Comment 10 for bug 1572630

There is an upstream fix for this issue.

First introduced upstream in branch : v4.6-rc1

---
commit 897bb0c7f1ea82d7cc882b19790b5e1df00ffc29
Author: Thomas Gleixner <email address hidden>
Date: Sat Mar 19 11:30:33 2016 +0100

    blk-mq: Use proper cpumask iterator

    queue_for_each_ctx() iterates over per_cpu variables under the assumption that
    the possible cpu mask cannot have holes. That's wrong as all cpumasks can have
    holes. In case there are holes the iteration ends up accessing uninitialized
    memory and crashing as a result.

    Replace the macro by a proper for_each_possible_cpu() loop and drop the unused
    macro blk_ctx_sum() which references queue_for_each_ctx().
---

Eric