derick at perkinstechnologies.com.au
Thu May 1 13:06:53 UTC 2003
Thank you for letting me know about the -O2 option.
I compiled the same file with the -O2 option included in the command line
and it worked.
I then added the -O2 option to the CFLAGS variable in:
And successfully completed the build. I had a quick look at:
But didn't see anything obviously wrong.
Thanks for your quick response.
Chris Johns wrote:
>Derick Hammond wrote:
>>I have tried the following, as I too wish to build my BSP outside of the
>>RTEMS source tree:
>>$ ../rtems-4.6.0pre3/configure --enable-multilib
>>$ make RTEMS_BSP="" all
>>But I had the following error during the make process:
>>m68k-rtems-gcc --pipe -DPACKAGE_NAME=\"rtems-cpukit-libmisc\"
>>4.6.0pre3\" -DPACKAGE_BUGREPORT=\"rtems-bugs at OARcorp.com\" -isystem
>>../../../lib/include -g -Wall -o o-optimize/capture-cli.o -c
>This is similar to the error reported in PR292 in the RTEMS bug database,
>but the line number has moved and the target was the cpu32. This is the
>default m68k variant which I think is the 68020.
>What is strange is I cannot see the -O2 option. I noticed the compiler's
>standard build optimisation level is now 2, that is I had -O2 as a gcc
>option. The PR292 has no -O as well.
>>I note that there is a difference in the tools you used to compile RTEMS.
>>Is there a reason to use gcc-3.2.3 over gcc-3.2.2?
>GCC 3.2.3 may help, but I think we should find out what is happening with
>the -O option first.
>>Any hints or tips to diagnose this problem would be greatly appreciated.
>>Redhat 7.3 - kernel-2.4.18-26.7.x
>Should be fine.
>>RTEMS Packages installed:
>I have just 2.57 but this should matter.
>I have 1.7.4, but this should not be a problem.
>I have these.
>I have gcc 3.2.3 and newlib 1.11.0.
> Chris Johns, cjohns at cybertec.com.au
More information about the users