Date: Thu, 07 Apr 2011 12:53:27 -0700 From: Julian Elischer <julian@freebsd.org> To: Marko Zec <zec@freebsd.org> Cc: svn-src-head@freebsd.org, svn-src-all@freebsd.org, Gleb Smirnoff <glebius@freebsd.org>, src-committers@freebsd.org Subject: Re: svn commit: r220416 - head/sys/netgraph Message-ID: <4D9E1637.1000200@freebsd.org> In-Reply-To: <201104071615.49994.zec@freebsd.org> References: <201104071140.p37BeAlE039248@svn.freebsd.org> <20110407135440.GX84445@FreeBSD.org> <201104071615.49994.zec@freebsd.org>
next in thread | previous in thread | raw e-mail | index | archive | help
On 4/7/11 7:15 AM, Marko Zec wrote: > On Thursday 07 April 2011 15:54:40 Gleb Smirnoff wrote: >> On Thu, Apr 07, 2011 at 11:40:10AM +0000, Marko Zec wrote: >> M> Author: zec >> M> Date: Thu Apr 7 11:40:10 2011 >> M> New Revision: 220416 >> M> URL: http://svn.freebsd.org/changeset/base/220416 >> M> >> M> Log: >> M> Properly unref ng_hub nodes on shutdown, so that we don't leak them. >> M> >> M> MFC after: 3 days >> M> >> M> Modified: >> M> head/sys/netgraph/ng_hub.c >> M> >> M> Modified: head/sys/netgraph/ng_hub.c >> M> >> =========================================================================== >> === M> --- head/sys/netgraph/ng_hub.c Thu Apr 7 11:13:50 2011 (r220415) M> >> +++ head/sys/netgraph/ng_hub.c Thu Apr 7 11:40:10 2011 (r220416) M> @@ >> -157,6 +157,8 @@ ng_hub_shutdown(node_p node) >> M> const priv_p priv = NG_NODE_PRIVATE(node); >> M> >> M> free(priv, M_NETGRAPH_HUB); >> M> + NG_NODE_SET_PRIVATE(node, NULL); >> M> + NG_NODE_UNREF(node); >> M> return (0); >> M> } >> >> Do we really need NG_NODE_SET_PRIVATE()? > Most probably not, though the majority of .shutdown method implementations in > sys/netgraph do include a call to NG_NODE_SET_PRIVATE(node, NULL), so I did a > C/P from one of those for consistency. If there's a consensus that > NG_NODE_SET_PRIVATE(node, NULL) in shutdown methods serves no useful purpose > then we should do a sweep across all the existing nodes... I'd rather do it than not do it. private data can refer to stuff that has been malloc'd and at one stage I had debug code that checked for NULL there to catch leaks there's no speed advantage to shutting down faster.. :-) > Marko >
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?4D9E1637.1000200>