Received-SPF: pass (google.com: domain of linux-nfs+bounces-1442-linux.lists.archive=gmail.com@vger.kernel.org designates 147.75.80.249 as permitted sender) client-ip=147.75.80.249;
From: Benjamin Coddington <bcodding@redhat.com>
To: Chuck Lever <cel@kernel.org>
Cc: linux-nfs@vger.kernel.org
Subject: Re: [PATCH RFC 00/13] NFSD backchannel fixes
Date: Thu, 25 Jan 2024 17:09:27 -0500
Message-ID: <0DCE1190-19FA-46BD-822D-6984F0B5B296@redhat.com>
In-Reply-To: <170619984210.2833.7173004255003914651.stgit@manet.1015granger.net>
References: <170619984210.2833.7173004255003914651.stgit@manet.1015granger.net>
Precedence: bulk
MIME-Version: 1.0
Content-Type: text/plain

On 25 Jan 2024, at 11:28, Chuck Lever wrote:

> The first three patches fix bugs that prevent NFSD's backchannel
> from reliably retransmitting after a client reconnects. These fixes
> might be appropriate for 6.8-rc.
>
> Following that are some new trace points that might be helpful for
> field troubleshooting.
>
> Then there are some minor clean-ups.
>
> I am still testing this series, and there is one msleep() call that
> needs some thought. Thoughts, comments, opinions, rotten fruit? You
> know the drill.
>
> ---
>
> Chuck Lever (13):
>       NFSD: Reset cb_seq_status after NFS4ERR_DELAY
>       NFSD: Reschedule CB operations when backchannel rpc_clnt is shut down
>       NFSD: Retransmit callbacks after client reconnects
>       NFSD: Add nfsd_seq4_status trace event
>       NFSD: Replace dprintks in nfsd4_cb_sequence_done()
>       NFSD: Rename nfsd_cb_state trace point
>       NFSD: Add callback operation lifetime trace points
>       SUNRPC: Remove EXPORT_SYMBOL_GPL for svc_process_bc()
>       NFSD: Remove unused @reason argument
>       NFSD: Replace comment with lockdep assertion
>       NFSD: Remove BUG_ON in nfsd4_process_cb_update()
>       SUNRPC: Remove stale comments
>       NFSD: Remove redundant cb_seq_status initialization
>
>
>  fs/nfsd/nfs4callback.c   |  81 +++++++++++++-------
>  fs/nfsd/nfs4state.c      |   1 +
>  fs/nfsd/trace.h          | 162 ++++++++++++++++++++++++++++++++++++++-
>  include/trace/misc/nfs.h |  34 ++++++++
>  net/sunrpc/svc.c         |   1 -
>  net/sunrpc/xprtsock.c    |   9 ---
>  6 files changed, 250 insertions(+), 38 deletions(-)


These are great, looking forward to see how 02/13 waits for reconnection.
Seems like a wait_on_bit or wait_on_var triggered from nfsd4_init_conn()
would do, but that's just my wild speculation.

Ben