Re: [PATCH v2] Remove support for thread events without PTRACE_EVENT_CLONE in GDB.

[Pedro Alves <palves at redhat dot com>]

On 12/10/2015 06:08 PM, Antoine Tremblay wrote:
> 
> 
> On 12/10/2015 12:59 PM, Pedro Alves wrote:
>> On 12/10/2015 05:57 PM, Antoine Tremblay wrote:
>>>
>>>
>>> On 12/10/2015 12:45 PM, Pedro Alves wrote:
>>>> On 12/10/2015 03:49 PM, Antoine Tremblay wrote:
>>>>
>>>>> Pushed in. (Without the dot at the end of the subject)
>>>>
>>>> This broke corethreads.exp though:
>>>>
>>>> Running /home/pedro/gdb/mygit/src/gdb/testsuite/gdb.threads/corethreads.exp ...
>>>> FAIL: gdb.threads/corethreads.exp: thread0 found
>>>> FAIL: gdb.threads/corethreads.exp: thread1 found
>>>>
>>>
>>> Right, somehow my buildbot did not see it :(
>>>
>>> I'm looking into it now.
>>>
>>
>> It's this hunk:
>>
>> @@ -1627,24 +1226,7 @@ find_new_threads_callback (const td_thrhandle_t *th_p, void *data)
>>     ptid = ptid_build (info->pid, ti.ti_lid, 0);
>>     tp = find_thread_ptid (ptid);
>>     if (tp == NULL || tp->priv == NULL)
>> -    {
>> -      if (attach_thread (ptid, th_p, &ti))
>> -       cb_data->new_threads += 1;
>> -      else
>> -       /* Problem attaching this thread; perhaps it exited before we
>> -          could attach it?
>> -          This could mean that the thread list inside glibc itself is in
>> -          inconsistent state, and libthread_db could go on looping forever
>> -          (observed with glibc-2.3.6).  To prevent that, terminate
>> -          iteration: thread_db_find_new_threads_2 will retry.  */
>> -       return 1;
>> -    }
>> -  else if (target_has_execution && !thread_db_use_events ())
>> -    {
>> -      /* Need to update this if not using the libthread_db events
>> -        (particularly, the TD_DEATH event).  */
>> -      update_thread_state (tp->priv, &ti);
>> -    }
>> +    thread_from_lwp (ptid);
>>
>>     return 0;
>>   }
>>
>> We can't use thread_from_lwp with core files.  As mentioned in a comment,
>> td_ta_map_lwp2thr uses ps_get_thread_area, but we can't use that
>> currently on core targets, as it uses ptrace directly.
>>
> Right, I think I can replace that directly with :
> 
>    if (tp == NULL || tp->priv == NULL)
>        record_thread (info, tp, ptid, th_p, &ti);
> 
> Sounds ok ? (it fixes the test)
> 

I think that's right.

> I'll run the full testsuite...

Thanks,
Pedro Alves