diff mbox series

[v2,01/20] fscrypt: expose fscrypt_nokey_name

Message ID b77f881449e67d61862ec28d3863fd9036978e8e.1662420176.git.sweettea-kernel@dorminy.me (mailing list archive)
State New, archived
Headers show
Series btrfs: add fscrypt integration | expand

Commit Message

Sweet Tea Dorminy Sept. 6, 2022, 12:35 a.m. UTC
From: Omar Sandoval <osandov@osandov.com>

btrfs stores its data structures, including filenames in directories, in
its own buffer implementation, struct extent_buffer, composed of
several non-contiguous pages. We could copy filenames into a
temporary buffer and use fscrypt_match_name() against that buffer, such
extensive memcpying would be expensive. Instead, exposing
fscrypt_nokey_name as in this change allows btrfs to recapitulate
fscrypt_match_name() using methods on struct extent_buffer instead of
dealing with a raw byte array.

Signed-off-by: Omar Sandoval <osandov@osandov.com>
Signed-off-by: Sweet Tea Dorminy <sweettea-kernel@dorminy.me>
---
 fs/crypto/fname.c       | 39 +--------------------------------------
 include/linux/fscrypt.h | 37 +++++++++++++++++++++++++++++++++++++
 2 files changed, 38 insertions(+), 38 deletions(-)

Comments

Josef Bacik Sept. 8, 2022, 1:41 p.m. UTC | #1
On Mon, Sep 05, 2022 at 08:35:16PM -0400, Sweet Tea Dorminy wrote:
> From: Omar Sandoval <osandov@osandov.com>
> 
> btrfs stores its data structures, including filenames in directories, in
> its own buffer implementation, struct extent_buffer, composed of
> several non-contiguous pages. We could copy filenames into a
> temporary buffer and use fscrypt_match_name() against that buffer, such
> extensive memcpying would be expensive. Instead, exposing
> fscrypt_nokey_name as in this change allows btrfs to recapitulate
> fscrypt_match_name() using methods on struct extent_buffer instead of
> dealing with a raw byte array.
> 
> Signed-off-by: Omar Sandoval <osandov@osandov.com>
> Signed-off-by: Sweet Tea Dorminy <sweettea-kernel@dorminy.me>

Reviewed-by: Josef Bacik <josef@toxicpanda.com>

Thanks,

Josef
diff mbox series

Patch

diff --git a/fs/crypto/fname.c b/fs/crypto/fname.c
index 12bd61d20f69..6c092a1533f7 100644
--- a/fs/crypto/fname.c
+++ b/fs/crypto/fname.c
@@ -14,7 +14,6 @@ 
 #include <linux/namei.h>
 #include <linux/scatterlist.h>
 #include <crypto/hash.h>
-#include <crypto/sha2.h>
 #include <crypto/skcipher.h>
 #include "fscrypt_private.h"
 
@@ -26,43 +25,7 @@ 
 #define FSCRYPT_FNAME_MIN_MSG_LEN 16
 
 /*
- * struct fscrypt_nokey_name - identifier for directory entry when key is absent
- *
- * When userspace lists an encrypted directory without access to the key, the
- * filesystem must present a unique "no-key name" for each filename that allows
- * it to find the directory entry again if requested.  Naively, that would just
- * mean using the ciphertext filenames.  However, since the ciphertext filenames
- * can contain illegal characters ('\0' and '/'), they must be encoded in some
- * way.  We use base64url.  But that can cause names to exceed NAME_MAX (255
- * bytes), so we also need to use a strong hash to abbreviate long names.
- *
- * The filesystem may also need another kind of hash, the "dirhash", to quickly
- * find the directory entry.  Since filesystems normally compute the dirhash
- * over the on-disk filename (i.e. the ciphertext), it's not computable from
- * no-key names that abbreviate the ciphertext using the strong hash to fit in
- * NAME_MAX.  It's also not computable if it's a keyed hash taken over the
- * plaintext (but it may still be available in the on-disk directory entry);
- * casefolded directories use this type of dirhash.  At least in these cases,
- * each no-key name must include the name's dirhash too.
- *
- * To meet all these requirements, we base64url-encode the following
- * variable-length structure.  It contains the dirhash, or 0's if the filesystem
- * didn't provide one; up to 149 bytes of the ciphertext name; and for
- * ciphertexts longer than 149 bytes, also the SHA-256 of the remaining bytes.
- *
- * This ensures that each no-key name contains everything needed to find the
- * directory entry again, contains only legal characters, doesn't exceed
- * NAME_MAX, is unambiguous unless there's a SHA-256 collision, and that we only
- * take the performance hit of SHA-256 on very long filenames (which are rare).
- */
-struct fscrypt_nokey_name {
-	u32 dirhash[2];
-	u8 bytes[149];
-	u8 sha256[SHA256_DIGEST_SIZE];
-}; /* 189 bytes => 252 bytes base64url-encoded, which is <= NAME_MAX (255) */
-
-/*
- * Decoded size of max-size no-key name, i.e. a name that was abbreviated using
+ * Decoded size of max-size nokey name, i.e. a name that was abbreviated using
  * the strong hash and thus includes the 'sha256' field.  This isn't simply
  * sizeof(struct fscrypt_nokey_name), as the padding at the end isn't included.
  */
diff --git a/include/linux/fscrypt.h b/include/linux/fscrypt.h
index 7d2f1e0f23b1..a236d8c6d0da 100644
--- a/include/linux/fscrypt.h
+++ b/include/linux/fscrypt.h
@@ -16,6 +16,7 @@ 
 #include <linux/fs.h>
 #include <linux/mm.h>
 #include <linux/slab.h>
+#include <crypto/sha2.h>
 #include <uapi/linux/fscrypt.h>
 
 /*
@@ -54,6 +55,42 @@  struct fscrypt_name {
 #define fname_name(p)		((p)->disk_name.name)
 #define fname_len(p)		((p)->disk_name.len)
 
+/*
+ * struct fscrypt_nokey_name - identifier for directory entry when key is absent
+ *
+ * When userspace lists an encrypted directory without access to the key, the
+ * filesystem must present a unique "no-key name" for each filename that allows
+ * it to find the directory entry again if requested.  Naively, that would just
+ * mean using the ciphertext filenames.  However, since the ciphertext filenames
+ * can contain illegal characters ('\0' and '/'), they must be encoded in some
+ * way.  We use base64url.  But that can cause names to exceed NAME_MAX (255
+ * bytes), so we also need to use a strong hash to abbreviate long names.
+ *
+ * The filesystem may also need another kind of hash, the "dirhash", to quickly
+ * find the directory entry.  Since filesystems normally compute the dirhash
+ * over the on-disk filename (i.e. the ciphertext), it's not computable from
+ * no-key names that abbreviate the ciphertext using the strong hash to fit in
+ * NAME_MAX.  It's also not computable if it's a keyed hash taken over the
+ * plaintext (but it may still be available in the on-disk directory entry);
+ * casefolded directories use this type of dirhash.  At least in these cases,
+ * each no-key name must include the name's dirhash too.
+ *
+ * To meet all these requirements, we base64url-encode the following
+ * variable-length structure.  It contains the dirhash, or 0's if the filesystem
+ * didn't provide one; up to 149 bytes of the ciphertext name; and for
+ * ciphertexts longer than 149 bytes, also the SHA-256 of the remaining bytes.
+ *
+ * This ensures that each no-key name contains everything needed to find the
+ * directory entry again, contains only legal characters, doesn't exceed
+ * NAME_MAX, is unambiguous unless there's a SHA-256 collision, and that we only
+ * take the performance hit of SHA-256 on very long filenames (which are rare).
+ */
+struct fscrypt_nokey_name {
+	u32 dirhash[2];
+	u8 bytes[149];
+	u8 sha256[SHA256_DIGEST_SIZE];
+}; /* 189 bytes => 252 bytes base64url-encoded, which is <= NAME_MAX (255) */
+
 /* Maximum value for the third parameter of fscrypt_operations.set_context(). */
 #define FSCRYPT_SET_CONTEXT_MAX_SIZE	40