From: Pavel Emelyanov <xemul@openvz.org>
To: Andrew Morton <akpm@linux-foundation.org>
Cc: Linux Kernel Mailing List <linux-kernel@vger.kernel.org>,
Paul Menage <menage@google.com>,
Sukadev Bhattiprolu <sukadev@us.ibm.com>,
Serge Hallyn <serue@us.ibm.com>
Subject: [PATCH 5/9] Make use of permissions, returned by kobj_lookup
Date: Wed, 05 Mar 2008 20:37:40 +0300 [thread overview]
Message-ID: <47CEDA64.1070506@openvz.org> (raw)
In-Reply-To: <47CED717.60406@openvz.org>
Now check the requesting permissions against the granted
(with the dev_t-to-kobj map) ones.
The tricky place is chrdev_open - it caches the struct cdev
on inode and thus, we have to perform lookup each time
if we are in a restricted mapping.
The task_cdev_map and task_bdev_map provide the map which
the current task is in, but now they just return NULL, which
means, that the task is not in any.
Signed-off-by: Pavel Emelyanov <xemul@openvz.org>
---
block/genhd.c | 8 +++++++-
fs/block_dev.c | 8 ++++++++
fs/char_dev.c | 18 ++++++++++++++++--
include/linux/devscontrol.h | 12 ++++++++++++
4 files changed, 43 insertions(+), 3 deletions(-)
create mode 100644 include/linux/devscontrol.h
diff --git a/block/genhd.c b/block/genhd.c
index 1d1d0f2..a619158 100644
--- a/block/genhd.c
+++ b/block/genhd.c
@@ -8,6 +8,7 @@
#include <linux/kdev_t.h>
#include <linux/kernel.h>
#include <linux/blkdev.h>
+#include <linux/devscontrol.h>
#include <linux/init.h>
#include <linux/spinlock.h>
#include <linux/seq_file.h>
@@ -212,10 +213,15 @@ void unlink_gendisk(struct gendisk *disk)
*/
struct gendisk *get_gendisk(dev_t devt, mode_t *mode, int *part)
{
+ struct kobj_map *map;
struct kobject *kobj;
struct device *dev;
- kobj = kobj_lookup(bdev_map, devt, mode, part);
+ map = task_bdev_map(current);
+ if (map == NULL)
+ map = bdev_map;
+
+ kobj = kobj_lookup(map, devt, mode, part);
if (kobj == NULL)
return NULL;
diff --git a/fs/block_dev.c b/fs/block_dev.c
index 00dda91..34dc607 100644
--- a/fs/block_dev.c
+++ b/fs/block_dev.c
@@ -945,6 +945,14 @@ static int do_open(struct block_device *bdev, struct file *file, int for_part)
bdput(bdev);
return ret;
}
+
+ if ((file->f_mode & mode) != file->f_mode) {
+ unlock_kernel();
+ bdput(bdev);
+ put_disk(disk);
+ return -EACCES;
+ }
+
owner = disk->fops->owner;
mutex_lock_nested(&bdev->bd_mutex, for_part);
diff --git a/fs/char_dev.c b/fs/char_dev.c
index dceb579..d042446 100644
--- a/fs/char_dev.c
+++ b/fs/char_dev.c
@@ -22,6 +22,8 @@
#include <linux/mutex.h>
#include <linux/backing-dev.h>
+#include <linux/devscontrol.h>
+
#ifdef CONFIG_KMOD
#include <linux/kmod.h>
#endif
@@ -361,19 +363,31 @@ static int chrdev_open(struct inode *inode, struct file *filp)
struct cdev *p;
struct cdev *new = NULL;
int ret = 0;
+ struct kobj_map *map;
+
+ map = task_cdev_map(current);
+ if (map == NULL)
+ map = cdev_map;
spin_lock(&cdev_lock);
p = inode->i_cdev;
- if (!p) {
+ if (!p || map != cdev_map) {
struct kobject *kobj;
int idx;
mode_t mode;
spin_unlock(&cdev_lock);
- kobj = kobj_lookup(cdev_map, inode->i_rdev, &mode, &idx);
+ kobj = kobj_lookup(map, inode->i_rdev, &mode, &idx);
if (!kobj)
return -ENXIO;
new = container_of(kobj, struct cdev, kobj);
+ BUG_ON(p != NULL && p != new);
+
+ if ((filp->f_mode & mode) != filp->f_mode) {
+ cdev_put(new);
+ return -EACCES;
+ }
+
spin_lock(&cdev_lock);
p = inode->i_cdev;
if (!p) {
diff --git a/include/linux/devscontrol.h b/include/linux/devscontrol.h
new file mode 100644
index 0000000..04c168b
--- /dev/null
+++ b/include/linux/devscontrol.h
@@ -0,0 +1,12 @@
+#ifndef __DEVS_CONTROL_H__
+#define __DEVS_CONTROL_H__
+static inline struct kobj_map *task_cdev_map(struct task_struct *tsk)
+{
+ return NULL;
+}
+
+static inline struct kobj_map *task_bdev_map(struct task_struct *tsk)
+{
+ return NULL;
+}
+#endif
--
1.5.3.4
next prev parent reply other threads:[~2008-03-05 17:51 UTC|newest]
Thread overview: 49+ messages / expand[flat|nested] mbox.gz Atom feed top
2008-03-05 17:23 [PATCH 0/9] Devices accessibility control group (v4) Pavel Emelyanov
2008-03-05 17:25 ` [PATCH 1/9] Avoid magic constants in drivers/base/map.c Pavel Emelyanov
2008-03-05 17:28 ` [PATCH 2/9] Cleanup the get_gendisk() a bit Pavel Emelyanov
2008-03-05 17:32 ` [PATCH 3/9] Add a mode on the struct probe Pavel Emelyanov
2008-03-05 17:34 ` [PATCH 4/9] Make kobj_lookup() return the mapping's permissions Pavel Emelyanov
2008-03-05 17:37 ` Pavel Emelyanov [this message]
2008-03-06 1:13 ` [PATCH 5/9] Make use of permissions, returned by kobj_lookup Andrew Morton
2008-03-06 8:48 ` Pavel Emelyanov
2008-03-07 9:22 ` Pavel Emelyanov
2008-03-07 9:35 ` Andrew Morton
2008-03-07 9:52 ` Pavel Emelyanov
2008-03-07 15:59 ` Greg KH
2008-03-07 16:38 ` Pavel Emelyanov
2008-03-07 17:01 ` Greg KH
2008-03-07 17:08 ` Al Viro
2008-03-07 17:35 ` Serge E. Hallyn
2008-03-07 17:57 ` Casey Schaufler
2008-03-07 18:30 ` Serge E. Hallyn
2008-03-07 19:46 ` Stephen Smalley
2008-03-07 20:57 ` Casey Schaufler
2008-03-07 21:32 ` Serge E. Hallyn
2008-03-07 18:14 ` Greg KH
2008-03-07 18:50 ` Serge E. Hallyn
2008-03-08 6:04 ` Greg KH
2008-03-08 21:47 ` Serge E. Hallyn
2008-03-09 3:15 ` Greg KH
2008-03-10 20:35 ` Serge E. Hallyn
2008-03-11 9:57 ` Pavel Emelyanov
2008-03-11 17:36 ` Greg KH
2008-03-12 8:26 ` Pavel Emelyanov
2008-03-12 13:09 ` Serge E. Hallyn
2008-03-12 13:18 ` Stephen Smalley
2008-03-12 13:27 ` Stephen Smalley
2008-03-12 14:18 ` Serge E. Hallyn
2008-03-12 14:15 ` Serge E. Hallyn
2008-03-12 16:21 ` Casey Schaufler
2008-03-12 13:36 ` Pavel Emelyanov
2008-03-05 17:40 ` [PATCH 6/9] Extend the drivers/base/map.c functionality Pavel Emelyanov
2008-03-05 17:43 ` [PATCH 7/9] Provide functions to manipulate char device mappings Pavel Emelyanov
2008-03-05 17:46 ` [PATCH 8/9] Provide functions to manipulate block " Pavel Emelyanov
2008-03-05 17:47 ` [PATCH 9/9] Devices accessibility control group itself Pavel Emelyanov
2008-03-06 2:02 ` Greg KH
2008-03-06 1:55 ` [PATCH 0/9] Devices accessibility control group (v4) Greg KH
2008-03-06 3:15 ` Serge E. Hallyn
2008-03-06 4:34 ` Greg KH
2008-03-06 8:36 ` Pavel Emelyanov
2008-03-07 4:58 ` Greg KH
2008-03-07 8:42 ` Pavel Machek
2008-03-07 8:54 ` Pavel Emelyanov
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=47CEDA64.1070506@openvz.org \
--to=xemul@openvz.org \
--cc=akpm@linux-foundation.org \
--cc=linux-kernel@vger.kernel.org \
--cc=menage@google.com \
--cc=serue@us.ibm.com \
--cc=sukadev@us.ibm.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox