Skip to contentSkip to conversation

fscrypt: reserve flags for hardware-wrapped keys feature

PATCH In review Eric Biggers 4 participants 4 messages Jan 16, 2020 – Jan 17, 2020 1 message not in the archive

Subsystems (4)
lore
Share & tools
lore

Eric Biggers

author maintainer
#0 lore raw

PATCH 1/1 fscrypt: reserve flags for hardware-wrapped keys feature

From: Eric Biggers <[email protected]>

Reserve flags for the hardware-wrapped keys feature which is being worked on [1]. FSCRYPT_POLICY_FLAG_HW_WRAPPED_KEY will denote that the encryption policy needs a hardware-wrapped key to be unlocked. FSCRYPT_ADD_KEY_FLAG_HW_WRAPPED will denote that the key being added is a hardware-wrapped key.

This reservation is tentative, and these codepoints may be reused if the feature is not upstreamed.

[1] https://android-review.googlesource.com/c/kernel/common/+/1200864

---
 Documentation/filesystems/fscrypt.rst | 5 +++--
 fs/crypto/keyring.c                   | 5 ++++-
 fs/crypto/policy.c                    | 4 +++-
 include/uapi/linux/fscrypt.h          | 9 ++++++---
 4 files changed, 16 insertions(+), 7 deletions(-)
Documentation/filesystems/fscrypt.rst +3 -2
index 9c53336d06a438..4c443d7b1fc6b5 100644
@@ -639,7 +639,8 @@ follows::
             struct fscrypt_key_specifier key_spec;
             __u32 raw_size;
             __u32 key_id;
-            __u32 __reserved[8];
+            __u32 flags;
+            __u32 __reserved[7];
             __u8 raw[];
     };
 
@@ -658,7 +659,7 @@ follows::
 
     struct fscrypt_provisioning_key_payload {
             __u32 type;
-            __u32 __reserved;
+            __u32 flags;
             __u8 raw[];
     };
 
fs/crypto/keyring.c +4 -1
index 098ff2e0f0bb41..fc27f5d08d7dbe 100644
@@ -477,7 +477,7 @@ static int fscrypt_provisioning_key_preparse(struct key_preparsed_payload *prep)
 	    payload->type != FSCRYPT_KEY_SPEC_TYPE_IDENTIFIER)
 		return -EINVAL;
 
-	if (payload->__reserved)
+	if (payload->flags)
 		return -EINVAL;
 
 	prep->payload.data[0] = kmemdup(payload, prep->datalen, GFP_KERNEL);
@@ -606,6 +606,9 @@ int fscrypt_ioctl_add_key(struct file *filp, void __user *_uarg)
 	if (!valid_key_spec(&arg.key_spec))
 		return -EINVAL;
 
+	if (arg.flags)
+		return -EINVAL;
+
 	if (memchr_inv(arg.__reserved, 0, sizeof(arg.__reserved)))
 		return -EINVAL;
 
fs/crypto/policy.c +3 -1
index f1cff83c151acf..36a2bb077b6910 100644
@@ -139,7 +139,9 @@ static bool fscrypt_supported_v2_policy(const struct fscrypt_policy_v2 *policy,
 		return false;
 	}
 
-	if (policy->flags & ~FSCRYPT_POLICY_FLAGS_VALID) {
+	if (policy->flags & ~(FSCRYPT_POLICY_FLAGS_PAD_MASK |
+			      FSCRYPT_POLICY_FLAG_DIRECT_KEY |
+			      FSCRYPT_POLICY_FLAG_IV_INO_LBLK_64)) {
 		fscrypt_warn(inode, "Unsupported encryption flags (0x%02x)",
 			     policy->flags);
 		return false;
include/uapi/linux/fscrypt.h +6 -3
index 0d8a6f47711c32..fad624a4c5feda 100644
@@ -19,7 +19,8 @@
 #define FSCRYPT_POLICY_FLAGS_PAD_MASK		0x03
 #define FSCRYPT_POLICY_FLAG_DIRECT_KEY		0x04
 #define FSCRYPT_POLICY_FLAG_IV_INO_LBLK_64	0x08
-#define FSCRYPT_POLICY_FLAGS_VALID		0x0F
+#define FSCRYPT_POLICY_FLAG_HW_WRAPPED_KEY	0x10
+#define FSCRYPT_POLICY_FLAGS_VALID		0x1F
 
 /* Encryption algorithms */
 #define FSCRYPT_MODE_AES_256_XTS		1
@@ -116,7 +117,7 @@ struct fscrypt_key_specifier {
  */
 struct fscrypt_provisioning_key_payload {
 	__u32 type;
-	__u32 __reserved;
+	__u32 flags;
 	__u8 raw[];
 };
 
@@ -125,7 +126,9 @@ struct fscrypt_add_key_arg {
 	struct fscrypt_key_specifier key_spec;
 	__u32 raw_size;
 	__u32 key_id;
-	__u32 __reserved[8];
+#define FSCRYPT_ADD_KEY_FLAG_HW_WRAPPED			0x00000001
+	__u32 flags;
+	__u32 __reserved[7];
 	__u8 raw[];
 };
 

base-commit: 2d8f7f119b0b2ce5e7ff0e8024b0763bf42b99c9

Signature · 1 line
2.25.0.rc1.283.g88dfdc4193-goog

A message that is not in the archive (sent off-list or not yet received): 20200116192008.35766-1-ebiggers-DgEjT+Ai2ygdnm+yROfE0A@public.gmane.org

#2 lore raw
Quoted 10 lines · Eric Biggers

From: Eric Biggers <[email protected]>

Reserve flags for the hardware-wrapped keys feature which is being worked on [1]. FSCRYPT_POLICY_FLAG_HW_WRAPPED_KEY will denote that the encryption policy needs a hardware-wrapped key to be unlocked. FSCRYPT_ADD_KEY_FLAG_HW_WRAPPED will denote that the key being added is a hardware-wrapped key.

This reservation is tentative, and these codepoints may be reused if the feature is not upstreamed.

NAK. While the feature itself sounds really useful we don't just reserve format bits for code not upstream.

maintainer Reviewed-by
#3 lore raw
Quoted 9 lines · Christoph Hellwig
Quoted 8 lines · Eric Biggers

From: Eric Biggers <[email protected]>

Reserve flags for the hardware-wrapped keys feature which is being worked on [1]. FSCRYPT_POLICY_FLAG_HW_WRAPPED_KEY will denote that the encryption policy needs a hardware-wrapped key to be unlocked. FSCRYPT_ADD_KEY_FLAG_HW_WRAPPED will denote that the key being added is a hardware-wrapped key.

Quoted 5 lines
Quoted 2 lines

This reservation is tentative, and these codepoints may be reused if the feature is not upstreamed.

NAK. While the feature itself sounds really useful we don't just reserve format bits for code not upstream.

I disagree; saving a codepoint to avoid accidental collision of a feature bit is a good and proper thing to do.

- Ted

______________________________________________________
Linux MTD discussion mailing list
http://lists.infradead.org/mailman/listinfo/linux-mtd/

#4 lore raw
Quoted 13 lines · Theodore Y. Ts'o
Quoted 12 lines · Christoph Hellwig
Quoted 10 lines · Eric Biggers

From: Eric Biggers <[email protected]>

Reserve flags for the hardware-wrapped keys feature which is being worked on [1]. FSCRYPT_POLICY_FLAG_HW_WRAPPED_KEY will denote that the encryption policy needs a hardware-wrapped key to be unlocked. FSCRYPT_ADD_KEY_FLAG_HW_WRAPPED will denote that the key being added is a hardware-wrapped key.

This reservation is tentative, and these codepoints may be reused if the feature is not upstreamed.

Quoted 7 lines
Quoted 2 lines

NAK. While the feature itself sounds really useful we don't just reserve format bits for code not upstream.

I disagree; saving a codepoint to avoid accidental collision of a feature bit is a good and proper thing to do.

What kind of "deadline" do you have for that feature to then be merged? I'm with Christoph here, we shouldn't be reserving bits for stuff not in mergable state, what's the rush?

thansk,

greg k-h