[1103] in linux-scsi channel archive
Fujitsu MO M2513A2: advices on switch setting?
daemon@ATHENA.MIT.EDU (Jose Manuel Cerqueira Esteves)
Sun Dec 15 11:51:10 1996
Date: Sun, 15 Dec 1996 16:49:25 +0000 (WET)
From: Jose Manuel Cerqueira Esteves <jmce@titania.cii.fc.ul.pt>
To: linux-scsi@vger.rutgers.edu
cc: Samuel R Martins Eleuterio <samuel@label2.ist.utl.pt>,
Eduardo Lopes <eduardolopes@alf4.cii.fc.ul.pt>
Greetings
I just installed a Fujitsu M2513A2 in a DECpc XL 590, a 90 MHz Pentium
with a NCR53c810 running Linux (for now still Debian 0.93R6, with the good
old 1.2.13 kernel... but to be upgraded soon).
The drive seems to be working fine (for now only with 230MB disks, of course;
probably it would be hard to adapt the existing patches for 2048
bytes/sector to so such an old kernel). But, due to my lack of knowledge
concerning SCSI and Linux drivers, I was not able to find out what would
be the best settings for some of the switches available, and would like to
know if anyone has some advice concerning them:
SW1-4 SCSI data bus parity check (checked [default] / not checked)
Using the default...
SW1-6 Device type mode (optical memory device / direct access device)
"This determines what response the drive will give to an INQUIRY command".
I tried both settings. Both hardware boot messages and Linux kernel
messages reported the type according to what I set... The most notorious
consequence at boot time was that, with the "direct access device" option
set, hardware tests (those of the PC itself) waited for some time (about
half a minute) when there was no disk in the drive (but no errors were
reported). Under Linux, it seemed to me, in a few writing tests of about
30MB of files, that the system was less responsive to other commands while
writing with the "optical device" option set, but since this was not a
rigorous test (and I did not disconnect the PC from the network), I am not
sure if it really meant something. But there were no errors creating
file systems, reading and writing with both settings.
SW2-1 Write cache mode (enabled / disabled [default])
"When enabled, a write error is reported at the completion of the next
command; this can speed up write times but do ensure your system will
respond correctly by knowing what data to resend".
Is it safe to enable write caching? Would it be convenient? (Perhaps the
latter is not obvious... I am asking because of some comments I saw
concerning hardware caching and Linux).
SW2-6 SAVE DATA POINTER message mode
OFF: SAVE DATA POINTER message posted before sending DISCONNECT
ON: No message sent before DISCONNECT message
The default option is the first one and I am using it... without knowing
its meaning.
Many thanks in advance
J. Esteves
