From mboxrd@z Thu Jan  1 00:00:00 1970
From: Al Viro <viro@ZenIV.linux.org.uk>
Subject: Re: [PATCH 16/17] get rid of the size argument of sock_sendmsg()
Date: Tue, 14 Apr 2015 17:44:28 +0100
Message-ID: <20150414164427.GA889@ZenIV.linux.org.uk>
References: <20150411211742.GJ889@ZenIV.linux.org.uk>
 <1428787108-13650-16-git-send-email-viro@ZenIV.linux.org.uk>
 <063D6719AE5E284EB5DD2968C1650D6D1CB1DDB2@AcuExch.aculab.com>
 <20150414163534.GZ889@ZenIV.linux.org.uk>
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Cc: "davem@davemloft.net" <davem@davemloft.net>,
	"netdev@vger.kernel.org" <netdev@vger.kernel.org>
To: David Laight <David.Laight@ACULAB.COM>
Return-path: <netdev-owner@vger.kernel.org>
Received: from zeniv.linux.org.uk ([195.92.253.2]:59961 "EHLO
	ZenIV.linux.org.uk" rhost-flags-OK-OK-OK-OK) by vger.kernel.org
	with ESMTP id S932896AbbDNQob (ORCPT
	<rfc822;netdev@vger.kernel.org>); Tue, 14 Apr 2015 12:44:31 -0400
Content-Disposition: inline
In-Reply-To: <20150414163534.GZ889@ZenIV.linux.org.uk>
Sender: netdev-owner@vger.kernel.org
List-ID: <netdev.vger.kernel.org>

On Tue, Apr 14, 2015 at 05:35:34PM +0100, Al Viro wrote:
> On Tue, Apr 14, 2015 at 04:25:24PM +0000, David Laight wrote:
> > From: Al Viro
> > > Sent: 11 April 2015 22:18
> > > it's equal to iov_iter_count(&msg->msg_iter) in all cases
> > 
> > I don't know whether this is guaranteed for iov[] that come from
> > userspace.
> >
> > In any case iov_iter_count() is non-trivial and you don't
> > really want to call it when unnecessary.
> 
> Really?
> 
> static inline size_t iov_iter_count(struct iov_iter *i)
> {
>         return i->count;
> }

Incidentally, have you even looked at the patch?  sock_write_iter():
-       res = sock_sendmsg(sock, &msg, iov_iter_count(from));
+       res = sock_sendmsg(sock, &msg);
sys_sendto():
-       err = sock_sendmsg(sock, &msg, iov_iter_count(&msg.msg_iter));
+       err = sock_sendmsg(sock, &msg);
__sys_sendmsg():
-       total_len = iov_iter_count(&msg_sys->msg_iter);
-       err = sock_sendmsg(sock, msg_sys, total_len);
+       err = sock_sendmsg(sock, msg_sys);
kernel_sendmsg():
        iov_iter_kvec(&msg->msg_iter, WRITE | ITER_KVEC, vec, num, size);
-       return sock_sendmsg(sock, msg, size);
+       return sock_sendmsg(sock, msg);
... and iov_iter_kvec() sets ->count to its last argument, so after the
first line we'll have iov_iter_count(&msg->msg_iter) == size.
svc_sendto():
-               if (sock_sendmsg(sock, &msg, 0) < 0)
+               if (sock_sendmsg(sock, &msg) < 0)
and msg->msg_iter is all-zeroes there, including ->msg_iter.count.

In cases when iov comes from userland we used to have iov_iter_count(...)
passed there; no need to even look its definition up.  It's an equivalent
transformation, no matter what...