Sun 11 Mar 2007 07:19:55 PM UTC, comment #8:
Just commit the release. I use it without any problems, and I just have change the way I initialize the mac filter to be independant from my target...
I can try to integrate some other modifications from Mace which can be interest some others developers, if they need them...
Please, send me any comments...
Hoping that will help Thomas and others...
|
Sun 04 Mar 2007 12:22:21 AM UTC, comment #7:
Thanks Frederic. I'll be doing extensive testing with IGMP over the course of the next 3 months, and will start with these patches as a baseline.
|
Mon 26 Feb 2007 06:24:51 PM UTC, comment #6:
I join to this post two files. First, a patch file produce with :
diff -urN lwip_cvs lwip_mine > IGMP_AND_RCVTIMEO.patch
I'm not an diff expert, so, I hope it's good. If not, I provide my full own copy in the Lwip.zip file.
I try to clean the code, but I let the SO_RCVTIMEO for UDP sockets in the code, and some others things (some patchs already included).
The most modifications are about the sys_mbox_fetch "timeout" parameter to allow to implement the SO_RCVTIMEO (it's a very simple modification, but very useful for DNS or SNTP protocol).
If you REALLY want, I can remove it and try to provide you two separates patchs, but adding it is not intrusive (on my target).
Email me if you want any comments about the modifications.
(file #12047, file #12048)
|
Mon 26 Feb 2007 05:02:40 PM UTC, comment #5:
This is not in icmp.h, but in igmp.h.
I don't know if the Mace version is fix, but in searching in forum archive, it doesn't seems.
My version is based on Mace for :
sockets.c/sockets.h
api_lib.c/api.h
api_msg.c/api_msg.h
but in lower layer, I use my own version based on Steve Reynolds for :
igmp.c/igmp.h
ip.c (main core modifications, tagged by LWIP_IGMP #ifdef/#endif)
I will try to provide you a "clean" patch file...
|
Mon 26 Feb 2007 03:57:32 PM UTC, comment #4:
Thanks Frederic. That should provide an excellent base for someone to come up with a working solution. You mentioned that some fixes are needed in the icmp.c code - are these already applied in the patch you have supplied, or are there extra ones that you know about?
Any volunteers to take this further and get the code into CVS?
|
Mon 26 Feb 2007 03:51:01 PM UTC, comment #3:
Hi Kieran,
This was not merged. Some fix are need in igmp.c code.
But you can find a first patch in this thread : "[lwip-users] Patch for IGMP V1/V2 and raw multicast support" (near 22/03/2006) started by Mace Gael. It was based on Steve Reynolds code.
I also got a such version (also based on Steve Reynolds code, and on Mace Gael modifications), with fix in igmp.c about mem allocation failure controls, and about a pbuf which is never freed... I have replace printf by LWIP_DEBUG (I have to add new define "formaters" like U32_F), some initializations...
But my own lwip copy got some others little modifications (mainly, a recvfrom timeout which need to add a "timeout" parameter to sys_mbox_fetch).
But except that, I'm synchronized with current CVS HEAD (and IGMP modifications are between #ifdef/#endif) + Christiaan's SNMP.
Do you want a diff?
I join you the Mace Geal patch, posted in forum in March 2006. I know that Christiaan Simons replied to this thread.
(file #12043, file #12044)
|
Mon 26 Feb 2007 03:08:05 PM UTC, comment #2:
Did this ever get merged in? Anyone know?
|
Thu 02 Oct 2003 08:15:08 PM UTC, comment #1:
I forgot that Steve Reynolds already did a good job supporting IGMP v2. We just never got around to merging it in the current lwIP-DEVEL tree.
http://www.sics.se/~adam/lwip/download/igmpfiles.zip
I will do just that, just to have the files available in CVS for us to merge.
|
Wed 17 Sep 2003 05:43:04 PM UTC, original submission:
I may start to implement IGMP as well as test drive
multicasting, depending on our specific needs. Any
feedback or joint effort is welcomed.
http://www.networksorcery.com/enp/protocol/igmp.htm
|