Dynamic Load Failure on Three MIPS BSPs

Chris Johns chrisj at rtems.org
Tue Oct 4 21:07:46 UTC 2016


On 05/10/2016 01:20, Joel Sherrill wrote:
> Well.. looking at the log, I think the answer is that rtems-syms is not
> processing the "-EL" option on the MIPS.

It looks to me like it is but rtems-syms is not handling it correctly.

> Thus the objects are always
> big endian on the MIPS but three BSPs use the -EL option.

>
> mips-rtems4.12-gcc -B../../../../../hurricane/lib/ -specs bsp_specs
> -qrtems -mips3 -G0 -EL -O2 -g -fomit-frame-pointer -ffunction-sections

I see the -EL here ...

> -fdata-sections -Wall -Wmissing-prototypes
> -Wimplicit-function-declaration -Wstrict-prototypes -Wnested-externs
>   -Wl,--gc-sections  -mips3 -G0 -EL   -o dl01.pre.exe init.o dl-load.o
> dl-tar.o
> mips-rtems4.12-objcopy -O srec dl01.pre.exe dl01.pre.srec1
> ../../../../../../tools/build/packhex < dl01.pre.srec1 > dl01.pre.ralf
> rm -f dl01.pre.srec1
> mips-rtems4.12-nm -g -n dl01.pre.exe > dl01.pre.num
> mips-rtems4.12-size dl01.pre.exe
>     text  data   bss   dec   hexfilename
>   167912 35392  9968213272 34118dl01.pre.exe
> rm -f dl01.pre.ralf
> mv dl01.pre.exe dl01.pre
> rtems-syms -e -c "-mips3 -G0 -EL -O2 -g -fomit-frame-pointer
             .... also here so RTEMS is OK.

This looks like a bug in rtems-syms.

I think rtemstoolkit/rld-cc.cpp needs to be taught about the flags being 
used here.

Chris


More information about the devel mailing list