Hi Lin & Chun-Yeow,

Thank you very much for your help and time. I'll install the patch and let
you know if it solved my problem.

Regards,
Siva.


>
> Message: 1
> Date: Mon, 11 Mar 2013 13:46:41 +0000
> From: Chaoxing Lin <[email protected]>
> To: "[email protected]" <[email protected]>
> Subject: RE: mpath dump showing a path with next_hop address
>         00:00:00:00:00:00
> Message-ID:
>         <[email protected]>
> Content-Type: text/plain; charset="iso-8859-1"
>
> I guess your mp is added to bridge since you mentioned "Ethernet" MAC.
> I saw the same problem and solved it long ago(meaning, I can't remember
> exact scenario to reproduce it).
>
> Below is the note when I fixed it. I hope it will help.
> I believe the mesh_hwmp.c maintainer would know what's going on after
> reading it.
>
>
> "The problem is that code always looks up DA from mpath table.
> Only when there is no entry in mpath table, it looks up DA from mpp table.
> If neither table has an entry for DA, code adds an empty entry with DA in
> mpath
> table and starts path request with the hope that this entry will be filled
> when
> path reply packet comes back. In this case(DA is not a MP, it's Ethernet
> MAC), there won't be path reply, which
> results in a dead entry in mpath table.
>
> This dead entry prevent future communication because code won't look up DA
> from
> mpp table any more (there is an entry in mpath table, although it's dead).
>
> The fix for the bug part is to delete dead entry in mpath table when DA is
> learned in mpp table. Now, valid entry can be found in mpp table and there
> is no
> need to do path request(meaning, no further dead entry later on)"
>
>
>
> From: [email protected] [mailto:
> [email protected]] On Behalf Of Sivateja Patibandla
> Sent: Saturday, March 09, 2013 3:19 PM
> To: [email protected]
> Subject: mpath dump showing a path with next_hop address 00:00:00:00:00:00
>
> Hi,
>
> Sometimes my mpath dump shows paths with next hop address
> as?00:00:00:00:00:00. And destination address as Ethernet's MAC address of
> the destination.?
> For example: Let's consider node A is trying to reach node B with wireless
> MAC: 00:15:6d:ec:33:Fa and Ethernet MAC: 00:15:6d:ed:33:Fa. And the mpath
> dump on node A is as follows.
>
> Dest Addr ? ? ? ? ? ? ? ? ? ? ?Next Hop ? ? ? ? ?
> 00:15:6d:ec:33:Fa ? ? ? ? ? 00:15:6d:ec:33:fa
> 00:15:6d:ed:33:Fa ? ? ? ? ? 00:00:00:00:00:00
>
> When the mpath dump show this kind of paths. B won't respond to ping from
> A. If I delete the second mapth by iw wlan0 mpath del 00:15:6d:ed:33:Fa
> then it starts perfectly working fine.
>
> Any comments on what's going on?
>
> Regards,
> Siva. ?
>
>
>
> ------------------------------
>
> Message: 2
> Date: Mon, 11 Mar 2013 23:45:59 +0800
> From: Yeoh Chun-Yeow <[email protected]>
> To: [email protected]
> Subject: Re: mpath dump showing a path with next_hop address
>         00:00:00:00:00:00
> Message-ID:
>         <
> caefj984q+qd125o+phbody2uqae6b+w0vcngvdf8gn5-aiq...@mail.gmail.com>
> Content-Type: text/plain; charset=ISO-8859-1
>
> Hi, Chaoxing
>
> Yes. You are correct. As pointed out, the following patch is intended
> to solve this:
>
> http://lists.open80211s.org/pipermail/devel/2013-February/004149.html
> http://www.spinics.net/lists/linux-wireless/msg103642.html
>
> ---
> Chun-Yeow
>
> On Mon, Mar 11, 2013 at 9:46 PM, Chaoxing Lin
> <[email protected]> wrote:
> > I guess your mp is added to bridge since you mentioned "Ethernet" MAC.
> > I saw the same problem and solved it long ago(meaning, I can't remember
> exact scenario to reproduce it).
> >
> > Below is the note when I fixed it. I hope it will help.
> > I believe the mesh_hwmp.c maintainer would know what's going on after
> reading it.
> >
> >
> > "The problem is that code always looks up DA from mpath table.
> > Only when there is no entry in mpath table, it looks up DA from mpp
> table.
> > If neither table has an entry for DA, code adds an empty entry with DA
> in mpath
> > table and starts path request with the hope that this entry will be
> filled when
> > path reply packet comes back. In this case(DA is not a MP, it's Ethernet
> MAC), there won't be path reply, which
> > results in a dead entry in mpath table.
> >
> > This dead entry prevent future communication because code won't look up
> DA from
> > mpp table any more (there is an entry in mpath table, although it's
> dead).
> >
> > The fix for the bug part is to delete dead entry in mpath table when DA
> is
> > learned in mpp table. Now, valid entry can be found in mpp table and
> there is no
> > need to do path request(meaning, no further dead entry later on)"
> >
> >
>
>
> ------------------------------
>
> _______________________________________________
> Devel mailing list
> [email protected]
> http://lists.open80211s.org/cgi-bin/mailman/listinfo/devel
>
>
> End of Devel Digest, Vol 13, Issue 13
> *************************************
>
_______________________________________________
Devel mailing list
[email protected]
http://lists.open80211s.org/cgi-bin/mailman/listinfo/devel

Reply via email to