This is the mail archive of the gdb-patches@sourceware.org mailing list for the GDB project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

[PATCH] Fix PR threads/19354: "info threads" error with multiple inferiors


Note: this applies on top of:
 [PATCH] Remove support for LinuxThreads and vendor 2.4 kernels w/ backported NPTL
 https://sourceware.org/ml/gdb-patches/2015-12/msg00214.html

We try to avoid using libthread_db.so to list threads in the inferior
when debugging live processes, but the code that decides whether to
use it decides incorrectly if you have more than one inferior, and the
current inferior doesn't have execution yet.  The result is visible
as:

 (gdb) add-inferior
 Added inferior 2
 (gdb) inferior 2
 [Switching to inferior 2 [<null>] (<noexec>)]
 (gdb) info inferiors
   Num  Description       Executable
   1    process 15397     /home/pedro/gdb/tests/threads
 * 2    <null>
 (gdb) info threads
 Cannot find new threads: generic error
 (gdb)

Fix this by checking whether each inferior has execution rather than
just the current inferior.

By moving the core updating to linux-nat.c's update_thread_list
implementation, this also ends up fixing the
lwp-last-seen-running-on-core updating in the case we're debugging a
program that uses raw clone rather than pthreads, as linux-thread-db.c
isn't pushed in the target stack in that scenario.

Tested on x86_64 Fedora 20.

gdb/ChangeLog:
2015-12-11  Pedro Alves  <palves@redhat.com>

	PR threads/19354
	* linux-nat.c (linux_nat_update_thread_list): Update process cores
	each lwp was last seen running on here.
	* linux-thread-db.c (update_thread_core): Delete.
	(thread_db_update_thread_list_td_ta_thr_iter): Rename to ...
	(thread_db_update_thread_list): ... this.  Skip inferiors with
	execution.  Also call the target beneath.
	(thread_db_update_thread_list): Delete.

gdb/testsuite/ChangeLog:
2015-12-11  Pedro Alves  <palves@redhat.com>

	PR threads/19354
	* gdb.multi/info-threads.exp: New file.
---
 gdb/linux-nat.c                          |  7 +++++
 gdb/linux-thread-db.c                    | 53 ++++++++++++--------------------
 gdb/testsuite/gdb.multi/info-threads.exp | 39 +++++++++++++++++++++++
 3 files changed, 65 insertions(+), 34 deletions(-)
 create mode 100644 gdb/testsuite/gdb.multi/info-threads.exp

diff --git a/gdb/linux-nat.c b/gdb/linux-nat.c
index af1b764..1f9bb47 100644
--- a/gdb/linux-nat.c
+++ b/gdb/linux-nat.c
@@ -3719,10 +3719,17 @@ linux_nat_thread_alive (struct target_ops *ops, ptid_t ptid)
 static void
 linux_nat_update_thread_list (struct target_ops *ops)
 {
+  struct lwp_info *lwp;
+
   /* We add/delete threads from the list as clone/exit events are
      processed, so just try deleting exited threads still in the
      thread list.  */
   delete_exited_threads ();
+
+  /* Update the processor core that each lwp/thread was last seen
+     running on.  */
+  ALL_LWPS (lwp)
+    lwp->core = linux_common_core_of_thread (lwp->ptid);
 }
 
 static char *
diff --git a/gdb/linux-thread-db.c b/gdb/linux-thread-db.c
index ce3f6a1..68b6988 100644
--- a/gdb/linux-thread-db.c
+++ b/gdb/linux-thread-db.c
@@ -1324,17 +1324,11 @@ thread_db_find_new_threads_1 (ptid_t ptid)
   thread_db_find_new_threads_2 (ptid, 0);
 }
 
-static int
-update_thread_core (struct lwp_info *info, void *closure)
-{
-  info->core = linux_common_core_of_thread (info->ptid);
-  return 0;
-}
-
-/* Update the thread list using td_ta_thr_iter.  */
+/* Implement the to_update_thread_list target method for this
+   target.  */
 
 static void
-thread_db_update_thread_list_td_ta_thr_iter (struct target_ops *ops)
+thread_db_update_thread_list (struct target_ops *ops)
 {
   struct thread_db_info *info;
   struct inferior *inf;
@@ -1356,34 +1350,25 @@ thread_db_update_thread_list_td_ta_thr_iter (struct target_ops *ops)
       if (thread == NULL || thread->executing)
 	continue;
 
+      /* It's best to avoid td_ta_thr_iter if possible.  That walks
+	 data structures in the inferior's address space that may be
+	 corrupted, or, if the target is running, the list may change
+	 while we walk it.  In the latter case, it's possible that a
+	 thread exits just at the exact time that causes GDB to get
+	 stuck in an infinite loop.  To avoid pausing all threads
+	 whenever the core wants to refresh the thread list, we
+	 instead use thread_from_lwp immediately when we see an LWP
+	 stop.  That uses thread_db entry points that do not walk
+	 libpthread's thread list, so should be safe, as well as more
+	 efficient.  */
+      if (target_has_execution_1 (thread->ptid))
+	continue;
+
       thread_db_find_new_threads_1 (thread->ptid);
     }
-}
 
-/* Implement the to_update_thread_list target method for this
-   target.  */
-
-static void
-thread_db_update_thread_list (struct target_ops *ops)
-{
-  /* It's best to avoid td_ta_thr_iter if possible.  That walks data
-     structures in the inferior's address space that may be corrupted,
-     or, if the target is running, the list may change while we walk
-     it.  In the latter case, it's possible that a thread exits just
-     at the exact time that causes GDB to get stuck in an infinite
-     loop.  To avoid pausing all threads whenever the core wants to
-     refresh the thread list, use thread_from_lwp immediately when we
-     see an LWP stop.  That uses  thread_db entry points that do not
-     walk libpthread's thread list, so should be safe, as well as
-     more efficient.  */
-  if (target_has_execution)
-    ops->beneath->to_update_thread_list (ops->beneath);
-  else
-    thread_db_update_thread_list_td_ta_thr_iter (ops);
-
-  if (target_has_execution)
-    iterate_over_lwps (minus_one_ptid /* iterate over all */,
-		       update_thread_core, NULL);
+  /* Give the beneath target a chance to do extra processing.  */
+  ops->beneath->to_update_thread_list (ops->beneath);
 }
 
 static char *
diff --git a/gdb/testsuite/gdb.multi/info-threads.exp b/gdb/testsuite/gdb.multi/info-threads.exp
new file mode 100644
index 0000000..c00b8d7
--- /dev/null
+++ b/gdb/testsuite/gdb.multi/info-threads.exp
@@ -0,0 +1,39 @@
+# Copyright 2009-2015 Free Software Foundation, Inc.
+
+# This program is free software; you can redistribute it and/or modify
+# it under the terms of the GNU General Public License as published by
+# the Free Software Foundation; either version 3 of the License, or
+# (at your option) any later version.
+#
+# This program is distributed in the hope that it will be useful,
+# but WITHOUT ANY WARRANTY; without even the implied warranty of
+# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+# GNU General Public License for more details.
+#
+# You should have received a copy of the GNU General Public License
+# along with this program.  If not, see <http://www.gnu.org/licenses/>.
+
+# Regression test for PR threads/19354.  If you have a threaded
+# inferior running (thread_db must be loaded), switch to another
+# inferior, one that is NOT running yet, and do "info threads", you
+# would get back an error.
+
+standard_testfile hello.c
+
+if { [prepare_for_testing "failed to prepare" ${testfile} ${srcfile} {pthreads debug}] } {
+    return -1
+}
+
+clean_restart ${testfile}
+
+if { ![runto_main] } then {
+    return -1
+}
+
+# Add another inferior.
+gdb_test "add-inferior" "Added inferior 2.*" "add empty inferior 2"
+gdb_test "inferior 2" "Switching to inferior 2.*" "switch to inferior 2"
+
+# "info threads" while inferior 1 has execution and inferior 2 is not
+# running yet should show inferior 1's thread, and give no error.
+gdb_test "info threads" "1    .* main .* at .*$srcfile:.*No selected thread.*"
-- 
1.9.3


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]