ipv6: report why a route was deleted in RTM_DELROUTE
Subsystems (5)
v5 cover ipv6: report why a route was deleted in RTM_DELROUTE
When the kernel deletes an IPv6 route on its own, the RTM_DELROUTE notification does not say why. User space cannot tell a route that expired from one the router explicitly withdrew, yet the two call for different reactions: an expired RA route means the router failed to refresh it in time, which points at a misconfigured or unreliable router and may warrant action such as disabling IPv6 on that network, while a zero-lifetime withdrawal is normal, RFC-compliant operation.
This is a general problem for any consumer device running Linux, especially on Wi-Fi networks, where multicast delivery is not guaranteed (e.g. frames can be lost around DTIM for clients in power save mode). The motivating case is Android: the userspace NetworkStack process listens on RTMGRP_IPV6_ROUTE and today treats any loss of the IPv6 default route as "router lost". To avoid the device repeatedly gaining and losing IPv6 connectivity on a badly configured network, when it detects the device is on a dual-stack network with working IPv4 connectivity, it defensively clears accept_ra_defrtr and restarts IPv6, so user space apps stop using broken global IPv6 connectivity while link-local IPv6 keeps working. That reaction is wrong if the route was withdrawn by a zero-lifetime RA (some ISPs do this intentionally for reconfiguration) - with accept_ra_defrtr off, IPv6 never recovers once the router advertises again. It is the right reaction if the route genuinely expired, since the router failed to refresh it in time.
Fixing this in user space is not practical: RTM_NEWROUTE carries the initial route lifetime (in rta_cacheinfo), but the kernel does not resend it when a later RA refreshes the lifetime. So distinguishing the cause of an RTM_DELROUTE from user space would mean opening a raw socket, listening to RAs, and tracking lifetimes independently, duplicating logic the kernel already has. Sending RTM_NEWROUTE on every RA lifetime refresh was also considered, but that would be spammy and is technically wrong, since a lifetime update does not add a new route.
This series proposes RTA_DEL_REASON instead: it tells user space why the route was deleted so it can react accordingly. In the Android case, NetworkStack would defensively disable global IPv6 only on RT_DEL_REASON_EXPIRED, and take no action on RT_DEL_REASON_RA_WITHDRAWN, since that is RFC-compliant behavior.
Patches 1 to 6 add RTA_DEL_REASON and enum rt_del_reason to the rtnetlink uAPI, thread the reason from the kernel-initiated IPv6 deletion paths down to the RTM_DELROUTE notification, and record the
cause: RT_DEL_REASON_EXPIRED for routes garbage collected after their RTF_EXPIRES lifetime ran out, and RT_DEL_REASON_RA_WITHDRAWN for default routes, prefix routes and RFC 4191 route information routes withdrawn by Router Advertisements. Patches 1 to 5 are no-ops on the wire; the attribute first appears in patch 6. The route addition path is not touched.
Patches 7 to 9 extend the rt-route Netlink spec with the route notifications and their multicast groups, split the newroute and delroute request attribute lists out of the shared getroute reply list, and add the new attribute and its enum.
Only kernel-initiated deletions that user space cannot otherwise explain are attributed. User-requested deletions are self-explanatory to the requester, so they carry no reason; the UAPI documents that absence and RT_DEL_REASON_UNSPEC must be treated identically, which keeps the door open for attributing more paths (nexthop removal cascades, device removal) later.
Patch 10 adds selftests covering all three producer paths: a GC-expired route, and a default route + PIO prefix route + RIO route advertised and then withdrawn by hand-crafted RAs over a raw ICMPv6 socket (no external RA tool needed), plus a check that user-requested deletions carry no attribute. The notifications are decoded with YNL, which also exercises the rt-route spec additions.
Changes since v4:
- Rename the payload values to RT_DEL_REASON_*, so they do not read as attribute ids. The attribute id is still RTA_DEL_REASON.
- Document that the reason value space is family-agnostic.
- Drop the skip_notify argument from ip6_del_rt_reason(); all callers passed false, and skipping the notification discards the reason.
- Drop the unused !CONFIG_IPV6 stub for ip6_del_rt_reason().
Changes since v3:
- Split the single kernel patch into six, one logical step each, per review.
- Use the enum throughout instead of a plain u32.
- Add inet6_rt_del_notify() and call it from fib6_del_route(), so the route addition path is unchanged.
- Move the route notifications and their multicast groups to their own spec patch.
- Split the newroute and delroute request attribute lists out of the getroute reply list in a separate spec patch, so the attribute addition is a one-line diff.
Changes since v2:
- Fix the patch 1 commit message, which still said u8.
- Keep del-reason out of the newroute and delroute request attribute lists in the rt-route spec, since the kernel rejects it in requests.
Changes since v1:
- Expand the motivation with the Android use case, per review request.
- Widen RTA_DEL_REASON from u8 to u32, per Netlink uAPI convention.
- Convert dict_keys to a set before the set difference in the selftest, for clarity.
Yuyang Huang (10):
ipv6: add ip6_del_rt_reason()
ipv6: propagate the route deletion reason to fib6_del_route()
ipv6: record the reason for kernel-initiated route deletions
ipv6: add a deletion reason argument to rt6_fill_node()
ipv6: expose the route deletion reason in RTM_DELROUTE
ipv6: add inet6_rt_del_notify()
netlink: specs: rt-route: add route notifications
netlink: specs: rt-route: split out the request attribute list
netlink: specs: rt-route: add the route deletion reason
selftests: net: verify RTA_DEL_REASON on route deletion
Documentation/netlink/specs/rt-route.yaml | 68 ++++++- include/net/ip6_fib.h | 5 +- include/net/ip6_route.h | 2 + include/uapi/linux/rtnetlink.h | 22 +++ net/ipv6/addrconf.c | 3 +- net/ipv6/ip6_fib.c | 14 +- net/ipv6/ndisc.c | 4 +- net/ipv6/route.c | 75 ++++++-- .../testing/selftests/net/lib/py/__init__.py | 4 +- tools/testing/selftests/net/lib/py/ynl.py | 7 +- tools/testing/selftests/net/rtnetlink.py | 181 +++++++++++++++++- 11 files changed, 349 insertions(+), 36 deletions(-)
--
2.43.0
PATCH v5 1/10 ipv6: add ip6_del_rt_reason()
Add RTA_DEL_REASON and enum rt_del_reason to the rtnetlink uAPI, and add ip6_del_rt_reason(), which takes the reason a route is being deleted. It has no skip_notify argument: a caller that records a deletion reason wants the notification that carries it.
The reason is unused for now. Subsequent patches propagate it to the deletion path and report it on RTM_DELROUTE.
--- include/net/ip6_route.h | 2 ++ include/uapi/linux/rtnetlink.h | 22 ++++++++++++++++++++++ net/ipv6/route.c | 8 ++++++++ 3 files changed, 32 insertions(+)
include/net/ip6_route.h +2 -0
@@ -128,6 +128,8 @@ int ip6_route_add(struct fib6_config *cfg, gfp_t gfp_flags,int ip6_ins_rt (struct net *net,struct fib6_info *f6i);#if IS_ENABLED (CONFIG_IPV6 )int ip6_del_rt (struct net *net,struct fib6_info *f6i,bool skip_notify); +int ip6_del_rt_reason (struct net *net,struct fib6_info *f6i, +enum rt_del_reason del_reason);#else static inline int ip6_del_rt (struct net *net,struct fib6_info *f6i,bool skip_notify)
include/uapi/linux/rtnetlink.h +22 -0
@@ -399,6 +399,7 @@ enum rtattr_type_t {RTA_DPORT ,RTA_NH_ID ,RTA_FLOWLABEL , +RTA_DEL_REASON , __RTA_MAX };
@@ -407,6 +408,27 @@ enum rtattr_type_t {#define RTM_RTA (r) ((struct rtattr*)(((char*)(r)) + NLMSG_ALIGN(sizeof(struct rtmsg))))#define RTM_PAYLOAD (n) NLMSG_PAYLOAD(n,sizeof(struct rtmsg)) +/* RTA_DEL_REASON: why the kernel deleted the route. u32. +* Emitted only on RTM_DELROUTE notifications, and only when the deletion +* path records a cause. Absence means either an older kernel or a +* deletion path that does not (yet) record its cause - consumers must +* treat "absent" and "unspec" identically. New causes may be appended. +* Currently only IPv6 deletion paths record a cause. +* +* The value space is family-agnostic: a value must never be +* reinterpreted per address family. A cause that only one family can +* produce still gets its own value rather than reusing another +* family's. +*/ +enum rt_del_reason { +RT_DEL_REASON_UNSPEC ,/* cause not recorded */ +RT_DEL_REASON_EXPIRED ,/* RTF_EXPIRES lifetime ran out (GC) */ +RT_DEL_REASON_RA_WITHDRAWN ,/* zero-lifetime RA / PIO / RIO */ + __RT_DEL_REASON_MAX +}; + +#define RT_DEL_REASON_MAX (__RT_DEL_REASON_MAX - 1) +/* RTM_MULTIPATH --- array of struct rtnexthop. * * "struct rtnexthop" describes all necessary nexthop information,
net/ipv6/route.c +8 -0
@@ -4004,6 +4004,14 @@ int ip6_del_rt(struct net *net, struct fib6_info *rt, bool skip_notify)return __ip6_del_rt (rt, &info); } +int ip6_del_rt_reason (struct net *net,struct fib6_info *rt, +enum rt_del_reason del_reason) +{ +struct nl_info info = { .nl_net = net }; + +return __ip6_del_rt (rt, &info); +} +static int __ip6_del_rt_siblings (struct fib6_info *rt,struct fib6_config *cfg) {struct nl_info *info = &cfg->fc_nlinfo;
Signature · 1 line
2.43.0
Ido Schimmel
maintainerRe: [PATCH net-next v5 01/10] ipv6: add ip6_del_rt_reason()
Quoted 20 lines · Yuyang Huang
Add
RTA_DEL_REASONand enumrt_del_reasonto the rtnetlink uAPI, and addip6_del_rt_reason(), which takes the reason a route is being deleted. It has noskip_notifyargument: a caller that records a deletion reason wants the notification that carries it.The reason is unused for now. Subsequent patches propagate it to the deletion path and report it on
RTM_DELROUTE.Signed-off-by Yuyang Huang <[email protected]>--- include/net/ip6_route.h | 2 ++ include/uapi/linux/rtnetlink.h | 22 ++++++++++++++++++++++ net/ipv6/route.c | 8 ++++++++ 3 files changed, 32 insertions(+)include/net/ip6_route.h +0
-0
include/net/ip6_route.h int ip6_route_add(struct fib6_config *cfg, gfp_t gfp_flags, · line 132
Quoted 5 lines
include/net/ip6_route.h +2
-0@@ -128,6 +128,8 @@ int ip6_route_add(struct fib6_config *cfg, gfp_t gfp_flags,int ip6_ins_rt (struct net *net,struct fib6_info *f6i);#if IS_ENABLED (CONFIG_IPV6 )int ip6_del_rt (struct net *net,struct fib6_info *f6i,bool skip_notify); +int ip6_del_rt_reason (struct net *net,struct fib6_info *f6i, +enum rt_del_reason del_reason);
Since the function is only invoked by IPv6 code there is no reason to put it inside the guard and it would also suppress the comment from Sashiko. You can place it near ip6_ins_rt().
Quoted 16 lines
include/net/ip6_route.h +0
-0#else static inline int ip6_del_rt (struct net *net,struct fib6_info *f6i,bool skip_notify)include/uapi/linux/rtnetlink.h +1
-0@@ -399,6 +399,7 @@ enum rtattr_type_t {RTA_DPORT ,RTA_NH_ID ,RTA_FLOWLABEL , +RTA_DEL_REASON , __RTA_MAX };
include/uapi/linux/rtnetlink.h enum rtattr_type_t { · line 416
Quoted 9 lines
include/uapi/linux/rtnetlink.h +6
-0@@ -407,6 +408,27 @@ enum rtattr_type_t {#define RTM_RTA (r) ((struct rtattr*)(((char*)(r)) + NLMSG_ALIGN(sizeof(struct rtmsg))))#define RTM_PAYLOAD (n) NLMSG_PAYLOAD(n,sizeof(struct rtmsg)) +/* RTA_DEL_REASON: why the kernel deleted the route. u32. +* Emitted only on RTM_DELROUTE notifications, and only when the deletion +* path records a cause. Absence means either an older kernel or a +* deletion path that does not (yet) record its cause - consumers must +* treat "absent" and "unspec" identically. New causes may be appended. +* Currently only IPv6 deletion paths record a cause.
Please add a note that this attribute is rejected by the kernel in requests, like you did in patch #9.
Quoted 40 lines
include/uapi/linux/rtnetlink.h +15
-0+* +* The value space is family-agnostic: a value must never be +* reinterpreted per address family. A cause that only one family can +* produce still gets its own value rather than reusing another +* family's. +*/ +enum rt_del_reason { +RT_DEL_REASON_UNSPEC ,/* cause not recorded */ +RT_DEL_REASON_EXPIRED ,/* RTF_EXPIRES lifetime ran out (GC) */ +RT_DEL_REASON_RA_WITHDRAWN ,/* zero-lifetime RA / PIO / RIO */ + __RT_DEL_REASON_MAX +}; + +#define RT_DEL_REASON_MAX (__RT_DEL_REASON_MAX - 1) +/* RTM_MULTIPATH --- array of struct rtnexthop. * * "struct rtnexthop" describes all necessary nexthop information, net/ipv6/route.c +8
-0@@ -4004,6 +4004,14 @@ int ip6_del_rt(struct net *net, struct fib6_info *rt, bool skip_notify)return __ip6_del_rt (rt, &info); } +int ip6_del_rt_reason (struct net *net,struct fib6_info *rt, +enum rt_del_reason del_reason) +{ +struct nl_info info = { .nl_net = net }; + +return __ip6_del_rt (rt, &info); +} +static int __ip6_del_rt_siblings (struct fib6_info *rt,struct fib6_config *cfg) {struct nl_info *info = &cfg->fc_nlinfo;Signature · 1 line
2.43.0
Re: [PATCH net-next v5 01/10] ipv6: add ip6_del_rt_reason()
Quoted 7 lines · Ido Schimmel
Since the function is only invoked by IPv6 code there is no reason to put it inside the guard and it would also suppress the comment from Sashiko. You can place it near
ip6_ins_rt().Please add a note that this attribute is rejected by the kernel in requests, like you did in patch #9.
Will fix them in patchset v6.
PATCH v5 2/10 ipv6: propagate the route deletion reason to fib6_del_route()
In reply to Yuyang HuangPass the deletion reason from ip6_del_rt_reason() down through __ip6_del_rt(), fib6_del() and into fib6_del_route(). All existing callers pass RT_DEL_REASON_UNSPEC.
fib6_del_route() ignores the reason until the notification path learns to report it.
--- include/net/ip6_fib.h | 3 ++- net/ipv6/ip6_fib.c | 12 +++++++----- net/ipv6/route.c | 19 +++++++++++-------- 3 files changed, 20 insertions(+), 14 deletions(-)
include/net/ip6_fib.h +2 -1
@@ -468,7 +468,8 @@ void fib6_clean_all_skip_notify(struct net *net,int fib6_add (struct fib6_node *root,struct fib6_info *rt,struct nl_info *info,struct netlink_ext_ack *extack); -int fib6_del (struct fib6_info *rt,struct nl_info *info); +int fib6_del (struct fib6_info *rt,struct nl_info *info, +enum rt_del_reason del_reason);static inline void rt6_get_prefsrc (const struct rt6_info *rt,struct in6_addr *addr)
net/ipv6/ip6_fib.c +7 -5
@@ -1967,7 +1967,8 @@ static struct fib6_node *fib6_repair_tree(struct net *net, }static void fib6_del_route (struct fib6_table *table,struct fib6_node *fn, -struct fib6_info __rcu **rtp,struct nl_info *info) +struct fib6_info __rcu **rtp,struct nl_info *info, +enum rt_del_reason del_reason) {struct fib6_info *leaf, *replace_rt =NULL ;struct fib6_walker *w;
@@ -2062,7 +2063,8 @@ static void fib6_del_route(struct fib6_table *table, struct fib6_node *fn, }/* Need to own table->tb6_lock */ -int fib6_del (struct fib6_info *rt,struct nl_info *info) +int fib6_del (struct fib6_info *rt,struct nl_info *info, +enum rt_del_reason del_reason) {struct net *net = info->nl_net;struct fib6_info __rcu **rtp;
@@ -2091,7 +2093,7 @@ int fib6_del(struct fib6_info *rt, struct nl_info *info)if (rt == cur) {if (fib6_requires_src (cur))fib6_routes_require_src_dec (info->nl_net); -fib6_del_route (table, fn, rtp, info); +fib6_del_route (table, fn, rtp, info, del_reason);return 0 ; } rtp_next = &cur->fib6_next;
@@ -2253,7 +2255,7 @@ static int fib6_clean_node(struct fib6_walker *w) res = c->func (rt, c->arg);if (res ==-1 ) { w->leaf = rt; - res =fib6_del (rt, &info); + res =fib6_del (rt, &info,RT_DEL_REASON_UNSPEC );if (res) {#if RT6_DEBUG >=2 pr_debug("%s: del failed: rt=%p@%p err=%d\n",
@@ -2401,7 +2403,7 @@ static void fib6_gc_table(struct net *net,hlist_for_each_entry_safe (rt, n, &tb6->tb6_gc_hlist, gc_link)if (fib6_age (rt, gc_args) ==-1 ) -fib6_del (rt, &info); +fib6_del (rt, &info,RT_DEL_REASON_UNSPEC ); }static void fib6_gc_all (struct net *net,struct fib6_gc_args *gc_args)
net/ipv6/route.c +11 -8
@@ -3973,7 +3973,8 @@ int ip6_route_add(struct fib6_config *cfg, gfp_t gfp_flags,return err; } -static int __ip6_del_rt (struct fib6_info *rt,struct nl_info *info) +static int __ip6_del_rt (struct fib6_info *rt,struct nl_info *info, +enum rt_del_reason del_reason) {struct net *net = info->nl_net;struct fib6_table *table;
@@ -3986,7 +3987,7 @@ static int __ip6_del_rt(struct fib6_info *rt, struct nl_info *info) table = rt->fib6_table;spin_lock_bh (&table->tb6_lock); - err =fib6_del (rt, info); + err =fib6_del (rt, info, del_reason);spin_unlock_bh (&table->tb6_lock); out:
@@ -4001,7 +4002,7 @@ int ip6_del_rt(struct net *net, struct fib6_info *rt, bool skip_notify) .skip_notify = skip_notify }; -return __ip6_del_rt (rt, &info); +return __ip6_del_rt (rt, &info,RT_DEL_REASON_UNSPEC ); } int ip6_del_rt_reason(struct net *net, struct fib6_info *rt,
@@ -4009,7 +4010,7 @@ int ip6_del_rt_reason(struct net *net, struct fib6_info *rt, {struct nl_info info = { .nl_net = net }; -return __ip6_del_rt (rt, &info); +return __ip6_del_rt (rt, &info, del_reason); }static int __ip6_del_rt_siblings (struct fib6_info *rt,struct fib6_config *cfg)
@@ -4072,13 +4073,13 @@ static int __ip6_del_rt_siblings(struct fib6_info *rt, struct fib6_config *cfg)list_for_each_entry_safe (sibling, next_sibling, &rt->fib6_siblings, fib6_siblings) { - err =fib6_del (sibling, info); + err =fib6_del (sibling, info,RT_DEL_REASON_UNSPEC );if (err) goto out_unlock; } } - err =fib6_del (rt, info); + err =fib6_del (rt, info,RT_DEL_REASON_UNSPEC ); out_unlock:spin_unlock_bh (&table->tb6_lock); out_put:
@@ -4204,7 +4205,8 @@ static int ip6_route_del(struct fib6_config *cfg,if (!fib6_info_hold_safe (rt))continue ; - err =__ip6_del_rt (rt, &cfg->fc_nlinfo); + err =__ip6_del_rt (rt, &cfg->fc_nlinfo, +RT_DEL_REASON_UNSPEC );break ; }if (cfg->fc_nh_id)
@@ -4223,7 +4225,8 @@ static int ip6_route_del(struct fib6_config *cfg,/* if gateway was specified only delete the one hop */ if (cfg->fc_flags &RTF_GATEWAY ) - err =__ip6_del_rt (rt, &cfg->fc_nlinfo); + err =__ip6_del_rt (rt, &cfg->fc_nlinfo, +RT_DEL_REASON_UNSPEC );else err =__ip6_del_rt_siblings (rt, cfg);break ;
Signature · 1 line
2.43.0
Re: [PATCH net-next v5 02/10] ipv6: propagate the route deletion reason to fib6_del_route()
Quoted 8 lines · Yuyang Huang
Pass the deletion reason from
ip6_del_rt_reason()down through__ip6_del_rt(),fib6_del()and intofib6_del_route(). All existing callers passRT_DEL_REASON_UNSPEC.
fib6_del_route()ignores the reason until the notification path learns to report it.Signed-off-by Yuyang Huang <[email protected]>
PATCH v5 3/10 ipv6: record the reason for kernel-initiated route deletions
In reply to Yuyang HuangRecord why the kernel deletes an IPv6 route on its own:
- RT_DEL_REASON_EXPIRED for routes reaped by the FIB6 garbage collector after their RTF_EXPIRES lifetime ran out.
- RT_DEL_REASON_RA_WITHDRAWN for default routes, prefix routes and RFC 4191 route information routes withdrawn by a zero-lifetime Router Advertisement.
Deleting a default route because its metric changed is not a withdrawal, so it keeps RT_DEL_REASON_UNSPEC.
--- net/ipv6/addrconf.c | 3 ++- net/ipv6/ip6_fib.c | 2 +- net/ipv6/ndisc.c | 4 +++- net/ipv6/route.c | 2 +- 4 files changed, 7 insertions(+), 4 deletions(-)
net/ipv6/addrconf.c +2 -1
@@ -2874,7 +2874,8 @@ void addrconf_prefix_rcv(struct net_device *dev, u8 *opt, int len, bool sllao)if (rt) {/* Autoconf prefix route */ if (valid_lft ==0 ) { -ip6_del_rt (net, rt, false); +ip6_del_rt_reason (net, rt, +RT_DEL_REASON_RA_WITHDRAWN ); rt =NULL ; }else { table = rt->fib6_table;
net/ipv6/ip6_fib.c +1 -1
@@ -2403,7 +2403,7 @@ static void fib6_gc_table(struct net *net,hlist_for_each_entry_safe (rt, n, &tb6->tb6_gc_hlist, gc_link)if (fib6_age (rt, gc_args) ==-1 ) -fib6_del (rt, &info,RT_DEL_REASON_UNSPEC ); +fib6_del (rt, &info,RT_DEL_REASON_EXPIRED ); }static void fib6_gc_all (struct net *net,struct fib6_gc_args *gc_args)
net/ipv6/ndisc.c +3 -1
@@ -1361,7 +1361,9 @@ static enum skb_drop_reason ndisc_router_discovery(struct sk_buff *skb) defrtr_usr_metric = in6_dev->cnf.ra_defrtr_metric;/* delete the route if lifetime is 0 or if metric needs change */ if (rt && (lifetime ==0 || rt->fib6_metric != defrtr_usr_metric)) { -ip6_del_rt (net, rt, false); +ip6_del_rt_reason (net, rt, + lifetime ==0 ?RT_DEL_REASON_RA_WITHDRAWN : +RT_DEL_REASON_UNSPEC ); rt =NULL ; }
net/ipv6/route.c +1 -1
@@ -1020,7 +1020,7 @@ int rt6_route_rcv(struct net_device *dev, u8 *opt, int len, gwaddr, dev);if (rt && !lifetime) { -ip6_del_rt (net, rt, false); +ip6_del_rt_reason (net, rt,RT_DEL_REASON_RA_WITHDRAWN ); rt =NULL ; }
Signature · 1 line
2.43.0
Re: [PATCH net-next v5 03/10] ipv6: record the reason for kernel-initiated route deletions
Quoted 8 lines · Yuyang Huang
Record why the kernel deletes an IPv6 route on its own:
-
RT_DEL_REASON_EXPIREDfor routes reaped by the FIB6 garbage collector after theirRTF_EXPIRESlifetime ran out.
-RT_DEL_REASON_RA_WITHDRAWNfor default routes, prefix routes and RFC 4191 route information routes withdrawn by a zero-lifetime Router Advertisement.
Quoted 4 lines
Deleting a default route because its metric changed is not a withdrawal, so it keeps
RT_DEL_REASON_UNSPEC.Signed-off-by Yuyang Huang <[email protected]>
PATCH v5 4/10 ipv6: add a deletion reason argument to rt6_fill_node()
In reply to Yuyang HuangAdd the deletion reason to rt6_fill_node() so that it can report it to user space. All callers pass RT_DEL_REASON_UNSPEC for now.
--- net/ipv6/route.c | 27 +++++++++++++++++---------- 1 file changed, 17 insertions(+), 10 deletions(-)
net/ipv6/route.c +17 -10
@@ -111,7 +111,7 @@ static int rt6_fill_node(struct net *net, struct sk_buff *skb, struct fib6_info *rt, struct dst_entry *dst, struct in6_addr *dest, struct in6_addr *src, int iif, int type, u32 portid, u32 seq, - unsigned int flags); + unsigned int flags, enum rt_del_reason del_reason);static struct rt6_info *rt6_find_cached_rt (const struct fib6_result *res,const struct in6_addr *daddr,const struct in6_addr *saddr);
@@ -4037,7 +4037,8 @@ static int __ip6_del_rt_siblings(struct fib6_info *rt, struct fib6_config *cfg)if (rt6_fill_node (net, skb, rt,NULL ,NULL ,NULL ,0 ,RTM_DELROUTE , - info->portid, seq,0 ) <0 ) { + info->portid, seq,0 , +RT_DEL_REASON_UNSPEC ) <0 ) {kfree_skb (skb); skb =NULL ; }else
@@ -5786,7 +5787,7 @@ static int rt6_fill_node(struct net *net, struct sk_buff *skb, struct fib6_info *rt, struct dst_entry *dst, struct in6_addr *dest, struct in6_addr *src, int iif, int type, u32 portid, u32 seq, - unsigned int flags) + unsigned int flags, enum rt_del_reason del_reason) { struct rt6_info *rt6 = dst_rt6_info(dst);struct rt6key *rt6_dst, *rt6_src;
@@ -6066,7 +6067,8 @@ static int rt6_nh_dump_exceptions(struct fib6_nh *nh, void *arg) &rt6_ex->rt6i->dst,NULL ,NULL ,0 ,RTM_NEWROUTE ,NETLINK_CB (dump->cb->skb).portid, - dump->cb->nlh->nlmsg_seq, w->flags); + dump->cb->nlh->nlmsg_seq, w->flags, + RT_DEL_REASON_UNSPEC);if (err)return err;
@@ -6114,7 +6116,8 @@ int rt6_dump_route(struct fib6_info *rt, void *p_arg, unsigned int skip)if (rt6_fill_node (net, arg->skb, rt,NULL ,NULL ,NULL ,0 ,RTM_NEWROUTE ,NETLINK_CB (arg->cb->skb).portid, - arg->cb->nlh->nlmsg_seq, flags)) { + arg->cb->nlh->nlmsg_seq, flags, +RT_DEL_REASON_UNSPEC )) {return 0 ; } count++;
@@ -6347,12 +6350,14 @@ static int inet6_rtm_getroute(struct sk_buff *in_skb, struct nlmsghdr *nlh, err =rt6_fill_node (net, skb, from,NULL ,NULL ,NULL , iif,RTM_NEWROUTE ,NETLINK_CB (in_skb).portid, - nlh->nlmsg_seq,0 ); + nlh->nlmsg_seq,0 , +RT_DEL_REASON_UNSPEC ); else err =rt6_fill_node (net, skb, from, dst, &fl6.daddr, &fl6.saddr, iif,RTM_NEWROUTE ,NETLINK_CB (in_skb).portid, - nlh->nlmsg_seq,0 ); + nlh->nlmsg_seq,0 , +RT_DEL_REASON_UNSPEC ); }else { err = -ENETUNREACH ; }
@@ -6388,7 +6393,8 @@ void inet6_rt_notify(int event, struct fib6_info *rt, struct nl_info *info, goto errout; err =rt6_fill_node (net, skb, rt,NULL ,NULL ,NULL ,0 , - event, info->portid, seq, nlm_flags); + event, info->portid, seq, nlm_flags, +RT_DEL_REASON_UNSPEC );if (err <0 ) {kfree_skb (skb);/* -EMSGSIZE implies needed space grew under us. */
@@ -6421,7 +6427,8 @@ void fib6_rt_update(struct net *net, struct fib6_info *rt, goto errout; err =rt6_fill_node (net, skb, rt,NULL ,NULL ,NULL ,0 , -RTM_NEWROUTE , info->portid, seq,NLM_F_REPLACE ); +RTM_NEWROUTE , info->portid, seq,NLM_F_REPLACE , +RT_DEL_REASON_UNSPEC );if (err <0 ) {/* -EMSGSIZE implies BUG in rt6_nlmsg_size() */ WARN_ON (err == -EMSGSIZE );
@@ -6474,7 +6481,7 @@ void fib6_info_hw_flags_set(struct net *net, struct fib6_info *f6i, } err =rt6_fill_node (net, skb, f6i,NULL ,NULL ,NULL ,0 ,RTM_NEWROUTE ,0 , -0 ,0 ); +0 ,0 ,RT_DEL_REASON_UNSPEC );if (err <0 ) {/* -EMSGSIZE implies BUG in rt6_nlmsg_size() */ WARN_ON (err == -EMSGSIZE );
Signature · 1 line
2.43.0
Re: [PATCH net-next v5 04/10] ipv6: add a deletion reason argument to rt6_fill_node()
Quoted 4 lines · Yuyang Huang
Add the deletion reason to
rt6_fill_node()so that it can report it to user space. All callers passRT_DEL_REASON_UNSPECfor now.Signed-off-by Yuyang Huang <[email protected]>
PATCH v5 5/10 ipv6: expose the route deletion reason in RTM_DELROUTE
In reply to Yuyang HuangEmit RTA_DEL_REASON from rt6_fill_node() when the deletion reason is not RT_DEL_REASON_UNSPEC, and reserve room for it in rt6_nlmsg_size().
Every caller still passes RT_DEL_REASON_UNSPEC.
--- net/ipv6/route.c | 5 +++++ 1 file changed, 5 insertions(+)
net/ipv6/route.c +5 -0
@@ -5751,6 +5751,7 @@ static size_t rt6_nlmsg_size(struct fib6_info *f6i) +nla_total_size (sizeof (struct rta_cacheinfo )) +nla_total_size (TCP_CA_NAME_MAX )/* RTAX_CC_ALGO */ +nla_total_size (1 )/* RTA_PREF */ + +nla_total_size (4 )/* RTA_DEL_REASON */ + nexthop_len; }
@@ -5966,6 +5967,10 @@ static int rt6_fill_node(struct net *net, struct sk_buff *skb,if (rtnl_put_cacheinfo (skb, dst,0 , expires, dst ? dst->error :0 ) <0 ) goto nla_put_failure; +if (del_reason !=RT_DEL_REASON_UNSPEC && +nla_put_u32 (skb,RTA_DEL_REASON , del_reason)) + goto nla_put_failure; +if (nla_put_u8 (skb,RTA_PREF ,IPV6_EXTRACT_PREF (rt6_flags))) goto nla_put_failure;
Signature · 1 line
2.43.0
Re: [PATCH net-next v5 05/10] ipv6: expose the route deletion reason in RTM_DELROUTE
Quoted 7 lines · Yuyang Huang
Emit
RTA_DEL_REASONfromrt6_fill_node()when the deletion reason is notRT_DEL_REASON_UNSPEC, and reserve room for it inrt6_nlmsg_size().Every caller still passes
RT_DEL_REASON_UNSPEC.Signed-off-by Yuyang Huang <[email protected]>
PATCH v5 6/10 ipv6: add inet6_rt_del_notify()
In reply to Yuyang HuangMove the body of inet6_rt_notify() to __inet6_rt_notify() and give it the deletion reason. inet6_rt_notify() keeps its prototype, so the route addition path does not change.
Add inet6_rt_del_notify() and call it from fib6_del_route(). RTA_DEL_REASON now reaches user space on RTM_DELROUTE for routes the kernel deleted on its own.
--- include/net/ip6_fib.h | 2 ++ net/ipv6/ip6_fib.c | 2 +- net/ipv6/route.c | 20 ++++++++++++++++---- 3 files changed, 19 insertions(+), 5 deletions(-)
include/net/ip6_fib.h +2 -0
@@ -533,6 +533,8 @@ static inline void fib6_rt_update(struct net *net, struct fib6_info *rt,#endif void inet6_rt_notify (int event,struct fib6_info *rt,struct nl_info *info,unsigned int flags); +void inet6_rt_del_notify (struct fib6_info *rt,struct nl_info *info, +enum rt_del_reason del_reason);void fib6_age_exceptions (struct fib6_info *rt,struct fib6_gc_args *gc_args,unsigned long now);
net/ipv6/ip6_fib.c +1 -1
@@ -2057,7 +2057,7 @@ static void fib6_del_route(struct fib6_table *table, struct fib6_node *fn,call_fib6_entry_notifiers_replace (net, replace_rt); }if (!info->skip_notify) -inet6_rt_notify (RTM_DELROUTE , rt, info,0 ); +inet6_rt_del_notify (rt, info, del_reason);fib6_info_release (rt); }
net/ipv6/route.c +16 -4
@@ -6378,8 +6378,9 @@ static int inet6_rtm_getroute(struct sk_buff *in_skb, struct nlmsghdr *nlh,return err; } -void inet6_rt_notify (int event,struct fib6_info *rt,struct nl_info *info, -unsigned int nlm_flags) +static void __inet6_rt_notify (int event,struct fib6_info *rt, +struct nl_info *info,unsigned int nlm_flags, +enum rt_del_reason del_reason) {struct net *net = info->nl_net;struct sk_buff *skb;
@@ -6398,8 +6399,7 @@ void inet6_rt_notify(int event, struct fib6_info *rt, struct nl_info *info, goto errout; err =rt6_fill_node (net, skb, rt,NULL ,NULL ,NULL ,0 , - event, info->portid, seq, nlm_flags, -RT_DEL_REASON_UNSPEC ); + event, info->portid, seq, nlm_flags, del_reason);if (err <0 ) {kfree_skb (skb);/* -EMSGSIZE implies needed space grew under us. */
@@ -6420,6 +6420,18 @@ void inet6_rt_notify(int event, struct fib6_info *rt, struct nl_info *info,rtnl_set_sk_err (net,RTNLGRP_IPV6_ROUTE , err); } +void inet6_rt_notify (int event,struct fib6_info *rt,struct nl_info *info, +unsigned int nlm_flags) +{ +__inet6_rt_notify (event, rt, info, nlm_flags,RT_DEL_REASON_UNSPEC ); +} + +void inet6_rt_del_notify (struct fib6_info *rt,struct nl_info *info, +enum rt_del_reason del_reason) +{ +__inet6_rt_notify (RTM_DELROUTE , rt, info,0 , del_reason); +} +void fib6_rt_update (struct net *net,struct fib6_info *rt,struct nl_info *info) {
Signature · 1 line
2.43.0
Re: [PATCH net-next v5 06/10] ipv6: add inet6_rt_del_notify()
Quoted 9 lines · Yuyang Huang
Move the body of
inet6_rt_notify()to__inet6_rt_notify()and give it the deletion reason.inet6_rt_notify()keeps its prototype, so the route addition path does not change.Add
inet6_rt_del_notify()and call it fromfib6_del_route().RTA_DEL_REASONnow reaches user space onRTM_DELROUTEfor routes the kernel deleted on its own.Signed-off-by Yuyang Huang <[email protected]>
PATCH v5 7/10 netlink: specs: rt-route: add route notifications
In reply to Yuyang HuangDeclare the RTM_NEWROUTE and RTM_DELROUTE notifications and the route multicast groups, so that generated clients can subscribe to route changes. Both notifications reuse the getroute reply attributes.
--- Documentation/netlink/specs/rt-route.yaml | 19 +++++++++++++++++++ 1 file changed, 19 insertions(+)
Documentation/netlink/specs/rt-route.yaml +19 -0
@@ -322,3 +322,22 @@ operations: request: value: 25 attributes: *all-route-attrs + - + name: newroute-ntf + doc: Notification about a created route. + value: 24 + notify: getroute + - + name: delroute-ntf + doc: Notification about a deleted route. + value: 25 + notify: getroute + +mcast-groups: + list: + - + name: rtnlgrp-ipv4-route + value: 7 + - + name: rtnlgrp-ipv6-route + value: 11
Signature · 1 line
2.43.0
PATCH v5 8/10 netlink: specs: rt-route: split out the request attribute list
In reply to Yuyang HuangThe newroute and delroute requests alias the same attribute list as the getroute reply, but requests and replies do not carry the same attributes. Give the requests their own list.
The two lists are identical today, so the generated code does not change.
--- Documentation/netlink/specs/rt-route.yaml | 32 +++++++++++++++++++++-- 1 file changed, 30 insertions(+), 2 deletions(-)
Documentation/netlink/specs/rt-route.yaml +30 -2
@@ -313,7 +313,35 @@ operations: do: request: value: 24 - attributes: *all-route-attrs + attributes: &route-req-attrs + - dst + - src + - iif + - oif + - gateway + - priority + - prefsrc + - metrics + - multipath + - flow + - cacheinfo + - table + - mark + - mfc-stats + - via + - newdst + - pref + - encap-type + - encap + - expires + - pad + - uid + - ttl-propagate + - ip-proto + - sport + - dport + - nh-id + - flowlabel - name: delroute doc: Delete an existing route
@@ -321,7 +349,7 @@ operations: do: request: value: 25 - attributes: *all-route-attrs + attributes: *route-req-attrs - name: newroute-ntf doc: Notification about a created route.
Signature · 1 line
2.43.0
PATCH v5 9/10 netlink: specs: rt-route: add the route deletion reason
In reply to Yuyang HuangAdd the del-reason attribute and its enum to the route attribute set, and to the getroute reply, which the route notifications reuse.
The attribute is absent from the newroute and delroute request lists. RTA_DEL_REASON is above strict_start_type in rtm_ipv6_policy, so encoding it in a request is rejected with -EINVAL.
--- Documentation/netlink/specs/rt-route.yaml | 17 +++++++++++++++++ 1 file changed, 17 insertions(+)
Documentation/netlink/specs/rt-route.yaml +17 -0
@@ -78,6 +78,15 @@ definitions: - name: rta-used type: u32 + - + name: del-reason + type: enum + name-prefix: rt-del-reason- + enum-name: rt-del-reason + entries: + - unspec + - expired + - ra-withdrawn attribute-sets: -
@@ -185,6 +194,10 @@ attribute-sets: type: u32 byte-order: big-endian display-hint: hex + - + name: del-reason + type: u32 + enum: del-reason - name: metrics name-prefix: rtax-
@@ -299,6 +312,7 @@ operations: - dport - nh-id - flowlabel + - del-reason dump: request: value: 26
@@ -313,6 +327,9 @@ operations: do: request: value: 24 + # del-reason is notification-only. It is above + # strict_start_type in rtm_ipv6_policy, so the kernel rejects + # it in requests. attributes: &route-req-attrs - dst - src
Signature · 1 line
2.43.0
PATCH v5 10/10 selftests: net: verify RTA_DEL_REASON on route deletion
In reply to Yuyang HuangExtend rtnetlink.py to check the reason reported in RTM_DELROUTE:
- expired: route with a 2s lifetime collected by the fib6 GC (gc_interval lowered like fib_tests.sh fib6_gc_test does);
- ra-withdrawn: a single RA advertises a default route (router lifetime), an on-link prefix route (RFC 4861 prefix information option) and a route information option route (RFC 4191), then a second RA withdraws all three with zero lifetimes; the RAs are crafted over a raw ICMPv6 socket so the test does not depend on an external RA tool;
- absence: a userspace deletion request records no cause and must not carry the attribute at all.
--- .../testing/selftests/net/lib/py/__init__.py | 4 +- tools/testing/selftests/net/lib/py/ynl.py | 7 +- tools/testing/selftests/net/rtnetlink.py | 181 +++++++++++++++++- 3 files changed, 187 insertions(+), 5 deletions(-)
tools/testing/selftests/net/lib/py/__init__.py +2 -2
@@ -17,7 +17,7 @@ from .utils import CmdExitFailure, fd_read_timeout, cmd, bkg, defer, \ wait_file, tool, tc from .bpf import bpf_map_set, bpf_map_dump, bpf_prog_map_ids from .ynl import NlError, NlctrlFamily, YnlFamily, \ - EthtoolFamily, NetdevFamily, RtnlFamily, RtnlAddrFamily + EthtoolFamily, NetdevFamily, RtnlFamily, RtnlAddrFamily, RtnlRouteFamily from .ynl import NetshaperFamily, DevlinkFamily, PSPFamily, Netlink __all__ = ["KSRC",
@@ -34,4 +34,4 @@ __all__ = ["KSRC", "NetdevSim", "NetdevSimDev", "NetshaperFamily", "DevlinkFamily", "PSPFamily", "NlError", "YnlFamily", "EthtoolFamily", "NetdevFamily", "RtnlFamily", - "NlctrlFamily", "RtnlAddrFamily", "Netlink"] + "NlctrlFamily", "RtnlAddrFamily", "RtnlRouteFamily", "Netlink"]
tools/testing/selftests/net/lib/py/ynl.py +6 -1
@@ -29,7 +29,7 @@ except ModuleNotFoundError as e: __all__ = [ "NlError", "NlPolicy", "Netlink", "YnlFamily", "SPEC_PATH", - "EthtoolFamily", "RtnlFamily", "RtnlAddrFamily", + "EthtoolFamily", "RtnlFamily", "RtnlAddrFamily", "RtnlRouteFamily", "NetdevFamily", "NetshaperFamily", "NlctrlFamily", "DevlinkFamily", "PSPFamily", ]
@@ -54,6 +54,11 @@ class RtnlAddrFamily(YnlFamily): super().__init__((SPEC_PATH / Path('rt-addr.yaml')).as_posix(), schema='', recv_size=recv_size) +class RtnlRouteFamily(YnlFamily): + def __init__(self, recv_size=0): + super().__init__((SPEC_PATH / Path('rt-route.yaml')).as_posix(), + schema='', recv_size=recv_size) + class NetdevFamily(YnlFamily): def __init__(self, recv_size=0): super().__init__((SPEC_PATH / Path('netdev.yaml')).as_posix(),
tools/testing/selftests/net/rtnetlink.py +179 -2
194 lines not on this page · raw message
Signature · 1 line
2.43.0
I checked the sashiko's comment:
https://patchwork.kernel.org/project/netdevbpf/patch/[email protected]/
Quoted 8 lines
Should this new UAPI attribute also be added to the Netlink YAML specification?
Is this new attribute missing from the netlink YAML specification?
should this new attribute be documented inDocumentation/netlink/specs/rt-route.yaml
Quoted 7 lines
Is the del-reason (
RTA_DEL_REASON) attribute missing from the YAML specification?Is there a missing patch in this series to actually report the deletion reason?
Does this
del_reasonactually get passed from the deletion path?
These are answered by later patches in the same series.
Quoted 2 lines
This is a pre-existing issue, but could this
RTM_NEWROUTEbe emitted out of order with respect to anRTM_DELROUTE?
Not related to the current series but might be worth sending a follow up separately.
Quoted 2 lines
Should this new enum and its entries include doc properties to describe the specific deletion reasons for the uAPI documentation?
Agreed, but it is a nit. If I need to send out patch series 6 due to adjusting later review comments, I will include the fix for this comment as well.
Quoted 5 lines
Is
RTA_DEL_REASONmissing fromrtm_ipv6_policyandrtm_ipv4_policy? [...] Should it be added to the policies asNLA_IGNOREorNLA_REJECT?
Will the newRTA_DEL_REASONattribute be rejected if user space echoes it back? [...] causing an ABI breakage.
The comment seems wrong. The attribute is kernel to user space notification-only, user space code should never set it.
Thanks,
Yuyang
Reply to comment in the other sashiko run:
https://netdev-ai.bots.linux.dev/sashiko/#/patchset/20260804014714.4362-1-sigefriedhyy%40gmail.com
Quoted 10 lines
Should the netlink socket returned here be closed by the callers? [...] Would a defer(rtnl.close) [...] make teardown explicit?
Same question for this raw ICMPv6 socket [...] Would a defer(sock.close) here be preferable?
is the stated reason for the retry loop accurate? [...]
addrconf_dad_begin()takes the early branch and never sends a DAD probe at all [...] Could the docstring describe that tentative window instead?
The comments above look valid, but I do not think they affect the correctness of the test code, so it does not seem worth sending a v6 just to fix them. If a v6 is needed for other reasons, I will fix them there.
Quoted 1 line
Should the #else branch also get a stub for
ip6_del_rt_reason()?
The ip6_del_rt() stub exists only because __remove_nexthop_fib() in net/ipv4/nexthop.c is obj-y and calls it with CONFIG_IPV6=n. The new helper has no caller outside net/ipv6/, so a stub for it would be dead code. I would rather add one when a caller needs it.
Thanks,
Yuyang
Ido Schimmel
maintainerQuoted 3 lines · Yuyang Huang
Reply to comment in the other sashiko run:
https://netdev-ai.bots.linux.dev/sashiko/#/patchset/20260804014714.4362-1-sigefriedhyy%40gmail.com
[...]
Quoted 6 lines
Quoted 1 line
Should the #else branch also get a stub for
ip6_del_rt_reason()?The
ip6_del_rt()stub exists only because__remove_nexthop_fib()innet/ipv4/nexthop.cis obj-y and calls it withCONFIG_IPV6=n. The new helper has no caller outsidenet/ipv6/, so a stub for it would be dead code. I would rather add one when a caller needs it.
But why put it under '#if IS_ENABLED(CONFIG_IPV6)' in the first place? Just move it above, next to ip6_ins_rt()...
Quoted 2 lines · Ido Schimmel
But why put it under '#if
IS_ENABLED(CONFIG_IPV6)' in the first place? Just move it above, next toip6_ins_rt()...
Thanks for the comment, will fix it in patchset v6.