[PATCH net-next-2.6] sfq: fix sfq stats handling

[PATCH net-next-2.6] sfq: fix sfq stats handling

[Eric Dumazet]

David, since this one fixes a bug, could you apply it before the pending
SFQ patch (sch_sfq: allow big packets and be fair), I'll respin this one
if necessary.

Thanks

[PATCH net-next-2.6] sfq: fix sfq class stats handling

sfq_walk() runs without qdisc lock. By the time it selects a non empty
hash slot and sfq_dump_class_stats() is run (with lock held), slot might
have been freed : We then access q->slots[SFQ_EMPTY_SLOT], out of
bounds, and crash in slot_queue_walk()

On previous kernels, bug is here but out of bounds qs[SFQ_DEPTH] and
allot[SFQ_DEPTH] are located in struct sfq_sched_data, so no illegal
memory access happens, only possibly wrong data reported to user.

Also, slot_dequeue_tail() should make sure slot skb chain is correctly
terminated, or sfq_dump_class_stats() can access freed skbs.

Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Cc: Jarek Poplawski <jarkao2@gmail.com>
---
 net/sched/sch_sfq.c |   16 +++++++++++-----
 1 files changed, 11 insertions(+), 5 deletions(-)

diff --git a/net/sched/sch_sfq.c b/net/sched/sch_sfq.c
index 13322e8..6a2f88f 100644
--- a/net/sched/sch_sfq.c
+++ b/net/sched/sch_sfq.c
@@ -281,6 +281,7 @@ static inline struct sk_buff *slot_dequeue_tail(struct sfq_slot *slot)
 	struct sk_buff *skb = slot->skblist_prev;
 
 	slot->skblist_prev = skb->prev;
+	skb->prev->next = (struct sk_buff *)slot;
 	skb->next = skb->prev = NULL;
 	return skb;
 }
@@ -608,14 +609,19 @@ static int sfq_dump_class_stats(struct Qdisc *sch, unsigned long cl,
 				struct gnet_dump *d)
 {
 	struct sfq_sched_data *q = qdisc_priv(sch);
-	const struct sfq_slot *slot = &q->slots[q->ht[cl - 1]];
-	struct gnet_stats_queue qs = { .qlen = slot->qlen };
-	struct tc_sfq_xstats xstats = { .allot = slot->allot };
+	sfq_index idx = q->ht[cl - 1];
+	struct gnet_stats_queue qs = { 0 };
+	struct tc_sfq_xstats xstats = { 0 };
 	struct sk_buff *skb;
 
-	slot_queue_walk(slot, skb)
-		qs.backlog += qdisc_pkt_len(skb);
+	if (idx != SFQ_EMPTY_SLOT) {
+		const struct sfq_slot *slot = &q->slots[idx];
 
+		xstats.allot = slot->allot;
+		qs.qlen = slot->qlen;
+		slot_queue_walk(slot, skb)
+			qs.backlog += qdisc_pkt_len(skb);
+	}
 	if (gnet_stats_copy_queue(d, &qs) < 0)
 		return -1;
 	return gnet_stats_copy_app(d, &xstats, sizeof(xstats));

Re: [PATCH net-next-2.6] sfq: fix sfq stats handling

[Jarek Poplawski]

On Wed, Dec 22, 2010 at 07:14:59AM +0100, Eric Dumazet wrote:
> David, since this one fixes a bug, could you apply it before the pending
> SFQ patch (sch_sfq: allow big packets and be fair), I'll respin this one
> if necessary.
> 
> Thanks
> 
> [PATCH net-next-2.6] sfq: fix sfq class stats handling
> 
> sfq_walk() runs without qdisc lock. By the time it selects a non empty
> hash slot and sfq_dump_class_stats() is run (with lock held), slot might
> have been freed : We then access q->slots[SFQ_EMPTY_SLOT], out of
> bounds, and crash in slot_queue_walk()
> 
> On previous kernels, bug is here but out of bounds qs[SFQ_DEPTH] and
> allot[SFQ_DEPTH] are located in struct sfq_sched_data, so no illegal
> memory access happens, only possibly wrong data reported to user.

Yes, nice catch!

> Also, slot_dequeue_tail() should make sure slot skb chain is correctly
> terminated, or sfq_dump_class_stats() can access freed skbs.

...and a good hint for code reusing ;-)

Thanks,
Jarek P.

> 
> Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
> Cc: Jarek Poplawski <jarkao2@gmail.com>
> ---
>  net/sched/sch_sfq.c |   16 +++++++++++-----
>  1 files changed, 11 insertions(+), 5 deletions(-)
> 
> diff --git a/net/sched/sch_sfq.c b/net/sched/sch_sfq.c
> index 13322e8..6a2f88f 100644
> --- a/net/sched/sch_sfq.c
> +++ b/net/sched/sch_sfq.c
> @@ -281,6 +281,7 @@ static inline struct sk_buff *slot_dequeue_tail(struct sfq_slot *slot)
>  	struct sk_buff *skb = slot->skblist_prev;
>  
>  	slot->skblist_prev = skb->prev;
> +	skb->prev->next = (struct sk_buff *)slot;
>  	skb->next = skb->prev = NULL;
>  	return skb;
>  }
> @@ -608,14 +609,19 @@ static int sfq_dump_class_stats(struct Qdisc *sch, unsigned long cl,
>  				struct gnet_dump *d)
>  {
>  	struct sfq_sched_data *q = qdisc_priv(sch);
> -	const struct sfq_slot *slot = &q->slots[q->ht[cl - 1]];
> -	struct gnet_stats_queue qs = { .qlen = slot->qlen };
> -	struct tc_sfq_xstats xstats = { .allot = slot->allot };
> +	sfq_index idx = q->ht[cl - 1];
> +	struct gnet_stats_queue qs = { 0 };
> +	struct tc_sfq_xstats xstats = { 0 };
>  	struct sk_buff *skb;
>  
> -	slot_queue_walk(slot, skb)
> -		qs.backlog += qdisc_pkt_len(skb);
> +	if (idx != SFQ_EMPTY_SLOT) {
> +		const struct sfq_slot *slot = &q->slots[idx];
>  
> +		xstats.allot = slot->allot;
> +		qs.qlen = slot->qlen;
> +		slot_queue_walk(slot, skb)
> +			qs.backlog += qdisc_pkt_len(skb);
> +	}
>  	if (gnet_stats_copy_queue(d, &qs) < 0)
>  		return -1;
>  	return gnet_stats_copy_app(d, &xstats, sizeof(xstats));
> 
> 

[PATCH net-next-2.6] sfq: fix slot_dequeue_head()

[Eric Dumazet]

Le mercredi 22 décembre 2010 à 07:32 +0000, Jarek Poplawski a écrit :
> > Also, slot_dequeue_tail() should make sure slot skb chain is correctly
> > terminated, or sfq_dump_class_stats() can access freed skbs.
> 
> ...and a good hint for code reusing ;-)

Yes, and of course same fix is needed in slot_dequeue_head(), as further
testing on my side made it pretty clear.

I was adding possibility to have more packets queued in SFQ (more
packets than max number of flows) and got unexpected crashes.

Reverting to net-next-2.6, I still got crashes. Oops.

[PATCH net-next-2.6] sfq: fix slot_dequeue_head()

slot_dequeue_head() should make sure slot skb chain is correct in both
ways, or we can crash if all possible flows are in use.

Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Cc: Jarek Poplawski <jarkao2@gmail.com>
---
 net/sched/sch_sfq.c |    1 +
 1 files changed, 1 insertion(+)

diff --git a/net/sched/sch_sfq.c b/net/sched/sch_sfq.c
index 6a2f88f..3977e56 100644
--- a/net/sched/sch_sfq.c
+++ b/net/sched/sch_sfq.c
@@ -292,6 +292,7 @@ static inline struct sk_buff *slot_dequeue_head(struct sfq_slot *slot)
 	struct sk_buff *skb = slot->skblist_next;
 
 	slot->skblist_next = skb->next;
+	skb->next->prev = (struct sk_buff *)slot;
 	skb->next = skb->prev = NULL;
 	return skb;
 }

Re: [PATCH net-next-2.6] sfq: fix slot_dequeue_head()

[Jarek Poplawski]

On Thu, Dec 30, 2010 at 04:02:48PM +0100, Eric Dumazet wrote:
> Le mercredi 22 décembre 2010 ?? 07:32 +0000, Jarek Poplawski a écrit :
> > > Also, slot_dequeue_tail() should make sure slot skb chain is correctly
> > > terminated, or sfq_dump_class_stats() can access freed skbs.
> > 
> > ...and a good hint for code reusing ;-)
> 
> Yes, and of course same fix is needed in slot_dequeue_head(), as further
> testing on my side made it pretty clear.
> 
> I was adding possibility to have more packets queued in SFQ (more
> packets than max number of flows) and got unexpected crashes.
> 
> Reverting to net-next-2.6, I still got crashes. Oops.
> 
> [PATCH net-next-2.6] sfq: fix slot_dequeue_head()
> 
> slot_dequeue_head() should make sure slot skb chain is correct in both
> ways, or we can crash if all possible flows are in use.

Nice scenario ;-) Of course, it's easy to guess I looked for something
like this after your previous fix and missed that :-| Btw, it looks
like slot_queue_init() could go back to sfq_init() now.

Jarek P.

> 
> Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
> Cc: Jarek Poplawski <jarkao2@gmail.com>
> ---
>  net/sched/sch_sfq.c |    1 +
>  1 files changed, 1 insertion(+)
> 
> diff --git a/net/sched/sch_sfq.c b/net/sched/sch_sfq.c
> index 6a2f88f..3977e56 100644
> --- a/net/sched/sch_sfq.c
> +++ b/net/sched/sch_sfq.c
> @@ -292,6 +292,7 @@ static inline struct sk_buff *slot_dequeue_head(struct sfq_slot *slot)
>  	struct sk_buff *skb = slot->skblist_next;
>  
>  	slot->skblist_next = skb->next;
> +	skb->next->prev = (struct sk_buff *)slot;
>  	skb->next = skb->prev = NULL;
>  	return skb;
>  }
> 
> 

Re: [PATCH net-next-2.6] sfq: fix slot_dequeue_head()

[Eric Dumazet]

Le jeudi 30 décembre 2010 à 18:49 +0100, Jarek Poplawski a écrit :

> Nice scenario ;-) Of course, it's easy to guess I looked for something
> like this after your previous fix and missed that :-| Btw, it looks
> like slot_queue_init() could go back to sfq_init() now.
> 

Indeed, I tested following combined patch and all is good.

Thanks Jarek !

[PATCH v2 net-next-2.6] sfq: fix slot_dequeue_head()

slot_dequeue_head() should make sure slot skb chain is correct in both
ways, or we can crash if all possible flows are in use.

Jarek pointed out slot_queue_init() can now be done in sfq_init() once,
instead each time a flow is setup.

Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Cc: Jarek Poplawski <jarkao2@gmail.com>
---
 net/sched/sch_sfq.c |    6 ++++--
 1 files changed, 4 insertions(+), 2 deletions(-)

diff --git a/net/sched/sch_sfq.c b/net/sched/sch_sfq.c
index 6a2f88f..02fa1dd 100644
--- a/net/sched/sch_sfq.c
+++ b/net/sched/sch_sfq.c
@@ -292,6 +292,7 @@ static inline struct sk_buff *slot_dequeue_head(struct sfq_slot *slot)
 	struct sk_buff *skb = slot->skblist_next;
 
 	slot->skblist_next = skb->next;
+	skb->next->prev = (struct sk_buff *)slot;
 	skb->next = skb->prev = NULL;
 	return skb;
 }
@@ -375,7 +376,6 @@ sfq_enqueue(struct sk_buff *skb, struct Qdisc *sch)
 		q->ht[hash] = x;
 		slot = &q->slots[x];
 		slot->hash = hash;
-		slot_queue_init(slot);
 	}
 
 	/* If selected queue has length q->limit, do simple tail drop,
@@ -533,8 +533,10 @@ static int sfq_init(struct Qdisc *sch, struct nlattr *opt)
 			return err;
 	}
 
-	for (i = 0; i < SFQ_SLOTS; i++)
+	for (i = 0; i < SFQ_SLOTS; i++) {
+		slot_queue_init(&q->slots[i]);
 		sfq_link(q, i);
+	}
 	return 0;
 }
 

Re: [PATCH net-next-2.6] sfq: fix slot_dequeue_head()

[David Miller]

From: Eric Dumazet <eric.dumazet@gmail.com>
Date: Thu, 30 Dec 2010 22:31:09 +0100

> [PATCH v2 net-next-2.6] sfq: fix slot_dequeue_head()
> 
> slot_dequeue_head() should make sure slot skb chain is correct in both
> ways, or we can crash if all possible flows are in use.
> 
> Jarek pointed out slot_queue_init() can now be done in sfq_init() once,
> instead each time a flow is setup.
> 
> Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>

Also applied, thanks.

Re: [PATCH net-next-2.6] sfq: fix sfq stats handling

[David Miller]

From: Eric Dumazet <eric.dumazet@gmail.com>
Date: Wed, 22 Dec 2010 07:14:59 +0100

> David, since this one fixes a bug, could you apply it before the pending
> SFQ patch (sch_sfq: allow big packets and be fair), I'll respin this one
> if necessary.
> 
> Thanks
> 
> [PATCH net-next-2.6] sfq: fix sfq class stats handling
> 
> sfq_walk() runs without qdisc lock. By the time it selects a non empty
> hash slot and sfq_dump_class_stats() is run (with lock held), slot might
> have been freed : We then access q->slots[SFQ_EMPTY_SLOT], out of
> bounds, and crash in slot_queue_walk()
> 
> On previous kernels, bug is here but out of bounds qs[SFQ_DEPTH] and
> allot[SFQ_DEPTH] are located in struct sfq_sched_data, so no illegal
> memory access happens, only possibly wrong data reported to user.
> 
> Also, slot_dequeue_tail() should make sure slot skb chain is correctly
> terminated, or sfq_dump_class_stats() can access freed skbs.
> 
> Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
> Cc: Jarek Poplawski <jarkao2@gmail.com>

Applied, thanks.