Is it safe to use wip-edison-3.19.5 branch?

[sandeepmvd]

Hi, Is it ok and is it safe to use wip-edison-3.19.5 branch for edison? When can we expect the official release?

[andy-shev]

I'm not sure WIP is a correct abbreviation there. Last commit dated end of May last year. You may instead try my unofficial eds branch (see my GH page, linux repository).

[sandeepmvd]

Thanks @andy-shev

The kernel on eds branch is 4.10. Are the edison patches ported to it? If yes, it would be awesome! :+1:

I will try and post the results.. Thanks...

[andy-shev]

@sandeep-mvd Not everything is working there. There are a lot of areas out of my scope. But some necessary minimum works fine. I would suggest to follow the discussion on https://communities.intel.com/thread/75472.

[agherzan]

I see now a branch that doesn't have the wip prefix. Is that supposed to mean that it is stable to be used?

[andy-shev]

@agherzan May I ask what is your use case for that old kernel?

[agherzan]

I don't have a usecase of an old kernel. I'm looking for a stable one that is officially supported.

[andy-shev]

Fair enough. As far as I know the latest that officially supported is based on v3.10.98.

[agherzan]

Is there any documentation on this? From the branching names 3.19 looks stable.

[andy-shev]

I dunno about documentation, I looked at the code, basically latest BSP for Edison. There is still v3.10.98 based kernel which makes my conclusion.

[agherzan]

So you are saying that edison-3.19 is not a stable branch, right?

[agherzan]

I'm really confused about 3.10.98. I try to apply aufs patches on top. but things fail because it seems this branch includes patches which were merged in later versions of kernel: https://github.com/torvalds/linux/commit/946e51f2bf37f1656916eb75bd0742ba33983c28 https://github.com/torvalds/linux/commit/eee5cc2702929fd41cce28058dc6d6717f723f87

The nice thing about edison-3.19 is that is already including aufs so it works out of the box but I don't know the dev status of this branch.

[htot]

edison-3.19.5 has last commit 4-jun-2015. wip-edison-3.19.5 has last commit 10-may-2016. Comparing them shows wip has about 64 additional commits in a one year period. You could say that both are pretty stable. Or dead.

[andy-shev]

@htot I like your ironic mood! :+1:

[agherzan]

@htot Sadly it sounds right :+1: Are there any maintainers of this repo who can confirm/inform?

[htot]

:-) I didn't want to hurt anybodies feelings. The difference between stable, finished code (no commits over time) and dead abandoned code (no commits over time) is only if it works for you or not. I just read this article https://www.linux.com/news/event/ELCE/2017/long-term-embedded-linux-maintenance-made-easier and concluded that it would probably be best that the work from this repository be integrated into upstream as soon as possible. I am really hoping that behind the scenes Intel developers are supporting Andy's work to bring Edison support into main line. That would probably be more productive than continue working on out-of-tree 3.10 or 3.19 (note that there is no preempt_rt for 3.19 and never will be, so for industrial use 3.19 will in most cases not be an appropriate choice).

[agherzan]

Currently I'm only interested in aufs support so 3.19 fixes my incompatibility but I totally agree with what you say.

[htot]

I don't know aufs. But when I applied preempt-rt to the 3.10.14 kernel I needed to disable aufs in the kernel config. So may be it has been backported into 3.10.14?

[agherzan]

Sounds like a mess...

[sandeepmvd]

Well with no option left, I finally had to ditch edison and move to beaglebone.. I can close this issue if no one has anything more to discuss.

[agherzan]

I think we can leave it open. Nothing was resolved.

[andy-shev]

@agherzan, I don't think it will be for any of v3.xx kernel. So, agree on @binary-nerd, issue can be closed.

[agherzan]

OK. We, at resin, are planning to drop support for this board anyway. So I guess I'm fine. (sad story)

[andy-shev]

@agherzan I dunno what resin means, though v4.12-rc1 of the kernel supports most of the stuff on Edison, which means any Linux distro can be run on it. Thus, the obvious question why are you so linked to Yocto based disto for this board?

[agherzan]

@andy-shev resin means https://resinos.io/ . And we use yocto to compile the same distro for multiple board types.

[andy-shev]

@agherzan I see. So, what prevents you to switch to vanilla kernel instead in meta-edison or whatever layer you have (I suppose you have layer per platform anyway) and build same distribution for all boards?

[andy-shev]

Since we have established https://github.com/edison-fw/meta-intel-edison for more than year, I think this question doesn't make sense anymore.