Messages in this thread | | | Date | Tue, 11 Aug 1998 09:51:32 +0400 | From | Savochkin Andrey Vladimirovich <> | Subject | Re: 100Mbps TCP stalls in 2.1.115 |
| |
Hi,
On Mon, Aug 10, 1998 at 10:01:46AM +1200, Chris Wedgwood wrote: [...] > I only have problems talking to some stacks, fancy stacks like recent > '95 stacks and perhaps '98 with have scaling, SACK, etc. perform very > well, not to fancy stacks like 2.0.x and NT blow goats. > > A breif (possible inaccurate summary): > 2.1.x Intel <-> Windows95 + IE4.01SP1 - rocks > 2.1.x Intel <-> 2.1.x Alpha - sucks > 2.1.x Intel <-> Windows NT 4sp3 - sucks > 2.1.x Intel <-> 2.0.x Intel - sucks > 2.1.x Alpha <-> Anything ? - rocks > > Sucks means <= 20k/sec ov a 10Mbit switch network. > > Rocks means sustains 800k or greater over the same wire. > > More details and trace available upon request. I'm also happy to > provide shell access (with root and tcpdump) to people I deem > trustworthy if this will help.
Windoze TCP stacks are buggy. Once I observed a windoze box sending full size segment in a closed window and expecting the segment to be rejected. If the receiver accepts the segment the whole connection stalls: windoze rejects all packets because they ack a segment which shouldn't be accepted from microsoft point of view. A nice decision, isn't it?
However, if you gave me access to your system or provided a tcpdump output I could spend some time debugging.
Best wishes Andrey V. Savochkin
- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu Please read the FAQ at http://www.altern.org/andrebalsa/doc/lkml-faq.html
| |