This is because of another binder change that delays handling certain events to improve latency - and the test had certain assumptions about the arrival time of these events it shouldn't have.

I don't have the kernel commit triggering this (am on mobile), but the test should be fixed instead - fix is here:

https://android-review.googlesource.com/563488

Martijn

On Dec 12, 2017 20:09, "Steve Muckle" <smuckle.linux@gmail.com> wrote:
On 12/11/2017 09:23 AM, Greg KH wrote:
On Sat, Dec 09, 2017 at 12:54:40AM +0000, Linaro QA wrote:
Summary
------------------------------------------------------------------------

git repo: https://android.googlesource.com/kernel/hikey-linaro
git branch: android-hikey-linaro-4.9
git commit: 71c123c37b45b13047137860c196f35d509cae8d
git describe: v4.9.67-638727-g71c123c37b45
Test details: https://qa-reports.linaro.org/lkft/android-hikey-linaro-4.9-aosp/build/v4.9.67-638727-g71c123c37b45

Regressions (compared to build v4.9.65-638704-gca272c3a4d62)
------------------------------------------------------------------------

hi6220-hikey - arm64:
   vts-test/arm64-v8a.VtsKernelBinderTest:
     * VtsKernelBinderTest.BinderDriverInterfaceTest.Transaction_IPC64_64bit

     * test src: not informed
   vts-test/armeabi-v7a.VtsKernelBinderTest:
     * VtsKernelBinderTest.BinderDriverInterfaceTest.Transaction_IPC64_32bit


What did we break in binder???

This worries me as there was a binder patch in the tree, any ideas?

This reproduced for me. I figured reverting this commit:

commit e312c3fce34d4d82f5a55ae7d89f4d9a961ef8df
Author:     Arnd Bergmann <arnd@arndb.de>
AuthorDate: Tue Sep 5 10:56:13 2017 +0200
Commit:     Martijn Coenen <maco@google.com>
CommitDate: Mon Nov 27 09:28:08 2017 -0800

    UPSTREAM: android: binder: fix type mismatch warning

would solve it, but in a two runs I've gotten one failure of Transaction_IPC64_64bit and one failure of Transaction_IPC64_32bit. Not sure what is going on yet.