mayer@cooper.cooper.EDU (Mayer Ilovitz ) (01/19/90)
Hi, The SCSI tape controllers used on our Tower 650 has a register that determines whether or not the bytes are swapped by the controller. By changing the value of this register you can avoid having to do a dd conv=swab. In Release 2.01.00 there is an undocumented command which can change this register's value. The command is /menu/obj/mtswap.o TAPEDEVICE -set VALUE TAPEDEVICE is the device name of your tape drive VALUE is: 0 for no swapping 256 for swapping There are two restrictions to this command; you must be root to execute it and you must have a tape in the drive for the command to work. On our system, the 150M drive is set to swab and the 45M drive is set to no swab. Mayer Ilovitz mayer@cooper.cooper.edu cmcl2!cooper!mayer phri!cooper!mayer mstan!cooper!mayer hombre!cooper!mayer
roe@sobmips.UUCP (r.peterson) (01/23/90)
From article <2053@cooper.cooper.EDU>, by mayer@cooper.cooper.EDU (Mayer Ilovitz ): > Hi, > > The SCSI tape controllers used on our Tower 650 has a register > that determines whether or not the bytes are swapped by the controller. This is incorrect. It implies that the controller has byte-swap capability _in hardware_. There are several problems with this - the "three-byte block" is the least of them. The hardware does not have this capability. > By changing the value of this register you can avoid having to do a > dd conv=swab. In Release 2.01.00 there is an undocumented command which can > change this register's value. The command is > > /menu/obj/mtswap.o TAPEDEVICE -set VALUE > What this command does is set an external device-driver variable that convinces the driver to swap the bytes for you. I realize the point may be moot - the user process sees the bytes in the correct order. However, you will pay a performance penalty. Don't expect things to run as fast, *especially* on write, as they might otherwise. -- One makes strong assumptions delving Roe Peterson into the beginning of the universe... {uunet,mcgill-vision}!sobeco!roe - Stephen Hawking, Cambridge