David, Charles, and everybody:
This is just a quick summary of my recent results. I tried running the usbtest endpoint-0 tests on two different computers, running the same kernel and talking to the same USB device. The difference was that the desktop system has an OHCI controller and the laptop has a UHCI controller.
With OHCI, test 10 ran perfectly. With UHCI it didn't, although my symptoms were quite different from what happened to Charles. Every single time, the first control transfer succeeded and the second one stalled. That happened even when I swapped the case labels in the usbtest source, so that the subtest 1 packet was sent first and the subtest 0 packet second. Clearly there's something wrong with the control transfer queueing in the UHCI driver. (It's vaguely possible that it might be related to the changes Duncan Sands has been working on -- I haven't had time to look at them yet.)
Last time I used test10 on UHCI it behaved better, FWIW. Which suggests something changed.
I still don't have more than a clue what the real problem is. But I did track down about a half-dozen other problems or things-to-change! I'll post more details tomorrow. Briefly: I've got some patches for usbtest, nuke_urbs() is used wrongly in usb.c, hcd_endpoint_disable() has an irq-related error, and hub_events() could stand to be written a little more defensively.
It would seem to be serving its purpose of helping to shake out bugs then! Glad to hear it, and I look forward to hearing about those other updates.
- Dave
Alan Stern
------------------------------------------------------- This SF.net email is sponsored by: VM Ware With VMware you can run multiple operating systems on a single machine. WITHOUT REBOOTING! Mix Linux / Windows / Novell virtual machines at the same time. Free trial click here: http://www.vmware.com/wl/offer/345/0 _______________________________________________ [EMAIL PROTECTED] To unsubscribe, use the last form field at: https://lists.sourceforge.net/lists/listinfo/linux-usb-devel
