Jiri Wiesner ee4b9c
From de6d25924c2a8c2988c6a385990cafbe742061bf Mon Sep 17 00:00:00 2001
Jiri Wiesner ee4b9c
From: Davide Caratti <dcaratti@redhat.com>
Jiri Wiesner ee4b9c
Date: Wed, 24 Nov 2021 17:14:40 +0100
Jiri Wiesner ee4b9c
Subject: [PATCH] net/sched: sch_ets: don't peek at classes beyond 'nbands'
Jiri Wiesner ee4b9c
Git-commit: de6d25924c2a8c2988c6a385990cafbe742061bf
Jiri Wiesner ee4b9c
Patch-mainline: v5.16-rc3
Jiri Wiesner ee4b9c
References: bsc#1207361
Jiri Wiesner ee4b9c
Jiri Wiesner ee4b9c
when the number of DRR classes decreases, the round-robin active list can
Jiri Wiesner ee4b9c
contain elements that have already been freed in ets_qdisc_change(). As a
Jiri Wiesner ee4b9c
consequence, it's possible to see a NULL dereference crash, caused by the
Jiri Wiesner ee4b9c
attempt to call cl->qdisc->ops->peek(cl->qdisc) when cl->qdisc is NULL:
Jiri Wiesner ee4b9c
Jiri Wiesner ee4b9c
 BUG: kernel NULL pointer dereference, address: 0000000000000018
Jiri Wiesner ee4b9c
 #PF: supervisor read access in kernel mode
Jiri Wiesner ee4b9c
 #PF: error_code(0x0000) - not-present page
Jiri Wiesner ee4b9c
 PGD 0 P4D 0
Jiri Wiesner ee4b9c
 Oops: 0000 [#1] PREEMPT SMP NOPTI
Jiri Wiesner ee4b9c
 CPU: 1 PID: 910 Comm: mausezahn Not tainted 5.16.0-rc1+ #475
Jiri Wiesner ee4b9c
 Hardware name: Red Hat KVM, BIOS 1.11.1-4.module+el8.1.0+4066+0f1aadab 04/01/2014
Jiri Wiesner ee4b9c
 RIP: 0010:ets_qdisc_dequeue+0x129/0x2c0 [sch_ets]
Jiri Wiesner ee4b9c
 Code: c5 01 41 39 ad e4 02 00 00 0f 87 18 ff ff ff 49 8b 85 c0 02 00 00 49 39 c4 0f 84 ba 00 00 00 49 8b ad c0 02 00 00 48 8b 7d 10 <48> 8b 47 18 48 8b 40 38 0f ae e8 ff d0 48 89 c3 48 85 c0 0f 84 9d
Jiri Wiesner ee4b9c
 RSP: 0000:ffffbb36c0b5fdd8 EFLAGS: 00010287
Jiri Wiesner ee4b9c
 RAX: ffff956678efed30 RBX: 0000000000000000 RCX: 0000000000000000
Jiri Wiesner ee4b9c
 RDX: 0000000000000002 RSI: ffffffff9b938dc9 RDI: 0000000000000000
Jiri Wiesner ee4b9c
 RBP: ffff956678efed30 R08: e2f3207fe360129c R09: 0000000000000000
Jiri Wiesner ee4b9c
 R10: 0000000000000001 R11: 0000000000000001 R12: ffff956678efeac0
Jiri Wiesner ee4b9c
 R13: ffff956678efe800 R14: ffff956611545000 R15: ffff95667ac8f100
Jiri Wiesner ee4b9c
 FS:  00007f2aa9120740(0000) GS:ffff95667b800000(0000) knlGS:0000000000000000
Jiri Wiesner ee4b9c
 CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
Jiri Wiesner ee4b9c
 CR2: 0000000000000018 CR3: 000000011070c000 CR4: 0000000000350ee0
Jiri Wiesner ee4b9c
 Call Trace:
Jiri Wiesner ee4b9c
  <TASK>
Jiri Wiesner ee4b9c
  qdisc_peek_dequeued+0x29/0x70 [sch_ets]
Jiri Wiesner ee4b9c
  tbf_dequeue+0x22/0x260 [sch_tbf]
Jiri Wiesner ee4b9c
  __qdisc_run+0x7f/0x630
Jiri Wiesner ee4b9c
  net_tx_action+0x290/0x4c0
Jiri Wiesner ee4b9c
  __do_softirq+0xee/0x4f8
Jiri Wiesner ee4b9c
  irq_exit_rcu+0xf4/0x130
Jiri Wiesner ee4b9c
  sysvec_apic_timer_interrupt+0x52/0xc0
Jiri Wiesner ee4b9c
  asm_sysvec_apic_timer_interrupt+0x12/0x20
Jiri Wiesner ee4b9c
 RIP: 0033:0x7f2aa7fc9ad4
Jiri Wiesner ee4b9c
 Code: b9 ff ff 48 8b 54 24 18 48 83 c4 08 48 89 ee 48 89 df 5b 5d e9 ed fc ff ff 0f 1f 00 66 2e 0f 1f 84 00 00 00 00 00 f3 0f 1e fa <53> 48 83 ec 10 48 8b 05 10 64 33 00 48 8b 00 48 85 c0 0f 85 84 00
Jiri Wiesner ee4b9c
 RSP: 002b:00007ffe5d33fab8 EFLAGS: 00000202
Jiri Wiesner ee4b9c
 RAX: 0000000000000002 RBX: 0000561f72c31460 RCX: 0000561f72c31720
Jiri Wiesner ee4b9c
 RDX: 0000000000000002 RSI: 0000561f72c31722 RDI: 0000561f72c31720
Jiri Wiesner ee4b9c
 RBP: 000000000000002a R08: 00007ffe5d33fa40 R09: 0000000000000014
Jiri Wiesner ee4b9c
 R10: 0000000000000000 R11: 0000000000000246 R12: 0000561f7187e380
Jiri Wiesner ee4b9c
 R13: 0000000000000000 R14: 0000000000000000 R15: 0000561f72c31460
Jiri Wiesner ee4b9c
  </TASK>
Jiri Wiesner ee4b9c
 Modules linked in: sch_ets sch_tbf dummy rfkill iTCO_wdt intel_rapl_msr iTCO_vendor_support intel_rapl_common joydev virtio_balloon lpc_ich i2c_i801 i2c_smbus pcspkr ip_tables xfs libcrc32c crct10dif_pclmul crc32_pclmul crc32c_intel ahci libahci ghash_clmulni_intel serio_raw libata virtio_blk virtio_console virtio_net net_failover failover sunrpc dm_mirror dm_region_hash dm_log dm_mod
Jiri Wiesner ee4b9c
 CR2: 0000000000000018
Jiri Wiesner ee4b9c
Jiri Wiesner ee4b9c
Ensuring that 'alist' was never zeroed [1] was not sufficient, we need to
Jiri Wiesner ee4b9c
remove from the active list those elements that are no more SP nor DRR.
Jiri Wiesner ee4b9c
Jiri Wiesner ee4b9c
[1] https://lore.kernel.org/netdev/60d274838bf09777f0371253416e8af71360bc08.1633609148.git.dcaratti@redhat.com/
Jiri Wiesner ee4b9c
Jiri Wiesner ee4b9c
v3: fix race between ets_qdisc_change() and ets_qdisc_dequeue() delisting
Jiri Wiesner ee4b9c
    DRR classes beyond 'nbands' in ets_qdisc_change() with the qdisc lock
Jiri Wiesner ee4b9c
    acquired, thanks to Cong Wang.
Jiri Wiesner ee4b9c
Jiri Wiesner ee4b9c
v2: when a NULL qdisc is found in the DRR active list, try to dequeue skb
Jiri Wiesner ee4b9c
    from the next list item.
Jiri Wiesner ee4b9c
Jiri Wiesner ee4b9c
Reported-by: Hangbin Liu <liuhangbin@gmail.com>
Jiri Wiesner ee4b9c
Fixes: dcc68b4d8084 ("net: sch_ets: Add a new Qdisc")
Jiri Wiesner ee4b9c
Signed-off-by: Davide Caratti <dcaratti@redhat.com>
Jiri Wiesner ee4b9c
Link: https://lore.kernel.org/r/7a5c496eed2d62241620bdbb83eb03fb9d571c99.1637762721.git.dcaratti@redhat.com
Jiri Wiesner ee4b9c
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Jiri Wiesner ee4b9c
Signed-off-by: Jiri Wiesner <jwiesner@suse.de>
Jiri Wiesner ee4b9c
---
Jiri Wiesner ee4b9c
 net/sched/sch_ets.c | 8 +++++---
Jiri Wiesner ee4b9c
 1 file changed, 5 insertions(+), 3 deletions(-)
Jiri Wiesner ee4b9c
Jiri Wiesner ee4b9c
diff --git a/net/sched/sch_ets.c b/net/sched/sch_ets.c
Jiri Wiesner ee4b9c
index 0eae9ff5edf6..e007fc75ef2f 100644
Jiri Wiesner ee4b9c
--- a/net/sched/sch_ets.c
Jiri Wiesner ee4b9c
+++ b/net/sched/sch_ets.c
Jiri Wiesner ee4b9c
@@ -665,12 +665,14 @@ static int ets_qdisc_change(struct Qdisc *sch, struct nlattr *opt,
Jiri Wiesner ee4b9c
 			q->classes[i].deficit = quanta[i];
Jiri Wiesner ee4b9c
 		}
Jiri Wiesner ee4b9c
 	}
Jiri Wiesner ee4b9c
+	for (i = q->nbands; i < oldbands; i++) {
Jiri Wiesner ee4b9c
+		qdisc_tree_flush_backlog(q->classes[i].qdisc);
Jiri Wiesner ee4b9c
+		if (i >= q->nstrict)
Jiri Wiesner ee4b9c
+			list_del(&q->classes[i].alist);
Jiri Wiesner ee4b9c
+	}
Jiri Wiesner ee4b9c
 	q->nstrict = nstrict;
Jiri Wiesner ee4b9c
 	memcpy(q->prio2band, priomap, sizeof(priomap));
Jiri Wiesner ee4b9c
 
Jiri Wiesner ee4b9c
-	for (i = q->nbands; i < oldbands; i++)
Jiri Wiesner ee4b9c
-		qdisc_tree_flush_backlog(q->classes[i].qdisc);
Jiri Wiesner ee4b9c
-
Jiri Wiesner ee4b9c
 	for (i = 0; i < q->nbands; i++)
Jiri Wiesner ee4b9c
 		q->classes[i].quantum = quanta[i];
Jiri Wiesner ee4b9c
 
Jiri Wiesner ee4b9c
-- 
Jiri Wiesner ee4b9c
2.35.3
Jiri Wiesner ee4b9c