...and please always use the correct list address so that "reply to all" works! :)
Ja
--- Begin Message ---Leonardo Pappagallo wrote: > Hello Jan, > > >>> I use Rtnet 0.8.3 and Rtai 3.1.The console tell me the same think that > you >>> have said me(memory leakage and the requirement to reboot). >>> I'm using the driver rt_3x59c. >>> >>> Do you have any idea? > >> Try the same with some other driver. The rt_3c59x is not well tested due >> to its fundamental latency issues. I would like to exclude problems here >>from core problems. > > I want to put in evidence, that I have the same problem with both drivers > rt_3c59x and rt_eepro100. > In my application, I have noticed that the behaviour of driver 3c59x is same > of eepro100, for this I think that there is another problem > that I have to reboot all my machines. > > do you have another idea to resolve this situation? As long as we don't know the reason, we cannot discuss a resolution. We first need to track it down by going step by step through the init stages of the rtnet script and calling rtnet stop afterwards until this warning shows up. Then we know what subsystem/what step triggers the problem. The steps are roughly: 1. load basic modules (rtnet+drivers) and do a rtifconfig up 2. load some application or do a rtping 3. load protocol modules and configure RTmac/TDMA via RTcfg (based on rtnet script - otherwise, this would be a too lengthy explanation) 4. load some application or to a rtping Check after each step if rtnet stop can clean up without generating that warning on the kernel console. > >>>> I have another question to ask you: >>>> >>>> I have noticed that I can't configure the slot 2 on my machine with >> driver >>>> rt_eepro100 infact when i see the file (/proc/rtnet/rtmac cat >> tdma_slots) >>>> the slot 2 doesn't appear also if i have put this in my file tdma.conf. >>>> While with other two devices (rt_3c59c) there isn't this problem. >>> Very strange as there is no relation between the Ethernet driver and the >>> precise RTmac/TDMA configuration. I guess this is some obscure side >>> effect of a major issue. Already tried to configure this slot manually >>> with tdmacfg? > > > Using command line tdmacfg I have resolved the situation. > Please post the tdma.conf which triggers this problem. Maybe we have a bug in the rtnet script. > Could you tell me what are the steps to configure RTnet manually without > using this command? : > 1) rtnet start (using rtnet.conf and tdma.conf) > 2) rtnet stop. > > I would like to do like this: > > insmod rtnet > insmod rt_<DRIVER> > insmod rtcfg (just for starting the timer) > rtifconfig rteth0 up <IP> > rtroute > .... > As said above, it's best to follow the rtnet script to see what happens. A rough overview: 1. rtnet+driver+protocols are loaded 2. interface is started 3. rtcfg is configured either to obtain config from the server or to distribute the existing data (e.g. tdma.conf) to the rtcfg clients 4. on TDMA slave, the received config is loaded into TDMA; on the master, this happens based on tdma.conf 5. when all slaves are up, rtcfg is used to signal the common startup to all nodes A lot of shell commands and also a bit shell magic is involved in this config distribution. Ask if you do not understand a particular line. > >>>> What do you think about this situation? Do you think that is it >> necessary to >>>> use the slot 2 for user traffic? >>>> >>> Well, the best thing would be if you could try to test on a more recent >>> platform. Some things have improved since 0.8.3 (although I don't see a >>> concrete fix related to your observations), and it would be easier to >>> track the issues on the latest release. Unfortunately, RTAI itself is >>> not in a reliable state regarding compilation and existing test results >>> right now, but Xenomai 2.0.x is well tested and should work. If not, we >>> do have a bug! > > > Thanks for this advertisement but i woud like to know If it is necessary to > use the slot 2 for user traffic. Is It ? > It is only required if some of your real-time applications plans to use it explicitly. If not, all of the real-time traffic go over slot 0 by default. See also README.rtmac. Jan
signature.asc
Description: OpenPGP digital signature
--- End Message ---
signature.asc
Description: OpenPGP digital signature
