Re: GCC toolchain with HW floating-point enabled

[Ilija Kocho <ilijak at siva dot com dot mk>]

Hi David.

We have a GCC (test) release with FPU support for Cortex-M4F in eCos
gnutools. You shall find information for download and installation here.
http://ecos.sourceware.org/ml/ecos-discuss/2012-06/msg00047.html

As this is still test release, your feedback will be appreciated.

Ilija

On 06.04.2013 12:23, Davide Pippa wrote:
> Hi!
>
> I've recently updated my ecos CVS, and made a new rom using the new hw
> floating point support (I'm compiling for TWR-K70F120M). The rom
> builds fine, but the application I used to work with complains during
> compilation:
>
> make all
> Building file: ../main.cpp
> Invoking: ARM devkitPro Windows GCC C++ Compiler
> arm-eabi-g++ -I"H:/Works/Arm/PyIde/roms/01009_TWR-K70F120M/include"
> -O0 -Wall -Wa,-adhlns="main.o.lst" -fno-exceptions -fno-rtti -c
> -fmessage-length=0 -MMD -MP -MF"main.d" -MT"main.d" -mcpu=cortex-m4
> -mthumb -mfloat-abi=hard -mfpu=vfpv4-d16 -g3 -gdwarf-2 -o "main.o"
> "../main.cpp"
> ../main.cpp: In function ‘void cyg_user_start()’:
> ../main.cpp:33:30: warning: deprecated conversion from string constant
> to ‘char*’ [-Wwrite-strings]
> ../main.cpp: In function ‘void main_thread_entry(cyg_addrword_t)’:
> ../main.cpp:80:87: warning: deprecated conversion from string constant
> to ‘char*’ [-Wwrite-strings]
> ../main.cpp: In function ‘void udpecho_thread_entry(void*)’:
> ../main.cpp:107:20: warning: unused variable ‘port’ [-Wunused-variable]
> ../main.cpp:110:9: warning: variable ‘j’ set but not used
> [-Wunused-but-set-variable]
> Finished building: ../main.cpp
>
> Building target: demo1.elf
> Invoking: ARM devkitPro Windows GCC C++ Linker
> arm-eabi-g++ -T"target.ld" -nostdlib
> -LH:/Works/Arm/PyIde/roms/01009_TWR-K70F120M/lib -Wl,-Map,demo1.map
> -mcpu=cortex-m4 -mthumb -mfloat-abi=hard -mfpu=vfpv4-d16 -g3 -gdwarf-2
> -o "demo1.elf"  ./main.o
> /cygdrive/h/Works/Arm/PyIde/tools/gnutools/arm-eabi/bin/../lib/gcc/arm-eabi/4.6.3/../../../../arm-eabi/bin/ld:
> error: demo1.elf uses VFP register arguments,
> /cygdrive/h/Works/Arm/PyIde/tools/gnutools/arm-eabi/bin/../lib/gcc/arm-eabi/4.6.3/libgcc.a(bpabi.o)
> does not
> makefile:59: recipe for target `demo1.elf' failed
> /cygdrive/h/Works/Arm/PyIde/tools/gnutools/arm-eabi/bin/../lib/gcc/arm-eabi/4.6.3/../../../../arm-eabi/bin/ld:
> failed to merge target specific data of file
> /cygdrive/h/Works/Arm/PyIde/tools/gnutools/arm-eabi/bin/../lib/gcc/arm-eabi/4.6.3/libgcc.a(bpabi.o)
> /cygdrive/h/Works/Arm/PyIde/tools/gnutools/arm-eabi/bin/../lib/gcc/arm-eabi/4.6.3/../../../../arm-eabi/bin/ld:
> error: demo1.elf uses VFP register arguments,
> /cygdrive/h/Works/Arm/PyIde/tools/gnutools/arm-eabi/bin/../lib/gcc/arm-eabi/4.6.3/libgcc.a(_divdi3.o)
> does not
> /cygdrive/h/Works/Arm/PyIde/tools/gnutools/arm-eabi/bin/../lib/gcc/arm-eabi/4.6.3/../../../../arm-eabi/bin/ld:
> failed to merge target specific data of file
> /cygdrive/h/Works/Arm/PyIde/tools/gnutools/arm-eabi/bin/../lib/gcc/arm-eabi/4.6.3/libgcc.a(_divdi3.o)
> /cygdrive/h/Works/Arm/PyIde/tools/gnutools/arm-eabi/bin/../lib/gcc/arm-eabi/4.6.3/../../../../arm-eabi/bin/ld:
> error: demo1.elf uses VFP register arguments,
> /cygdrive/h/Works/Arm/PyIde/tools/gnutools/arm-eabi/bin/../lib/gcc/arm-eabi/4.6.3/libgcc.a(_udivdi3.o)
> does not
> /cygdrive/h/Works/Arm/PyIde/tools/gnutools/arm-eabi/bin/../lib/gcc/arm-eabi/4.6.3/../../../../arm-eabi/bin/ld:
> failed to merge target specific data of file
> /cygdrive/h/Works/Arm/PyIde/tools/gnutools/arm-eabi/bin/../lib/gcc/arm-eabi/4.6.3/libgcc.a(_udivdi3.o)
> collect2: ld returned 1 exit status
> make: *** [demo1.elf] Error 1
>
> The application is built with HW fpu enabled, but it seems that
> libgcc.a is not floating-point aware (sort-of?)... is there any new
> release of the 4.6.3 gcc toolchain built with floating point support?
> Or is there any guide to rebuild the 4.6.3 gcc toolchain myself?
>
> Thanks!
>
> Pyper.
>


-- 
Before posting, please read the FAQ: http://ecos.sourceware.org/fom/ecos
and search the list archive: http://ecos.sourceware.org/ml/ecos-discuss