[coreboot] Linux 4.0.x iomem regression

Aaron Durbin adurbin at google.com
Thu Sep 3 13:42:58 CET 2015


thanks

On Thu, Sep 3, 2015 at 5:53 AM, Torsten Duwe <duwe at lst.de> wrote:
> JFYI, I got stuck with kernel 3.19 this spring, and Linux-4.0 didn't work
> properly on my coreboot machine. Only now I've had time to look into this.
>
> To make it short: the fault is in the Linux PCI/ACPI code; see mainline
> commit 2c62e8492, which fixes this in 4.1-rc3. You may want to skim through
> https://lkml.org/lkml/2015/4/3/608 to get an idea of this mess.
>
> " I did it because Windows apparently does that ..."
>
> I'm not sure if and why proprietary BIOSes are unaffected.
> Just to let you all know in case you're bisecting across 4.0 linux kernels.

Thanks for the heads up. I'm sure this won't be last time either.
We're definitely in the age of tight coupling between kernel and
firmware. ACPI is not the abstraction it was intended to be. Both
sides have to agree on interpretation and tables provided.  For most
people who are used to be able to take an x86 linux live cd and boot
w/ things working, that is not longer the case any more.

-Aaron



More information about the coreboot mailing list