diff mbox

[V3] EVM: Allow userland to permit modification of EVM-protected metadata

Message ID 20171102221005.10190-1-mjg59@google.com (mailing list archive)
State New, archived
Headers show

Commit Message

Matthew Garrett Nov. 2, 2017, 10:10 p.m. UTC
When EVM is enabled it forbids modification of metadata are protected by
EVM unless there is already a valid EVM signature. If any modification
is made, the kernel will then generate a new EVM HMAC. However, this
does not map well on use cases which use only asymmetric EVM signatures,
as in this scenario the kernel is unable to generate new signatures.

This patch extends the /sys/kernel/security/evm interface to allow
userland to request that modification of these xattrs be permitted. This
is only permitted if no keys have already been loaded. In this
configuration, modifying the metadata will invalidate the EVM appraisal
on the file in question. This allows packaging systems to write out new
files, set the relevant extended attributes and then move them into
place.

There's also some refactoring of the use of evm_initialized in order to
avoid heading down codepaths that assume there's a key available.

Signed-off-by: Matthew Garrett <mjg59@google.com>
---
 Documentation/ABI/testing/evm      | 51 +++++++++++++++++++++++++-------------
 security/integrity/evm/evm.h       |  5 +++-
 security/integrity/evm/evm_main.c  | 32 ++++++++++++++++++------
 security/integrity/evm/evm_secfs.c | 14 +++++++++++
 4 files changed, 77 insertions(+), 25 deletions(-)

Comments

Mimi Zohar Nov. 3, 2017, 4:03 p.m. UTC | #1
On Thu, 2017-11-02 at 15:10 -0700, Matthew Garrett wrote:
> When EVM is enabled it forbids modification of metadata are protected by
> EVM unless there is already a valid EVM signature. If any modification
> is made, the kernel will then generate a new EVM HMAC. However, this
> does not map well on use cases which use only asymmetric EVM signatures,
> as in this scenario the kernel is unable to generate new signatures.
> 
> This patch extends the /sys/kernel/security/evm interface to allow
> userland to request that modification of these xattrs be permitted. This
> is only permitted if no keys have already been loaded. In this
> configuration, modifying the metadata will invalidate the EVM appraisal
> on the file in question. This allows packaging systems to write out new
> files, set the relevant extended attributes and then move them into
> place.
> 
> There's also some refactoring of the use of evm_initialized in order to
> avoid heading down codepaths that assume there's a key available.
> 
> Signed-off-by: Matthew Garrett <mjg59@google.com>
> ---
>  Documentation/ABI/testing/evm      | 51 +++++++++++++++++++++++++-------------
>  security/integrity/evm/evm.h       |  5 +++-
>  security/integrity/evm/evm_main.c  | 32 ++++++++++++++++++------
>  security/integrity/evm/evm_secfs.c | 14 +++++++++++
>  4 files changed, 77 insertions(+), 25 deletions(-)
> 
> diff --git a/Documentation/ABI/testing/evm b/Documentation/ABI/testing/evm
> index d2782afb0d96..b0f54fe8d0c6 100644
> --- a/Documentation/ABI/testing/evm
> +++ b/Documentation/ABI/testing/evm
> @@ -14,28 +14,45 @@ Description:
>  		generated either locally or remotely using an
>  		asymmetric key. These keys are loaded onto root's
>  		keyring using keyctl, and EVM is then enabled by
> -		echoing a value to <securityfs>/evm:
> +		echoing a value to <securityfs>/evm made up of the
> +		following bits:
> 
> -		1: enable HMAC validation and creation
> -		2: enable digital signature validation
> -		3: enable HMAC and digital signature validation and HMAC
> -		   creation
> +		Bit	  Effect
> +		0	  Enable HMAC validation and creation

The code and documentation do not seem to be in sync.  Dracut is
currently using 1 to indicate the HMAC key has been loaded.  

> +		1	  Enable digital signature validation
> +		2	  Permit modification of EVM-protected metadata at
> +			  runtime. Not supported if HMAC validation and
> +			  creation is enabled.
> +		31	  Disable further runtime modification of EVM policy
> 
> -		Further writes will be blocked if HMAC support is enabled or
> -		if bit 32 is set:
> +		For example:
> 
> -		echo 0x80000002 ><securityfs>/evm
> +		echo 1 ><securityfs>/evm
> 
> -		will enable digital signature validation and block
> -		further writes to <securityfs>/evm.
> +		will enable HMAC validation and creation
> 
> -		Until this is done, EVM can not create or validate the
> -		'security.evm' xattr, but returns INTEGRITY_UNKNOWN.
> -		Loading keys and signaling EVM should be done as early
> -		as possible.  Normally this is done in the initramfs,
> -		which has already been measured as part of the trusted
> -		boot.  For more information on creating and loading
> -		existing trusted/encrypted keys, refer to:
> +		echo 0x80000003 ><securityfs>/evm
> +
> +		will enable HMAC and digital signature validation and
> +		HMAC creation and disable all further modification of policy.
> +
> +		echo 0x80000006 ><securityfs>/evm

4 is missing above.

> +
> +		will enable digital signature validation, permit
> +		modification of EVM-protected metadata and
> +		disable all further modification of policy
> +
> +		Note that once a key has been loaded, it will no longer be
> +		possible to enable metadata modification.
> +
> +		Until key loading has been signaled EVM can not create
> +		or validate the 'security.evm' xattr, but returns
> +		INTEGRITY_UNKNOWN.  Loading keys and signaling EVM
> +		should be done as early as possible.  Normally this is
> +		done in the initramfs, which has already been measured
> +		as part of the trusted boot.  For more information on
> +		creating and loading existing trusted/encrypted keys,
> +		refer to:
>  		Documentation/keys-trusted-encrypted.txt. Both dracut
>  		(via 97masterkey and 98integrity) and systemd (via
>  		core/ima-setup) have support for loading keys at boot
> diff --git a/security/integrity/evm/evm.h b/security/integrity/evm/evm.h
> index 946efffcc389..6c63db19fbcf 100644
> --- a/security/integrity/evm/evm.h
> +++ b/security/integrity/evm/evm.h
> @@ -23,9 +23,12 @@
> 
>  #define EVM_INIT_HMAC	0x0001
>  #define EVM_INIT_X509	0x0002
> +#define EVM_ALLOW_METADATA_WRITES	0x0004
>  #define EVM_SETUP       0x80000000 /* userland has signaled key load */

Sorry for not commenting earlier on this bit's naming, but it's
confusing.  Perhaps something like EVM_SETUP_COMPLETED/FINISHED?

> 
> -#define EVM_INIT_MASK (EVM_INIT_HMAC | EVM_INIT_X509 | EVM_SETUP)
> +#define EVM_KEY_MASK (EVM_INIT_HMAC | EVM_INIT_X509)
> +#define EVM_INIT_MASK (EVM_INIT_HMAC | EVM_INIT_X509 | EVM_SETUP | \
> +		       EVM_ALLOW_METADATA_WRITES)
> 
>  extern int evm_initialized;
>  extern char *evm_hmac;
> diff --git a/security/integrity/evm/evm_main.c b/security/integrity/evm/evm_main.c
> index 675a835b6d6d..0788fd08b509 100644
> --- a/security/integrity/evm/evm_main.c
> +++ b/security/integrity/evm/evm_main.c
> @@ -73,6 +73,11 @@ static void __init evm_init_config(void)
>  	pr_info("HMAC attrs: 0x%x\n", evm_hmac_attrs);
>  }
> 
> +static bool evm_key_loaded(void)
> +{
> +	return (bool)(evm_initialized & EVM_KEY_MASK);
> +}
> +/

>  static int evm_find_protected_xattrs(struct dentry *dentry)
>  {
>  	struct inode *inode = d_backing_inode(dentry);
> @@ -243,7 +248,7 @@ enum integrity_status evm_verifyxattr(struct dentry *dentry,
>  				      void *xattr_value, size_t xattr_value_len,
>  				      struct integrity_iint_cache *iint)
>  {
> -	if (!evm_initialized || !evm_protected_xattr(xattr_name))
> +	if (!evm_key_loaded() || !evm_protected_xattr(xattr_name))
>  		return INTEGRITY_UNKNOWN;
> 
>  	if (!iint) {
> @@ -267,7 +272,7 @@ static enum integrity_status evm_verify_current_integrity(struct dentry *dentry)
>  {
>  	struct inode *inode = d_backing_inode(dentry);
> 
> -	if (!evm_initialized || !S_ISREG(inode->i_mode) || evm_fixmode)
> +	if (!evm_key_loaded() || !S_ISREG(inode->i_mode) || evm_fixmode)
>  		return 0;
>  	return evm_verify_hmac(dentry, NULL, NULL, 0, NULL);
>  }
> @@ -302,6 +307,7 @@ static int evm_protect_xattr(struct dentry *dentry, const char *xattr_name,
>  			return 0;
>  		goto out;
>  	}
> +
>  	evm_status = evm_verify_current_integrity(dentry);
>  	if (evm_status == INTEGRITY_NOXATTRS) {
>  		struct integrity_iint_cache *iint;
> @@ -348,6 +354,10 @@ int evm_inode_setxattr(struct dentry *dentry, const char *xattr_name,
>  {
>  	const struct evm_ima_xattr_data *xattr_data = xattr_value;
> 
> +	/* Policy permits modification of the protected xattrs */
> +	if (evm_initialized & EVM_ALLOW_METADATA_WRITES)
> +		return 0;
> +
>  	if (strcmp(xattr_name, XATTR_NAME_EVM) == 0) {
>  		if (!xattr_value_len)
>  			return -EINVAL;
> @@ -369,6 +379,10 @@ int evm_inode_setxattr(struct dentry *dentry, const char *xattr_name,
>   */
>  int evm_inode_removexattr(struct dentry *dentry, const char *xattr_name)
>  {
> +	/* Policy permits modification of the protected xattrs */
> +	if (evm_initialized & EVM_ALLOW_METADATA_WRITES)
> +		return 0;
> +
>  	return evm_protect_xattr(dentry, xattr_name, NULL, 0);
>  }
> 
> @@ -397,8 +411,8 @@ static void evm_reset_status(struct inode *inode)
>  void evm_inode_post_setxattr(struct dentry *dentry, const char *xattr_name,
>  			     const void *xattr_value, size_t xattr_value_len)
>  {
> -	if (!evm_initialized || (!evm_protected_xattr(xattr_name)
> -				 && !posix_xattr_acl(xattr_name)))
> +	if (!evm_key_loaded() || (!evm_protected_xattr(xattr_name)
> +				  && !posix_xattr_acl(xattr_name)))
>  		return;
> 
>  	evm_reset_status(dentry->d_inode);
> @@ -418,7 +432,7 @@ void evm_inode_post_setxattr(struct dentry *dentry, const char *xattr_name,
>   */
>  void evm_inode_post_removexattr(struct dentry *dentry, const char *xattr_name)
>  {
> -	if (!evm_initialized || !evm_protected_xattr(xattr_name))
> +	if (!evm_key_loaded() || !evm_protected_xattr(xattr_name))
>  		return;
> 
>  	evm_reset_status(dentry->d_inode);
> @@ -435,6 +449,10 @@ int evm_inode_setattr(struct dentry *dentry, struct iattr *attr)
>  	unsigned int ia_valid = attr->ia_valid;
>  	enum integrity_status evm_status;
> 
> +	/* Policy permits modification of the protected attrs */

Could we indicate that there is no HMAC key loaded.

> +	if (evm_initialized & EVM_ALLOW_METADATA_WRITES)
> +		return 0;
> +
>  	if (!(ia_valid & (ATTR_MODE | ATTR_UID | ATTR_GID)))
>  		return 0;
>  	evm_status = evm_verify_current_integrity(dentry);
> @@ -460,7 +478,7 @@ int evm_inode_setattr(struct dentry *dentry, struct iattr *attr)
>   */
>  void evm_inode_post_setattr(struct dentry *dentry, int ia_valid)
>  {
> -	if (!evm_initialized)
> +	if (!evm_key_loaded())
>  		return;
> 
>  	if (ia_valid & (ATTR_MODE | ATTR_UID | ATTR_GID))
> @@ -477,7 +495,7 @@ int evm_inode_init_security(struct inode *inode,
>  	struct evm_ima_xattr_data *xattr_data;
>  	int rc;
> 
> -	if (!evm_initialized || !evm_protected_xattr(lsm_xattr->name))
> +	if (!evm_key_loaded() || !evm_protected_xattr(lsm_xattr->name))
>  		return 0;
> 
>  	xattr_data = kzalloc(sizeof(*xattr_data), GFP_NOFS);
> diff --git a/security/integrity/evm/evm_secfs.c b/security/integrity/evm/evm_secfs.c
> index 319cf16d6603..ee8e3abb7dbb 100644
> --- a/security/integrity/evm/evm_secfs.c
> +++ b/security/integrity/evm/evm_secfs.c
> @@ -75,6 +75,14 @@ static ssize_t evm_write_key(struct file *file, const char __user *buf,
>  	if (!i || (i & ~EVM_INIT_MASK) != 0)
>  		return -EINVAL;
> 
> +	/* Don't allow a request to freshly enable metadata writes if
> +	 * keys are loaded.
> +	 */
> +	if ((i & EVM_ALLOW_METADATA_WRITES) &&
> +	    ((evm_initialized & EVM_KEY_MASK) != 0) &&
> +	    !(evm_initialized & EVM_ALLOW_METADATA_WRITES))

Ok, not sure that the "(evm_initialized & EVM_ALLOW_METADATA_WRITES)"
is needed, but it doesn't hurt.

Mimi

> +		return -EPERM;
> +
>  	if (i & EVM_INIT_HMAC) {
>  		ret = evm_init_key();
>  		if (ret != 0)
> @@ -85,6 +93,12 @@ static ssize_t evm_write_key(struct file *file, const char __user *buf,
> 
>  	evm_initialized |= i;
> 
> +	/* Don't allow protected metadata modification if a symmetric key
> +	 * is loaded
> +	 */
> +	if (evm_initialized & EVM_INIT_HMAC)
> +		evm_initialized &= ~(EVM_ALLOW_METADATA_WRITES);
> +
>  	return count;
>  }
>
Matthew Garrett Nov. 3, 2017, 6:49 p.m. UTC | #2
On Fri, Nov 3, 2017 at 9:03 AM, Mimi Zohar <zohar@linux.vnet.ibm.com> wrote:
> On Thu, 2017-11-02 at 15:10 -0700, Matthew Garrett wrote:
>> -             1: enable HMAC validation and creation
>> -             2: enable digital signature validation
>> -             3: enable HMAC and digital signature validation and HMAC
>> -                creation
>> +             Bit       Effect
>> +             0         Enable HMAC validation and creation
>
> The code and documentation do not seem to be in sync.  Dracut is
> currently using 1 to indicate the HMAC key has been loaded.

I've changed from describing the raw values to the bits they
correspond to, so bit 0 corresponds to a value of 1. I can switch back
to describing the raw values instead?

>>
>> +     /* Policy permits modification of the protected attrs */
>
> Could we indicate that there is no HMAC key loaded.

In the comment, or in kernel output?
>> +     /* Don't allow a request to freshly enable metadata writes if
>> +      * keys are loaded.
>> +      */
>> +     if ((i & EVM_ALLOW_METADATA_WRITES) &&
>> +         ((evm_initialized & EVM_KEY_MASK) != 0) &&
>> +         !(evm_initialized & EVM_ALLOW_METADATA_WRITES))
>
> Ok, not sure that the "(evm_initialized & EVM_ALLOW_METADATA_WRITES)"
> is needed, but it doesn't hurt.

Goal here was to allow:

echo 6 >evm
echo 7 >evm

to work without an error, but I guess that's not a big deal.
Mimi Zohar Nov. 3, 2017, 7:16 p.m. UTC | #3
On Fri, 2017-11-03 at 11:49 -0700, Matthew Garrett wrote:
> On Fri, Nov 3, 2017 at 9:03 AM, Mimi Zohar <zohar@linux.vnet.ibm.com> wrote:
> > On Thu, 2017-11-02 at 15:10 -0700, Matthew Garrett wrote:
> >> -             1: enable HMAC validation and creation
> >> -             2: enable digital signature validation
> >> -             3: enable HMAC and digital signature validation and HMAC
> >> -                creation
> >> +             Bit       Effect
> >> +             0         Enable HMAC validation and creation
> >
> > The code and documentation do not seem to be in sync.  Dracut is
> > currently using 1 to indicate the HMAC key has been loaded.
> 
> I've changed from describing the raw values to the bits they
> correspond to, so bit 0 corresponds to a value of 1. I can switch back
> to describing the raw values instead?

Ok, bits are fine.

> >>
> >> +     /* Policy permits modification of the protected attrs */
> >
> > Could we indicate that there is no HMAC key loaded.

Just as a reminder in the comment.

> In the comment, or in kernel output?
> >> +     /* Don't allow a request to freshly enable metadata writes if
> >> +      * keys are loaded.
> >> +      */
> >> +     if ((i & EVM_ALLOW_METADATA_WRITES) &&
> >> +         ((evm_initialized & EVM_KEY_MASK) != 0) &&
> >> +         !(evm_initialized & EVM_ALLOW_METADATA_WRITES))
> >
> > Ok, not sure that the "(evm_initialized & EVM_ALLOW_METADATA_WRITES)"
> > is needed, but it doesn't hurt.
> 
> Goal here was to allow:
> 
> echo 6 >evm
> echo 7 >evm
> 
> t work without an error, but I guess that's not a big deal.
> 
That's fine.

Thanks!
diff mbox

Patch

diff --git a/Documentation/ABI/testing/evm b/Documentation/ABI/testing/evm
index d2782afb0d96..b0f54fe8d0c6 100644
--- a/Documentation/ABI/testing/evm
+++ b/Documentation/ABI/testing/evm
@@ -14,28 +14,45 @@  Description:
 		generated either locally or remotely using an
 		asymmetric key. These keys are loaded onto root's
 		keyring using keyctl, and EVM is then enabled by
-		echoing a value to <securityfs>/evm:
+		echoing a value to <securityfs>/evm made up of the
+		following bits:
 
-		1: enable HMAC validation and creation
-		2: enable digital signature validation
-		3: enable HMAC and digital signature validation and HMAC
-		   creation
+		Bit	  Effect
+		0	  Enable HMAC validation and creation
+		1	  Enable digital signature validation
+		2	  Permit modification of EVM-protected metadata at
+			  runtime. Not supported if HMAC validation and
+			  creation is enabled.
+		31	  Disable further runtime modification of EVM policy
 
-		Further writes will be blocked if HMAC support is enabled or
-		if bit 32 is set:
+		For example:
 
-		echo 0x80000002 ><securityfs>/evm
+		echo 1 ><securityfs>/evm
 
-		will enable digital signature validation and block
-		further writes to <securityfs>/evm.
+		will enable HMAC validation and creation
 
-		Until this is done, EVM can not create or validate the
-		'security.evm' xattr, but returns INTEGRITY_UNKNOWN.
-		Loading keys and signaling EVM should be done as early
-		as possible.  Normally this is done in the initramfs,
-		which has already been measured as part of the trusted
-		boot.  For more information on creating and loading
-		existing trusted/encrypted keys, refer to:
+		echo 0x80000003 ><securityfs>/evm
+
+		will enable HMAC and digital signature validation and
+		HMAC creation and disable all further modification of policy.
+
+		echo 0x80000006 ><securityfs>/evm
+
+		will enable digital signature validation, permit
+		modification of EVM-protected metadata and
+		disable all further modification of policy
+
+		Note that once a key has been loaded, it will no longer be
+		possible to enable metadata modification.
+
+		Until key loading has been signaled EVM can not create
+		or validate the 'security.evm' xattr, but returns
+		INTEGRITY_UNKNOWN.  Loading keys and signaling EVM
+		should be done as early as possible.  Normally this is
+		done in the initramfs, which has already been measured
+		as part of the trusted boot.  For more information on
+		creating and loading existing trusted/encrypted keys,
+		refer to:
 		Documentation/keys-trusted-encrypted.txt. Both dracut
 		(via 97masterkey and 98integrity) and systemd (via
 		core/ima-setup) have support for loading keys at boot
diff --git a/security/integrity/evm/evm.h b/security/integrity/evm/evm.h
index 946efffcc389..6c63db19fbcf 100644
--- a/security/integrity/evm/evm.h
+++ b/security/integrity/evm/evm.h
@@ -23,9 +23,12 @@ 
 
 #define EVM_INIT_HMAC	0x0001
 #define EVM_INIT_X509	0x0002
+#define EVM_ALLOW_METADATA_WRITES	0x0004
 #define EVM_SETUP       0x80000000 /* userland has signaled key load */
 
-#define EVM_INIT_MASK (EVM_INIT_HMAC | EVM_INIT_X509 | EVM_SETUP)
+#define EVM_KEY_MASK (EVM_INIT_HMAC | EVM_INIT_X509)
+#define EVM_INIT_MASK (EVM_INIT_HMAC | EVM_INIT_X509 | EVM_SETUP | \
+		       EVM_ALLOW_METADATA_WRITES)
 
 extern int evm_initialized;
 extern char *evm_hmac;
diff --git a/security/integrity/evm/evm_main.c b/security/integrity/evm/evm_main.c
index 675a835b6d6d..0788fd08b509 100644
--- a/security/integrity/evm/evm_main.c
+++ b/security/integrity/evm/evm_main.c
@@ -73,6 +73,11 @@  static void __init evm_init_config(void)
 	pr_info("HMAC attrs: 0x%x\n", evm_hmac_attrs);
 }
 
+static bool evm_key_loaded(void)
+{
+	return (bool)(evm_initialized & EVM_KEY_MASK);
+}
+
 static int evm_find_protected_xattrs(struct dentry *dentry)
 {
 	struct inode *inode = d_backing_inode(dentry);
@@ -243,7 +248,7 @@  enum integrity_status evm_verifyxattr(struct dentry *dentry,
 				      void *xattr_value, size_t xattr_value_len,
 				      struct integrity_iint_cache *iint)
 {
-	if (!evm_initialized || !evm_protected_xattr(xattr_name))
+	if (!evm_key_loaded() || !evm_protected_xattr(xattr_name))
 		return INTEGRITY_UNKNOWN;
 
 	if (!iint) {
@@ -267,7 +272,7 @@  static enum integrity_status evm_verify_current_integrity(struct dentry *dentry)
 {
 	struct inode *inode = d_backing_inode(dentry);
 
-	if (!evm_initialized || !S_ISREG(inode->i_mode) || evm_fixmode)
+	if (!evm_key_loaded() || !S_ISREG(inode->i_mode) || evm_fixmode)
 		return 0;
 	return evm_verify_hmac(dentry, NULL, NULL, 0, NULL);
 }
@@ -302,6 +307,7 @@  static int evm_protect_xattr(struct dentry *dentry, const char *xattr_name,
 			return 0;
 		goto out;
 	}
+
 	evm_status = evm_verify_current_integrity(dentry);
 	if (evm_status == INTEGRITY_NOXATTRS) {
 		struct integrity_iint_cache *iint;
@@ -348,6 +354,10 @@  int evm_inode_setxattr(struct dentry *dentry, const char *xattr_name,
 {
 	const struct evm_ima_xattr_data *xattr_data = xattr_value;
 
+	/* Policy permits modification of the protected xattrs */
+	if (evm_initialized & EVM_ALLOW_METADATA_WRITES)
+		return 0;
+
 	if (strcmp(xattr_name, XATTR_NAME_EVM) == 0) {
 		if (!xattr_value_len)
 			return -EINVAL;
@@ -369,6 +379,10 @@  int evm_inode_setxattr(struct dentry *dentry, const char *xattr_name,
  */
 int evm_inode_removexattr(struct dentry *dentry, const char *xattr_name)
 {
+	/* Policy permits modification of the protected xattrs */
+	if (evm_initialized & EVM_ALLOW_METADATA_WRITES)
+		return 0;
+
 	return evm_protect_xattr(dentry, xattr_name, NULL, 0);
 }
 
@@ -397,8 +411,8 @@  static void evm_reset_status(struct inode *inode)
 void evm_inode_post_setxattr(struct dentry *dentry, const char *xattr_name,
 			     const void *xattr_value, size_t xattr_value_len)
 {
-	if (!evm_initialized || (!evm_protected_xattr(xattr_name)
-				 && !posix_xattr_acl(xattr_name)))
+	if (!evm_key_loaded() || (!evm_protected_xattr(xattr_name)
+				  && !posix_xattr_acl(xattr_name)))
 		return;
 
 	evm_reset_status(dentry->d_inode);
@@ -418,7 +432,7 @@  void evm_inode_post_setxattr(struct dentry *dentry, const char *xattr_name,
  */
 void evm_inode_post_removexattr(struct dentry *dentry, const char *xattr_name)
 {
-	if (!evm_initialized || !evm_protected_xattr(xattr_name))
+	if (!evm_key_loaded() || !evm_protected_xattr(xattr_name))
 		return;
 
 	evm_reset_status(dentry->d_inode);
@@ -435,6 +449,10 @@  int evm_inode_setattr(struct dentry *dentry, struct iattr *attr)
 	unsigned int ia_valid = attr->ia_valid;
 	enum integrity_status evm_status;
 
+	/* Policy permits modification of the protected attrs */
+	if (evm_initialized & EVM_ALLOW_METADATA_WRITES)
+		return 0;
+
 	if (!(ia_valid & (ATTR_MODE | ATTR_UID | ATTR_GID)))
 		return 0;
 	evm_status = evm_verify_current_integrity(dentry);
@@ -460,7 +478,7 @@  int evm_inode_setattr(struct dentry *dentry, struct iattr *attr)
  */
 void evm_inode_post_setattr(struct dentry *dentry, int ia_valid)
 {
-	if (!evm_initialized)
+	if (!evm_key_loaded())
 		return;
 
 	if (ia_valid & (ATTR_MODE | ATTR_UID | ATTR_GID))
@@ -477,7 +495,7 @@  int evm_inode_init_security(struct inode *inode,
 	struct evm_ima_xattr_data *xattr_data;
 	int rc;
 
-	if (!evm_initialized || !evm_protected_xattr(lsm_xattr->name))
+	if (!evm_key_loaded() || !evm_protected_xattr(lsm_xattr->name))
 		return 0;
 
 	xattr_data = kzalloc(sizeof(*xattr_data), GFP_NOFS);
diff --git a/security/integrity/evm/evm_secfs.c b/security/integrity/evm/evm_secfs.c
index 319cf16d6603..ee8e3abb7dbb 100644
--- a/security/integrity/evm/evm_secfs.c
+++ b/security/integrity/evm/evm_secfs.c
@@ -75,6 +75,14 @@  static ssize_t evm_write_key(struct file *file, const char __user *buf,
 	if (!i || (i & ~EVM_INIT_MASK) != 0)
 		return -EINVAL;
 
+	/* Don't allow a request to freshly enable metadata writes if
+	 * keys are loaded.
+	 */
+	if ((i & EVM_ALLOW_METADATA_WRITES) &&
+	    ((evm_initialized & EVM_KEY_MASK) != 0) &&
+	    !(evm_initialized & EVM_ALLOW_METADATA_WRITES))
+		return -EPERM;
+
 	if (i & EVM_INIT_HMAC) {
 		ret = evm_init_key();
 		if (ret != 0)
@@ -85,6 +93,12 @@  static ssize_t evm_write_key(struct file *file, const char __user *buf,
 
 	evm_initialized |= i;
 
+	/* Don't allow protected metadata modification if a symmetric key
+	 * is loaded
+	 */
+	if (evm_initialized & EVM_INIT_HMAC)
+		evm_initialized &= ~(EVM_ALLOW_METADATA_WRITES);
+
 	return count;
 }