On Tuesday 12 October 2004 9:41 am, Russell King wrote: > > I suspect that problem isn't associated with Lothar's code at all; > > a lot of folk have been running into problems like that lately. > > Actually it is. After running the SL811HS against the SA1110 UDC driver, > the problem comes out immediately. The SL811HS driver is _NOT_ following > the protocol for bringing a device out of "default suspend" state. > > Sure, it's resetting the bus, but then it's waiting for about 10ms before > enabling SOFs, by which time the device has given up completely and gone > back to suspend mode. > > With this kind of bug, this driver could never have worked.
Hmm, that sounds like a longtanding issue in OHCI that I finally fixed not long ago. Fix is in the MM tree; it finally mattered on some hardware. Basically initialization was split into two routines, and a 2 msec timeout spanned both ... preemptibly! It was amazing to me that it ever worked at all reliably. The fix was obvious. > I have it almost working now, mainly by unhooking the reset device > and device detection from the "OHCI emulation". > > As I've said previously, I think the whole design of this driver is > flawed by the mere fact that it tries to pretend to be an OHCI device > when it isn't. That leads to a hell of a lot of extra code and > complexity - which directly leads to these types of bugs. That was one of my concerns. Another is that the OHCI code is starting to talk to a heck of a lot of hardware, and having codepaths that "real OHCI" (or even the big-endian type) can't use would be bad for reliability. On the flip side, I think the OHCI code is a probably the best current model for "all that an HCD must do". Writing an HCD is still unfortunately a bit of a black art, and the SL811 seems to need more help from software than "real OHCI" ... writing one from scratch is probably not for the faint hearted. Sounds like it might be best if SL811 just did a copy/modify of the relevant chunks of OHCI code, and diverged. That'd make it easier to merge into some BK tree too, helping sort out the version issues. - Dave ------------------------------------------------------- This SF.net email is sponsored by: IT Product Guide on ITManagersJournal Use IT products in your business? Tell us what you think of them. Give us Your Opinions, Get Free ThinkGeek Gift Certificates! Click to find out more http://productguide.itmanagersjournal.com/guidepromo.tmpl _______________________________________________ [EMAIL PROTECTED] To unsubscribe, use the last form field at: https://lists.sourceforge.net/lists/listinfo/linux-usb-devel