giomba
/
nes-proj
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

Turns out that with cxmac, clearing the header after sending a packet is not enough to make fragmentation work. Instead, we allocate a queuebuf and restore the packet after from the queuebuf after sending it.

This commit is contained in:
adamdunkels 2010-02-06 09:42:49 +00:00
parent 494e692d0b
commit b6647b5630
1 changed files with 15 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

16
core/net/sicslowpan.c
View File

@ -32,7 +32,7 @@
*
* This file is part of the Contiki operating system.
*
* $Id: sicslowpan.c,v 1.13 2010/02/06 07:49:58 adamdunkels Exp $
* $Id: sicslowpan.c,v 1.14 2010/02/06 09:42:49 adamdunkels Exp $
*/
/**
* \file
@ -1180,6 +1180,7 @@ output(uip_lladdr_t *localdest)
if(uip_len - uncomp_hdr_len > MAC_MAX_PAYLOAD - rime_hdr_len) {
#if SICSLOWPAN_CONF_FRAG
struct queuebuf *q;
/*
* The outbound IPv6 packet is too large to fit into a single 15.4
* packet, so we fragment it into multiple packets and send them.
@ -1212,7 +1213,14 @@ output(uip_lladdr_t *localdest)
memcpy(rime_ptr + rime_hdr_len,
(void *)UIP_IP_BUF + uncomp_hdr_len, rime_payload_len);
packetbuf_set_datalen(rime_payload_len + rime_hdr_len);
q = queuebuf_new_from_packetbuf();
if(q == NULL) {
PRINTFO("could not allocate queuebuf for first fragment, dropping packet\n");
return 0;
}
send_packet(&dest);
queuebuf_to_packetbuf(q);
queuebuf_free(q);
/* set processed_ip_len to what we already sent from the IP payload*/
processed_ip_len = rime_payload_len + uncomp_hdr_len;
@ -1243,7 +1251,13 @@ output(uip_lladdr_t *localdest)
memcpy(rime_ptr + rime_hdr_len,
(void *)UIP_IP_BUF + processed_ip_len, rime_payload_len);
packetbuf_set_datalen(rime_payload_len + rime_hdr_len);
if(q == NULL) {
PRINTFO("could not allocate queuebuf, dropping fragment\n");
return 0;
}
send_packet(&dest);
queuebuf_to_packetbuf(q);
queuebuf_free(q);
processed_ip_len += rime_payload_len;
}