lkml.org 
[lkml]   [1996]   [Sep]   [9]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectInteresting SCSI messages
Yesterday I was messing around with a large file (just over 0.5 Gb - a CDROM
image) and giving my scsi subsystem a thorough workout. While copying the file
from one disk to another and back again I got lots of messages on the console
of the form

Warning - running *really* short on DMA buffers

From what I read of the source this means that I could have potentially done
a larger request in one hit but ran out of DMA memory - harmless in terms of
safety, but potentially damaging to performance.

Also, just once while I was trying to do some other stuff as well while doing
these large copies, I got

scsi0 : DANGER : abort_connected() called

which sounds rather more serious. However, the resulting copy was identical
with the source image so it didn't appear to cause any damage.

Should the first message appear at the default debug level, or should it be
relegated to, say, info or debug status? And what does the DANGER message
actually mean?

Oh, this is using the 53c7,8xx driver (Drew's, not the BSD one) on a standard
810 board, with one drive being a Quantum Fireball and the other (the target
when the DANGER message appeared) being an old Seagate ST41650 (which is a full
height 5.25" drive from 1992, which still manages 2.6MB/s sustained). I did
not have 10MHz mode enabled (lots of external cables). Doing anything at all
on the system while the copy was running seemed to provoke lots of the Warning
messages, which probably makes sense (more outstanding commands etc).

Just my 2 cents - this is a stable kernel and we don't want to alarm the
punters too much now do we :-)

David

\
 
 \ /
  Last update: 2005-03-22 13:37    [W:0.059 / U:0.020 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site