fscrypt: add support for hardware-wrapped keys

Add support for hardware-wrapped keys to fscrypt. Such keys are
protected from certain attacks, such as cold boot attacks. For more
information, see the "Hardware-wrapped keys" section of
Documentation/block/inline-encryption.rst.

To support hardware-wrapped keys in fscrypt, we allow the fscrypt master
keys to be hardware-wrapped. File contents encryption is done by
passing the wrapped key to the inline encryption hardware via
blk-crypto. Other fscrypt operations such as filenames encryption
continue to be done by the kernel, using the "software secret" which the
hardware derives. For more information, see the documentation which
this patch adds to Documentation/filesystems/fscrypt.rst.

Note that this feature doesn't require any filesystem-specific changes.
However it does depend on inline encryption support, and thus currently
it is only applicable to ext4 and f2fs.

The version of this feature introduced by this patch is mostly
equivalent to the version that has existed downstream in the Android
Common Kernels since 2020. However, a couple fixes are included.
First, the flags field in struct fscrypt_add_key_arg is now placed in
the proper location. Second, key identifiers for HW-wrapped keys are
now derived using a distinct HKDF context byte; this fixes a bug where a
raw key could have the same identifier as a HW-wrapped key. Note that
as a result of these fixes, the version of this feature introduced by
this patch is not UAPI or on-disk format compatible with the version in
the Android Common Kernels, though the divergence is limited to just
those specific fixes. This version should be used going forwards.

This patch has been heavily rewritten from the original version by
Gaurav Kashyap <[e-mail]> and
Barani Muthukumaran <[e-mail]>.

Tested-by: Bartosz Golaszewski <[e-mail]> # sm8650
Link: https://lore.kernel.org/r/[e-mail]
Signed-off-by: Eric Biggers <[e-mail]>

Documentation/filesystems/fscrypt.rst

···
 --------------
 
 fscrypt (and storage encryption in general) can only provide limited
-protection, if any at all, against online attacks.  In detail:
+protection against online attacks.  In detail:
 
 Side-channel attacks
 ~~~~~~~~~~~~~~~~~~~~
···
 be enforced by kernel *code* and therefore would be largely redundant
 with the wide variety of access control mechanisms already available.)
 
-Kernel memory compromise
-~~~~~~~~~~~~~~~~~~~~~~~~
+Read-only kernel memory compromise
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
 
-An attacker who compromises the system enough to read from arbitrary
-memory, e.g. by mounting a physical attack or by exploiting a kernel
-security vulnerability, can compromise all encryption keys that are
-currently in use.
+Unless `hardware-wrapped keys`_ are used, an attacker who gains the
+ability to read from arbitrary kernel memory, e.g. by mounting a
+physical attack or by exploiting a kernel security vulnerability, can
+compromise all fscrypt keys that are currently in-use.  This also
+extends to cold boot attacks; if the system is suddenly powered off,
+keys the system was using may remain in memory for a short time.
 
-However, fscrypt allows encryption keys to be removed from the kernel,
-which may protect them from later compromise.
+However, if hardware-wrapped keys are used, then the fscrypt master
+keys and file contents encryption keys (but not other types of fscrypt
+subkeys such as filenames encryption keys) are protected from
+compromises of arbitrary kernel memory.
+
+In addition, fscrypt allows encryption keys to be removed from the
+kernel, which may protect them from later compromise.
 
 In more detail, the FS_IOC_REMOVE_ENCRYPTION_KEY ioctl (or the
 FS_IOC_REMOVE_ENCRYPTION_KEY_ALL_USERS ioctl) can wipe a master
···
   accelerator hardware (if used by the crypto API to implement any of
   the algorithms), or in other places not explicitly considered here.
 
+Full system compromise
+~~~~~~~~~~~~~~~~~~~~~~
+
+An attacker who gains "root" access and/or the ability to execute
+arbitrary kernel code can freely exfiltrate data that is protected by
+any in-use fscrypt keys.  Thus, usually fscrypt provides no meaningful
+protection in this scenario.  (Data that is protected by a key that is
+absent throughout the entire attack remains protected, modulo the
+limitations of key removal mentioned above in the case where the key
+was removed prior to the attack.)
+
+However, if `hardware-wrapped keys`_ are used, such attackers will be
+unable to exfiltrate the master keys or file contents keys in a form
+that will be usable after the system is powered off.  This may be
+useful if the attacker is significantly time-limited and/or
+bandwidth-limited, so they can only exfiltrate some data and need to
+rely on a later offline attack to exfiltrate the rest of it.
+
 Limitations of v1 policies
 ~~~~~~~~~~~~~~~~~~~~~~~~~~
 
···
 
 Key hierarchy
 =============
+
+Note: this section assumes the use of raw keys rather than
+hardware-wrapped keys.  The use of hardware-wrapped keys modifies the
+key hierarchy slightly.  For details, see `Hardware-wrapped keys`_.
 
 Master Keys
 -----------
···
             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[];
     };
 
···
 
     struct fscrypt_provisioning_key_payload {
             __u32 type;
-            __u32 __reserved;
+            __u32 flags;
             __u8 raw[];
     };
 
···
   Alternatively, if ``key_id`` is nonzero, this field must be 0, since
   in that case the size is implied by the specified Linux keyring key.
 
-- ``key_id`` is 0 if the raw key is given directly in the ``raw``
-  field.  Otherwise ``key_id`` is the ID of a Linux keyring key of
-  type "fscrypt-provisioning" whose payload is
-  struct fscrypt_provisioning_key_payload whose ``raw`` field contains
-  the raw key and whose ``type`` field matches ``key_spec.type``.
-  Since ``raw`` is variable-length, the total size of this key's
-  payload must be ``sizeof(struct fscrypt_provisioning_key_payload)``
-  plus the raw key size.  The process must have Search permission on
-  this key.
+- ``key_id`` is 0 if the key is given directly in the ``raw`` field.
+  Otherwise ``key_id`` is the ID of a Linux keyring key of type
+  "fscrypt-provisioning" whose payload is struct
+  fscrypt_provisioning_key_payload whose ``raw`` field contains the
+  key, whose ``type`` field matches ``key_spec.type``, and whose
+  ``flags`` field matches ``flags``.  Since ``raw`` is
+  variable-length, the total size of this key's payload must be
+  ``sizeof(struct fscrypt_provisioning_key_payload)`` plus the number
+  of key bytes.  The process must have Search permission on this key.
 
-  Most users should leave this 0 and specify the raw key directly.
-  The support for specifying a Linux keyring key is intended mainly to
+  Most users should leave this 0 and specify the key directly.  The
+  support for specifying a Linux keyring key is intended mainly to
   allow re-adding keys after a filesystem is unmounted and re-mounted,
-  without having to store the raw keys in userspace memory.
+  without having to store the keys in userspace memory.
+
+- ``flags`` contains optional flags from ``<linux/fscrypt.h>``:
+
+  - FSCRYPT_ADD_KEY_FLAG_HW_WRAPPED: This denotes that the key is a
+    hardware-wrapped key.  See `Hardware-wrapped keys`_.  This flag
+    can't be used if FSCRYPT_KEY_SPEC_TYPE_DESCRIPTOR is used.
 
 - ``raw`` is a variable-length field which must contain the actual
   key, ``raw_size`` bytes long.  Alternatively, if ``key_id`` is
-  nonzero, then this field is unused.
+  nonzero, then this field is unused.  Note that despite being named
+  ``raw``, if FSCRYPT_ADD_KEY_FLAG_HW_WRAPPED is specified then it
+  will contain a wrapped key, not a raw key.
 
 For v2 policy keys, the kernel keeps track of which user (identified
 by effective user ID) added the key, and only allows the key to be
···
 FS_IOC_ADD_ENCRYPTION_KEY may also be used to add a v2 policy key
 *again*, even if it's already added by other user(s).  In this case,
 FS_IOC_ADD_ENCRYPTION_KEY will just install a claim to the key for the
-current user, rather than actually add the key again (but the raw key
-must still be provided, as a proof of knowledge).
+current user, rather than actually add the key again (but the key must
+still be provided, as a proof of knowledge).
 
 FS_IOC_ADD_ENCRYPTION_KEY returns 0 if either the key or a claim to
 the key was either added or already exists.
···
 
 - ``EACCES``: FSCRYPT_KEY_SPEC_TYPE_DESCRIPTOR was specified, but the
   caller does not have the CAP_SYS_ADMIN capability in the initial
-  user namespace; or the raw key was specified by Linux key ID but the
+  user namespace; or the key was specified by Linux key ID but the
   process lacks Search permission on the key.
+- ``EBADMSG``: invalid hardware-wrapped key
 - ``EDQUOT``: the key quota for this user would be exceeded by adding
   the key
 - ``EINVAL``: invalid key size or key specifier type, or reserved bits
   were set
-- ``EKEYREJECTED``: the raw key was specified by Linux key ID, but the
-  key has the wrong type
-- ``ENOKEY``: the raw key was specified by Linux key ID, but no key
-  exists with that ID
+- ``EKEYREJECTED``: the key was specified by Linux key ID, but the key
+  has the wrong type
+- ``ENOKEY``: the key was specified by Linux key ID, but no key exists
+  with that ID
 - ``ENOTTY``: this type of filesystem does not implement encryption
 - ``EOPNOTSUPP``: the kernel was not configured with encryption
   support for this filesystem, or the filesystem superblock has not
-  had encryption enabled on it
+  had encryption enabled on it; or a hardware wrapped key was specified
+  but the filesystem does not support inline encryption or the hardware
+  does not support hardware-wrapped keys
 
 Legacy method
 ~~~~~~~~~~~~~
···
 These ioctls don't work on keys that were added via the legacy
 process-subscribed keyrings mechanism.
 
-Before using these ioctls, read the `Kernel memory compromise`_
-section for a discussion of the security goals and limitations of
-these ioctls.
+Before using these ioctls, read the `Online attacks`_ section for a
+discussion of the security goals and limitations of these ioctls.
 
 FS_IOC_REMOVE_ENCRYPTION_KEY
 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
···
 (e.g. support for the needed encryption algorithm and data unit size)
 and where blk-crypto-fallback is unusable.  (For blk-crypto-fallback
 to be usable, it must be enabled in the kernel configuration with
-CONFIG_BLK_INLINE_ENCRYPTION_FALLBACK=y.)
+CONFIG_BLK_INLINE_ENCRYPTION_FALLBACK=y, and the file must be
+protected by a raw key rather than a hardware-wrapped key.)
 
 Currently fscrypt always uses the filesystem block size (which is
 usually 4096 bytes) as the data unit size.  Therefore, it can only use
···
 
 Inline encryption doesn't affect the ciphertext or other aspects of
 the on-disk format, so users may freely switch back and forth between
-using "inlinecrypt" and not using "inlinecrypt".
+using "inlinecrypt" and not using "inlinecrypt".  An exception is that
+files that are protected by a hardware-wrapped key can only be
+encrypted/decrypted by the inline encryption hardware and therefore
+can only be accessed when the "inlinecrypt" mount option is used.  For
+more information about hardware-wrapped keys, see below.
+
+Hardware-wrapped keys
+---------------------
+
+fscrypt supports using *hardware-wrapped keys* when the inline
+encryption hardware supports it.  Such keys are only present in kernel
+memory in wrapped (encrypted) form; they can only be unwrapped
+(decrypted) by the inline encryption hardware and are temporally bound
+to the current boot.  This prevents the keys from being compromised if
+kernel memory is leaked.  This is done without limiting the number of
+keys that can be used and while still allowing the execution of
+cryptographic tasks that are tied to the same key but can't use inline
+encryption hardware, e.g. filenames encryption.
+
+Note that hardware-wrapped keys aren't specific to fscrypt; they are a
+block layer feature (part of *blk-crypto*).  For more details about
+hardware-wrapped keys, see the block layer documentation at
+:ref:`Documentation/block/inline-encryption.rst
+<hardware_wrapped_keys>`.  The rest of this section just focuses on
+the details of how fscrypt can use hardware-wrapped keys.
+
+fscrypt supports hardware-wrapped keys by allowing the fscrypt master
+keys to be hardware-wrapped keys as an alternative to raw keys.  To
+add a hardware-wrapped key with `FS_IOC_ADD_ENCRYPTION_KEY`_,
+userspace must specify FSCRYPT_ADD_KEY_FLAG_HW_WRAPPED in the
+``flags`` field of struct fscrypt_add_key_arg and also in the
+``flags`` field of struct fscrypt_provisioning_key_payload when
+applicable.  The key must be in ephemerally-wrapped form, not
+long-term wrapped form.
+
+Some limitations apply.  First, files protected by a hardware-wrapped
+key are tied to the system's inline encryption hardware.  Therefore
+they can only be accessed when the "inlinecrypt" mount option is used,
+and they can't be included in portable filesystem images.  Second,
+currently the hardware-wrapped key support is only compatible with
+`IV_INO_LBLK_64 policies`_ and `IV_INO_LBLK_32 policies`_, as it
+assumes that there is just one file contents encryption key per
+fscrypt master key rather than one per file.  Future work may address
+this limitation by passing per-file nonces down the storage stack to
+allow the hardware to derive per-file keys.
+
+Implementation-wise, to encrypt/decrypt the contents of files that are
+protected by a hardware-wrapped key, fscrypt uses blk-crypto,
+attaching the hardware-wrapped key to the bio crypt contexts.  As is
+the case with raw keys, the block layer will program the key into a
+keyslot when it isn't already in one.  However, when programming a
+hardware-wrapped key, the hardware doesn't program the given key
+directly into a keyslot but rather unwraps it (using the hardware's
+ephemeral wrapping key) and derives the inline encryption key from it.
+The inline encryption key is the key that actually gets programmed
+into a keyslot, and it is never exposed to software.
+
+However, fscrypt doesn't just do file contents encryption; it also
+uses its master keys to derive filenames encryption keys, key
+identifiers, and sometimes some more obscure types of subkeys such as
+dirhash keys.  So even with file contents encryption out of the
+picture, fscrypt still needs a raw key to work with.  To get such a
+key from a hardware-wrapped key, fscrypt asks the inline encryption
+hardware to derive a cryptographically isolated "software secret" from
+the hardware-wrapped key.  fscrypt uses this "software secret" to key
+its KDF to derive all subkeys other than file contents keys.
+
+Note that this implies that the hardware-wrapped key feature only
+protects the file contents encryption keys.  It doesn't protect other
+fscrypt subkeys such as filenames encryption keys.
 
 Direct I/O support
 ==================

fs/crypto/fscrypt_private.h

···
 #define _FSCRYPT_PRIVATE_H
 
 #include <linux/fscrypt.h>
+#include <linux/minmax.h>
 #include <linux/siphash.h>
 #include <crypto/hash.h>
 #include <linux/blk-crypto.h>
···
  * absolute minimum, which applies when only 128-bit encryption is used.
  */
 #define FSCRYPT_MIN_KEY_SIZE	16
+
+/* Maximum size of a raw fscrypt master key */
+#define FSCRYPT_MAX_RAW_KEY_SIZE	64
+
+/* Maximum size of a hardware-wrapped fscrypt master key */
+#define FSCRYPT_MAX_HW_WRAPPED_KEY_SIZE	BLK_CRYPTO_MAX_HW_WRAPPED_KEY_SIZE
+
+/* Maximum size of an fscrypt master key across both key types */
+#define FSCRYPT_MAX_ANY_KEY_SIZE \
+	MAX(FSCRYPT_MAX_RAW_KEY_SIZE, FSCRYPT_MAX_HW_WRAPPED_KEY_SIZE)
+
+/*
+ * FSCRYPT_MAX_KEY_SIZE is defined in the UAPI header, but the addition of
+ * hardware-wrapped keys has made it misleading as it's only for raw keys.
+ * Don't use it in kernel code; use one of the above constants instead.
+ */
+#undef FSCRYPT_MAX_KEY_SIZE
 
 #define FSCRYPT_CONTEXT_V1	1
 #define FSCRYPT_CONTEXT_V2	2
···
  * outputs are unique and cryptographically isolated, i.e. knowledge of one
  * output doesn't reveal another.
  */
-#define HKDF_CONTEXT_KEY_IDENTIFIER	1 /* info=<empty>		*/
+#define HKDF_CONTEXT_KEY_IDENTIFIER_FOR_RAW_KEY	1 /* info=<empty>	*/
 #define HKDF_CONTEXT_PER_FILE_ENC_KEY	2 /* info=file_nonce		*/
 #define HKDF_CONTEXT_DIRECT_KEY		3 /* info=mode_num		*/
 #define HKDF_CONTEXT_IV_INO_LBLK_64_KEY	4 /* info=mode_num||fs_uuid	*/
 #define HKDF_CONTEXT_DIRHASH_KEY	5 /* info=file_nonce		*/
 #define HKDF_CONTEXT_IV_INO_LBLK_32_KEY	6 /* info=mode_num||fs_uuid	*/
 #define HKDF_CONTEXT_INODE_HASH_KEY	7 /* info=<empty>		*/
+#define HKDF_CONTEXT_KEY_IDENTIFIER_FOR_HW_WRAPPED_KEY \
+					8 /* info=<empty>		*/
 
 int fscrypt_hkdf_expand(const struct fscrypt_hkdf *hkdf, u8 context,
 			const u8 *info, unsigned int infolen,
···
 
 /* inline_crypt.c */
 #ifdef CONFIG_FS_ENCRYPTION_INLINE_CRYPT
-int fscrypt_select_encryption_impl(struct fscrypt_inode_info *ci);
+int fscrypt_select_encryption_impl(struct fscrypt_inode_info *ci,
+				   bool is_hw_wrapped_key);
 
 static inline bool
 fscrypt_using_inline_encryption(const struct fscrypt_inode_info *ci)
···
 }
 
 int fscrypt_prepare_inline_crypt_key(struct fscrypt_prepared_key *prep_key,
-				     const u8 *raw_key,
+				     const u8 *key_bytes, size_t key_size,
+				     bool is_hw_wrapped,
 				     const struct fscrypt_inode_info *ci);
 
 void fscrypt_destroy_inline_crypt_key(struct super_block *sb,
 				      struct fscrypt_prepared_key *prep_key);
+
+int fscrypt_derive_sw_secret(struct super_block *sb,
+			     const u8 *wrapped_key, size_t wrapped_key_size,
+			     u8 sw_secret[BLK_CRYPTO_SW_SECRET_SIZE]);
 
 /*
  * Check whether the crypto transform or blk-crypto key has been allocated in
···
 
 #else /* CONFIG_FS_ENCRYPTION_INLINE_CRYPT */
 
-static inline int fscrypt_select_encryption_impl(struct fscrypt_inode_info *ci)
+static inline int fscrypt_select_encryption_impl(struct fscrypt_inode_info *ci,
+						 bool is_hw_wrapped_key)
 {
 	return 0;
 }
···
 
 static inline int
 fscrypt_prepare_inline_crypt_key(struct fscrypt_prepared_key *prep_key,
-				 const u8 *raw_key,
+				 const u8 *key_bytes, size_t key_size,
+				 bool is_hw_wrapped,
 				 const struct fscrypt_inode_info *ci)
 {
 	WARN_ON_ONCE(1);
···
 fscrypt_destroy_inline_crypt_key(struct super_block *sb,
 				 struct fscrypt_prepared_key *prep_key)
 {
+}
+
+static inline int
+fscrypt_derive_sw_secret(struct super_block *sb,
+			 const u8 *wrapped_key, size_t wrapped_key_size,
+			 u8 sw_secret[BLK_CRYPTO_SW_SECRET_SIZE])
+{
+	fscrypt_warn(NULL, "kernel doesn't support hardware-wrapped keys");
+	return -EOPNOTSUPP;
 }
 
 static inline bool
···
 struct fscrypt_master_key_secret {
 
 	/*
-	 * For v2 policy keys: HKDF context keyed by this master key.
-	 * For v1 policy keys: not set (hkdf.hmac_tfm == NULL).
+	 * The KDF with which subkeys of this key can be derived.
+	 *
+	 * For v1 policy keys, this isn't applicable and won't be set.
+	 * Otherwise, this KDF will be keyed by this master key if
+	 * ->is_hw_wrapped=false, or by the "software secret" that hardware
+	 * derived from this master key if ->is_hw_wrapped=true.
 	 */
 	struct fscrypt_hkdf	hkdf;
 
 	/*
-	 * Size of the raw key in bytes.  This remains set even if ->raw was
+	 * True if this key is a hardware-wrapped key; false if this key is a
+	 * raw key (i.e. a "software key").  For v1 policy keys this will always
+	 * be false, as v1 policy support is a legacy feature which doesn't
+	 * support newer functionality such as hardware-wrapped keys.
+	 */
+	bool			is_hw_wrapped;
+
+	/*
+	 * Size of the key in bytes.  This remains set even if ->bytes was
 	 * zeroized due to no longer being needed.  I.e. we still remember the
 	 * size of the key even if we don't need to remember the key itself.
 	 */
 	u32			size;
 
-	/* For v1 policy keys: the raw key.  Wiped for v2 policy keys. */
-	u8			raw[FSCRYPT_MAX_KEY_SIZE];
+	/*
+	 * The bytes of the key, when still needed.  This can be either a raw
+	 * key or a hardware-wrapped key, as indicated by ->is_hw_wrapped.  In
+	 * the case of a raw, v2 policy key, there is no need to remember the
+	 * actual key separately from ->hkdf so this field will be zeroized as
+	 * soon as ->hkdf is initialized.
+	 */
+	u8			bytes[FSCRYPT_MAX_ANY_KEY_SIZE];
 
 } __randomize_layout;
 

fs/crypto/hkdf.c

···
 // SPDX-License-Identifier: GPL-2.0
 /*
- * This is used to derive keys from the fscrypt master keys.
+ * This is used to derive keys from the fscrypt master keys (or from the
+ * "software secrets" which hardware derives from the fscrypt master keys, in
+ * the case that the fscrypt master keys are hardware-wrapped keys).
  *
  * Copyright 2019 Google LLC
  */

fs/crypto/inline_crypt.c

···
 }
 
 /* Enable inline encryption for this file if supported. */
-int fscrypt_select_encryption_impl(struct fscrypt_inode_info *ci)
+int fscrypt_select_encryption_impl(struct fscrypt_inode_info *ci,
+				   bool is_hw_wrapped_key)
 {
 	const struct inode *inode = ci->ci_inode;
 	struct super_block *sb = inode->i_sb;
···
 	crypto_cfg.crypto_mode = ci->ci_mode->blk_crypto_mode;
 	crypto_cfg.data_unit_size = 1U << ci->ci_data_unit_bits;
 	crypto_cfg.dun_bytes = fscrypt_get_dun_bytes(ci);
-	crypto_cfg.key_type = BLK_CRYPTO_KEY_TYPE_RAW;
+	crypto_cfg.key_type = is_hw_wrapped_key ?
+		BLK_CRYPTO_KEY_TYPE_HW_WRAPPED : BLK_CRYPTO_KEY_TYPE_RAW;
 
 	devs = fscrypt_get_devices(sb, &num_devs);
 	if (IS_ERR(devs))
···
 }
 
 int fscrypt_prepare_inline_crypt_key(struct fscrypt_prepared_key *prep_key,
-				     const u8 *raw_key,
+				     const u8 *key_bytes, size_t key_size,
+				     bool is_hw_wrapped,
 				     const struct fscrypt_inode_info *ci)
 {
 	const struct inode *inode = ci->ci_inode;
 	struct super_block *sb = inode->i_sb;
 	enum blk_crypto_mode_num crypto_mode = ci->ci_mode->blk_crypto_mode;
+	enum blk_crypto_key_type key_type = is_hw_wrapped ?
+		BLK_CRYPTO_KEY_TYPE_HW_WRAPPED : BLK_CRYPTO_KEY_TYPE_RAW;
 	struct blk_crypto_key *blk_key;
 	struct block_device **devs;
 	unsigned int num_devs;
···
 	if (!blk_key)
 		return -ENOMEM;
 
-	err = blk_crypto_init_key(blk_key, raw_key, ci->ci_mode->keysize,
-				  BLK_CRYPTO_KEY_TYPE_RAW, crypto_mode,
-				  fscrypt_get_dun_bytes(ci),
+	err = blk_crypto_init_key(blk_key, key_bytes, key_size, key_type,
+				  crypto_mode, fscrypt_get_dun_bytes(ci),
 				  1U << ci->ci_data_unit_bits);
 	if (err) {
 		fscrypt_err(inode, "error %d initializing blk-crypto key", err);
···
 		kfree(devs);
 	}
 	kfree_sensitive(blk_key);
+}
+
+/*
+ * Ask the inline encryption hardware to derive the software secret from a
+ * hardware-wrapped key.  Returns -EOPNOTSUPP if hardware-wrapped keys aren't
+ * supported on this filesystem or hardware.
+ */
+int fscrypt_derive_sw_secret(struct super_block *sb,
+			     const u8 *wrapped_key, size_t wrapped_key_size,
+			     u8 sw_secret[BLK_CRYPTO_SW_SECRET_SIZE])
+{
+	int err;
+
+	/* The filesystem must be mounted with -o inlinecrypt. */
+	if (!(sb->s_flags & SB_INLINECRYPT)) {
+		fscrypt_warn(NULL,
+			     "%s: filesystem not mounted with inlinecrypt\n",
+			     sb->s_id);
+		return -EOPNOTSUPP;
+	}
+
+	err = blk_crypto_derive_sw_secret(sb->s_bdev, wrapped_key,
+					  wrapped_key_size, sw_secret);
+	if (err == -EOPNOTSUPP)
+		fscrypt_warn(NULL,
+			     "%s: block device doesn't support hardware-wrapped keys\n",
+			     sb->s_id);
+	return err;
 }
 
 bool __fscrypt_inode_uses_inline_crypto(const struct inode *inode)

fs/crypto/keyring.c

···
 					struct key_preparsed_payload *prep)
 {
 	/*
-	 * We just charge FSCRYPT_MAX_KEY_SIZE bytes to the user's key quota for
-	 * each key, regardless of the exact key size.  The amount of memory
+	 * We just charge FSCRYPT_MAX_RAW_KEY_SIZE bytes to the user's key quota
+	 * for each key, regardless of the exact key size.  The amount of memory
 	 * actually used is greater than the size of the raw key anyway.
 	 */
-	return key_payload_reserve(key, FSCRYPT_MAX_KEY_SIZE);
+	return key_payload_reserve(key, FSCRYPT_MAX_RAW_KEY_SIZE);
 }
 
 static void fscrypt_user_key_describe(const struct key *key, struct seq_file *m)
···
 	int err;
 
 	if (key_spec->type == FSCRYPT_KEY_SPEC_TYPE_IDENTIFIER) {
-		err = fscrypt_init_hkdf(&secret->hkdf, secret->raw,
-					secret->size);
+		u8 sw_secret[BLK_CRYPTO_SW_SECRET_SIZE];
+		u8 *kdf_key = secret->bytes;
+		unsigned int kdf_key_size = secret->size;
+		u8 keyid_kdf_ctx = HKDF_CONTEXT_KEY_IDENTIFIER_FOR_RAW_KEY;
+
+		/*
+		 * For raw keys, the fscrypt master key is used directly as the
+		 * fscrypt KDF key.  For hardware-wrapped keys, we have to pass
+		 * the master key to the hardware to derive the KDF key, which
+		 * is then only used to derive non-file-contents subkeys.
+		 */
+		if (secret->is_hw_wrapped) {
+			err = fscrypt_derive_sw_secret(sb, secret->bytes,
+						       secret->size, sw_secret);
+			if (err)
+				return err;
+			kdf_key = sw_secret;
+			kdf_key_size = sizeof(sw_secret);
+			/*
+			 * To avoid weird behavior if someone manages to
+			 * determine sw_secret and add it as a raw key, ensure
+			 * that hardware-wrapped keys and raw keys will have
+			 * different key identifiers by deriving their key
+			 * identifiers using different KDF contexts.
+			 */
+			keyid_kdf_ctx =
+				HKDF_CONTEXT_KEY_IDENTIFIER_FOR_HW_WRAPPED_KEY;
+		}
+		err = fscrypt_init_hkdf(&secret->hkdf, kdf_key, kdf_key_size);
+		/*
+		 * Now that the KDF context is initialized, the raw KDF key is
+		 * no longer needed.
+		 */
+		memzero_explicit(kdf_key, kdf_key_size);
 		if (err)
 			return err;
 
-		/*
-		 * Now that the HKDF context is initialized, the raw key is no
-		 * longer needed.
-		 */
-		memzero_explicit(secret->raw, secret->size);
-
 		/* Calculate the key identifier */
-		err = fscrypt_hkdf_expand(&secret->hkdf,
-					  HKDF_CONTEXT_KEY_IDENTIFIER, NULL, 0,
+		err = fscrypt_hkdf_expand(&secret->hkdf, keyid_kdf_ctx, NULL, 0,
 					  key_spec->u.identifier,
 					  FSCRYPT_KEY_IDENTIFIER_SIZE);
 		if (err)
···
 	return do_add_master_key(sb, secret, key_spec);
 }
 
+/*
+ * Validate the size of an fscrypt master key being added.  Note that this is
+ * just an initial check, as we don't know which ciphers will be used yet.
+ * There is a stricter size check later when the key is actually used by a file.
+ */
+static inline bool fscrypt_valid_key_size(size_t size, u32 add_key_flags)
+{
+	u32 max_size = (add_key_flags & FSCRYPT_ADD_KEY_FLAG_HW_WRAPPED) ?
+		       FSCRYPT_MAX_HW_WRAPPED_KEY_SIZE :
+		       FSCRYPT_MAX_RAW_KEY_SIZE;
+
+	return size >= FSCRYPT_MIN_KEY_SIZE && size <= max_size;
+}
+
 static int fscrypt_provisioning_key_preparse(struct key_preparsed_payload *prep)
 {
 	const struct fscrypt_provisioning_key_payload *payload = prep->data;
 
-	if (prep->datalen < sizeof(*payload) + FSCRYPT_MIN_KEY_SIZE ||
-	    prep->datalen > sizeof(*payload) + FSCRYPT_MAX_KEY_SIZE)
+	if (prep->datalen < sizeof(*payload))
+		return -EINVAL;
+
+	if (!fscrypt_valid_key_size(prep->datalen - sizeof(*payload),
+				    payload->flags))
 		return -EINVAL;
 
 	if (payload->type != FSCRYPT_KEY_SPEC_TYPE_DESCRIPTOR &&
 	    payload->type != FSCRYPT_KEY_SPEC_TYPE_IDENTIFIER)
 		return -EINVAL;
 
-	if (payload->__reserved)
+	if (payload->flags & ~FSCRYPT_ADD_KEY_FLAG_HW_WRAPPED)
 		return -EINVAL;
 
 	prep->payload.data[0] = kmemdup(payload, prep->datalen, GFP_KERNEL);
···
 };
 
 /*
- * Retrieve the raw key from the Linux keyring key specified by 'key_id', and
- * store it into 'secret'.
+ * Retrieve the key from the Linux keyring key specified by 'key_id', and store
+ * it into 'secret'.
  *
- * The key must be of type "fscrypt-provisioning" and must have the field
- * fscrypt_provisioning_key_payload::type set to 'type', indicating that it's
- * only usable with fscrypt with the particular KDF version identified by
- * 'type'.  We don't use the "logon" key type because there's no way to
- * completely restrict the use of such keys; they can be used by any kernel API
- * that accepts "logon" keys and doesn't require a specific service prefix.
+ * The key must be of type "fscrypt-provisioning" and must have the 'type' and
+ * 'flags' field of the payload set to the given values, indicating that the key
+ * is intended for use for the specified purpose.  We don't use the "logon" key
+ * type because there's no way to completely restrict the use of such keys; they
+ * can be used by any kernel API that accepts "logon" keys and doesn't require a
+ * specific service prefix.
  *
  * The ability to specify the key via Linux keyring key is intended for cases
  * where userspace needs to re-add keys after the filesystem is unmounted and
- * re-mounted.  Most users should just provide the raw key directly instead.
+ * re-mounted.  Most users should just provide the key directly instead.
  */
-static int get_keyring_key(u32 key_id, u32 type,
+static int get_keyring_key(u32 key_id, u32 type, u32 flags,
 			   struct fscrypt_master_key_secret *secret)
 {
 	key_ref_t ref;
···
 		goto bad_key;
 	payload = key->payload.data[0];
 
-	/* Don't allow fscrypt v1 keys to be used as v2 keys and vice versa. */
-	if (payload->type != type)
+	/*
+	 * Don't allow fscrypt v1 keys to be used as v2 keys and vice versa.
+	 * Similarly, don't allow hardware-wrapped keys to be used as
+	 * non-hardware-wrapped keys and vice versa.
+	 */
+	if (payload->type != type || payload->flags != flags)
 		goto bad_key;
 
 	secret->size = key->datalen - sizeof(*payload);
-	memcpy(secret->raw, payload->raw, secret->size);
+	memcpy(secret->bytes, payload->raw, secret->size);
 	err = 0;
 	goto out_put;
 
···
 		return -EACCES;
 
 	memset(&secret, 0, sizeof(secret));
+
+	if (arg.flags) {
+		if (arg.flags & ~FSCRYPT_ADD_KEY_FLAG_HW_WRAPPED)
+			return -EINVAL;
+		if (arg.key_spec.type != FSCRYPT_KEY_SPEC_TYPE_IDENTIFIER)
+			return -EINVAL;
+		secret.is_hw_wrapped = true;
+	}
+
 	if (arg.key_id) {
 		if (arg.raw_size != 0)
 			return -EINVAL;
-		err = get_keyring_key(arg.key_id, arg.key_spec.type, &secret);
+		err = get_keyring_key(arg.key_id, arg.key_spec.type, arg.flags,
+				      &secret);
 		if (err)
 			goto out_wipe_secret;
 	} else {
-		if (arg.raw_size < FSCRYPT_MIN_KEY_SIZE ||
-		    arg.raw_size > FSCRYPT_MAX_KEY_SIZE)
+		if (!fscrypt_valid_key_size(arg.raw_size, arg.flags))
 			return -EINVAL;
 		secret.size = arg.raw_size;
 		err = -EFAULT;
-		if (copy_from_user(secret.raw, uarg->raw, secret.size))
+		if (copy_from_user(secret.bytes, uarg->raw, secret.size))
 			goto out_wipe_secret;
 	}
 
···
 static void
 fscrypt_get_test_dummy_secret(struct fscrypt_master_key_secret *secret)
 {
-	static u8 test_key[FSCRYPT_MAX_KEY_SIZE];
+	static u8 test_key[FSCRYPT_MAX_RAW_KEY_SIZE];
 
-	get_random_once(test_key, FSCRYPT_MAX_KEY_SIZE);
+	get_random_once(test_key, sizeof(test_key));
 
 	memset(secret, 0, sizeof(*secret));
-	secret->size = FSCRYPT_MAX_KEY_SIZE;
-	memcpy(secret->raw, test_key, FSCRYPT_MAX_KEY_SIZE);
+	secret->size = sizeof(test_key);
+	memcpy(secret->bytes, test_key, sizeof(test_key));
 }
 
 int fscrypt_get_test_dummy_key_identifier(
···
 
 	fscrypt_get_test_dummy_secret(&secret);
 
-	err = fscrypt_init_hkdf(&secret.hkdf, secret.raw, secret.size);
+	err = fscrypt_init_hkdf(&secret.hkdf, secret.bytes, secret.size);
 	if (err)
 		goto out;
-	err = fscrypt_hkdf_expand(&secret.hkdf, HKDF_CONTEXT_KEY_IDENTIFIER,
+	err = fscrypt_hkdf_expand(&secret.hkdf,
+				  HKDF_CONTEXT_KEY_IDENTIFIER_FOR_RAW_KEY,
 				  NULL, 0, key_identifier,
 				  FSCRYPT_KEY_IDENTIFIER_SIZE);
 out:

fs/crypto/keysetup.c

···
 	struct crypto_skcipher *tfm;
 
 	if (fscrypt_using_inline_encryption(ci))
-		return fscrypt_prepare_inline_crypt_key(prep_key, raw_key, ci);
+		return fscrypt_prepare_inline_crypt_key(prep_key, raw_key,
+							ci->ci_mode->keysize,
+							false, ci);
 
 	tfm = fscrypt_allocate_skcipher(ci->ci_mode, raw_key, ci->ci_inode);
 	if (IS_ERR(tfm))
···
 	struct fscrypt_mode *mode = ci->ci_mode;
 	const u8 mode_num = mode - fscrypt_modes;
 	struct fscrypt_prepared_key *prep_key;
-	u8 mode_key[FSCRYPT_MAX_KEY_SIZE];
+	u8 mode_key[FSCRYPT_MAX_RAW_KEY_SIZE];
 	u8 hkdf_info[sizeof(mode_num) + sizeof(sb->s_uuid)];
 	unsigned int hkdf_infolen = 0;
+	bool use_hw_wrapped_key = false;
 	int err;
 
 	if (WARN_ON_ONCE(mode_num > FSCRYPT_MODE_MAX))
 		return -EINVAL;
+
+	if (mk->mk_secret.is_hw_wrapped && S_ISREG(inode->i_mode)) {
+		/* Using a hardware-wrapped key for file contents encryption */
+		if (!fscrypt_using_inline_encryption(ci)) {
+			if (sb->s_flags & SB_INLINECRYPT)
+				fscrypt_warn(ci->ci_inode,
+					     "Hardware-wrapped key required, but no suitable inline encryption capabilities are available");
+			else
+				fscrypt_warn(ci->ci_inode,
+					     "Hardware-wrapped keys require inline encryption (-o inlinecrypt)");
+			return -EINVAL;
+		}
+		use_hw_wrapped_key = true;
+	}
 
 	prep_key = &keys[mode_num];
 	if (fscrypt_is_key_prepared(prep_key, ci)) {
···
 
 	if (fscrypt_is_key_prepared(prep_key, ci))
 		goto done_unlock;
+
+	if (use_hw_wrapped_key) {
+		err = fscrypt_prepare_inline_crypt_key(prep_key,
+						       mk->mk_secret.bytes,
+						       mk->mk_secret.size, true,
+						       ci);
+		if (err)
+			goto out_unlock;
+		goto done_unlock;
+	}
 
 	BUILD_BUG_ON(sizeof(mode_num) != 1);
 	BUILD_BUG_ON(sizeof(sb->s_uuid) != 16);
···
 {
 	int err;
 
+	if (mk->mk_secret.is_hw_wrapped &&
+	    !(ci->ci_policy.v2.flags & (FSCRYPT_POLICY_FLAG_IV_INO_LBLK_64 |
+					FSCRYPT_POLICY_FLAG_IV_INO_LBLK_32))) {
+		fscrypt_warn(ci->ci_inode,
+			     "Hardware-wrapped keys are only supported with IV_INO_LBLK policies");
+		return -EINVAL;
+	}
+
 	if (ci->ci_policy.v2.flags & FSCRYPT_POLICY_FLAG_DIRECT_KEY) {
 		/*
 		 * DIRECT_KEY: instead of deriving per-file encryption keys, the
···
 		   FSCRYPT_POLICY_FLAG_IV_INO_LBLK_32) {
 		err = fscrypt_setup_iv_ino_lblk_32_key(ci, mk);
 	} else {
-		u8 derived_key[FSCRYPT_MAX_KEY_SIZE];
+		u8 derived_key[FSCRYPT_MAX_RAW_KEY_SIZE];
 
 		err = fscrypt_hkdf_expand(&mk->mk_secret.hkdf,
 					  HKDF_CONTEXT_PER_FILE_ENC_KEY,
···
 	struct fscrypt_master_key *mk;
 	int err;
 
-	err = fscrypt_select_encryption_impl(ci);
-	if (err)
-		return err;
-
 	err = fscrypt_policy_to_key_spec(&ci->ci_policy, &mk_spec);
 	if (err)
 		return err;
···
 		if (ci->ci_policy.version != FSCRYPT_POLICY_V1)
 			return -ENOKEY;
 
+		err = fscrypt_select_encryption_impl(ci, false);
+		if (err)
+			return err;
+
 		/*
 		 * As a legacy fallback for v1 policies, search for the key in
 		 * the current task's subscribed keyrings too.  Don't move this
···
 		goto out_release_key;
 	}
 
+	err = fscrypt_select_encryption_impl(ci, mk->mk_secret.is_hw_wrapped);
+	if (err)
+		goto out_release_key;
+
 	switch (ci->ci_policy.version) {
 	case FSCRYPT_POLICY_V1:
-		err = fscrypt_setup_v1_file_key(ci, mk->mk_secret.raw);
+		if (WARN_ON_ONCE(mk->mk_secret.is_hw_wrapped)) {
+			/*
+			 * This should never happen, as adding a v1 policy key
+			 * that is hardware-wrapped isn't allowed.
+			 */
+			err = -EINVAL;
+			goto out_release_key;
+		}
+		err = fscrypt_setup_v1_file_key(ci, mk->mk_secret.bytes);
 		break;
 	case FSCRYPT_POLICY_V2:
 		err = fscrypt_setup_v2_file_key(ci, mk, need_dirhash_key);

fs/crypto/keysetup_v1.c

···
 	payload = (const struct fscrypt_key *)ukp->data;
 
 	if (ukp->datalen != sizeof(struct fscrypt_key) ||
-	    payload->size < 1 || payload->size > FSCRYPT_MAX_KEY_SIZE) {
+	    payload->size < 1 || payload->size > sizeof(payload->raw)) {
 		fscrypt_warn(NULL,
 			     "key with description '%s' has invalid payload",
 			     key->description);
···
 	const struct fscrypt_mode	*dk_mode;
 	struct fscrypt_prepared_key	dk_key;
 	u8				dk_descriptor[FSCRYPT_KEY_DESCRIPTOR_SIZE];
-	u8				dk_raw[FSCRYPT_MAX_KEY_SIZE];
+	u8				dk_raw[FSCRYPT_MAX_RAW_KEY_SIZE];
 };
 
 static void free_direct_key(struct fscrypt_direct_key *dk)

include/uapi/linux/fscrypt.h

···
  */
 struct fscrypt_provisioning_key_payload {
 	__u32 type;
-	__u32 __reserved;
+	__u32 flags;
 	__u8 raw[];
 };
 
···
 	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[];
 };