[lwip-users] Corrupt output packets with multiple TCP protocols active

[Pettinato, Jim]

Hello all,

 

I've been seeing an odd problem with my product and was hoping someone might have bumped into something similar or could give me some ideas as to possible culprits.

 

I'm running lwIP 1.2.0 using the raw API with an SMSC 91C111 MAC/PHY. The interface works flawlessly for extended periods when just Modbus/TCP is running, but when a second protocol is activated (HTTP, FTP, custom apps, etc.) I begin to see periodic checksum errors with my outgoing response packets on Wireshark. Upon examining the error packets I can see they've been corrupted somewhere along the way. They are sometimes short, sometimes long, and sometimes it appears data from two separate apps has been mangled together. The corruption is not just limited to the TCP payload data - I've seen trashed seq / ack numbers, etc.

 

I've been concentrating mostly on the weird 91C111 buffer allocation area of my code, but in several days of probing haven't really seen a hole in my low-level driver. I've tried some changes to that code with no effect also.

 

Any ideas??

 

Thanks in advance,

 

- Jim

 

__

 

James M. Pettinato, Jr.

Software Engineer

E: jim.pettinato@fmcti.com | P: 814 898 5250

FMC Technologies Measurement Solutions Inc.
1602 Wagner Avenue | Erie PA | 16510 USA
Phone: 814 898 5000 | Fax: 814 899-3414
www.fmctechnologies.com