Distribution: Debian 3.1 Sarge Hardware Environment: AMD Duron 1100, QDI Kinetiz 7E (KT-133), 256MB SDRAM, Seagate 60GB HDD, TEAC D16G DVDRW, Pinnacle soundcard, Rtl8139 NIC, STB Velocity 128, USB scanner Microtek (unplugged from electricity) Software Environment: Problem Description: I have Paralell port ZIP drive 100 connected to EPP LPT. With recent kernel version, when I try to mount (as user), I get an error "Special device /dev/sda4 does not exist". And now the mystery begins: I do (as root)# mount /dev/sda -t vfat /media/zip I get strange error about not valid filesystem. Very strange because there is no device at /dev/sda either! So I should get an "no such device" error, no just complaint about filesystem! (The filesystem IS vfat on the ZIP however. But as I said, the ZIP drive doesn't sit on /dev/sda anyway, and no other device does) More strange, after that step, I am suddenly able to mount zip (as user) at /dev/sda4 and normally use it! I discovered, that if there is a medium inserted during boot process, problem does not appear. Steps to reproduce:
Created attachment 5475 [details] Here's my custom kernel config I use
I hear change in sound of ZIP! If I don't let the medium in the drive during boot, and then I insert the medium, the drive cliks twice! Anybody who works with ZIP knows the click sound that appears when U insert the medium (probably the startup of motors and movement of heads because of initial medium check or so). So, the sound is doubled somehow when the bug is going to appear. I don't know the structure of the driver, but for me being not kernel hacker it looks like some bad value sits somewhere in kernel variables (SCSI, ATAPI or whatever) when computer boots without medium inserted in ZIP drive.
When I run LILO after ZIP problems, I get "Fatal: open /dev/sda does no exist" or whatever. Then I try to run LILO again, and it is OK. The error happens whenever in the same session (=until reboot)
With 2.6.13.1 the bug seems to dissapear. Is it accidentally or knowlingly? :o) Should it be marked as "Resolved", or better let it be "NEW" some time?
If the bug is still absent from 2.6.15 then yes, please mark this as resolved.
Hmm, in 2.6.15, bug is there.
I bought a new MP3 player Emgeton E5Cult. It serves as USB Mass Storage device too, 1GB, nice machine. Hence I discovered a new kind of behaviour of this bug (kernel 2.6.17.13): I plugged the Player to USB, mounted as /dev/sdb1. Now the ZIP cannot be mounted at usual /dev/sda4, "does not exist" message. $ lsscsi -gv [0:0:6:0] disk IOMEGA ZIP 100 D.09 - /dev/sg0 dir: /sys/devices/platform/host0/target0:0:6/0:0:6:0 [1:0:0:0] disk SigmaTel MSCN 0100 - /dev/sg1 dir: /sys/devices/pci0000:00/0000:00:07.2/usb1/1-2/1-2:1.0/host1/target1:0:0/1:0:0:0 I unmount Player and unplug from USB. ZIP still cannot be mounted at /dev/sda4. $ lsscsi -gv [0:0:6:0] disk IOMEGA ZIP 100 D.09 - /dev/sg0 dir: /sys/devices/platform/host0/target0:0:6/0:0:6:0 And now the obligatory part of the bug's behaviour: I rename the ZIP in /etc/fstab to from /dev/sda4 to /dev/sda and attemp to mount. I get the strange error "wrong fs type". Renamed back to /dev/sda4 and now mounted flawlessly! I could understand if the ZIP just "migrated" because of kinda position re-ordering after the USB device was plugged in. However, the Player plugged to USB is usually assigned to /dev/sdb1 here (with or without the ZIP mounted), and ZIP is _always_ at /dev/sda4. And it seems, that ZIP (sda4) really disappears from /dev directory when player is plugged to USB for the first time. Couldn't also understand, why it is still blocked after unplugging the USB, and why does it recover itself after unsuccessful attemp to mount /dev/sda.
Peter, Please confirm if the problem still exists with new kernels (2.6.22+, recent one was rc7). Have you moved up to newer distro? This can be a user level hot plug issue (udev, hald agents that provide "plug and play" device capabilities). Thanks.
Please reopen this bug if it's still present with kernel 2.6.22.