On Thu, 7 Mar 2019 00:32:09 +0100 Dominique Martinet [off-list ref] wrote:

Andrew Morton wrote on Wed, Mar 06, 2019:

quoted

On Wed, 6 Mar 2019 23:48:03 +0100 (CET) Jiri Kosina [off-list ref] wrote:

quoted

3/3 is actually waiting for your decision, see

	https://lore.kernel.org/lkml/20190212063643.GL15609@dhcp22.suse.cz/ (local)

I pity anyone who tried to understand this code by reading this code. 
Can we please get some careful commentary in there explaining what is
going on, and why things are thus?

I guess the [3/3] change makes sense, although it's unclear whether
anyone really needs it?  5.0 was released with 574823bfab8 ("Change
mincore() to count "mapped" pages rather than "cached" pages") so we'll
have a release cycle to somewhat determine how much impact 574823bfab8
has on users.  How about I queue up [3/3] and we reevaluate its
desirability in a couple of months?

FWIW,

574823bfab8 has been reverted in 30bac164aca750, included in 5.0-rc4, so
the controversial change has only been there from 5.0-rc1 to 5.0-rc3

Ah, OK, thanks, I misread.

Linus, do you have thoughts on
http://lkml.kernel.org/r/20190130124420.1834-4-vbabka@suse.cz ?