Re: Status of v4.9

[Christoph Biedl <linux-kernel.bfrz@manchmal.in-ulm.de>]

[ That was in January ]

Christoph Biedl wrote...

> Greg Kroah-Hartman wrote...
>
> > If you need to point a manager at something, point them at my blog post,
> > that is why I wrote it.  And test the heck out of it starting now!
>
> The only thing that requires checking is the X server in a KVM guest
> that fails to start. But since I haven got around to take a closer look
> yet, I preferred not the bother anybody for the time being.

After a round tuit finally arrived: This is not a regression in the kernel,
just a stricter behaviour some userland cannot cope with (yet).

It was introduced in

    commit 21266be9ed542f13436bd9c75316d43e1e84f6ae
    Author: Dan Williams <dan.j.williams@intel.com>
    Date:   Thu Nov 19 18:19:29 2015 -0800

        arch: consolidate CONFIG_STRICT_DEVM in lib/Kconfig.debug

and manifests only if CONFIG_IO_STRICT_DEVMEM is set. Then some X
servers will report "Internal error: could not map framebuffer range
(22, Invalid argument)" or even happily segfault. It's beyond my
knowledge whether they can handle that situation. In my opinion, they
should, if ever possible.

    Christoph