On 06.03.2026 18:21, Mark Michelson wrote: > On Fri, Mar 6, 2026 at 4:51 AM Ales Musil <[email protected]> wrote: >> >> >> On Thu, Mar 5, 2026 at 6:44 PM Mark Michelson via dev >> <[email protected]> wrote: >>> Hi Alexandra, thanks for the patch! >>> >>> I have a couple of small notes below. They're so minor, that I think >>> that a maintainer can fix them when merging, so there is no need for a >>> v4. >>> >>> Acked-by: Mark Michelson <[email protected]> >>> >>> On Mon, Mar 2, 2026 at 7:35 AM Alexandra Rukomoinikova >>> <[email protected]> wrote: >>>> The learned routes hash previously only used route parameters (prefix, >>>> nexthop, origin, route_table). This caused collisions when identical >>>> routes were learned through different transit switches, leading to >>>> infinite route re-learning. >>>> >>>> While such configurations are invalid due to overlapping IP addressing >>>> - ICshould handle them without looping. >>>> >>>> Fix by including the IC route UUID in the hash to distinguish routes >>>> with identical parameters but different origins. As a result, routers >>>> will now learn the correct number of prefixes as ECMP routes, making >>>> the misconfiguration visible to the user. >>>> >>>> Signed-off-by: Alexandra Rukomoinikova <[email protected]> >>>> --- >>>> v1 --> v2: fixed typo in the tests >>>> v2 --> v3: fixed hash calculation >>>> --- >>>> ic/ovn-ic.c | 56 ++++++++++++++-------------- >>>> tests/ovn-ic.at | 98 +++++++++++++++++++++++++++++++++++++++++++++++++ >>>> 2 files changed, 127 insertions(+), 27 deletions(-) >>>> >>>> diff --git a/ic/ovn-ic.c b/ic/ovn-ic.c >>>> index 95d73cb4b..7d2f9442a 100644 >>>> --- a/ic/ovn-ic.c >>>> +++ b/ic/ovn-ic.c >>>> @@ -1286,6 +1286,7 @@ struct ic_route_info { >>>> const char *origin; >>>> const char *route_table; >>>> const char *route_tag; >>>> + struct uuid ic_route_uuid; >>>> >>>> const struct nbrec_logical_router *nb_lr; >>>> >>>> @@ -1305,9 +1306,11 @@ struct ic_route_info { >>>> static uint32_t >>>> ic_route_hash(const struct in6_addr *prefix, unsigned int plen, >>>> const struct in6_addr *nexthop, const char *origin, >>>> - const char *route_table) >>>> + const char *route_table, const struct uuid *ic_route_uuid) >>>> { >>>> - uint32_t basis = hash_bytes(prefix, sizeof *prefix, (uint32_t)plen); >>>> + uint32_t basis = ic_route_uuid ? uuid_hash(ic_route_uuid) : 0; >>>> + basis = hash_bytes(prefix, sizeof *prefix, basis); >>>> + basis = hash_int((uint32_t) plen, basis); >>>> basis = hash_string(origin, basis); >>>> basis = hash_string(route_table, basis); >>>> return hash_bytes(nexthop, sizeof *nexthop, basis); >>>> @@ -1316,18 +1319,22 @@ ic_route_hash(const struct in6_addr *prefix, >>>> unsigned int plen, >>>> static struct ic_route_info * >>>> ic_route_find(struct hmap *routes, const struct in6_addr *prefix, >>>> unsigned int plen, const struct in6_addr *nexthop, >>>> - const char *origin, const char *route_table, uint32_t hash) >>>> + const char *origin, const char *route_table, >>>> + const struct uuid *ic_route_uuid, uint32_t hash) >>>> { >>>> struct ic_route_info *r; >>>> if (!hash) { >>>> - hash = ic_route_hash(prefix, plen, nexthop, origin, route_table); >>>> + hash = ic_route_hash(prefix, plen, nexthop, origin, route_table, >>>> + ic_route_uuid); >>>> } >>>> HMAP_FOR_EACH_WITH_HASH (r, node, hash, routes) { >>>> if (ipv6_addr_equals(&r->prefix, prefix) && >>>> r->plen == plen && >>>> ipv6_addr_equals(&r->nexthop, nexthop) && >>>> !strcmp(r->origin, origin) && >>>> - !strcmp(r->route_table ? r->route_table : "", route_table)) { >>>> + !strcmp(r->route_table ? r->route_table : "", route_table) && >>>> + (!ic_route_uuid || uuid_equals(&r->ic_route_uuid, >>>> + ic_route_uuid))) { >>>> return r; >>>> } >>>> } >>>> @@ -1370,7 +1377,8 @@ parse_route(const char *s_prefix, const char >>>> *s_nexthop, >>>> static bool >>>> add_to_routes_learned(struct hmap *routes_learned, >>>> const struct nbrec_logical_router_static_route >>>> *nb_route, >>>> - const struct nbrec_logical_router *nb_lr) >>>> + const struct nbrec_logical_router *nb_lr, >>>> + const struct uuid *ic_route_uuid) >>>> { >>>> struct in6_addr prefix, nexthop; >>>> unsigned int plen; >>>> @@ -1379,8 +1387,11 @@ add_to_routes_learned(struct hmap *routes_learned, >>>> return false; >>>> } >>>> const char *origin = smap_get_def(&nb_route->options, "origin", ""); >>>> + >>>> + uint32_t hash = ic_route_hash(&prefix, plen, &nexthop, origin, >>>> + nb_route->route_table, ic_route_uuid); >>>> if (ic_route_find(routes_learned, &prefix, plen, &nexthop, origin, >>>> - nb_route->route_table, 0)) { >>>> + nb_route->route_table, ic_route_uuid, hash)) { >>>> /* Route was added to learned on previous iteration. */ >>>> return true; >>>> } >>>> @@ -1393,9 +1404,9 @@ add_to_routes_learned(struct hmap *routes_learned, >>>> ic_route->origin = origin; >>>> ic_route->route_table = nb_route->route_table; >>>> ic_route->nb_lr = nb_lr; >>>> - hmap_insert(routes_learned, &ic_route->node, >>>> - ic_route_hash(&prefix, plen, &nexthop, origin, >>>> - nb_route->route_table)); >>>> + memcpy(&ic_route->ic_route_uuid, ic_route_uuid, sizeof(struct uuid)); >>> There's no need for memcpy here. You can just do a struct assignment: >>> >>> ic_route->ic_route_uuid = *ic_route_uuid; >>> >>>> + hmap_insert(routes_learned, &ic_route->node, hash); >>>> + >>>> return true; >>>> } >>>> >>>> @@ -1562,10 +1573,11 @@ add_to_routes_ad(struct hmap *routes_ad, const >>>> struct in6_addr prefix, >>>> route_table = ""; >>>> } >>>> >>>> - uint hash = ic_route_hash(&prefix, plen, &nexthop, origin, >>>> route_table); >>>> + uint hash = ic_route_hash(&prefix, plen, &nexthop, origin, >>>> + route_table, NULL); >>>> >>>> if (!ic_route_find(routes_ad, &prefix, plen, &nexthop, origin, >>>> route_table, >>>> - hash)) { >>>> + NULL, hash)) { >>>> struct ic_route_info *ic_route = xzalloc(sizeof *ic_route); >>>> ic_route->prefix = prefix; >>>> ic_route->plen = plen; >>>> @@ -2154,20 +2166,9 @@ sync_learned_routes(struct ic_context *ctx, >>>> struct ic_route_info *route_learned >>>> = ic_route_find(&ic_lr->routes_learned, &prefix, plen, >>>> &nexthop, isb_route->origin, >>>> - isb_route->route_table, 0); >>>> + isb_route->route_table, >>>> + &isb_route->header_.uuid, 0); >>>> if (route_learned) { >>>> - /* Sync external-ids */ >>>> - struct uuid ext_id; >>>> - smap_get_uuid(&route_learned->nb_route->external_ids, >>>> - "ic-learned-route", &ext_id); >>>> - if (!uuid_equals(&ext_id, &isb_route->header_.uuid)) { >>>> - char *uuid_s = >>>> - xasprintf(UUID_FMT, >>>> - UUID_ARGS(&isb_route->header_.uuid)); >>>> - >>>> nbrec_logical_router_static_route_update_external_ids_setkey( >>>> - route_learned->nb_route, "ic-learned-route", >>>> uuid_s); >>>> - free(uuid_s); >>>> - } >>>> hmap_remove(&ic_lr->routes_learned, >>>> &route_learned->node); >>>> free(route_learned); >>>> } else { >>>> @@ -2275,7 +2276,7 @@ advertise_routes(struct ic_context *ctx, >>>> } >>>> struct ic_route_info *route_adv = >>>> ic_route_find(routes_ad, &prefix, plen, &nexthop, >>>> - isb_route->origin, isb_route->route_table, 0); >>>> + isb_route->origin, isb_route->route_table, >>>> NULL, 0); >>>> if (!route_adv) { >>>> /* Delete the extra route from IC-SB. */ >>>> VLOG_DBG("Delete route %s -> %s from IC-SB, which is not >>>> found" >>>> @@ -2349,7 +2350,8 @@ build_ts_routes_to_adv(struct ic_context *ctx, >>>> if (smap_get_uuid(&nb_route->external_ids, "ic-learned-route", >>>> &isb_uuid)) { >>>> /* It is a learned route */ >>>> - if (!add_to_routes_learned(&ic_lr->routes_learned, nb_route, >>>> lr)) { >>>> + if (!add_to_routes_learned(&ic_lr->routes_learned, nb_route, >>>> lr, >>>> + &isb_uuid)) { >>>> static struct vlog_rate_limit rl = >>>> VLOG_RATE_LIMIT_INIT(5, 1); >>>> VLOG_WARN_RL(&rl, "Bad format of learned route in NB: " >>>> "%s -> %s. Delete it.", nb_route->ip_prefix, >>>> diff --git a/tests/ovn-ic.at b/tests/ovn-ic.at >>>> index 1a826aa1c..c633cdeb4 100644 >>>> --- a/tests/ovn-ic.at >>>> +++ b/tests/ovn-ic.at >>>> @@ -4594,3 +4594,101 @@ OVN_CLEANUP_IC([az1], [az2]) >>>> >>>> AT_CLEANUP >>>> ]) >>>> + >>>> +OVN_FOR_EACH_NORTHD([ >>>> +AT_SETUP([ovn-ic -- check loop with annonce duplicated prefixes - >>>> multiple ts]) >>> s/annonce/announce/ >>> >>> The same mistake happens a few other times in this test as well. >>> >>> >>> >>>> + >>>> +ovn_init_ic_db >>>> +ovn-ic-nbctl ts-add ts11 >>>> +ovn-ic-nbctl ts-add ts12 >>>> + >>>> +for i in 1 2; do >>>> + ovn_start az$i >>>> + ovn_as az$i >>>> + >>>> + # Enable route learning at AZ level >>>> + check ovn-nbctl set nb_global . options:ic-route-learn=true >>>> + # Enable route advertising at AZ level >>>> + check ovn-nbctl set nb_global . options:ic-route-adv=true >>>> +done >>>> + >>>> +for i in 1 2; do >>>> + ovn_as az$i >>>> + >>>> + check ovn-nbctl lr-add lr$i >>>> + >>>> + check ovn-nbctl lrp-add lr$i lr-lrp-$i aa:aa:aa:aa:a1:0$i >>>> 169.254.101.1/24 >>>> + check ovn-nbctl lsp-add-router-port ts11 lsp-az$i lr-lrp-$i >>>> + >>>> + # Create subnet for annonce. >>>> + check ovn-nbctl lrp-add lr$i lr-lrp-subnet-$i aa:aa:aa:aa:a2:0$i >>>> 192.168.0.$i/24 >>>> + >>>> + # Create one more lrp-ts port through another transit switch. >>>> + check ovn-nbctl lrp-add lr$i lr-lrp-dup$i aa:aa:aa:aa:a3:01 >>>> 169.254.101.1/24 >>>> + check ovn-nbctl lsp-add-router-port ts12 lsp-dup-az$i lr-lrp-dup$i >>>> +done >>>> + >>>> +OVS_WAIT_FOR_OUTPUT([ovn_as az1 ovn-nbctl lr-route-list lr1 | grep >>>> 192.168 | >>>> + grep learned | awk '{print $1, $2}' | sort ], [0], >>>> [dnl >>>> +192.168.0.0/24 169.254.101.1 >>>> +192.168.0.0/24 169.254.101.1 >>>> +]) >>>> + >>>> +OVN_CLEANUP_IC([az1], [az2]) >>>> +AT_CLEANUP >>>> +]) >>>> + >>>> +OVN_FOR_EACH_NORTHD([ >>>> +AT_SETUP([ovn-ic -- check loop with annonce duplicated prefixes - single >>>> ts]) >>>> + >>>> +ovn_init_ic_db >>>> + >>>> +for i in 1 2 3; do >>>> + ovn_start az$i >>>> + ovn_as az$i >>>> + >>>> + # Enable route learning at AZ level >>>> + check ovn-nbctl set nb_global . options:ic-route-learn=true >>>> + # Enable route advertising at AZ level >>>> + check ovn-nbctl set nb_global . options:ic-route-adv=true >>>> +done >>>> + >>>> +# Create new transit switches and LRs. Test topology is next: >>>> +# logical router (lr12) >>>> +# | >>>> +# logical router (lr11) - / transit switch (ts11) \ - logical router >>>> (lr13) >>>> +# >>>> + >>>> +# Create lr11, lr13, ts11 and connect them >>>> +for i in 1 3; do >>>> + ovn_as az$i >>>> + >>>> + check ovn-nbctl lr-add lr1$i >>>> + check ovn-ic-nbctl --wait=sb --may-exist ts-add ts11 >>>> + >>>> + # Create LRP and connect to TS >>>> + check ovn-nbctl lrp-add lr1$i lrp-lr1$i-ts11 aa:aa:aa:aa:a1:0$i >>>> 169.254.101.1/24 >>>> + check ovn-nbctl lsp-add-router-port ts11 lsp-ts11-lr1$i lrp-lr1$i-ts11 >>>> +done >>>> + >>>> +# Create lr12 and connect it to ts11 >>>> +ovn_as az2 >>>> +check ovn-nbctl lr-add lr12 >>>> + >>>> +# Create LRP and connect to TS >>>> +check ovn-nbctl lrp-add lr12 lrp-lr12-ts11 aa:aa:aa:aa:a1:03 >>>> 169.254.101.2/24 >>>> +check ovn-nbctl lsp-add-router-port ts11 lsp-lr12-ts11 lrp-lr12-ts11 >>>> + >>>> +# Create directly-connected route in lr11 >>>> +check ovn_as az1 ovn-nbctl lrp-add lr11 lrp-lr11 aa:aa:aa:aa:bb:01 >>>> "192.168.0.1/24" >>>> +check ovn_as az3 ovn-nbctl lrp-add lr13 lrp-lr13 aa:aa:aa:aa:bb:03 >>>> "192.168.0.1/24" >>>> + >>>> +OVS_WAIT_FOR_OUTPUT([ovn_as az2 ovn-nbctl lr-route-list lr12 | grep >>>> 192.168 | >>>> + grep learned | awk '{print $1, $2}' | sort ], [0], >>>> [dnl >>>> +192.168.0.0/24 169.254.101.1 >>>> +192.168.0.0/24 169.254.101.1 >>>> +]) >>>> + >>>> +OVN_CLEANUP_IC([az1], [az2], [az3]) >>>> +AT_CLEANUP >>>> +]) >>>> -- >>>> 2.48.1 >>>> >>>> _______________________________________________ >>>> dev mailing list >>>> [email protected] >>>> https://mail.openvswitch.org/mailman/listinfo/ovs-dev >>>> >>> _______________________________________________ >>> dev mailing list >>> [email protected] >>> https://mail.openvswitch.org/mailman/listinfo/ovs-dev >> >> Thank you Alexandra and Mark, >> >> I'm slightly confused; the commit message says it's a fix, but there isn't a >> Fixes tag. Also based on the commit message about invalid configuration I >> don't think this needs a backport, please let me know if that's not the >> case. I've fixed the nits and applied to main. >> >> Regards, >> Ales > Hi Ales, > > A specific invalid configuration can cause `ovn-nbctl lr-route-list` > to loop infinitely, listing the same routes forever. If you try to run > the first new test added in this patch on an unpatched OVN main, then > the test fails because it continuously lists "192.168.0.0/24 > 169.254.101.1" in its output. This patch is ensuring that this invalid > configuration allows `ovn-nbctl lr-route-list` to work as expected. > > I tried to find what an appropriate "Fixes" tag would be, but I had > trouble pinpointing it. The issue of hash collisions for identical > routes appears to have existed since route advertising was added to > ovn-ic (57b347c55 ("ovn-ic: Route advertisement.")). I believe this > patch needs to be backported all the way down to branch-24.03. I > confirmed that if I apply the first test from this patch to > branch-24.03 and running it there, I see the same infinite loop > behavior as I see in unpatched main. > Sorry for the noise, I don't think I can read. I thought my patch wasn't working on 24.03, and after checking that it was working, I realized I must have misread your email =)
-- regards, Alexandra. _______________________________________________ dev mailing list [email protected] https://mail.openvswitch.org/mailman/listinfo/ovs-dev
