[PATCH] covoar: Add symbol set reader and ELF data parser to covoar.

[Chris Johns]

On 30/04/2018 15:38, Cillian O'Donnell wrote:
> On Sun, 29 Apr 2018, 23:30 Chris Johns, <chrisj at rtems.org
> <mailto:chrisj at rtems.org>> wrote:
> 
>     On 29/04/2018 20:50, Cillian O'Donnell wrote:
>     > On 28 April 2018 at 08:08, Chris Johns <chrisj at rtems.org
>     <mailto:chrisj at rtems.org>
>     > <mailto:chrisj at rtems.org <mailto:chrisj at rtems.org>>> wrote:
>     >     +      for (const std::string set : sets) {
>     >     +        if (verbose)
>     >     +          std::cerr << " Symbol set: " << set << std::endl;
>     >     +        const rld::config::section& set_section =
>     config.get_section(set);
>     >     +        rld::strings libs;
>     >     +        rld::config::parse_items (set_section, "libraries", libs, true);
>     >     +        for (std::string lib : libs) {
>     >     +          lib = rld::find_replace(lib, "@BUILD-TARGET@", buildTarget);
>     >     +          lib = rld::find_replace(lib, "@BSP@", buildBSP);
>     >
>     >
>     > Problem here. It tries to find the find the library but doesn't have the top
>     > build directory.
> 
>     Yes, this is correct. I have detected the build path in `main` but I decided to
>     not use it yet and I have not decided if it gets used at all.
> 
> 
> Ah ok, you were expecting this.
> 

Yes and I am sorry it fell out this way. I did not want to create more churn
than we needed.

> 
>     I have not looked at or worked on the qemu side of things, that is next. I am
>     not sure how this effects things so at this point in time you need to be at the
>     top of the build tree to work. This is not a bad thing, I think covoar is
>     leaving temporary files about the place.
> 
> 
> Yes the .cov files could be left lying around in the event of a crash. They are
> generated while the tests are run, so before covoar starts and then need to hang
> around for covoar's use later on. They probably need to be handled by the tester.
> 

I am OK with the .cov files being there, that is valid out just as I am OK with
the reports from covoar once generated, however internal files are not great. I
know there has been work on this in covoar but I think there are some other
places. I think they will be plugged when Joel and I move the tool to DWARF info.

> 
>     >     -  while ((opt = getopt(argc, argv, "1:L:e:c:g:E:f:s:T:O:p:v:d")) != -1) {
>     >     +  while ((opt = getopt(argc, argv, "1:L:e:c:g:E:f:s:S:T:O:p:vd")) !=
>     -1) {
>     >  
>     > Missing semicolon between v and d option.
> 
>     There is no argument to the -v option. With the ':' the option parsing get
>     stuffed up.
> 
> 
> Ok, my mistake.
> 

OK, I will push these changes soon and we can carry on from there.

First I want to get an updated libelf plus adding the elftoolchain libdwarf done
and pushed.

Chris