The default 4 is really small, and imposes very sequential processing of
reconnections in case we lose all sessions. Nowadays we easily have
memory for a lot more.
controlc doesn't mean we will send more messages, but only that some of them
are pending ack. If we have already sent them and the peer has just not
acked them yet, controlnull() would not send a ZLB, and once the peer
has acked our control messages, we wouldn't send a ZLB with an updated Nr.
Previously, when we got a call to be forwarded to another LNS and we didn't
have a tunnel to it yet, we would wait for a PAP/CHAP re-send to trigger the
creation of a session in the fresh tunnel. We do not have to wait, we can
trigger that on SCCCN ack.
Retransmissions are done very slowly, so we'd really better store the
Out-of-Order messages so that we can catch up quickly once we get the
missing piece.
Creating/destroying interfaces etc. does take some time. When e.g. receiving
a lot of sessions as new slave, we don't want to stay stuck creating
hundreds of interfaces while we are already receiving control messages
that we have to forward to master not too late. Switching kernel
acceleration can wait a bit most of the time.
fmtaddr formats the string to a static buffer that must be selected
manually, but this log message was re-using the same buffer.
Use another index for 2nd IP so we get to know which is what.
Fixes: 2429969bd04d ("Add L2TP offloading support")
Otherwise with kernel acceleration we would only have the same /64 route
on all ppp devices. We need separate /96 routes on each ppp device for
packets of the /96 subprefix to be routed to the correct session.
If we are a master and we don't have peers, there is no use keeping the
sessions alive, when we'll restart we will have forgotten them all.
Better gracefully close them, so they can be quickly respawn on restart,
instead of having to wait for timeouts.
The RFC indeed say that we should send them periodically. We were
previously only sending them along LCP echo replies, but echo requests
are typically sent only when there is no trafic, which RA need to be
sent even when there is trafic.
The size of the memset was just wrong, but the data has already been
zeroed in initdata so we can just remove it.
# This is the commit message #2:
# squash! initippool: fix initialization of ip_address_pool
When acting as LNS, we have to answer challenges with our own secret, but
when acting as LAC, we have to answer challenges with the LNS secret, not
ours.
send/recvchalresponse should be freed before early returns in the function,
as there are code paths where they would be allocated before these.
Note free is no-op on null pointer so checking for non-null value is useless.
Reported-by: Coverity#375342
