David Brownell wrote: > > This fix enhances the scope of the existing OHCI_QUIRK_ZFMICRO flag: > > 1. A watchdog routine periodically scans the OHCI structures to check > for orphaned TDs. In these cases the TD is taken back from the > controller and completed normally.
Of course, just after I verified and signed off on the final version of this patch, a problem has cropped up. This just showed up in the log of one of my test systems: user.warn kernel: ohci_hcd 0000:00:13.0: Reclaiming orphaned TD c3f41380 user.err kernel: ohci_hcd 0000:00:13.0: bad entry 3f41381 The first message indicates that my watchdog was triggered and the TD was taken back from the HC. The second message seems to indicate that the HC did, in fact, later complete the TD. I can think of two explanations for this: 1. We aren't waiting long enough between the time the TD is popped off the queue and the time it's taken back through the donelist. If that's the case, I don't know how long would be reasonable. 2. The logic is too naive. The quirk assumes that if HeadP and DoneP remain equal (meaning that the HC's queue is empty) for at least one frame, but the HCD still sees an outstanding TD, then that TD must be lost. But it may be that even though the queue is still empty, there were intervening transactions in the meantime (it was momentarily nonempty, but is now empty again). In that case, I could fix it by verifying that the second time around, if HeadP and TailP are still equal, they point to a *different* dummy than they did before. I don't know how that comparison would be done - a simple address check may not be sufficient, because the DMA allocation code will give out the same addresses for new TDs after the old ones are retired. Either way, this patch in its current form isn't safe. Mike ------------------------------------------------------------------------- This SF.net email is sponsored by: Splunk Inc. Still grepping through log files to find problems? Stop. Now Search log events and configuration files using AJAX and a browser. Download your FREE copy of Splunk now >> http://get.splunk.com/ _______________________________________________ linux-usb-devel@lists.sourceforge.net To unsubscribe, use the last form field at: https://lists.sourceforge.net/lists/listinfo/linux-usb-devel
