Date: Mon, 22 Jun 2009 15:10:57 -0700 From: Kip Macy <kmacy@freebsd.org> To: Andre Oppermann <andre@freebsd.org> Cc: svn-src-head@freebsd.org, svn-src-all@freebsd.org, src-committers@freebsd.org, Sam Leffler <sam@freebsd.org> Subject: Re: svn commit: r194643 - head/sys/kern Message-ID: <3c1674c90906221510s9b62e25w3a5d41c21bd512e1@mail.gmail.com> In-Reply-To: <200906221935.n5MJZdLJ050266@svn.freebsd.org> References: <200906221935.n5MJZdLJ050266@svn.freebsd.org>
next in thread | previous in thread | raw e-mail | index | archive | help
On Mon, Jun 22, 2009 at 12:35 PM, Andre Oppermann<andre@freebsd.org> wrote: > Author: andre > Date: Mon Jun 22 19:35:39 2009 > New Revision: 194643 > URL: http://svn.freebsd.org/changeset/base/194643 > > Log: > =A0Update m_demote: > =A0- remove HT_HEADER test (MT_HEADER =3D=3D MT_DATA for some time now) > =A0- be more pedantic about m_nextpkt in other than first mbuf > =A0- update m_flags to be retained > > Modified: > =A0head/sys/kern/uipc_mbuf.c > > Modified: head/sys/kern/uipc_mbuf.c > =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D > --- head/sys/kern/uipc_mbuf.c =A0 Mon Jun 22 19:09:48 2009 =A0 =A0 =A0 = =A0(r194642) > +++ head/sys/kern/uipc_mbuf.c =A0 Mon Jun 22 19:35:39 2009 =A0 =A0 =A0 = =A0(r194643) > @@ -320,11 +320,13 @@ m_demote(struct mbuf *m0, int all) > =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0m->m_flags &=3D ~M_PKTHDR; > =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0bzero(&m->m_pkthdr, sizeof= (struct pkthdr)); > =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0} > - =A0 =A0 =A0 =A0 =A0 =A0 =A0 if (m->m_type =3D=3D MT_HEADER) > - =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 m->m_type =3D MT_DATA; > - =A0 =A0 =A0 =A0 =A0 =A0 =A0 if (m !=3D m0 && m->m_nextpkt !=3D NULL) > + =A0 =A0 =A0 =A0 =A0 =A0 =A0 if (m !=3D m0 && m->m_nextpkt !=3D NULL) { > + =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 KASSERT(m->m_nextpkt =3D=3D= NULL, > + =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 ("%s: m_nextpkt not= NULL", __func__)); > + =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 m_freem(m->m_nextpkt); > =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0m->m_nextpkt =3D NULL; > - =A0 =A0 =A0 =A0 =A0 =A0 =A0 m->m_flags =3D m->m_flags & (M_EXT|M_EOR|M_= RDONLY|M_FREELIST); > + =A0 =A0 =A0 =A0 =A0 =A0 =A0 } > + =A0 =A0 =A0 =A0 =A0 =A0 =A0 m->m_flags =3D m->m_flags & (M_EXT|M_RDONLY= |M_FREELIST|M_NOFREE); > =A0 =A0 =A0 =A0} > Freeing an mbuf that shouldn't be there is not a safe change. You don't know that m_nextpkt isn't pointing at some random value in memory and that doing so isn't going to lead to some inexplicable crash some time later. This is not a good strategy from a support standpoint. I've cc'd sam in case he feels differently. Cheers, Kip
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?3c1674c90906221510s9b62e25w3a5d41c21bd512e1>