A patch for the bit script

Michael P. Collins on korat mcollins at hawkeye.sps.mot.com
Thu Jun 29 01:42:32 UTC 2000

Hans Peter Jepsen wrote:

> The attached patch for the script bit includes three fixes/enhancements:

  I was unable to access the attachment.  Perhaps you could appended it
as text rather than attach it as an encoded binary.

> <item 2>
>    This fixes a problem I had, where build of gcc or newlib failed, because
> it failed to find the ranlib for target.

  I encountered this also.  I had previously built the RTEMS 4.0.0 tools,
so the first time I attempted to run the "bit" script, m68k-rtems-ranlib
(which was the old executable but in my search path) complained about
file types.  I presume this is due to the m68k COFF-to-ELF transition.

  Removing the old executables from my path (actually, just the symbolic
links to them from "/usr/local/bin") caused a different failure, as now
m68k-rtems-ranlib was not found.  In order to get the "bit" script to
run to completion without error, I had to first create symlinks from
the eventual path of the m68k-rtems executables (those which would be
built by the script itself) to one of the directories in my search

  I encountered another problem when running the "bit_rtems" script.
Before I could get the script to do much of anything, I had to create
a symbolic link in my tools directory from "rtems-4.5.0-beta3a" to
"4.5.0-beta3a".  There is apparently an incorrect variable assignment
somewhere in the script, however I did not take the time to track it
down.  It would seem unlikely that I'm the first to report this problem.

  Once the difficulties mentioned above were addressed, I was able to
successfully build the efi332 BSP.  I would have preferred to build
all of the m68k BSP's, to be more rigorous, but there was insufficient
space in the 1GB(!) partition I allocated for this process.

  On to the task of porting my existing BSP...

					-- MC --

More information about the users mailing list