Hello Kaustubh,
1) auto-summarization can create problems if two routers announce to a third router that they can reach net 10.0.0.0/8 because the backbone lan is in another major network for example the backbone lan is 172.16.20.0/24.
R3 will see two paths to net 10.0.0.0/8 and it will try to do load balancing.
But R1 can reach only some subnets within net 10/8 and so happens for R2.
Auto-summarization at major network boundaries cause problems if a major network is partitioned, from the point of view of routers that haven't an interface on the partitioned major network multiple paths to it exist as in the example above and so routing can fail.
Modern protocols like EIGRP, RIPv2 and BGP allows to disable auto-summary and they are usually configured in this way.
No auto-summary at major network boundary (class A,B,C) exists in link state protocols like OSPF and IS-IS
2) a router performs a recursion when the next hop for a route is part of another route:
example
B [200/0] 10.90.10.0/24 via 10.80.0.25
and then
sh ip route 10.80.0.25
O [110/1500] 10.80.0.25/32 via ...
in this case a network has been learned in iBGP (AD 200) and the packets for net 10.90.10.0/24 have a BGP next hop of 10.80.0.25
This is a loopback address on a remote router.
Packets for the net 10.90.10.0/24 follow the same path as packets for 10.80.0.25
This is recursion. On this a lot modern architectures and techologies have built their services: for example MPLS, MPLS VPN.
Hope to help
Giuseppe
