Messages in this thread | | | Date | Mon, 3 Aug 1998 02:10:24 -0400 (EDT) | From | Derrick J Brashear <> | Subject | Re: 2.0 scsi BLIST_SINGLELUN and MBR-7 |
| |
On Fri, 31 Jul 1998, Jon Lewis wrote:
> On Thu, 30 Jul 1998, Derrick J Brashear wrote: > > then cat /cdc/1/somebigfile > /dev/null & > > cat /cdc/2/someotherfile > /dev/null, what happens? in my case, if i kill > > things fast enough i'm ok, but within less than 60 seconds if i let it go, > > the machine's hung > > What's your SCSI card? I just did exactly this, and it's wildly > inefficient...but doesn't crash. Each cat just takes turns. strace and > my ears tell me that cat does just a few reads, then the CD changer > swaps CD's and the other cat does a few reads, then the changer swaps > again...and so on, and so on. Again...this was with a MBR-7, U34f, and > kernel 2.0.31 (too lazy/busy to upgrade my home box).
Doesn't work for me (sorta).
in one window: cat cd6/doors/albums/doors/11.mp3 > /dev/null
in the other: cat cd5/police/albums/every_breath_you_take_the_classics/14.mp3 > /dev/null
both lose: cat: cd5/police/albums/every_breath_you_take_the_classics/14.mp3: I/O error cat: cd6/doors/albums/doors/11.mp3: I/O error
which isn't a crash, but isn't what you see either
both discs become inaccessible from then on until unmounted and remounted.
i augmented scsi.c to notify me when we get to the return NULL case in the single_lun stanza in request_queueable; it wasn't called.
i tried catting a file to /dev/null and simultaneously ls -lR'ing another disc, and that had the behavior you observed, i.e. things taking turns.
I'm going to try a few more things and see if I can again get it to crash, but the I/O error isn't my kettle of tea either.
-D
- 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.altern.org/andrebalsa/doc/lkml-faq.html
| |