Merge branch 'kallsyms-restrictions'

Merge /proc/kallsyms pointer value restrictions.

Instead of using %pK, and making it about root access (at the wrong
time, no less), make the whole choice of whether to show the actual
pointer value be very explicit to the kallsyms code.

In particular, we can now default to not doing so, and yet avoid
annoying kernel profiling by actually looking at whether kernel
profiling is allowed or not (by default it is not).

This is all mostly preparation for the real "let's stop leaking kernel
addresses" work that Tobin Harding is working on.

Small steps.

* kallsyms-restrictions:
  stop using '%pK' for /proc/kallsyms pointer values

Author: torvalds

kernel/kallsyms.c

@@ -480,6 +480,7 @@ struct kallsym_iter {
 	char name[KSYM_NAME_LEN];
 	char module_name[MODULE_NAME_LEN];
 	int exported;
+	int show_value;
 };
 
 static int get_ksymbol_mod(struct kallsym_iter *iter)
@@ -580,14 +581,23 @@ static void s_stop(struct seq_file *m, void *p)
 {
 }
 
+#ifndef CONFIG_64BIT
+# define KALLSYM_FMT "%08lx"
+#else
+# define KALLSYM_FMT "%016lx"
+#endif
+
 static int s_show(struct seq_file *m, void *p)
 {
+	unsigned long value;
 	struct kallsym_iter *iter = m->private;
 
 	/* Some debugging symbols have no name.  Ignore them. */
 	if (!iter->name[0])
 		return 0;
 
+	value = iter->show_value ? iter->value : 0;
+
 	if (iter->module_name[0]) {
 		char type;
 
@@ -597,10 +607,10 @@ static int s_show(struct seq_file *m, void *p)
 		 */
 		type = iter->exported ? toupper(iter->type) :
 					tolower(iter->type);
-		seq_printf(m, "%pK %c %s\t[%s]\n", (void *)iter->value,
+		seq_printf(m, KALLSYM_FMT " %c %s\t[%s]\n", value,
 			   type, iter->name, iter->module_name);
 	} else
-		seq_printf(m, "%pK %c %s\n", (void *)iter->value,
+		seq_printf(m, KALLSYM_FMT " %c %s\n", value,
 			   iter->type, iter->name);
 	return 0;
 }
@@ -612,6 +622,40 @@ static const struct seq_operations kallsyms_op = {
 	.show = s_show
 };
 
+static inline int kallsyms_for_perf(void)
+{
+#ifdef CONFIG_PERF_EVENTS
+	extern int sysctl_perf_event_paranoid;
+	if (sysctl_perf_event_paranoid <= 1)
+		return 1;
+#endif
+	return 0;
+}
+
+/*
+ * We show kallsyms information even to normal users if we've enabled
+ * kernel profiling and are explicitly not paranoid (so kptr_restrict
+ * is clear, and sysctl_perf_event_paranoid isn't set).
+ *
+ * Otherwise, require CAP_SYSLOG (assuming kptr_restrict isn't set to
+ * block even that).
+ */
+static int kallsyms_show_value(void)
+{
+	switch (kptr_restrict) {
+	case 0:
+		if (kallsyms_for_perf())
+			return 1;
+	/* fallthrough */
+	case 1:
+		if (has_capability_noaudit(current, CAP_SYSLOG))
+			return 1;
+	/* fallthrough */
+	default:
+		return 0;
+	}
+}
+
 static int kallsyms_open(struct inode *inode, struct file *file)
 {
 	/*
@@ -625,6 +669,7 @@ static int kallsyms_open(struct inode *inode, struct file *file)
 		return -ENOMEM;
 	reset_iter(iter, 0);
 
+	iter->show_value = kallsyms_show_value();
 	return 0;
 }