module: trim exception table on init free.

It's theoretically possible that there are exception table entries
which point into the (freed) init text of modules.  These could cause
future problems if other modules get loaded into that memory and cause
an exception as we'd see the wrong fixup.  The only case I know of is
kvm-intel.ko (when CONFIG_CC_OPTIMIZE_FOR_SIZE=n).

Amerigo fixed this long-standing FIXME in the x86 version, but this
patch is more general.

This implements trim_init_extable(); most archs are simple since they
use the standard lib/extable.c sort code.  Alpha and IA64 use relative
addresses in their fixups, so thier trimming is a slight variation.

Sparc32 is unique; it doesn't seem to define ARCH_HAS_SORT_EXTABLE,
yet it defines its own sort_extable() which overrides the one in lib.
It doesn't sort, so we have to mark deleted entries instead of
actually trimming them.

Inspired-by: Amerigo Wang <amwang@redhat.com>
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Cc: linux-alpha@vger.kernel.org
Cc: sparclinux@vger.kernel.org
Cc: linux-ia64@vger.kernel.org
diff --git a/arch/alpha/mm/extable.c b/arch/alpha/mm/extable.c
index 62dc379..813c9b6 100644
--- a/arch/alpha/mm/extable.c
+++ b/arch/alpha/mm/extable.c
@@ -48,6 +48,27 @@
 	     cmp_ex, swap_ex);
 }
 
+#ifdef CONFIG_MODULES
+/*
+ * Any entry referring to the module init will be at the beginning or
+ * the end.
+ */
+void trim_init_extable(struct module *m)
+{
+	/*trim the beginning*/
+	while (m->num_exentries &&
+	       within_module_init(ex_to_addr(&m->extable[0]), m)) {
+		m->extable++;
+		m->num_exentries--;
+	}
+	/*trim the end*/
+	while (m->num_exentries &&
+	       within_module_init(ex_to_addr(&m->extable[m->num_exentries-1]),
+				  m))
+		m->num_exentries--;
+}
+#endif /* CONFIG_MODULES */
+
 const struct exception_table_entry *
 search_extable(const struct exception_table_entry *first,
 	       const struct exception_table_entry *last,
diff --git a/arch/ia64/mm/extable.c b/arch/ia64/mm/extable.c
index 71c50dd..e95d5ad 100644
--- a/arch/ia64/mm/extable.c
+++ b/arch/ia64/mm/extable.c
@@ -53,6 +53,32 @@
 	     cmp_ex, swap_ex);
 }
 
+static inline unsigned long ex_to_addr(const struct exception_table_entry *x)
+{
+	return (unsigned long)&x->insn + x->insn;
+}
+
+#ifdef CONFIG_MODULES
+/*
+ * Any entry referring to the module init will be at the beginning or
+ * the end.
+ */
+void trim_init_extable(struct module *m)
+{
+	/*trim the beginning*/
+	while (m->num_exentries &&
+	       within_module_init(ex_to_addr(&m->extable[0]), m)) {
+		m->extable++;
+		m->num_exentries--;
+	}
+	/*trim the end*/
+	while (m->num_exentries &&
+	       within_module_init(ex_to_addr(&m->extable[m->num_exentries-1]),
+				  m))
+		m->num_exentries--;
+}
+#endif /* CONFIG_MODULES */
+
 const struct exception_table_entry *
 search_extable (const struct exception_table_entry *first,
 		const struct exception_table_entry *last,
diff --git a/arch/sparc/include/asm/uaccess_32.h b/arch/sparc/include/asm/uaccess_32.h
index 47d5619..8303ac4 100644
--- a/arch/sparc/include/asm/uaccess_32.h
+++ b/arch/sparc/include/asm/uaccess_32.h
@@ -17,6 +17,9 @@
 
 #ifndef __ASSEMBLY__
 
+#define ARCH_HAS_SORT_EXTABLE
+#define ARCH_HAS_SEARCH_EXTABLE
+
 /* Sparc is not segmented, however we need to be able to fool access_ok()
  * when doing system calls from kernel mode legitimately.
  *
diff --git a/arch/sparc/mm/extable.c b/arch/sparc/mm/extable.c
index 16cc289..a61c349 100644
--- a/arch/sparc/mm/extable.c
+++ b/arch/sparc/mm/extable.c
@@ -28,6 +28,10 @@
 	 *	word 3: last insn address + 4 bytes
 	 *	word 4: fixup code address
 	 *
+	 * Deleted entries are encoded as:
+	 *	word 1: unused
+	 *	word 2: -1
+	 *
 	 * See asm/uaccess.h for more details.
 	 */
 
@@ -39,6 +43,10 @@
 			continue;
 		}
 
+		/* A deleted entry; see trim_init_extable */
+		if (walk->fixup == -1)
+			continue;
+
 		if (walk->insn == value)
 			return walk;
 	}
@@ -57,6 +65,27 @@
         return NULL;
 }
 
+#ifdef CONFIG_MODULES
+/* We could memmove them around; easier to mark the trimmed ones. */
+void trim_init_extable(struct module *m)
+{
+	unsigned int i;
+	bool range;
+
+	for (i = 0; i < m->num_exentries; i += range ? 2 : 1) {
+		range = m->extable[i].fixup == 0;
+
+		if (within_module_init(m->extable[i].insn, m)) {
+			m->extable[i].fixup = -1;
+			if (range)
+				m->extable[i+1].fixup = -1;
+		}
+		if (range)
+			i++;
+	}
+}
+#endif /* CONFIG_MODULES */
+
 /* Special extable search, which handles ranges.  Returns fixup */
 unsigned long search_extables_range(unsigned long addr, unsigned long *g2)
 {
diff --git a/include/linux/module.h b/include/linux/module.h
index a8f2c0a..a7bc6e7b 100644
--- a/include/linux/module.h
+++ b/include/linux/module.h
@@ -77,6 +77,7 @@
 void sort_extable(struct exception_table_entry *start,
 		  struct exception_table_entry *finish);
 void sort_main_extable(void);
+void trim_init_extable(struct module *m);
 
 #ifdef MODULE
 #define MODULE_GENERIC_TABLE(gtype,name)			\
diff --git a/kernel/module.c b/kernel/module.c
index 35f7de0..e4ab36c 100644
--- a/kernel/module.c
+++ b/kernel/module.c
@@ -2455,6 +2455,7 @@
 	mutex_lock(&module_mutex);
 	/* Drop initial reference. */
 	module_put(mod);
+	trim_init_extable(mod);
 	module_free(mod, mod->module_init);
 	mod->module_init = NULL;
 	mod->init_size = 0;
diff --git a/lib/extable.c b/lib/extable.c
index 179c087..4cac81e 100644
--- a/lib/extable.c
+++ b/lib/extable.c
@@ -39,7 +39,26 @@
 	sort(start, finish - start, sizeof(struct exception_table_entry),
 	     cmp_ex, NULL);
 }
-#endif
+
+#ifdef CONFIG_MODULES
+/*
+ * If the exception table is sorted, any referring to the module init
+ * will be at the beginning or the end.
+ */
+void trim_init_extable(struct module *m)
+{
+	/*trim the beginning*/
+	while (m->num_exentries && within_module_init(m->extable[0].insn, m)) {
+		m->extable++;
+		m->num_exentries--;
+	}
+	/*trim the end*/
+	while (m->num_exentries &&
+		within_module_init(m->extable[m->num_exentries-1].insn, m))
+		m->num_exentries--;
+}
+#endif /* CONFIG_MODULES */
+#endif /* !ARCH_HAS_SORT_EXTABLE */
 
 #ifndef ARCH_HAS_SEARCH_EXTABLE
 /*