> Hi,
>
> On Mon, 26 Apr 1999 08:45:27 -0400 (EDT), "Richard B. Johnson"
> <root@chaos.analogic.com> said:
>
> > Once I tried to put a CD-ROM on IDE on this machine. A disaster
> > occurred when I tried to copy some files from the CD-ROM to a SCSI-
> > Disk. The disaster was so complete that there were not any "known"
> > file-systems left on the machine. The contents of all four SCSI
> > disks were completely trashed, having been overwritten with a
> > pattern of "|||||||||||||" (forever).
>
> > I suggest you do not use an IDE disk with SCSI. There seems to
> > be something about IDE that doesn't "like" SCSI boards.
>
> There's no such problem with scsi, as thousands upon thousands of
> Linux users can testify. Installing from an IDE cdrom to local scsi
> is something I do regularly. There may be hw problems with specific
> motherboards or controllers, but in principle there is absolutely no
> fundamental incompatibility between IDE and scsi.
>
The fundamental problem between using the IDE/EIDE device and a SCSI
device is that many IDE device/controller combinations corrupt the
bus. This is particularly evident when using motherboards that have
both built-in SCSI (aic7xxx), and IDE such as this "Thunder board".
Attempts to copy from an IDE to/from the SCSI may (read will) result
in extreme data corruption. In other words, so evident that there will
be no possibility of thinking that it's just casual.
We attempted to find out why and determined that the IDE interface
just qualifies an address and R/W and slams its data on the bus
ignoring any timing specifications whatsoever. Of course the last
device to write data wins, but the currents that flow when two
devices are writing opposite-polarity bits causes ringing when
the deselected device stops writing. These ring-down glitches _are_
data. Notgood(tm).
Cheers,
Dick Johnson
***** FILE SYSTEM WAS MODIFIED *****
Penguin : Linux version 2.2.6 on an i686 machine (400.59 BogoMips).
Warning : It's hard to remain at the trailing edge of technology.
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/