netdev - Re: [PATCH v7 02/11] net: add ndo to setup/query xdp prog in adapter rx

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-Id: <20160711.234553.1347801109908530356.davem@davemloft.net>
Date:	Mon, 11 Jul 2016 23:45:53 -0700 (PDT)
From:	David Miller <davem@...emloft.net>
To:	bblanco@...mgrid.com
Cc:	netdev@...r.kernel.org, jhs@...atatu.com,
	saeedm@....mellanox.co.il, kafai@...com, brouer@...hat.com,
	as754m@....com, alexei.starovoitov@...il.com, gerlitz.or@...il.com,
	john.fastabend@...il.com, hannes@...essinduktion.org,
	tgraf@...g.ch, tom@...bertland.com, daniel@...earbox.net
Subject: Re: [PATCH v7 02/11] net: add ndo to setup/query xdp prog in
 adapter rx

From: Brenden Blanco <bblanco@...mgrid.com>
Date: Mon, 11 Jul 2016 23:35:00 -0700

> On Mon, Jul 11, 2016 at 11:12:24PM -0700, David Miller wrote:
>> From: Brenden Blanco <bblanco@...mgrid.com>
>> Date: Mon, 11 Jul 2016 14:29:49 -0700
>> 
>> > +	if (fd >= 0) {
>> > +		prog = bpf_prog_get_type(fd, BPF_PROG_TYPE_XDP);
>> > +		if (IS_ERR(prog))
>> > +			return PTR_ERR(prog);
>> > +	}
>> > +
>> > +	xdp.command = XDP_SETUP_PROG;
>> > +	xdp.prog = prog;
>> > +	err = ops->ndo_xdp(dev, &xdp);
>> > +	if (err < 0 && prog)
>> > +		bpf_prog_put(prog);
>> 
>> I don't understand the reference counting on 'prog' here.
>> 
>> The enumeration documentation states that no matter what, the passed
>> in prog doesn't need to be mangaged by the ->ndo_xdp() method.
>> 
>> Therefore, if that is true, we must always put the 'prog' here if it
>> is non-NULL.  Regardless of the 'err' value.
>> 
> The documentation is unclear then. What I _meant_ to say is that the callee
> is not responsible for putting the program on error, but on success it
> takes ownership of the reference. In context of that, does the code make
> sense? Is there a more conventional way of handling this?

Then please document it more clearly, thanks.