I hate to say this but I want to run Android in Pyra


so android should just run on a vm - no porting except maybe the virtualization software - so i'm kindof in favour of that. but i agree that we shouid focus on improving the pyra stock os, not android, not bsd. i hate android actually.

how's the virtualization support on OMAP5?
I know the feature can be enabled when compiling the kernel and I did that once on the devboard, but had no use cases for it at the time.
 
Last edited:
It is interresting to read.
I am one of the People who want Android too for the Pyra.
But maybe we will find another Ways to get Android running inside the Pyra.

There exist Software like Anbox or Shashlik.
Maybe with one of them we get Android Apps run inside the PyraOS.

But most they are simply too big and too complicated to Compile.
All my Compiling Trys failed and andbox from the Repo definly do not work on the Pyra.


I do dream of a Pyra MultiOS.
One OS where you have
- ARMLinux as Base
- X86 Symbiotic with PtitSeb Box86
- and a Software to run Android Apps native inside this Linux

It will be great when Someone get this Anbox,Shashlik or one of the many Android Wrappers running on the Pyra.

And all Whishes are solved :)
 
it's less of a kernel and more of a hardware thing
It's both surely? To enable virtualisation, you need a processor with the right execution modes and instructions, and you need the kernel to support it, because the instructions aren't available to userspace. But even I'd agree that most of the time it's easier to swap out you kernel that it is to swap out a chip. But at the very least if Trashy reckons there's kernels out there that do support virualisation on A15 Cortex, that does suggest the chip is good.
 
it's less of a kernel and more of a hardware thing
No, like levi says it's a Kernel thing, the OMAP5's ARM Cortex A15s have KVM support, it just needs to be enabled in the kernel config before being compiled. I went through the motions of this a while ago, I just never got around to figure out all the userspace nonsense that I needed to do to get QEMU to utilize it. I don't believe this is anything aTc enables in the kernel that the Pyra currently uses.
 
Let me share my thoughts.
The glory future of the pyra will surely be based on the new RK3568 SOC.
However, it will take a few more years until everything is ready. So it's sad that the there are still so many issues with the OPMAP5 SOC.
Until we get the new processor board, I would still love to see any improvements with the current hardware.
Even if it is based on older Android versions or older Linux kernels, it would be great to have a OMAP5 based Pyra that makes more use of the processor. So any increase in performance or having more features like better power saving routines is well desired!
 
Even if it is based on older Android versions or older Linux kernels, it would be great to have a OMAP5 based Pyra that makes more use of the processor.
Debian already runs an old enough kernel (albeit LTS I think) That's had all of the currently discovered holes fixed in, at least all those that apply to such an old kernel. If you pick any older one and use the kernel sources to build it you'll be opening yourself up to issues.
 
Isnt GBA Fullspeed? I thought it were, i ditnt noticed it to be slow.
Yes it's fast enough but it's an early arm device so virtualization instead of emulation might work, for certain instructions. just an idea. maybe not practical and the pundits will surely disagree.
 
Don't forget the GBA is a two core device. It contains an ARM7TDMI chip as well as a Sharp chip which runs the 8088 instruction set. However, the Sharp chip is there mainly so it can run GB/GBC games, and I think GBA games only use it for it's additional musical tone generator. Perhaps that can be simulated more easily.
 
Debian already runs an old enough kernel (albeit LTS I think) That's had all of the currently discovered holes fixed in, at least all those that apply to such an old kernel. If you pick any older one and use the kernel sources to build it you'll be opening yourself up to issues.
Pyra uses kernel 5.6 from 2020. Quite new for a processor discontinued (for the consumer market) in 2012 IMHO. I was thinking of a linux environment from Texas Instruments as a base when the OMAP5 still was kind of supported...
So if I understand correctly, when someone would try to build a linux with kernel 3.17.8 for example, you run into nearly unsolvable issues due to the Pyras other components?
 
Last edited:
Well yesterday was friday so not that much time for Pyra Stuff,
Also this emulator needs quite some preparation if you want to have it “legal” , as you have to dump all your Arcade Boards, or buy some Arcade Machines first ..
you could also have it easer but this is would be a Sinn and would bring you whit the highway to hell

I will have a look this evening ^^
 
Well yesterday was friday so not that much time for Pyra Stuff,
Also this emulator needs quite some preparation if you want to have it “legal” , as you have to dump all your Arcade Boards, or buy some Arcade Machines first ..
you could also have it easer but this is would be a Sinn and would bring you whit the highway to hell

I will have a look this evening ^^
Don't ask for Android to get new emulator if you dan't have the ROMS.
 
I wasn’t the one who asked for Android,
Also my intention for the Pandora was only the Gameboy Emulator.. ,
I don’t want to start a discussion about the Roms the users got where dumped or download ..
 
So, after 2 pages of debate "we need Android to have more emulator", I took some (a lot) time to build PFBneo, to add CPS1/2/3 emulator to the repo....
Conclusion: 0 downloads after 12h.
Yep, but wait two months please as I haven't my unit yet :p
More seriously, thanks for the good work ! I'm happy seeing the repo growing day after days...I don't know how many Pyras are on the wild yet, but my interest is also fresh as at beginning. Sure my learning curve will increase a lot in the future...
 
So, after 2 pages of debate "we need Android to have more emulator", I took some (a lot) time to build PFBneo, to add CPS1/2/3 emulator to the repo....
Conclusion: 0 downloads after 12h.
i wanted to so bad..

But my question was answered: omap5 virtualization support is crap, but for wrong reason: the kernel. software issues are usually much more tractable than hardware ones. why can't we have arm32 KVM support via (3rd party) kernel module? probably not worth the effort, because the kernel internals have changed. some stupid reason like that related to technology churn.
 
Back
Top