On Wed, 19 May 2010 20:22:46 -0700 "Paul E. McKenney" <paulmck@xxxxxxxxxxxxxxxxxx> wrote: > On Wed, May 19, 2010 at 09:19:11PM -0500, Clark Williams wrote: > > Thomas/Paul, > > > > Resumed my laptop (T60, 64-bit kernel) and after about five > > minutes of doing nothing but surfing with firefox, got the following > > RCU stall notification from a 2.6.33.4-rt20 kernel: > > > > INFO: RCU detected CPU 1 stall (t=0 jiffies) > > Hmmm... "t=0 jiffies". I wonder if the hibernation/resumption confused > the stall-warning code. My first thought was that an RCU grace period > had spanned the hibernation, but that would not be consistent with it > waiting five minutes after resumption to splat. > > I will look into this. > Don't take that "five minutes" as anything approaching some degree of accuracy. It could have been quicker; I was distracted and not even touching the keyboard when a loud beep sounded and I had to poke around a bit to realize that an NMI backtrace had triggered. Depending on your timeout values, I think that it's quite reasonable to suspect that an RCU spanned the suspend/resume. Clark
Attachment:
signature.asc
Description: PGP signature
