in hub_clear_tt_buffer, the value USB_DIR_IN | USB_RECIP_OTHER is used for bmRequestType, when section 11.24.2 of the spec says it should really be USB_RT_PORT, or 00100011b. This would explain why I can't clear some TT errors. I'll see if this gets better with the correct value in there.
Kevin
On Friday, October 10, 2003, at 05:22 PM, David Brownell wrote:
Kevin Owen wrote:On Thursday, October 9, 2003, at 07:50 PM, David Brownell wrote:Kevin Owen wrote:I was thinking more down at the post-transaction translator level, where the sITDs and sIRQ transactions have to share the 8 microframes. Even if sITDS come first in the schedule, you still have to fit them into the same TT pipeline. Also, sITD and sIRQ share the same 90% restriction, that is they have to leave 10% of the 1.1 b/w for control transfers.
A good portion of the devices I'm trying to support with USB 2.0 are audio devices. These device's isochronous transfers would have to share the EHCI schedule with interrupt transfers, which would become very complicated, especially given interrupt transfer's beta status under the
Actually less than you might think, since the EHCI spec says sITDs always
need to go at the front of the schedule. Or do you read that differently?
So the only real complication is making sure it stays in-budget.
I keep hoping someone will dive into split iso...
Both are budgeting issues. One is the "is this TT available now" issue, the other is the 90% rule.
Interrupt transactions are unnecessary for hubs, all they do is poll for a change. You need a control transfer afterwards anyway to find out what changed. If I just use control transfers to poll hubs, I have the full 90% (or more) of the 1.1 bus for isoc transactions. I would really like to get 3 audio devices, 48khz stereo in/out on a 1.1 bus, this bumps up close to the 90% limit.
What ISO packet size are you assuming?
Yes. But you'll need at least some part of that patch to handle this.It is our intention to move to 2.6 ASAP, but I have to get lots of audio devices working this month, so I have to use some sort of hybrid kernel.
Or at any rate it'd awkward to rely on me to supply anything ... :)
The 2.6 and 2.4 drivers are essentially identical.
Using python and ioctls, its a 30 second change to experiment with
One of these days I'll get more into UML ... :)
- Dave
------------------------------------------------------- This SF.net email is sponsored by: IBM Linux Tutorials. Become an expert in LINUX or just sharpen your skills. Sign up for IBM's Free Linux Tutorials. Learn everything from the bash shell to sys admin. Click now! http://ads.osdn.com/?ad_id=1278&alloc_id=3371&op=click _______________________________________________ [EMAIL PROTECTED] To unsubscribe, use the last form field at: https://lists.sourceforge.net/lists/listinfo/linux-usb-devel
