Comment # 9 on bug 1048860 from Franck Bui

(In reply to Bart Van Assche from comment #8)
> Franck, the kernel should provide a mechanism but not implement a policy so
> a kernel patch that changes the default scheduler for single-queue block
> devices probably wouldn't be accepted upstream.

The rule file doesn't look to implement any policy but only changes the (IO
scheduler) default for SSD.

And we do that unconditionally from userspace, so either the choice made by the
kernel is incorrect for that type of block device or either the rule file is. 

> How about this updated patch
> that does not touch the scheduler assigned to multiqueue block devices?

I don't know and I think the kernel guys should be the one who review and ACK
such changes. That's the reason why I think the rule file should be maintained
by the kernel team.