[2/2,v2] ieee802154: assembly of 6LoWPAN fragments improvement
diff mbox

Message ID 20180717152620.GB22664@Rafael-Mac.intra.sownet.nl
State Under Review
Headers show

Commit Message

Rafael Vuijk July 17, 2018, 3:26 p.m. UTC
6LoWPAN reassembly fragment overlap checks.

Signed-off-by: Rafael Vuijk <r.vuijk@sownet.nl>

--
To unsubscribe from this list: send the line "unsubscribe linux-wpan" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Comments

Alexander Aring July 18, 2018, 2:30 p.m. UTC | #1
Hi,

On Tue, Jul 17, 2018 at 05:26:20PM +0200, Rafael Vuijk wrote:
> 6LoWPAN reassembly fragment overlap checks.
> 
> Signed-off-by: Rafael Vuijk <r.vuijk@sownet.nl>
> --- ./net/ieee802154/6lowpan/reassembly.c	2018-02-20 11:10:06.000000000 +0100
> +++ ./net/ieee802154/6lowpan/reassembly.c	2018-02-21 09:13:29.000000000 +0100
> @@ -179,6 +170,13 @@ static int lowpan_frag_queue(struct lowp
>  	}
>  
>  found:
> +	/* Current fragment overlaps with previous fragment? */
> +	if (prev && (lowpan_802154_cb(prev)->d_offset << 3) + prev->len > offset)
> +		goto err;
> +	/* Current fragment overlaps with next fragment? */
> +	if (next && offset + skb->len > lowpan_802154_cb(next)->d_offset << 3)
> +		goto err;
> +

I have some thought of mine when seeing this code. The function is
separated into two phases:

phase 0:
 - finding the missing piece of fragment skb in the right order
 - if found goto found:

phase 1:
 - everything after found to reassemble everything if we have everything
   together.


This patch moves parts which are belongs to "phase 0" to "phase 1". I
think the general idea in this algorithmn is to simple don't make checks
on invalid things at all. If "broken" fragments are inside the "fragment
bucket" then simple leave it there. There exists a garbage collector,
controlled by some expire parameter to drop them then (60 seconds by
default).

Important is that the code should never run lowpan_frag_reasm() when
something is not right in the fragments.

- Alex
--
To unsubscribe from this list: send the line "unsubscribe linux-wpan" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Stefan Schmidt July 24, 2018, 1:04 p.m. UTC | #2
Hello.

On 17.07.2018 17:26, Rafael Vuijk wrote:
> 6LoWPAN reassembly fragment overlap checks.

A bit more explanation would not hurd here. :-)

> Signed-off-by: Rafael Vuijk <r.vuijk@sownet.nl>
> --- ./net/ieee802154/6lowpan/reassembly.c	2018-02-20 11:10:06.000000000 +0100
> +++ ./net/ieee802154/6lowpan/reassembly.c	2018-02-21 09:13:29.000000000 +0100
> @@ -179,6 +170,13 @@ static int lowpan_frag_queue(struct lowp
>  	}
>  
>  found:
> +	/* Current fragment overlaps with previous fragment? */
> +	if (prev && (lowpan_802154_cb(prev)->d_offset << 3) + prev->len > offset)> +		goto err;
> +	/* Current fragment overlaps with next fragment? */
> +	if (next && offset + skb->len > lowpan_802154_cb(next)->d_offset << 3)

To be honest, I am not sure I would get the operator priorities right in
my head for this line without looking it up. :-) Could we get some () to
help with the operator priorities? Even if that means we need to break
this line into two.

> +		goto err;
> +
>  	/* Insert this fragment in the chain of fragments. */
>  	skb->next = next;
>  	if (!next)

regards
Stefan Schmidt
--
To unsubscribe from this list: send the line "unsubscribe linux-wpan" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Stefan Schmidt July 24, 2018, 1:08 p.m. UTC | #3
Hello.

On 18.07.2018 16:30, Alexander Aring wrote:
> Hi,
> 
> On Tue, Jul 17, 2018 at 05:26:20PM +0200, Rafael Vuijk wrote:
>> 6LoWPAN reassembly fragment overlap checks.
>>
>> Signed-off-by: Rafael Vuijk <r.vuijk@sownet.nl>
>> --- ./net/ieee802154/6lowpan/reassembly.c	2018-02-20 11:10:06.000000000 +0100
>> +++ ./net/ieee802154/6lowpan/reassembly.c	2018-02-21 09:13:29.000000000 +0100
>> @@ -179,6 +170,13 @@ static int lowpan_frag_queue(struct lowp
>>  	}
>>  
>>  found:
>> +	/* Current fragment overlaps with previous fragment? */
>> +	if (prev && (lowpan_802154_cb(prev)->d_offset << 3) + prev->len > offset)
>> +		goto err;
>> +	/* Current fragment overlaps with next fragment? */
>> +	if (next && offset + skb->len > lowpan_802154_cb(next)->d_offset << 3)
>> +		goto err;
>> +
> 
> I have some thought of mine when seeing this code. The function is
> separated into two phases:
> 
> phase 0:
>  - finding the missing piece of fragment skb in the right order
>  - if found goto found:
> 
> phase 1:
>  - everything after found to reassemble everything if we have everything
>    together.
> 
> 
> This patch moves parts which are belongs to "phase 0" to "phase 1". I
> think the general idea in this algorithmn is to simple don't make checks
> on invalid things at all. If "broken" fragments are inside the "fragment
> bucket" then simple leave it there. There exists a garbage collector,
> controlled by some expire parameter to drop them then (60 seconds by
> default).

I think Rafael looked around for the best place of these checks. The
algorithm is buggy in this regard as they have seen in their testing and
packet traces. If you have a better place for this check I am pretty
sure they would be happy to test it and come back to you if it fixes
their issue.

If we find no better place I will just use this patch and move forward
to have at least the practical issue fixed.

regards
Stefan Schmidt
--
To unsubscribe from this list: send the line "unsubscribe linux-wpan" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Patch
diff mbox

--- ./net/ieee802154/6lowpan/reassembly.c	2018-02-20 11:10:06.000000000 +0100
+++ ./net/ieee802154/6lowpan/reassembly.c	2018-02-21 09:13:29.000000000 +0100
@@ -179,6 +170,13 @@  static int lowpan_frag_queue(struct lowp
 	}
 
 found:
+	/* Current fragment overlaps with previous fragment? */
+	if (prev && (lowpan_802154_cb(prev)->d_offset << 3) + prev->len > offset)
+		goto err;
+	/* Current fragment overlaps with next fragment? */
+	if (next && offset + skb->len > lowpan_802154_cb(next)->d_offset << 3)
+		goto err;
+
 	/* Insert this fragment in the chain of fragments. */
 	skb->next = next;
 	if (!next)