On Thu, 9 Jun 2005 10:27:27 -0400 (EDT), Alan Stern <[EMAIL PROTECTED]> wrote:
> On Thu, 9 Jun 2005, Christian Pernegger wrote:
> (Which won't be much help until Pete can decode it for us...)
OK, here is the interesting part of the trace.
Up to this point, something hits the UPS with control messages every 4ms
(sounds crazy, but here it is: the daemon is obviously out of control).
Then:
> [... gazillions of identical control requests ...]
> c21ed160 2336264501 C Ci:002:00 0 3 D
> c21ed160 2336264571 S Ci:002:00 -115 8 D
> c21ed160 2336267491 C Ci:002:00 0 3 D
> c21ed160 2336267525 S Ci:002:00 -115 8 D
> c21ed1e0 2336269495 C Ii:002:01 0 3 D
<---- Interrupt callback from ep1
> c21ed1e0 2336269506 S Ii:002:01 -115 3 D
<---- Interrupt URB resubmitted
> c21ed160 2336271492 C Ci:002:00 0 3 D
<---- Another 4ms, another control request
> c21ed160 2336271526 S Ci:002:00 -115 8 D
<---- a small pause of 6ms
> c21ed1e0 2336277499 C Ii:002:01 0 2 D
<---- Interrupt returned with size 2
> c21ed1e0 2336277514 S Ii:002:01 -115 3 D
> c21ed160 2336278491 C Ci:002:00 0 3 D
> c21ed160 2336278523 S Ci:002:00 -115 8 D
> c21ed160 2336283507 C Ci:002:00 0 3 D
> c21ed160 2336283581 S Ci:002:00 -115 8 D
<---- Looks like periodic control fetches resumed
> c21ed1e0 2336285496 C Ii:002:01 0 3 D
> c21ed1e0 2336285515 S Ii:002:01 -115 3 D
> c21ed160 2336286494 C Ci:002:00 0 3 D
> c21ed160 2336286525 S Ci:002:00 -115 8 D
> c21ed160 2336291501 C Ci:002:00 0 3 D
> c21ed160 2336291538 S Ci:002:00 -115 8 D
<---- Submission of control request
> c21ed1e0 2336293503 C Ii:002:01 0 2 D
> c21ed1e0 2336293519 S Ii:002:01 -115 3 D
<---- Interrupt - callback received (2 bytes), URB resubmitted
> c21ed160 2336298501 C Ci:002:00 0 3 D
<---- Control request succeeded, 3 bytes returned as usual
> c21ed160 2336298536 S Ci:002:00 -115 8 D
> c21ed160 2336303500 C Ci:002:00 0 3 D
> [... skipping a few of the same control requests ...]
> c21ed160 2336518545 S Ci:002:00 -115 8 D
> c21ed160 2336523515 C Ci:002:00 0 3 D
> c21ed160 2336523562 S Ci:002:00 -115 8 D
<---- Control submission without completion
> c21ed1e0 2337061533 C Ii:002:01 0 3 D
> c21ed1e0 2337061550 S Ii:002:01 -115 3 D
<----- Pretty big gap, 3.5s
> c21ed1e0 2340357664 C Ii:002:01 0 3 D
> c21ed1e0 2340357681 S Ii:002:01 -115 3 D
>From now on it's only interrupt requests ending, about every 8ms or so.
This also sounds pretty wild.
Unfortunately, current usbmon cannot tell if usbfs tried to unlink
the stuck control. I'll work on it.
Also, these 'D' tags mean that USB_NO_TRANSFER_DMA_MAP was set. In such
case, usbmon does not peek on the data, out of concern for the ownership
transfer for cache lines. It probably should do it at least on x86.
-- Pete
-------------------------------------------------------
This SF.Net email is sponsored by: NEC IT Guy Games. How far can you shotput
a projector? How fast can you ride your desk chair down the office luge track?
If you want to score the big prize, get to know the little guy.
Play to win an NEC 61" plasma display: http://www.necitguy.com/?r=20
_______________________________________________
linux-usb-devel@lists.sourceforge.net
To unsubscribe, use the last form field at:
https://lists.sourceforge.net/lists/listinfo/linux-usb-devel
