On 5/20/21 6:22 PM, Simon Marchi wrote:
> It's quite annoying that separate debug info files are represented by
> "objfile"s...
> 

Yeah, indeed.

And you could even say that that's fine, but question whether they
should be in the objfiles list, that is, have the default behaviour of:
...
   for (objfile *the_objfile : pspace->objfiles ())
     {
+      /* Skip separate debug objects.  */
+      if (the_objfile->separate_debug_objfile_backlink != nullptr)
+       continue;
...
without having to specify this, and if you need to access the separate
debuginfo files, use the specific iterator for this.

>> ...
>> but consequently we'll have two jit breakpoints, so we also make sure we don't
>> set a jit breakpoint on separate debug objects like libLLVM.so.10.debug.
>>
>> Tested on x86_64-linux.
> 
> Does that fix some test when running the testsuite with the fission
> board or something like that?

No, though I ran into another problem :) while playing around with the
jit test-cases and target boards  (
https://sourceware.org/bugzilla/show_bug.cgi?id=27893 ).  [ And yet
another problem while looking into the missing DIE problem (
https://sourceware.org/bugzilla/show_bug.cgi?id=27894 ).

>  I think it would be important for this to
> be tested.
> 

I wrote this target board file, which does trigger this assert in the
jit test-cases, and verified that the patch fixes all of them.

During testing of the target board, I ran into another assert in
gdb.python/py-objfile.exp, will file that as well.  WDYT?

Thanks,
- Tom