RE: [PATCH v5 03/15] gdb, gdbserver: support dlmopen()

["Metzger, Markus T" <markus.t.metzger@intel.com>]

Thanks for your review, Kevin,


>> diff --git a/gdb/solib-svr4.c b/gdb/solib-svr4.c
>> index 6222f79a7a0..1f2abcf16ef 100644
>[...]
>> @@ -1697,17 +1872,34 @@ solist_update_incremental (struct svr4_info *info,
>CORE_ADDR lm)
>>        if (!svr4_current_sos_via_xfer_libraries (&library_list, annex))
>>  	return 0;
>>
>> -      tail->next = library_list.head;
>> +      /* We expect gdbserver to provide updates for the namespace that
>> +	 contains LM, which whould be this namespace...  */
>> +      so_list *sos = nullptr;
>> +      if (library_list.solib_lists.find (debug_base)
>> +	  != library_list.solib_lists.end ())
>> +	std::swap (sos, library_list.solib_lists[debug_base]);
>> +      if (sos == nullptr)
>> +	{
>> +	  /* ...or for the special zero namespace for earlier versions...  */
>> +	  if (library_list.solib_lists.find (0)
>> +	      != library_list.solib_lists.end ())
>> +	    std::swap (sos, library_list.solib_lists[0]);
>[...]
>
>I was puzzled for a while by the uses of std:swap() in the above
>snippets of code.  I checked the rest of the GDB sources and found
>that (at least in GDB) this is not a common idiom.  I'd appreciate
>it if you could add a few lines of explanation for at least one
>of them.

Do those comments help?

diff --git a/gdb/solib-svr4.c b/gdb/solib-svr4.c
index 1f2abcf16ef..b231888b001 100644
--- a/gdb/solib-svr4.c
+++ b/gdb/solib-svr4.c
@@ -1346,6 +1346,10 @@ svr4_current_sos_direct (struct svr4_info *info)
          && (library_list.solib_lists[0] == nullptr))
        library_list.solib_lists.erase (0);
 
+      /* Replace the (empty) solib_lists in INFO with the one generated
+        from the target.  We don't want to copy it on assignment and then
+        delete the original afterwards, so let's just swap the
+        internals.  */
       std::swap (info->solib_lists, library_list.solib_lists);
       return;
     }
@@ -1872,7 +1876,11 @@ solist_update_incremental (svr4_info *info, CORE_ADDR debug_base,
       if (!svr4_current_sos_via_xfer_libraries (&library_list, annex))
        return 0;
 
-      /* We expect gdbserver to provide updates for the namespace that
+      /* Get the so list from the target.  We replace the list in the
+         target response so we can easily check that the response only
+         covers one namespace.
+
+        We expect gdbserver to provide updates for the namespace that
         contains LM, which whould be this namespace...  */
       so_list *sos = nullptr;
       if (library_list.solib_lists.find (debug_base)


>> diff --git a/gdb/testsuite/lib/gdb.exp b/gdb/testsuite/lib/gdb.exp
>> index 6aef39e5e02..98715fdc87e 100644
>> --- a/gdb/testsuite/lib/gdb.exp
>> +++ b/gdb/testsuite/lib/gdb.exp
>> @@ -2451,6 +2451,102 @@ proc skip_shlib_tests {} {
>>      return 1
>>  }
>>
>> +# Return 1 if we should skip dlmopen tests, 0 if we should not.
>> +
>> +proc skip_dlmopen_tests {} {
>> +    global srcdir subdir gdb_prompt inferior_exited_re
>> +
[...]
>> +    verbose "$me:  returning $skip_dlmopen_tests" 2
>> +    return $skip_dlmopen_tests
>> +}
>> +
>>  # Return 1 if we should skip tui related tests.
>
>I'm wondering if this can be a gdb_caching_proc?  At the moment, there's
>probably not much point since I think it's only used by the new test.
>But if we end up with more tests which use it, it might be worthwhile.

Makes sense.

thanks,
markus.
Intel Deutschland GmbH
Registered Address: Am Campeon 10, 85579 Neubiberg, Germany
Tel: +49 89 99 8853-0, www.intel.de <http://www.intel.de>
Managing Directors: Christin Eisenschmid, Sharon Heck, Tiffany Doon Silva  
Chairperson of the Supervisory Board: Nicole Lau
Registered Office: Munich
Commercial Register: Amtsgericht Muenchen HRB 186928