On 2024/9/25 15:51, Jan Kara wrote:
On Wed 25-09-24 09:56:24, Hongbo Li wrote:The `fs_lookup_param` did not consider the relative path for block device. When we mount ext4 with `journal_path` option using relative path, `param->dirfd` was not set which will cause mounting error. This can be reproduced easily like this: mke2fs -F -O journal_dev $JOURNAL_DEV -b 4096 100M mkfs.ext4 -F -J device=$JOURNAL_DEV -b 4096 $FS_DEV cd /dev; mount -t ext4 -o journal_path=`basename $JOURNAL_DEV` $FS_DEV $MNT Fixes: 461c3af045d3 ("ext4: Change handle_mount_opt() to use fs_parameter") Suggested-by: Christian Brauner <brauner@xxxxxxxxxx> Signed-off-by: Hongbo Li <lihongbo22@xxxxxxxxxx> --- v2: - Change the journal_path parameter as string not bdev, and determine the relative path situation inside fs_lookup_param. - Add Suggested-by. v1: https://lore.kernel.org/all/20240527-mahlen-packung-3fe035ab390d@brauner/ --- fs/ext4/super.c | 4 ++-- fs/fs_parser.c | 3 +++ 2 files changed, 5 insertions(+), 2 deletions(-) diff --git a/fs/ext4/super.c b/fs/ext4/super.c index 16a4ce704460..cd23536ce46e 100644 --- a/fs/ext4/super.c +++ b/fs/ext4/super.c @@ -1744,7 +1744,7 @@ static const struct fs_parameter_spec ext4_param_specs[] = { fsparam_u32 ("min_batch_time", Opt_min_batch_time), fsparam_u32 ("max_batch_time", Opt_max_batch_time), fsparam_u32 ("journal_dev", Opt_journal_dev), - fsparam_bdev ("journal_path", Opt_journal_path), + fsparam_string ("journal_path", Opt_journal_path),Why did you change this? As far as I can see the only effect would be that empty path will not be allowed (which makes sense) but that seems like an independent change which would deserve a comment in the changelog? Or am I missing something?
The fsparam_bdev serves no purpose here, you're right, empty path will not be allowed for `journal_path` option. And all changes are made to fix the issues (journal_path options changed) introduced by the previous new mount api conversion.
fsparam_flag ("journal_checksum", Opt_journal_checksum), fsparam_flag ("nojournal_checksum", Opt_nojournal_checksum), fsparam_flag ("journal_async_commit",Opt_journal_async_commit), @@ -2301,7 +2301,7 @@ static int ext4_parse_param(struct fs_context *fc, struct fs_parameter *param) return -EINVAL; }- error = fs_lookup_param(fc, param, 1, LOOKUP_FOLLOW, &path);+ error = fs_lookup_param(fc, param, true, LOOKUP_FOLLOW, &path); if (error) { ext4_msg(NULL, KERN_ERR, "error: could not find " "journal device path"); diff --git a/fs/fs_parser.c b/fs/fs_parser.c index 24727ec34e5a..2ae296764b69 100644 --- a/fs/fs_parser.c +++ b/fs/fs_parser.c @@ -156,6 +156,9 @@ int fs_lookup_param(struct fs_context *fc, f = getname_kernel(param->string); if (IS_ERR(f)) return PTR_ERR(f); + /* for relative path */ + if (f->name[0] != '/') + param->dirfd = AT_FDCWD;What Al meant is that you can do simply: param->dirfd = AT_FDCWD; and everything will work the same because 'dfd' is ignored for absolute pathnames in path_init().
ok, seems reasonable. Thanks, Hongbo
Honza
