[PATCH] KEYS: put keyring if install_session_keyring_to_cred() fails

[PATCH] KEYS: put keyring if install_session_keyring_to_cred() fails

[Eric Biggers]

From: Eric Biggers <ebiggers@google.com>

In join_session_keyring(), if install_session_keyring_to_cred() were to
fail, we would leak the keyring reference, just like in the bug fixed by
commit 23567fd052a9 ("KEYS: Fix keyring ref leak in
join_session_keyring()").  Fortunately this cannot happen currently, but
we really should be more careful.  Do this by adding and using a new
error label at which the keyring reference is dropped.

Signed-off-by: Eric Biggers <ebiggers@google.com>
---
 security/keys/process_keys.c | 7 ++++---
 1 file changed, 4 insertions(+), 3 deletions(-)

diff --git a/security/keys/process_keys.c b/security/keys/process_keys.c
index b6fdd22205b1..fc60f998c74d 100644
--- a/security/keys/process_keys.c
+++ b/security/keys/process_keys.c
@@ -799,15 +799,14 @@ long join_session_keyring(const char *name)
 		ret = PTR_ERR(keyring);
 		goto error2;
 	} else if (keyring == new->session_keyring) {
-		key_put(keyring);
 		ret = 0;
-		goto error2;
+		goto error3;
 	}
 
 	/* we've got a keyring - now to install it */
 	ret = install_session_keyring_to_cred(new, keyring);
 	if (ret < 0)
-		goto error2;
+		goto error3;
 
 	commit_creds(new);
 	mutex_unlock(&key_session_mutex);
@@ -817,6 +816,8 @@ long join_session_keyring(const char *name)
 okay:
 	return ret;
 
+error3:
+	key_put(keyring);
 error2:
 	mutex_unlock(&key_session_mutex);
 error:
-- 
2.12.1

Re: [PATCH] KEYS: put keyring if install_session_keyring_to_cred() fails

[David Howells]

Eric Biggers <ebiggers3@gmail.com> wrote:

> +error3:
> +	key_put(keyring);
>  error2:
>  	mutex_unlock(&key_session_mutex);

I would prefer the put to be outside of the locked section.  How about the
attached instead?

David
---
commit 5f9e6fc4fdbbf9ada9bd03640b1c3ca50a3e5415
Author: Eric Biggers <ebiggers@google.com>
Date:   Sat Apr 1 11:54:16 2017 -0700

    KEYS: put keyring if install_session_keyring_to_cred() fails
    
    In join_session_keyring(), if install_session_keyring_to_cred() were to
    fail, we would leak the keyring reference, just like in the bug fixed by
    commit 23567fd052a9 ("KEYS: Fix keyring ref leak in
    join_session_keyring()").  Fortunately this cannot happen currently, but
    we really should be more careful.  Do this by adding and using a new
    error label at which the keyring reference is dropped.
    
    Signed-off-by: Eric Biggers <ebiggers@google.com>
    Signed-off-by: David Howells <dhowells@redhat.com>

diff --git a/security/keys/process_keys.c b/security/keys/process_keys.c
index 44451af828c0..8e98e5d1b5fb 100644
--- a/security/keys/process_keys.c
+++ b/security/keys/process_keys.c
@@ -793,21 +793,20 @@ long join_session_keyring(const char *name)
 			KEY_ALLOC_IN_QUOTA, NULL, NULL);
 		if (IS_ERR(keyring)) {
 			ret = PTR_ERR(keyring);
-			goto error2;
+			goto error_unlock_no_keyring;
 		}
 	} else if (IS_ERR(keyring)) {
 		ret = PTR_ERR(keyring);
-		goto error2;
+		goto error_unlock_no_keyring;
 	} else if (keyring == new->session_keyring) {
-		key_put(keyring);
 		ret = 0;
-		goto error2;
+		goto error_unlock;
 	}
 
 	/* we've got a keyring - now to install it */
 	ret = install_session_keyring_to_cred(new, keyring);
 	if (ret < 0)
-		goto error2;
+		goto error_unlock;
 
 	commit_creds(new);
 	mutex_unlock(&key_session_mutex);
@@ -817,8 +816,11 @@ long join_session_keyring(const char *name)
 okay:
 	return ret;
 
-error2:
+error_unlock_no_keyring:
+	keyring = NULL;
+error_unlock:
 	mutex_unlock(&key_session_mutex);
+	key_put(keyring);
 error:
 	abort_creds(new);
 	return ret;

Re: [PATCH] KEYS: put keyring if install_session_keyring_to_cred() fails

[Eric Biggers]

Hi David,

On Mon, Apr 03, 2017 at 04:42:05PM +0100, David Howells wrote:
> Eric Biggers <ebiggers3@gmail.com> wrote:
> 
> > +error3:
> > +	key_put(keyring);
> >  error2:
> >  	mutex_unlock(&key_session_mutex);
> 
> I would prefer the put to be outside of the locked section.  How about the
> attached instead?

[...]

> diff --git a/security/keys/process_keys.c b/security/keys/process_keys.c
> index 44451af828c0..8e98e5d1b5fb 100644
> --- a/security/keys/process_keys.c
> +++ b/security/keys/process_keys.c
> @@ -793,21 +793,20 @@ long join_session_keyring(const char *name)
>  			KEY_ALLOC_IN_QUOTA, NULL, NULL);
>  		if (IS_ERR(keyring)) {
>  			ret = PTR_ERR(keyring);
> -			goto error2;
> +			goto error_unlock_no_keyring;
>  		}
>  	} else if (IS_ERR(keyring)) {
>  		ret = PTR_ERR(keyring);
> -		goto error2;
> +		goto error_unlock_no_keyring;
>  	} else if (keyring == new->session_keyring) {
> -		key_put(keyring);
>  		ret = 0;
> -		goto error2;
> +		goto error_unlock;
>  	}
>  
>  	/* we've got a keyring - now to install it */
>  	ret = install_session_keyring_to_cred(new, keyring);
>  	if (ret < 0)
> -		goto error2;
> +		goto error_unlock;
>  
>  	commit_creds(new);
>  	mutex_unlock(&key_session_mutex);
> @@ -817,8 +816,11 @@ long join_session_keyring(const char *name)
>  okay:
>  	return ret;
>  
> -error2:
> +error_unlock_no_keyring:
> +	keyring = NULL;
> +error_unlock:
>  	mutex_unlock(&key_session_mutex);
> +	key_put(keyring);
>  error:
>  	abort_creds(new);
>  	return ret;

I guess I'm okay with either patch, but I think jumping to the error labels in
the way I proposed is more straightforward since it keeps things in the logical
order.  And I doubt there's much advantage in moving key_put() outside of the
mutex, given that key_session_mutex is only used for joining session keyrings so
shouldn't be contended, and key_put() hardly does any work anyway.

- Eric