From: Michal Soltys <redacted>
Date: Wed,  3 Aug 2016 00:44:53 +0200

Changes since v1:

- in the first patch's commit message, reference old patch using kernel.org's
  historical tree (instead of using gmane)
- in the second patch we just remove variable

Notes regarding patch #1/2:

This patch syncs virtual times with fair service curve and fixes a very old
subtle bug.

The detailed explanation is in the commit message. Additionally
I've made an illustration to help understand the issue better:

http://imgur.com/a/N8uMC

See the example at the bottom of the commit message - Am1_3 and Am2_3 is what
should happen with such queue setup, Am1_3real and Am2_3real is what actually
happens due to rtsc_min() calculating minimum from corrected and uncorrected
curves.

I applied this series to net-next, thanks Michal.