Jari Ruusu <jariruusu@xxxxxxxxxxxxxxxxxxxxx> wrote:
> I believe that sets up an environment variable and knoppix init scripts
> use
> that variable to load requested keyboard mapping file.
Since all distros (beside some embedded systems) have a console built-in (kernel CONFIG_VT) I wonder why this console can´t be build with UTF8 support and german key-mapping.
> I hope that you understand that the UUID thing works only on unencrypted
> partitions. For encrypted partitions, the file system superblock is
> encrypted, and probing will always fail.
>
> My personal opinion is that the folks who made that "go with UUID"
> decision,
> made a really bad decision. Probing all devices makes spun-down devices to
> spin-up, and depending on device type, that annoyance can range from minor
> annoyance to "accelerated death of the device" [1]. Normal wear and
> eventual
> failure in normal use is ok, but anyone who thinks that "accelerated death
> of the device" is ok because of fucked-up design, must be out of his mind.
True, I was a bit puzzled at first, but UUID/Label means a string somewhere on unencrypted space. For my usb-boot-stick that would be ok. I have to use 2 initrds since attaching a second SATA drive changes the usb-stick to shift /lib from /dev/sdb1 to /dev/sdc1. UUID/Label might avoid this.
A very important point are spun-down disks! I use to send drives to sleep with "hdparm -y /dev/sdX" to keep the system calm and cool. Some system events awake them which is very annoying. Manufacturers guarantee at least 50.000 spin-ups so do you really mean this would harm the drive? "Polling" is a bad thing, I remember my second blue-sparkling stick from SanDisk. 256 mb and one day the blue light vanished for an RMA. Most likely some SuSE versions polled usb-storage permanently, at least I think this is why it broke.
> Peter,
> If I remember correctly, recent linux kernels include "ub" driver that is
> sort of dumbed-down version of USB. The main advantage is that /dev/ub*
> naming does not "mix" with /dev/sd* devices. Disadvantage may be lower
> performance. Don't know if it works at all, but loop-AES' build-initrd.sh
> script supports /dev/ub* devices. If you try it, please let me know if it
> works.
Yes, the build-initrd.sh script lists /dev/ub* and in /dev on Kubuntu there is plenty more!
usb1-5, usbdev1.1_ep00, usbdev1.2_ep01, and the like.
Honestly, I didn´t see these before and have no glue what they are made for. Issuing a command like:
mount -v /dev/usbdev1.1_ep81 /mnt/
mount: you didn't specify a filesystem type for /dev/usbdev1.1_ep81
I will try all types mentioned in /etc/filesystems or /proc/filesystems
Trying msdos
mount: /dev/usbdev1.1_ep81 is not a block device
does nothing useful. Plugging in an ubs-storage device always results in /dev/sdXX on SuSE and Ubuntu. My recent SuSE kernel even addresses IDE drives with /dev/sdc and changed to /dev/sr0 for DVD-drive. This makes no sense and even reduces available functionality in connection with hdparm.
> [1] Does anyone remember IOMEGA ZIP drives, famous for their
> click-of-death
> syndrome?
Oh yes, and how I remember that! In the late 90s I supposed it would be a good idea to buy one of these, with a SCSI controller card. Never again! Fat hard disk drives from Western Digital are the only good thing. Since Samsungs F1 HD753LJ seems to have CRC errors, the WD5000AACS is my favorite one. Two platters and 5400 rpm will make an end to my bleeding ears.
Kind regards,
Peter
--
GMX FreeMail: 1 GB Postfach, 5 E-Mail-Adressen, 10 Free SMS.
Alle Infos und kostenlose Anmeldung: http://www.gmx.net/de/go/freemail
-
Linux-crypto: cryptography in and on the Linux system
Archive: http://mail.nl.linux.org/linux-crypto/
