[Devel] Re: Getting the new RxRPC patches upstream

[David Howells]

David Miller <davem at davemloft.net> wrote:

> Is it possible for your changes to be purely networking
> and not need those changes outside of the networking?

See my latest patchset release.  I've reduced the dependencies on
non-networking changes to:

 (1) Oleg Nesterov's patch to change cancel_delayed_work() to use del_timer()
     rather than del_timer_sync() [patch 02/16].

     This patch can be discarded without compilation failure at the expense of
     making AFS slightly less efficient. It also makes AF_RXRPC slightly less
     efficient, but only in the rmmod path.

 (2) A symbol export in the keyring stuff plus a proliferation of the types
     available in the struct key::type_data union [patch 03/16].  This does
     not conflict with any other patches that I know about.

 (3) A symbol export in the timer stuff [patch 04/16].

Everything else that remains after the reduction is confined to the AF_RXRPC
or AFS code, save for a couple of networking patches in my patchset that you
already have and I just need to make the thing compile.

I'm not sure that I can make the AF_RXRPC patches totally independent of the
AFS patches as the two sets need to interleave since the last AF_RXRPC patch
deletes the old RxRPC code - which the old AFS code depends on.

David
_______________________________________________
Containers mailing list
Containers at lists.linux-foundation.org
https://lists.linux-foundation.org/mailman/listinfo/containers