I think we got this bug
in the current gentoo-sources-4.9.16 stable gentoo kernel.
I'm using autofs in many of my systems and I'm getting a number of this hanging kernel threads, which can not be killed and may lead, eventually, to fill the kernel process table.
I haven't see this problem before and I'm pretty sure it was not present in the previous gentoo-sources-4.4.39 stable kernel I've used in the last couple of months. This Fedora bug is the only reference to this bug I've found so far.
For what I can understand, upstream patches are available for kernel 4.10.13, but, for what I'm reading, they are still under review.
seems to apply correctly to the gentoo-sources-4.9.16 stable kernel.
I had been able to build and boot a 4.9.16 kernel with these patches applied and nfs server seems to work correctly.
Only time may say if these patches actually solve the issue without introducing some new problem.
My last comment on this topic.
The patches seems to have solved the problem when applied to "NFS clients".
In 3 days I haven't seen any other hanging callback thread and no other evident problem so far in 3 of my NFS Gentoo systems
By the way: I don't think an "emerge --info" output would make any difference for such a problem.
Regards, G. Vitillaro
(In reply to Giuseppe Vitillaro from comment #2)
> My last comment on this topic.
> The patches seems to have solved the problem when applied to "NFS clients".
> In 3 days I haven't seen any other hanging callback thread and no other
> evident problem so far in 3 of my NFS Gentoo systems
Thanks for reporting this bug.
> By the way: I don't think an "emerge --info" output would make any
> difference for such a problem.
emerge --info is something that we ask by default to add to a bug request and that would be better to add in all bugs request (Gentoo kernel or not).
But yes in this case would be of little difference.
> Regards, G. Vitillaro
Patches are included in recent kernels.