linux-cifs.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: ronnie sahlberg <ronniesahlberg@gmail.com>
To: Pavel Shilovsky <piastryyy@gmail.com>
Cc: Ronnie Sahlberg <lsahlber@redhat.com>,
	linux-cifs <linux-cifs@vger.kernel.org>,
	Steve French <smfrench@gmail.com>,
	Pavel Shilovsky <pshilov@microsoft.com>
Subject: Re: [PATCH 5/6] cifs: add a timeout argument to wait_for_free_credits
Date: Fri, 8 Mar 2019 12:39:40 +1000	[thread overview]
Message-ID: <CAN05THR89uDK36mLCBwWAsm2piK2q3kkBJL_vDnnKdp+kXWEuQ@mail.gmail.com> (raw)
In-Reply-To: <CAKywueRFjXs160fV2m-jsNPtLnBvqaHWJmJZ+SN5+yhjbtVmtA@mail.gmail.com>

On Thu, Mar 7, 2019 at 4:33 AM Pavel Shilovsky <piastryyy@gmail.com> wrote:
>
> вт, 5 мар. 2019 г. в 21:45, Ronnie Sahlberg <lsahlber@redhat.com>:
> >
> > A negative timeout is the same as the current behaviour, i.e. no timeout.
> >
> > Signed-off-by: Ronnie Sahlberg <lsahlber@redhat.com>
> > ---
> >  fs/cifs/transport.c | 40 ++++++++++++++++++++++++++++++----------
> >  1 file changed, 30 insertions(+), 10 deletions(-)
> >
> > diff --git a/fs/cifs/transport.c b/fs/cifs/transport.c
> > index f4ec9635dab2..3263e8b3a57d 100644
> > --- a/fs/cifs/transport.c
> > +++ b/fs/cifs/transport.c
> > @@ -478,11 +478,18 @@ smb_send(struct TCP_Server_Info *server, struct smb_hdr *smb_buffer,
> >
> >  static int
> >  wait_for_free_credits(struct TCP_Server_Info *server, const int num_credits,
> > -                     const int flags, unsigned int *instance)
> > +                     const int timeout, const int flags,
> > +                     unsigned int *instance)
> >  {
> >         int rc;
> >         int *credits;
> >         int optype;
> > +       long int t;
> > +
> > +       if (timeout < 0)
> > +               t = MAX_JIFFY_OFFSET;
> > +       else
> > +               t = msecs_to_jiffies(timeout);
> >
> >         optype = flags & CIFS_OP_MASK;
> >
> > @@ -507,11 +514,16 @@ wait_for_free_credits(struct TCP_Server_Info *server, const int num_credits,
> >                 if (*credits < num_credits) {
> >                         spin_unlock(&server->req_lock);
> >                         cifs_num_waiters_inc(server);
> > -                       rc = wait_event_killable(server->request_q,
> > -                               has_credits(server, credits, num_credits));
> > +                       rc = wait_event_killable_timeout(server->request_q,
> > +                               has_credits(server, credits, num_credits), t);
> >                         cifs_num_waiters_dec(server);
> > -                       if (rc)
> > -                               return rc;
> > +                       if (!rc) {
> > +                               cifs_dbg(VFS, "wait timed out after %d ms\n",
> > +                                        timeout);
> > +                               return -EAGAIN;
>
> It would be hard to distinguish different EAGAIN reasons in the upper
> layers. As of now we use EAGAIN as retryable error indicating the
> upper layer to retry. This case is different because the retry most
> likely won't help and the sequential fallback is needed. In
> compound_send_recv() we return -ENOTSUPP if there is not enough
> credits to satisfy compounding and the number of requests in flight
> doesn't promise to get new credits in future.
>
> I suggest to return the same error code here (-ENOTSUPP). Probably
> this particular error code isn't ideal choice and suggestions to
> change it are welcome.

I will do that change.  ENOTSUPP might not be ideal but should be fine for now.
If/when we need to take special action and start serializing these
requests we can
pick a better errno.


>
> > +                       }
> > +                       if (rc == -ERESTARTSYS)
> > +                               return -ERESTARTSYS;
> >                         spin_lock(&server->req_lock);
> >                 } else {
> >                         if (server->tcpStatus == CifsExiting) {
> > @@ -537,12 +549,19 @@ wait_for_free_credits(struct TCP_Server_Info *server, const int num_credits,
> >                             *credits <= MAX_COMPOUND) {
> >                                 spin_unlock(&server->req_lock);
> >                                 cifs_num_waiters_inc(server);
> > -                               rc = wait_event_killable(server->request_q,
> > +                               rc = wait_event_killable_timeout(
> > +                                       server->request_q,
> >                                         has_credits(server, credits,
> > -                                                   MAX_COMPOUND + 1));
> > +                                                   MAX_COMPOUND + 1),
> > +                                       t);
> >                                 cifs_num_waiters_dec(server);
> > -                               if (rc)
> > -                                       return rc;
> > +                               if (!rc) {
> > +                                       cifs_dbg(VFS, "wait timed out after %d ms\n",
> > +                                                timeout);
> > +                                       return -EAGAIN;
> > +                               }
> > +                               if (rc == -ERESTARTSYS)
> > +                                       return -ERESTARTSYS;
> >                                 spin_lock(&server->req_lock);
> >                                 continue;
> >                         }
> > @@ -569,7 +588,8 @@ static int
> >  wait_for_free_request(struct TCP_Server_Info *server, const int flags,
> >                       unsigned int *instance)
> >  {
> > -       return wait_for_free_credits(server, 1, flags, instance);
> > +       return wait_for_free_credits(server, 1, -1, flags,
> > +                                    instance);
> >  }
> >
> >  int
> > --
> > 2.13.6
> >
>
> --
> Best regards,
> Pavel Shilovsky

  reply	other threads:[~2019-03-08  2:39 UTC|newest]

Thread overview: 14+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2019-03-06  4:15 [PATCH 0/6] cifs: simplify handling of credits for compounds Ronnie Sahlberg
2019-03-06  4:15 ` [PATCH 1/6] cifs: change wait_for_free_request() to take flags as argument Ronnie Sahlberg
2019-03-06  4:15 ` [PATCH 2/6] cifs: pass flags down into wait_for_free_credits() Ronnie Sahlberg
2019-03-06  4:15 ` [PATCH 3/6] cifs: wait_for_free_credits() make it possible to wait for >=1 credits Ronnie Sahlberg
2019-03-06  4:15 ` [PATCH 4/6] cifs: prevent starvation in wait_for_free_credits for multi-credit requests Ronnie Sahlberg
2019-03-06  4:15 ` [PATCH 5/6] cifs: add a timeout argument to wait_for_free_credits Ronnie Sahlberg
2019-03-06 17:32   ` Pavel Shilovsky
2019-03-08  2:39     ` ronnie sahlberg [this message]
2019-03-06  4:15 ` [PATCH 6/6] cifs: simplify how we handle credits in compond_send_recv() Ronnie Sahlberg
2019-03-06 17:26   ` Pavel Shilovsky
2019-03-08  2:58 [PATCH 0/6] cifs: simplify handling of credits for compounds Ronnie Sahlberg
2019-03-08  2:58 ` [PATCH 5/6] cifs: add a timeout argument to wait_for_free_credits Ronnie Sahlberg
2019-03-09  0:58   ` Pavel Shilovsky
     [not found]     ` <CAH2r5ms3METKgX9sMet44r0RKdVAEe_x=vuB_6JZobJn95x4_Q@mail.gmail.com>
2019-03-09  1:06       ` Pavel Shilovsky
2019-03-10  2:21     ` Steve French

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=CAN05THR89uDK36mLCBwWAsm2piK2q3kkBJL_vDnnKdp+kXWEuQ@mail.gmail.com \
    --to=ronniesahlberg@gmail.com \
    --cc=linux-cifs@vger.kernel.org \
    --cc=lsahlber@redhat.com \
    --cc=piastryyy@gmail.com \
    --cc=pshilov@microsoft.com \
    --cc=smfrench@gmail.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).