[OE-core] lots of WARNING: QA Issue: Bit size did not match (32 to 0) warnings, what do they mean?

[Mark Hatle]

On 3/3/11 9:45 AM, Koen Kooi wrote:
> Hi,
> 
> I started noticing a lot of warnings like this:
> 
> WARNING: QA Issue: Bit size did not match (32 to 0) on /work/armv7a-angstrom-linux-gnueabi/udev-165-r2/packages-split/udev/sbin/udevd

If the file is an ELF file, there are specific checks that are performed,
machine type, os type.. and bit size.  (I suspect the QA message for all of them
probably need to be improved.)

What I find interesting is it came back as "0".  0 indicates that the elf
comparison loader was never run.. 0 is "unknown".  (32 and 64 are obvious values...)

The code that determines the size is in meta/lib/oe/qa.py.  The normal use of
the code is from the meta/classes/insane.bbclass function.  The fact the
comparison returned a 0 indicates that the qa.py functions did not run.  This
could be a bug in meta/lib/oe/qa.py, or it could be that the routine call in
package_qa_walk (in insane.bbclass) failed for some reason and something went
wrong with the exception handler?

> What do they mean and how can I (help to) fix them?

First thing to do is check the binary that is being referenced and verify that
it is infact a valid ARM EABI binary.  Assuming it is file, we need to track
down the code likely starting in qa.py and figure out why it didn't determine
the bit size.

--Mark

> regards,
> 
> Koen
> _______________________________________________
> Openembedded-core mailing list
> Openembedded-core at lists.openembedded.org
> http://lists.linuxtogo.org/cgi-bin/mailman/listinfo/openembedded-core