[EMAIL PROTECTED] wrote: > I think it's pratically impossible to remove/cancel/erase/unlink > an URB submition from its completition handler so that the handler > itself will _never_ be called.
I think that didn't translate right to English ... if the handler weren't called in the first place, it couldn't unlink from there. And unlinking should be guaranteeing the completion will be called with -ECONNRESET status. (Because the only kind of unlink you can use from a completion handler is the async kind.) So "never" is not going to happen > I tried everything but the system crashes with no kernel messages (!) > and the usb_unlink_urb is the problem. I tried putting urb->next = NULL I think you said you were using one of the UHCI drivers ... what happened when you tried the other one? Or using OHCI hardware? There have in the past been bugs in at least "usb-uhci" in that area ... using "uhci" sometimes improved things. What happened when you got rid of all use of urb->next, and used a simpler strategy for managing your (seemingly periodic) urbs? - Dave > Oh, ok, the system crashes ..so the transfer is stopped automatically > ...yeah, but it is not exactly what i want :-) ------------------------------------------------------- This SF.NET email is sponsored by: SourceForge Enterprise Edition + IBM + LinuxWorld = Something 2 See! http://www.vasoftware.com _______________________________________________ [EMAIL PROTECTED] To unsubscribe, use the last form field at: https://lists.sourceforge.net/lists/listinfo/linux-usb-devel
