<html>
<head>
<style>
.hmmessage P
{
margin:0px;
padding:0px
}
body.hmmessage
{
FONT-SIZE: 9pt;
FONT-FAMILY:Tahoma
}
</style>
</head>
<body class='hmmessage'>
Dear Robert<BR><BR>> Let me guess. <BR>> You are using a base system from "Base System Builder Wizard"?<BR>> EDK 9.2i. Default syntheses/P&R options.<BR><BR>
I am using EDK 10.1 evaluation version (60 days) from Xilinx. I don't have experience on EDK 9.2 and previously I was working on EDK 8.2. <BR>
<BR>> I have seen a massive packet loss problems on my ML403 and two other<BR>> boards I have with an FX60.<BR>> <BR>> This is probably a hardware problem.<BR>> Xilinx has acknowledged an LL_TEMAC problem to me but has not provided a<BR>> fix. I have heard that things are better with EDK/ISE-10.1 but I have<BR>> not tested it.<BR><BR>
I got the following message from my NFS server PC: <BR>
<BR><PRE>Apr 1 15:08:34 linux kernel: UDP: bad checksum. From 192.168.0.4:839 to
192.168.0.3:627 ulen 48
Apr 1 15:08:40 linux kernel: UDP: bad checksum. From 192.168.0.4:839 to
192.168.0.3:627 ulen 48
Apr 1 15:08:49 linux kernel: UDP: bad checksum. From 192.168.0.4:839 to
192.168.0.3:627 ulen 48
Apr 1 15:13:23 linux kernel: nfsd: last server has exited
Apr 1 15:13:23 linux kernel: nfsd: unexporting all filesystems
Apr 1 15:13:23 linux rpc.mountd: Caught signal 15, un-registering and
exiting.
Apr 1 15:13:24 linux kernel: NFSD: Using /var/lib/nfs/v4recovery as the
NFSv4 state recovery directory
Apr 1 15:13:24 linux kernel: NFSD: recovery directory
/var/lib/nfs/v4recovery doesn't exist
Apr 1 15:13:24 linux kernel: NFSD: starting 90-second grace period
Apr 1 16:02:57 linux syslog-ng[3700]: STATS: dropped 0
Apr 1 17:02:57 linux syslog-ng[3700]: STATS: dropped 0
Apr 1 18:02:58 linux syslog-ng[3700]: STATS: dropped 0</PRE>
I shows "UDP bad checksum" from my board to the PC. I don't know if it has something to do with the package loss problem. <BR>
<BR>> The vendor of one of my boards (Pico) has fixed the problem on the FX60<BR>> by highly constraining the timing of the LL_TEMAC in map/PR. On my<BR>> ML403 I used similar constraints and it fixed the problem, but only if<BR>> the device is plugged into a GigE switch. The problem is still there<BR>> with the same .bit file on a 100-T switch. <BR><BR>
I connect the board with the PC point-to-point with gigabit ethernet. It should be even better than a Gigabit switch, right? <BR>
<BR>
BR<BR>
Ming<BR><br /><hr />使用新一代 Windows Live Messenger 轻松交流和共享！ <a href='http://messenger.live.cn/' target='_new'>立即体验！</a></body>
</html>