Using rtl-host in covor - some questions

[Chris Johns]

On 13/08/2014 7:03 pm, Krzysztof Mięsowicz wrote:
> Hi,
>
> I think we have small misunderstanding here :-) I'm currently trying to
> move symbol generation to covoar - previously symbol list was generated
> in shell script do_coverage.sh and passed to covoar in configuration
> file or as a command line argument. What we want now is to allow covoar
> to generate symbols. Chris suggested that use of code in rtl-host repo
> to manage symbols (it seems to use elf libraries inside), but this was
> said to be out of the scope of this project - I am going to do this
> after end of SOCIS. But as the first approximation we decided to invoke
> nm from covoar, using functions from rld-process[cpp, h] and parse its
> output in covoar. I have it almost done, but currently I added building
> rtems-syms from rtl-host as library and use it in covoar. I don't know
> if this is right solution and how should it be done better :-)

Yes this was the discussion and it still make sense if you want to 
follow this path.

On the other hand ... :) .... once you link to librld.a you can with a 
small amount of code open an ELF file and read the symbols into a symbol 
table and get at the data that way. This is how the rtems-ld works when 
linking against an RTEMS kernel base image.

Chris

>
>
> 2014-08-13 0:14 GMT+02:00 Joel Sherrill <joel.sherrill at oarcorp.com
> <mailto:joel.sherrill at oarcorp.com>>:
>
>
>     On 8/12/2014 6:58 AM, Krzysztof Mięsowicz wrote:
>      > Hi,
>      >
>      > I'm currently working on adding symbol generation to covoar. I'm
>     going
>      > to use rtl::process::execute function to run nm from covoar (as
>      > suggested by Chris). I do not know exactly how should I use this in
>      > covoar. Should I build rtl-host as a library and link it to
>     covoar? Or
>      > maybe there is another, better option?
>      >
>     covoar is in C++ and you would be invoking Python instead of nm from C++
>     and still producing something that the C++ has to read.
>
>     Ian Taylor suggested using nm over the elf libraries because the output
>     of nm was stable. But converting that code to use an elf reading library
>     directly would likely be a better solution.
>
>     If that's the use of nm you are talking about. :)
>
>     Waiting to hear from Chris.
>
>      > Thanks in advance for replies :)
>      >
>      > Regards,
>      > Krzysztof
>
>     --
>     Joel Sherrill, Ph.D.             Director of Research & Development
>     joel.sherrill at OARcorp.com        On-Line Applications Research
>     Ask me about RTEMS: a free RTOS  Huntsville AL 35805
>     Support Available (256) 722-9985 <tel:%28256%29%20722-9985>
>
>