[Debian-eeepc-devel] Network data corruption on atl1e

[Vlad Cerny]

Hello,

I and my friend experience issues with data corruption in
network transfers on EEE 901 and EEE 1000. We found them by copying
of encrypted files over NFS/CIFS ending up with disastrous results
(we also saw scp errors and other issues).

The issues seem to be similar as described in the thread at
http://kerneltrap.org/mailarchive/linux-netdev/2008/12/24/4486694

When I restart my EEE901 laptop, boot Debian lenny (Debian 2.6.26 kernel,
or compiled 2.6.28.8/2.6.29.3), mount NFS/CIFS volume, and calculate
an MD5 of an 1.5 GB file the checksum will be incorrect!! There seems
to be approx. 120 byte block difference at a random offset.

Removing TSO support from the atl1e driver does not help. When I boot WinXP,
mount SMB drive and calculate MD5 it works correctly!!

I have not found more information about this ugly problem. It makes the
laptop unusable. Is anyone experiencing similar problems on the Attansic
atl1e ethernet? Or does the network work reliably for someone?
(which kernel/driver?)

Thanks,

Vlad Cerny
<vlad.cerny at gmail.com>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.alioth.debian.org/pipermail/debian-eeepc-devel/attachments/20090521/69172cb8/attachment.htm>