error source mac in sended ethernet packet

Hi Ketty Lu,

Could you please provide more details on the issue you observe?

Is this issue in u-boot, linux kernel or user space?

Are you using DM816x TI EVM or custom board?

Are you using the latest code base from arago for u-boot and linux kernel?

Regards,
Pavel

We used DM816x TI EVM,
Our linux kernel is "Linux version 2.6.37+",

we use switch learn the more than one mac in the port which connected to DM8168 ,except the DM8168 MAC,there are other source mac in there MACs,
we use tcpdump catch ethernet packet in DM8168, alse find error source mac in sended ethernet packet,it is not the DM8168 MAC.
we use 14 DM8168 run 12 hours, 5 DM8168 can be find have sended error source mac.

We kill all our process,
use iperl test,the probelm still exited.

We add log infomation in function "emac_dev_xmit" in file "davinci_emac.c",want to find the error mac ,
but only find the local mac,did not find error mac.

Ketty Lu,

ketty Lu said:

Our linux kernel is "Linux version 2.6.37+",

Can you try with the latest kernel version for DM816x DVR RDK kernel, available at the below:

DM8168 TI EVM has two external ethernet ports EMAC0/CPGMAC0/J10 and EMAC1/CPGMAC1/J14. Which one do you use?

Does your u-boot set valid EMAC address? Could you please provide me your u-boot console output, same as below:

Net:   <ethaddr> not set. Reading from E-fuse
Detected MACID:40:5f:c2:63:a2:30
Ethernet PHY: GENERIC @ 0x01
DaVinci EMAC
Hit any key to stop autoboot:  0
TI8168_EVM#md 0x48140630 1
48140630: 000030a2    .0..
TI8168_EVM#md 0x48140634 1
48140634: 63c25f40    @_.c
TI8168_EVM#md 0x48140638 1
48140638: 000031a2    .1..
TI8168_EVM#md 0x4814063C 1
4814063c: 63c25f40    @_.c
TI8168_EVM#print ethaddr
ethaddr=40:5f:c2:63:a2:30
TI8168_EVM#

BR
Pavel

Not only the mac is error,the all IP head had lost,we use wireshark cacth error net packet like this:
" 36 37 38 39 30 31 32 33 34 35 36 37 38 39 30 31
32 33 34 35 36 37 38 39 30 31 32 33 34 35 36 37
38 39 30 31 32 33 34 35 36 37 38 39 30 31 32 33
34 35 36 37 38 39 30 31 32 33 34 35 36 37 38 39
2e 30 2d 34 35 34 34 2e 30 20 73 65 63 20 20 34
2e 30 31 20 47 42 79 74 65 73 20 20 33 34 2e 35
20 47 62 69 74 73 2f 73 65 63 0d 0a "
and the right net packet link this,
" 00 06 f6 7b 9c c8 90 59 af 45 f4 8c 08 00 45 00
05 da 86 1c 40 00 40 11 c7 7a 0a 01 18 4c ac 10
19 1f 81 60 13 89 05 c6 3a 3b 00 00 00 5d 54 89
45 0e 00 08 98 6b 00 00 00 00 00 00 00 01 00 00
13 89 00 00 00 00 05 5d 4a 80 ff 67 69 80 36 37
38 39 30 31 32 33 34 35 36 37 38 39 30 31 32 33
34 35 36 37 38 39 30 31 32 33 34 35 36 37 38 39
30 31 32 33 34 35 36 37 38 39 30 31 32 33 34 35
36 37 38 39 30 31 32 33 34 35 36 37 38 39 30 31
32 33 34 35 36 37 38 39 30 31 32 33 34 35 36 37 "
clear,the error net packet lost 78 bytes ip head.

we use EMAC0,
uboot info is:
VT8168#
VT8168#
VT8168# print
bootargs1=console=ttyO2,115200n8 rootwait=1 ro ubi.mtd=6,2048 rootfstype=ubifs root=ubi0:rootfs init=/init mem=512M vram=20M notifyk.vpssm3_sva=0xbfb00000 vmalloc=476M ddr_mem=2048M atr
bootargs2=console=ttyO2,115200n8 rootwait=1 ro ubi.mtd=8,2048 rootfstype=ubifs root=ubi0:rootfs init=/init mem=512M vram=20M notifyk.vpssm3_sva=0xbfb00000 vmalloc=476M ddr_mem=2048M atr
bootcmd1=nand read 80009000 0x1c0100 260000; bootm 80009000
bootcmd2=nand read 80009000 0x2240100 260000; bootm 80009000
bootdelay=3
baudrate=115200
ipaddr=10.1.24.189
serverip=10.1.24.120
autoload=no
ethaddr=90:59:af:45:f4:8c
ethact=DaVinci EMAC
stdin=serial
stdout=serial
stderr=serial
bootargs=console=ttyO2,115200n8 rootwait=1 ro ubi.mtd=8,2048 rootfstype=ubifs root=ubi0:rootfs init=/init mem=512M vram=20M notifyk.vpssm3_sva=0xbfb00000 vmalloc=476M ddr_mem=2048M atr uboot-v2.1(Jun 04 2014-03:45:03) usb
bootcmd=nand read 80009000 0x2240100 260000;bootm 80009000;reset;
ver=U-Boot 2010.06-svn2678 (Jun 04 2014 - 03:45:03)

Environment size: 1000/131068 bytes

Ketty Lu,

Did this discussion go offline and you reached a conclusion? Has there been any new information on this for you?

Were you able to run the uboot console commands that Pavel listed above, and show that output?

Regards,
RandyP

Ketty Lu,

I found that you continued this on a different thread here and had a discussion with Merry Christmas. That thread also did not seem to come to a conclusion, so I am still concerned with this.

You saw this problem on an EVM with two ports connected to a switch. That should be very helpful for duplicating the issue anywhere.

Were you able to update to the latest Linux kernel from Pavel's link, above? Did that make any difference?

Regards,
RandyP