rpl: refactor send_dao to use pktbuf properly #4202
Conversation
cgundogan
added
Type: enhancement
cgundogan
added
the
CI: ready for build
|
nice one, ripping the |
| size += sizeof(gnrc_rpl_opt_dodag_conf_t); | ||
| if (dodag->prefix_info_requested) { | ||
| if ((pkt = _dio_prefix_info_build(pkt, dodag)) == NULL) { | ||
| return; |
There was a problem hiding this comment.
OT: shouldn't we do something if the packet cannot be created here and in the further cases below, at least return an error to the caller?
There was a problem hiding this comment.
the pkt is release in the _*_build funcitons if it wasn't possible to allocate a new buffer
There was a problem hiding this comment.
I know, but RPL wrongly assumes now it successfully sent a DIO, and in consequence it will be eventually deleted from the parent sets of its children if this situation sustains
There was a problem hiding this comment.
Well, there is not much I can do if the pktbuf is full (and stays full for further DIOs). I guess it is good then that children choose another parent
is this an ACK? (: |
|
not yet, but close to ;) |
|
tested on native, works, travis is happy, ACK'n'Go |
rpl: refactor send_dao to use pktbuf properly
The first commit refactors the sending DIO code to make proper use of the pktbuf API instead of pointer arithmetic.
The second commit adds a missing pktbuf release in the sending DAO code.