Viewing: ext4-dir-entry-len.patch
commit 409935cc7094bf0d98b392d8ddc9a016f02749eb
Author: Artem Blagodarenko <ablagodarenko@thelustrecollective.com>
Date: Mon Jan 19 09:37:24 2026 +0300
LU-19847 ldiskfs: dirent and fscrypt+case_insensitive
EXT4 occupies space in dirent just after the name with
a hash for simultaneous fscrypt and casefold support.
It was discussed with the EXT4 community that it is
possible to move the hash to dirdata. It could be the
second (or third, if 64-bit inode count) user of
dirdata.
At the same time, the hash placed after the file name
should also be supported.
This patch makes LDISKFS ready for such a hash present
in both variants. While it is not currently possible
to enable fscrypt + case_insensitive on LDISKFS, it is
useful to verify that LUFID works well with such code.
fscrypt + case_insensitive support with dirdata
enabled is checked in a special xfstest sent to EXT4
with the dirdata patch.
Test-Parameters: clientdistro=el10.0 serverdistro=el10.0
Test-Parameters: clientdistro=el10.1 serverdistro=el10.1
Signed-off-by: Artem Blagodarenko <ablagodarenko@thelustrecollective.com>
Change-Id: Ia9874396037a24b494dd3cfa7208e10a366f5eb3
Reviewed-on: https://review.whamcloud.com/c/fs/lustre-release/+/64439
Tested-by: jenkins <devops@whamcloud.com>
Reviewed-by: Oleg Drokin <green@whamcloud.com>
Tested-by: Maloo <maloo@whamcloud.com>
Reviewed-by: Jian Yu <yujian@whamcloud.com>
Reviewed-by: Andreas Dilger <adilger@thelustrecollective.com>
---
Patch 1 of 3: ext4-dir-entry-len.patch
---
fs/ext4/ext4.h | 99 +++++++++++++++++++++++++++++++++++++++++++++++--
fs/ext4/namei.c | 12 +++---
2 files changed, 102 insertions(+), 9 deletions(-)
diff --git a/fs/ext4/ext4.h b/fs/ext4/ext4.h
--- a/fs/ext4/ext4.h
+++ b/fs/ext4/ext4.h
@@ -1213,6 +1213,7 @@ struct ext4_inode_info {
* Mount flags set via mount options or defaults
*/
#define EXT4_MOUNT_NO_MBCACHE 0x00001 /* Do not use mbcache */
+#define EXT4_MOUNT_DIRDATA 0x00002 /* Data in directory entries */
#define EXT4_MOUNT_GRPID 0x00004 /* Create files with directory's group */
#define EXT4_MOUNT_DEBUG 0x00008 /* Some debugging messages */
#define EXT4_MOUNT_ERRORS_CONT 0x00010 /* Continue on errors */
@@ -2244,6 +2245,7 @@ EXT4_FEATURE_INCOMPAT_FUNCS(casefold, CASEFOLD)
EXT4_FEATURE_INCOMPAT_FLEX_BG| \
EXT4_FEATURE_INCOMPAT_EA_INODE| \
EXT4_FEATURE_INCOMPAT_MMP | \
+ EXT4_FEATURE_INCOMPAT_DIRDATA | \
EXT4_FEATURE_INCOMPAT_INLINE_DATA | \
EXT4_FEATURE_INCOMPAT_ENCRYPT | \
EXT4_FEATURE_INCOMPAT_CASEFOLD | \
@@ -2481,6 +2483,35 @@ struct ext4_dir_entry_tail {
#define EXT4_FT_SYMLINK 7
#define EXT4_FT_MAX 8
+#define EXT4_FT_MASK 0xf
+
+#if EXT4_FT_MAX > EXT4_FT_MASK
+#error "conflicting EXT4_FT_MAX and EXT4_FT_MASK"
+#endif
+
+/*
+ * d_type has 4 unused bits, so it can hold four types data. these different
+ * type of data (e.g. lustre data, high 32 bits of 64-bit inode number) can be
+ * stored, in flag order, after file-name in ext4 dirent.
+*/
+/*
+ * These flags is added to d_type if ext4 dirent has extra data after
+ * filename. This data length is variable and length is stored in first byte
+ * of data. Data starts after filename NUL byte.
+ */
+#define EXT4_DIRENT_LUFID 0x10
+#define EXT4_DIRENT_INO64 0x20
+#define EXT4_DIRENT_CFHASH 0x40
+
+struct ext4_dirent_data_header {
+ /* length of this header + the whole data blob */
+ __u8 ddh_length;
+} __packed;
+
+struct ext4_dirent_hash {
+ struct ext4_dirent_data_header dh_header;
+ struct ext4_dir_entry_hash dh_hash;
+} __packed;
#define EXT4_FT_DIR_CSUM 0xDE
@@ -2492,6 +2523,17 @@ struct ext4_dir_entry_tail {
#define EXT4_DIR_PAD 4
#define EXT4_DIR_ROUND (EXT4_DIR_PAD - 1)
#define EXT4_MAX_REC_LEN ((1<<16)-1)
+#define EXT4_DIR_REC_LEN_(name_len, i_dir) \
+ ext4_dir_rec_len((name_len), (i_dir))
+#define EXT4_DIR_ENTRY_LEN_(de, i_dir) \
+ (EXT4_DIR_REC_LEN_((de)->name_len + ext4_get_dirent_data_len(de), \
+ (i_dir)))
+/* ldiskfs */
+#define EXT4_DIR_REC_LEN(name_len, i_dir) EXT4_DIR_REC_LEN_((name_len), (i_dir))
+#define EXT4_DIR_ENTRY_LEN(de, i_dir) EXT4_DIR_ENTRY_LEN_((de), (i_dir))
+/* lustre osd_handler compat -- ifdef LDISKFS_DIR_REC_LEN_WITH_DIR */
+#define EXT4_DIR_REC_LEN_WITH_DIR 1
+#define __EXT4_DIR_REC_LEN(name_len) EXT4_DIR_REC_LEN_((name_len), NULL)
/*
* The rec_len is dependent on the type of directory. Directories that are
@@ -2499,10 +2541,10 @@ struct ext4_dir_entry_tail {
* ext4_extended_dir_entry_2. For all entries related to '.' or '..' you should
* pass NULL for dir, as those entries do not use the extra fields.
*/
-static inline unsigned int ext4_dir_rec_len(__u8 name_len,
+static inline unsigned int ext4_dir_rec_len(__u32 name_len,
const struct inode *dir)
{
- int rec_len = (name_len + 8 + EXT4_DIR_ROUND);
+ __u32 rec_len = (name_len + 8 + EXT4_DIR_ROUND);
if (dir && ext4_hash_in_dirent(dir))
rec_len += sizeof(struct ext4_dir_entry_hash);
@@ -2942,10 +2984,16 @@ static const unsigned char ext4_filetype_table[] = {
static inline unsigned char get_dtype(struct super_block *sb, int filetype)
{
- if (!ext4_has_feature_filetype(sb) || filetype >= EXT4_FT_MAX)
+ int fl_index = filetype & EXT4_FT_MASK;
+
+ if (!ext4_has_feature_filetype(sb) || fl_index >= EXT4_FT_MAX)
return DT_UNKNOWN;
- return ext4_filetype_table[filetype];
+ if (!test_opt(sb, DIRDATA))
+ return ext4_filetype_table[fl_index];
+
+ return (ext4_filetype_table[fl_index]) |
+ (filetype & ~EXT4_FT_MASK);
}
extern int ext4_check_all_de(struct inode *dir, struct buffer_head *bh,
void *buf, int buf_size);
@@ -3948,6 +3996,49 @@ static inline void ext4_clear_io_unwritten_flag(ext4_io_end_t *io_end)
io_end->flag &= ~EXT4_IO_END_UNWRITTEN;
}
+#define ext4_dirdata_next(ddh) \
+ (struct ext4_dirent_data_header *)((char *)ddh + ddh->ddh_length)
+/*
+ * Compute the total directory entry data length.
+ * This includes the filename and an implicit NUL terminator (always present),
+ * and optional extensions. Each extension has a bit set in the high 4 bits of
+ * de->file_type, and the extension length is the first byte in each entry.
+ */
+static inline int ext4_get_dirent_data_len(struct ext4_dir_entry_2 *de)
+{
+ struct ext4_dirent_data_header *ddh =
+ (struct ext4_dirent_data_header *)de->name + de->name_len +
+ 1 /* NUL terminator */;
+ __u8 extra_data_flags = (de->file_type & ~EXT4_FT_MASK) >> 4;
+ struct ext4_dir_entry_tail *t = (struct ext4_dir_entry_tail *)de;
+ int dlen = 0;
+
+ if (!t->det_reserved_zero1 &&
+ le16_to_cpu(t->det_rec_len) ==
+ sizeof(struct ext4_dir_entry_tail) &&
+ !t->det_reserved_zero2 &&
+ t->det_reserved_ft == EXT4_FT_DIR_CSUM)
+ return 0;
+
+ while (extra_data_flags) {
+ if (extra_data_flags & 1) {
+ dlen += ddh->ddh_length + (dlen == 0);
+ ddh = ext4_dirdata_next(ddh);
+ }
+ extra_data_flags >>= 1;
+ }
+ return dlen;
+}
+
+/* Inline function to calculate directory entry length */
+static inline unsigned int ext4_dir_entry_len(struct ext4_dir_entry_2 *de,
+ const struct inode *dir)
+{
+ unsigned int name_len = de->name_len + ext4_get_dirent_data_len(de);
+
+ return ext4_dir_rec_len(name_len, dir);
+}
+
extern const struct iomap_ops ext4_iomap_ops;
extern const struct iomap_ops ext4_iomap_overwrite_ops;
extern const struct iomap_ops ext4_iomap_report_ops;
diff --git a/fs/ext4/namei.c b/fs/ext4/namei.c
--- a/fs/ext4/namei.c
+++ b/fs/ext4/namei.c
@@ -525,13 +525,15 @@ ext4_next_entry(struct ext4_dir_entry_2 *p, unsigned long blocksize)
* Future: use high four bits of block for coalesce-on-delete flags
* Mask them off for now.
*/
-struct dx_root_info *dx_get_dx_info(struct ext4_dir_entry_2 *de)
+static struct dx_root_info *dx_get_dx_info(struct ext4_dir_entry_2 *de)
{
/* get dotdot first */
- de = (struct ext4_dir_entry_2 *)((char *)de + EXT4_DIR_REC_LEN(1));
+ de = (struct ext4_dir_entry_2 *)((char *)de +
+ ext4_dir_entry_len(de, NULL));
/* dx root info is after dotdot entry */
- de = (struct ext4_dir_entry_2 *)((char *)de + EXT4_DIR_REC_LEN(2));
+ de = (struct ext4_dir_entry_2 *)((char *)de +
+ ext4_dir_entry_len(de, NULL));
return (struct dx_root_info *)de;
}
@@ -2367,7 +2369,7 @@ out_frames:
return retval;
}
-/* update ".." entry */
+/* update ".." entry, try to expand the entry if necessary */
static int ext4_update_dotdot(handle_t *handle, struct dentry *dentry,
struct inode *inode)
{
@@ -3054,7 +3056,7 @@ int ext4_init_dirblock(handle_t *handle, struct inode *inode,
ext4_set_de_type(inode->i_sb, de, S_IFDIR);
if (inline_buf) {
de->rec_len = ext4_rec_len_to_disk(
- ext4_dir_rec_len(de->name_len, NULL),
+ ext4_dir_rec_len(de, NULL),
blocksize);
de = ext4_next_entry(de, blocksize);
header_size = (char *)de - bh->b_data;
--
