Greg Kroah-Hartman wrote...

> I dropped e1c3743e1a20 ("powerpc/tm: Set MSR[TS] just prior to recheckpoint")
> from the stable trees, which is what I was told was the commit that was
> causing the problems by Christoph and Breno (on to: now).
>
> Was that not the offending commit?  If so, what one was?
>
> totally confused,

Sorry for not getting back to you earlier. Building yesterday's
release (v4.19.14) *failed*, bisect led to

| commit a9935a12768851762089fda8e5a9daaf0231808e (HEAD)
| Author: Breno Leitao <leitao@debian.org>
| Date:   Mon Nov 26 18:12:00 2018 -0200
|
|     powerpc/tm: Unset MSR[TS] if not recheckpointing

Reverting that commit seems to be sufficient, build passes then.

Additionally, neither 4.20 nor 5.0-rc1 show this problem. The
| commit 6f5b9f018f4c7686fd944d920209d1382d320e4e upstream.
builds as well, so I'll try to find the missing prerequisite next.

An example .config is attached.

    Christoph