From: Mathias Nyman > Sent: 24 August 2015 10:48 > On 23.08.2015 09:37, Vincent Pelletier wrote: > > > > I have one machine which fails to get a working USB bus on at least > > every other boot. This machine having no PS2 ports, bus failure means > > no keyboard, and a forced power cycle (sadly, there is no accessible > > reset button) to try again. > > > > Do the usb devices work normally after a successful boot? > (no suspicious xhci entries in dmesg) > > > This is with vanilla kernel as of v4.1.4 tag, but I do not know if this > > is a regression (new machine). > > > > Visible USB-related boot messages (boot with "quiet" kernel parameter > > to avoid scrolling, messages typed from a screenshot): > > > > [ 7.25] xhci_hcd 0000:00:14.0: Command completion event does not match command > > This is not very common, when we queue a command for xhci we write the command to > to the command ring, and also add it to a linked list. We get an event for each completed > command. the event should match the command in the list. Probably worth determining which xhci controller hardware is being used. I've seen issues with (I think) the asmedia one where it would only process one command per ring of the doorbell. This generated timeouts when multiple requests got queued (doesn't happen often, but the Ge driver didn't wait for all responses). After that each later command actioned one of the queued ones - giving confusing results. I didn't try patching that one, I've still still got the host, but not the ethernet card. (Easy enough to ring the doorbell again just in case.) David -- To unsubscribe from this list: send the line "unsubscribe linux-usb" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
