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

Pat LaVarre wrote:
> Mark K:
>
>> Can you please tell me what you mean by "With my PC, Knoppix
>> Linux-2.4.26 consistently reads hundreds of MiB thru FireWire at
>> within three percent of that rate." I do nt understand what this
>> sentence means or how you got the data.
>
>
> I meant that I burned 2004-05-17 Knoppix 3.4 from linuxiso.org to CD-R,
> rebooted, and tried something much like:
>
> uname -msr
> sudo blockdev --flushbufs /dev/scd0
> sudo dd if=/dev/scd0 bs=1M skip=0 count=1 >/dev/null
>
> uname -msr told me 2.4.26 and `dc` divided B by s for me. I got 2.6.6
> instead only if I answered boot: knoppix26. I got "aborting sbp2
> command" whenever I tried the kernel.org sources instead.

OK, this is a bit beyond me, but I'm guessing that 'count=1' tells you
the number of bytes dd moves? If so, where do you get the time used to
do the copy?

>
> I'm guessing that further detail makes my English understandable?

That's perfectly OK. My French, Spanish and Hindi are far worse!

<SNIP>

>
>> Post back the output of hdparm -tT or some other data.
>
>
> Please tell me again this is still useful, if it is, and gladly I will.
> (Else I have no experience of hdparm.)

Yes, please post back the output of

hdparm -tT /dev/scd0

under both 2.4.26 and 2.6.6.

>
>>> How do I build 2.6 Linux from kernel.org that does read at speed without
>>> "aborting sbp2 command"?
>>> ...
>>
>> The common answer around here is that:
>>
>> 1) The 2.4 series 1394 drivers work better than the 2.6 series drivers
>>
>> 2) The 1394 drivers have never been optimized for performance
>
>
> Thanks for defining normal for me, yes that's exactly what I see in
> Knoppix.

Well, I don't remember seeing good performance numbers in a while so we
can just work from what you have taken. That's what matters to you anyway.

>
> By contrast, in kernel.org source, I see nothing work. Is that also
> normal?

That I don't know although it may be that some distro's test and modify
1394 to somehow improve it but I have no knowledge. kernel.org is what
the 1394 developers write.

>
> Can I help the kernel.org source in any way?

Maybe once we agree on and understand the data.

- Mark

-------------------------------------------------------
This SF.Net email is sponsored by BEA Weblogic Workshop
FREE Java Enterprise J2EE developer tools!
Get your free copy of BEA WebLogic Workshop 8.1 today.
http://ads.osdn.com/?ad_id=5047&alloc_id=10808&op=click
_______________________________________________
mailing list Linux1394-user@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/linux1394-user
Received on Thu Sep 2 01:09:00 2004