Date: Fri, 8 Jan 2016 10:24:46 -0800 From: Gleb Smirnoff <glebius@FreeBSD.org> To: "Jonathan T. Looney" <jtl@freebsd.org> Cc: David Wolfskill <david@catwhisker.org>, current@freebsd.org Subject: Re: panic: sbappendstream 1 [head/amd64 @r293419] Message-ID: <20160108182446.GZ1906@FreeBSD.org> In-Reply-To: <D2B531DB.4F113%jlooney@juniper.net> References: <20160108140518.GM2781@albert.catwhisker.org> <D2B531DB.4F113%jlooney@juniper.net>
next in thread | previous in thread | raw e-mail | index | archive | help
On Fri, Jan 08, 2016 at 09:34:23AM -0500, Jonathan T. Looney wrote: J> The likely suspect here looks like r293405, which changed uipc_send() to J> use sbappendstream_locked() instead of sbappend_locked(). J> J> However, I can't explain *why* that change is causing this problem without J> further investigation. That is because sbappendstream() has invariant that socket buffer doesn't have any records in it. But, if control data is sent, which is possible for AF_UNIX, a record is made. Thus, we can't used optimized version for AF_UNIX, only for AF_INET. -- Totus tuus, Glebius.
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20160108182446.GZ1906>