I had something similar with the token ring drivers and sockets. The
network code drops packets where the sk_buff size is greater than the
socket buffer size. On NFS (which defaults to 2048 byte packets) this was
a killer as the t/r default mtu is 4096. Fix in the driver is to test the
packet size and copy it over to a correctly sized sk_buff as most packets
are easily less than the 2048 byte range. (ugly, but works)
Mike
Brent Clements <bclem@rice.edu>
Sent by: linux-net-owner@vger.kernel.org
03/05/03 03:40 PM
To: "David S. Miller" <davem@redhat.com>
cc: linux-net@vger.kernel.org
Subject: Re: Strange network issues using jumbo packets.
Getting the following errors when the card freeze occurs.
tg3: tg3_stop_block timed out, ofs=3400 enable_bit=2
tg3: tg3_stop_block timed out, ofs=2400 enable_bit=2
Anybody have a clue?
Thanks,
Brent
On Wed, 2003-03-05 at 14:06, David S. Miller wrote:
> From: Brent Clements <bclem@rice.edu>
> Date: 05 Mar 2003 13:04:18 -0600
>
> We move both linux machines over to mtu's of 9000 and then remount
the
> directories, but the machine that we are writing from or reading to
will
> just die...ie its network port freezes.
>
> Test with something other than NFS, for example do an FTP
> transfer.
-
: send the line "unsubscribe linux-net" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
-
: send the line "unsubscribe linux-net" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
