Comment 24 for bug 1653327

Alright, so I noticed that 4.9-rc1 is still good, so I did a reverse commit bisect between 4.8 and r.9-rc1. Here was the final result:

2a5708409e4e05446eb1a89ecb48641d6fd5d5a9 is the first bad commit
commit 2a5708409e4e05446eb1a89ecb48641d6fd5d5a9
Author: Lv Zheng <email address hidden>
Date: Wed Sep 7 16:50:21 2016 +0800

    ACPI / EC: Fix a gap that ECDT EC cannot handle EC events

    It is possible to register _Qxx from namespace and use the ECDT EC to
    perform event handling. The reported bug reveals that Windows is using ECDT
    in this way in case the namespace EC is not present. This patch facilitates
    Linux to support ECDT in this way.

    Link: https://bugzilla.kernel.org/show_bug.cgi?id=115021
    Reported-and-tested-by: Luya Tshimbalanga <email address hidden>
    Tested-by: Jonh Henderson <email address hidden>
    Reviewed-by: Peter Wu <email address hidden>
    Signed-off-by: Lv Zheng <email address hidden>
    Signed-off-by: Rafael J. Wysocki <email address hidden>

The parent of that commit (which I am assuming is the "good commit") is:

commit 46922d2a3aff5122253d97e64500801c08f4f2c0
Author: Lv Zheng <email address hidden>
Date: Wed Sep 7 16:50:14 2016 +0800

    ACPI / EC: Fix a memory leakage issue in acpi_ec_add()

    When the handler installation failed, there was no code to free the
    allocated EC device. This patch fixes this memory leakage issue.

    Link: https://bugzilla.kernel.org/show_bug.cgi?id=115021
    Reported-and-tested-by: Luya Tshimbalanga <email address hidden>
    Tested-by: Jonh Henderson <email address hidden>
    Signed-off-by: Lv Zheng <email address hidden>
    Signed-off-by: Rafael J. Wysocki <email address hidden>

Let me know if you need anything else.