| To: | netdev@xxxxxxxxxxx |
|---|---|
| Subject: | Windows XP zero initial timestamps |
| From: | "David S. Miller" <davem@xxxxxxxxxx> |
| Date: | Tue, 28 Oct 2003 00:58:42 -0800 |
| Cc: | kuznet@xxxxxxxxxxxxx, b_wydrowski@xxxxxxxxxxxx, casey@xxxxxxxxxx, ccarter@xxxxxxxxxxx |
| Sender: | netdev-bounce@xxxxxxxxxxx |
This is for the people who brought up the zero timestamp issue with Windows XP. It may be allowed by RFC1323, but always using a zero initial timestamp breaks PAWS. PAWS depends upon each end-host using a single host-wide timestamp source for all connections. I can't even imagine what XP is trying to accomplish by using a zero-based timestamp for all connections. But there is no doubt that it breaks PAWS. The best we could ever do is to allow zero initial timestamps but to disable PAWS when using timestamps with such systems. Someone should contact the windows networking coders, I doubt they broke PAWS intentionally. |
| <Prev in Thread] | Current Thread | [Next in Thread> |
|---|---|---|
| ||
| Previous by Date: | Re: Linux 2.6.0-test9, Tommy Christensen |
|---|---|
| Next by Date: | Re: [PATCH] IPV6: inappropriate usage of inet{,6}_sk(), David S. Miller |
| Previous by Thread: | use of Prequeue, an7 |
| Next by Thread: | Re: Announce: NetKeeper Firewall For Linux, David S. Miller |
| Indexes: | [Date] [Thread] [Top] [All Lists] |