Hi all, I have an annoying problem which took my whole week and I have no solution in my hand still.
I have many lwIP devices(6 to 12) which are connected to a PC through a hub. I'm transferring ~66 kbytes/sec(992 bytes at every 16msec) from each device to the PC, all the devices running at the same time. In some place at some random time, maybe 5 mins or 5 hours, data transfer is getting currupted such that for a period of approx. 200 msecs no messages(packets) flying around. Then after 200 msecs, devices continue sending packets to the PC. On the ethereal log, it's seen that this happens right after an arp query by the PC. (or some other broadcast packet) In my port, sys_arch_protect is implemented by disabling interrupts. I'm using raw api(TCP) with lwIP 1.1.0. I suspect that this may be a phy issue. I tried to switch to half-duplex mode with no help. Since I see lost TCP segments in the ethereal log, I suspect that lwIP is sending packets but they are not reaching the other end for some reason. Where do yo think the problem can be? In my application code, my port's ethernet driver, in my physical layer? Thanks for your help, Caglar AKYUZ P.S. : Some test results: 1) Problem is not happening on a direct cable connection(without a hub) 2) Problem is happening with various hubs. 3) If I use a Linux PC, problem is rarely seen, with a Windows PC the problem is more frequent. 4) All the devices losing connection at the same time. 5) If I use static arp entries in Linux, all goes well. I can not test this on Windows since Windows has a strange static arp understanding. _______________________________________________ lwip-users mailing list [email protected] http://lists.nongnu.org/mailman/listinfo/lwip-users
