From 98d2b43081972abeb5bb5a087bc3e3197531c46e Mon Sep 17 00:00:00 2001 From: Miklos Szeredi Date: Wed, 25 Oct 2023 16:01:59 +0200 Subject: add unique mount ID If a mount is released then its mnt_id can immediately be reused. This is bad news for user interfaces that want to uniquely identify a mount. Implementing a unique mount ID is trivial (use a 64bit counter). Unfortunately userspace assumes 32bit size and would overflow after the counter reaches 2^32. Introduce a new 64bit ID alongside the old one. Initialize the counter to 2^32, this guarantees that the old and new IDs are never mixed up. Signed-off-by: Miklos Szeredi Link: https://lore.kernel.org/r/20231025140205.3586473-2-mszeredi@redhat.com Reviewed-by: Ian Kent Signed-off-by: Christian Brauner --- include/uapi/linux/stat.h | 1 + 1 file changed, 1 insertion(+) (limited to 'include/uapi/linux') diff --git a/include/uapi/linux/stat.h b/include/uapi/linux/stat.h index 7cab2c65d3d7..2f2ee82d5517 100644 --- a/include/uapi/linux/stat.h +++ b/include/uapi/linux/stat.h @@ -154,6 +154,7 @@ struct statx { #define STATX_BTIME 0x00000800U /* Want/got stx_btime */ #define STATX_MNT_ID 0x00001000U /* Got stx_mnt_id */ #define STATX_DIOALIGN 0x00002000U /* Want/got direct I/O alignment info */ +#define STATX_MNT_ID_UNIQUE 0x00004000U /* Want/got extended stx_mount_id */ #define STATX__RESERVED 0x80000000U /* Reserved for future struct statx expansion */ -- cgit v1.2.3 From 46eae99ef73302f9fb3dddcd67c374b3dffe8fd6 Mon Sep 17 00:00:00 2001 From: Miklos Szeredi Date: Wed, 25 Oct 2023 16:02:02 +0200 Subject: add statmount(2) syscall Add a way to query attributes of a single mount instead of having to parse the complete /proc/$PID/mountinfo, which might be huge. Lookup the mount the new 64bit mount ID. If a mount needs to be queried based on path, then statx(2) can be used to first query the mount ID belonging to the path. Design is based on a suggestion by Linus: "So I'd suggest something that is very much like "statfsat()", which gets a buffer and a length, and returns an extended "struct statfs" *AND* just a string description at the end." The interface closely mimics that of statx. Handle ASCII attributes by appending after the end of the structure (as per above suggestion). Pointers to strings are stored in u64 members to make the structure the same regardless of pointer size. Strings are nul terminated. Link: https://lore.kernel.org/all/CAHk-=wh5YifP7hzKSbwJj94+DZ2czjrZsczy6GBimiogZws=rg@mail.gmail.com/ Signed-off-by: Miklos Szeredi Link: https://lore.kernel.org/r/20231025140205.3586473-5-mszeredi@redhat.com Reviewed-by: Ian Kent [Christian Brauner : various minor changes] Signed-off-by: Christian Brauner --- include/uapi/linux/mount.h | 53 ++++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 53 insertions(+) (limited to 'include/uapi/linux') diff --git a/include/uapi/linux/mount.h b/include/uapi/linux/mount.h index bb242fdcfe6b..afdf4f2f6672 100644 --- a/include/uapi/linux/mount.h +++ b/include/uapi/linux/mount.h @@ -138,4 +138,57 @@ struct mount_attr { /* List of all mount_attr versions. */ #define MOUNT_ATTR_SIZE_VER0 32 /* sizeof first published struct */ + +/* + * Structure for getting mount/superblock/filesystem info with statmount(2). + * + * The interface is similar to statx(2): individual fields or groups can be + * selected with the @mask argument of statmount(). Kernel will set the @mask + * field according to the supported fields. + * + * If string fields are selected, then the caller needs to pass a buffer that + * has space after the fixed part of the structure. Nul terminated strings are + * copied there and offsets relative to @str are stored in the relevant fields. + * If the buffer is too small, then EOVERFLOW is returned. The actually used + * size is returned in @size. + */ +struct statmount { + __u32 size; /* Total size, including strings */ + __u32 __spare1; + __u64 mask; /* What results were written */ + __u32 sb_dev_major; /* Device ID */ + __u32 sb_dev_minor; + __u64 sb_magic; /* ..._SUPER_MAGIC */ + __u32 sb_flags; /* SB_{RDONLY,SYNCHRONOUS,DIRSYNC,LAZYTIME} */ + __u32 fs_type; /* [str] Filesystem type */ + __u64 mnt_id; /* Unique ID of mount */ + __u64 mnt_parent_id; /* Unique ID of parent (for root == mnt_id) */ + __u32 mnt_id_old; /* Reused IDs used in proc/.../mountinfo */ + __u32 mnt_parent_id_old; + __u64 mnt_attr; /* MOUNT_ATTR_... */ + __u64 mnt_propagation; /* MS_{SHARED,SLAVE,PRIVATE,UNBINDABLE} */ + __u64 mnt_peer_group; /* ID of shared peer group */ + __u64 mnt_master; /* Mount receives propagation from this ID */ + __u64 propagate_from; /* Propagation from in current namespace */ + __u32 mnt_root; /* [str] Root of mount relative to root of fs */ + __u32 mnt_point; /* [str] Mountpoint relative to current root */ + __u64 __spare2[50]; + char str[]; /* Variable size part containing strings */ +}; + +struct mnt_id_req { + __u64 mnt_id; + __u64 request_mask; +}; + +/* + * @mask bits for statmount(2) + */ +#define STATMOUNT_SB_BASIC 0x00000001U /* Want/got sb_... */ +#define STATMOUNT_MNT_BASIC 0x00000002U /* Want/got mnt_... */ +#define STATMOUNT_PROPAGATE_FROM 0x00000004U /* Want/got propagate_from */ +#define STATMOUNT_MNT_ROOT 0x00000008U /* Want/got mnt_root */ +#define STATMOUNT_MNT_POINT 0x00000010U /* Want/got mnt_point */ +#define STATMOUNT_FS_TYPE 0x00000020U /* Want/got fs_type */ + #endif /* _UAPI_LINUX_MOUNT_H */ -- cgit v1.2.3 From b4c2bea8ceaa50cd42a8f73667389d801a3ecf2d Mon Sep 17 00:00:00 2001 From: Miklos Szeredi Date: Wed, 25 Oct 2023 16:02:03 +0200 Subject: add listmount(2) syscall Add way to query the children of a particular mount. This is a more flexible way to iterate the mount tree than having to parse /proc/self/mountinfo. Lookup the mount by the new 64bit mount ID. If a mount needs to be queried based on path, then statx(2) can be used to first query the mount ID belonging to the path. Return an array of new (64bit) mount ID's. Without privileges only mounts are listed which are reachable from the task's root. Folded into this patch are several later improvements. Keeping them separate would make the history pointlessly confusing: * Recursive listing of mounts is the default now (cf. [1]). * Remove explicit LISTMOUNT_UNREACHABLE flag (cf. [1]) and fail if mount is unreachable from current root. This also makes permission checking consistent with statmount() (cf. [3]). * Start listing mounts in unique mount ID order (cf. [2]) to allow continuing listmount() from a midpoint. * Allow to continue listmount(). The @request_mask parameter is renamed and to @param to be usable by both statmount() and listmount(). If @param is set to a mount id then listmount() will continue listing mounts from that id on. This allows listing mounts in multiple listmount invocations without having to resize the buffer. If @param is zero then the listing starts from the beginning (cf. [4]). * Don't return EOVERFLOW, instead return the buffer size which allows to detect a full buffer as well (cf. [4]). Signed-off-by: Miklos Szeredi Link: https://lore.kernel.org/r/20231025140205.3586473-6-mszeredi@redhat.com Reviewed-by: Ian Kent Link: https://lore.kernel.org/r/20231128160337.29094-2-mszeredi@redhat.com [1] (folded) Link: https://lore.kernel.org/r/20231128160337.29094-3-mszeredi@redhat.com [2] (folded) Link: https://lore.kernel.org/r/20231128160337.29094-4-mszeredi@redhat.com [3] (folded) Link: https://lore.kernel.org/r/20231128160337.29094-5-mszeredi@redhat.com [4] (folded) [Christian Brauner : various smaller fixes] Signed-off-by: Christian Brauner --- include/uapi/linux/mount.h | 14 +++++++++++++- 1 file changed, 13 insertions(+), 1 deletion(-) (limited to 'include/uapi/linux') diff --git a/include/uapi/linux/mount.h b/include/uapi/linux/mount.h index afdf4f2f6672..dc9a0112d819 100644 --- a/include/uapi/linux/mount.h +++ b/include/uapi/linux/mount.h @@ -176,9 +176,16 @@ struct statmount { char str[]; /* Variable size part containing strings */ }; +/* + * Structure for passing mount ID and miscellaneous parameters to statmount(2) + * and listmount(2). + * + * For statmount(2) @param represents the request mask. + * For listmount(2) @param represents the last listed mount id (or zero). + */ struct mnt_id_req { __u64 mnt_id; - __u64 request_mask; + __u64 param; }; /* @@ -191,4 +198,9 @@ struct mnt_id_req { #define STATMOUNT_MNT_POINT 0x00000010U /* Want/got mnt_point */ #define STATMOUNT_FS_TYPE 0x00000020U /* Want/got fs_type */ +/* + * Special @mnt_id values that can be passed to listmount + */ +#define LSMT_ROOT 0xffffffffffffffff /* root mount */ + #endif /* _UAPI_LINUX_MOUNT_H */ -- cgit v1.2.3 From 35e27a5744131996061e6e323f1bcb4c827ae867 Mon Sep 17 00:00:00 2001 From: Christian Brauner Date: Wed, 29 Nov 2023 12:27:15 +0100 Subject: fs: keep struct mnt_id_req extensible Make it extensible so that we have the liberty to reuse it in future mount-id based apis. Treat zero size as the first published struct. Signed-off-by: Christian Brauner --- include/uapi/linux/mount.h | 5 +++++ 1 file changed, 5 insertions(+) (limited to 'include/uapi/linux') diff --git a/include/uapi/linux/mount.h b/include/uapi/linux/mount.h index dc9a0112d819..ad5478dbad00 100644 --- a/include/uapi/linux/mount.h +++ b/include/uapi/linux/mount.h @@ -184,10 +184,15 @@ struct statmount { * For listmount(2) @param represents the last listed mount id (or zero). */ struct mnt_id_req { + __u32 size; + __u32 spare; __u64 mnt_id; __u64 param; }; +/* List of all mnt_id_req versions. */ +#define MNT_ID_REQ_SIZE_VER0 24 /* sizeof first published struct */ + /* * @mask bits for statmount(2) */ -- cgit v1.2.3