[PATCH] [4/4] /dev/random: Remove RNDGETPOOL ioctl

[PATCH] [4/4] /dev/random: Remove RNDGETPOOL ioctl

[Theodore Ts'o]

Recently, someone has kvetched that RNDGETPOOL is a "security
vulnerability".  Never mind that it is superuser only, and with
superuser privs you could load a nasty kernel module, or read the
entropy pool out of /dev/mem directly, but they are nevertheless still
spreading FUD.

In any case, no one is using it (it was there for debugging purposes
only), so we can remove it as dead code.

patch-random-4-remove-rndgetpool

--- random.c	2004/08/19 22:50:19	1.4
+++ random.c	2004/08/19 22:50:43	1.5
@@ -1741,10 +1741,9 @@
 random_ioctl(struct inode * inode, struct file * file,
 	     unsigned int cmd, unsigned long arg)
 {
-	int *tmp, size, ent_count;
+	int size, ent_count;
 	int __user *p = (int __user *)arg;
 	int retval;
-	unsigned long flags;
 	
 	switch (cmd) {
 	case RNDGETENTCNT:
@@ -1765,40 +1764,6 @@
 		if (random_state->entropy_count >= random_read_wakeup_thresh)
 			wake_up_interruptible(&random_read_wait);
 		return 0;
-	case RNDGETPOOL:
-		if (!capable(CAP_SYS_ADMIN))
-			return -EPERM;
-		if (get_user(size, p) ||
-		    put_user(random_state->poolinfo.poolwords, p++))
-			return -EFAULT;
-		if (size < 0)
-			return -EFAULT;
-		if (size > random_state->poolinfo.poolwords)
-			size = random_state->poolinfo.poolwords;
-
-		/* prepare to atomically snapshot pool */
-
-		tmp = kmalloc(size * sizeof(__u32), GFP_KERNEL);
-
-		if (!tmp)
-			return -ENOMEM;
-
-		spin_lock_irqsave(&random_state->lock, flags);
-		ent_count = random_state->entropy_count;
-		memcpy(tmp, random_state->pool, size * sizeof(__u32));
-		spin_unlock_irqrestore(&random_state->lock, flags);
-
-		if (!copy_to_user(p, tmp, size * sizeof(__u32))) {
-			kfree(tmp);
-			return -EFAULT;
-		}
-
-		kfree(tmp);
-
-		if(put_user(ent_count, p++))
-			return -EFAULT;
-
-		return 0;
 	case RNDADDENTROPY:
 		if (!capable(CAP_SYS_ADMIN))
 			return -EPERM;

Re: [PATCH] [4/4] /dev/random: Remove RNDGETPOOL ioctl

[Matt Mackall]

On Fri, Aug 20, 2004 at 12:57:23AM -0400, Theodore Ts'o wrote:
> 
> Recently, someone has kvetched that RNDGETPOOL is a "security
> vulnerability".  Never mind that it is superuser only, and with
> superuser privs you could load a nasty kernel module, or read the
> entropy pool out of /dev/mem directly, but they are nevertheless still
> spreading FUD.

While such concerns are a bit exaggerated, the ioctl isn't in fact
very useful: it only gets one of the pools. In other words, it's been
obsolete even for debugging purposes since we went to two pools.

The pool resize ioctl is still racy in a painful way and ought to be
axed as well.

-- 
Mathematics is the supreme nostalgia of our time.