sched/autogroup: Fix possible Spectre-v1 indexing for sched_prio_to_weight[]

> kernel/sched/autogroup.c:230 proc_sched_autogroup_set_nice() warn: potential spectre issue 'sched_prio_to_weight'

Userspace controls @nice, sanitize the array index.
Reported-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: <stable@kernel.org>
Signed-off-by: Ingo Molnar <mingo@kernel.org>

CVE-2017-5753

(backported from commit 354d7793)
[juergh:
 - Adjusted context.
 - Modified kernel/sched/auto_group.c instead of kernel/sched/autogroup.c.]
Signed-off-by: Juerg Haefliger <juergh@canonical.com>
Acked-by: Stefan Bader <stefan.bader@canonical.com>
Acked-by: Kleber Sacilotto de Souza <kleber.souza@canonical.com>
Signed-off-by: Kleber Sacilotto de Souza <kleber.souza@canonical.com>

kernel/sched/auto_group.c

@@ -6,6 +6,7 @@
 #include <linux/utsname.h>
 #include <linux/security.h>
 #include <linux/export.h>
+#include <linux/nospec.h>
 
 unsigned int __read_mostly sysctl_sched_autogroup_enabled = 1;
 static struct autogroup autogroup_default;
@@ -193,7 +194,7 @@ int proc_sched_autogroup_set_nice(struct task_struct *p, int nice)
 {
 	static unsigned long next = INITIAL_JIFFIES;
 	struct autogroup *ag;
-	int err;
+	int err, idx;
 
 	if (nice < MIN_NICE || nice > MAX_NICE)
 		return -EINVAL;
@@ -213,7 +214,9 @@ int proc_sched_autogroup_set_nice(struct task_struct *p, int nice)
 	ag = autogroup_task_get(p);
 
 	down_write(&ag->lock);
-	err = sched_group_set_shares(ag->tg, prio_to_weight[nice + 20]);
+
+	idx = array_index_nospec(nice + 20, 40);
+	err = sched_group_set_shares(ag->tg, prio_to_weight[idx]);
 	if (!err)
 		ag->nice = nice;
 	up_write(&ag->lock);