* [PATCH net] net: sched: correct flower port blocking
@ 2020-02-13 20:28 Jason Baron
2020-02-13 22:50 ` Cong Wang
0 siblings, 1 reply; 7+ messages in thread
From: Jason Baron @ 2020-02-13 20:28 UTC (permalink / raw)
To: davem, jiri, xiyou.wangcong, jhs; +Cc: netdev, soukjin.bae, Eric Dumazet
tc flower rules that are based on src or dst port blocking are sometimes
ineffective due to uninitialized stack data. __skb_flow_dissect() extracts
ports from the skb for tc flower to match against. However, the port
dissection is not done when when the FLOW_DIS_IS_FRAGMENT bit is set in
key_control->flags. All callers of __skb_flow_dissect(), zero-out the
key_control field except for fl_classify() as used by the flower
classifier. Thus, the FLOW_DIS_IS_FRAGMENT may be set on entry to
__skb_flow_dissect(), since key_control is allocated on the stack
and may not be initialized.
Since key_basic and key_control are present for all flow keys, let's
make sure they are initialized.
Fixes: 62230715fd24 ("flow_dissector: do not dissect l4 ports for fragments")
Co-developed-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: Jason Baron <jbaron@akamai.com>
---
include/linux/skbuff.h | 8 ++++++++
net/sched/cls_flower.c | 1 +
2 files changed, 9 insertions(+)
diff --git a/include/linux/skbuff.h b/include/linux/skbuff.h
index ca8806b..f953bfa 100644
--- a/include/linux/skbuff.h
+++ b/include/linux/skbuff.h
@@ -1288,6 +1288,14 @@ bool __skb_flow_dissect(const struct net *net,
void *data, __be16 proto, int nhoff, int hlen,
unsigned int flags);
+static inline void
+skb_flow_dissect_init_key(struct flow_dissector_key_control *key_control,
+ struct flow_dissector_key_basic *key_basic)
+{
+ memset(key_control, 0, sizeof(*key_control));
+ memset(key_basic, 0, sizeof(*key_basic));
+}
+
static inline bool skb_flow_dissect(const struct sk_buff *skb,
struct flow_dissector *flow_dissector,
void *target_container, unsigned int flags)
diff --git a/net/sched/cls_flower.c b/net/sched/cls_flower.c
index f9c0d1e..b0a534b 100644
--- a/net/sched/cls_flower.c
+++ b/net/sched/cls_flower.c
@@ -305,6 +305,7 @@ static int fl_classify(struct sk_buff *skb, const struct tcf_proto *tp,
struct cls_fl_filter *f;
list_for_each_entry_rcu(mask, &head->masks, list) {
+ skb_flow_dissect_init_key(&skb_key.control, &skb_key.basic);
fl_clear_masked_range(&skb_key, mask);
skb_flow_dissect_meta(skb, &mask->dissector, &skb_key);
--
2.7.4
^ permalink raw reply related [flat|nested] 7+ messages in thread
* Re: [PATCH net] net: sched: correct flower port blocking
2020-02-13 20:28 [PATCH net] net: sched: correct flower port blocking Jason Baron
@ 2020-02-13 22:50 ` Cong Wang
2020-02-14 16:20 ` [PATCH v2 " Jason Baron
0 siblings, 1 reply; 7+ messages in thread
From: Cong Wang @ 2020-02-13 22:50 UTC (permalink / raw)
To: Jason Baron
Cc: David Miller, Jiri Pirko, Jamal Hadi Salim,
Linux Kernel Network Developers, soukjin.bae, Eric Dumazet
On Thu, Feb 13, 2020 at 12:31 PM Jason Baron <jbaron@akamai.com> wrote:
>
> tc flower rules that are based on src or dst port blocking are sometimes
> ineffective due to uninitialized stack data. __skb_flow_dissect() extracts
> ports from the skb for tc flower to match against. However, the port
> dissection is not done when when the FLOW_DIS_IS_FRAGMENT bit is set in
> key_control->flags. All callers of __skb_flow_dissect(), zero-out the
> key_control field except for fl_classify() as used by the flower
> classifier. Thus, the FLOW_DIS_IS_FRAGMENT may be set on entry to
> __skb_flow_dissect(), since key_control is allocated on the stack
> and may not be initialized.
>
> Since key_basic and key_control are present for all flow keys, let's
> make sure they are initialized.
>
> Fixes: 62230715fd24 ("flow_dissector: do not dissect l4 ports for fragments")
> Co-developed-by: Eric Dumazet <edumazet@google.com>
> Signed-off-by: Eric Dumazet <edumazet@google.com>
> Signed-off-by: Jason Baron <jbaron@akamai.com>
> ---
> include/linux/skbuff.h | 8 ++++++++
> net/sched/cls_flower.c | 1 +
> 2 files changed, 9 insertions(+)
>
> diff --git a/include/linux/skbuff.h b/include/linux/skbuff.h
> index ca8806b..f953bfa 100644
> --- a/include/linux/skbuff.h
> +++ b/include/linux/skbuff.h
> @@ -1288,6 +1288,14 @@ bool __skb_flow_dissect(const struct net *net,
> void *data, __be16 proto, int nhoff, int hlen,
> unsigned int flags);
>
> +static inline void
> +skb_flow_dissect_init_key(struct flow_dissector_key_control *key_control,
> + struct flow_dissector_key_basic *key_basic)
> +{
> + memset(key_control, 0, sizeof(*key_control));
> + memset(key_basic, 0, sizeof(*key_basic));
> +}
I think this function has nothing to do with skb, therefore it fits
better in include/net/flow_dissector.h? And remove skb prefix from
its name too.
Other than this:
Acked-by: Cong Wang <xiyou.wangcong@gmail.com>
Thanks.
^ permalink raw reply [flat|nested] 7+ messages in thread
* [PATCH v2 net] net: sched: correct flower port blocking
2020-02-13 22:50 ` Cong Wang
@ 2020-02-14 16:20 ` Jason Baron
2020-02-17 3:08 ` David Miller
0 siblings, 1 reply; 7+ messages in thread
From: Jason Baron @ 2020-02-14 16:20 UTC (permalink / raw)
To: davem, jiri, xiyou.wangcong, jhs; +Cc: netdev, soukjin.bae, Eric Dumazet
tc flower rules that are based on src or dst port blocking are sometimes
ineffective due to uninitialized stack data. __skb_flow_dissect() extracts
ports from the skb for tc flower to match against. However, the port
dissection is not done when when the FLOW_DIS_IS_FRAGMENT bit is set in
key_control->flags. All callers of __skb_flow_dissect(), zero-out the
key_control field except for fl_classify() as used by the flower
classifier. Thus, the FLOW_DIS_IS_FRAGMENT may be set on entry to
__skb_flow_dissect(), since key_control is allocated on the stack
and may not be initialized.
Since key_basic and key_control are present for all flow keys, let's
make sure they are initialized.
Fixes: 62230715fd24 ("flow_dissector: do not dissect l4 ports for fragments")
Co-developed-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: Eric Dumazet <edumazet@google.com>
Acked-by: Cong Wang <xiyou.wangcong@gmail.com>
Signed-off-by: Jason Baron <jbaron@akamai.com>
---
v2:
-move rename to flow_dissector_init_keys() amd move to
flow_dissector.h (Cong Wang)
include/net/flow_dissector.h | 8 ++++++++
net/sched/cls_flower.c | 1 +
2 files changed, 9 insertions(+)
diff --git a/include/net/flow_dissector.h b/include/net/flow_dissector.h
index d93017a..4bd4931 100644
--- a/include/net/flow_dissector.h
+++ b/include/net/flow_dissector.h
@@ -349,4 +349,12 @@ struct bpf_flow_dissector {
void *data_end;
};
+static inline void
+flow_dissector_init_keys(struct flow_dissector_key_control *key_control,
+ struct flow_dissector_key_basic *key_basic)
+{
+ memset(key_control, 0, sizeof(*key_control));
+ memset(key_basic, 0, sizeof(*key_basic));
+}
+
#endif
diff --git a/net/sched/cls_flower.c b/net/sched/cls_flower.c
index f9c0d1e..b783254 100644
--- a/net/sched/cls_flower.c
+++ b/net/sched/cls_flower.c
@@ -305,6 +305,7 @@ static int fl_classify(struct sk_buff *skb, const struct tcf_proto *tp,
struct cls_fl_filter *f;
list_for_each_entry_rcu(mask, &head->masks, list) {
+ flow_dissector_init_keys(&skb_key.control, &skb_key.basic);
fl_clear_masked_range(&skb_key, mask);
skb_flow_dissect_meta(skb, &mask->dissector, &skb_key);
--
2.7.4
^ permalink raw reply related [flat|nested] 7+ messages in thread
* Re: [PATCH v2 net] net: sched: correct flower port blocking
2020-02-14 16:20 ` [PATCH v2 " Jason Baron
@ 2020-02-17 3:08 ` David Miller
2020-02-17 3:18 ` David Miller
0 siblings, 1 reply; 7+ messages in thread
From: David Miller @ 2020-02-17 3:08 UTC (permalink / raw)
To: jbaron; +Cc: jiri, xiyou.wangcong, jhs, netdev, soukjin.bae, edumazet
From: Jason Baron <jbaron@akamai.com>
Date: Fri, 14 Feb 2020 11:20:24 -0500
> tc flower rules that are based on src or dst port blocking are sometimes
> ineffective due to uninitialized stack data. __skb_flow_dissect() extracts
> ports from the skb for tc flower to match against. However, the port
> dissection is not done when when the FLOW_DIS_IS_FRAGMENT bit is set in
> key_control->flags. All callers of __skb_flow_dissect(), zero-out the
> key_control field except for fl_classify() as used by the flower
> classifier. Thus, the FLOW_DIS_IS_FRAGMENT may be set on entry to
> __skb_flow_dissect(), since key_control is allocated on the stack
> and may not be initialized.
>
> Since key_basic and key_control are present for all flow keys, let's
> make sure they are initialized.
>
> Fixes: 62230715fd24 ("flow_dissector: do not dissect l4 ports for fragments")
> Co-developed-by: Eric Dumazet <edumazet@google.com>
> Signed-off-by: Eric Dumazet <edumazet@google.com>
> Acked-by: Cong Wang <xiyou.wangcong@gmail.com>
> Signed-off-by: Jason Baron <jbaron@akamai.com>
Applied and queued up for -stable.
^ permalink raw reply [flat|nested] 7+ messages in thread
* Re: [PATCH v2 net] net: sched: correct flower port blocking
2020-02-17 3:08 ` David Miller
@ 2020-02-17 3:18 ` David Miller
2020-02-17 20:38 ` [PATCH v3 " Jason Baron
0 siblings, 1 reply; 7+ messages in thread
From: David Miller @ 2020-02-17 3:18 UTC (permalink / raw)
To: jbaron; +Cc: jiri, xiyou.wangcong, jhs, netdev, soukjin.bae, edumazet
From: David Miller <davem@davemloft.net>
Date: Sun, 16 Feb 2020 19:08:12 -0800 (PST)
> From: Jason Baron <jbaron@akamai.com>
> Date: Fri, 14 Feb 2020 11:20:24 -0500
>
>> tc flower rules that are based on src or dst port blocking are sometimes
>> ineffective due to uninitialized stack data. __skb_flow_dissect() extracts
>> ports from the skb for tc flower to match against. However, the port
>> dissection is not done when when the FLOW_DIS_IS_FRAGMENT bit is set in
>> key_control->flags. All callers of __skb_flow_dissect(), zero-out the
>> key_control field except for fl_classify() as used by the flower
>> classifier. Thus, the FLOW_DIS_IS_FRAGMENT may be set on entry to
>> __skb_flow_dissect(), since key_control is allocated on the stack
>> and may not be initialized.
>>
>> Since key_basic and key_control are present for all flow keys, let's
>> make sure they are initialized.
>>
>> Fixes: 62230715fd24 ("flow_dissector: do not dissect l4 ports for fragments")
>> Co-developed-by: Eric Dumazet <edumazet@google.com>
>> Signed-off-by: Eric Dumazet <edumazet@google.com>
>> Acked-by: Cong Wang <xiyou.wangcong@gmail.com>
>> Signed-off-by: Jason Baron <jbaron@akamai.com>
>
> Applied and queued up for -stable.
Actually this doesn't even compile:
In file included from drivers/net/ethernet/mellanox/mlx5/core/en_tc.c:33:
./include/net/flow_dissector.h: In function ‘flow_dissector_init_keys’:
./include/net/flow_dissector.h:355:2: error: implicit declaration of function ‘memset’ [-Werror=implicit-function-declaration]
memset(key_control, 0, sizeof(*key_control));
^~~~~~
./include/net/flow_dissector.h:355:2: warning: incompatible implicit declaration of built-in function ‘memset’
./include/net/flow_dissector.h:355:2: note: include ‘<string.h>’ or provide a declaration of ‘memset’
./include/net/flow_dissector.h:9:1:
+#include <string.h>
./include/net/flow_dissector.h:355:2:
memset(key_control, 0, sizeof(*key_control));
^~~~~~
^ permalink raw reply [flat|nested] 7+ messages in thread
* [PATCH v3 net] net: sched: correct flower port blocking
2020-02-17 3:18 ` David Miller
@ 2020-02-17 20:38 ` Jason Baron
2020-02-18 5:33 ` David Miller
0 siblings, 1 reply; 7+ messages in thread
From: Jason Baron @ 2020-02-17 20:38 UTC (permalink / raw)
To: davem, jiri, xiyou.wangcong, jhs; +Cc: netdev, soukjin.bae, Eric Dumazet
tc flower rules that are based on src or dst port blocking are sometimes
ineffective due to uninitialized stack data. __skb_flow_dissect() extracts
ports from the skb for tc flower to match against. However, the port
dissection is not done when when the FLOW_DIS_IS_FRAGMENT bit is set in
key_control->flags. All callers of __skb_flow_dissect(), zero-out the
key_control field except for fl_classify() as used by the flower
classifier. Thus, the FLOW_DIS_IS_FRAGMENT may be set on entry to
__skb_flow_dissect(), since key_control is allocated on the stack
and may not be initialized.
Since key_basic and key_control are present for all flow keys, let's
make sure they are initialized.
Fixes: 62230715fd24 ("flow_dissector: do not dissect l4 ports for fragments")
Co-developed-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: Eric Dumazet <edumazet@google.com>
Acked-by: Cong Wang <xiyou.wangcong@gmail.com>
Signed-off-by: Jason Baron <jbaron@akamai.com>
---
v3:
-add include of string.h for memset() (David Miller)
v2:
-move rename to flow_dissector_init_keys() amd move to
flow_dissector.h (Cong Wang)
include/net/flow_dissector.h | 9 +++++++++
net/sched/cls_flower.c | 1 +
2 files changed, 10 insertions(+)
diff --git a/include/net/flow_dissector.h b/include/net/flow_dissector.h
index d93017a..e03827f 100644
--- a/include/net/flow_dissector.h
+++ b/include/net/flow_dissector.h
@@ -5,6 +5,7 @@
#include <linux/types.h>
#include <linux/in6.h>
#include <linux/siphash.h>
+#include <linux/string.h>
#include <uapi/linux/if_ether.h>
struct sk_buff;
@@ -349,4 +350,12 @@ struct bpf_flow_dissector {
void *data_end;
};
+static inline void
+flow_dissector_init_keys(struct flow_dissector_key_control *key_control,
+ struct flow_dissector_key_basic *key_basic)
+{
+ memset(key_control, 0, sizeof(*key_control));
+ memset(key_basic, 0, sizeof(*key_basic));
+}
+
#endif
diff --git a/net/sched/cls_flower.c b/net/sched/cls_flower.c
index f9c0d1e..b783254 100644
--- a/net/sched/cls_flower.c
+++ b/net/sched/cls_flower.c
@@ -305,6 +305,7 @@ static int fl_classify(struct sk_buff *skb, const struct tcf_proto *tp,
struct cls_fl_filter *f;
list_for_each_entry_rcu(mask, &head->masks, list) {
+ flow_dissector_init_keys(&skb_key.control, &skb_key.basic);
fl_clear_masked_range(&skb_key, mask);
skb_flow_dissect_meta(skb, &mask->dissector, &skb_key);
--
2.7.4
^ permalink raw reply related [flat|nested] 7+ messages in thread
* Re: [PATCH v3 net] net: sched: correct flower port blocking
2020-02-17 20:38 ` [PATCH v3 " Jason Baron
@ 2020-02-18 5:33 ` David Miller
0 siblings, 0 replies; 7+ messages in thread
From: David Miller @ 2020-02-18 5:33 UTC (permalink / raw)
To: jbaron; +Cc: jiri, xiyou.wangcong, jhs, netdev, soukjin.bae, edumazet
From: Jason Baron <jbaron@akamai.com>
Date: Mon, 17 Feb 2020 15:38:09 -0500
> tc flower rules that are based on src or dst port blocking are sometimes
> ineffective due to uninitialized stack data. __skb_flow_dissect() extracts
> ports from the skb for tc flower to match against. However, the port
> dissection is not done when when the FLOW_DIS_IS_FRAGMENT bit is set in
> key_control->flags. All callers of __skb_flow_dissect(), zero-out the
> key_control field except for fl_classify() as used by the flower
> classifier. Thus, the FLOW_DIS_IS_FRAGMENT may be set on entry to
> __skb_flow_dissect(), since key_control is allocated on the stack
> and may not be initialized.
>
> Since key_basic and key_control are present for all flow keys, let's
> make sure they are initialized.
>
> Fixes: 62230715fd24 ("flow_dissector: do not dissect l4 ports for fragments")
> Co-developed-by: Eric Dumazet <edumazet@google.com>
> Signed-off-by: Eric Dumazet <edumazet@google.com>
> Acked-by: Cong Wang <xiyou.wangcong@gmail.com>
> Signed-off-by: Jason Baron <jbaron@akamai.com>
Applied and queued up for -stable, thanks Jason.
^ permalink raw reply [flat|nested] 7+ messages in thread
end of thread, other threads:[~2020-02-18 5:33 UTC | newest]
Thread overview: 7+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2020-02-13 20:28 [PATCH net] net: sched: correct flower port blocking Jason Baron
2020-02-13 22:50 ` Cong Wang
2020-02-14 16:20 ` [PATCH v2 " Jason Baron
2020-02-17 3:08 ` David Miller
2020-02-17 3:18 ` David Miller
2020-02-17 20:38 ` [PATCH v3 " Jason Baron
2020-02-18 5:33 ` David Miller
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.