[strongSwan-dev] [PATCH 0/2] Add reference counting to child_sa
Martin Willi
martin at strongswan.org
Tue May 18 12:34:22 CEST 2010
> Our send/receive_delay testing hooks are insufficient to simulate this,
> I'll have to add some conditional packet delays to reproduce it.
I have added a simple conditional packet delay mechanism, it allows us
to delay some specific packets and reproduce this issue.
> I'll try to reproduce this condition
I configured box with:
strongswan.conf:
send_delay = 1000
receive_delay = 5000
receive_delay_type = 36
receive_delay_request = no
ipsec.conf:
lifetime=30s
margintime=11s
rekeyfuzz=0%
For xob, I used:
strongswan.conf:
send_delay = 1000
ipsec.conf:
lifetime=30s
margintime=10s
rekeyfuzz=0%
This will trigger a simultaneous rekey in ~20s, but will delay the
problematic CREATE_CHILD_SA response a little more. With this
configuration, I could reproduce the problem.
> I think it should be possible to catch this in the collide() check.
I additionally checked in a patch that fixes the problem for this
configuration for both cases, if xob wins the nonce compare, but also if
it loses.
Patches attached, please verify.
Best regards
Martin
-------------- next part --------------
A non-text attachment was scrubbed...
Name: 0001-Added-simple-conditional-packet-send-delay.patch
Type: text/x-patch
Size: 2323 bytes
Desc: not available
URL: <http://lists.strongswan.org/pipermail/dev/attachments/20100518/b00cf42e/attachment.bin>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: 0002-Added-simple-conditional-packet-receive-delay.patch
Type: text/x-patch
Size: 2857 bytes
Desc: not available
URL: <http://lists.strongswan.org/pipermail/dev/attachments/20100518/b00cf42e/attachment-0001.bin>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: 0003-Handle-collisions-between-rekey-and-the-following-de.patch
Type: text/x-patch
Size: 3964 bytes
Desc: not available
URL: <http://lists.strongswan.org/pipermail/dev/attachments/20100518/b00cf42e/attachment-0002.bin>
More information about the Dev
mailing list