| Message ID | 20230323060451.24280-1-vladimir@nikishkin.pw (mailing list archive) |
|---|---|
| State | Superseded |
| Delegated to: | David Ahern |
| Headers | show |
| Series | [iproute2-next,v1,1/1,v1] ip-link: add support for nolocalbypass in vxlan | expand |
| Context | Check | Description |
|---|---|---|
| netdev/tree_selection | success | Not a local patch |
Hi Vladimir, For the subject prefix, [PATCH iproute2-next] is enough for the v1 patch. On Thu, Mar 23, 2023 at 02:04:51PM +0800, Vladimir Nikishkin wrote: > Add userspace support for the nolocalbypass vxlan netlink > attribute. With nolocalbypass, if an entry is pointing to the > local machine, but the system driver is not listening on this > port, the driver will not drop packets, but will forward them > to the userspace network stack instead. > > This commit has a corresponding patch in the net-next list. > > Signed-off-by: Vladimir Nikishkin <vladimir@nikishkin.pw> > --- > include/uapi/linux/if_link.h | 1 + > ip/iplink_vxlan.c | 18 ++++++++++++++++++ > man/man8/ip-link.8.in | 8 ++++++++ > 3 files changed, 27 insertions(+) > > diff --git a/include/uapi/linux/if_link.h b/include/uapi/linux/if_link.h > index d61bd32d..fd390b40 100644 > --- a/include/uapi/linux/if_link.h > +++ b/include/uapi/linux/if_link.h > @@ -824,6 +824,7 @@ enum { > IFLA_VXLAN_TTL_INHERIT, > IFLA_VXLAN_DF, > IFLA_VXLAN_VNIFILTER, /* only applicable with COLLECT_METADATA mode */ > + IFLA_VXLAN_LOCALBYPASS, > __IFLA_VXLAN_MAX > }; > #define IFLA_VXLAN_MAX (__IFLA_VXLAN_MAX - 1) There is no need to include the uapi header. Stephen will sync it with upstream. Hi Stephen, should we add this note to the README.devel? > diff --git a/ip/iplink_vxlan.c b/ip/iplink_vxlan.c > index c7e0e1c4..17fa5cf7 100644 > --- a/ip/iplink_vxlan.c > +++ b/ip/iplink_vxlan.c > @@ -276,6 +276,12 @@ static int vxlan_parse_opt(struct link_util *lu, int argc, char **argv, > } else if (!matches(*argv, "noudpcsum")) { > check_duparg(&attrs, IFLA_VXLAN_UDP_CSUM, *argv, *argv); > addattr8(n, 1024, IFLA_VXLAN_UDP_CSUM, 0); > + } else if (!matches(*argv, "localbypass")) { > + check_duparg(&attrs, IFLA_VXLAN_LOCALBYPASS, *argv, *argv); > + addattr8(n, 1024, IFLA_VXLAN_LOCALBYPASS, 1); > + } else if (!matches(*argv, "nolocalbypass")) { > + check_duparg(&attrs, IFLA_VXLAN_LOCALBYPASS, *argv, *argv); > + addattr8(n, 1024, IFLA_VXLAN_LOCALBYPASS, 0); matches is deparated, please use strcmp instead. Thanks Hangbin
Hangbin Liu <liuhangbin@gmail.com> writes: > Hi Vladimir, > > For the subject prefix, [PATCH iproute2-next] is enough for the v1 patch. > > On Thu, Mar 23, 2023 at 02:04:51PM +0800, Vladimir Nikishkin wrote: >> Add userspace support for the nolocalbypass vxlan netlink >> attribute. With nolocalbypass, if an entry is pointing to the >> local machine, but the system driver is not listening on this >> port, the driver will not drop packets, but will forward them >> to the userspace network stack instead. >> >> This commit has a corresponding patch in the net-next list. >> >> Signed-off-by: Vladimir Nikishkin <vladimir@nikishkin.pw> >> --- >> include/uapi/linux/if_link.h | 1 + >> ip/iplink_vxlan.c | 18 ++++++++++++++++++ >> man/man8/ip-link.8.in | 8 ++++++++ >> 3 files changed, 27 insertions(+) >> >> diff --git a/include/uapi/linux/if_link.h b/include/uapi/linux/if_link.h >> index d61bd32d..fd390b40 100644 >> --- a/include/uapi/linux/if_link.h >> +++ b/include/uapi/linux/if_link.h >> @@ -824,6 +824,7 @@ enum { >> IFLA_VXLAN_TTL_INHERIT, >> IFLA_VXLAN_DF, >> IFLA_VXLAN_VNIFILTER, /* only applicable with COLLECT_METADATA mode */ >> + IFLA_VXLAN_LOCALBYPASS, >> __IFLA_VXLAN_MAX >> }; >> #define IFLA_VXLAN_MAX (__IFLA_VXLAN_MAX - 1) > > There is no need to include the uapi header. Stephen will sync it with upstream. > > Hi Stephen, should we add this note to the README.devel? > Without this change, my code does not compile. I ended up modifying the header, but not adding it to git. Is this the correct way of doing it? >> diff --git a/ip/iplink_vxlan.c b/ip/iplink_vxlan.c >> index c7e0e1c4..17fa5cf7 100644 >> --- a/ip/iplink_vxlan.c >> +++ b/ip/iplink_vxlan.c >> @@ -276,6 +276,12 @@ static int vxlan_parse_opt(struct link_util *lu, int argc, char **argv, >> } else if (!matches(*argv, "noudpcsum")) { >> check_duparg(&attrs, IFLA_VXLAN_UDP_CSUM, *argv, *argv); >> addattr8(n, 1024, IFLA_VXLAN_UDP_CSUM, 0); >> + } else if (!matches(*argv, "localbypass")) { >> + check_duparg(&attrs, IFLA_VXLAN_LOCALBYPASS, *argv, *argv); >> + addattr8(n, 1024, IFLA_VXLAN_LOCALBYPASS, 1); >> + } else if (!matches(*argv, "nolocalbypass")) { >> + check_duparg(&attrs, IFLA_VXLAN_LOCALBYPASS, *argv, *argv); >> + addattr8(n, 1024, IFLA_VXLAN_LOCALBYPASS, 0); > > matches is deparated, please use strcmp instead. Why is strcmp recommended, not strncmp? I remember strcmp being frowned upon for some potential memory bounds violations. > > Thanks > Hangbin
On Wed, Apr 05, 2023 at 01:21:17PM +0800, Vladimir Nikishkin wrote: > > There is no need to include the uapi header. Stephen will sync it with upstream. > > > > Hi Stephen, should we add this note to the README.devel? > > > > Without this change, my code does not compile. I ended up modifying the > header, but not adding it to git. Is this the correct way of doing it? Yes, that's what I did. > > >> diff --git a/ip/iplink_vxlan.c b/ip/iplink_vxlan.c > >> index c7e0e1c4..17fa5cf7 100644 > >> --- a/ip/iplink_vxlan.c > >> +++ b/ip/iplink_vxlan.c > >> @@ -276,6 +276,12 @@ static int vxlan_parse_opt(struct link_util *lu, int argc, char **argv, > >> } else if (!matches(*argv, "noudpcsum")) { > >> check_duparg(&attrs, IFLA_VXLAN_UDP_CSUM, *argv, *argv); > >> addattr8(n, 1024, IFLA_VXLAN_UDP_CSUM, 0); > >> + } else if (!matches(*argv, "localbypass")) { > >> + check_duparg(&attrs, IFLA_VXLAN_LOCALBYPASS, *argv, *argv); > >> + addattr8(n, 1024, IFLA_VXLAN_LOCALBYPASS, 1); > >> + } else if (!matches(*argv, "nolocalbypass")) { > >> + check_duparg(&attrs, IFLA_VXLAN_LOCALBYPASS, *argv, *argv); > >> + addattr8(n, 1024, IFLA_VXLAN_LOCALBYPASS, 0); > > > > matches is deparated, please use strcmp instead. > > Why is strcmp recommended, not strncmp? I remember strcmp being frowned > upon for some potential memory bounds violations. We can't limit the string length as the parameter may have same prefix. e.g. when you have 2 parameters "beef" and "beefsalad". Hangbin
diff --git a/include/uapi/linux/if_link.h b/include/uapi/linux/if_link.h index d61bd32d..fd390b40 100644 --- a/include/uapi/linux/if_link.h +++ b/include/uapi/linux/if_link.h @@ -824,6 +824,7 @@ enum { IFLA_VXLAN_TTL_INHERIT, IFLA_VXLAN_DF, IFLA_VXLAN_VNIFILTER, /* only applicable with COLLECT_METADATA mode */ + IFLA_VXLAN_LOCALBYPASS, __IFLA_VXLAN_MAX }; #define IFLA_VXLAN_MAX (__IFLA_VXLAN_MAX - 1) diff --git a/ip/iplink_vxlan.c b/ip/iplink_vxlan.c index c7e0e1c4..17fa5cf7 100644 --- a/ip/iplink_vxlan.c +++ b/ip/iplink_vxlan.c @@ -276,6 +276,12 @@ static int vxlan_parse_opt(struct link_util *lu, int argc, char **argv, } else if (!matches(*argv, "noudpcsum")) { check_duparg(&attrs, IFLA_VXLAN_UDP_CSUM, *argv, *argv); addattr8(n, 1024, IFLA_VXLAN_UDP_CSUM, 0); + } else if (!matches(*argv, "localbypass")) { + check_duparg(&attrs, IFLA_VXLAN_LOCALBYPASS, *argv, *argv); + addattr8(n, 1024, IFLA_VXLAN_LOCALBYPASS, 1); + } else if (!matches(*argv, "nolocalbypass")) { + check_duparg(&attrs, IFLA_VXLAN_LOCALBYPASS, *argv, *argv); + addattr8(n, 1024, IFLA_VXLAN_LOCALBYPASS, 0); } else if (!matches(*argv, "udp6zerocsumtx")) { check_duparg(&attrs, IFLA_VXLAN_UDP_ZERO_CSUM6_TX, *argv, *argv); @@ -613,6 +619,18 @@ static void vxlan_print_opt(struct link_util *lu, FILE *f, struct rtattr *tb[]) } } + if (tb[IFLA_VXLAN_LOCALBYPASS]) { + __u8 localbypass = rta_getattr_u8(tb[IFLA_VXLAN_LOCALBYPASS]); + + if (is_json_context()) { + print_bool(PRINT_ANY, "localbypass", NULL, localbypass); + } else { + if (!localbypass) + fputs("no", f); + fputs("localbypass ", f); + } + } + if (tb[IFLA_VXLAN_UDP_ZERO_CSUM6_TX]) { __u8 csum6 = rta_getattr_u8(tb[IFLA_VXLAN_UDP_ZERO_CSUM6_TX]); diff --git a/man/man8/ip-link.8.in b/man/man8/ip-link.8.in index c8c65657..c78dc9dd 100644 --- a/man/man8/ip-link.8.in +++ b/man/man8/ip-link.8.in @@ -623,6 +623,8 @@ the following additional arguments are supported: ] [ .RB [ no ] udpcsum ] [ +.RB [ no ] localbypass +] [ .RB [ no ] udp6zerocsumtx ] [ .RB [ no ] udp6zerocsumrx @@ -727,6 +729,12 @@ are entered into the VXLAN device forwarding database. .RB [ no ] udpcsum - specifies if UDP checksum is calculated for transmitted packets over IPv4. +.sp +.RB [ no ] localbypass +- if fdb destination is local, but there is no corresponding vni, forward packets +to the userspace network stack. Supposedly, there may be a userspace process +listening for these packets. + .sp .RB [ no ] udp6zerocsumtx - skip UDP checksum calculation for transmitted packets over IPv6.
Add userspace support for the nolocalbypass vxlan netlink attribute. With nolocalbypass, if an entry is pointing to the local machine, but the system driver is not listening on this port, the driver will not drop packets, but will forward them to the userspace network stack instead. This commit has a corresponding patch in the net-next list. Signed-off-by: Vladimir Nikishkin <vladimir@nikishkin.pw> --- include/uapi/linux/if_link.h | 1 + ip/iplink_vxlan.c | 18 ++++++++++++++++++ man/man8/ip-link.8.in | 8 ++++++++ 3 files changed, 27 insertions(+)