On 2018-01-16 17:14, Henning Schild wrote:
> Hi,
> 
> these patches are an attempt to get the contributions from Jonas in a
> shape that can be merged. This first version was tested on x86 and arm
> 32, i want it to work on arm64 as well but still had some trouble there.
> 
> So this first attempt is just to collect some feedback about the
> general approach.
> 
> Jonas:
> I kept you as author for one patch, and i added your signed-off, please
> confirm that that was ok. If you want to test it, please also update
> ivshmem-guest-code to what i pushed there today.
> 
> There are still a few points:
> - the pci_cfg_base could become a general cmdline arg for systems doing
>   mmio-pci, right now this is part of the demo, could maybe done once
>   another demo needs it as well
> - for my tests i abused the -linux-cells because we have ready and
>   working configs with an ivshmem-pair for those, that is why i dropped
>   the check for the !VETH, i am even tempted to revert the patch that
>   introduced that, just for sharing configs
>   Ivshmem-net should not get confused, since the demo-code does not
>   touch the state-register
> - i kind of dislike the #ifdef-spaghetti that i ended up with ...
> 

Ralf ran into problems with his comm region patches for ARM, and I just
realized that they will affect the bare-metal ivshmem demos on that
archs as well:

So far we run ARM inmates with MMU disabled, which implies that caches
are off. But that also breaks coherence with cores which have caches
enabled - like the root cell or when in hypervisor mode. We have to
enable the MMU (I've sent Ralf already a never submitted changeset that
did this), and then we have to implement and use map_range() on ARM as well.

Jan

-- 
Siemens AG, Corporate Technology, CT RDA IOT SES-DE
Corporate Competence Center Embedded Linux

-- 
You received this message because you are subscribed to the Google Groups 
"Jailhouse" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to [email protected].
For more options, visit https://groups.google.com/d/optout.

Reply via email to