Did the bigger buffers improve performance and lessen
the following problems?
> There are some disquieting symptoms, however, which I
> don't know whether to attribute to SANE, sg, the SCSI card,
> or the scanner, and on which perhaps someone can shed some light.
>
> Symptoms:
>
> in all configurations, interrupts seem to be disabled, or the system
> seems to hang, for inordinate periods (several seconds) during a scan,
> especially when the scan is first initiated, but also during each scan
> segment.
My advansys 5140 (ISA aha1540 clone) slows anything else on the
same SCSI bus cause my UMAX 1220S doesn't support 'disconnects',
but the rest of my system goes ok (eg my advansys 940UW with my
root file system on it).
> At least some configurations of the system can be made to crash
> by trying other activity during a scan. I have been reluctant to gather
> too many statistics about just what activies cause system crashes however,
> but small scans seem typically to complete OK.
>
> Running xtops during a scan shows cpu usage 99%idle after starting
> xscanimage, but before a scan is initiated, at 75% system, 25% user
> during the scan, and continuing at 75/25 even after the scan has completed,
> only returning to 99% idle when xscanimage is exited.
Time taken to DMA/block_move data to and from the SCSI device
and any polling that the aha152x driver does will _not_
show on a command prefixed with "time ".
> Is this behavior normal for SANE, or this card or scanner, or do I just
> have something configured wrong.
Does 'cat /proc/interrupts' show a reasonable count
of interrupts on the aha152x driver?
> The scanner is the only device on the SCSI bus, which is (at least
> nominally) properly terminated, and which has worked fine for DISKS and
> CDROMS.
>
> The aha152x driver for the SCSI card is slightly ersatz on 2.2.10,
> however, since that version doesn't recognize the 1520B/1522B, I added
> the recognition entry from the 2.0.37 driver to the 2.2.10 driver, which
> seems to make it work fine otherwise.
Perhaps you could post that info/patch through to
the aha152x maintainer who is
Juergen Fischer <fischer@et-inf.fho-emden.de>.
> The card is configured with aha152x=0x340,11,7,1 in LILO, with sg, etc. as
> modules. Behavior was similar with aha152x as a module.
>
> I think this card has FIFO polled data transfer, which if properly implemented
> should avoid the symptoms noted above, so I suspect the driver/card interface.
>
> The configuration file mustek.conf is:
>
> option strip-height 1
> scsi MUSTEK
>
> /proc/scsi/scsi is:
>
> Attached devices:
> Host: scsi0 Channel: 00 Id: 05 Lun: 00
> Vendor: MUSTEK Model: MFS-06000CX Rev: 2.20
> Type: Scanner ANSI SCSI revision: ffffffff
Interesting SCSI version ...
Doug Gilbert
-- Source code, list archive, and docs: http://www.mostang.com/sane/ To unsubscribe: echo unsubscribe sane-devel | mail majordomo@mostang.com