Hi,

Thanks for the help and sorry for taking too long to answer back. Below
my comments:

Alan Stern wrote:
>>     * I bought and tried with a new USB PCI card instead of the
>>       motherboard-attached USB ports. The chipset of the PCI card is
>>       NEC. The chipset of the motherboard is VIA (although later saw
>>       that lsusb -v shows as NEC also)
>>     
>
> Does the same problem occur with the PCI-card controller?  With each of 
> your devices?
>   
Yes - I mean, "I thought it ocurred, with all devices". What happens is
that the situation is not exactly as I pictured it in the last email
(probably because some non-metodological testing)

What happened with the VIA motherboard USB ports was that them got
stalled. With the NEC PCI ports where that the transfer was not steady,
so working at something like 15MB/s for a while, then stopping for a
time, then back at ~15MB/s, averaging something like 1-2MB/s, but
without any warning/error in the logs.
>>     * I tried with low speed USB devices attached at the same time, and
>>       without them
>>     
>
> Have you tried using different USB cables?
>   
Yes, I did - but I did it again now and it helps somehow. The issue (I
think the trigger of the problems) is a 4port hub attached to the VIA
USB bus. In my first post I said that there were no hubs attached to the
port, as I was using a diferent connector in the USB for the test than
for the hub. Also, those two connectors were attached to the motherboard
through two different connectors, so I assumed that there were diferent
busses, but they aren't. So now, knowing that all the mb USB 2.0 ports
are just one bus, I detached the hub and tried again, using those USB
cables with that "anti noise cylinder" at the ends (sorry, don't know
the name in english - hope you understand what i'm talking about), and
I've been unable to reproduce the problem.

>> I thoght it could be just a matter of usbmon doing *something* enough to
>> make the USB bus stable, so I repeated the tests with everything loaded,
>> but without actually capturing the bus to a file, and the test failed. I
>> even tried to "/cat /sys/kernel/debug/usbmon/5t >/dev/null/" but it
>> still fails.
>>     
>
> usbmon doesn't do anything to the bus other than inserting delays at 
> various spots (while usbmon does its work).  Of course, the delays 
> themselves may be enough to change the behavior, although I would expect 
> those same delays to be present also in the "cat >/dev/null" test.
>
> What happens if you do:
>
>       dd if=/sys/kernel/debug/usbmon/5t obs=10M of=file
> ?
>   
TBH, having found a working solution (remove the hub + use "anti noise"
cables), I am a bit lazy about doing more tests. However, if anyone
reading this post is in the same situation that the one I described and
needs me to do the test, or if any linux-usb-devel is particularly
interested in these tests, let me know and I will set up them again.

Cheers,
k

-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys - and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
_______________________________________________
Linux-usb-users@lists.sourceforge.net
To unsubscribe, use the last form field at:
https://lists.sourceforge.net/lists/listinfo/linux-usb-users

Reply via email to