Date: Fri, 17 Apr 2026 11:58:09 +0200 From: Kristof Provost <kp@FreeBSD.org> To: Andrew Gallatin <gallatin@FreeBSD.org> Cc: src-committers@FreeBSD.org, dev-commits-src-all@FreeBSD.org, dev-commits-src-main@FreeBSD.org Subject: Re: git: 956acdce0505 - main - loopback: Account for packet drops Message-ID: <CAEBDB5C-26CC-4F4F-8B27-2E0CC25AAFF2@FreeBSD.org> In-Reply-To: <69de99dd.422ab.2b84d627@gitrepo.freebsd.org>
index | next in thread | previous in thread | raw e-mail
On 14 Apr 2026, at 21:47, Andrew Gallatin wrote: > The branch main has been updated by gallatin: > > URL: https://cgit.FreeBSD.org/src/commit/?id=956acdce0505ca8028d287d3b44789c623c8f37e > > commit 956acdce0505ca8028d287d3b44789c623c8f37e > Author: Andrew Gallatin <gallatin@FreeBSD.org> > AuthorDate: 2026-04-14 19:43:28 +0000 > Commit: Andrew Gallatin <gallatin@FreeBSD.org> > CommitDate: 2026-04-14 19:43:28 +0000 > > loopback: Account for packet drops > > Make loopback packet drops more obvious by reporting them > in interface stats visable via netstat -ni > Since loopback uses netisr, packets can be dropped if the > netisr queue overflows. These drops are visible via > netisr -Q, but its not an obvious place to look. > > Differential Revision: https://reviews.freebsd.org/D56356 > Reviewed by: glebius, tuexen > Sponsored by: Netflix > --- > sys/net/if_loop.c | 10 +++++++--- > 1 file changed, 7 insertions(+), 3 deletions(-) > > diff --git a/sys/net/if_loop.c b/sys/net/if_loop.c > index 37309260a0d3..2ff265d5d1e7 100644 > --- a/sys/net/if_loop.c > +++ b/sys/net/if_loop.c > @@ -350,9 +350,13 @@ if_simloop(struct ifnet *ifp, struct mbuf *m, int af, int hlen) > m_freem(m); > return (EAFNOSUPPORT); > } > - if_inc_counter(ifp, IFCOUNTER_IPACKETS, 1); > - if_inc_counter(ifp, IFCOUNTER_IBYTES, m->m_pkthdr.len); > - netisr_queue(isr, m); /* mbuf is free'd on failure. */ > + if (netisr_queue(isr, m) == 0) { > + if_inc_counter(ifp, IFCOUNTER_IPACKETS, 1); > + if_inc_counter(ifp, IFCOUNTER_IBYTES, m->m_pkthdr.len); I’m seeing reproducible KASAN use-after-free warnings here (running the pf tests). Which makes sense, because after the netisr_queue() call we no longer own the mbuf. I think we need something like this: diff --git a/sys/net/if_loop.c b/sys/net/if_loop.c index 2ff265d5d1e7..33ddd3a8540e 100644 --- a/sys/net/if_loop.c +++ b/sys/net/if_loop.c @@ -276,6 +276,7 @@ int if_simloop(struct ifnet *ifp, struct mbuf *m, int af, int hlen) { int isr; + int32_t len; M_ASSERTPKTHDR(m); m_tag_delete_nonpersistent(m); @@ -350,9 +351,10 @@ if_simloop(struct ifnet *ifp, struct mbuf *m, int af, int hlen) m_freem(m); return (EAFNOSUPPORT); } + len = m->m_pkthdr.len; if (netisr_queue(isr, m) == 0) { if_inc_counter(ifp, IFCOUNTER_IPACKETS, 1); - if_inc_counter(ifp, IFCOUNTER_IBYTES, m->m_pkthdr.len); + if_inc_counter(ifp, IFCOUNTER_IBYTES, len); } else { /* mbuf is free'd on failure. */ if_inc_counter(ifp, IFCOUNTER_IQDROPS, 1); Best regards, Kristofhome | help
Want to link to this message? Use this
URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?CAEBDB5C-26CC-4F4F-8B27-2E0CC25AAFF2>