Re: usb suspend backtraces.
Alan Stern <[email protected]>
| Newsgroups | gmane.linux.usb.devel |
|---|---|
| Message-ID | <[email protected]> |
On Mon, 16 Jul 2007, Dave Jones wrote: > On Mon, Jul 16, 2007 at 03:22:17PM -0400, Alan Stern wrote: > > > > I'm not sure how that 5-1.4 relates to the above map. > > > > > > oh, there's one other case, where it prints the same thing for usb 4-2: > > > > All right, it turns out there really is a minor bug that shows up only > > with the autosuspend delay set to 0. The patch below fixes it (should > > you happen to care). > > Ah, so that's why I was seeing those WARN_ON's ? Yep. The old code incremented the parent hub's count of unsuspended children _after_ registering the child device -- but at the end of the registration procedure the child gets autosuspended. With a delay of 0, the child was suspending before the parent's count could be incremented; hence the suspend caused the count to go negative briefly. The patch fixes the problem by incrementing the parent's count _before_ registering the child (and then decrementing it afterward if the registration failed). Alan Stern ------------------------------------------------------------------------- This SF.net email is sponsored by DB2 Express Download DB2 Express C - the FREE version of DB2 express and take control of your XML. No limits. Just data. Click to get it now. http://sourceforge.net/powerbar/db2/ _______________________________________________ [email protected] To unsubscribe, use the last form field at: https://lists.sourceforge.net/lists/listinfo/linux-usb-devel