I wondered about my system performance slowing down (-15% with an old
Pentium 133) if I insert USB devices which use an interrupt endpoint (e.g.
USB <-> Ethernet Adaptor, USB <-> Bluetooth Stick). This happens if I just
set the interfaces to UP, no data transfer at all.

Digging more into detail with these devices I found, that the PCI bus load
grows heavily because of periodic DMA cycles (in a 8 mikro seconds
period). I measured this in several PCs with OHCI USB and different
kernels (2.4.21 and 2.6.7) and I found no differences.

I took an USB analyser and found that there are a lot of NAK's on the
referring USB's interrupt pipe (in the same 8 usec period). It seems to be
that the interrupt pipe gets round about 117 NAK's from the device per
milli second. Is this a correct behaviour? Regarding USB I am far away
from being an expert. Maybe one of you can help me out of my confusion.
However, the OHCI spec says that the minimal polling rate should be 1 ms,
thats all I know. Why these flood of NAKs?

If I am right that the USB host controller floods the PCI bus with all
these NAKs: Is it possible to stop this by some settings for this
endpoint? It wastes a lot of PCI bandwidth.

Thank you in advance for your help ...

Christian Gohmann



-------------------------------------------------------
This SF.Net email sponsored by Black Hat Briefings & Training.
Attend Black Hat Briefings & Training, Las Vegas July 24-29 - 
digital self defense, top technical experts, no vendor pitches, 
unmatched networking opportunities. Visit www.blackhat.com
_______________________________________________
[EMAIL PROTECTED]
To unsubscribe, use the last form field at:
https://lists.sourceforge.net/lists/listinfo/linux-usb-devel

Reply via email to