[PATCH] ext4: Improve feature checking
Sean Anderson
seanga2 at gmail.com
Mon Jul 1 02:23:00 CEST 2024
On 6/30/24 17:25, Richard Weinberger wrote:
> Evaluate the filesystem incompat and ro_compat bit fields to judge
> whether the filesystem can be read or written.
> For the read side only a scary warning is shown so far.
> I'd love to about mounting too, but I fear this will break some setups
I think you are missing a verb in the first half of your sentence
> where the driver works by chance.
>
> Signed-off-by: Richard Weinberger <richard at nod.at>
> ---
> After facing ext4 write corruptions and read errors I checked
> the ext4 implementation briefly.
> Hopefully this patch helps other users to detect earlier that
> they're using ext4 features which are not supported by u-boot.
>
> To make feature checking possible I had to figure what this
> implementation actually supports.
> I hope I got them right, feedback is welcome!
>
> Thanks,
> //richard
> ---
> fs/ext4/ext4_common.c | 8 +++++++
> fs/ext4/ext4_write.c | 12 ++++++++--
> include/ext4fs.h | 55 ++++++++++++++++++++++++++++++++++++++++++-
> 3 files changed, 72 insertions(+), 3 deletions(-)
>
> diff --git a/fs/ext4/ext4_common.c b/fs/ext4/ext4_common.c
> index 2ff0dca249..7148d35ee0 100644
> --- a/fs/ext4/ext4_common.c
> +++ b/fs/ext4/ext4_common.c
> @@ -2386,6 +2386,14 @@ int ext4fs_mount(void)
> fs->inodesz = 128;
> fs->gdsize = 32;
> } else {
> + int missing = __le32_to_cpu(data->sblock.feature_incompat) & \
> + ~(EXT4_FEATURE_INCOMPAT_SUPP | \
> + EXT4_FEATURE_INCOMPAT_SUPP_LAZY_RO);
> +
> + if (missing)
> + log_err("fs uses incompatible features: %08x, failures *are* expected!\n",
> + missing);
> +
I think it is a bit unclear to the user what their action should be. Maybe something like
printf("EXT4 incompatible features: %08x, ignoring...\n")
and then maybe add a comment like what you have in the commit message.
> debug("EXT4 features COMPAT: %08x INCOMPAT: %08x RO_COMPAT: %08x\n",
> __le32_to_cpu(data->sblock.feature_compatibility),
> __le32_to_cpu(data->sblock.feature_incompat),
> diff --git a/fs/ext4/ext4_write.c b/fs/ext4/ext4_write.c
> index d057f6b5a7..4aae3c5f7f 100644
> --- a/fs/ext4/ext4_write.c
> +++ b/fs/ext4/ext4_write.c
> @@ -869,6 +869,7 @@ int ext4fs_write(const char *fname, const char *buffer,
> ALLOC_CACHE_ALIGN_BUFFER(char, filename, 256);
> bool store_link_in_inode = false;
> memset(filename, 0x00, 256);
> + int missing_feat;
>
> if (type != FILETYPE_REG && type != FILETYPE_SYMLINK)
> return -1;
> @@ -882,8 +883,15 @@ int ext4fs_write(const char *fname, const char *buffer,
> return -1;
> }
>
> - if (le32_to_cpu(fs->sb->feature_ro_compat) & EXT4_FEATURE_RO_COMPAT_METADATA_CSUM) {
> - printf("Unsupported feature metadata_csum found, not writing.\n");
> + missing_feat = le32_to_cpu(fs->sb->feature_incompat) & ~EXT4_FEATURE_INCOMPAT_SUPP;
> + if (missing_feat) {
> + log_err("Unsupported features found %08x, not writing.\n", missing_feat);
> + return -1;
> + }
> +
> + missing_feat = le32_to_cpu(fs->sb->feature_ro_compat) & ~EXT4_FEATURE_RO_COMPAT_SUPP;
> + if (missing_feat) {
> + log_err("Unsupported RO compat features found %08x, not writing.\n", missing_feat);
> return -1;
> }
>
> diff --git a/include/ext4fs.h b/include/ext4fs.h
> index d96edfd057..01b66f469f 100644
> --- a/include/ext4fs.h
> +++ b/include/ext4fs.h
> @@ -34,12 +34,65 @@ struct disk_partition;
> #define EXT4_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/
> #define EXT4_EXTENTS_FL 0x00080000 /* Inode uses extents */
> #define EXT4_EXT_MAGIC 0xf30a
> -#define EXT4_FEATURE_RO_COMPAT_GDT_CSUM 0x0010
> +
> +#define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER 0x0001
> +#define EXT4_FEATURE_RO_COMPAT_LARGE_FILE 0x0002
> +#define EXT4_FEATURE_RO_COMPAT_BTREE_DIR 0x0004
> +#define EXT4_FEATURE_RO_COMPAT_HUGE_FILE 0x0008
> +#define EXT4_FEATURE_RO_COMPAT_GDT_CSUM 0x0010
> +#define EXT4_FEATURE_RO_COMPAT_DIR_NLINK 0x0020
> +#define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE 0x0040
> +#define EXT4_FEATURE_RO_COMPAT_QUOTA 0x0100
> +#define EXT4_FEATURE_RO_COMPAT_BIGALLOC 0x0200
> #define EXT4_FEATURE_RO_COMPAT_METADATA_CSUM 0x0400
> +
> +#define EXT4_FEATURE_INCOMPAT_FILETYPE 0x0002
> +#define EXTE_FEATURE_INCOMPAT_RECOVER 0x0004
> #define EXT4_FEATURE_INCOMPAT_EXTENTS 0x0040
> #define EXT4_FEATURE_INCOMPAT_64BIT 0x0080
> +#define EXT4_FEATURE_INCOMPAT_MMP 0x0100
> +#define EXT4_FEATURE_INCOMPAT_FLEX_BG 0x0200
> +#define EXT4_FEATURE_INCOMPAT_CSUM_SEED 0x2000
> +#define EXT4_FEATURE_INCOMPAT_ENCRYPT 0x10000
> +
> #define EXT4_INDIRECT_BLOCKS 12
>
> +/*
> + * Incompat features supported by this implementation.
> + */
> +#define EXT4_FEATURE_INCOMPAT_SUPP ( EXT4_FEATURE_INCOMPAT_FILETYPE \
> + | EXTE_FEATURE_INCOMPAT_RECOVER \
EXT4
> + | EXT4_FEATURE_INCOMPAT_EXTENTS \
> + | EXT4_FEATURE_INCOMPAT_64BIT \
> + | EXT4_FEATURE_INCOMPAT_FLEX_BG \
> + )
Operators should go at the end of the line. So like
#define EXT4_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE | \
EXT4_FEATURE_INCOMPAT_RECOVER | \
EXT4_FEATURE_INCOMPAT_EXTENTS | \
EXT4_FEATURE_INCOMPAT_64BIT | \
EXT4_FEATURE_INCOMPAT_FLEX_BG)
> +/*
> + * Incompat features supported by this implementation only in a lazy
> + * way, good enough for reading files.
> + *
> + * - Multi mount protection (mmp) is not supported, but for read-only
> + * we get away with it.
> + * - Same fore metadata_csum_seed and metadata_csum.
nit: for
> + * - The implementation has also no clue about fscrypt, but it can read
> + * unencrypted files. Reading encrypted files will read garbage.
> + */
> +#define EXT4_FEATURE_INCOMPAT_SUPP_LAZY_RO ( EXT4_FEATURE_INCOMPAT_MMP \
> + | EXT4_FEATURE_INCOMPAT_CSUM_SEED \
> + | EXT4_FEATURE_INCOMPAT_ENCRYPT \
> + )
ditto
> +/*
> + * Read-only compat features we support.
> + * If unknown ro compat features are detected, writing to the fs is denied.
> + */
> +#define EXT4_FEATURE_RO_COMPAT_SUPP ( EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER \
> + | EXT4_FEATURE_RO_COMPAT_LARGE_FILE \
> + | EXT4_FEATURE_RO_COMPAT_HUGE_FILE \
> + | EXT4_FEATURE_RO_COMPAT_DIR_NLINK \
> + | EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE \
> + )
ditto
> #define EXT4_BG_INODE_UNINIT 0x0001
> #define EXT4_BG_BLOCK_UNINIT 0x0002
> #define EXT4_BG_INODE_ZEROED 0x0004
Anyway, the substance of this patch is good. Just needs a little cosmetic cleanup.
--Sean
More information about the U-Boot
mailing list