Received-SPF: pass (google.com: best guess record for domain of linux-nfs-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) client-ip=209.132.180.67;
Content-Type: text/plain;
        charset=us-ascii
Mime-Version: 1.0 (Mac OS X Mail 12.4 \(3445.104.11\))
Subject: Re: [PATCH 1/2] SUNRPC: Fix up backchannel slot table accounting
From:   Chuck Lever <chuck.lever@oracle.com>
In-Reply-To: <97e9839faef3d1bc901d4ced3d0cf2e0bf2a0bd1.camel@hammerspace.com>
Date:   Wed, 17 Jul 2019 14:12:56 -0400
Cc:     Linux NFS Mailing List <linux-nfs@vger.kernel.org>
Content-Transfer-Encoding: quoted-printable
Message-Id: <F138F86C-84FB-4FEA-8240-FDAD6FD4CE38@oracle.com>
References: <20190716200157.38583-1-trond.myklebust@hammerspace.com>
 <99A569FB-DD7F-4547-AB06-FEB5DABA8488@oracle.com>
 <97e9839faef3d1bc901d4ced3d0cf2e0bf2a0bd1.camel@hammerspace.com>
To:     Trond Myklebust <trondmy@hammerspace.com>
Sender: linux-nfs-owner@vger.kernel.org
Precedence: bulk


> On Jul 17, 2019, at 1:19 PM, Trond Myklebust <trondmy@hammerspace.com> =
wrote:
>=20
> On Wed, 2019-07-17 at 09:55 -0400, Chuck Lever wrote:
>> Hi Trond-
>>=20
>>> On Jul 16, 2019, at 4:01 PM, Trond Myklebust <trondmy@gmail.com>
>>> wrote:
>>>=20
>>> Add a per-transport maximum limit in the socket case, and add
>>> helpers to allow the NFSv4 code to discover that limit.
>>=20
>> For RDMA, the number of credits is permitted to change during the
>> life
>> of the connection, so this is not a fixed limit for such transports.
>=20
> This is defining a maximum value, which is used for backchannel =
session
> slot negotiation.
>=20
>>=20
>> And, AFAICT, it's not necessary to know the transport's limit. The
>> lesser of the NFS backchannel and RPC/RDMA reverse credit limit will
>> be used.
>=20
> The server needs to know how many requests it can send in parallel on
> the back channel. If it sends too many, which it can and will do on
> TCP, then we currently break the connection, and so callbacks end up
> being dropped or missed altogether.

IIUC, RPC/RDMA has a fixed maximum constant. Would you like a patch at
some point that advertises that constant via ->bc_num_slots ?


--
Chuck Lever