On Thu, 10 Jul 2014, Sasha Levin wrote: > On 07/10/2014 01:55 PM, Hugh Dickins wrote: > >> And finally, (not) holding the i_mmap_mutex: > > I don't understand what prompts you to show this particular task. > > I imagine the dump shows lots of other tasks which are waiting to get an > > i_mmap_mutex, and quite a lot of other tasks which are neither waiting > > for nor holding an i_mmap_mutex. > > > > Why are you showing this one in particular? Because it looks like the > > one you fingered yesterday? But I didn't see a good reason to finger > > that one either. > > There are a few more tasks like this one, my criteria was tasks that lockdep > claims were holding i_mmap_mutex, but are actually not. You and Vlastimil enlightened me yesterday that lockdep shows tasks as holding i_mmap_mutex when they are actually waiting to get i_mmap_mutex. Hundreds of those in yesterday's log, hundreds of them in today's. The full log you've sent (thanks) is for a different run from the one you showed in today's mail. No problem with that, except when I assert that trinity-c190 in today's mail is just like trinity-c402 in yesterday's, a task caught at one stage of exit_mmap in the stack dumps, then a later stage of exit_mmap in the locks held dumps, I'm guessing rather than confirming from the log. There's nothing(?) interesting about those tasks, they're just tasks we have been lucky to catch a moment before they reach the i_mmap_mutex hang affecting the majority. > > One new thing that I did notice is that since trinity spins a lot of new children > to test out things like execve() which would kill said children, there tends to > be a rather large amount of new tasks created and killed constantly. > > So if you look at the bottom of the new log (attached), you'll see that there > are quite a few "trinity-subchild" processes trying to die, unsuccessfully. Lots of those in yesterday's log too: waiting to get i_mmap_mutex. I'll pore over the new log. It does help to know that its base kernel is more stable: thanks so much. But whether I can work out any more... Hugh -- To unsubscribe from this list: send the line "unsubscribe stable" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html