Hi Sarah, Andiry, I finally have some logs of those issues that I reported before. It takes a more complicated USB setup than before to show these issues. What I have now is this: +----+ +----+ +----+ |Snps| |USB3|---> USB3 SS BOT device |USB2|---> USB2 HS webcam |xhci|--->|SS |---------------------------->|FS | |ctlr| |hub |---> USB3 SS BOT device |hub | +----+ +----+ +----+ I am using a script to do file transfers to/from the BOT devices, and using Cheese to play video from the webcam. I am using Linus' 2.6.36 kernel from yesterday, plus a modified version of the debugging patch that Sarah created for this issue, plus John's hub patch. I have 3 logs with 3 different issues. I am attaching the first log, plus the modified debugging patch, to this email. I will send the other 2 logs as replies to this email. event_seg-is-null.log is only a partial log, unfortunately. It doesn't contain the initial instance of the problem, so I'm not sure if there is enough info to tell what went wrong. The next 2 logs are complete. I would appreciate if you can glean any clues to the problem from these logs. Thanks. -- Paul > -----Original Message----- > From: Paul Zimmerman > Sent: Tuesday, August 24, 2010 11:20 AM > To: 'Andiry Xu' > Cc: Sarah Sharp; linux-usb@xxxxxxxxxxxxxxx > Subject: RE: [PATCH][RFC] xhci fixes > > > From: Andiry Xu [mailto:andiry.xu@xxxxxxx] > > <snip> > > > But I haven't seen td bd415150 and following tds on the endpoint ring. > > I'm afraid the full endpoint ring is not printed, because I allocate 8 > > segments for isoc endpoint, but there is only one segment showed in the > > log. Perhaps you only print one segment of isoc endpoint ring? > > > > With your patch, every time when event_seg is NULL, you break out and > > returns. But event_seg will be NULL when a missed service interval event > > is encountered and it's quite normal for isoc transfer. Your patch may > > workaround the issue but it does not help to find the root cause. And > > suppose the do-while will not run infinity. ep->skip is cleared when > > irq_handler found the td on the ring, or when the td_list is empty. > > Since urb_enqueue needs to acquire the lock which is hold by > > irq_handler, it cannot insert tds to the ring. Eventually the td_list is > > empty, and ep->skip is clear. > > Hi Andiry, > > I have some good/bad news (depending on how you look at it). > > I have updated my debug code to print all the ring segments instead of just > one. > > However, our xHCI controller hardware was just updated to the latest version. > With this version, the "event_seg is NULL" case happens *much* less frequently > than before. So far, I have only been able to reproduce it once in over an > hour of intensive testing. Unfortunately, the debug messages had already been > flushed from the log by the time I realized it had occurred. > > Previously, it would happen after just a few startup/shutdown cycles of the > webcam app. > > So it's possible that the problematic case I was seeing was related to some > sort of hardware issue that has now been resolved. > > I will keep trying to capture a dmesg log from when it does happen, just to be > sure. > > -- > Paul
Attachment:
xhci-dbg.patch
Description: xhci-dbg.patch
Attachment:
event_seg-is-null.log.gz
Description: event_seg-is-null.log.gz
