Date: Mon, 22 Apr 2013 14:48:47 -0400 From: Juan Mojica <jmojica@gmail.com> To: Gleb Smirnoff <glebius@freebsd.org> Cc: FreeBSD Net <freebsd-net@freebsd.org> Subject: Re: ARP: Error Message in if_ether.c "arprequest: cannot find matching address" Message-ID: <CAPKuH-wQnyxdddMozTN328V3U_GLpiCOp-nuKiYEKXO-FWhmVg@mail.gmail.com> In-Reply-To: <20130422095759.GC76816@FreeBSD.org> References: <CAPKuH-w4ugf4KSWBKMoURbNsSrs1O-XNmWdn47VKn9MkCOgRDw@mail.gmail.com> <20130422095759.GC76816@FreeBSD.org>
next in thread | previous in thread | raw e-mail | index | archive | help
Thanks Gleb. I do not believe this path can triggered by an incoming
frame. I am planning on adding a KASSERT here to get a core when we
exercise this code in our debug images.
-Juan
On Mon, Apr 22, 2013 at 5:57 AM, Gleb Smirnoff <glebius@freebsd.org> wrote:
> On Wed, Apr 17, 2013 at 08:46:42AM -0400, Juan Mojica wrote:
> J> We manage to hit the following message with some regularity.
> J>
> J> arprequest: cannot find matching address
> J>
> J> The code shows a printf:
> J>
> J> printf("%s: cannot find matching address\n", __func__);
> J>
> J>
> J> Any reason this is a printf and not a
> J>
> J> log(LOG_ERR,
> J>
> J> The only things I can come up with are:
> J>
> J> a) it is a really severe and should be printed out, which if that is the
> J> case why isn't there an assert there?
> J> b) whoops, that should probably be a log(LOG_ERR,
> J> On our end we need to figure out exactly why we're intermittently
> hitting
> J> this patch of code.
>
> Can you please try this patch? Let's see what's going on.
>
> printf()ing in kernel is especially unsafe when the event can be triggered
> remotely. The arprequest() is called on output path of a packet, however
> I'm not sure that it can't be triggered by incoming packet.
>
> --
> Totus tuus, Glebius.
>
--
Juan Mojica
Email: jmojica@gmail.com
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?CAPKuH-wQnyxdddMozTN328V3U_GLpiCOp-nuKiYEKXO-FWhmVg>