On Tue, Aug 08, 2017 at 06:34:01PM +0200, Mike Galbraith wrote:

On Tue, 2017-08-08 at 09:22 -0700, Greg KH wrote:

quoted

On Sun, Jul 30, 2017 at 03:50:15PM +0200, Oleksandr Natalenko wrote:

quoted

Hello Mike et al.

On neděle 30. července 2017 7:12:31 CEST Mike Galbraith wrote:

quoted

FWIW, first thing I'd do is update that 4.12.0 to 4.12.4, and see if
stable fixed it.

My build already includes v4.12.4.

quoted

If not, I'd find these two commits irresistible.

5f042e7cbd9eb blk-mq: Include all present CPUs in the default queue mapping
4b855ad37194f blk-mq: Create hctx for each present CPU

I've applied these 2 commits, and cannot reproduce the issue anymore. Looks 
like a perfect hit, thanks!

quoted

'course applying random upstream bits does come with some risk, trying
a kernel already containing them has less "entertainment" potential. 

Should you consider applying them to v4.12.x stable series? CC'ing Greg just 
in case.

I can queue these up if I get an ack from the developers/maintainers
that it is ok to do so...

{hint}

{hint++}

Those commits take Steven Rostedt's hotplug stress script runtime down
from 4 _minutes_ down to 7 seconds for my RT tree, so I'm rather hoping
you hear an "ACK" too.

Oh, nice!

Should these go back farther than 4.12?  Looks like they apply cleanly
to 4.9, didn't look older than that...

thanks,

greg k-h