Hello,
I would use strlcpy() rather than strncpy() which is almost considered deprecated within the kernel (see http://lwn.net/Articles/33812/ ). Here, it could cause hlpr->name not to be NULL terminated, which could lead to problems later on, depending how this string is read.
Thanks for maintening this RTSP conntrack code!
Fab
Wed Dec 26 21:35:55 2012
Fabsays:
Also, it seems the fix is needed since 3.5, not 3.6.
Fri Jan 4 23:41:41 2013
Fabsays:
Hello,
since kernel 3.7, IPv6 NAT has come, and the related netfilter code has changed towards genericness.
The RTSP module doesn't build/work anymore on this version, so I adapted it to the new 3.7 code. Please see the following two patches:
Sorry, I'm using a git branch to maintain a RTSP patch on the kernel code, so I can compile it straight within the kernel (no module).
However, I believe these two patches should be fairly easy to apply on the module code.
Fab
Sat Jan 5 10:54:54 2013
Michael Guntschesays:
Thanks for the patches Fab.
It did not apply cleanly to my tree but I fixed that. The problem is that I can't even compile test this right now. I'll try to upload a new patch today, maybe you can try if it works.
Discussion
I would use strlcpy() rather than strncpy() which is almost considered deprecated within the kernel (see http://lwn.net/Articles/33812/ ). Here, it could cause hlpr->name not to be NULL terminated, which could lead to problems later on, depending how this string is read.
Thanks for maintening this RTSP conntrack code!
Fab
since kernel 3.7, IPv6 NAT has come, and the related netfilter code has changed towards genericness.
The RTSP module doesn't build/work anymore on this version, so I adapted it to the new 3.7 code. Please see the following two patches:
http://sebsauvage.net/paste/?1a65c4ba38de6d70#iQijKE4kSLTOfMQggEkeCBXb1z+xMi7tTvgEOUTghvk=
http://sebsauvage.net/paste/?07b73b3916b6f976#4R+LpKmrk9IUKNSGdzK7NbuKg+S54vfpHW4E96Y6Y6s=
Sorry, I'm using a git branch to maintain a RTSP patch on the kernel code, so I can compile it straight within the kernel (no module).
However, I believe these two patches should be fairly easy to apply on the module code.
Fab
It did not apply cleanly to my tree but I fixed that. The problem is that I can't even compile test this right now. I'll try to upload a new patch today, maybe you can try if it works.