Re: khbud stuck in D state

[Pete Zaitcev <zaitcev@xxxxxxxxxx>]

On Sun, 18 Nov 2007 22:18:17 -0500 (EST), Alan Stern <stern@xxxxxxxxxxxxxxxxxxx> wrote:

> About all I can say is that the problem seems to be caused by a missing 
> hardware I/O completion event. [...]

Coincidently, I'm working on a situation like that in OHCI (since May!).
For some reason Sun x4100/x4200, AMD 8111, seems especially suceptible.
It was reported on 2.6.9, but as far as I can tell, there were
no changes which could address it since then. The tell-tale sign
is the "bad entry" message, when a TD is not found in the hash.
It is replaced by "no hash" these days, and relegated to debug-only.

After examining the code, I think that ohci_dbg was a bad idea.
When the "no hash" situation occurs, we may have an unfixable hard
hang for an URB (because the list processing is abandoned when this
happens and thus all completed TDs behind the lost one are also lost).
This situation must not happen and we ought to keep tabs on it. We
should bump the "no hash" message up to a normal printk with KERN_ERR.

-- Pete

-------------------------------------------------------------------------
This SF.net email is sponsored by: Microsoft
Defy all challenges. Microsoft(R) Visual Studio 2005.
http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/
_______________________________________________
linux-usb-devel@xxxxxxxxxxxxxxxxxxxxx
To unsubscribe, use the last form field at:
https://lists.sourceforge.net/lists/listinfo/linux-usb-devel