Hi，
I use SJA1105 and TJA1100 on my board, but I have some problems.
The block diagram of my system is as follows, MCU1 and MCU2 run different platforms (RTOS and Linux), and all of them support TCP/IP stack.
H/W Configurations of TJA1100 are as follows:
CONFIG0: HIGH, PHY configured as Master
CONFIG1: LOW, Managed operation
CONFIG3/2: LOW HIGH, RMII mode (external 50MHz oscillator) <-- SJA1105 supplies 50MHz
When I configure the TJA1100 (by MCU1) as a master, Media Converter as a slave, link is OK.
Then I use PC to ping MCU1 or MCU2, results are also OK.
Dumped register data from TJA1100 when link down and link up are as follows.
When I configure the TJA1100 as a slave, Media Converter as a master, link is OK too.
But when I power on my board, and TJA1100 turns to Normal Mode at the first time, it spends a lot of time (> 1.5min) to establish the first link. After this first link up, I disconnect the cable between TJA1100 and Media Converter, link down immediately. Then I connect the cable again, link up immediately.
Then I use PC to ping MCU1 or MCU2, results are NG.
Dumped register data from TJA1100 when link down and link up are as follows.
I try to use Wireshark to find out the reason why ping fails.
I compare the communication logs when TJA1100 works as a master and slave.
I found that when TJA1100 works as a slave, 9-byte extra random data are added to the tail of an ethernet package, it causes ARP reply package be dropped by the net card of PC, and ping failure.
(Due to the same results when I use MCU1 and MCU2, It’s hard to consider that there are bugs in MCU1 or MCU2 software)
As a comparison, ARP reply package is as follows when TJA1100 works as a master.
The Questions I want to ask are as follows:
- Why the first link up spends a lot of time when TJA1100 works as a slave ?
- What may cause the 9-byte extra random data in the ethernet package ?
Thank you.
