[dpdk-dev,v2] net/tap: fix eBPF handling of non-RSS flows

Message ID 1517841642-10429-2-git-send-email-ophirmu@mellanox.com
State Accepted, archived
Delegated to: Ferruh Yigit
Headers show

Checks

Context Check Description
ci/Intel-compilation success Compilation OK
ci/checkpatch success coding style OK

Commit Message

Ophir Munk Feb. 5, 2018, 2:40 p.m.
The eBPF classifier (section "cls_q" in tap_bpf_program.c) is tracing
marked packets in which skb->cb[1] contains an RSS queue number, and
redirects those packets to the matched queue.
It is expected that skb->cb[1] has been previously set with a valid RSS
queue number during an eBPF action (section "l3_l4" in tap_bpf_program.c).
However, for non-RSS flows, skb->cb[1] may contain a random unset value,
which could falsely be interpreted as a valid RSS queue.
To avoid this potential error, tap_bpf_program.c has been updated as
follows:
1. After calculating the RSS queue number, it is added a unique offset in
order to uniquely identify it as a valid RSS queue number.
2. After matching an RSS queue to a packet, skb->cb[1] is set to 0.

Fixes: cdc07e83bb24 ("net/tap: add eBPF program file")
Fixes: aabe70df73a3 ("net/tap: add eBPF bytes code")

Signed-off-by: Ophir Munk <ophirmu@mellanox.com>
---
 drivers/net/tap/tap_bpf_insns.h   | 23 +++++++++++++----------
 drivers/net/tap/tap_bpf_program.c |  9 ++++++---
 2 files changed, 19 insertions(+), 13 deletions(-)

Comments

Pascal Mazon Feb. 5, 2018, 2:51 p.m. | #1
The mitigation is good enough, random packets are more likely to have
cb[1] == 0 than something above 7cafe800.

Acked-by: Pascal Mazon <pascal.mazon@6wind.com>

On 05/02/2018 15:40, Ophir Munk wrote:
> The eBPF classifier (section "cls_q" in tap_bpf_program.c) is tracing
> marked packets in which skb->cb[1] contains an RSS queue number, and
> redirects those packets to the matched queue.
> It is expected that skb->cb[1] has been previously set with a valid RSS
> queue number during an eBPF action (section "l3_l4" in tap_bpf_program.c).
> However, for non-RSS flows, skb->cb[1] may contain a random unset value,
> which could falsely be interpreted as a valid RSS queue.
> To avoid this potential error, tap_bpf_program.c has been updated as
> follows:
> 1. After calculating the RSS queue number, it is added a unique offset in
> order to uniquely identify it as a valid RSS queue number.
> 2. After matching an RSS queue to a packet, skb->cb[1] is set to 0.
>
> Fixes: cdc07e83bb24 ("net/tap: add eBPF program file")
> Fixes: aabe70df73a3 ("net/tap: add eBPF bytes code")
>
> Signed-off-by: Ophir Munk <ophirmu@mellanox.com>
> ---
>  drivers/net/tap/tap_bpf_insns.h   | 23 +++++++++++++----------
>  drivers/net/tap/tap_bpf_program.c |  9 ++++++---
>  2 files changed, 19 insertions(+), 13 deletions(-)
>
> diff --git a/drivers/net/tap/tap_bpf_insns.h b/drivers/net/tap/tap_bpf_insns.h
> index c406f78..89873b6 100644
> --- a/drivers/net/tap/tap_bpf_insns.h
> +++ b/drivers/net/tap/tap_bpf_insns.h
> @@ -6,17 +6,20 @@
>  
>  /* bpf_insn array matching cls_q section. See tap_bpf_program.c file */
>  struct bpf_insn cls_q_insns[] = {
> -	{0x61,    1,    1,       52, 0x00000000},
> -	{0x18,    2,    0,        0, 0xdeadbeef},
> +	{0x61,    2,    1,       52, 0x00000000},
> +	{0x18,    3,    0,        0, 0xdeadbeef},
>  	{0x00,    0,    0,        0, 0x00000000},
> -	{0x63,   10,    2,       -4, 0x00000000},
> -	{0x61,    2,   10,       -4, 0x00000000},
> -	{0x07,    2,    0,        0, 0x00000001},
> -	{0x67,    2,    0,        0, 0x00000020},
> -	{0x77,    2,    0,        0, 0x00000020},
> -	{0xb7,    0,    0,        0, 0xffffffff},
> -	{0x1d,    1,    2,        1, 0x00000000},
> +	{0x63,   10,    3,       -4, 0x00000000},
>  	{0xb7,    0,    0,        0, 0x00000000},
> +	{0x61,    3,   10,       -4, 0x00000000},
> +	{0x07,    3,    0,        0, 0x7cafe800},
> +	{0x67,    3,    0,        0, 0x00000020},
> +	{0x77,    3,    0,        0, 0x00000020},
> +	{0x5d,    2,    3,        4, 0x00000000},
> +	{0xb7,    2,    0,        0, 0x00000000},
> +	{0x63,    1,    2,       52, 0x00000000},
> +	{0x18,    0,    0,        0, 0xffffffff},
> +	{0x00,    0,    0,        0, 0x00000000},
>  	{0x95,    0,    0,        0, 0x00000000},
>  };
>  
> @@ -1685,7 +1688,7 @@ struct bpf_insn l3_l4_hash_insns[] = {
>  	{0x4f,    3,    2,        0, 0x00000000},
>  	{0x67,    3,    0,        0, 0x00000010},
>  	{0x4f,    3,    1,        0, 0x00000000},
> -	{0x07,    3,    0,        0, 0x00000001},
> +	{0x07,    3,    0,        0, 0x7cafe800},
>  	{0x63,    5,    3,       52, 0x00000000},
>  	{0xb7,    7,    0,        0, 0x00000001},
>  	{0xbf,    0,    7,        0, 0x00000000},
> diff --git a/drivers/net/tap/tap_bpf_program.c b/drivers/net/tap/tap_bpf_program.c
> index 848c50b..8abb3b7 100644
> --- a/drivers/net/tap/tap_bpf_program.c
> +++ b/drivers/net/tap/tap_bpf_program.c
> @@ -28,10 +28,10 @@
>  		((b) & 0xff))
>  
>  /*
> - * The queue number is offset by 1, to distinguish packets that have
> - * gone through this rule (skb->cb[1] != 0) from others.
> + * The queue number is offset by a unique QUEUE_OFFSET, to distinguish
> + * packets that have gone through this rule (skb->cb[1] != 0) from others.
>   */
> -#define QUEUE_OFFSET		1
> +#define QUEUE_OFFSET		0x7cafe800
>  #define PIN_GLOBAL_NS		2
>  
>  #define KEY_IDX			0
> @@ -63,6 +63,9 @@ match_q(struct __sk_buff *skb)
>  
>  	if (queue != match_queue)
>  		return TC_ACT_OK;
> +
> +	/* queue match */
> +	skb->cb[1] = 0;
>  	return TC_ACT_UNSPEC;
>  }
>
Ferruh Yigit Feb. 5, 2018, 6:10 p.m. | #2
On 2/5/2018 2:51 PM, Pascal Mazon wrote:

> On 05/02/2018 15:40, Ophir Munk wrote:
>> The eBPF classifier (section "cls_q" in tap_bpf_program.c) is tracing
>> marked packets in which skb->cb[1] contains an RSS queue number, and
>> redirects those packets to the matched queue.
>> It is expected that skb->cb[1] has been previously set with a valid RSS
>> queue number during an eBPF action (section "l3_l4" in tap_bpf_program.c).
>> However, for non-RSS flows, skb->cb[1] may contain a random unset value,
>> which could falsely be interpreted as a valid RSS queue.
>> To avoid this potential error, tap_bpf_program.c has been updated as
>> follows:
>> 1. After calculating the RSS queue number, it is added a unique offset in
>> order to uniquely identify it as a valid RSS queue number.
>> 2. After matching an RSS queue to a packet, skb->cb[1] is set to 0.
>>
>> Fixes: cdc07e83bb24 ("net/tap: add eBPF program file")
>> Fixes: aabe70df73a3 ("net/tap: add eBPF bytes code")
>>
>> Signed-off-by: Ophir Munk <ophirmu@mellanox.com>

> The mitigation is good enough, random packets are more likely to have
> cb[1] == 0 than something above 7cafe800.
>
> Acked-by: Pascal Mazon <pascal.mazon@6wind.com>

Applied to dpdk-next-net/master, thanks.

Patch

diff --git a/drivers/net/tap/tap_bpf_insns.h b/drivers/net/tap/tap_bpf_insns.h
index c406f78..89873b6 100644
--- a/drivers/net/tap/tap_bpf_insns.h
+++ b/drivers/net/tap/tap_bpf_insns.h
@@ -6,17 +6,20 @@ 
 
 /* bpf_insn array matching cls_q section. See tap_bpf_program.c file */
 struct bpf_insn cls_q_insns[] = {
-	{0x61,    1,    1,       52, 0x00000000},
-	{0x18,    2,    0,        0, 0xdeadbeef},
+	{0x61,    2,    1,       52, 0x00000000},
+	{0x18,    3,    0,        0, 0xdeadbeef},
 	{0x00,    0,    0,        0, 0x00000000},
-	{0x63,   10,    2,       -4, 0x00000000},
-	{0x61,    2,   10,       -4, 0x00000000},
-	{0x07,    2,    0,        0, 0x00000001},
-	{0x67,    2,    0,        0, 0x00000020},
-	{0x77,    2,    0,        0, 0x00000020},
-	{0xb7,    0,    0,        0, 0xffffffff},
-	{0x1d,    1,    2,        1, 0x00000000},
+	{0x63,   10,    3,       -4, 0x00000000},
 	{0xb7,    0,    0,        0, 0x00000000},
+	{0x61,    3,   10,       -4, 0x00000000},
+	{0x07,    3,    0,        0, 0x7cafe800},
+	{0x67,    3,    0,        0, 0x00000020},
+	{0x77,    3,    0,        0, 0x00000020},
+	{0x5d,    2,    3,        4, 0x00000000},
+	{0xb7,    2,    0,        0, 0x00000000},
+	{0x63,    1,    2,       52, 0x00000000},
+	{0x18,    0,    0,        0, 0xffffffff},
+	{0x00,    0,    0,        0, 0x00000000},
 	{0x95,    0,    0,        0, 0x00000000},
 };
 
@@ -1685,7 +1688,7 @@  struct bpf_insn l3_l4_hash_insns[] = {
 	{0x4f,    3,    2,        0, 0x00000000},
 	{0x67,    3,    0,        0, 0x00000010},
 	{0x4f,    3,    1,        0, 0x00000000},
-	{0x07,    3,    0,        0, 0x00000001},
+	{0x07,    3,    0,        0, 0x7cafe800},
 	{0x63,    5,    3,       52, 0x00000000},
 	{0xb7,    7,    0,        0, 0x00000001},
 	{0xbf,    0,    7,        0, 0x00000000},
diff --git a/drivers/net/tap/tap_bpf_program.c b/drivers/net/tap/tap_bpf_program.c
index 848c50b..8abb3b7 100644
--- a/drivers/net/tap/tap_bpf_program.c
+++ b/drivers/net/tap/tap_bpf_program.c
@@ -28,10 +28,10 @@ 
 		((b) & 0xff))
 
 /*
- * The queue number is offset by 1, to distinguish packets that have
- * gone through this rule (skb->cb[1] != 0) from others.
+ * The queue number is offset by a unique QUEUE_OFFSET, to distinguish
+ * packets that have gone through this rule (skb->cb[1] != 0) from others.
  */
-#define QUEUE_OFFSET		1
+#define QUEUE_OFFSET		0x7cafe800
 #define PIN_GLOBAL_NS		2
 
 #define KEY_IDX			0
@@ -63,6 +63,9 @@  match_q(struct __sk_buff *skb)
 
 	if (queue != match_queue)
 		return TC_ACT_OK;
+
+	/* queue match */
+	skb->cb[1] = 0;
 	return TC_ACT_UNSPEC;
 }