sdd
> dosfsck 3.0.6, 04 Oct 2009, FAT32, LFN
> /dev/sdd: 1 files, 2606/61664 clusters
>
> now, this could be a problem with udev, the kernel, or elsewhere.
(Just for the record, you meant /dev/sdc here, right? In the other
places, you used /dev/sdc and here you used /dev/sdd---just trying to
check if you're not seeing different devices).
That being said, I don't know about the ramification of these problems.
Do you have any "real" problems with that? I mean, like data written on
the wrong place?
> 4. the created symlinks under /var/run/usbmount for the two sticks are
> identical (/var/run/usbmount/Ut165_USB_Flash_Disk), at least with my
> modified scripts, and that leeds to confusion
>
> I guess the easiest way to reproduce this is to 'dd' one stick to the
> other, but I didn't test that.
>
> May be so that using ID_SERIAL_SHORT would be a partial workaround, but I
> didn't try that either, yet.
Yes, that could be a solution. I don't really use the /var/run/ symlinks
myself, but updating them is prone to introducing all sorts of race
conditions.
Anyway, I'm thinking of releasing the new version of usbmount as-is from
the SVN repository, since it contains a versioned dependency on udev
that is necessary so that we don't break things (IOW, this must be
pushed as soon as possible).
If you have any small patch, please let me know. I'm a little bit short
on time and, if you would like, you could be a co-maintainer of the
package.
Regards, Rogério Brito.
--
Rogério Brito : rbr...@{ime.usp.br,gmail.com} : GPG key 1024D/7C2CAEB8
http://rb.doesntexist.org : Packages for LaTeX : algorithms.berlios.de
DebianQA: http://qa.debian.org/developer.php?login=rbrito%40ime.usp.br
--
To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
