Re: Is EMC Working on RTL 2.2 at all?

[Fred Proctor <frederick.proctor-at-nist.gov>]

EMC folks,

I just built the EMC for RTL 2.2 yesterday. Here are my observations:

1. I used Red Hat 6.2, which is kernel 2.2.14.

2. I got the prepatched kernel source from rtlinux.org, for the 2.2a
patch. I think this may be a patched 2.2.13 kernel; anyone know? I'm
using linux_2_2_14/rtlinux_2_2 as the plat names for this.

3. The default build (do nothing during the "make xconfig", just save
and quit) compiles for SMP. This will define APIC_BASE, or something
like this, that is needed at some point during the EMC compile. If
you're getting this error during the EMC compile, set the SMP option
during "make xconfig". There's a way to compile a uniprocessor kernel
and not get this error, but I forget what we had to do and Will is out
at the moment.

4. The latest EMC release had a problem with steppermod.o and the PERIOD
flag. During the insmod, a "parm_PERIOD" error popped up. This was due
to a missing line in emcstepmot.c, which is correct in emcmot.c. Just
above init_module(), you should have:

MODULE_PARM(PERIOD, "i");
MODULE_PARM(SHMEM_BASE_ADDRESS, "l");
MODULE_PARM(STG_BASE_ADDRESS, "h");
MODULE_PARM(PARPORT_IO_ADDRESS, "l");

The first one, for PERIOD, was omitted in emcstepmot.c. I put it in but
it's not in the release yet.

I built this for a laptop and was able to bring up the EMC, take it out
of estop, and jog. I was running emcfreqmot.o, the frequency stepper
algorithm, and I got a following error. The P gain was set high, 1000 I
think, and the numbers moved a small amount before the following error
kicked in. I'll have to check this again. If anyone notices this, let me
know.

I'll get a new release up for 2.2.14 as soon as Will's disk vertebrae
slip back into place.

--Fred