On Wed, 3 Jun 2015 13:07:34 -0400
Chuck Lever <chuck.lever@xxxxxxxxxx> wrote:
>
> On Jun 3, 2015, at 11:01 AM, Jeff Layton <jlayton@xxxxxxxxxxxxxxx> wrote:
>
> > On Wed, 3 Jun 2015 10:48:10 -0400
> > Trond Myklebust <trond.myklebust@xxxxxxxxxxxxxxx> wrote:
> >
> >> On Wed, Jun 3, 2015 at 10:43 AM, Jeff Layton <jlayton@xxxxxxxxxxxxxxx> wrote:
> >>> RDMA xprts don't have a sock_xprt, but an rdma_xprt, so the
> >>> xs_swapper_enable/disable functions will likely oops when fed an RDMA
> >>> xprt. Turn these functions into rpc_xprt_ops so that that doesn't
> >>> occur. For now the RDMA versions are no-ops.
> >>>
> >>> Cc: Chuck Lever <chuck.lever@xxxxxxxxxx>
> >>> Signed-off-by: Jeff Layton <jeff.layton@xxxxxxxxxxxxxxx>
> >>> ---
> >>> include/linux/sunrpc/xprt.h | 16 ++++++++++++++--
> >>> net/sunrpc/clnt.c | 4 ++--
> >>> net/sunrpc/xprtrdma/transport.c | 15 ++++++++++++++-
> >>> net/sunrpc/xprtsock.c | 31 +++++++++++++++++++++++++------
> >>> 4 files changed, 55 insertions(+), 11 deletions(-)
> >>>
> >>> diff --git a/include/linux/sunrpc/xprt.h b/include/linux/sunrpc/xprt.h
> >>> index 26b1624128ec..7eb58610eb94 100644
> >>> --- a/include/linux/sunrpc/xprt.h
> >>> +++ b/include/linux/sunrpc/xprt.h
> >>> @@ -133,6 +133,8 @@ struct rpc_xprt_ops {
> >>> void (*close)(struct rpc_xprt *xprt);
> >>> void (*destroy)(struct rpc_xprt *xprt);
> >>> void (*print_stats)(struct rpc_xprt *xprt, struct seq_file *seq);
> >>> + int (*enable_swap)(struct rpc_xprt *xprt);
> >>> + void (*disable_swap)(struct rpc_xprt *xprt);
> >>> };
> >>>
> >>> /*
> >>> @@ -327,6 +329,18 @@ static inline __be32 *xprt_skip_transport_header(struct rpc_xprt *xprt, __be32 *
> >>> return p + xprt->tsh_size;
> >>> }
> >>>
> >>> +static inline int
> >>> +xprt_enable_swap(struct rpc_xprt *xprt)
> >>> +{
> >>> + return xprt->ops->enable_swap(xprt);
> >>> +}
> >>> +
> >>> +static inline void
> >>> +xprt_disable_swap(struct rpc_xprt *xprt)
> >>> +{
> >>> + xprt->ops->disable_swap(xprt);
> >>> +}
> >>> +
> >>> /*
> >>> * Transport switch helper functions
> >>> */
> >>> @@ -345,8 +359,6 @@ void xprt_release_rqst_cong(struct rpc_task *task);
> >>> void xprt_disconnect_done(struct rpc_xprt *xprt);
> >>> void xprt_force_disconnect(struct rpc_xprt *xprt);
> >>> void xprt_conditional_disconnect(struct rpc_xprt *xprt, unsigned int cookie);
> >>> -int xs_swapper_enable(struct rpc_xprt *xprt);
> >>> -void xs_swapper_disable(struct rpc_xprt *xprt);
> >>>
> >>> bool xprt_lock_connect(struct rpc_xprt *, struct rpc_task *, void *);
> >>> void xprt_unlock_connect(struct rpc_xprt *, void *);
> >>> diff --git a/net/sunrpc/clnt.c b/net/sunrpc/clnt.c
> >>> index 804a75e71e84..60d1835edb26 100644
> >>> --- a/net/sunrpc/clnt.c
> >>> +++ b/net/sunrpc/clnt.c
> >>> @@ -2492,7 +2492,7 @@ retry:
> >>> goto retry;
> >>> }
> >>>
> >>> - ret = xs_swapper_enable(xprt);
> >>> + ret = xprt_enable_swap(xprt);
> >>> xprt_put(xprt);
> >>> }
> >>> return ret;
> >>> @@ -2519,7 +2519,7 @@ retry:
> >>> goto retry;
> >>> }
> >>>
> >>> - xs_swapper_disable(xprt);
> >>> + xprt_disable_swap(xprt);
> >>> xprt_put(xprt);
> >>> }
> >>> }
> >>> diff --git a/net/sunrpc/xprtrdma/transport.c b/net/sunrpc/xprtrdma/transport.c
> >>> index 54f23b1be986..e7a157754095 100644
> >>> --- a/net/sunrpc/xprtrdma/transport.c
> >>> +++ b/net/sunrpc/xprtrdma/transport.c
> >>> @@ -682,6 +682,17 @@ static void xprt_rdma_print_stats(struct rpc_xprt *xprt, struct seq_file *seq)
> >>> r_xprt->rx_stats.bad_reply_count);
> >>> }
> >>>
> >>> +static int
> >>> +xprt_rdma_enable_swap(struct rpc_xprt *xprt)
> >>> +{
> >>> + return 0;
> >>
> >> Shouldn't the function be returning an error here? What does swapon
> >> expect if the device you are trying to enable doesn't support swap?
> >>
> >
> >
> > Chuck suggested making these no-ops for RDMA for now.
>
> I did indeed. What I meant was that you needn’t worry too much right now
> about how swap-on-NFS/RDMA is supposed to work, just make it not crash, and
> someone (maybe me) will look at it later to ensure it is working correctly
> and then we can claim it is supported. Sorry I was not clear.
>
> > I'm fine with
> > returning an error, but is it really an error? Maybe RDMA doesn't need
> > any special setup for swapping?
>
> This sounds a little snarky, but we don’t know for sure that nothing is
> needed until it is tested and reviewed. I think it’s reasonable to assume
> it doesn’t work 100% until we have positive confirmation that it does work.
>
> Maybe add a comment to that effect in these new xprt methods? And I would
> have it return something like ENOSYS.
>
> Likewise, consider the same return code here:
>
> +#if IS_ENABLED(CONFIG_SUNRPC_SWAP)
> +int rpc_clnt_swap_activate(struct rpc_clnt *clnt);
> +void rpc_clnt_swap_deactivate(struct rpc_clnt *clnt);
> +#else
> +static inline int
> +rpc_clnt_swap_activate(struct rpc_clnt *clnt)
> +{
> + return 0;
> ^^^^
> +}
>
> I’m not familiar enough with the swapon administrative interface to know if
> “swapping on this device is not supported” is a reasonable and expected
> failure mode for swapon. So maybe I’m just full of turtles.
>
No worries. I'm fine with returning an error if this stuff is disabled.
The manpage seems to indicate that EINVAL is the right error code to
use, but I'll see if I can verify that.
I'll need to look over the code a little more...
--
Jeff Layton <jlayton@xxxxxxxxxxxxxxx>
--
To unsubscribe from this list: send the line "unsubscribe linux-nfs" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
