OpenE2K
/
linux
6
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity

random: Fix crashes with sparse node ids 

On a system with sparse node ids, eg. a powerpc system with 4 nodes
numbered like so:

  node   0: [mem 0x0000000000000000-0x00000007ffffffff]
  node   1: [mem 0x0000000800000000-0x0000000fffffffff]
  node  16: [mem 0x0000001000000000-0x00000017ffffffff]
  node  17: [mem 0x0000001800000000-0x0000001fffffffff]

The code in rand_initialize() will allocate 4 pointers for the pool
array, and initialise them correctly.

However when go to use the pool, in eg. extract_crng(), we use the
numa_node_id() to index into the array. For the higher numbered node ids
this leads to random memory corruption, depending on what was kmalloc'ed
adjacent to the pool array.

Fix it by using nr_node_ids to size the pool array.

Fixes: 1e7f583af6 ("random: make /dev/urandom scalable for silly userspace programs")
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
This commit is contained in:
Michael Ellerman 2016-07-31 00:23:08 +10:00 committed by Linus Torvalds
parent 7f155c7026
commit dd0f0cf58a
1 changed files with 2 additions and 3 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

5
drivers/char/random.c
View File

@ -249,6 +249,7 @@
#include <linux/genhd.h>
#include <linux/interrupt.h>
#include <linux/mm.h>
#include <linux/nodemask.h>
#include <linux/spinlock.h>
#include <linux/kthread.h>
#include <linux/percpu.h>
@ -1656,7 +1657,6 @@ static int rand_initialize(void)
{
#ifdef CONFIG_NUMA
int i;
int num_nodes = num_possible_nodes();
struct crng_state *crng;
struct crng_state **pool;
#endif
@ -1666,8 +1666,7 @@ static int rand_initialize(void)
crng_initialize(&primary_crng);
#ifdef CONFIG_NUMA
pool = kmalloc(num_nodes * sizeof(void *),
GFP_KERNEL|__GFP_NOFAIL|__GFP_ZERO);
pool = kcalloc(nr_node_ids, sizeof(*pool), GFP_KERNEL|__GFP_NOFAIL);
for_each_online_node(i) {
crng = kmalloc_node(sizeof(struct crng_state),
GFP_KERNEL | __GFP_NOFAIL, i);