Linux build

[Duane Gustavus]

Hello Folks,

I have a Robomind 68332 card that has been mentioned on this list
before.  I am using Linux (RH 7.1) and have installed the GNU tools
for 68k architecture using the supplied rpms (ie in /opt/rtems).  I
have downloaded and unpacked the source in /opt/rtems to make
/opt/rtems/rtems-4.5.0.  I have made an /opt/rtems/build-rtems
directory, cd'd into it and run ../rtems-4.5.0/configure
--target=m68k-rtems --disable-itron --disable-networking.  This all
seems to work correctly.

For a make command, I am using make RTEMS_BPS="efi332".  I realize
this will probably not build code loadable in my Robomind card, but it
will verify that I can build RTEMS for a 68332 with the tools (a big
step in the right direction).  The make goes for quite a bit building
lots of modules without problems.  Then I get what appears to be some
configuration error:

make[5]: Entering directory `/export/rtems/build-rtems/m68k-rtems/c/efi332/exec/posix/src'
mkdir o-optimize
/opt/rtems/bin/m68k-rtems-gcc --pipe -B../../../../../../efi332/lib/ -specs bsp_specs -qrtems -g -Wall -ansi -fasm    -mcpu32     -c -D__RTEMS_INSIDE__ -O4 -fomit-frame-pointer -o o-optimize/adasupp.o ../../../../../../../rtems-4.5.0/c/src/exec/posix/src/adasupp.c
In file included from ../../../../../../efi332/lib/include/sched.h:20,
                 from /usr/include/pthread.h:20,
                 from ../../../../../../../rtems-4.5.0/c/src/exec/posix/src/adasupp.c:7:
/opt/rtems/lib/gcc-lib/m68k-rtems/2.95.2/../../../../m68k-rtems/include/sys/sched.h:44: field `ss_replenish_period' has incomplete type
/opt/rtems/lib/gcc-lib/m68k-rtems/2.95.2/../../../../m68k-rtems/include/sys/sched.h:46: field `ss_initial_budget' has incomplete type

I have looked at the sched.h file, and the ss_replenish_period and
ss_initial_budget don't appear to be erroneous.  It looks suspicious
to me that the /usr/include/pthread.h file is being included, but I
don't really know what is going on there. BTW, the various reporting
of /opt/rtems and /export/rtems above is caused by the fact that
/opt/rtems in a symbolic link to /export/rtems (where the free disk
space is).  On the theory that this might have something to do with
the efi332 configuration, I tried one of the other BSPs mentioned in
the configure exit message.  The result was the same, failing on the
same file (with the exception of being in the mvme136 path).

Up to this point everything has worked perfectly, so I am hoping this
is not too big an issue (I am following the online docs and have
checked them twice to see if I can spot what I'm doing wrong, but no
joy).  I would certainly appreciate useful suggestions as how to
proceed with the make.

On another issue, the archives show some discussion of a BSP for the
Robomind card (even pointing out that the efi332 BSP was a good
starting point).  Did anyone ever make a useable BSP for this card?  I
think RTEMS might be a great platform for a free robot environment if
some inexpensive cards are supported for it.  I am willing to help
with the MRM board if others are interested.

Thanks for any info you can provide,
Duane Gustavus
Denton, TX