On Mon, Jun 22, 2009 at 12:12:56PM -0700, Dan Smith wrote:
> This patch adds AF_INET c/r support based on the framework established in
> my AF_UNIX patch. I've tested it by checkpointing a single app with a
> pair of sockets connected over loopback.
>
> I expect a pile of comments :)
>
> A couple points about the operation:
>
> 1. In order to properly hook up the established sockets with the matching
> listening parent socket, I added a new list to the ckpt_ctx and run the
> parent attachment in the deferqueue at the end of the restart process.
> 2. I don't do anything to redirect or freeze traffic flowing to or from the
> remote system (to prevent a RST from breaking things). I expect that
> userspace will bring down a veth device or freeze traffic to the remote
> system to handle this case.
>
> Cc: Oren Laaden <orenl@xxxxxxxxxxxxxxx>
> Cc: Alexey Dobriyan <adobriyan@xxxxxxxxx>
> Signed-off-by: Dan Smith <danms@xxxxxxxxxx>
> ---
> checkpoint/sys.c | 2 +
> include/linux/checkpoint_hdr.h | 1 +
> include/linux/checkpoint_types.h | 2 +
> include/linux/socket.h | 95 ++++++++++
> net/checkpoint.c | 369 +++++++++++++++++++++++++++++++++-----
> 5 files changed, 428 insertions(+), 41 deletions(-)
>
> diff --git a/checkpoint/sys.c b/checkpoint/sys.c
> index 38a5299..b6f18ea 100644
> --- a/checkpoint/sys.c
> +++ b/checkpoint/sys.c
> @@ -242,6 +242,8 @@ static struct ckpt_ctx *ckpt_ctx_alloc(int fd, unsigned long uflags,
> INIT_LIST_HEAD(&ctx->pgarr_pool);
> init_waitqueue_head(&ctx->waitq);
>
> + INIT_LIST_HEAD(&ctx->listen_sockets);
> +
> err = -EBADF;
> ctx->file = fget(fd);
> if (!ctx->file)
> diff --git a/include/linux/checkpoint_hdr.h b/include/linux/checkpoint_hdr.h
> index 46285f8..0a19767 100644
> --- a/include/linux/checkpoint_hdr.h
> +++ b/include/linux/checkpoint_hdr.h
> @@ -87,6 +87,7 @@ enum {
> CKPT_HDR_SOCKET_BUFFERS,
> CKPT_HDR_SOCKET_BUFFER,
> CKPT_HDR_SOCKET_UN,
> + CKPT_HDR_SOCKET_IN,
Nano-nit: I think it's easier to read if we call these CKPT_HDR_SOCKET_UNIX
and CKPT_HDR_SOCKET_INET. The extra 2 characters are worth that. This also
applies to the struct names later in the patch.
Cheers,
-Matt Helsley
_______________________________________________
Containers mailing list
Containers@xxxxxxxxxxxxxxxxxxxxxxxxxx
https://lists.linux-foundation.org/mailman/listinfo/containers
