[PATCH 2.6] fix use-after-free in ds.c
From: Daniel Ritz (daniel.ritz_at_gmx.ch)
Date: 12/28/03
- Previous message: DervishD: "Re: best AMD motherboard for Linux"
- Next in thread: Russell King: "Re: [PATCH 2.6] fix use-after-free in ds.c"
- Reply: Russell King: "Re: [PATCH 2.6] fix use-after-free in ds.c"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
To: Russell King <rmk@arm.linux.org.uk>, "linux-pcmcia" <linux-pcmcia@lists.infradead.org>, "linux-kernel" <linux-kernel@vger.kernel.org> Date: Sun, 28 Dec 2003 18:52:01 +0100
fixes a use-after-free in ds.c seen when doing:
# cardctl eject
# rmmod yenta_socket
# killall cardmgr
this is russell king's patch from october plus removes some unused vars.
(original see: http://lists.infradead.org/pipermail/linux-pcmcia/2003-October/000303.html )
against 2.6.0, please enqueue.
--- 1.39/drivers/pcmcia/ds.c Sat Sep 27 11:11:05 2003
+++ edited/drivers/pcmcia/ds.c Sun Dec 28 17:30:54 2003
@@ -51,6 +51,8 @@
#include <linux/list.h>
#include <linux/workqueue.h>
+#include <asm/atomic.h>
+
#include <pcmcia/version.h>
#include <pcmcia/cs_types.h>
#include <pcmcia/cs.h>
@@ -95,10 +97,12 @@
int event_head, event_tail;
event_t event[MAX_EVENTS];
struct user_info_t *next;
+ struct pcmcia_bus_socket *socket;
} user_info_t;
/* Socket state information */
struct pcmcia_bus_socket {
+ atomic_t refcount;
client_handle_t handle;
int state;
user_info_t *user;
@@ -113,6 +117,7 @@
#define SOCKET_PRESENT 0x01
#define SOCKET_BUSY 0x02
#define SOCKET_REMOVAL_PENDING 0x10
+#define SOCKET_DEAD 0x80
/*====================================================================*/
@@ -137,6 +142,24 @@
static struct pcmcia_driver * get_pcmcia_driver (dev_info_t *dev_info);
static struct pcmcia_bus_socket * get_socket_info_by_nr(unsigned int nr);
+static void pcmcia_put_bus_socket(struct pcmcia_bus_socket *s)
+{
+ if (atomic_dec_and_test(&s->refcount))
+ kfree(s);
+}
+
+static struct pcmcia_bus_socket *pcmcia_get_bus_socket(int nr)
+{
+ struct pcmcia_bus_socket *s;
+
+ s = get_socket_info_by_nr(nr);
+ if (s) {
+ WARN_ON(atomic_read(&s->refcount) == 0);
+ atomic_inc(&s->refcount);
+ }
+ return s;
+}
+
/**
* pcmcia_register_driver - register a PCMCIA driver with the bus core
*
@@ -501,7 +524,7 @@
DEBUG(0, "ds_open(socket %d)\n", i);
- s = get_socket_info_by_nr(i);
+ s = pcmcia_get_bus_socket(i);
if (!s)
return -ENODEV;
@@ -517,6 +540,7 @@
user->event_tail = user->event_head = 0;
user->next = s->user;
user->user_magic = USER_MAGIC;
+ user->socket = s;
s->user = user;
file->private_data = user;
@@ -529,20 +553,17 @@
static int ds_release(struct inode *inode, struct file *file)
{
- socket_t i = iminor(inode);
struct pcmcia_bus_socket *s;
user_info_t *user, **link;
DEBUG(0, "ds_release(socket %d)\n", i);
- s = get_socket_info_by_nr(i);
- if (!s)
- return 0;
-
user = file->private_data;
if (CHECK_USER(user))
goto out;
+ s = user->socket;
+
/* Unlink user data structure */
if ((file->f_flags & O_ACCMODE) != O_RDONLY)
s->state &= ~SOCKET_BUSY;
@@ -554,6 +575,7 @@
*link = user->next;
user->user_magic = 0;
kfree(user);
+ pcmcia_put_bus_socket(s);
out:
return 0;
} /* ds_release */
@@ -563,7 +585,6 @@
static ssize_t ds_read(struct file *file, char *buf,
size_t count, loff_t *ppos)
{
- socket_t i = iminor(file->f_dentry->d_inode);
struct pcmcia_bus_socket *s;
user_info_t *user;
@@ -572,14 +593,14 @@
if (count < 4)
return -EINVAL;
- s = get_socket_info_by_nr(i);
- if (!s)
- return -ENODEV;
-
user = file->private_data;
if (CHECK_USER(user))
return -EIO;
+ s = user->socket;
+ if (s->state & SOCKET_DEAD)
+ return -EIO;
+
if (queue_empty(user)) {
interruptible_sleep_on(&s->queue);
if (signal_pending(current))
@@ -594,7 +615,6 @@
static ssize_t ds_write(struct file *file, const char *buf,
size_t count, loff_t *ppos)
{
- socket_t i = iminor(file->f_dentry->d_inode);
struct pcmcia_bus_socket *s;
user_info_t *user;
@@ -605,14 +625,14 @@
if ((file->f_flags & O_ACCMODE) == O_RDONLY)
return -EBADF;
- s = get_socket_info_by_nr(i);
- if (!s)
- return -ENODEV;
-
user = file->private_data;
if (CHECK_USER(user))
return -EIO;
+ s = user->socket;
+ if (s->state & SOCKET_DEAD)
+ return -EIO;
+
if (s->req_pending) {
s->req_pending--;
get_user(s->req_result, (int *)buf);
@@ -629,19 +649,19 @@
/* No kernel lock - fine */
static u_int ds_poll(struct file *file, poll_table *wait)
{
- socket_t i = iminor(file->f_dentry->d_inode);
struct pcmcia_bus_socket *s;
user_info_t *user;
DEBUG(2, "ds_poll(socket %d)\n", i);
- s = get_socket_info_by_nr(i);
- if (!s)
- return POLLERR;
-
user = file->private_data;
if (CHECK_USER(user))
return POLLERR;
+ s = user->socket;
+ /*
+ * We don't check for a dead socket here since that
+ * will send cardmgr into an endless spin.
+ */
poll_wait(file, &s->queue, wait);
if (!queue_empty(user))
return POLLIN | POLLRDNORM;
@@ -653,17 +673,21 @@
static int ds_ioctl(struct inode * inode, struct file * file,
u_int cmd, u_long arg)
{
- socket_t i = iminor(inode);
struct pcmcia_bus_socket *s;
u_int size;
int ret, err;
ds_ioctl_arg_t buf;
+ user_info_t *user;
DEBUG(2, "ds_ioctl(socket %d, %#x, %#lx)\n", i, cmd, arg);
- s = get_socket_info_by_nr(i);
- if (!s)
- return -ENODEV;
+ user = file->private_data;
+ if (CHECK_USER(user))
+ return -EIO;
+
+ s = user->socket;
+ if (s->state & SOCKET_DEAD)
+ return -EIO;
size = (cmd & IOCSIZE_MASK) >> IOCSIZE_SHIFT;
if (size > sizeof(ds_ioctl_arg_t)) return -EINVAL;
@@ -833,6 +857,7 @@
if(!s)
return -ENOMEM;
memset(s, 0, sizeof(struct pcmcia_bus_socket));
+ atomic_set(&s->refcount, 1);
/*
* Ugly. But we want to wait for the socket threads to have started up.
@@ -894,7 +919,8 @@
pcmcia_deregister_client(socket->pcmcia->handle);
- kfree(socket->pcmcia);
+ socket->pcmcia->state |= SOCKET_DEAD;
+ pcmcia_put_bus_socket(socket->pcmcia);
socket->pcmcia = NULL;
return;
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
- Previous message: DervishD: "Re: best AMD motherboard for Linux"
- Next in thread: Russell King: "Re: [PATCH 2.6] fix use-after-free in ds.c"
- Reply: Russell King: "Re: [PATCH 2.6] fix use-after-free in ds.c"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Relevant Pages
|
|
