RTlinux 2.2, mbuff 0.7.1 and Linux 2.2.14

[Philip N Daly <pnd@noao.edu>]

Hi,

(This message is posted to both www.rtlinux.org and www.realtimelinux.org).

Has anyone got mbuff 0.7.1 working under SMP with RTlinux 2.2?

Although the system is RedHat 6.1, I started off with a squeaky clean 
2.2.14 kernel from ftp.kernel.org. I installed RTLinux 2.2 and the examples 
all work fine. I installed mbuff 0.7.1 and the examples work fine.

Now, I upgraded my mbuffex tester and here's what happens:

1. Using Kmem/Umem, everything works as expected. Kmem is a pure Linux
   re-scheduling module and Umem the user side application. They share
   two memory buffers just fine. So it looks like mbuff works ok in
   pure Linux kernel space etc.

2. I upgraded Rmem.c to be used by RTLinux and here's what happens 
   (after insrtl and insmod of mbuff.o, of course):

   a) on at least one occassion, the systems hangs straight after the
	  insmod of the real-time mbuffex module;

   b) on two occassions, I have been able to insmod the real-time
	  mbuffex module and run the test to successful completion. The 
	  3rd insmod test hangs the system;

   c) on one occasion, the insmod blew the system away to a console (non-X)
	  state with the repeated message: "hda: lost interrupt".
      (Is this related to ide0 irq 14 problem via xosview? Unlikely but
	   suspicious, I think)

Of course, there's nothing in the log file to help.

If you want to take a look at the interim code, the URL is
  ftp://orion.tuc.noao.edu/pub/pnd/mbuffex-0.7.1.tgz

Thanks in advance,


+==================================================================+
 Phil Daly, NOAO/AURA, 950 N. Cherry Avenue, Tucson AZ 85719, U S A
 E-mail: pdaly@noao.edu  V-mail: (520) 318 8438  Fax: (520) 318 8360