Re: linux-next: umh: fix processed error when UMH_WAIT_PROC is used seems to break linux bridge on s390x (bisected)

From: Luis Chamberlain
Date: Fri Jun 26 2020 - 07:50:47 EST


On Fri, Jun 26, 2020 at 11:40:08AM +0000, Luis Chamberlain wrote:
> Andrew, can you please revert these two for now:
>
> selftests: simplify kmod failure value
> umh: fix processed error when UMH_WAIT_PROC is used
>
> Later, we'll add Christoph's simplier kernel wait, and make the change
> directly there to catpure the right error. That still won't fix this reported
> issue, but it will be cleaner and will go tested by Christian Borntraeger
> before.

However, note that the only consideration to make here against this
approach of the fix later going in with the simpler kernel wait is
if this actually is fixing a real issue, and if we'd want this going to
stable.

We should for sure revert though, so Andrew please do proceed to revert
or drop those two patches alone for now.

It was unclear to me if this should go to stable given I was not sure
how severe that NFS case mentioned on the commit log was, and no one on
the NFS side has replied about that yet, however there may be other
areas where code inspection of callsites was not sufficient to find the
real critical areas.

I'm now very curious what this issue that Christian with bridge on s390
found will be.

Luis