Skip to contentSkip to conversation

FAILED: patch "[PATCH] net/sched: ets: use old 'nbands' while purging unused classes" failed to apply to 5.10-stable tree

PATCH New [email protected] 2 participants 4 messages Aug 18, 2025 – Aug 19, 2025

Subsystems (3)
lore
Share & tools
lore

[email protected]

author maintainer
#0 lore raw

The patch below does not apply to the 5.10-stable tree.
If someone wants it applied there, or to any other stable or longterm tree, then please email the backport, including the original git commit id to <[email protected]>.

To reproduce the conflict and resubmit, you may use the following commands:

git fetch https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/ linux-5.10.y
git checkout FETCH_HEAD
git cherry-pick -x 87c6efc5ce9c126ae4a781bc04504b83780e3650
# <resolve conflicts, build, test, etc.>
git commit -s
git send-email --to '<[email protected]>' --in-reply-to '2025081858-blurred-unlinked-7eae@gregkh' --subject-prefix 'PATCH 5.10.y' HEAD^..

Possible dependencies:

thanks,

greg k-h

------------------ original commit in Linus's tree ------------------

From 87c6efc5ce9c126ae4a781bc04504b83780e3650 Mon Sep 17 00:00:00 2001
From: Davide Caratti <[email protected]>
Date: Tue, 12 Aug 2025 18:40:29 +0200
Subject: [PATCH] net/sched: ets: use old 'nbands' while purging unused classes

Shuang reported sch_ets test-case [1] crashing in ets_class_qlen_notify() after recent changes from Lion [2]. The problem is: in ets_qdisc_change() we purge unused DWRR queues; the value of 'q->nbands' is the new one, and the cleanup should be done with the old one. The problem is here since my first attempts to fix ets_qdisc_change(), but it surfaced again after the recent qdisc len accounting fixes. Fix it purging idle DWRR queues before assigning a new value of 'q->nbands', so that all purge operations find a consistent configuration:

- old 'q->nbands' because it's needed by ets_class_find()
- old 'q->nstrict' because it's needed by ets_class_is_strict()

 BUG: kernel NULL pointer dereference, address: 0000000000000000
 #PF: supervisor read access in kernel mode
 #PF: error_code(0x0000) - not-present page
 PGD 0 P4D 0
 Oops: Oops: 0000 [#1] SMP NOPTI
 CPU: 62 UID: 0 PID: 39457 Comm: tc Kdump: loaded Not tainted 6.12.0-116.el10.x86_64 #1 PREEMPT(voluntary)
 Hardware name: Dell Inc. PowerEdge R640/06DKY5, BIOS 2.12.2 07/09/2021
 RIP: 0010:__list_del_entry_valid_or_report+0x4/0x80
 Code: ff 4c 39 c7 0f 84 39 19 8e ff b8 01 00 00 00 c3 cc cc cc cc 66 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 f3 0f 1e fa <48> 8b 17 48 8b 4f 08 48 85 d2 0f 84 56 19 8e ff 48 85 c9 0f 84 ab
 RSP: 0018:ffffba186009f400 EFLAGS: 00010202
 RAX: 00000000000000d6 RBX: 0000000000000000 RCX: 0000000000000004
 RDX: ffff9f0fa29b69c0 RSI: 0000000000000000 RDI: 0000000000000000
 RBP: ffffffffc12c2400 R08: 0000000000000008 R09: 0000000000000004
 R10: ffffffffffffffff R11: 0000000000000004 R12: 0000000000000000
 R13: ffff9f0f8cfe0000 R14: 0000000000100005 R15: 0000000000000000
 FS:  00007f2154f37480(0000) GS:ffff9f269c1c0000(0000) knlGS:0000000000000000
 CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
 CR2: 0000000000000000 CR3: 00000001530be001 CR4: 00000000007726f0
 DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
 DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400
 PKRU: 55555554
 Call Trace:
  <TASK>
  ets_class_qlen_notify+0x65/0x90 [sch_ets]
  qdisc_tree_reduce_backlog+0x74/0x110
  ets_qdisc_change+0x630/0xa40 [sch_ets]
  __tc_modify_qdisc.constprop.0+0x216/0x7f0
  tc_modify_qdisc+0x7c/0x120
  rtnetlink_rcv_msg+0x145/0x3f0
  netlink_rcv_skb+0x53/0x100
  netlink_unicast+0x245/0x390
  netlink_sendmsg+0x21b/0x470
  ____sys_sendmsg+0x39d/0x3d0
  ___sys_sendmsg+0x9a/0xe0
  __sys_sendmsg+0x7a/0xd0
  do_syscall_64+0x7d/0x160
  entry_SYSCALL_64_after_hwframe+0x76/0x7e
 RIP: 0033:0x7f2155114084
 Code: 89 02 b8 ff ff ff ff eb bb 66 2e 0f 1f 84 00 00 00 00 00 0f 1f 00 f3 0f 1e fa 80 3d 25 f0 0c 00 00 74 13 b8 2e 00 00 00 0f 05 <48> 3d 00 f0 ff ff 77 54 c3 0f 1f 00 48 83 ec 28 89 54 24 1c 48 89
 RSP: 002b:00007fff1fd7a988 EFLAGS: 00000202 ORIG_RAX: 000000000000002e
 RAX: ffffffffffffffda RBX: 0000560ec063e5e0 RCX: 00007f2155114084
 RDX: 0000000000000000 RSI: 00007fff1fd7a9f0 RDI: 0000000000000003
 RBP: 00007fff1fd7aa60 R08: 0000000000000010 R09: 000000000000003f
 R10: 0000560ee9b3a010 R11: 0000000000000202 R12: 00007fff1fd7aae0
 R13: 000000006891ccde R14: 0000560ec063e5e0 R15: 00007fff1fd7aad0
  </TASK>

[1] https://lore.kernel.org/netdev/e08c7f4a6882f260011909a868311c6e9b54f3e4.1639153474.git.dcaratti@redhat.com/
[2] https://lore.kernel.org/netdev/[email protected]/

Fixes 103406b38c60 ("net/sched: Always pass notifications when child class becomes empty")
Fixes c062f2a0b04d ("net/sched: sch_ets: don't remove idle classes from the round-robin list")
Fixes dcc68b4d8084 ("net: sch_ets: Add a new Qdisc")
net/sched/sch_ets.c +6 -5
index 037f764822b9..82635dd2cfa5 100644
@@ -651,6 +651,12 @@ static int ets_qdisc_change(struct Qdisc *sch, struct nlattr *opt,
 
 	sch_tree_lock(sch);
 
+	for (i = nbands; i < oldbands; i++) {
+		if (i >= q->nstrict && q->classes[i].qdisc->q.qlen)
+			list_del_init(&q->classes[i].alist);
+		qdisc_purge_queue(q->classes[i].qdisc);
+	}
+
 	WRITE_ONCE(q->nbands, nbands);
 	for (i = nstrict; i < q->nstrict; i++) {
 		if (q->classes[i].qdisc->q.qlen) {
@@ -658,11 +664,6 @@ static int ets_qdisc_change(struct Qdisc *sch, struct nlattr *opt,
 			q->classes[i].deficit = quanta[i];
 		}
 	}
-	for (i = q->nbands; i < oldbands; i++) {
-		if (i >= q->nstrict && q->classes[i].qdisc->q.qlen)
-			list_del_init(&q->classes[i].alist);
-		qdisc_purge_queue(q->classes[i].qdisc);
-	}
 	WRITE_ONCE(q->nstrict, nstrict);
 	memcpy(q->prio2band, priomap, sizeof(priomap));
 
maintainer
#1 lore raw

PATCH 1/3 net/sched: sch_ets: properly init all active DRR list handles

From: Davide Caratti <[email protected]>

[ Upstream commit 454d3e1ae057a1e09a15905b06b860f60d6c14d0 ]

leaf classes of ETS qdiscs are served in strict priority or deficit round robin (DRR), depending on the value of 'nstrict'. Since this value can be changed while traffic is running, we need to be sure that the active list of DRR classes can be updated at any time, so:

1) call INIT_LIST_HEAD(&alist) on all leaf classes in .init(), before the first packet hits any of them.
2) ensure that 'alist' is not overwritten with zeros when a leaf class is no more strict priority nor DRR (i.e. array elements beyond 'nbands').

Stable-dep-of: 87c6efc5ce9c ("net/sched: ets: use old 'nbands' while purging unused classes")

---
 net/sched/sch_ets.c | 12 +++++++++---
 1 file changed, 9 insertions(+), 3 deletions(-)
net/sched/sch_ets.c +9 -3
index 4f4da11a2c77..b9e35a0a60cf 100644
@@ -666,7 +666,6 @@ static int ets_qdisc_change(struct Qdisc *sch, struct nlattr *opt,
 
 	q->nbands = nbands;
 	for (i = nstrict; i < q->nstrict; i++) {
-		INIT_LIST_HEAD(&q->classes[i].alist);
 		if (q->classes[i].qdisc->q.qlen) {
 			list_add_tail(&q->classes[i].alist, &q->active);
 			q->classes[i].deficit = quanta[i];
@@ -694,7 +693,11 @@ static int ets_qdisc_change(struct Qdisc *sch, struct nlattr *opt,
 	ets_offload_change(sch);
 	for (i = q->nbands; i < oldbands; i++) {
 		qdisc_put(q->classes[i].qdisc);
-		memset(&q->classes[i], 0, sizeof(q->classes[i]));
+		q->classes[i].qdisc = NULL;
+		q->classes[i].quantum = 0;
+		q->classes[i].deficit = 0;
+		memset(&q->classes[i].bstats, 0, sizeof(q->classes[i].bstats));
+		memset(&q->classes[i].qstats, 0, sizeof(q->classes[i].qstats));
 	}
 	return 0;
 }
@@ -703,7 +706,7 @@ static int ets_qdisc_init(struct Qdisc *sch, struct nlattr *opt,
 			  struct netlink_ext_ack *extack)
 {
 	struct ets_sched *q = qdisc_priv(sch);
-	int err;
+	int err, i;
 
 	if (!opt)
 		return -EINVAL;
@@ -713,6 +716,9 @@ static int ets_qdisc_init(struct Qdisc *sch, struct nlattr *opt,
 		return err;
 
 	INIT_LIST_HEAD(&q->active);
+	for (i = 0; i < TCQ_ETS_MAX_BANDS; i++)
+		INIT_LIST_HEAD(&q->classes[i].alist);
+
 	return ets_qdisc_change(sch, opt, extack);
 }
 
Signature · 1 line
2.50.1
maintainer
#2 lore raw

PATCH 2/3 net_sched: sch_ets: implement lockless ets_dump()

From: Eric Dumazet <[email protected]>

[ Upstream commit c5f1dde7f731e7bf2e7c169ca42cb4989fc2f8b9 ]

Instead of relying on RTNL, ets_dump() can use READ_ONCE() annotations, paired with WRITE_ONCE() ones in ets_change().

Stable-dep-of: 87c6efc5ce9c ("net/sched: ets: use old 'nbands' while purging unused classes")

---
 net/sched/sch_ets.c | 25 ++++++++++++++-----------
 1 file changed, 14 insertions(+), 11 deletions(-)
net/sched/sch_ets.c +14 -11
index b9e35a0a60cf..5c9a72f1e1f2 100644
@@ -664,7 +664,7 @@ static int ets_qdisc_change(struct Qdisc *sch, struct nlattr *opt,
 
 	sch_tree_lock(sch);
 
-	q->nbands = nbands;
+	WRITE_ONCE(q->nbands, nbands);
 	for (i = nstrict; i < q->nstrict; i++) {
 		if (q->classes[i].qdisc->q.qlen) {
 			list_add_tail(&q->classes[i].alist, &q->active);
@@ -676,11 +676,11 @@ static int ets_qdisc_change(struct Qdisc *sch, struct nlattr *opt,
 			list_del_init(&q->classes[i].alist);
 		qdisc_purge_queue(q->classes[i].qdisc);
 	}
-	q->nstrict = nstrict;
+	WRITE_ONCE(q->nstrict, nstrict);
 	memcpy(q->prio2band, priomap, sizeof(priomap));
 
 	for (i = 0; i < q->nbands; i++)
-		q->classes[i].quantum = quanta[i];
+		WRITE_ONCE(q->classes[i].quantum, quanta[i]);
 
 	for (i = oldbands; i < q->nbands; i++) {
 		q->classes[i].qdisc = queues[i];
@@ -694,7 +694,7 @@ static int ets_qdisc_change(struct Qdisc *sch, struct nlattr *opt,
 	for (i = q->nbands; i < oldbands; i++) {
 		qdisc_put(q->classes[i].qdisc);
 		q->classes[i].qdisc = NULL;
-		q->classes[i].quantum = 0;
+		WRITE_ONCE(q->classes[i].quantum, 0);
 		q->classes[i].deficit = 0;
 		memset(&q->classes[i].bstats, 0, sizeof(q->classes[i].bstats));
 		memset(&q->classes[i].qstats, 0, sizeof(q->classes[i].qstats));
@@ -751,6 +751,7 @@ static int ets_qdisc_dump(struct Qdisc *sch, struct sk_buff *skb)
 	struct ets_sched *q = qdisc_priv(sch);
 	struct nlattr *opts;
 	struct nlattr *nest;
+	u8 nbands, nstrict;
 	int band;
 	int prio;
 	int err;
@@ -763,21 +764,22 @@ static int ets_qdisc_dump(struct Qdisc *sch, struct sk_buff *skb)
 	if (!opts)
 		goto nla_err;
 
-	if (nla_put_u8(skb, TCA_ETS_NBANDS, q->nbands))
+	nbands = READ_ONCE(q->nbands);
+	if (nla_put_u8(skb, TCA_ETS_NBANDS, nbands))
 		goto nla_err;
 
-	if (q->nstrict &&
-	    nla_put_u8(skb, TCA_ETS_NSTRICT, q->nstrict))
+	nstrict = READ_ONCE(q->nstrict);
+	if (nstrict && nla_put_u8(skb, TCA_ETS_NSTRICT, nstrict))
 		goto nla_err;
 
-	if (q->nbands > q->nstrict) {
+	if (nbands > nstrict) {
 		nest = nla_nest_start(skb, TCA_ETS_QUANTA);
 		if (!nest)
 			goto nla_err;
 
-		for (band = q->nstrict; band < q->nbands; band++) {
+		for (band = nstrict; band < nbands; band++) {
 			if (nla_put_u32(skb, TCA_ETS_QUANTA_BAND,
-					q->classes[band].quantum))
+					READ_ONCE(q->classes[band].quantum)))
 				goto nla_err;
 		}
 
@@ -789,7 +791,8 @@ static int ets_qdisc_dump(struct Qdisc *sch, struct sk_buff *skb)
 		goto nla_err;
 
 	for (prio = 0; prio <= TC_PRIO_MAX; prio++) {
-		if (nla_put_u8(skb, TCA_ETS_PRIOMAP_BAND, q->prio2band[prio]))
+		if (nla_put_u8(skb, TCA_ETS_PRIOMAP_BAND,
+			       READ_ONCE(q->prio2band[prio])))
 			goto nla_err;
 	}
 
Signature · 1 line
2.50.1
maintainer
#3 lore raw

PATCH 3/3 net/sched: ets: use old 'nbands' while purging unused classes

In reply to Sasha Levin

From: Davide Caratti <[email protected]>

[ Upstream commit 87c6efc5ce9c126ae4a781bc04504b83780e3650 ]

Shuang reported sch_ets test-case [1] crashing in ets_class_qlen_notify() after recent changes from Lion [2]. The problem is: in ets_qdisc_change() we purge unused DWRR queues; the value of 'q->nbands' is the new one, and the cleanup should be done with the old one. The problem is here since my first attempts to fix ets_qdisc_change(), but it surfaced again after the recent qdisc len accounting fixes. Fix it purging idle DWRR queues before assigning a new value of 'q->nbands', so that all purge operations find a consistent configuration:

- old 'q->nbands' because it's needed by ets_class_find()
- old 'q->nstrict' because it's needed by ets_class_is_strict()

 BUG: kernel NULL pointer dereference, address: 0000000000000000
 #PF: supervisor read access in kernel mode
 #PF: error_code(0x0000) - not-present page
 PGD 0 P4D 0
 Oops: Oops: 0000 [#1] SMP NOPTI
 CPU: 62 UID: 0 PID: 39457 Comm: tc Kdump: loaded Not tainted 6.12.0-116.el10.x86_64 #1 PREEMPT(voluntary)
 Hardware name: Dell Inc. PowerEdge R640/06DKY5, BIOS 2.12.2 07/09/2021
 RIP: 0010:__list_del_entry_valid_or_report+0x4/0x80
 Code: ff 4c 39 c7 0f 84 39 19 8e ff b8 01 00 00 00 c3 cc cc cc cc 66 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 f3 0f 1e fa <48> 8b 17 48 8b 4f 08 48 85 d2 0f 84 56 19 8e ff 48 85 c9 0f 84 ab
 RSP: 0018:ffffba186009f400 EFLAGS: 00010202
 RAX: 00000000000000d6 RBX: 0000000000000000 RCX: 0000000000000004
 RDX: ffff9f0fa29b69c0 RSI: 0000000000000000 RDI: 0000000000000000
 RBP: ffffffffc12c2400 R08: 0000000000000008 R09: 0000000000000004
 R10: ffffffffffffffff R11: 0000000000000004 R12: 0000000000000000
 R13: ffff9f0f8cfe0000 R14: 0000000000100005 R15: 0000000000000000
 FS:  00007f2154f37480(0000) GS:ffff9f269c1c0000(0000) knlGS:0000000000000000
 CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
 CR2: 0000000000000000 CR3: 00000001530be001 CR4: 00000000007726f0
 DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
 DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400
 PKRU: 55555554
 Call Trace:
  <TASK>
  ets_class_qlen_notify+0x65/0x90 [sch_ets]
  qdisc_tree_reduce_backlog+0x74/0x110
  ets_qdisc_change+0x630/0xa40 [sch_ets]
  __tc_modify_qdisc.constprop.0+0x216/0x7f0
  tc_modify_qdisc+0x7c/0x120
  rtnetlink_rcv_msg+0x145/0x3f0
  netlink_rcv_skb+0x53/0x100
  netlink_unicast+0x245/0x390
  netlink_sendmsg+0x21b/0x470
  ____sys_sendmsg+0x39d/0x3d0
  ___sys_sendmsg+0x9a/0xe0
  __sys_sendmsg+0x7a/0xd0
  do_syscall_64+0x7d/0x160
  entry_SYSCALL_64_after_hwframe+0x76/0x7e
 RIP: 0033:0x7f2155114084
 Code: 89 02 b8 ff ff ff ff eb bb 66 2e 0f 1f 84 00 00 00 00 00 0f 1f 00 f3 0f 1e fa 80 3d 25 f0 0c 00 00 74 13 b8 2e 00 00 00 0f 05 <48> 3d 00 f0 ff ff 77 54 c3 0f 1f 00 48 83 ec 28 89 54 24 1c 48 89
 RSP: 002b:00007fff1fd7a988 EFLAGS: 00000202 ORIG_RAX: 000000000000002e
 RAX: ffffffffffffffda RBX: 0000560ec063e5e0 RCX: 00007f2155114084
 RDX: 0000000000000000 RSI: 00007fff1fd7a9f0 RDI: 0000000000000003
 RBP: 00007fff1fd7aa60 R08: 0000000000000010 R09: 000000000000003f
 R10: 0000560ee9b3a010 R11: 0000000000000202 R12: 00007fff1fd7aae0
 R13: 000000006891ccde R14: 0000560ec063e5e0 R15: 00007fff1fd7aad0
  </TASK>

[1] https://lore.kernel.org/netdev/e08c7f4a6882f260011909a868311c6e9b54f3e4.1639153474.git.dcaratti@redhat.com/
[2] https://lore.kernel.org/netdev/[email protected]/

Fixes 103406b38c60 ("net/sched: Always pass notifications when child class becomes empty")
Fixes c062f2a0b04d ("net/sched: sch_ets: don't remove idle classes from the round-robin list")
Fixes dcc68b4d8084 ("net: sch_ets: Add a new Qdisc")
---
 net/sched/sch_ets.c | 11 ++++++-----
 1 file changed, 6 insertions(+), 5 deletions(-)
net/sched/sch_ets.c +6 -5
index 5c9a72f1e1f2..e38879e59872 100644
@@ -664,6 +664,12 @@ static int ets_qdisc_change(struct Qdisc *sch, struct nlattr *opt,
 
 	sch_tree_lock(sch);
 
+	for (i = nbands; i < oldbands; i++) {
+		if (i >= q->nstrict && q->classes[i].qdisc->q.qlen)
+			list_del_init(&q->classes[i].alist);
+		qdisc_purge_queue(q->classes[i].qdisc);
+	}
+
 	WRITE_ONCE(q->nbands, nbands);
 	for (i = nstrict; i < q->nstrict; i++) {
 		if (q->classes[i].qdisc->q.qlen) {
@@ -671,11 +677,6 @@ static int ets_qdisc_change(struct Qdisc *sch, struct nlattr *opt,
 			q->classes[i].deficit = quanta[i];
 		}
 	}
-	for (i = q->nbands; i < oldbands; i++) {
-		if (i >= q->nstrict && q->classes[i].qdisc->q.qlen)
-			list_del_init(&q->classes[i].alist);
-		qdisc_purge_queue(q->classes[i].qdisc);
-	}
 	WRITE_ONCE(q->nstrict, nstrict);
 	memcpy(q->prio2band, priomap, sizeof(priomap));
 
Signature · 1 line
2.50.1