Re: [PATCH] rnbd-srv: remove 'dir' argument from rnbd_srv_rdma_ev

[Jinpu Wang <jinpu.wang@ionos.com>]

On Fri, Aug 26, 2022 at 1:38 PM Guoqing Jiang <guoqing.jiang@linux.dev> wrote:
>
>
>
> On 8/26/22 7:29 PM, Jinpu Wang wrote:
> > On Fri, Aug 26, 2022 at 1:26 PM Guoqing Jiang <guoqing.jiang@linux.dev> wrote:
> >>
> >>
> >> On 8/26/22 6:48 PM, Jinpu Wang wrote:
> >>> On Fri, Aug 26, 2022 at 10:11 AM Guoqing Jiang <guoqing.jiang@linux.dev> wrote:
> >>>> Since all callers (process_{read,write}) set id->dir, no need to
> >>>> pass 'dir' again.
> >>>>
> >>>> Signed-off-by: Guoqing Jiang <guoqing.jiang@linux.dev>
> >>>> ---
> >>>>    drivers/block/rnbd/rnbd-srv.c          | 9 ++++-----
> >>>>    drivers/block/rnbd/rnbd-srv.h          | 1 +
> >>>>    drivers/infiniband/ulp/rtrs/rtrs-srv.c | 4 ++--
> >>>>    drivers/infiniband/ulp/rtrs/rtrs.h     | 3 +--
> >>>>    4 files changed, 8 insertions(+), 9 deletions(-)
> >>>>
> >>>> diff --git a/drivers/block/rnbd/rnbd-srv.c b/drivers/block/rnbd/rnbd-srv.c
> >>>> index 3f6c268e04ef..9600715f1029 100644
> >>>> --- a/drivers/block/rnbd/rnbd-srv.c
> >>>> +++ b/drivers/block/rnbd/rnbd-srv.c
> >>>> @@ -368,10 +368,9 @@ static int process_msg_sess_info(struct rnbd_srv_session *srv_sess,
> >>>>                                    const void *msg, size_t len,
> >>>>                                    void *data, size_t datalen);
> >>>>
> >>>> -static int rnbd_srv_rdma_ev(void *priv,
> >>>> -                           struct rtrs_srv_op *id, int dir,
> >>>> -                           void *data, size_t datalen, const void *usr,
> >>>> -                           size_t usrlen)
> >>>> +static int rnbd_srv_rdma_ev(void *priv, struct rtrs_srv_op *id,
> >>>> +                           void *data, size_t datalen,
> >>>> +                           const void *usr, size_t usrlen)
> >>>>    {
> >>>>           struct rnbd_srv_session *srv_sess = priv;
> >>>>           const struct rnbd_msg_hdr *hdr = usr;
> >>>> @@ -398,7 +397,7 @@ static int rnbd_srv_rdma_ev(void *priv,
> >>>>                   break;
> >>>>           default:
> >>>>                   pr_warn("Received unexpected message type %d with dir %d from session %s\n",
> >>>> -                       type, dir, srv_sess->sessname);
> >>>> +                       type, id->dir, srv_sess->sessname);
> >>>>                   return -EINVAL;
> >>>>           }
> >>>>
> >>>> diff --git a/drivers/block/rnbd/rnbd-srv.h b/drivers/block/rnbd/rnbd-srv.h
> >>>> index 081bceaf4ae9..5a0ef6c2b5c7 100644
> >>>> --- a/drivers/block/rnbd/rnbd-srv.h
> >>>> +++ b/drivers/block/rnbd/rnbd-srv.h
> >>>> @@ -14,6 +14,7 @@
> >>>>    #include <linux/kref.h>
> >>>>
> >>>>    #include <rtrs.h>
> >>>> +#include <rtrs-srv.h>
> >>> why do we need this?
> >> Otherwise, compiler complains
> >>
> >> drivers/block/rnbd/rnbd-srv.c: In function ‘rnbd_srv_rdma_ev’:
> >> drivers/block/rnbd/rnbd-srv.c:400:33: error: invalid use of undefined
> >> type ‘struct rtrs_srv_op’
> >>     400 |                         type, id->dir, srv_sess->sessname);
> >>
> >> Thanks,
> >> Guoqing
> > ah, okay, this reminds me, why we have dir there, we don't want to
> > export too much detail regarding the rtrs_srv_op to
> > rnbd-server, it is supposed to be transparent  to rnbd-srv.
>
> What is the issue with more details are exported from rtrs-srv? Both of
> the modules
> are run in the same machine.
with including rtrs-srv.h, the code size is bigger, not to mention
many unnecessary
info are exported to rnbd-srv module, ideally we want to have
rnbd/rtrs layered properly
with clear separation.

>
> And I guess we can just pass parameters with register after remove an
> argument,
> otherwise need to push/pop stack with more than six parameters for x64.
I doubt it makes any notable performance change.
>
> Thanks,
> Guoqing
Thx!