batman-adv: Migration from staging to net

Hi,

the batman-adv developers had a smaller discussion what we wanted to achieve in drivers/staging and if we have done everything we wanted to do. We think that everything we wanted to do in staging is finished (the TODO file is empty since a while), but that is just our opinion.

So we wanted to ask you first if you see any showstopper which make batman-adv insufficient for net before we would ask the net maintainer. Our personal favorite kernel for the move would be 2.6.37. Before that we would have 2.6.35 (which is not completely correct using the sysfs infrastructure) and at least some RCs of 2.6.36 (which moves the remaining wrong stuff from sysfs to debugfs).

Open patches for 2.6.36 are currently (may change)
* batman-adv: Mark locally used symbols as static (removed some dead code)
* batman-adv: 32bit sequence number and TTL for broadcasts
* batman-adv: Add bonding functionality
* batman-adv: bonding and interface alternating
* batman-adv: record route for ICMP messages
* ...

Nearly all of them are features and not really things which makes it more suitable for the kernel. Open things are for example
* Re-Integration of some debugging infrastructure (currently not decided which technology to use)
* virtualized meshing (I am not sure if that was the same fancy term the actual developer used)
* forward error correction
* mesh optimized gateway support
* protocol optimizations

Again I wouldn't think that one of the mentioned points would make it better suitable for the kernel (but of course makes batman-adv better in general terms).

So is there anything you would want us to change before you would say that we could try a review by the net maintainer?

To the merge process itself. We should prepare a patch which adds everything in net/batman-adv/ (and Documentation/ABI/testing/, MAINTAINERS) and then send it to [email protected]/David S. Miller. After it was merged we just send you a patch which removes drivers/staging/batman-adv. Is that correct (just searched for an example and only found benet[1, 2])?

thanks,
Sven

[1] http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=shortlog;h=6b7c5b947c671a96e39f9526a5fd70c178b8dfd1
[2] http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=shortlog;h=d0573facf21d1e5cfbc1ddac272b7592722e6c01

If everything is now moved into debugfs, and everything in sysfs is only "one-value-per-file" then I don't really have an objection.

I would say let the networking developers start reviewing it.

good luck,

greg k-h

Greg KH wrote:
[...]

Thanks for the response.

Can please also make some comments about the last part of the mail (regarding the merge process):

thanks,
Sven

Oops, sorry about that.

Yes, a new patch would be easiest for the network developers to review. When it gets accepted we can then delete the drivers/staging/ code.

thanks,

greg k-h