Skip to contentSkip to conversation

splice: missing user pointer access verification (CVE-2008-0009/10)

PATCH In review Greg KH 2 participants 4 messages Feb 8, 2008 5 messages not in the archive VFS

lore
Share & tools
lore

Greg KH

author
#5 lore raw

PATCH 1/1 splice: missing user pointer access verification (CVE-2008-0009/10)

In reply to a message not in the archive

From: Jens Axboe <[email protected]>

vmsplice_to_user() must always check the user pointer and length with access_ok() before copying. Likewise, for the slow path of copy_from_user_mmap_sem() we need to check that we may read from the user region.

---

Linus, this fixes a security hole in splice that is now public. I have it queued up for the .23 and .24 -stable releases as well.

 fs/splice.c |    8 ++++++++
 1 files changed, 8 insertions(+), 0 deletions(-)
fs/splice.c +8 -0
index 4ee49e8..14e2262 100644
@@ -1179,6 +1179,9 @@ static int copy_from_user_mmap_sem(void *dst, const void __user *src, size_t n)
 {
 	int partial;
 
+	if (!access_ok(VERIFY_READ, src, n))
+		return -EFAULT;
+
 	pagefault_disable();
 	partial = __copy_from_user_inatomic(dst, src, n);
 	pagefault_enable();
@@ -1387,6 +1390,11 @@ static long vmsplice_to_user(struct file *file, const struct iovec __user *iov,
 			break;
 		}
 
+		if (unlikely(!access_ok(VERIFY_WRITE, base, len))) {
+			error = -EFAULT;
+			break;
+		}
+
 		sd.len = 0;
 		sd.total_len = len;
 		sd.flags = flags;
Signature · 5 lines
1.5.4.22.g7a20


-- 
Jens Axboe
#6 lore raw

greg it's for .22 or the splice is changed between .22 and .23?

Quoted 58 lines · Greg KH

From: Jens Axboe <[email protected]>

vmsplice_to_user() must always check the user pointer and length with access_ok() before copying. Likewise, for the slow path of copy_from_user_mmap_sem() we need to check that we may read from the user region.

---

Linus, this fixes a security hole in splice that is now public. I have it queued up for the .23 and .24 -stable releases as well.

 fs/splice.c |    8 ++++++++
 1 files changed, 8 insertions(+), 0 deletions(-)
fs/splice.c +3 -0
index 4ee49e8..14e2262 100644
@@ -1179,6 +1179,9 @@ static int copy_from_user_mmap_sem(void *dst, const
 void __user *src, size_t n)
 {
 	int partial;
 
+	if (!access_ok(VERIFY_READ, src, n))
+		return -EFAULT;
+
 	pagefault_disable();
 	partial = __copy_from_user_inatomic(dst, src, n);
fs/splice.c +5 -0
 	pagefault_enable();
@@ -1387,6 +1390,11 @@ static long vmsplice_to_user(struct file *file, const
 struct iovec __user *iov,
 			break;
 		}
 
+		if (unlikely(!access_ok(VERIFY_WRITE, base, len))) {
+			error = -EFAULT;
+			break;
+		}
+
 		sd.len = 0;
 		sd.total_len = len;

sd.flags = flags;
--
1.5.4.22.g7a20

--
Jens Axboe
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/

Signature · 2 lines
Thanks,
Oliver
#7 lore raw

hmm, when I good see, this is not for .22, and it (vmsplice_to_user) is came with .23

Quoted 68 lines · Oliver Pinter

greg it's for .22 or the splice is changed between .22 and .23?

fs/splice.c static long vmsplice_to_user(struct file *file,

Quoted 34 lines · Greg KH

From: Jens Axboe <[email protected]>

vmsplice_to_user() must always check the user pointer and length with access_ok() before copying. Likewise, for the slow path of copy_from_user_mmap_sem() we need to check that we may read from the user region.

---

Linus, this fixes a security hole in splice that is now public. I have it queued up for the .23 and .24 -stable releases as well.

 fs/splice.c |    8 ++++++++
 1 files changed, 8 insertions(+), 0 deletions(-)
fs/splice.c +3 -0
index 4ee49e8..14e2262 100644
@@ -1179,6 +1179,9 @@ static int copy_from_user_mmap_sem(void *dst, const
 void __user *src, size_t n)
 {
 	int partial;
 
+	if (!access_ok(VERIFY_READ, src, n))
+		return -EFAULT;
+
 	pagefault_disable();
 	partial = __copy_from_user_inatomic(dst, src, n);
fs/splice.c +0 -0
 	pagefault_enable();
@@ -1387,6 +1390,11 @@ static long vmsplice_to_user(struct file *file,

const

Quoted 16 lines
fs/splice.c +5 -0
 struct iovec __user *iov,
 			break;
 		}
 
+		if (unlikely(!access_ok(VERIFY_WRITE, base, len))) {
+			error = -EFAULT;
+			break;
+		}
+
 		sd.len = 0;
 		sd.total_len = len;

sd.flags = flags;
--
1.5.4.22.g7a20

Quoted 8 lines

--
Jens Axboe
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/

--
Thanks,
Oliver

Signature · 2 lines
Thanks,
Oliver
author
#8 lore raw
In reply to Oliver Pinter
Quoted 1 line · Oliver Pinter

greg it's for .22 or the splice is changed between .22 and .23?

splice changed for .23 and this only affects .23 and older kernels, so .22 and older kernels do not have issues.

thanks,

greg k-h