[...]
The diff looks mostly ok, though checking for S_ISLNK from stat(2) seems useless, and I'm not quite sure why the chmod is done *after* writing the config file rather than upfront.
I have added the patches to fix the permission issues as done per upstream but keeping in mind no to change too much things on my own. I have also updated the patches to make it works against the rc2 but just a bit. For the client, the permissions should not be overwritten at the end of the function but rather set when created as you mentionned : I think you are right.
For the S_ISLNK, I have to check further.I am going to check all that and try to make the 2.0.3 release build on MIPs.
Regards, Franck