[Valgrind-developers] valgrind: r3825 - trunk/coregrind/m_scheduler

[<sv...@va...>]

Author: sewardj
Date: 2005-05-31 18:24:49 +0100 (Tue, 31 May 2005)
New Revision: 3825

Modified:
   trunk/coregrind/m_scheduler/scheduler.c
Log:
Completely zero out all the thread records at initialisation.  As far
as I can see, deallocate_LGDTs_for_thread were being called on
uninitialised thread records.



Modified: trunk/coregrind/m_scheduler/scheduler.c
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D
--- trunk/coregrind/m_scheduler/scheduler.c	2005-05-31 15:41:42 UTC (rev =
3824)
+++ trunk/coregrind/m_scheduler/scheduler.c	2005-05-31 17:24:49 UTC (rev =
3825)
@@ -632,8 +632,12 @@
    VG_(sema_init)(&run_sema);
=20
    for (i =3D 0 /* NB; not 1 */; i < VG_N_THREADS; i++) {
-      VG_(threads)[i].sig_queue            =3D NULL;
=20
+      /* Paranoia .. completely zero it out. */
+      VG_(memset)( & VG_(threads)[i], 0, sizeof( VG_(threads)[i] ) );
+
+      VG_(threads)[i].sig_queue =3D NULL;
+
       VGO_(os_state_init)(&VG_(threads)[i]);
       mostly_clear_thread_record(i);
=20