Re: "aborting sbp2 command" expected in kernel.org FireWire maybe

Mark K:

> > dmesg ...
>
> sr0: scsi3-mmc drive: 125x/125x caddy
> Uniform CD-ROM driver Revision: 3.20
> Attached scsi CD-ROM sr0 at scsi1, channel 0, id 0, lun 0
> ...
> Maybe this is very standard but I do not own any peripherals that tell
> me they are SCSI-3...

Sorry, I think that "scsi3" dmesg token isn't literal. Instead that
token comes misleadingly from places like:

http://lxr.linux.no/search?v=2.6.8.1&string=scsi3
http://lxr.linux.no/source/drivers/scsi/sr.c?v=2.6.8.1#L838

I see that dmesg text actually only means a DVD/CD that made mode page
x2A "Capabilities" available. Maybe t10.org didn't get around to
defining that mode page until after they had begun defining SCSI 3 to
supersede SCSI 2. I hear that page became useful at least as CD-R
began to supersede CD-ROM i.e. many many years ago now.

> What's a scsi3-mmc drive?

I think "SCSI" in this context means SCSI over FireWire i.e. normative
Annex B of SBP-2.

I think "MMC" in this context means what it means at
http://t10.org/scsi-3.htm i.e. DVD/CD i.e. SCSI op x12 "INQUIRY" data
byte mask x1F PDT ("peripheral device type") = x05, not one of x 0E 07
04 00 as in HDD.

> Are you having this abort problem with all drives, or wih only this
> drive?

Before now I had not much tried any other FireWire drive, except for my
Apple FireWire Titanium target HDD that died. I'm still trying to
understand what's normal in Linux FireWire. Only after establishing
that some of us feel the need for serialize_io is abnormal have I
turned to borrowing other devices from friends for the sake of
comparison.

Thanks to the help given me here, I now have two reasons to believe
that Knoppix thinks the sbp2 need for serialize_io is and has been
normal since 2.4 thru 2.6, not abnormal.

1) I see the Knoppix 2.6 /etc/modules.conf includes the
in-2.6-misspelled line:

options sbp2 sbp2_serialize_io=1

2) Naturally, I then also see the Knoppix 2.6 boot complain:

$ dmesg | cat -n | egrep -i '1394|sbp2'
    277 ieee1394: Initialized config rom entry `ip1394'
    278 ohci1394: $Rev: 1223 $ Ben Collins
    279 ohci1394: fw-host0: Unexpected PCI resource length of 1000!
    280 ohci1394: fw-host0: OHCI-1394 1.0 (PCI): IRQ=[19]
MMIO=[ff8ff000-ff8ff7ff] Max Packet=[2048]
    281 sbp2: Unknown parameter `sbp2_serialize_io'
    282 sbp2: $Rev: 1219 $ Ben Collins
    283 ieee1394: Host added: ID:BUS[0-00:1023] GUID[000199000003ba26]
$

> What happens if you use an older hard drive?

The one sample I tried looks fine, up to 200 MiB, reading or writing.

> I'm suspecting that some part of the SBP2 command tables might not be
> very happy talking to a very new type of drive.

Ouch. To distinguish between sbp2 etc. choking over DVD/CD from sbp2
etc. choking over specifically my device, I think I next need to go
find another FireWire DVD/CD to try?

> Is there a reason why serialize_io=1 is a problem for you?

Not a problem for me. Only a problem for the people after me who have
an equally hard time learning to spell serialize_io without the sbp2_
prefix. This issue is a priority for me only because people here are
making the time to teach me how to help, not because I particularly
need to discover now if the device or the host or both are at fault and
why.

> Can you not add that parameter to your modules.conf file and then it
> would boot and work correctly?

I think yes. I will try that and report back.

> Possibly the drive has firmware problems, ...
> or possibly updates?

Could be. Works in Mac & Windows, so I need more data to claim this
credibly enough to interest firmware folk of the manufacturer, I think.

> Possibly the drive has ... updates?

Not yet. This model appeared in like August 2004, no updates yet.

> understandably frustrated I'm sure.

Not at all. Instead I'm delighted to see you make time to teach me how
to help debug the host and/or the device here.

> Help me understand. I got this from your blog.

Thanks for making the time to say so. I'm trying to learn what subset
of data I can helpfully forward here, and I know blog-city.com doesn't
talk to at least some Konqueror unless one thinks to try turning off
Browser Identification.

Pat LaVarre
http://linux-pel.blog-city.com/read/796163.htm

-------------------------------------------------------
This SF.Net email is sponsored by: YOU BE THE JUDGE. Be one of 170
Project Admins to receive an Apple iPod Mini FREE for your judgement on
who ports your project to Linux PPC the best. Sponsored by IBM.
Deadline: Sept. 24. Go here: http://sf.net/ppc_contest.php
_______________________________________________
mailing list Linux1394-user@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/linux1394-user
Received on Fri Sep 17 01:52:00 2004