linux/net/sched
Toke Høiland-Jørgensen cde71a5677 sched: sch_cake: fix bulk flow accounting logic for host fairness
commit 546ea84d07 upstream.

In sch_cake, we keep track of the count of active bulk flows per host,
when running in dst/src host fairness mode, which is used as the
round-robin weight when iterating through flows. The count of active
bulk flows is updated whenever a flow changes state.

This has a peculiar interaction with the hash collision handling: when a
hash collision occurs (after the set-associative hashing), the state of
the hash bucket is simply updated to match the new packet that collided,
and if host fairness is enabled, that also means assigning new per-host
state to the flow. For this reason, the bulk flow counters of the
host(s) assigned to the flow are decremented, before new state is
assigned (and the counters, which may not belong to the same host
anymore, are incremented again).

Back when this code was introduced, the host fairness mode was always
enabled, so the decrement was unconditional. When the configuration
flags were introduced the *increment* was made conditional, but
the *decrement* was not. Which of course can lead to a spurious
decrement (and associated wrap-around to U16_MAX).

AFAICT, when host fairness is disabled, the decrement and wrap-around
happens as soon as a hash collision occurs (which is not that common in
itself, due to the set-associative hashing). However, in most cases this
is harmless, as the value is only used when host fairness mode is
enabled. So in order to trigger an array overflow, sch_cake has to first
be configured with host fairness disabled, and while running in this
mode, a hash collision has to occur to cause the overflow. Then, the
qdisc has to be reconfigured to enable host fairness, which leads to the
array out-of-bounds because the wrapped-around value is retained and
used as an array index. It seems that syzbot managed to trigger this,
which is quite impressive in its own right.

This patch fixes the issue by introducing the same conditional check on
decrement as is used on increment.

The original bug predates the upstreaming of cake, but the commit listed
in the Fixes tag touched that code, meaning that this patch won't apply
before that.

Fixes: 7126399299 ("sch_cake: Make the dual modes fairer")
Reported-by: syzbot+7fe7b81d602cc1e6b94d@syzkaller.appspotmail.com
Signed-off-by: Toke Høiland-Jørgensen <toke@redhat.com>
Link: https://patch.msgid.link/20240903160846.20909-1-toke@redhat.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2024-09-12 11:07:44 +02:00
..
act_api.c net/sched: act_api: fix possible infinite loop in tcf_idr_check_alloc() 2024-07-05 09:14:30 +02:00
act_bpf.c flow_offload: fill flags to action structure 2023-02-22 12:57:10 +01:00
act_connmark.c flow_offload: fill flags to action structure 2023-02-22 12:57:10 +01:00
act_csum.c net_sched: refactor TC action init API 2021-08-02 10:24:38 +01:00
act_ct.c sched: act_ct: take care of padding in struct zones_ht_key 2024-08-19 05:45:31 +02:00
act_ctinfo.c net/sched: act_ctinfo: use percpu stats 2023-02-22 12:57:10 +01:00
act_gact.c net_sched: refactor TC action init API 2021-08-02 10:24:38 +01:00
act_gate.c flow_offload: fill flags to action structure 2023-02-22 12:57:10 +01:00
act_ife.c flow_offload: fill flags to action structure 2023-02-22 12:57:10 +01:00
act_ipt.c net/sched: act_ipt: add sanity checks on table name and hook locations 2023-07-23 13:47:28 +02:00
act_meta_mark.c
act_meta_skbprio.c
act_meta_skbtcindex.c
act_mirred.c net/sched: act_mirred: Add carrier check 2023-05-17 11:50:17 +02:00
act_mpls.c net/sched: act_mpls: fix action bind logic 2023-03-11 13:57:30 +01:00
act_nat.c flow_offload: fill flags to action structure 2023-02-22 12:57:10 +01:00
act_pedit.c net/sched: act_pedit: Add size check for TCA_PEDIT_PARMS_EX 2023-07-23 13:47:29 +02:00
act_police.c net: sched: act_police: fix sparse errors in tcf_police_dump() 2023-06-14 11:13:03 +02:00
act_sample.c net/sched: act_sample: fix action bind logic 2023-03-11 13:57:30 +01:00
act_simple.c flow_offload: fill flags to action structure 2023-02-22 12:57:10 +01:00
act_skbedit.c flow_offload: fill flags to action structure 2023-02-22 12:57:10 +01:00
act_skbmod.c net/sched: act_skbmod: prevent kernel-infoleak 2024-04-10 16:19:38 +02:00
act_tunnel_key.c net_sched: refactor TC action init API 2021-08-02 10:24:38 +01:00
act_vlan.c net_sched: refactor TC action init API 2021-08-02 10:24:38 +01:00
cls_api.c net/sched: cls_api: Fix lockup on flushing explicitly created chain 2023-06-21 15:59:18 +02:00
cls_basic.c net_sched: refactor TC action init API 2021-08-02 10:24:38 +01:00
cls_bpf.c net: sched: cls_bpf: Undo tcf_bind_filter in case of an error 2023-07-27 08:47:00 +02:00
cls_cgroup.c net_sched: refactor TC action init API 2021-08-02 10:24:38 +01:00
cls_flow.c net_sched: refactor TC action init API 2021-08-02 10:24:38 +01:00
cls_flower.c net/sched: flower: Ensure both minimum and maximum ports are specified 2023-07-23 13:47:45 +02:00
cls_fw.c net/sched: cls_fw: No longer copy tcf_result on update to avoid use-after-free 2023-08-11 15:13:52 +02:00
cls_matchall.c net_sched: refactor TC action init API 2021-08-02 10:24:38 +01:00
cls_route.c net/sched: cls_route: No longer copy tcf_result on update to avoid use-after-free 2023-08-11 15:13:52 +02:00
cls_u32.c net: sched: cls_u32: Fix allocation size in u32_init() 2023-11-08 17:26:45 +01:00
em_canid.c
em_cmp.c
em_ipset.c
em_ipt.c
em_meta.c
em_nbyte.c net: sched: Return the correct errno code 2021-02-06 11:15:28 -08:00
em_text.c net: sched: em_text: fix possible memory leak in em_text_destroy() 2024-01-15 18:51:12 +01:00
em_u32.c
ematch.c net_sched: reject TCF_EM_SIMPLE case for complex ematch module 2022-12-31 13:14:39 +01:00
Kconfig net/sched: Retire dsmark qdisc 2024-03-01 13:21:42 +01:00
Makefile net/sched: Retire dsmark qdisc 2024-03-01 13:21:42 +01:00
sch_api.c net/sched: Refactor qdisc_graft() for ingress and clsact Qdiscs 2024-03-01 13:21:54 +01:00
sch_blackhole.c
sch_cake.c sched: sch_cake: fix bulk flow accounting logic for host fairness 2024-09-12 11:07:44 +02:00
sch_cbs.c
sch_choke.c net: sched: delete duplicate cleanup of backlog and qlen 2022-10-29 10:12:57 +02:00
sch_codel.c
sch_drr.c net: sched: delete duplicate cleanup of backlog and qlen 2022-10-29 10:12:57 +02:00
sch_etf.c net: sched: delete duplicate cleanup of backlog and qlen 2022-10-29 10:12:57 +02:00
sch_ets.c net: sched: delete duplicate cleanup of backlog and qlen 2022-10-29 10:12:57 +02:00
sch_fifo.c net_sched: fix NULL deref in fifo_set_limit() 2021-10-01 14:59:10 -07:00
sch_fq_codel.c net: sched: delete duplicate cleanup of backlog and qlen 2022-10-29 10:12:57 +02:00
sch_fq_pie.c net/sched: fq_pie: avoid stalls in fq_pie_timer() 2023-09-19 12:22:58 +02:00
sch_fq.c net/sched: sch_fq: fix integer overflow of "credit" 2023-05-11 23:00:31 +09:00
sch_frag.c net/sched: Extend qdisc control block with tc control block 2022-01-05 12:42:33 +01:00
sch_generic.c net/sched: fix netdevice reference leaks in attach_default_qdiscs() 2022-09-08 12:28:02 +02:00
sch_gred.c net: sched: Fix spelling mistakes 2021-05-31 22:44:56 -07:00
sch_hfsc.c net/sched: sch_hfsc: upgrade 'rt' to 'sc' when it becomes a inner curve 2023-10-25 11:58:57 +02:00
sch_hhf.c
sch_htb.c net: sched: sch: Fix off by one in htb_activate_prios() 2023-02-22 12:57:11 +01:00
sch_ingress.c net/sched: Reserve TC_H_INGRESS (TC_H_CLSACT) for ingress (clsact) Qdiscs 2023-06-09 10:32:17 +02:00
sch_mq.c net: sched: update default qdisc visibility after Tx queue cnt changes 2021-11-18 19:16:10 +01:00
sch_mqprio.c net/sched: mqprio: Add length check for TCA_MQPRIO_{MAX/MIN}_RATE64 2023-08-03 10:22:36 +02:00
sch_multiq.c net: sched: sch_multiq: fix possible OOB write in multiq_tune() 2024-07-05 09:14:06 +02:00
sch_netem.c sch/netem: fix use after free in netem_dequeue 2024-09-12 11:07:42 +02:00
sch_pie.c
sch_plug.c net: sched: sch_qfq: Fix UAF in qfq_dequeue() 2023-09-19 12:22:59 +02:00
sch_prio.c net: sched: delete duplicate cleanup of backlog and qlen 2022-10-29 10:12:57 +02:00
sch_qfq.c net: sched: sch_qfq: Fix UAF in qfq_dequeue() 2023-09-19 12:22:59 +02:00
sch_red.c net: sched: Fix use after free in red_enqueue() 2022-11-10 18:15:28 +01:00
sch_sfb.c net: sched: sfb: fix null pointer access issue when sfb_init() fails 2022-10-29 10:12:57 +02:00
sch_sfq.c net/sched: store the last executed chain also for clsact egress 2021-07-29 22:17:37 +01:00
sch_skbprio.c net: sched: delete duplicate cleanup of backlog and qlen 2022-10-29 10:12:57 +02:00
sch_taprio.c net/sched: taprio: always validate TCA_TAPRIO_ATTR_PRIOMAP 2024-07-05 09:14:07 +02:00
sch_tbf.c net: sched: delete duplicate cleanup of backlog and qlen 2022-10-29 10:12:57 +02:00
sch_teql.c net: sched: delete duplicate cleanup of backlog and qlen 2022-10-29 10:12:57 +02:00