Ray Henry wrote: > On Wed, 2008-12-03 at 10:16 -0800, Peter C. Wallace wrote: > >> Actually we started a 7I43 with Luminary Micro ARM Ethernet interface CPU >> back >> in May or so but custom work kept me from finishing it. I will start again >> soon. >> >> As John says the real problem is getting realtime Ethernet support on the >> EMC >> host. >> >> As I said before maybe the way to ease into this is just support 1 or a few >> Ethernet chips, and require the user to have a add-in PCI/PCIe Ethernet card >> with the required chip. >> > > This sounds like just the right approach. We should pick a PCI ethernet > card with a chip we can work with and begin developing a HAL driver so > that when you get the ethernet version of the 7I43 working we have a way > to begin testing. > > Any thoughts on a specific card to begin testing with? > The Arm 7 chips would be fine for the stuff I want to do with them. There are a few Arm 7 chips that have 10/100 ethernet support built into the chip. On Linux, a standard network stack that was accessible from the real time environment would work in the case I'm thinking of. The target device would only speak when spoken to, it would always be idle when receiving a command so it should be able to respond quite quickly, like within 10 us or so. The scheme I am thinking of is the regular dispatch of the servo thread sends a request to the target device, it reads encoder counts and switch condition and sends back a response packet. The PID calculates new velocities and sends a command packet, there's no need to reply to that. This would allow standard protocol stacks to be used on the Arm target CPU.
So, the only thing I don't know how to do is get a protocol stack that is accessible by an RT thread. rtnet is a time-sliced variant of ethernet, and requires at the very least modified protocol stacks on all senders, and there might be hardware incompatibilities on some chips. It might be, in fact, that the standard protocol stack would be fine, the way I have described the scheme above, I know only a LITTLE about rtnet. rtnet's time slot assignment is to prevent any form of collision or net monopolizing, but the protocol I describe is VERY limited, and should have the same effect. Jon ------------------------------------------------------------------------- This SF.Net email is sponsored by the Moblin Your Move Developer's challenge Build the coolest Linux based applications with Moblin SDK & win great prizes Grand prize is a trip for two to an Open Source event anywhere in the world http://moblin-contest.org/redirect.php?banner_id=100&url=/ _______________________________________________ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users