diff options
author | Xin Long | 2017-04-15 16:00:28 +0200 |
---|---|---|
committer | David S. Miller | 2017-04-18 19:39:50 +0200 |
commit | d0f025e611581169c81e3c0fc44b5133d12748dd (patch) | |
tree | a4328d92cba9958abe1472ba10d2f91423d57201 /include/uapi/xen/gntdev.h | |
parent | sctp: process duplicated strreset out and addstrm out requests correctly (diff) | |
download | kernel-qcow2-linux-d0f025e611581169c81e3c0fc44b5133d12748dd.tar.gz kernel-qcow2-linux-d0f025e611581169c81e3c0fc44b5133d12748dd.tar.xz kernel-qcow2-linux-d0f025e611581169c81e3c0fc44b5133d12748dd.zip |
sctp: process duplicated strreset in and addstrm in requests correctly
This patch is to fix the replay attack issue for strreset and addstrm in
requests.
When a duplicated strreset in or addstrm in request is received, reply it
with bad seqno if it's seqno < asoc->strreset_inseq - 2, and reply it with
the result saved in asoc if it's seqno >= asoc->strreset_inseq - 2.
For strreset in or addstrm in request, if the receiver side processes it
successfully, a strreset out or addstrm out request(as a response for that
request) will be sent back to peer. reconf_time will retransmit the out
request even if it's lost.
So when receiving a duplicated strreset in or addstrm in request and it's
result was performed, it shouldn't reply this request, but drop it instead.
Signed-off-by: Xin Long <lucien.xin@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Diffstat (limited to 'include/uapi/xen/gntdev.h')
0 files changed, 0 insertions, 0 deletions