Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) client-ip=23.128.96.18;
From:   David Laight <David.Laight@ACULAB.COM>
To:     'Matthew Wilcox' <willy@infradead.org>,
        Jakub Kicinski <kuba@kernel.org>
CC:     Shoaib Rao <rao.shoaib@oracle.com>,
        "linux-kernel@vger.kernel.org" <linux-kernel@vger.kernel.org>,
        "linux-api@vger.kernel.org" <linux-api@vger.kernel.org>,
        "netdev@vger.kernel.org" <netdev@vger.kernel.org>,
        "David S. Miller" <davem@davemloft.net>,
        "andy.rudoff@intel.com" <andy.rudoff@intel.com>
Subject: RE: [PATCH] af_unix: Allow Unix sockets to raise SIGURG
Thread-Topic: [PATCH] af_unix: Allow Unix sockets to raise SIGURG
Thread-Index: AQHW9oaDqv48RJCczk2sHQ3JiT7IG6o/Q1OQ
Date:   Fri, 29 Jan 2021 23:47:14 +0000
Message-ID: <ee13e83b22b7411c97a2a961015343d1@AcuMS.aculab.com>
References: <20210122150638.210444-1-willy@infradead.org>
 <20210125153650.18c84b1a@kicinski-fedora-pc1c0hjn.dhcp.thefacebook.com>
 <23fc3de2-7541-04c9-a56f-4006a7dc773f@oracle.com>
 <20210129110605.54df8409@kicinski-fedora-pc1c0hjn.dhcp.thefacebook.com>
 <a21dc26a-87dc-18c8-b8bd-24f9797afbad@oracle.com>
 <20210129120250.269c366d@kicinski-fedora-pc1c0hjn.dhcp.thefacebook.com>
 <cef52fb0-43cb-9038-7e48-906b58b356b6@oracle.com>
 <20210129121837.467280fb@kicinski-fedora-pc1c0hjn.dhcp.thefacebook.com>
 <e1047be3-2d53-49d3-67b4-a2a99e0c0f0f@oracle.com>
 <20210129131820.4b97fdeb@kicinski-fedora-pc1c0hjn.dhcp.thefacebook.com>
 <20210129213217.GD308988@casper.infradead.org>
In-Reply-To: <20210129213217.GD308988@casper.infradead.org>
Accept-Language: en-GB, en-US
MIME-Version: 1.0
Content-Language: en-US
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8BIT
Precedence: bulk

> I'd encourage anyone thinking about "using OOB" to read
> https://tools.ietf.org/html/rfc6093 first.  Basically, TCP does not
> actually provide an OOB mechanism, and frankly Unix sockets shouldn't
> try either.

OOB data maps much better onto ISO transport 'expedited data'
than anything in a bytestream protocol like TCP.
There you can send a message (it is message oriented) that isn't
subject to normal data flow control.
The length is limited (IIRC 32 bytes) and expedited data has
its own credit of one, but can overtake (and is expected to
overtake) flow control blocked normal data.

All TCP provides is a byte sequence number for OOB data.
This is just a marker in the bytestream.
It really doesn't map onto the socket OOB data data all.

	David

-
Registered Address Lakeside, Bramley Road, Mount Farm, Milton Keynes, MK1 1PT, UK
Registration No: 1397386 (Wales)