Fix handling of static TLS in dlopen'ed objects

When dynamically loading a library along with several dependencies, calls to
_dl_add_to_slotinfo and _dl_update_slotinfo can become intermixed. As a
consequence, _dl_update_slotinfo will update the generation counter of the dtv
although not all of the slots belonging to that generation have been added.
Subsequent calls to _dl_add_to_slotinfo will add more slots to the same
generation, for which no storage will be allocated, as the dtv generation
checks will claim no work is necessary. This will lead to uninitialized dtv
entries and will likely cause a SIGSEGV when thread local variables are
accessed.

8 files changed, 132 insertions, 30 deletions

diff --git a/ChangeLog b/ChangeLog
index 26c431771a..a11ee6cc21 100644
--- a/ChangeLog
+++ b/ChangeLog
@@ -1,5 +1,15 @@
 2011-05-14  Ulrich Drepper  <drepper@gmail.com>
 
+	[BZ #12453]
+	* elf/dl-open.c (dl_open_worker): Delay calls to _dl_update_slotinfo
+	until all modules are registered in the DTV.
+	* elf/Makefile: Add rules to build and run tst-tls19.
+	* elf/tst-tls19.c: New file.
+	* elf/tst-tls19mod1.c: New file.
+	* elf/tst-tls19mod2.c: New file.
+	* elf/tst-tls19mod3.c: New file.
+	Patch mostly by Martin von Gagern <Martin.vGagern@gmx.net>.
+
 	[BZ #12083]
 	* sysdeps/pthread/aio_misc.c (__aio_init): Compute optim.aio_num
 	correctly.
diff --git a/NEWS b/NEWS
index 4214c3c426..1c5d020eed 100644
--- a/NEWS
+++ b/NEWS
@@ -11,10 +11,10 @@ Version 2.14
 
   386, 11257, 11258, 11487, 11532, 11578, 11653, 11668, 11724, 11945, 11947,
   12052, 12083, 12158, 12178, 12200, 12346, 12393, 12420, 12432, 12445,
-  12449, 12454, 12460, 12469, 12489, 12509, 12510, 12511, 12518, 12527,
-  12541, 12545, 12551, 12582, 12583, 12587, 12597, 12601, 12611, 12625,
-  12626, 12631, 12650, 12653, 12655, 12660, 12681, 12685, 12711, 12713,
-  12714, 12717, 12723, 12724, 12734, 12738
+  12449, 12453, 12454, 12460, 12469, 12489, 12509, 12510, 12511, 12518,
+  12527, 12541, 12545, 12551, 12582, 12583, 12587, 12597, 12601, 12611,
+  12625, 12626, 12631, 12650, 12653, 12655, 12660, 12681, 12685, 12711,
+  12713, 12714, 12717, 12723, 12724, 12734, 12738
 
 * The RPC implementation in libc is obsoleted.  Old programs keep working
   but new programs cannot be linked with the routines in libc anymore.
diff --git a/elf/Makefile b/elf/Makefile
index 8d9657d4c2..6efb86c7e9 100644
--- a/elf/Makefile
+++ b/elf/Makefile
@@ -76,6 +76,7 @@ distribute	:= rtld-Rules \
 		   tst-tlsmod12.c tst-tls10.h tst-alignmod.c tst-alignmod2.c \
 		   circlemod1.c circlemod1a.c circlemod2.c circlemod2a.c \
 		   circlemod3.c circlemod3a.c nodlopenmod2.c \
+		   tst-tls19mod1.c tst-tls19mod2.c tst-tls19mod3.c \
 		   tls-macros.h \
 		   reldep8mod1.c reldep8mod2.c reldep8mod3.c \
 		   nodel2mod1.c nodel2mod2.c nodel2mod3.c \
@@ -194,7 +195,7 @@ tests += loadtest restest1 preloadtest loadfail multiload origtest resolvfail \
 	 restest2 next dblload dblunload reldep5 reldep6 reldep7 reldep8 \
 	 circleload1 tst-tls3 tst-tls4 tst-tls5 tst-tls6 tst-tls7 tst-tls8 \
 	 tst-tls10 tst-tls11 tst-tls12 tst-tls13 tst-tls14 tst-tls15 \
-	 tst-tls16 tst-tls17 tst-tls18 tst-tls-dlinfo \
+	 tst-tls16 tst-tls17 tst-tls18 tst-tls19 tst-tls-dlinfo \
 	 tst-align tst-align2 $(tests-execstack-$(have-z-execstack)) \
 	 tst-dlmodcount tst-dlopenrpath tst-deep1 \
 	 tst-dlmopen1 tst-dlmopen2 tst-dlmopen3 \
@@ -240,6 +241,7 @@ modules-names = testobj1 testobj2 testobj3 testobj4 testobj5 testobj6 \
 		tst-tlsmod13 tst-tlsmod13a tst-tlsmod14a tst-tlsmod14b \
 		tst-tlsmod15a tst-tlsmod15b tst-tlsmod16a tst-tlsmod16b \
 		$(tlsmod17a-modules) tst-tlsmod17b $(tlsmod18a-modules) \
+		tst-tls19mod1 tst-tls19mod2 tst-tls19mod3 \
 		circlemod1 circlemod1a circlemod2 circlemod2a \
 		circlemod3 circlemod3a \
 		reldep8mod1 reldep8mod2 reldep8mod3 \
@@ -525,6 +527,8 @@ $(objpfx)tst-tlsmod13a.so: $(objpfx)tst-tlsmod13.so
 # For tst-tls9-static, make sure the modules it dlopens have libc.so in DT_NEEDED
 $(objpfx)tst-tlsmod5.so: $(common-objpfx)libc.so
 $(objpfx)tst-tlsmod6.so: $(common-objpfx)libc.so
+$(objpfx)tst-tls19mod1.so: $(objpfx)tst-tls19mod2.so $(objpfx)tst-tls19mod3.so
+$(objpfx)tst-tls19mod3.so: $(objpfx)ld.so
 $(objpfx)reldep8mod3.so: $(objpfx)reldep8mod1.so $(objpfx)reldep8mod2.so
 $(objpfx)nodel2mod3.so: $(objpfx)nodel2mod1.so $(objpfx)nodel2mod2.so
 $(objpfx)reldep9mod2.so: $(objpfx)reldep9mod1.so
@@ -822,6 +826,9 @@ $(patsubst %,$(objpfx)%.os,$(tlsmod18a-modules)): $(objpfx)tst-tlsmod18a%.os : t
 	$(compile-command.c) -DN=$*
 $(patsubst %,$(objpfx)%.so,$(tlsmod18a-modules)): $(objpfx)tst-tlsmod18a%.so: $(objpfx)ld.so
 
+$(objpfx)tst-tls19: $(libdl)
+$(objpfx)tst-tls19.out: $(objpfx)tst-tls19mod1.so
+
 CFLAGS-tst-align.c = $(stack-align-test-flags)
 CFLAGS-tst-align2.c = $(stack-align-test-flags)
 CFLAGS-tst-alignmod.c = $(stack-align-test-flags)
diff --git a/elf/dl-open.c b/elf/dl-open.c
index cf8e8cc671..8d90b56dca 100644
--- a/elf/dl-open.c
+++ b/elf/dl-open.c
@@ -1,5 +1,5 @@
 /* Load a shared object at runtime, relocate it, and run its initializer.
-   Copyright (C) 1996-2007, 2009, 2010 Free Software Foundation, Inc.
+   Copyright (C) 1996-2007, 2009, 2010, 2011 Free Software Foundation, Inc.
    This file is part of the GNU C Library.
 
    The GNU C Library is free software; you can redistribute it and/or
@@ -347,6 +347,7 @@ dl_open_worker (void *a)
   /* If the file is not loaded now as a dependency, add the search
      list of the newly loaded object to the scope.  */
   bool any_tls = false;
+  unsigned int first_static_tls = new->l_searchlist.r_nlist;
   for (unsigned int i = 0; i < new->l_searchlist.r_nlist; ++i)
     {
       struct link_map *imap = new->l_searchlist.r_list[i];
@@ -425,30 +426,9 @@ dl_open_worker (void *a)
 	     might have to increase its size.  */
 	  _dl_add_to_slotinfo (imap);
 
-	  if (imap->l_need_tls_init)
-	    {
-	      /* For static TLS we have to allocate the memory here
-		 and now.  This includes allocating memory in the DTV.
-		 But we cannot change any DTV other than our own. So,
-		 if we cannot guarantee that there is room in the DTV
-		 we don't even try it and fail the load.
-
-		 XXX We could track the minimum DTV slots allocated in
-		 all threads.  */
-	      if (! RTLD_SINGLE_THREAD_P && imap->l_tls_modid > DTV_SURPLUS)
-		_dl_signal_error (0, "dlopen", NULL, N_("\
-cannot load any more object with static TLS"));
-
-	      imap->l_need_tls_init = 0;
-#ifdef SHARED
-	      /* Update the slot information data for at least the
-		 generation of the DSO we are allocating data for.  */
-	      _dl_update_slotinfo (imap->l_tls_modid);
-#endif
-
-	      GL(dl_init_static_tls) (imap);
-	      assert (imap->l_need_tls_init == 0);
-	    }
+	  if (imap->l_need_tls_init
+	      && first_static_tls == new->l_searchlist.r_nlist)
+	    first_static_tls = i;
 
 	  /* We have to bump the generation counter.  */
 	  any_tls = true;
@@ -460,6 +440,40 @@ cannot load any more object with static TLS"));
     _dl_fatal_printf (N_("\
 TLS generation counter wrapped!  Please report this."));
 
+  /* We need a second pass for static tls data, because _dl_update_slotinfo
+     must not be run while calls to _dl_add_to_slotinfo are still pending. */
+  for (unsigned int i = first_static_tls; i < new->l_searchlist.r_nlist; ++i)
+    {
+      struct link_map *imap = new->l_searchlist.r_list[i];
+
+      if (imap->l_need_tls_init
+	  && ! imap->l_init_called
+	  && imap->l_tls_blocksize > 0)
+	{
+	  /* For static TLS we have to allocate the memory here and
+	     now.  This includes allocating memory in the DTV.  But we
+	     cannot change any DTV other than our own. So, if we
+	     cannot guarantee that there is room in the DTV we don't
+	     even try it and fail the load.
+
+	     XXX We could track the minimum DTV slots allocated in
+	     all threads.  */
+	  if (! RTLD_SINGLE_THREAD_P && imap->l_tls_modid > DTV_SURPLUS)
+	    _dl_signal_error (0, "dlopen", NULL, N_("\
+cannot load any more object with static TLS"));
+
+	  imap->l_need_tls_init = 0;
+#ifdef SHARED
+	  /* Update the slot information data for at least the
+	     generation of the DSO we are allocating data for.  */
+	  _dl_update_slotinfo (imap->l_tls_modid);
+#endif
+
+	  GL(dl_init_static_tls) (imap);
+	  assert (imap->l_need_tls_init == 0);
+	}
+    }
+
   /* Run the initializer functions of new objects.  */
   _dl_init (new, args->argc, args->argv, args->env);
 
diff --git a/elf/tst-tls19.c b/elf/tst-tls19.c
new file mode 100644
index 0000000000..acbc1d6964
--- /dev/null
+++ b/elf/tst-tls19.c
@@ -0,0 +1,27 @@
+// BZ 12453
+#include <stdio.h>
+#include <dlfcn.h>
+
+
+static int
+do_test (void)
+{
+  void* dl = dlopen ("tst-tls19mod1.so", RTLD_LAZY | RTLD_GLOBAL);
+  if (dl == NULL)
+    {
+      printf ("Error loading tst-tls19mod1.so: %s\n", dlerror ());
+      return 1;
+    }
+
+  int (*fn) (void) = dlsym (dl, "foo");
+  if (fn == NULL)
+    {
+      printf("Error obtaining symbol foo\n");
+      return 1;
+    }
+
+  return fn ();
+}
+
+#define TEST_FUNCTION do_test ()
+#include "../test-skeleton.c"
diff --git a/elf/tst-tls19mod1.c b/elf/tst-tls19mod1.c
new file mode 100644
index 0000000000..2790097ae5
--- /dev/null
+++ b/elf/tst-tls19mod1.c
@@ -0,0 +1,15 @@
+#include <stdio.h>
+
+extern int bar (void);
+extern int baz (void);
+
+int
+foo (void)
+{
+  int v1 = bar ();
+  int v2 = baz ();
+
+  printf ("bar=%d, baz=%d\n", v1, v2);
+
+  return v1 != 666 || v2 != 42;
+}
diff --git a/elf/tst-tls19mod2.c b/elf/tst-tls19mod2.c
new file mode 100644
index 0000000000..cae702f67c
--- /dev/null
+++ b/elf/tst-tls19mod2.c
@@ -0,0 +1,13 @@
+static int __thread tbar __attribute__ ((tls_model ("initial-exec"))) = 666;
+
+void
+setter (int a)
+{
+  tbar = a;
+}
+
+int
+bar (void)
+{
+  return tbar;
+}
diff --git a/elf/tst-tls19mod3.c b/elf/tst-tls19mod3.c
new file mode 100644
index 0000000000..e7b28016b3
--- /dev/null
+++ b/elf/tst-tls19mod3.c
@@ -0,0 +1,16 @@
+#include <stdio.h>
+
+static int __thread tbaz __attribute__ ((tls_model ("local-dynamic"))) = 42;
+
+void
+setter2 (int a)
+{
+  tbaz = a;
+}
+
+int
+baz (void)
+{
+  printf ("&tbaz=%p\n", &tbaz);
+  return tbaz;
+}