Reply to this email directly, view it on GitHub, or unsubscribe.
You are receiving this because you are subscribed to this thread.
I found some failures in the stress test of MSRP messages and prompted the following error log：
DEBUG: msrp [msrp_netio.c:98]: msrp_relay(): To-Path has only one URI -- nowehere to forward；
After adding the log embedding point, it is found that there is a problem in the logic of adding the linked list of "_msrp_cmap_head - > cslots". The node with a smaller hash value will be added behind the node with a larger "citemid" value, so that the linked list is not arranged strictly in the ascending order of "citemid", resulting in the failure of subsequent queries such as "msrp_cmap_lookup", and thus the normal relay cannot be performed.
37(81) ERROR: msrp [msrp_cmap.c:291]: msrp_cmap_save(): _msrp_cmap_head->cslots[492932] item citemid is [2183628164], sessionid is [xxxxxxxx]
37(81) ERROR: msrp [msrp_cmap.c:291]: msrp_cmap_save(): _msrp_cmap_head->cslots[492932] item citemid is [2174190980], sessionid is [xxxxxxxx]
2183628164>2174190980, out of order.
source code:
for(itb=_msrp_cmap_head->cslots[idx].first; itb; itb=itb->next)
{
if(itb->citemid>it->citemid || itb->next==NULL) {
if(itb->next==NULL) {
itb->next=it;
it->prev = itb;
} else {
it->next = itb;
if(itb->prev==NULL) {
_msrp_cmap_head->cslots[idx].first = it;
} else {
itb->prev->next = it;
}
it->prev = itb->prev;
itb->prev = it;
}
break;
}
}
(paste your debugging data here)
(paste your log messages here)
(paste your sip traffic here)
Tested against kamailio 5.4.x and actually all versions are affected`
fixed code:
for(itb=_msrp_cmap_head->cslots[idx].first; itb; itb=itb->next)
{
if(itb->citemid>it->citemid || itb->next==NULL) {
//modifed code
if(itb->next==NULL && (itb->citemid < it->citemid)) {
itb->next=it;
it->prev = itb;
} else {
it->next = itb;
if(itb->prev==NULL) {
_msrp_cmap_head->cslots[idx].first = it;
} else {
itb->prev->next = it;
}
it->prev = itb->prev;
itb->prev = it;
}
break;
}
}
Ubuntu 16.04
—
Reply to this email directly, view it on GitHub, or unsubscribe.
You are receiving this because you are subscribed to this thread.