Could people with scsi tapes please try the following:
dd if=/dev/urandom of=testfile bs=1024 count=20000
tar cvf /dev/st0 testfile
mv testfile testfile.good
tar xvf /dev/st0
diff testfile.good testfile
I always get different files and I want to know if its somewhere in my
hardware or in the scsi-tape support.
My setup is:
- asus PCI/I-486SP3 motherboard with ncr53c810 on board
- hp C1553A DAT tape (Rev: 9503)
- conner cfp 1060s (with patched firmware)
- both, the original linux driver and the bsd-driver port show this error
- scsi tape as module
I doubt that the tape is defect, because (as far as I know) DATs always check
the data after it is written, so I assume that the data is already wrong
when it reaches the tape. The scsi bus seems reliable to me because the
harddisc works fine and the bsd ncrdriver uses parity (and I got no parity
errors so far). The conclusion is that there is a bug somewhere in the
linux scsi-tape code, but I'm not sure.
Matthias
O .---------------. .___________. O
/\/ . `. m_sattle@ ,' / \ +FAX . \/\
__..--- ' /\/ | `._________,' | (___)/ * * \(___) \/ \ ` ---..__
""---__ \/`. | informatik. | / | \ +49 (0)6333 ,'\/ __---""
`.. / | .uni-kl.de | | `...' | -65079 \ ...'
`---------------' `._____.'
--> Don't take life too seriously -- you'll never get out of it alive. <--