Roadmap 2023

[Norman Feske]

Hi Alex,

thank you for sharing your perspective and interests.

Your engagement with the nitty-gritty details of PC platforms - like 
power management, Intel Gen12, and suspend/resume - is a crucial 
counterweight to the PinePhone theme. From your posting, I sense that 
these interests are intrinsic, which I find very fortunate.

>> What directions are you most excited about?
> 
> Privately, most exciting was that the Seoul VMM is now enjoyable on a 
> day-to-day basis, because of the way better integration with Sculpt 
> thanks to the new Virtio model support for gpu, input and audio. With 
> that I have an enjoyable and maintainable and easy and quick upgradable 
> path for up-to-date Firefox and Thunderbird VMs on Sculpt. Beside that, 
> I worked on several proof of concept topics to improve further life on 
> Sculpt, e.g. dynamic CPU power & frequency adjustments and monitoring 
> for Intel HWP (work notebook) and AMD p-states (home desktop), with 
> which one can steer indirectly the fan noise. The other life improvement 
> was the principle support of the Intel Display Driver (as proof of 
> concept) to support beside the mirror mode also the extended mode, to 
> have one large screen across several attached monitors.
> 
>> Which topics do you deem as interesting to explore yourself?
>> Do you already happen to have rough plans about 2023?
> 
> - Continuing support for ACPI suspend & resume to Genode' base-hw (x86_64)
> - Device & driver restart and/or resume support after ACPI wakeup -> 
> precondition for advanced/useful scenarios like Sculpt OS (x86)
> - native Qemu port for cross architecture development
> - Further Seoul VMM cultivation
> - AMD display and GPU support
> - make some proof of concept work ready for Genode upstream

This plan paints a very consistent picture, most of which I will happily 
reflect on the road map. Your Qemu ambition is a nice surprise. Count me 
in as a user. :)

Regarding the AMD display/GPU support, I'm hesitant when it comes to the 
official road map, for two reasons:

First, given that the Intel GPU support was a multi-year effort, I'm 
very much afraid of the potential fragmentation that would come by 
opening up a second line of development. I wonder, is this realistic at 
all without the focus granted by a well-funded project?

Second, I'm concerned about raising unrealistic public expectations. If 
we mention AMD on our official road map, observers may draw the wrong 
conclusion regarding the maturity of Genode on such hardware. Given the 
amount of quality assurance we spend on Intel (including the daily use 
of Sculpt on our laptops), I cannot see us achieving the same quality 
for AMD. In the worst case, it could become an expensive hobby that 
leaves people disappointed nevertheless.

So whereas I am fully supportive of further cultivating the use of 
Genode on AMD driven by curiosity, I personally want avoid an 
uncomfortable sense of responsibility about it. How do you think we 
should go about it?

Cheers
Norman

-- 
Dr.-Ing. Norman Feske
Genode Labs

https://www.genode-labs.com · https://genode.org

Genode Labs GmbH · Amtsgericht Dresden · HRB 28424 · Sitz Dresden
Geschäftsführer: Dr.-Ing. Norman Feske, Christian Helmuth