• State
  • Locked Locked
  • Replies 9 replies
  • Subscribers 106 subscribers
  • Views 739 views
  • Users 0 members are here
Support feedback
Options
Options
Similar topics

This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

hard float libWdrSupport_ti.a libGlbceSupport_ti.a libraries?

Stefano Antonelli
Intellectual 520 points

Hello,


We're trying to use the linaro 4.7 toolchain to compile the ipnc rdk with armhf so that we can use our debian root filesystem.  Everything compiles ok, but linking fails because of libWdrSupport_ti.a and libGlbceSupport_ti.a libraries.


Is there source code available for these libraries?  If not, can TI provide these libraries compiled with the linaro toolchain with hard float turned on?

Thanks.

  • 0
    TI__Guru* 80190 points

    Hello,

    I will notify the IPNC team to help here.

    Best Regards,

    Margarita

  • 0 in reply to Margarita Gashova
    TI__Prodigy 730 points

    Hi

    Could you please provide logs of linking errors.
     
    Regards,
    Mahesh
  • 0 in reply to Mahesha Mahadevappa
    Intellectual 430 points

    Here is the error messages:

    /opt/linaro/gcc-linaro-arm-linux-gnueabihf-4.7-2013.04-20130415_linux/bin/../lib/gcc/arm-linux-gnueabihf/4.7.3/../../../../arm-linux-gnueabihf/bin/ld: error: /home/david/build/8039/1.0.7/brown/products/grange/vidsrv/src/build/ipnc_rdk/../ipnc_rdk/ipnc_mcfw/bin/ti814x/bin//algoApp.out uses VFP register arguments, /home/david/build/8039/1.0.7/brown/products/grange/vidsrv/src/build/ipnc_rdk/../ipnc_rdk/ipnc_mcfw/mcfw/src_linux/links/alg/glbce/lib/libGlbceSupport_ti.a(TI_glbce.o) does not

    ...

    /opt/linaro/gcc-linaro-arm-linux-gnueabihf-4.7-2013.04-20130415_linux/bin/../lib/gcc/arm-linux-gnueabihf/4.7.3/../../../../arm-linux-gnueabihf/bin/ld: error: /home/david/build/8039/1.0.7/brown/products/grange/vidsrv/src/build/ipnc_rdk/../ipnc_rdk/ipnc_mcfw/bin/ti814x/bin//algoApp.out uses VFP register arguments, /home/david/build/8039/1.0.7/brown/products/grange/vidsrv/src/build/ipnc_rdk/../ipnc_rdk/ipnc_mcfw/mcfw/src_linux/links/alg/wdr/lib/libWdrSupport_ti.a(TI_glbce.o) does not

    ...

  • 0 in reply to David Liu1
    Prodigy 20 points
    I'm having the same error while trying to compile our code with a newer toolchain that has NEON hard floating points enabled. Anything you can do to help here would be greatly appreciated!

    Thanks,
    -Shawn
  • 0 in reply to Shawn Patterson
    TI__Guru**** 170625 points
    Hello,

    If you want to build the uImage (kernel) and modules, for DM81xx devices only software floating point toolchain is available:

    http://processors.wiki.ti.com/index.php/TI81XX_PSP_User_Guide#Toolchain

    If you want to build a debian package for the kernel (.deb file), armhf should be working. Note that Linaro armhf is supported for newer devices like Sitara, DRA7, OMAP5

    See also if the below links would be in help:

    http://omappedia.org/wiki/Ubuntu_kernel_for_OMAP4
    https://launchpad.net/linaro-toolchain-binaries

    BR
    Pavel
  • 0 in reply to Pavel Botev
    Intellectual 430 points

    Pavel,

    Thanks for the reply.  DM81xx has a normal ARM Cortex-A8 core with the capacity to do hard float.  We need this capacity for some algorithm running on our products and we have tested everything else with no any issue.  We do not need TI to support the toolchain as we've done it by ourselves.  What we ask is just to recompile the two binary static libraries, or release the source code so we can recompile.

    Regards,

    David

  • 0 in reply to David Liu1
    Prodigy 20 points

    Pavel,

    We are in the same situation as David.  We have the rest of our code base working with a compiler that supports hard floating points and we need to use hard floating for our application.  Any help would be greatly appreciated.

    Thanks!

    -Shawn

  • 0 in reply to David Liu1
    TI__Guru**** 170625 points

    David,

    David Liu1 said:
    What we ask is just to recompile the two binary static libraries, or release the source code so we can recompile.

    I would suggest you to check in our C/C++ compiler forum, if the team there have something to share.

    BR
    Pavel

  • 0 in reply to Pavel Botev
    Intellectual 430 points
    It might be possible to add some gcc options to bypass the link error for the stock application, but it won't help if we need to link other library compiled with different FP option, or run the executable on binary distro like debian armhf.

    We had the same issue on other modules, like cmem, and fixed them by recompiling the source code. It probably takes only a few minutes to compile the code for the 20+ C files in the two libraries. I can submit a Makefile if it helps.

    By the way, the new compiler can even detect some out of array boundary issues in mcfw_api files, which are ignored by old arago toolchain. There is really no harm but benefits to compile the source code with a different toolchain.