[PATCH] gdb: change thread_info::name to unique_xmalloc_ptr, add helper function

[PATCH] gdb: change thread_info::name to unique_xmalloc_ptr, add helper function

[Simon Marchi]

From: Simon Marchi <simon.marchi@polymtl.ca>

This started out as changing thread_info::name to a unique_xmalloc_ptr.
That showed that almost all users of that field had the same logic to
get a thread's name: use thread_info::name if non-nullptr, else ask the
target.  Factor out this logic in a new thread_name free function.  Make
the field private (rename to m_name) and add some accessors.

Change-Id: Iebdd95f4cd21fbefc505249bd1d05befc466a2fc
---
 gdb/breakpoint.c          |  3 +--
 gdb/gdbthread.h           | 30 ++++++++++++++++++++++++++----
 gdb/infrun.c              |  4 +---
 gdb/python/py-infthread.c |  9 ++-------
 gdb/thread.c              | 36 +++++++++++++++++++++++-------------
 5 files changed, 53 insertions(+), 29 deletions(-)

diff --git a/gdb/breakpoint.c b/gdb/breakpoint.c
index f6c9683aecf..aab1c50acdc 100644
--- a/gdb/breakpoint.c
+++ b/gdb/breakpoint.c
@@ -4565,13 +4565,12 @@ maybe_print_thread_hit_breakpoint (struct ui_out *uiout)
 
   if (show_thread_that_caused_stop ())
     {
-      const char *name;
       struct thread_info *thr = inferior_thread ();
 
       uiout->text ("Thread ");
       uiout->field_string ("thread-id", print_thread_id (thr));
 
-      name = thr->name != NULL ? thr->name : target_thread_name (thr);
+      const char *name = thread_name (thr);
       if (name != NULL)
 	{
 	  uiout->text (" \"");
diff --git a/gdb/gdbthread.h b/gdb/gdbthread.h
index 9c178f531d9..2990d8a9b28 100644
--- a/gdb/gdbthread.h
+++ b/gdb/gdbthread.h
@@ -232,7 +232,6 @@ class thread_info : public refcounted_object,
 {
 public:
   explicit thread_info (inferior *inf, ptid_t ptid);
-  ~thread_info ();
 
   bool deletable () const;
 
@@ -283,9 +282,21 @@ class thread_info : public refcounted_object,
   /* The inferior this thread belongs to.  */
   struct inferior *inf;
 
-  /* The name of the thread, as specified by the user.  This is NULL
-     if the thread does not have a user-given name.  */
-  char *name = NULL;
+  /* The user-given name of the thread.
+
+     Returns nullptr if the thread does not have a user-given name.  */
+  const char *name () const
+  {
+    return m_name.get ();
+  }
+
+  /* Set the user-given name of the thread.
+
+     Pass nullptr to clear the name.  */
+  void set_name (gdb::unique_xmalloc_ptr<char> name)
+  {
+    m_name = std::move (name);
+  }
 
   /* True means the thread is executing.  Note: this is different
      from saying that there is an active target and we are stopped at
@@ -491,6 +502,11 @@ class thread_info : public refcounted_object,
   /* State of inferior thread to restore after GDB is done with an inferior
      call.  See `struct thread_suspend_state'.  */
   thread_suspend_state m_suspend;
+
+  /* The user-give name of the thread.
+
+     Nullptr if the thread does not have a user-given name.  */
+  gdb::unique_xmalloc_ptr<char> m_name;
 };
 
 using thread_info_resumed_with_pending_wait_status_node
@@ -921,4 +937,10 @@ extern void print_selected_thread_frame (struct ui_out *uiout,
    alive anymore.  */
 extern void thread_select (const char *tidstr, class thread_info *thr);
 
+/* Return THREAD's name.
+
+   If THREAD has a user-given name, return it.  Otherwise, query the thread's
+   target to get the name.  May return nullptr.  */
+extern const char *thread_name (thread_info *thread);
+
 #endif /* GDBTHREAD_H */
diff --git a/gdb/infrun.c b/gdb/infrun.c
index 694bbe665f4..877c3f02b66 100644
--- a/gdb/infrun.c
+++ b/gdb/infrun.c
@@ -8186,12 +8186,10 @@ print_signal_received_reason (struct ui_out *uiout, enum gdb_signal siggnal)
     ;
   else if (show_thread_that_caused_stop ())
     {
-      const char *name;
-
       uiout->text ("\nThread ");
       uiout->field_string ("thread-id", print_thread_id (thr));
 
-      name = thr->name != NULL ? thr->name : target_thread_name (thr);
+      const char *name = thread_name (thr);
       if (name != NULL)
 	{
 	  uiout->text (" \"");
diff --git a/gdb/python/py-infthread.c b/gdb/python/py-infthread.c
index 3315af5eb59..f43e175f62c 100644
--- a/gdb/python/py-infthread.c
+++ b/gdb/python/py-infthread.c
@@ -66,14 +66,10 @@ static PyObject *
 thpy_get_name (PyObject *self, void *ignore)
 {
   thread_object *thread_obj = (thread_object *) self;
-  const char *name;
 
   THPY_REQUIRE_VALID (thread_obj);
 
-  name = thread_obj->thread->name;
-  if (name == NULL)
-    name = target_thread_name (thread_obj->thread);
-
+  const char *name = thread_name (thread_obj->thread);
   if (name == NULL)
     Py_RETURN_NONE;
 
@@ -115,8 +111,7 @@ thpy_set_name (PyObject *self, PyObject *newvalue, void *ignore)
 	return -1;
     }
 
-  xfree (thread_obj->thread->name);
-  thread_obj->thread->name = name.release ();
+  thread_obj->thread->set_name (std::move (name));
 
   return 0;
 }
diff --git a/gdb/thread.c b/gdb/thread.c
index a82fb49140a..f8e6a64eb1e 100644
--- a/gdb/thread.c
+++ b/gdb/thread.c
@@ -302,11 +302,6 @@ thread_info::thread_info (struct inferior *inf_, ptid_t ptid_)
   this->m_suspend.waitstatus.kind = TARGET_WAITKIND_IGNORE;
 }
 
-thread_info::~thread_info ()
-{
-  xfree (this->name);
-}
-
 /* See gdbthread.h.  */
 
 bool
@@ -1001,7 +996,7 @@ thread_target_id_str (thread_info *tp)
 {
   std::string target_id = target_pid_to_str (tp->ptid);
   const char *extra_info = target_extra_thread_info (tp);
-  const char *name = tp->name != nullptr ? tp->name : target_thread_name (tp);
+  const char *name = thread_name (tp);
 
   if (extra_info != nullptr && name != nullptr)
     return string_printf ("%s \"%s\" (%s)", target_id.c_str (), name,
@@ -1143,9 +1138,7 @@ print_thread_info_1 (struct ui_out *uiout, const char *requested_threads,
 	      if (extra_info != nullptr)
 		uiout->field_string ("details", extra_info);
 
-	      const char *name = (tp->name != nullptr
-				  ? tp->name
-				  : target_thread_name (tp));
+	      const char *name = thread_name (tp);
 	      if (name != NULL)
 		uiout->field_string ("name", name);
 	    }
@@ -1838,8 +1831,7 @@ thread_name_command (const char *arg, int from_tty)
   arg = skip_spaces (arg);
 
   info = inferior_thread ();
-  xfree (info->name);
-  info->name = arg ? xstrdup (arg) : NULL;
+  info->set_name (arg != nullptr ? make_unique_xstrdup (arg) : nullptr);
 }
 
 /* Find thread ids with a name, target pid, or extra info matching ARG.  */
@@ -1866,10 +1858,10 @@ thread_find_command (const char *arg, int from_tty)
     {
       switch_to_inferior_no_thread (tp->inf);
 
-      if (tp->name != NULL && re_exec (tp->name))
+      if (tp->name () != nullptr && re_exec (tp->name ()))
 	{
 	  printf_filtered (_("Thread %s has name '%s'\n"),
-			   print_thread_id (tp), tp->name);
+			   print_thread_id (tp), tp->name ());
 	  match++;
 	}
 
@@ -2013,6 +2005,24 @@ update_thread_list (void)
   update_threads_executing ();
 }
 
+/* See gdbthread.h.  */
+
+const char *
+thread_name (thread_info *thread)
+{
+  /* Use the manually set name if there is one.  */
+  const char *name = thread->name ();
+  if (name != nullptr)
+    return name;
+
+  /* Otherwise, ask the target.  Ensure we query the right target stack.  */
+  scoped_restore_current_thread restore_thread;
+  if (thread->inf != current_inferior ())
+    switch_to_inferior_no_thread (thread->inf);
+
+  return target_thread_name (thread);
+}
+
 /* Return a new value for the selected thread's id.  Return a value of
    0 if no thread is selected.  If GLOBAL is true, return the thread's
    global number.  Otherwise return the per-inferior number.  */
-- 
2.33.0

Re: [PATCH] gdb: change thread_info::name to unique_xmalloc_ptr, add helper function

[Tom Tromey]

>>>>> "Simon" == Simon Marchi via Gdb-patches <gdb-patches@sourceware.org> writes:

Simon> This started out as changing thread_info::name to a unique_xmalloc_ptr.
Simon> That showed that almost all users of that field had the same logic to
Simon> get a thread's name: use thread_info::name if non-nullptr, else ask the
Simon> target.  Factor out this logic in a new thread_name free function.  Make
Simon> the field private (rename to m_name) and add some accessors.

Thank you.

Simon> +  /* The user-give name of the thread.

Missing an "n" on "given".

Simon> +/* Return THREAD's name.
Simon> +
Simon> +   If THREAD has a user-given name, return it.  Otherwise, query the thread's
Simon> +   target to get the name.  May return nullptr.  */
Simon> +extern const char *thread_name (thread_info *thread);

It seems to me that this should probably be the only caller of
target_thread_name.  So, what about putting the extra logic into
target_thread_name instead of having a new function?

One idea here may be to keep target_thread_name as a simple wrapper for
the target method.  But in that case, I think target_thread_name's
declaration should be updated to say something like "don't call this,
use thread_name instead".

Other than this issue, this all looks great to me.

Tom

Re: [PATCH] gdb: change thread_info::name to unique_xmalloc_ptr, add helper function

[Simon Marchi]

On 2021-09-23 14:15, Tom Tromey wrote:
>>>>>> "Simon" == Simon Marchi via Gdb-patches <gdb-patches@sourceware.org> writes:
> 
> Simon> This started out as changing thread_info::name to a unique_xmalloc_ptr.
> Simon> That showed that almost all users of that field had the same logic to
> Simon> get a thread's name: use thread_info::name if non-nullptr, else ask the
> Simon> target.  Factor out this logic in a new thread_name free function.  Make
> Simon> the field private (rename to m_name) and add some accessors.
> 
> Thank you.
> 
> Simon> +  /* The user-give name of the thread.
> 
> Missing an "n" on "given".

Fixed.

> 
> Simon> +/* Return THREAD's name.
> Simon> +
> Simon> +   If THREAD has a user-given name, return it.  Otherwise, query the thread's
> Simon> +   target to get the name.  May return nullptr.  */
> Simon> +extern const char *thread_name (thread_info *thread);
> 
> It seems to me that this should probably be the only caller of
> target_thread_name.  So, what about putting the extra logic into
> target_thread_name instead of having a new function?
> 
> One idea here may be to keep target_thread_name as a simple wrapper for
> the target method.  But in that case, I think target_thread_name's
> declaration should be updated to say something like "don't call this,
> use thread_name instead".

I prefer the latter, keeping target_thread_name a simple wrapper.  I
added this to target_thread_name's doc:

   You likely don't want to call this function, but use the thread_name
   function instead, which prefers the user-given thread name, if set.  */

> Other than this issue, this all looks great to me.

Thanks, I'll push the patch with those fixed.

Simon