[RFA]: Additional dead thread patch

[RFA]: Additional dead thread patch

[Jeff Johnston]

[-- Attachment #1: Type: text/plain, Size: 725 bytes --]

After updating to current sources today, I have noticed a regression in 
print-threads.exp which is due to some missing logic in my June 4th patch.  The 
problem is a timing issue (it does not occur for my June 4th build).  There is a 
gdb_assert in thread_from_lwp() which gets triggered because the th_valid flag 
has not been set for a zombie thread's thread_info struct.

This patch fills in the zombie thread's thread_info private area since the 
information is indeed valid and available.

Ok to commit?

-- Jeff J.

2004-06-07  Jeff Johnston  <jjohnstn@redhat.com>

         * thread-db.c (thread_get_info_callback): Fill in the thread_info
         struct if one exists, even if we are dealing with a zombie thread.


[-- Attachment #2: thread-db.patch --]
[-- Type: text/plain, Size: 1039 bytes --]

Index: thread-db.c
===================================================================
RCS file: /cvs/src/src/gdb/thread-db.c,v
retrieving revision 1.41
diff -u -p -r1.41 thread-db.c
--- thread-db.c	4 Jun 2004 21:28:15 -0000	1.41
+++ thread-db.c	7 Jun 2004 21:35:57 -0000
@@ -275,12 +275,18 @@ thread_get_info_callback (const td_thrha
   thread_info = find_thread_pid (thread_ptid);
 
   /* In the case of a zombie thread, don't continue.  We don't want to
-     attach to it thinking it is a new thread and we don't want to mark
-     it as valid.  */
+     attach to it thinking it is a new thread.  */
   if (ti.ti_state == TD_THR_UNKNOWN || ti.ti_state == TD_THR_ZOMBIE)
     {
       if (infop != NULL)
         *(struct thread_info **) infop = thread_info;
+      if (thread_info != NULL)
+	{
+	  memcpy (&thread_info->private->th, thp, sizeof (*thp));
+	  thread_info->private->th_valid = 1;
+	  memcpy (&thread_info->private->ti, &ti, sizeof (ti));
+	  thread_info->private->ti_valid = 1;
+	}
       return TD_THR_ZOMBIE;
     }
 

Re: [RFA]: Additional dead thread patch

[Daniel Jacobowitz]

On Mon, Jun 07, 2004 at 05:53:47PM -0400, Jeff Johnston wrote:
> After updating to current sources today, I have noticed a regression in 
> print-threads.exp which is due to some missing logic in my June 4th patch.  
> The problem is a timing issue (it does not occur for my June 4th build).  
> There is a gdb_assert in thread_from_lwp() which gets triggered because the 
> th_valid flag has not been set for a zombie thread's thread_info struct.
> 
> This patch fills in the zombie thread's thread_info private area since the 
> information is indeed valid and available.
> 
> Ok to commit?
> 
> -- Jeff J.
> 
> 2004-06-07  Jeff Johnston  <jjohnstn@redhat.com>
> 
>         * thread-db.c (thread_get_info_callback): Fill in the thread_info
>         struct if one exists, even if we are dealing with a zombie thread.
> 

Yes, this is OK.

-- 
Daniel Jacobowitz

Re: [RFA]: Additional dead thread patch

[Jeff Johnston]

Daniel Jacobowitz wrote:
> On Mon, Jun 07, 2004 at 05:53:47PM -0400, Jeff Johnston wrote:
> 
>>After updating to current sources today, I have noticed a regression in 
>>print-threads.exp which is due to some missing logic in my June 4th patch.  
>>The problem is a timing issue (it does not occur for my June 4th build).  
>>There is a gdb_assert in thread_from_lwp() which gets triggered because the 
>>th_valid flag has not been set for a zombie thread's thread_info struct.
>>
>>This patch fills in the zombie thread's thread_info private area since the 
>>information is indeed valid and available.
>>
>>Ok to commit?
>>
>>-- Jeff J.
>>
>>2004-06-07  Jeff Johnston  <jjohnstn@redhat.com>
>>
>>        * thread-db.c (thread_get_info_callback): Fill in the thread_info
>>        struct if one exists, even if we are dealing with a zombie thread.
>>
> 
> 
> Yes, this is OK.
>

Patch checked in.  Thanks.

-- Jeff J.