[patch net 0/2] net: sched: couple of fixes on ingress/clsact init error path

From: Jiri Pirko <redacted>

Jiri Pirko (2):
  net: sched: fix clsact init error path
  net: sched: fix static key imbalance in case of ingress/clsact_init
    error

 net/sched/cls_api.c     |  4 ++--
 net/sched/sch_ingress.c | 15 ++++++---------
 2 files changed, 8 insertions(+), 11 deletions(-)

-- 
2.9.5

From: Jiri Pirko <redacted>

Since in qdisc_create, the destroy op is called when init fails, we
don't do cleanup in init and leave it up to destroy.
This fixes use-after-free when trying to put already freed block.

Fixes: 6e40cf2d4dee ("net: sched: use extended variants of block_get/put in ingress and clsact qdiscs")
Signed-off-by: Jiri Pirko <redacted>
---
 net/sched/cls_api.c     | 4 ++--
 net/sched/sch_ingress.c | 6 +-----
 2 files changed, 3 insertions(+), 7 deletions(-)

diff --git a/net/sched/cls_api.c b/net/sched/cls_api.c
index f40256a..b91ea03 100644
--- a/net/sched/cls_api.c
+++ b/net/sched/cls_api.c

@@ -351,6 +351,8 @@ void tcf_block_put_ext(struct tcf_block *block, struct Qdisc *q,
 {
 	struct tcf_chain *chain;
 
+	if (!block)
+		return;
 	/* Hold a refcnt for all chains, except 0, so that they don't disappear
 	 * while we are iterating.
 	 */

@@ -377,8 +379,6 @@ void tcf_block_put(struct tcf_block *block)
 {
 	struct tcf_block_ext_info ei = {0, };
 
-	if (!block)
-		return;
 	tcf_block_put_ext(block, block->q, &ei);
 }
 

diff --git a/net/sched/sch_ingress.c b/net/sched/sch_ingress.c
index 5ecc38f..5e1cd2e 100644
--- a/net/sched/sch_ingress.c
+++ b/net/sched/sch_ingress.c

@@ -190,7 +190,7 @@ static int clsact_init(struct Qdisc *sch, struct nlattr *opt)
 
 	err = tcf_block_get_ext(&q->egress_block, sch, &q->egress_block_info);
 	if (err)
-		goto err_egress_block_get;
+		return err;
 
 	net_inc_ingress_queue();
 	net_inc_egress_queue();

@@ -198,10 +198,6 @@ static int clsact_init(struct Qdisc *sch, struct nlattr *opt)
 	sch->flags |= TCQ_F_CPUSTATS;
 
 	return 0;
-
-err_egress_block_get:
-	tcf_block_put_ext(q->ingress_block, sch, &q->ingress_block_info);
-	return err;
 }
 
 static void clsact_destroy(struct Qdisc *sch)

-- 
2.9.5

From: Jiri Pirko <redacted>

Move static key increments to the beginning of the init function
so they pair 1:1 with decrements in ingress/clsact_destroy,
which is called in case ingress/clsact_init fails.

Fixes: 6529eaba33f0 ("net: sched: introduce tcf block infractructure")
Signed-off-by: Jiri Pirko <redacted>
---
 net/sched/sch_ingress.c | 9 +++++----
 1 file changed, 5 insertions(+), 4 deletions(-)

diff --git a/net/sched/sch_ingress.c b/net/sched/sch_ingress.c
index 5e1cd2e..fc1286f 100644
--- a/net/sched/sch_ingress.c
+++ b/net/sched/sch_ingress.c

@@ -68,6 +68,8 @@ static int ingress_init(struct Qdisc *sch, struct nlattr *opt)
 	struct net_device *dev = qdisc_dev(sch);
 	int err;
 
+	net_inc_ingress_queue();
+
 	mini_qdisc_pair_init(&q->miniqp, sch, &dev->miniq_ingress);
 
 	q->block_info.binder_type = TCF_BLOCK_BINDER_TYPE_CLSACT_INGRESS;

@@ -78,7 +80,6 @@ static int ingress_init(struct Qdisc *sch, struct nlattr *opt)
 	if (err)
 		return err;
 
-	net_inc_ingress_queue();
 	sch->flags |= TCQ_F_CPUSTATS;
 
 	return 0;

@@ -172,6 +173,9 @@ static int clsact_init(struct Qdisc *sch, struct nlattr *opt)
 	struct net_device *dev = qdisc_dev(sch);
 	int err;
 
+	net_inc_ingress_queue();
+	net_inc_egress_queue();
+
 	mini_qdisc_pair_init(&q->miniqp_ingress, sch, &dev->miniq_ingress);
 
 	q->ingress_block_info.binder_type = TCF_BLOCK_BINDER_TYPE_CLSACT_INGRESS;

@@ -192,9 +196,6 @@ static int clsact_init(struct Qdisc *sch, struct nlattr *opt)
 	if (err)
 		return err;
 
-	net_inc_ingress_queue();
-	net_inc_egress_queue();
-
 	sch->flags |= TCQ_F_CPUSTATS;
 
 	return 0;

-- 
2.9.5

On Fri, Dec 15, 2017 at 3:40 AM, Jiri Pirko [off-list ref] wrote:

From: Jiri Pirko <redacted>

Since in qdisc_create, the destroy op is called when init fails, we
don't do cleanup in init and leave it up to destroy.
This fixes use-after-free when trying to put already freed block.

Fixes: 6e40cf2d4dee ("net: sched: use extended variants of block_get/put in ingress and clsact qdiscs")
Signed-off-by: Jiri Pirko <redacted>

Acked-by: Cong Wang <redacted>

On Fri, Dec 15, 2017 at 3:40 AM, Jiri Pirko [off-list ref] wrote:

From: Jiri Pirko <redacted>

Move static key increments to the beginning of the init function
so they pair 1:1 with decrements in ingress/clsact_destroy,
which is called in case ingress/clsact_init fails.

Fixes: 6529eaba33f0 ("net: sched: introduce tcf block infractructure")
Signed-off-by: Jiri Pirko <redacted>

Acked-by: Cong Wang <redacted>

From: Jiri Pirko <jiri@resnulli.us>
Date: Fri, 15 Dec 2017 12:40:12 +0100

From: Jiri Pirko <redacted>

Since in qdisc_create, the destroy op is called when init fails, we
don't do cleanup in init and leave it up to destroy.
This fixes use-after-free when trying to put already freed block.

Fixes: 6e40cf2d4dee ("net: sched: use extended variants of block_get/put in ingress and clsact qdiscs")
Signed-off-by: Jiri Pirko <redacted>

Applied.

From: Jiri Pirko <jiri@resnulli.us>
Date: Fri, 15 Dec 2017 12:40:13 +0100

From: Jiri Pirko <redacted>

Move static key increments to the beginning of the init function
so they pair 1:1 with decrements in ingress/clsact_destroy,
which is called in case ingress/clsact_init fails.

Fixes: 6529eaba33f0 ("net: sched: introduce tcf block infractructure")
Signed-off-by: Jiri Pirko <redacted>

Applied and queued up for -stable.