Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > linux.kernel > #1550682 > unrolled thread

[PATCH] drivers: char: mem: Fix thinko in kmem address checks

Back to article view | Back to linux.kernel

Contents

  [PATCH] drivers: char: mem: Fix thinko in kmem address checks Robin Murphy <robin.murphy@arm.com> - 2017-01-04 12:40 +0100
    Re: [PATCH] drivers: char: mem: Fix thinko in kmem address checks Greg KH <gregkh@linuxfoundation.org> - 2017-01-04 12:50 +0100
      Re: [PATCH] drivers: char: mem: Fix thinko in kmem address checks Greg KH <gregkh@linuxfoundation.org> - 2017-01-04 18:30 +0100
    [PATCH v2] drivers: char: mem: Fix thinkos in kmem address checks Robin Murphy <robin.murphy@arm.com> - 2017-01-05 18:20 +0100
      Re: [PATCH v2] drivers: char: mem: Fix thinkos in kmem address checks Greg KH <gregkh@linuxfoundation.org> - 2017-01-10 18:50 +0100
        Re: [PATCH v2] drivers: char: mem: Fix thinkos in kmem address checks "Jason A. Donenfeld" <Jason@zx2c4.com> - 2017-01-11 20:20 +0100

#1550682 — [PATCH] drivers: char: mem: Fix thinko in kmem address checks

When borrowing the pfn_valid() check from mmap_kmem(), somebody managed
to get physical and virtual addresses spectacularly muddled up, such
that we've ended up with checks for one being the other. Whilst this
does indeed prevent out-of-bounds accesses crashing, on most systems it
also prevents the more desirable use-case of working at all ever.

Check the *virtual* offset correctly for what it is.

Reported-by: Jason A. Donenfeld <Jason@zx2c4.com>
CC: stable@vger.kernel.org
Fixes: 148a1bc84398 ("drivers: char: mem: Check {read,write}_kmem() addresses")
Signed-off-by: Robin Murphy <robin.murphy@arm.com>
---
 drivers/char/mem.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/drivers/char/mem.c b/drivers/char/mem.c
index 5bb1985ec484..bdc6a4018604 100644
--- a/drivers/char/mem.c
+++ b/drivers/char/mem.c
@@ -381,7 +381,7 @@ static ssize_t read_kmem(struct file *file, char __user *buf,
 	char *kbuf; /* k-addr because vread() takes vmlist_lock rwlock */
 	int err = 0;
 
-	if (!pfn_valid(PFN_DOWN(p)))
+	if (!virt_addr_valid(p))
 		return -EIO;
 
 	read = 0;
@@ -512,7 +512,7 @@ static ssize_t write_kmem(struct file *file, const char __user *buf,
 	char *kbuf; /* k-addr because vwrite() takes vmlist_lock rwlock */
 	int err = 0;
 
-	if (!pfn_valid(PFN_DOWN(p)))
+	if (!virt_addr_valid(p))
 		return -EIO;
 
 	if (p < (unsigned long) high_memory) {
-- 
2.10.2.dirty

[toc] | [next] | [standalone]

#1550694

On Wed, Jan 04, 2017 at 11:37:49AM +0000, Robin Murphy wrote:
> When borrowing the pfn_valid() check from mmap_kmem(), somebody managed

"sombody"?  :)

> to get physical and virtual addresses spectacularly muddled up, such
> that we've ended up with checks for one being the other. Whilst this
> does indeed prevent out-of-bounds accesses crashing, on most systems it
> also prevents the more desirable use-case of working at all ever.
> 
> Check the *virtual* offset correctly for what it is.
> 
> Reported-by: Jason A. Donenfeld <Jason@zx2c4.com>
> CC: stable@vger.kernel.org
> Fixes: 148a1bc84398 ("drivers: char: mem: Check {read,write}_kmem() addresses")
> Signed-off-by: Robin Murphy <robin.murphy@arm.com>
> ---
>  drivers/char/mem.c | 4 ++--
>  1 file changed, 2 insertions(+), 2 deletions(-)
> 
> diff --git a/drivers/char/mem.c b/drivers/char/mem.c
> index 5bb1985ec484..bdc6a4018604 100644
> --- a/drivers/char/mem.c
> +++ b/drivers/char/mem.c
> @@ -381,7 +381,7 @@ static ssize_t read_kmem(struct file *file, char __user *buf,
>  	char *kbuf; /* k-addr because vread() takes vmlist_lock rwlock */
>  	int err = 0;
>  
> -	if (!pfn_valid(PFN_DOWN(p)))
> +	if (!virt_addr_valid(p))
>  		return -EIO;
>  
>  	read = 0;
> @@ -512,7 +512,7 @@ static ssize_t write_kmem(struct file *file, const char __user *buf,
>  	char *kbuf; /* k-addr because vwrite() takes vmlist_lock rwlock */
>  	int err = 0;
>  
> -	if (!pfn_valid(PFN_DOWN(p)))
> +	if (!virt_addr_valid(p))
>  		return -EIO;
>  
>  	if (p < (unsigned long) high_memory) {

Jason, can you verify this fixes your test case?

thanks,

greg k-h

[toc] | [prev] | [next] | [standalone]

#1551006

On Wed, Jan 04, 2017 at 12:45:32PM +0100, Greg KH wrote:
> On Wed, Jan 04, 2017 at 11:37:49AM +0000, Robin Murphy wrote:
> > When borrowing the pfn_valid() check from mmap_kmem(), somebody managed
> 
> "sombody"?  :)
> 
> > to get physical and virtual addresses spectacularly muddled up, such
> > that we've ended up with checks for one being the other. Whilst this
> > does indeed prevent out-of-bounds accesses crashing, on most systems it
> > also prevents the more desirable use-case of working at all ever.
> > 
> > Check the *virtual* offset correctly for what it is.
> > 
> > Reported-by: Jason A. Donenfeld <Jason@zx2c4.com>
> > CC: stable@vger.kernel.org
> > Fixes: 148a1bc84398 ("drivers: char: mem: Check {read,write}_kmem() addresses")
> > Signed-off-by: Robin Murphy <robin.murphy@arm.com>
> > ---
> >  drivers/char/mem.c | 4 ++--
> >  1 file changed, 2 insertions(+), 2 deletions(-)
> > 
> > diff --git a/drivers/char/mem.c b/drivers/char/mem.c
> > index 5bb1985ec484..bdc6a4018604 100644
> > --- a/drivers/char/mem.c
> > +++ b/drivers/char/mem.c
> > @@ -381,7 +381,7 @@ static ssize_t read_kmem(struct file *file, char __user *buf,
> >  	char *kbuf; /* k-addr because vread() takes vmlist_lock rwlock */
> >  	int err = 0;
> >  
> > -	if (!pfn_valid(PFN_DOWN(p)))
> > +	if (!virt_addr_valid(p))
> >  		return -EIO;
> >  
> >  	read = 0;
> > @@ -512,7 +512,7 @@ static ssize_t write_kmem(struct file *file, const char __user *buf,
> >  	char *kbuf; /* k-addr because vwrite() takes vmlist_lock rwlock */
> >  	int err = 0;
> >  
> > -	if (!pfn_valid(PFN_DOWN(p)))
> > +	if (!virt_addr_valid(p))
> >  		return -EIO;
> >  
> >  	if (p < (unsigned long) high_memory) {
> 
> Jason, can you verify this fixes your test case?

Well, it fails kbuild testing, so can you try it again?

thanks,

greg k-h

[toc] | [prev] | [next] | [standalone]

#1552133 — [PATCH v2] drivers: char: mem: Fix thinkos in kmem address checks

When borrowing the pfn_valid() check from mmap_kmem(), somebody managed
to get physical and virtual addresses spectacularly muddled up, such
that we've ended up with checks for one being the other. Whilst this
does indeed prevent out-of-bounds accesses crashing, on most systems
it also prevents the more desirable use-case of working at all ever.

Check the *virtual* offset correctly for what it is. Furthermore, do
so in the right place - a read or write may span multiple pages, so a
single up-front check is insufficient. High memory accesses already
have a similar validity check just before the copy_to_user() call, so
just make the low memory path fully consistent with that.

Reported-by: Jason A. Donenfeld <Jason@zx2c4.com>
CC: stable@vger.kernel.org
Fixes: 148a1bc84398 ("drivers: char: mem: Check {read,write}_kmem() addresses")
Signed-off-by: Robin Murphy <robin.murphy@arm.com>
---

Third time lucky... And if there's some other problem with this one then
I guess we may as well just go ahead with Jason's revert, forget the whole
thing, and let 'cat /dev/kmem' go back to crashing on non-x86 :)

Robin.

 drivers/char/mem.c | 10 ++++------
 1 file changed, 4 insertions(+), 6 deletions(-)

diff --git a/drivers/char/mem.c b/drivers/char/mem.c
index 5bb1985ec484..6d9cc2d39d22 100644
--- a/drivers/char/mem.c
+++ b/drivers/char/mem.c
@@ -381,9 +381,6 @@ static ssize_t read_kmem(struct file *file, char __user *buf,
 	char *kbuf; /* k-addr because vread() takes vmlist_lock rwlock */
 	int err = 0;
 
-	if (!pfn_valid(PFN_DOWN(p)))
-		return -EIO;
-
 	read = 0;
 	if (p < (unsigned long) high_memory) {
 		low_count = count;
@@ -412,6 +409,8 @@ static ssize_t read_kmem(struct file *file, char __user *buf,
 			 * by the kernel or data corruption may occur
 			 */
 			kbuf = xlate_dev_kmem_ptr((void *)p);
+			if (!virt_addr_valid(kbuf))
+				return -ENXIO;
 
 			if (copy_to_user(buf, kbuf, sz))
 				return -EFAULT;
@@ -482,6 +481,8 @@ static ssize_t do_write_kmem(unsigned long p, const char __user *buf,
 		 * corruption may occur.
 		 */
 		ptr = xlate_dev_kmem_ptr((void *)p);
+		if (!virt_addr_valid(ptr))
+			return -ENXIO;
 
 		copied = copy_from_user(ptr, buf, sz);
 		if (copied) {
@@ -512,9 +513,6 @@ static ssize_t write_kmem(struct file *file, const char __user *buf,
 	char *kbuf; /* k-addr because vwrite() takes vmlist_lock rwlock */
 	int err = 0;
 
-	if (!pfn_valid(PFN_DOWN(p)))
-		return -EIO;
-
 	if (p < (unsigned long) high_memory) {
 		unsigned long to_write = min_t(unsigned long, count,
 					       (unsigned long)high_memory - p);
-- 
2.10.2.dirty

[toc] | [prev] | [next] | [standalone]

#1555769 — Re: [PATCH v2] drivers: char: mem: Fix thinkos in kmem address checks

On Thu, Jan 05, 2017 at 05:15:01PM +0000, Robin Murphy wrote:
> When borrowing the pfn_valid() check from mmap_kmem(), somebody managed
> to get physical and virtual addresses spectacularly muddled up, such
> that we've ended up with checks for one being the other. Whilst this
> does indeed prevent out-of-bounds accesses crashing, on most systems
> it also prevents the more desirable use-case of working at all ever.
> 
> Check the *virtual* offset correctly for what it is. Furthermore, do
> so in the right place - a read or write may span multiple pages, so a
> single up-front check is insufficient. High memory accesses already
> have a similar validity check just before the copy_to_user() call, so
> just make the low memory path fully consistent with that.
> 
> Reported-by: Jason A. Donenfeld <Jason@zx2c4.com>
> CC: stable@vger.kernel.org
> Fixes: 148a1bc84398 ("drivers: char: mem: Check {read,write}_kmem() addresses")
> Signed-off-by: Robin Murphy <robin.murphy@arm.com>

Jason, did this patch fix your issue?

thanks,

greg k-h

> ---
> 
> Third time lucky... And if there's some other problem with this one then
> I guess we may as well just go ahead with Jason's revert, forget the whole
> thing, and let 'cat /dev/kmem' go back to crashing on non-x86 :)
> 
> Robin.
> 
>  drivers/char/mem.c | 10 ++++------
>  1 file changed, 4 insertions(+), 6 deletions(-)
> 
> diff --git a/drivers/char/mem.c b/drivers/char/mem.c
> index 5bb1985ec484..6d9cc2d39d22 100644
> --- a/drivers/char/mem.c
> +++ b/drivers/char/mem.c
> @@ -381,9 +381,6 @@ static ssize_t read_kmem(struct file *file, char __user *buf,
>  	char *kbuf; /* k-addr because vread() takes vmlist_lock rwlock */
>  	int err = 0;
>  
> -	if (!pfn_valid(PFN_DOWN(p)))
> -		return -EIO;
> -
>  	read = 0;
>  	if (p < (unsigned long) high_memory) {
>  		low_count = count;
> @@ -412,6 +409,8 @@ static ssize_t read_kmem(struct file *file, char __user *buf,
>  			 * by the kernel or data corruption may occur
>  			 */
>  			kbuf = xlate_dev_kmem_ptr((void *)p);
> +			if (!virt_addr_valid(kbuf))
> +				return -ENXIO;
>  
>  			if (copy_to_user(buf, kbuf, sz))
>  				return -EFAULT;
> @@ -482,6 +481,8 @@ static ssize_t do_write_kmem(unsigned long p, const char __user *buf,
>  		 * corruption may occur.
>  		 */
>  		ptr = xlate_dev_kmem_ptr((void *)p);
> +		if (!virt_addr_valid(ptr))
> +			return -ENXIO;
>  
>  		copied = copy_from_user(ptr, buf, sz);
>  		if (copied) {
> @@ -512,9 +513,6 @@ static ssize_t write_kmem(struct file *file, const char __user *buf,
>  	char *kbuf; /* k-addr because vwrite() takes vmlist_lock rwlock */
>  	int err = 0;
>  
> -	if (!pfn_valid(PFN_DOWN(p)))
> -		return -EIO;
> -
>  	if (p < (unsigned long) high_memory) {
>  		unsigned long to_write = min_t(unsigned long, count,
>  					       (unsigned long)high_memory - p);
> -- 
> 2.10.2.dirty

[toc] | [prev] | [next] | [standalone]

#1556853 — Re: [PATCH v2] drivers: char: mem: Fix thinkos in kmem address checks

Tested-by: Jason A. Donenfeld <Jason@zx2c4.com>

My extract-keys [1] utility works again:

$ sudo ./extract-keys wg0
0x08d3c7a3 IbbC5O9g2dTZBeLQoivmKZtdSGqufU+M2s2oaFzMfYA=
0xdf917af6 3cUm2mXNaZVzWDmJIZ9XlPpvJ63GH1uyUwKLN1g1Dhs=

[1] https://git.zx2c4.com/WireGuard/tree/contrib/examples/extract-keys

[toc] | [prev] | [standalone]

Back to top | Article view | linux.kernel

csiph-web