Re: [PATCH 1/2] gdb: don't return an exited thread from any_non_exited_thread_of_inferior

[Simon Marchi <simon.marchi@efficios.com>]

On 2026-10-02 07:12, Matthieu Longo wrote:
> On 17/09/2026 21:16, Simon Marchi wrote:
>> any_non_exited_thread_of_inferior prefers to return the selected thread:
>>
>>   /* Prefer the current thread, if there's one.  */
>>   if (inf == current_inferior () && inferior_ptid != null_ptid)
>>     return inferior_thread ();
>>
>> ... but the selected thread can be an exited thread.  This can happen in
>> non-stop when the selected thread exits, in which case it is kept around
>> but marked as exited (see thread_info::deletable).
>>
>> Check the current thread's state before returning it, and fall through
>> to the existing loop otherwise.
>>
>> Note that we also have any_live_thread_of_inferior, which is similar but
>> differs in subtle ways.  We should probably check if we can merge them
>> at some point.
>>
>> Change-Id: I620ee200037c6a3fd3d59b96f16750d6b962c724
>> ---
>>  gdb/thread.c | 8 ++++++--
>>  1 file changed, 6 insertions(+), 2 deletions(-)
>>
>> diff --git a/gdb/thread.c b/gdb/thread.c
>> index 0f86f6efe380..e14254619e34 100644
>> --- a/gdb/thread.c
>> +++ b/gdb/thread.c
>> @@ -670,9 +670,13 @@ any_non_exited_thread_of_inferior (inferior *inf)
>>  {
>>    gdb_assert (inf->pid != 0);
>>  
>> -  /* Prefer the current thread, if there's one.  */
>> +  /* Prefer the current thread, if there's one and it hasn't exited.  */
>>    if (inf == current_inferior () && inferior_ptid != null_ptid)
>> -    return inferior_thread ();
>> +    {
>> +      if (thread_info *curr_thr = inferior_thread ();
>> +	  curr_thr->state () != THREAD_EXITED)
>> +	return curr_thr;
>> +    }
>>  
>>    for (thread_info &tp : inf->non_exited_threads ())
>>      return &tp;
>>
>> base-commit: 0595410d007e5f628e14717cd268e557e166675f
> 
> Hi Simon,
> 
> Given that there was no further feedback, would it be possible to merge this patch and the following
> one ? Or is it required to get an approval from another maintainer ?
> 
> Matthieu

Since it's been a couple of weeks and you tested it, I'll go ahead and
push it, thanks.

Simon