Roland Tollenaar wrote:
Hi Jan,100% the same. 101% the same even because all the other hardware is identical too. And may I add, exactly the same software as well. From operating system to application.Perhaps I have not been entirely clear in my email. But the application that generally uses rteth0 does not get hold of it either when it cannot be found in /proc/rtnet/devices. It appears that the application did have hold of rteth0 but lost it. When checking /proc/rtnet/devices after the application no longer has access to it, it appears to be gone.Advice would be appreciated.That's weird. The /proc output is built from iterating over all possible device numbers, looking the structure up via rtdev_get_by_index(). So if /proc tells you that there is no rteth0 anymore while the rest of RTnet still uses it, something must have corrupted rtnet_devices[]. Maybe someThe application code has been operational for 7 months on a number of separate but identical machines and never displayed this behaviour before. Only one of the machines is now playing up with this.Same hardware, specifically same NIC and driver?Maybe you want to check now if someone deregisters the device. You could add a WARN_ON(1); into rt_unregister_rtnetdev(). After the device disappeared, you should either have a stack dump in your kernel logI can try this. But it would have to be a process that does this. The machine is running embedded and we did not put a "deregister NIC" button in the GUI -:). Off hand I can think of no process that could come close to deregistering the device.Hmmm...so you think that the chances of this being a hardware fault (in the NIC) are slim?or... we really see a memory corruption. :-/
No, I'm not excluding this possibility. That's why I'm interested in the stack backtrace: if Linux happens to decide that the PCI device is no longer present, it may trigger a removal which, in the end, causes the driver (which one, BTW?) to unregister itself from RTnet. I haven't seen this in real live before, though. Are there no suspicious messages in the kernel log?
Jan
signature.asc
Description: OpenPGP digital signature
------------------------------------------------------------------------- This SF.net email is sponsored by the 2008 JavaOne(SM) Conference Register now and save $200. Hurry, offer ends at 11:59 p.m., Monday, April 7! Use priority code J8TLD2. http://ad.doubleclick.net/clk;198757673;13503038;p?http://java.sun.com/javaone
_______________________________________________ RTnet-users mailing list RTnet-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/rtnet-users
