Re: [PATCH] Add a pair of system calls to make extended file stats available [ver #3]

Re: [PATCH] Add a pair of system calls to make extended file stats available [ver #3]

[Michael Kerrisk]

Hi David,

[Please CC linux-api@ on patches that change the API/ABI]

On Thu, Jul 1, 2010 at 1:36 AM, David Howells <dhowells-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org> wrote:
> Add a pair of system calls to make extended file stats available, including
> file creation time, inode version and data version where available through the
> underlying filesystem.

Just some random thoughts here. I've not tried to guess the overhead
of these ideas...

* Include information from the "inode_info" structure, most notably
i_flags, but perhaps other info as well.

* Return a bit mask indicating the presence of additional information
associated with the i-node. Here, I am thinking of flags that indicate
that the file has any of the following: capabilities, an ACL, and
extended attributes (obviously a superset of the previous). I could
imagine some apps that, having got the xstat info, would be interested
to obtain some of this other info.

Obviously, the above only make sense if the overhead of providing the
extra information is low.

> [This depends on the previously posted pair of patches to (a) constify a number
>  of syscall string and buffer arguments and (b) rearrange AFS's use of
>  i_version and i_generation].
>
> The following structures are defined for their use:
>
>        struct xstat_parameters {
>                unsigned long long      request_mask;
>        };
>
>        struct xstat_dev {
>                unsigned int            major, minor;
>        };
>
>        struct xstat_time {
>                unsigned long long      tv_sec, tv_nsec;
>        };
>
>        struct xstat {
>                unsigned int            st_mode;
>                unsigned int            st_nlink;
>                unsigned int            st_uid;
>                unsigned int            st_gid;
>                struct xstat_dev        st_rdev;
>                struct xstat_dev        st_dev;
>                struct xstat_time       st_atime;
>                struct xstat_time       st_mtime;
>                struct xstat_time       st_ctime;
>                struct xstat_time       st_btime;
>                unsigned long long      st_ino;
>                unsigned long long      st_size;
>                unsigned long long      st_blksize;
>                unsigned long long      st_blocks;
>                unsigned long long      st_gen;
>                unsigned long long      st_data_version;
>                unsigned long long      st_result_mask;
>                unsigned long long      st_extra_results[0];
>        };
>
> where st_btime is the file creation time, st_gen is the inode generation
> (i_generation), st_data_version is the data version number (i_version),
> request_mask and st_result_mask are bitmasks of data desired/provided and
> st_extra_results[] is where as-yet undefined fields are appended.
>
> The defined bits in request_mask and st_result_mask are:
>
>        XSTAT_REQUEST_MODE              Want/got st_mode
>        XSTAT_REQUEST_NLINK             Want/got st_nlink
>        XSTAT_REQUEST_UID               Want/got st_uid
>        XSTAT_REQUEST_GID               Want/got st_gid
>        XSTAT_REQUEST_RDEV              Want/got st_rdev
>        XSTAT_REQUEST_ATIME             Want/got st_atime
>        XSTAT_REQUEST_MTIME             Want/got st_mtime
>        XSTAT_REQUEST_CTIME             Want/got st_ctime
>        XSTAT_REQUEST_INO               Want/got st_ino
>        XSTAT_REQUEST_SIZE              Want/got st_size
>        XSTAT_REQUEST_BLOCKS            Want/got st_blocks
>        XSTAT_REQUEST__BASIC_STATS      The stuff in the normal stat struct
>        XSTAT_REQUEST_BTIME             Want/got st_btime
>        XSTAT_REQUEST_GEN               Want/got st_gen
>        XSTAT_REQUEST_DATA_VERSION      Want/got st_data_version
>        XSTAT_REQUEST__EXTENDED_STATS   The stuff in the xstat struct
>        XSTAT_REQUEST__ALL_STATS        The defined set of requestables
>
> The system calls are:
>
>        ssize_t ret = xstat(int dfd,
>                            const char *filename,
>                            unsigned flags,
>                            const struct xstat_parameters *params,
>                            struct xstat *buffer,
>                            size_t buflen);
>
>        ssize_t ret = fxstat(unsigned fd,
>                             unsigned flags,
>                             const struct xstat_parameters *params,
>                             struct xstat *buffer,
>                             size_t buflen);
>
>
> The dfd, filename, flags and fd parameters indicate the file to query.  There
> is no equivalent of lstat() as that can be emulated with xstat() by passing
> AT_SYMLINK_NOFOLLOW in flags.
>
> AT_FORCE_ATTR_SYNC can also be set in flags.  This will require a network
> filesystem to synchronise its attributes with the server.
>
> When the system call is executed, the request_mask bitmask is read from the
> parameter block to work out what the user is requesting.  If params is NULL,
> then request_mask will be assumed to be XSTAT_REQUEST__GET_ANYWAY.
>
> The request_mask should be set by the caller to specify extra results that the
> caller may desire.  These come in a number of classes:
>
>  (0) dev, blksize.
>
>     These are local data and are always available.
>
>  (1) mode, nlinks, uid, gid, [amc]time, ino, size, blocks.
>
>     These will be returned whether the caller asks for them or not.  The
>     corresponding bits in result_mask will be set to indicate their presence.
>
>     If the caller didn't ask for them, then they may be approximated.  For
>     example, NFS won't waste any time updating them from the server, unless as
>     a byproduct of updating something requested.
>
>  (2) rdev.
>
>     As for class (1), but this won't be returned if the file is not a blockdev
>     or chardev.  The bit will be cleared if the value is not returned.
>
>  (3) File creation time, inode generation and data version.
>
>     These will be returned if available whether the caller asked for them or
>     not.  The corresponding bits in result_mask will be set or cleared as
>     appropriate to indicate their presence.
>
>     If the caller didn't ask for them, then they may be approximated.  For
>     example, NFS won't waste any time updating them from the server, unless
>     as a byproduct of updating something requested.
>
>  (4) Extra results.
>
>     These will only be returned if the caller asked for them by setting their
>     bits in request_mask.  They will be placed in the buffer after the xstat
>     struct in ascending result_mask bit order.  Any bit set in request_mask
>     mask will be left set in result_mask if the result is available and
>     cleared otherwise.
>
>     The pointer into the results list will be rounded up to the nearest 8-byte
>     boundary after each result is written in.  The size of each extra result
>     is specific to the definition for that result.
>
>     No extra results are currently defined.
>
> If the buffer is insufficiently big, the syscall returns the amount of space it
> will need to write the complete result set and returns a partial result in the
> buffer.
>
> At the moment, this will only work on x86_64 as it requires system calls to be
> wired up.
>
>
> ===========
> FILESYSTEMS
> ===========
>
> The following filesystems have been modified to make use of this facility:
>
>  (*) Ext4.  This will return the creation time and inode version number for all
>     files.  It will, however, only return the data version number for
>     directories unless the I_VERSION option is set on the filesystem.
>
>  (*) AFS.  This will return the vnode ID uniquifier as the inode version and
>     the AFS data version number as the data version.  There is no file
>     creation time available.
>
>     AFS should go to the server if AT_FORCE_ATTR_SYNC is specified.
>
>  (*) NFS.  This will return the change attribute if NFSv4 only.  No other extra
>     values are returned at this time.
>
>     If AT_FORCE_ATTR_SYNC is set or mtime, ctime or data_version (NFSv4 only)
>     are asked for then the outstanding writes will be written to the server
>     first.
>
>     If AT_FORCE_ATTR_SYNC is set or atime is requested then the attributes
>     will be reread unconditionally, otherwise if any of data version (NFSv4
>     only) XSTAT_REQUEST__BASIC_STATS are requested, then the attributes will
>     be reread if the cached attributes have expired.
>
>
> =======
> TESTING
> =======
>
> The following test program can be used to test the xstat system call:
>
>        #define _GNU_SOURCE
>        #define _ATFILE_SOURCE
>        #include <stdio.h>
>        #include <stdlib.h>
>        #include <string.h>
>        #include <unistd.h>
>        #include <fcntl.h>
>        #include <time.h>
>        #include <sys/syscall.h>
>        #include <sys/stat.h>
>        #include <sys/types.h>
>
>        #define AT_FORCE_ATTR_SYNC      0x800
>
>        struct xstat_parameters {
>                unsigned long long      request_mask;
>        #define XSTAT_REQUEST_MODE              0x00000001ULL
>        #define XSTAT_REQUEST_NLINK             0x00000002ULL
>        #define XSTAT_REQUEST_UID               0x00000004ULL
>        #define XSTAT_REQUEST_GID               0x00000008ULL
>        #define XSTAT_REQUEST_RDEV              0x00000010ULL
>        #define XSTAT_REQUEST_ATIME             0x00000020ULL
>        #define XSTAT_REQUEST_MTIME             0x00000040ULL
>        #define XSTAT_REQUEST_CTIME             0x00000080ULL
>        #define XSTAT_REQUEST_INO               0x00000100ULL
>        #define XSTAT_REQUEST_SIZE              0x00000200ULL
>        #define XSTAT_REQUEST_BLOCKS            0x00000400ULL
>        #define XSTAT_REQUEST__BASIC_STATS      0x000007ffULL
>        #define XSTAT_REQUEST_BTIME             0x00000800ULL
>        #define XSTAT_REQUEST_GEN               0x00001000ULL
>        #define XSTAT_REQUEST_DATA_VERSION      0x00002000ULL
>        #define XSTAT_REQUEST__EXTENDED_STATS   0x00003fffULL
>        #define XSTAT_REQUEST__ALL_STATS        0x00003fffULL
>        };
>
>        struct xstat_dev {
>                unsigned int    major;
>                unsigned int    minor;
>        };
>
>        struct xstat_time {
>                unsigned long long      tv_sec;
>                unsigned long long      tv_nsec;
>        };
>
>        struct xstat {
>                unsigned int            st_mode;
>                unsigned int            st_nlink;
>                unsigned int            st_uid;
>                unsigned int            st_gid;
>                struct xstat_dev        st_rdev;
>                struct xstat_dev        st_dev;
>                struct xstat_time       st_atim;
>                struct xstat_time       st_mtim;
>                struct xstat_time       st_ctim;
>                struct xstat_time       st_btim;
>                unsigned long long      st_ino;
>                unsigned long long      st_size;
>                unsigned long long      st_blksize;
>                unsigned long long      st_blocks;
>                unsigned long long      st_gen;
>                unsigned long long      st_data_version;
>                unsigned long long      st_result_mask;
>                unsigned long long      st_extra_results[0];
>        };
>
>        #define __NR_xstat                              300
>        #define __NR_fxstat                             301
>
>        static __attribute__((unused))
>        ssize_t xstat(int dfd, const char *filename, unsigned flags,
>                      struct xstat_parameters *params,
>                      struct xstat *buffer, size_t bufsize)
>        {
>                return syscall(__NR_xstat, dfd, filename, flags,
>                               params, buffer, bufsize);
>        }
>
>        static __attribute__((unused))
>        ssize_t fxstat(int fd, unsigned flags,
>                       struct xstat_parameters *params,
>                       struct xstat *buffer, size_t bufsize)
>        {
>                return syscall(__NR_fxstat, fd, flags,
>                               params, buffer, bufsize);
>        }
>
>        static void print_time(const char *field, const struct xstat_time *xstm)
>        {
>                struct tm tm;
>                time_t tim;
>                char buffer[100];
>                int len;
>
>                tim = xstm->tv_sec;
>                if (!localtime_r(&tim, &tm)) {
>                        perror("localtime_r");
>                        exit(1);
>                }
>                len = strftime(buffer, 100, "%F %T", &tm);
>                if (len == 0) {
>                        perror("strftime");
>                        exit(1);
>                }
>                printf("%s", field);
>                fwrite(buffer, 1, len, stdout);
>                printf(".%09llu", xstm->tv_nsec);
>                len = strftime(buffer, 100, "%z", &tm);
>                if (len == 0) {
>                        perror("strftime2");
>                        exit(1);
>                }
>                fwrite(buffer, 1, len, stdout);
>                printf("\n");
>        }
>
>        static void dump_xstat(struct xstat *xst)
>        {
>                char buffer[256], ft;
>
>                printf("results=%llx\n", xst->st_result_mask);
>
>                printf(" ");
>                if (xst->st_result_mask & XSTAT_REQUEST_SIZE)
>                        printf(" Size: %-15llu", xst->st_size);
>                if (xst->st_result_mask & XSTAT_REQUEST_BLOCKS)
>                        printf(" Blocks: %-10llu", xst->st_blocks);
>                printf(" IO Block: %-6llu ", xst->st_blksize);
>                if (xst->st_result_mask & XSTAT_REQUEST_MODE) {
>                        switch (xst->st_mode & S_IFMT) {
>                        case S_IFIFO:   printf(" FIFO\n");                      ft = 'p'; break;
>                        case S_IFCHR:   printf(" character special file\n");    ft = 'c'; break;
>                        case S_IFDIR:   printf(" directory\n");                 ft = 'd'; break;
>                        case S_IFBLK:   printf(" block special file\n");        ft = 'b'; break;
>                        case S_IFREG:   printf(" regular file\n");              ft = '-'; break;
>                        case S_IFLNK:   printf(" symbolic link\n");             ft = 'l'; break;
>                        case S_IFSOCK:  printf(" socket\n");                    ft = 's'; break;
>                        default:
>                                printf("unknown type (%o)\n", xst->st_mode & S_IFMT);
>                                ft = '?';
>                                break;
>                        }
>                }
>
>                sprintf(buffer, "%02x:%02x", xst->st_dev.major, xst->st_dev.minor);
>                printf("Device: %-15s", buffer);
>                if (xst->st_result_mask & XSTAT_REQUEST_INO)
>                        printf(" Inode: %-11llu", xst->st_ino);
>                if (xst->st_result_mask & XSTAT_REQUEST_SIZE)
>                        printf(" Links: %-5u", xst->st_nlink);
>                if (xst->st_result_mask & XSTAT_REQUEST_RDEV)
>                        printf(" Device type: %u,%u",
>                               xst->st_rdev.major, xst->st_rdev.minor);
>                printf("\n");
>
>                if (xst->st_result_mask & XSTAT_REQUEST_MODE)
>                        printf("Access: (%04o/%c%c%c%c%c%c%c%c%c%c)  ",
>                               xst->st_mode & 07777,
>                               ft,
>                               xst->st_mode & S_IRUSR ? 'r' : '-',
>                               xst->st_mode & S_IWUSR ? 'w' : '-',
>                               xst->st_mode & S_IXUSR ? 'x' : '-',
>                               xst->st_mode & S_IRGRP ? 'r' : '-',
>                               xst->st_mode & S_IWGRP ? 'w' : '-',
>                               xst->st_mode & S_IXGRP ? 'x' : '-',
>                               xst->st_mode & S_IROTH ? 'r' : '-',
>                               xst->st_mode & S_IWOTH ? 'w' : '-',
>                               xst->st_mode & S_IXOTH ? 'x' : '-');
>                if (xst->st_result_mask & XSTAT_REQUEST_UID)
>                        printf("Uid: %d   \n", xst->st_uid);
>                if (xst->st_result_mask & XSTAT_REQUEST_GID)
>                        printf("Gid: %u\n", xst->st_gid);
>
>                if (xst->st_result_mask & XSTAT_REQUEST_ATIME)
>                        print_time("Access: ", &xst->st_atim);
>                if (xst->st_result_mask & XSTAT_REQUEST_MTIME)
>                        print_time("Modify: ", &xst->st_mtim);
>                if (xst->st_result_mask & XSTAT_REQUEST_CTIME)
>                        print_time("Change: ", &xst->st_ctim);
>                if (xst->st_result_mask & XSTAT_REQUEST_BTIME)
>                        print_time("Create: ", &xst->st_btim);
>
>                if (xst->st_result_mask & XSTAT_REQUEST_GEN)
>                        printf("Inode version: %llxh\n", xst->st_gen);
>                if (xst->st_result_mask & XSTAT_REQUEST_DATA_VERSION)
>                        printf("Data version: %llxh\n", xst->st_data_version);
>        }
>
>        int main(int argc, char **argv)
>        {
>                struct xstat_parameters params;
>                struct xstat xst;
>                int ret, atflag = AT_SYMLINK_NOFOLLOW;
>
>                unsigned long long query =
>                        XSTAT_REQUEST__BASIC_STATS |
>                        XSTAT_REQUEST_BTIME |
>                        XSTAT_REQUEST_GEN |
>                        XSTAT_REQUEST_DATA_VERSION;
>
>                for (argv++; *argv; argv++) {
>                        if (strcmp(*argv, "-F") == 0) {
>                                atflag |= AT_FORCE_ATTR_SYNC;
>                                continue;
>                        }
>                        if (strcmp(*argv, "-L") == 0) {
>                                atflag &= ~AT_SYMLINK_NOFOLLOW;
>                                continue;
>                        }
>                        if (strcmp(*argv, "-O") == 0) {
>                                query &= ~XSTAT_REQUEST__BASIC_STATS;
>                                continue;
>                        }
>
>                        memset(&xst, 0xbf, sizeof(xst));
>                        params.request_mask = query;
>                        ret = xstat(AT_FDCWD, *argv, atflag, &params, &xst, sizeof(xst));
>                        printf("xstat(%s) = %d\n", *argv, ret);
>                        if (ret < 0) {
>                                perror(*argv);
>                                exit(1);
>                        }
>
>                        dump_xstat(&xst);
>                }
>                return 0;
>        }
>
> Just compile and run, passing it paths to the files you want to examine:
>
>        [root@andromeda ~]# /tmp/xstat -O /dev/tty
>        xstat(/dev/tty) = 152
>        results=7ff
>          Size: 0               Blocks: 0          IO Block: 4096    character special file
>        Device: 00:0f           Inode: 246         Links: 1     Device type: 5,0
>        Access: (0666/crw-rw-rw-)  Uid: 0
>        Gid: 5
>        Access: 2010-06-30 16:25:01.813517001+0100
>        Modify: 2010-06-30 16:25:01.813517001+0100
>        Change: 2010-06-30 16:25:01.813517001+0100
>
>        [root@andromeda ~]# /tmp/xstat /var/cache/fscache/cache/
>        xstat(/var/cache/fscache/cache/) = 152
>        results=3fef
>          Size: 4096            Blocks: 16         IO Block: 4096    directory
>        Device: 08:06           Inode: 130561      Links: 3
>        Access: (0700/drwx------)  Uid: 0
>        Gid: 0
>        Access: 2010-06-29 18:16:33.680703545+0100
>        Modify: 2010-06-29 18:16:20.132786632+0100
>        Change: 2010-06-29 18:16:20.132786632+0100
>        Create: 2010-06-25 15:17:39.471199293+0100
>        Inode version: f585ab70h
>        Data version: 2h
>
> Signed-off-by: David Howells <dhowells-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org>
> ---
>
>  arch/x86/include/asm/unistd_32.h |    4 +
>  arch/x86/include/asm/unistd_64.h |    4 +
>  fs/afs/inode.c                   |   11 +-
>  fs/ecryptfs/inode.c              |    2
>  fs/ext4/ext4.h                   |    2
>  fs/ext4/file.c                   |    2
>  fs/ext4/inode.c                  |   27 +++++-
>  fs/ext4/namei.c                  |    2
>  fs/ext4/symlink.c                |    2
>  fs/nfs/inode.c                   |   46 +++++++---
>  fs/nfsd/nfs3proc.c               |    2
>  fs/nfsd/nfs3xdr.c                |    4 +
>  fs/nfsd/nfs4xdr.c                |    4 +
>  fs/nfsd/nfsproc.c                |    6 +
>  fs/nfsd/nfsxdr.c                 |    2
>  fs/stat.c                        |  175 ++++++++++++++++++++++++++++++++++----
>  include/linux/fcntl.h            |    1
>  include/linux/fs.h               |    2
>  include/linux/stat.h             |  103 ++++++++++++++++++++++
>  include/linux/syscalls.h         |    9 ++
>  20 files changed, 368 insertions(+), 42 deletions(-)
>
> diff --git a/arch/x86/include/asm/unistd_32.h b/arch/x86/include/asm/unistd_32.h
> index beb9b5f..a9953cc 100644
> --- a/arch/x86/include/asm/unistd_32.h
> +++ b/arch/x86/include/asm/unistd_32.h
> @@ -343,10 +343,12 @@
>  #define __NR_rt_tgsigqueueinfo 335
>  #define __NR_perf_event_open   336
>  #define __NR_recvmmsg          337
> +#define __NR_xstat             338
> +#define __NR_fxstat            339
>
>  #ifdef __KERNEL__
>
> -#define NR_syscalls 338
> +#define NR_syscalls 340
>
>  #define __ARCH_WANT_IPC_PARSE_VERSION
>  #define __ARCH_WANT_OLD_READDIR
> diff --git a/arch/x86/include/asm/unistd_64.h b/arch/x86/include/asm/unistd_64.h
> index ff4307b..c90d240 100644
> --- a/arch/x86/include/asm/unistd_64.h
> +++ b/arch/x86/include/asm/unistd_64.h
> @@ -663,6 +663,10 @@ __SYSCALL(__NR_rt_tgsigqueueinfo, sys_rt_tgsigqueueinfo)
>  __SYSCALL(__NR_perf_event_open, sys_perf_event_open)
>  #define __NR_recvmmsg                          299
>  __SYSCALL(__NR_recvmmsg, sys_recvmmsg)
> +#define __NR_xstat                             300
> +__SYSCALL(__NR_xstat, sys_xstat)
> +#define __NR_fxstat                            301
> +__SYSCALL(__NR_fxstat, sys_fxstat)
>
>  #ifndef __NO_STUBS
>  #define __ARCH_WANT_OLD_READDIR
> diff --git a/fs/afs/inode.c b/fs/afs/inode.c
> index ee3190a..f624c5a 100644
> --- a/fs/afs/inode.c
> +++ b/fs/afs/inode.c
> @@ -300,16 +300,17 @@ error_unlock:
>  /*
>  * read the attributes of an inode
>  */
> -int afs_getattr(struct vfsmount *mnt, struct dentry *dentry,
> -                     struct kstat *stat)
> +int afs_getattr(struct vfsmount *mnt, struct dentry *dentry, struct kstat *stat)
>  {
> -       struct inode *inode;
> -
> -       inode = dentry->d_inode;
> +       struct inode *inode = dentry->d_inode;
>
>        _enter("{ ino=%lu v=%u }", inode->i_ino, inode->i_generation);
>
>        generic_fillattr(inode, stat);
> +
> +       stat->result_mask |= XSTAT_REQUEST_GEN | XSTAT_REQUEST_DATA_VERSION;
> +       stat->gen = inode->i_generation;
> +       stat->data_version = inode->i_version;
>        return 0;
>  }
>
> diff --git a/fs/ecryptfs/inode.c b/fs/ecryptfs/inode.c
> index 31ef525..0b02272 100644
> --- a/fs/ecryptfs/inode.c
> +++ b/fs/ecryptfs/inode.c
> @@ -994,6 +994,8 @@ int ecryptfs_getattr(struct vfsmount *mnt, struct dentry *dentry,
>        struct kstat lower_stat;
>        int rc;
>
> +       lower_stat.query_flags = stat->query_flags;
> +       lower_stat.request_mask = stat->request_mask | XSTAT_REQUEST_BLOCKS;
>        rc = vfs_getattr(ecryptfs_dentry_to_lower_mnt(dentry),
>                         ecryptfs_dentry_to_lower(dentry), &lower_stat);
>        if (!rc) {
> diff --git a/fs/ext4/ext4.h b/fs/ext4/ext4.h
> index 19a4de5..96823f3 100644
> --- a/fs/ext4/ext4.h
> +++ b/fs/ext4/ext4.h
> @@ -1571,6 +1571,8 @@ extern int  ext4_write_inode(struct inode *, struct writeback_control *);
>  extern int  ext4_setattr(struct dentry *, struct iattr *);
>  extern int  ext4_getattr(struct vfsmount *mnt, struct dentry *dentry,
>                                struct kstat *stat);
> +extern int  ext4_file_getattr(struct vfsmount *mnt, struct dentry *dentry,
> +                               struct kstat *stat);
>  extern void ext4_delete_inode(struct inode *);
>  extern int  ext4_sync_inode(handle_t *, struct inode *);
>  extern void ext4_dirty_inode(struct inode *);
> diff --git a/fs/ext4/file.c b/fs/ext4/file.c
> index 5313ae4..18c29ab 100644
> --- a/fs/ext4/file.c
> +++ b/fs/ext4/file.c
> @@ -150,7 +150,7 @@ const struct file_operations ext4_file_operations = {
>  const struct inode_operations ext4_file_inode_operations = {
>        .truncate       = ext4_truncate,
>        .setattr        = ext4_setattr,
> -       .getattr        = ext4_getattr,
> +       .getattr        = ext4_file_getattr,
>  #ifdef CONFIG_EXT4_FS_XATTR
>        .setxattr       = generic_setxattr,
>        .getxattr       = generic_getxattr,
> diff --git a/fs/ext4/inode.c b/fs/ext4/inode.c
> index 42272d6..f9a730a 100644
> --- a/fs/ext4/inode.c
> +++ b/fs/ext4/inode.c
> @@ -5550,12 +5550,33 @@ err_out:
>  int ext4_getattr(struct vfsmount *mnt, struct dentry *dentry,
>                 struct kstat *stat)
>  {
> -       struct inode *inode;
> -       unsigned long delalloc_blocks;
> +       struct inode *inode = dentry->d_inode;
>
> -       inode = dentry->d_inode;
>        generic_fillattr(inode, stat);
>
> +       stat->result_mask |= XSTAT_REQUEST_BTIME;
> +       stat->btime.tv_sec = EXT4_I(inode)->i_crtime.tv_sec;
> +       stat->btime.tv_nsec = EXT4_I(inode)->i_crtime.tv_nsec;
> +
> +       if (inode->i_ino != EXT4_ROOT_INO) {
> +               stat->result_mask |= XSTAT_REQUEST_GEN;
> +               stat->gen = inode->i_generation;
> +       }
> +       if (S_ISDIR(inode->i_mode) || test_opt(inode->i_sb, I_VERSION)) {
> +               stat->result_mask |= XSTAT_REQUEST_DATA_VERSION;
> +               stat->data_version = inode->i_version;
> +       }
> +       return 0;
> +}
> +
> +int ext4_file_getattr(struct vfsmount *mnt, struct dentry *dentry,
> +                     struct kstat *stat)
> +{
> +       struct inode *inode = dentry->d_inode;
> +       unsigned long delalloc_blocks;
> +
> +       ext4_getattr(mnt, dentry, stat);
> +
>        /*
>         * We can't update i_blocks if the block allocation is delayed
>         * otherwise in the case of system crash before the real block
> diff --git a/fs/ext4/namei.c b/fs/ext4/namei.c
> index a43e661..0f776c7 100644
> --- a/fs/ext4/namei.c
> +++ b/fs/ext4/namei.c
> @@ -2542,6 +2542,7 @@ const struct inode_operations ext4_dir_inode_operations = {
>        .mknod          = ext4_mknod,
>        .rename         = ext4_rename,
>        .setattr        = ext4_setattr,
> +       .getattr        = ext4_getattr,
>  #ifdef CONFIG_EXT4_FS_XATTR
>        .setxattr       = generic_setxattr,
>        .getxattr       = generic_getxattr,
> @@ -2554,6 +2555,7 @@ const struct inode_operations ext4_dir_inode_operations = {
>
>  const struct inode_operations ext4_special_inode_operations = {
>        .setattr        = ext4_setattr,
> +       .getattr        = ext4_getattr,
>  #ifdef CONFIG_EXT4_FS_XATTR
>        .setxattr       = generic_setxattr,
>        .getxattr       = generic_getxattr,
> diff --git a/fs/ext4/symlink.c b/fs/ext4/symlink.c
> index ed9354a..d8fe7fb 100644
> --- a/fs/ext4/symlink.c
> +++ b/fs/ext4/symlink.c
> @@ -35,6 +35,7 @@ const struct inode_operations ext4_symlink_inode_operations = {
>        .follow_link    = page_follow_link_light,
>        .put_link       = page_put_link,
>        .setattr        = ext4_setattr,
> +       .getattr        = ext4_getattr,
>  #ifdef CONFIG_EXT4_FS_XATTR
>        .setxattr       = generic_setxattr,
>        .getxattr       = generic_getxattr,
> @@ -47,6 +48,7 @@ const struct inode_operations ext4_fast_symlink_inode_operations = {
>        .readlink       = generic_readlink,
>        .follow_link    = ext4_follow_link,
>        .setattr        = ext4_setattr,
> +       .getattr        = ext4_getattr,
>  #ifdef CONFIG_EXT4_FS_XATTR
>        .setxattr       = generic_setxattr,
>        .getxattr       = generic_getxattr,
> diff --git a/fs/nfs/inode.c b/fs/nfs/inode.c
> index 099b351..8c6de96 100644
> --- a/fs/nfs/inode.c
> +++ b/fs/nfs/inode.c
> @@ -495,11 +495,21 @@ void nfs_setattr_update_inode(struct inode *inode, struct iattr *attr)
>  int nfs_getattr(struct vfsmount *mnt, struct dentry *dentry, struct kstat *stat)
>  {
>        struct inode *inode = dentry->d_inode;
> +       unsigned force = stat->query_flags & AT_FORCE_ATTR_SYNC;
>        int need_atime = NFS_I(inode)->cache_validity & NFS_INO_INVALID_ATIME;
>        int err;
>
> -       /* Flush out writes to the server in order to update c/mtime.  */
> -       if (S_ISREG(inode->i_mode)) {
> +       if (NFS_SERVER(inode)->nfs_client->rpc_ops->version < 4)
> +               stat->request_mask &= ~XSTAT_REQUEST_DATA_VERSION;
> +
> +       /* Flush out writes to the server in order to update c/mtime
> +        * or data version if the user wants them */
> +       if ((force || stat->request_mask & (XSTAT_REQUEST_MTIME |
> +                                           XSTAT_REQUEST_CTIME |
> +                                           XSTAT_REQUEST_DATA_VERSION
> +                                           )) &&
> +           S_ISREG(inode->i_mode)
> +           ) {
>                err = filemap_write_and_wait(inode->i_mapping);
>                if (err)
>                        goto out;
> @@ -514,18 +524,30 @@ int nfs_getattr(struct vfsmount *mnt, struct dentry *dentry, struct kstat *stat)
>         *  - NFS never sets MS_NOATIME or MS_NODIRATIME so there is
>         *    no point in checking those.
>         */
> -       if ((mnt->mnt_flags & MNT_NOATIME) ||
> -           ((mnt->mnt_flags & MNT_NODIRATIME) && S_ISDIR(inode->i_mode)))
> +       if (!(stat->request_mask & XSTAT_REQUEST_ATIME) ||
> +           (mnt->mnt_flags & MNT_NOATIME) ||
> +           ((mnt->mnt_flags & MNT_NODIRATIME) && S_ISDIR(inode->i_mode)))
>                need_atime = 0;
>
> -       if (need_atime)
> -               err = __nfs_revalidate_inode(NFS_SERVER(inode), inode);
> -       else
> -               err = nfs_revalidate_inode(NFS_SERVER(inode), inode);
> -       if (!err) {
> -               generic_fillattr(inode, stat);
> -               stat->ino = nfs_compat_user_ino64(NFS_FILEID(inode));
> +       if (force || stat->request_mask & (XSTAT_REQUEST__BASIC_STATS |
> +                                          XSTAT_REQUEST_DATA_VERSION)
> +           ) {
> +               if (force || need_atime)
> +                       err = __nfs_revalidate_inode(NFS_SERVER(inode), inode);
> +               else
> +                       err = nfs_revalidate_inode(NFS_SERVER(inode), inode);
> +               if (err)
> +                       goto out;
>        }
> +
> +       generic_fillattr(inode, stat);
> +       stat->ino = nfs_compat_user_ino64(NFS_FILEID(inode));
> +
> +       if (stat->request_mask & XSTAT_REQUEST_DATA_VERSION) {
> +               stat->data_version = NFS_I(inode)->change_attr;
> +               stat->result_mask |= XSTAT_REQUEST_DATA_VERSION;
> +       }
> +
>  out:
>        return err;
>  }
> @@ -770,7 +792,7 @@ int nfs_revalidate_inode(struct nfs_server *server, struct inode *inode)
>  static int nfs_invalidate_mapping(struct inode *inode, struct address_space *mapping)
>  {
>        struct nfs_inode *nfsi = NFS_I(inode);
> -
> +
>        if (mapping->nrpages != 0) {
>                int ret = invalidate_inode_pages2(mapping);
>                if (ret < 0)
> diff --git a/fs/nfsd/nfs3proc.c b/fs/nfsd/nfs3proc.c
> index 3d68f45..310ff05 100644
> --- a/fs/nfsd/nfs3proc.c
> +++ b/fs/nfsd/nfs3proc.c
> @@ -55,6 +55,8 @@ nfsd3_proc_getattr(struct svc_rqst *rqstp, struct nfsd_fhandle  *argp,
>        if (nfserr)
>                RETURN_STATUS(nfserr);
>
> +       resp->stat.query_flags = 0;
> +       resp->stat.request_mask = XSTAT_REQUEST__EXTENDED_STATS;
>        err = vfs_getattr(resp->fh.fh_export->ex_path.mnt,
>                          resp->fh.fh_dentry, &resp->stat);
>        nfserr = nfserrno(err);
> diff --git a/fs/nfsd/nfs3xdr.c b/fs/nfsd/nfs3xdr.c
> index 2a533a0..eaa3c3b 100644
> --- a/fs/nfsd/nfs3xdr.c
> +++ b/fs/nfsd/nfs3xdr.c
> @@ -205,6 +205,8 @@ encode_post_op_attr(struct svc_rqst *rqstp, __be32 *p, struct svc_fh *fhp)
>                int err;
>                struct kstat stat;
>
> +               stat.query_flags = 0;
> +               stat.request_mask = XSTAT_REQUEST__EXTENDED_STATS;
>                err = vfs_getattr(fhp->fh_export->ex_path.mnt, dentry, &stat);
>                if (!err) {
>                        *p++ = xdr_one;         /* attributes follow */
> @@ -257,6 +259,8 @@ void fill_post_wcc(struct svc_fh *fhp)
>        if (fhp->fh_post_saved)
>                printk("nfsd: inode locked twice during operation.\n");
>
> +       fhp->fh_post_attr.query_flags = 0;
> +       fhp->fh_post_attr.request_mask = XSTAT_REQUEST__EXTENDED_STATS;
>        err = vfs_getattr(fhp->fh_export->ex_path.mnt, fhp->fh_dentry,
>                        &fhp->fh_post_attr);
>        fhp->fh_post_change = fhp->fh_dentry->d_inode->i_version;
> diff --git a/fs/nfsd/nfs4xdr.c b/fs/nfsd/nfs4xdr.c
> index ac17a70..e9d1b59 100644
> --- a/fs/nfsd/nfs4xdr.c
> +++ b/fs/nfsd/nfs4xdr.c
> @@ -1769,6 +1769,8 @@ nfsd4_encode_fattr(struct svc_fh *fhp, struct svc_export *exp,
>                        goto out;
>        }
>
> +       stat.query_flags = 0;
> +       stat.request_mask = XSTAT_REQUEST__EXTENDED_STATS;
>        err = vfs_getattr(exp->ex_path.mnt, dentry, &stat);
>        if (err)
>                goto out_nfserr;
> @@ -2139,6 +2141,8 @@ out_acl:
>                                if (path.dentry != path.mnt->mnt_root)
>                                        break;
>                        }
> +                       stat.query_flags = 0;
> +                       stat.request_mask = XSTAT_REQUEST__EXTENDED_STATS;
>                        err = vfs_getattr(path.mnt, path.dentry, &stat);
>                        path_put(&path);
>                        if (err)
> diff --git a/fs/nfsd/nfsproc.c b/fs/nfsd/nfsproc.c
> index a047ad6..7c0e74b 100644
> --- a/fs/nfsd/nfsproc.c
> +++ b/fs/nfsd/nfsproc.c
> @@ -26,6 +26,8 @@ static __be32
>  nfsd_return_attrs(__be32 err, struct nfsd_attrstat *resp)
>  {
>        if (err) return err;
> +       resp->stat.query_flags = 0;
> +       resp->stat.request_mask = XSTAT_REQUEST__EXTENDED_STATS;
>        return nfserrno(vfs_getattr(resp->fh.fh_export->ex_path.mnt,
>                                    resp->fh.fh_dentry,
>                                    &resp->stat));
> @@ -34,6 +36,8 @@ static __be32
>  nfsd_return_dirop(__be32 err, struct nfsd_diropres *resp)
>  {
>        if (err) return err;
> +       resp->stat.query_flags = 0;
> +       resp->stat.request_mask = XSTAT_REQUEST__EXTENDED_STATS;
>        return nfserrno(vfs_getattr(resp->fh.fh_export->ex_path.mnt,
>                                    resp->fh.fh_dentry,
>                                    &resp->stat));
> @@ -150,6 +154,8 @@ nfsd_proc_read(struct svc_rqst *rqstp, struct nfsd_readargs *argp,
>                                  &resp->count);
>
>        if (nfserr) return nfserr;
> +       resp->stat.query_flags = 0;
> +       resp->stat.request_mask = XSTAT_REQUEST__EXTENDED_STATS;
>        return nfserrno(vfs_getattr(resp->fh.fh_export->ex_path.mnt,
>                                    resp->fh.fh_dentry,
>                                    &resp->stat));
> diff --git a/fs/nfsd/nfsxdr.c b/fs/nfsd/nfsxdr.c
> index 4ce005d..a595fb6 100644
> --- a/fs/nfsd/nfsxdr.c
> +++ b/fs/nfsd/nfsxdr.c
> @@ -197,6 +197,8 @@ encode_fattr(struct svc_rqst *rqstp, __be32 *p, struct svc_fh *fhp,
>  __be32 *nfs2svc_encode_fattr(struct svc_rqst *rqstp, __be32 *p, struct svc_fh *fhp)
>  {
>        struct kstat stat;
> +       stat.query_flags = 0;
> +       stat.request_mask = XSTAT_REQUEST__EXTENDED_STATS;
>        vfs_getattr(fhp->fh_export->ex_path.mnt, fhp->fh_dentry, &stat);
>        return encode_fattr(rqstp, p, fhp, &stat);
>  }
> diff --git a/fs/stat.c b/fs/stat.c
> index 12e90e2..2fb1527 100644
> --- a/fs/stat.c
> +++ b/fs/stat.c
> @@ -33,6 +33,9 @@ void generic_fillattr(struct inode *inode, struct kstat *stat)
>        stat->size = i_size_read(inode);
>        stat->blocks = inode->i_blocks;
>        stat->blksize = (1 << inode->i_blkbits);
> +       stat->result_mask |= XSTAT_REQUEST__BASIC_STATS & ~XSTAT_REQUEST_RDEV;
> +       if (unlikely(S_ISBLK(stat->mode) || S_ISCHR(stat->mode)))
> +               stat->result_mask |= XSTAT_REQUEST_RDEV;
>  }
>
>  EXPORT_SYMBOL(generic_fillattr);
> @@ -42,6 +45,8 @@ int vfs_getattr(struct vfsmount *mnt, struct dentry *dentry, struct kstat *stat)
>        struct inode *inode = dentry->d_inode;
>        int retval;
>
> +       stat->result_mask = 0;
> +
>        retval = security_inode_getattr(mnt, dentry);
>        if (retval)
>                return retval;
> @@ -55,41 +60,64 @@ int vfs_getattr(struct vfsmount *mnt, struct dentry *dentry, struct kstat *stat)
>
>  EXPORT_SYMBOL(vfs_getattr);
>
> -int vfs_fstat(unsigned int fd, struct kstat *stat)
> +/*
> + * VFS entrypoint to get extended stats by file descriptor
> + */
> +int vfs_fxstat(unsigned int fd, int flags, struct kstat *stat)
>  {
>        struct file *f = fget(fd);
>        int error = -EBADF;
>
> +       if (flags & ~KSTAT_QUERY_FLAGS)
> +               return -EINVAL;
> +       stat->query_flags = flags;
> +
>        if (f) {
>                error = vfs_getattr(f->f_path.mnt, f->f_path.dentry, stat);
>                fput(f);
>        }
>        return error;
>  }
> +EXPORT_SYMBOL(vfs_fxstat);
> +
> +int vfs_fstat(unsigned int fd, struct kstat *stat)
> +{
> +       stat->request_mask = XSTAT_REQUEST__BASIC_STATS;
> +       return vfs_fxstat(fd, 0, stat);
> +}
>  EXPORT_SYMBOL(vfs_fstat);
>
> -int vfs_fstatat(int dfd, const char __user *filename, struct kstat *stat,
> -               int flag)
> +/*
> + * VFS entrypoint to get extended stats by filename
> + */
> +int vfs_xstat(int dfd, const char __user *filename, int flags,
> +             struct kstat *stat)
>  {
>        struct path path;
> -       int error = -EINVAL;
> -       int lookup_flags = 0;
> +       int error, lookup_flags;
>
> -       if ((flag & ~AT_SYMLINK_NOFOLLOW) != 0)
> -               goto out;
> +       if (flags & ~(AT_SYMLINK_NOFOLLOW | KSTAT_QUERY_FLAGS))
> +               return -EINVAL;
>
> -       if (!(flag & AT_SYMLINK_NOFOLLOW))
> -               lookup_flags |= LOOKUP_FOLLOW;
> +       stat->query_flags = flags;
> +       lookup_flags = (flags & AT_SYMLINK_NOFOLLOW) ? 0 : LOOKUP_FOLLOW;
>
>        error = user_path_at(dfd, filename, lookup_flags, &path);
> -       if (error)
> -               goto out;
> -
> -       error = vfs_getattr(path.mnt, path.dentry, stat);
> -       path_put(&path);
> -out:
> +       if (!error) {
> +               error = vfs_getattr(path.mnt, path.dentry, stat);
> +               path_put(&path);
> +       }
>        return error;
>  }
> +EXPORT_SYMBOL(vfs_xstat);
> +
> +int vfs_fstatat(int dfd, const char __user *filename, struct kstat *stat,
> +               int flags)
> +{
> +       stat->request_mask = XSTAT_REQUEST__BASIC_STATS;
> +       stat->query_flags = 0;
> +       return vfs_xstat(dfd, filename, flags, stat);
> +}
>  EXPORT_SYMBOL(vfs_fstatat);
>
>  int vfs_stat(const char __user *name, struct kstat *stat)
> @@ -115,7 +143,7 @@ static int cp_old_stat(struct kstat *stat, struct __old_kernel_stat __user * sta
>  {
>        static int warncount = 5;
>        struct __old_kernel_stat tmp;
> -
> +
>        if (warncount > 0) {
>                warncount--;
>                printk(KERN_WARNING "VFS: Warning: %s using old stat() call. Recompile your binary.\n",
> @@ -140,7 +168,7 @@ static int cp_old_stat(struct kstat *stat, struct __old_kernel_stat __user * sta
>  #if BITS_PER_LONG == 32
>        if (stat->size > MAX_NON_LFS)
>                return -EOVERFLOW;
> -#endif
> +#endif
>        tmp.st_size = stat->size;
>        tmp.st_atime = stat->atime.tv_sec;
>        tmp.st_mtime = stat->mtime.tv_sec;
> @@ -222,7 +250,7 @@ static int cp_new_stat(struct kstat *stat, struct stat __user *statbuf)
>  #if BITS_PER_LONG == 32
>        if (stat->size > MAX_NON_LFS)
>                return -EOVERFLOW;
> -#endif
> +#endif
>        tmp.st_size = stat->size;
>        tmp.st_atime = stat->atime.tv_sec;
>        tmp.st_mtime = stat->mtime.tv_sec;
> @@ -408,6 +436,117 @@ SYSCALL_DEFINE4(fstatat64, int, dfd, const char __user *, filename,
>  }
>  #endif /* __ARCH_WANT_STAT64 */
>
> +/*
> + * Get the xstat parameters if supplied
> + */
> +static int xstat_get_params(struct xstat_parameters __user *_params,
> +                           struct kstat *stat)
> +{
> +       struct xstat_parameters params;
> +
> +       memset(stat, 0xde, sizeof(*stat));      // DEBUGGING
> +
> +       if (_params) {
> +               if (copy_from_user(&params, _params, sizeof(params)) != 0)
> +                       return -EFAULT;
> +               stat->request_mask =
> +                       params.request_mask & XSTAT_REQUEST__ALL_STATS;
> +       } else {
> +               stat->request_mask = XSTAT_REQUEST__EXTENDED_STATS;
> +       }
> +       stat->result_mask = 0;
> +       return 0;
> +}
> +
> +/*
> + * copy the extended stats to userspace and return the amount of data written
> + * into the buffer
> + */
> +static long xstat_set_result(struct kstat *stat,
> +                            struct xstat __user *buffer, size_t bufsize)
> +{
> +       struct xstat tmp;
> +       size_t copy;
> +
> +       /* transfer the fixed results */
> +       memset(&tmp, 0, sizeof(tmp));
> +       tmp.st_result_mask      = stat->result_mask;
> +       tmp.st_mode             = stat->mode;
> +       tmp.st_nlink            = stat->nlink;
> +       tmp.st_uid              = stat->uid;
> +       tmp.st_gid              = stat->gid;
> +       tmp.st_blksize          = stat->blksize;
> +       tmp.st_rdev.major       = MAJOR(stat->rdev);
> +       tmp.st_rdev.minor       = MINOR(stat->rdev);
> +       tmp.st_dev.major        = MAJOR(stat->dev);
> +       tmp.st_dev.minor        = MINOR(stat->dev);
> +       tmp.st_atime.tv_sec     = stat->atime.tv_sec;
> +       tmp.st_atime.tv_nsec    = stat->atime.tv_nsec;
> +       tmp.st_mtime.tv_sec     = stat->mtime.tv_sec;
> +       tmp.st_mtime.tv_nsec    = stat->mtime.tv_nsec;
> +       tmp.st_ctime.tv_sec     = stat->ctime.tv_sec;
> +       tmp.st_ctime.tv_nsec    = stat->ctime.tv_nsec;
> +       tmp.st_ino              = stat->ino;
> +       tmp.st_size             = stat->size;
> +       tmp.st_blocks           = stat->blocks;
> +
> +       if (tmp.st_result_mask & XSTAT_REQUEST_BTIME) {
> +               tmp.st_btime.tv_sec     = stat->btime.tv_sec;
> +               tmp.st_btime.tv_nsec    = stat->btime.tv_nsec;
> +       }
> +       if (tmp.st_result_mask & XSTAT_REQUEST_GEN)
> +               tmp.st_gen              = stat->gen;
> +       if (tmp.st_result_mask & XSTAT_REQUEST_DATA_VERSION)
> +               tmp.st_data_version     = stat->data_version;
> +
> +       copy = sizeof(tmp);
> +       if (copy > bufsize)
> +               copy = bufsize;
> +       if (copy_to_user(buffer, &tmp, copy) != 0)
> +               return -EFAULT;
> +       return sizeof(tmp);
> +}
> +
> +/*
> + * System call to get extended stats by path
> + */
> +SYSCALL_DEFINE6(xstat,
> +               int, dfd, const char __user *, filename, unsigned, atflag,
> +               struct xstat_parameters __user *, params,
> +               struct xstat __user *, buffer, size_t, bufsize)
> +{
> +       struct kstat stat;
> +       int error;
> +
> +       error = xstat_get_params(params, &stat);
> +       if (error != 0)
> +               return error;
> +       error = vfs_xstat(dfd, filename, atflag, &stat);
> +       if (error)
> +               return error;
> +       return xstat_set_result(&stat, buffer, bufsize);
> +}
> +
> +/*
> + * System call to get extended stats by file descriptor
> + */
> +SYSCALL_DEFINE5(fxstat, unsigned int, fd, unsigned int, flags,
> +               struct xstat_parameters __user *, params,
> +               struct xstat __user *, buffer, size_t, bufsize)
> +{
> +       struct kstat stat;
> +       int error;
> +
> +       error = xstat_get_params(params, &stat);
> +       if (error < 0)
> +               return error;
> +       error = vfs_fxstat(fd, flags, &stat);
> +       if (error)
> +               return error;
> +
> +       return xstat_set_result(&stat, buffer, bufsize);
> +}
> +
>  /* Caller is here responsible for sufficient locking (ie. inode->i_lock) */
>  void __inode_add_bytes(struct inode *inode, loff_t bytes)
>  {
> diff --git a/include/linux/fcntl.h b/include/linux/fcntl.h
> index afc00af..bcf8083 100644
> --- a/include/linux/fcntl.h
> +++ b/include/linux/fcntl.h
> @@ -45,6 +45,7 @@
>  #define AT_REMOVEDIR           0x200   /* Remove directory instead of
>                                            unlinking file.  */
>  #define AT_SYMLINK_FOLLOW      0x400   /* Follow symbolic links.  */
> +#define AT_FORCE_ATTR_SYNC     0x800   /* Force the attributes to be sync'd with the server */
>
>  #ifdef __KERNEL__
>
> diff --git a/include/linux/fs.h b/include/linux/fs.h
> index a18bcea..9ce2119 100644
> --- a/include/linux/fs.h
> +++ b/include/linux/fs.h
> @@ -2343,6 +2343,8 @@ extern int vfs_stat(const char __user *, struct kstat *);
>  extern int vfs_lstat(const char __user *, struct kstat *);
>  extern int vfs_fstat(unsigned int, struct kstat *);
>  extern int vfs_fstatat(int , const char __user *, struct kstat *, int);
> +extern int vfs_xstat(int, const char __user *, int, struct kstat *);
> +extern int vfs_xfstat(unsigned int, struct kstat *);
>
>  extern int do_vfs_ioctl(struct file *filp, unsigned int fd, unsigned int cmd,
>                    unsigned long arg);
> diff --git a/include/linux/stat.h b/include/linux/stat.h
> index 611c398..e0b89e4 100644
> --- a/include/linux/stat.h
> +++ b/include/linux/stat.h
> @@ -46,6 +46,99 @@
>
>  #endif
>
> +/*
> + * Extended stat structures
> + */
> +struct xstat_parameters {
> +       /* Query request/result mask
> +        *
> +        * Bits should be set in request_mask to request particular items
> +        * before calling xstat() or fxstat().
> +        *
> +        * For each item in the set XSTAT_REQUEST__EXTENDED_STATS:
> +        *
> +        * - if not available at all, the bit will be cleared before returning
> +        *   and the field will be cleared; otherwise,
> +        *
> +        * - if AT_FORCE_ATTR_SYNC is set, then the datum will be synchronised
> +        *   to the server and the bit will be set on return; otherwise,
> +        *
> +        * - if requested, the datum will be synchronised to a server or other
> +        *   hardware if out of date before being returned, and the bit will be
> +        *   set on return; otherwise,
> +        *
> +        * - if not requested, but available in approximate form without any
> +        *   effort, it will be filled in anyway, and the bit will be set upon
> +        *   return (it might not be up to date, however, and no attempt will
> +        *   be made to synchronise the internal state first); otherwise,
> +        *
> +        * - the bit will be cleared before returning, and the field will be
> +         *   cleared.
> +        *
> +        * For each item not in the set XSTAT_REQUEST__EXTENDED_STATS
> +        *
> +        * - if not available at all, the bit will be cleared, and no result
> +         *   data will be returned; otherwise,
> +        *
> +        * - if requested, the datum will be synchronised to a server or other
> +        *   hardware before being appended if necessary, and the bit will be
> +        *   set on return; otherwise,
> +        *
> +        * - the bit will be cleared, and no result data will be returned.
> +        *
> +        * Items in XSTAT_REQUEST__BASIC_STATS may be marked unavailable on
> +        * return, but they will have a value installed for compatibility
> +        * purposes.
> +        */
> +       unsigned long long      request_mask;
> +#define XSTAT_REQUEST_MODE             0x00000001ULL   /* want/got st_mode */
> +#define XSTAT_REQUEST_NLINK            0x00000002ULL   /* want/got st_nlink */
> +#define XSTAT_REQUEST_UID              0x00000004ULL   /* want/got st_uid */
> +#define XSTAT_REQUEST_GID              0x00000008ULL   /* want/got st_gid */
> +#define XSTAT_REQUEST_RDEV             0x00000010ULL   /* want/got st_rdev */
> +#define XSTAT_REQUEST_ATIME            0x00000020ULL   /* want/got st_atime */
> +#define XSTAT_REQUEST_MTIME            0x00000040ULL   /* want/got st_mtime */
> +#define XSTAT_REQUEST_CTIME            0x00000080ULL   /* want/got st_ctime */
> +#define XSTAT_REQUEST_INO              0x00000100ULL   /* want/got st_ino */
> +#define XSTAT_REQUEST_SIZE             0x00000200ULL   /* want/got st_size */
> +#define XSTAT_REQUEST_BLOCKS           0x00000400ULL   /* want/got st_blocks */
> +#define XSTAT_REQUEST__BASIC_STATS     0x000007ffULL   /* the stuff in the normal stat struct */
> +#define XSTAT_REQUEST_BTIME            0x00000800ULL   /* want/got st_btime */
> +#define XSTAT_REQUEST_GEN              0x00001000ULL   /* want/got st_gen */
> +#define XSTAT_REQUEST_DATA_VERSION     0x00002000ULL   /* want/got st_data_version */
> +#define XSTAT_REQUEST__EXTENDED_STATS  0x00003fffULL   /* the stuff in the xstat struct */
> +#define XSTAT_REQUEST__ALL_STATS       0x00003fffULL   /* the defined set of requestables */
> +};
> +
> +struct xstat_dev {
> +       unsigned int            major, minor;
> +};
> +
> +struct xstat_time {
> +       unsigned long long      tv_sec, tv_nsec;
> +};
> +
> +struct xstat {
> +       unsigned int            st_mode;        /* file mode */
> +       unsigned int            st_nlink;       /* number of hard links */
> +       unsigned int            st_uid;         /* user ID of owner */
> +       unsigned int            st_gid;         /* group ID of owner */
> +       struct xstat_dev        st_rdev;        /* device ID of special file */
> +       struct xstat_dev        st_dev;         /* ID of device containing file */
> +       struct xstat_time       st_atime;       /* last access time */
> +       struct xstat_time       st_mtime;       /* last data modification time */
> +       struct xstat_time       st_ctime;       /* last attribute change time */
> +       struct xstat_time       st_btime;       /* file creation time */
> +       unsigned long long      st_ino;         /* inode number */
> +       unsigned long long      st_size;        /* file size */
> +       unsigned long long      st_blksize;     /* block size for filesystem I/O */
> +       unsigned long long      st_blocks;      /* number of 512-byte blocks allocated */
> +       unsigned long long      st_gen;         /* inode generation number */
> +       unsigned long long      st_data_version; /* data version number */
> +       unsigned long long      st_result_mask; /* what requests were written */
> +       unsigned long long      st_extra_results[0]; /* extra requested results */
> +};
> +
>  #ifdef __KERNEL__
>  #define S_IRWXUGO      (S_IRWXU|S_IRWXG|S_IRWXO)
>  #define S_IALLUGO      (S_ISUID|S_ISGID|S_ISVTX|S_IRWXUGO)
> @@ -67,14 +160,20 @@ struct kstat {
>        uid_t           uid;
>        gid_t           gid;
>        dev_t           rdev;
> +       unsigned int    query_flags;            /* operational flags */
> +#define KSTAT_QUERY_FLAGS (AT_FORCE_ATTR_SYNC)
>        loff_t          size;
> -       struct timespec  atime;
> +       struct timespec atime;
>        struct timespec mtime;
>        struct timespec ctime;
> +       struct timespec btime;                  /* file creation time */
>        unsigned long   blksize;
>        unsigned long long      blocks;
> +       u64             request_mask;           /* what fields the user asked for */
> +       u64             result_mask;            /* what fields the user got */
> +       u64             gen;                    /* inode generation */
> +       u64             data_version;
>  };
>
>  #endif
> -
>  #endif
> diff --git a/include/linux/syscalls.h b/include/linux/syscalls.h
> index 8812a63..5d68b4c 100644
> --- a/include/linux/syscalls.h
> +++ b/include/linux/syscalls.h
> @@ -44,6 +44,8 @@ struct shmid_ds;
>  struct sockaddr;
>  struct stat;
>  struct stat64;
> +struct xstat_parameters;
> +struct xstat;
>  struct statfs;
>  struct statfs64;
>  struct __sysctl_args;
> @@ -824,4 +826,11 @@ asmlinkage long sys_mmap_pgoff(unsigned long addr, unsigned long len,
>                        unsigned long fd, unsigned long pgoff);
>  asmlinkage long sys_old_mmap(struct mmap_arg_struct __user *arg);
>
> +asmlinkage long sys_xstat(int, const char __user *, unsigned,
> +                         struct xstat_parameters __user *,
> +                         struct xstat __user *, size_t);
> +asmlinkage long sys_fxstat(unsigned, unsigned,
> +                          struct xstat_parameters __user *,
> +                          struct xstat __user *, size_t);
> +
>  #endif
>
> --
> To unsubscribe from this list: send the line "unsubscribe linux-fsdevel" in
> the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
>



-- 
Michael Kerrisk Linux man-pages maintainer;
http://www.kernel.org/doc/man-pages/
Author of "The Linux Programming Interface", http://blog.man7.org/

Re: [PATCH] Add a pair of system calls to make extended file stats available [ver #3]

[Andreas Dilger]

On 2010-07-01, at 23:36, Michael Kerrisk wrote:
> * Include information from the "inode_info" structure, most notably
> i_flags, but perhaps other info as well.

This one is actually pretty interesting, though instead of exporting the i_flags directly (the S_* flags), it would be much better to export the FS_*_FL values.  The FS_*_FL values (e.g. FS_IMMUTABLE_FL) are already exposed to userspace via FS_IOC_{GET,SET}FLAGS and are stored on disk in ext2/3/4, so are guaranteed never to change.  The S_* flags DO in fact change between kernel releases.


Cheers, Andreas

Re: [PATCH] Add a pair of system calls to make extended file stats available [ver #3]

[Michael Kerrisk]

hi David,

On Fri, Jul 2, 2010 at 7:36 AM, Michael Kerrisk <mtk.manpages@gmail.com> wrote:
> Hi David,
>
> [Please CC linux-api@ on patches that change the API/ABI]
>
> On Thu, Jul 1, 2010 at 1:36 AM, David Howells <dhowells@redhat.com> wrote:
>> Add a pair of system calls to make extended file stats available, including
>> file creation time, inode version and data version where available through the
>> underlying filesystem.
>
> Just some random thoughts here. I've not tried to guess the overhead
> of these ideas...
>
> * Include information from the "inode_info" structure, most notably
> i_flags, but perhaps other info as well.

I see you put a patch for the above for comment. Thanks.

> * Return a bit mask indicating the presence of additional information
> associated with the i-node. Here, I am thinking of flags that indicate
> that the file has any of the following: capabilities, an ACL, and
> extended attributes (obviously a superset of the previous). I could
> imagine some apps that, having got the xstat info, would be interested
> to obtain some of this other info.

What did you think about the above idea?

Cheers,

Michael


> Obviously, the above only make sense if the overhead of providing the
> extra information is low.
>
>> [This depends on the previously posted pair of patches to (a) constify a number
>>  of syscall string and buffer arguments and (b) rearrange AFS's use of
>>  i_version and i_generation].
>>
>> The following structures are defined for their use:
>>
>>        struct xstat_parameters {
>>                unsigned long long      request_mask;
>>        };
>>
>>        struct xstat_dev {
>>                unsigned int            major, minor;
>>        };
>>
>>        struct xstat_time {
>>                unsigned long long      tv_sec, tv_nsec;
>>        };
>>
>>        struct xstat {
>>                unsigned int            st_mode;
>>                unsigned int            st_nlink;
>>                unsigned int            st_uid;
>>                unsigned int            st_gid;
>>                struct xstat_dev        st_rdev;
>>                struct xstat_dev        st_dev;
>>                struct xstat_time       st_atime;
>>                struct xstat_time       st_mtime;
>>                struct xstat_time       st_ctime;
>>                struct xstat_time       st_btime;
>>                unsigned long long      st_ino;
>>                unsigned long long      st_size;
>>                unsigned long long      st_blksize;
>>                unsigned long long      st_blocks;
>>                unsigned long long      st_gen;
>>                unsigned long long      st_data_version;
>>                unsigned long long      st_result_mask;
>>                unsigned long long      st_extra_results[0];
>>        };
>>
>> where st_btime is the file creation time, st_gen is the inode generation
>> (i_generation), st_data_version is the data version number (i_version),
>> request_mask and st_result_mask are bitmasks of data desired/provided and
>> st_extra_results[] is where as-yet undefined fields are appended.
>>
>> The defined bits in request_mask and st_result_mask are:
>>
>>        XSTAT_REQUEST_MODE              Want/got st_mode
>>        XSTAT_REQUEST_NLINK             Want/got st_nlink
>>        XSTAT_REQUEST_UID               Want/got st_uid
>>        XSTAT_REQUEST_GID               Want/got st_gid
>>        XSTAT_REQUEST_RDEV              Want/got st_rdev
>>        XSTAT_REQUEST_ATIME             Want/got st_atime
>>        XSTAT_REQUEST_MTIME             Want/got st_mtime
>>        XSTAT_REQUEST_CTIME             Want/got st_ctime
>>        XSTAT_REQUEST_INO               Want/got st_ino
>>        XSTAT_REQUEST_SIZE              Want/got st_size
>>        XSTAT_REQUEST_BLOCKS            Want/got st_blocks
>>        XSTAT_REQUEST__BASIC_STATS      The stuff in the normal stat struct
>>        XSTAT_REQUEST_BTIME             Want/got st_btime
>>        XSTAT_REQUEST_GEN               Want/got st_gen
>>        XSTAT_REQUEST_DATA_VERSION      Want/got st_data_version
>>        XSTAT_REQUEST__EXTENDED_STATS   The stuff in the xstat struct
>>        XSTAT_REQUEST__ALL_STATS        The defined set of requestables
>>
>> The system calls are:
>>
>>        ssize_t ret = xstat(int dfd,
>>                            const char *filename,
>>                            unsigned flags,
>>                            const struct xstat_parameters *params,
>>                            struct xstat *buffer,
>>                            size_t buflen);
>>
>>        ssize_t ret = fxstat(unsigned fd,
>>                             unsigned flags,
>>                             const struct xstat_parameters *params,
>>                             struct xstat *buffer,
>>                             size_t buflen);
>>
>>
>> The dfd, filename, flags and fd parameters indicate the file to query.  There
>> is no equivalent of lstat() as that can be emulated with xstat() by passing
>> AT_SYMLINK_NOFOLLOW in flags.
>>
>> AT_FORCE_ATTR_SYNC can also be set in flags.  This will require a network
>> filesystem to synchronise its attributes with the server.
>>
>> When the system call is executed, the request_mask bitmask is read from the
>> parameter block to work out what the user is requesting.  If params is NULL,
>> then request_mask will be assumed to be XSTAT_REQUEST__GET_ANYWAY.
>>
>> The request_mask should be set by the caller to specify extra results that the
>> caller may desire.  These come in a number of classes:
>>
>>  (0) dev, blksize.
>>
>>     These are local data and are always available.
>>
>>  (1) mode, nlinks, uid, gid, [amc]time, ino, size, blocks.
>>
>>     These will be returned whether the caller asks for them or not.  The
>>     corresponding bits in result_mask will be set to indicate their presence.
>>
>>     If the caller didn't ask for them, then they may be approximated.  For
>>     example, NFS won't waste any time updating them from the server, unless as
>>     a byproduct of updating something requested.
>>
>>  (2) rdev.
>>
>>     As for class (1), but this won't be returned if the file is not a blockdev
>>     or chardev.  The bit will be cleared if the value is not returned.
>>
>>  (3) File creation time, inode generation and data version.
>>
>>     These will be returned if available whether the caller asked for them or
>>     not.  The corresponding bits in result_mask will be set or cleared as
>>     appropriate to indicate their presence.
>>
>>     If the caller didn't ask for them, then they may be approximated.  For
>>     example, NFS won't waste any time updating them from the server, unless
>>     as a byproduct of updating something requested.
>>
>>  (4) Extra results.
>>
>>     These will only be returned if the caller asked for them by setting their
>>     bits in request_mask.  They will be placed in the buffer after the xstat
>>     struct in ascending result_mask bit order.  Any bit set in request_mask
>>     mask will be left set in result_mask if the result is available and
>>     cleared otherwise.
>>
>>     The pointer into the results list will be rounded up to the nearest 8-byte
>>     boundary after each result is written in.  The size of each extra result
>>     is specific to the definition for that result.
>>
>>     No extra results are currently defined.
>>
>> If the buffer is insufficiently big, the syscall returns the amount of space it
>> will need to write the complete result set and returns a partial result in the
>> buffer.
>>
>> At the moment, this will only work on x86_64 as it requires system calls to be
>> wired up.
>>
>>
>> ===========
>> FILESYSTEMS
>> ===========
>>
>> The following filesystems have been modified to make use of this facility:
>>
>>  (*) Ext4.  This will return the creation time and inode version number for all
>>     files.  It will, however, only return the data version number for
>>     directories unless the I_VERSION option is set on the filesystem.
>>
>>  (*) AFS.  This will return the vnode ID uniquifier as the inode version and
>>     the AFS data version number as the data version.  There is no file
>>     creation time available.
>>
>>     AFS should go to the server if AT_FORCE_ATTR_SYNC is specified.
>>
>>  (*) NFS.  This will return the change attribute if NFSv4 only.  No other extra
>>     values are returned at this time.
>>
>>     If AT_FORCE_ATTR_SYNC is set or mtime, ctime or data_version (NFSv4 only)
>>     are asked for then the outstanding writes will be written to the server
>>     first.
>>
>>     If AT_FORCE_ATTR_SYNC is set or atime is requested then the attributes
>>     will be reread unconditionally, otherwise if any of data version (NFSv4
>>     only) XSTAT_REQUEST__BASIC_STATS are requested, then the attributes will
>>     be reread if the cached attributes have expired.
>>
>>
>> =======
>> TESTING
>> =======
>>
>> The following test program can be used to test the xstat system call:
>>
>>        #define _GNU_SOURCE
>>        #define _ATFILE_SOURCE
>>        #include <stdio.h>
>>        #include <stdlib.h>
>>        #include <string.h>
>>        #include <unistd.h>
>>        #include <fcntl.h>
>>        #include <time.h>
>>        #include <sys/syscall.h>
>>        #include <sys/stat.h>
>>        #include <sys/types.h>
>>
>>        #define AT_FORCE_ATTR_SYNC      0x800
>>
>>        struct xstat_parameters {
>>                unsigned long long      request_mask;
>>        #define XSTAT_REQUEST_MODE              0x00000001ULL
>>        #define XSTAT_REQUEST_NLINK             0x00000002ULL
>>        #define XSTAT_REQUEST_UID               0x00000004ULL
>>        #define XSTAT_REQUEST_GID               0x00000008ULL
>>        #define XSTAT_REQUEST_RDEV              0x00000010ULL
>>        #define XSTAT_REQUEST_ATIME             0x00000020ULL
>>        #define XSTAT_REQUEST_MTIME             0x00000040ULL
>>        #define XSTAT_REQUEST_CTIME             0x00000080ULL
>>        #define XSTAT_REQUEST_INO               0x00000100ULL
>>        #define XSTAT_REQUEST_SIZE              0x00000200ULL
>>        #define XSTAT_REQUEST_BLOCKS            0x00000400ULL
>>        #define XSTAT_REQUEST__BASIC_STATS      0x000007ffULL
>>        #define XSTAT_REQUEST_BTIME             0x00000800ULL
>>        #define XSTAT_REQUEST_GEN               0x00001000ULL
>>        #define XSTAT_REQUEST_DATA_VERSION      0x00002000ULL
>>        #define XSTAT_REQUEST__EXTENDED_STATS   0x00003fffULL
>>        #define XSTAT_REQUEST__ALL_STATS        0x00003fffULL
>>        };
>>
>>        struct xstat_dev {
>>                unsigned int    major;
>>                unsigned int    minor;
>>        };
>>
>>        struct xstat_time {
>>                unsigned long long      tv_sec;
>>                unsigned long long      tv_nsec;
>>        };
>>
>>        struct xstat {
>>                unsigned int            st_mode;
>>                unsigned int            st_nlink;
>>                unsigned int            st_uid;
>>                unsigned int            st_gid;
>>                struct xstat_dev        st_rdev;
>>                struct xstat_dev        st_dev;
>>                struct xstat_time       st_atim;
>>                struct xstat_time       st_mtim;
>>                struct xstat_time       st_ctim;
>>                struct xstat_time       st_btim;
>>                unsigned long long      st_ino;
>>                unsigned long long      st_size;
>>                unsigned long long      st_blksize;
>>                unsigned long long      st_blocks;
>>                unsigned long long      st_gen;
>>                unsigned long long      st_data_version;
>>                unsigned long long      st_result_mask;
>>                unsigned long long      st_extra_results[0];
>>        };
>>
>>        #define __NR_xstat                              300
>>        #define __NR_fxstat                             301
>>
>>        static __attribute__((unused))
>>        ssize_t xstat(int dfd, const char *filename, unsigned flags,
>>                      struct xstat_parameters *params,
>>                      struct xstat *buffer, size_t bufsize)
>>        {
>>                return syscall(__NR_xstat, dfd, filename, flags,
>>                               params, buffer, bufsize);
>>        }
>>
>>        static __attribute__((unused))
>>        ssize_t fxstat(int fd, unsigned flags,
>>                       struct xstat_parameters *params,
>>                       struct xstat *buffer, size_t bufsize)
>>        {
>>                return syscall(__NR_fxstat, fd, flags,
>>                               params, buffer, bufsize);
>>        }
>>
>>        static void print_time(const char *field, const struct xstat_time *xstm)
>>        {
>>                struct tm tm;
>>                time_t tim;
>>                char buffer[100];
>>                int len;
>>
>>                tim = xstm->tv_sec;
>>                if (!localtime_r(&tim, &tm)) {
>>                        perror("localtime_r");
>>                        exit(1);
>>                }
>>                len = strftime(buffer, 100, "%F %T", &tm);
>>                if (len == 0) {
>>                        perror("strftime");
>>                        exit(1);
>>                }
>>                printf("%s", field);
>>                fwrite(buffer, 1, len, stdout);
>>                printf(".%09llu", xstm->tv_nsec);
>>                len = strftime(buffer, 100, "%z", &tm);
>>                if (len == 0) {
>>                        perror("strftime2");
>>                        exit(1);
>>                }
>>                fwrite(buffer, 1, len, stdout);
>>                printf("\n");
>>        }
>>
>>        static void dump_xstat(struct xstat *xst)
>>        {
>>                char buffer[256], ft;
>>
>>                printf("results=%llx\n", xst->st_result_mask);
>>
>>                printf(" ");
>>                if (xst->st_result_mask & XSTAT_REQUEST_SIZE)
>>                        printf(" Size: %-15llu", xst->st_size);
>>                if (xst->st_result_mask & XSTAT_REQUEST_BLOCKS)
>>                        printf(" Blocks: %-10llu", xst->st_blocks);
>>                printf(" IO Block: %-6llu ", xst->st_blksize);
>>                if (xst->st_result_mask & XSTAT_REQUEST_MODE) {
>>                        switch (xst->st_mode & S_IFMT) {
>>                        case S_IFIFO:   printf(" FIFO\n");                      ft = 'p'; break;
>>                        case S_IFCHR:   printf(" character special file\n");    ft = 'c'; break;
>>                        case S_IFDIR:   printf(" directory\n");                 ft = 'd'; break;
>>                        case S_IFBLK:   printf(" block special file\n");        ft = 'b'; break;
>>                        case S_IFREG:   printf(" regular file\n");              ft = '-'; break;
>>                        case S_IFLNK:   printf(" symbolic link\n");             ft = 'l'; break;
>>                        case S_IFSOCK:  printf(" socket\n");                    ft = 's'; break;
>>                        default:
>>                                printf("unknown type (%o)\n", xst->st_mode & S_IFMT);
>>                                ft = '?';
>>                                break;
>>                        }
>>                }
>>
>>                sprintf(buffer, "%02x:%02x", xst->st_dev.major, xst->st_dev.minor);
>>                printf("Device: %-15s", buffer);
>>                if (xst->st_result_mask & XSTAT_REQUEST_INO)
>>                        printf(" Inode: %-11llu", xst->st_ino);
>>                if (xst->st_result_mask & XSTAT_REQUEST_SIZE)
>>                        printf(" Links: %-5u", xst->st_nlink);
>>                if (xst->st_result_mask & XSTAT_REQUEST_RDEV)
>>                        printf(" Device type: %u,%u",
>>                               xst->st_rdev.major, xst->st_rdev.minor);
>>                printf("\n");
>>
>>                if (xst->st_result_mask & XSTAT_REQUEST_MODE)
>>                        printf("Access: (%04o/%c%c%c%c%c%c%c%c%c%c)  ",
>>                               xst->st_mode & 07777,
>>                               ft,
>>                               xst->st_mode & S_IRUSR ? 'r' : '-',
>>                               xst->st_mode & S_IWUSR ? 'w' : '-',
>>                               xst->st_mode & S_IXUSR ? 'x' : '-',
>>                               xst->st_mode & S_IRGRP ? 'r' : '-',
>>                               xst->st_mode & S_IWGRP ? 'w' : '-',
>>                               xst->st_mode & S_IXGRP ? 'x' : '-',
>>                               xst->st_mode & S_IROTH ? 'r' : '-',
>>                               xst->st_mode & S_IWOTH ? 'w' : '-',
>>                               xst->st_mode & S_IXOTH ? 'x' : '-');
>>                if (xst->st_result_mask & XSTAT_REQUEST_UID)
>>                        printf("Uid: %d   \n", xst->st_uid);
>>                if (xst->st_result_mask & XSTAT_REQUEST_GID)
>>                        printf("Gid: %u\n", xst->st_gid);
>>
>>                if (xst->st_result_mask & XSTAT_REQUEST_ATIME)
>>                        print_time("Access: ", &xst->st_atim);
>>                if (xst->st_result_mask & XSTAT_REQUEST_MTIME)
>>                        print_time("Modify: ", &xst->st_mtim);
>>                if (xst->st_result_mask & XSTAT_REQUEST_CTIME)
>>                        print_time("Change: ", &xst->st_ctim);
>>                if (xst->st_result_mask & XSTAT_REQUEST_BTIME)
>>                        print_time("Create: ", &xst->st_btim);
>>
>>                if (xst->st_result_mask & XSTAT_REQUEST_GEN)
>>                        printf("Inode version: %llxh\n", xst->st_gen);
>>                if (xst->st_result_mask & XSTAT_REQUEST_DATA_VERSION)
>>                        printf("Data version: %llxh\n", xst->st_data_version);
>>        }
>>
>>        int main(int argc, char **argv)
>>        {
>>                struct xstat_parameters params;
>>                struct xstat xst;
>>                int ret, atflag = AT_SYMLINK_NOFOLLOW;
>>
>>                unsigned long long query =
>>                        XSTAT_REQUEST__BASIC_STATS |
>>                        XSTAT_REQUEST_BTIME |
>>                        XSTAT_REQUEST_GEN |
>>                        XSTAT_REQUEST_DATA_VERSION;
>>
>>                for (argv++; *argv; argv++) {
>>                        if (strcmp(*argv, "-F") == 0) {
>>                                atflag |= AT_FORCE_ATTR_SYNC;
>>                                continue;
>>                        }
>>                        if (strcmp(*argv, "-L") == 0) {
>>                                atflag &= ~AT_SYMLINK_NOFOLLOW;
>>                                continue;
>>                        }
>>                        if (strcmp(*argv, "-O") == 0) {
>>                                query &= ~XSTAT_REQUEST__BASIC_STATS;
>>                                continue;
>>                        }
>>
>>                        memset(&xst, 0xbf, sizeof(xst));
>>                        params.request_mask = query;
>>                        ret = xstat(AT_FDCWD, *argv, atflag, &params, &xst, sizeof(xst));
>>                        printf("xstat(%s) = %d\n", *argv, ret);
>>                        if (ret < 0) {
>>                                perror(*argv);
>>                                exit(1);
>>                        }
>>
>>                        dump_xstat(&xst);
>>                }
>>                return 0;
>>        }
>>
>> Just compile and run, passing it paths to the files you want to examine:
>>
>>        [root@andromeda ~]# /tmp/xstat -O /dev/tty
>>        xstat(/dev/tty) = 152
>>        results=7ff
>>          Size: 0               Blocks: 0          IO Block: 4096    character special file
>>        Device: 00:0f           Inode: 246         Links: 1     Device type: 5,0
>>        Access: (0666/crw-rw-rw-)  Uid: 0
>>        Gid: 5
>>        Access: 2010-06-30 16:25:01.813517001+0100
>>        Modify: 2010-06-30 16:25:01.813517001+0100
>>        Change: 2010-06-30 16:25:01.813517001+0100
>>
>>        [root@andromeda ~]# /tmp/xstat /var/cache/fscache/cache/
>>        xstat(/var/cache/fscache/cache/) = 152
>>        results=3fef
>>          Size: 4096            Blocks: 16         IO Block: 4096    directory
>>        Device: 08:06           Inode: 130561      Links: 3
>>        Access: (0700/drwx------)  Uid: 0
>>        Gid: 0
>>        Access: 2010-06-29 18:16:33.680703545+0100
>>        Modify: 2010-06-29 18:16:20.132786632+0100
>>        Change: 2010-06-29 18:16:20.132786632+0100
>>        Create: 2010-06-25 15:17:39.471199293+0100
>>        Inode version: f585ab70h
>>        Data version: 2h
>>
>> Signed-off-by: David Howells <dhowells@redhat.com>
>> ---
>>
>>  arch/x86/include/asm/unistd_32.h |    4 +
>>  arch/x86/include/asm/unistd_64.h |    4 +
>>  fs/afs/inode.c                   |   11 +-
>>  fs/ecryptfs/inode.c              |    2
>>  fs/ext4/ext4.h                   |    2
>>  fs/ext4/file.c                   |    2
>>  fs/ext4/inode.c                  |   27 +++++-
>>  fs/ext4/namei.c                  |    2
>>  fs/ext4/symlink.c                |    2
>>  fs/nfs/inode.c                   |   46 +++++++---
>>  fs/nfsd/nfs3proc.c               |    2
>>  fs/nfsd/nfs3xdr.c                |    4 +
>>  fs/nfsd/nfs4xdr.c                |    4 +
>>  fs/nfsd/nfsproc.c                |    6 +
>>  fs/nfsd/nfsxdr.c                 |    2
>>  fs/stat.c                        |  175 ++++++++++++++++++++++++++++++++++----
>>  include/linux/fcntl.h            |    1
>>  include/linux/fs.h               |    2
>>  include/linux/stat.h             |  103 ++++++++++++++++++++++
>>  include/linux/syscalls.h         |    9 ++
>>  20 files changed, 368 insertions(+), 42 deletions(-)
>>
>> diff --git a/arch/x86/include/asm/unistd_32.h b/arch/x86/include/asm/unistd_32.h
>> index beb9b5f..a9953cc 100644
>> --- a/arch/x86/include/asm/unistd_32.h
>> +++ b/arch/x86/include/asm/unistd_32.h
>> @@ -343,10 +343,12 @@
>>  #define __NR_rt_tgsigqueueinfo 335
>>  #define __NR_perf_event_open   336
>>  #define __NR_recvmmsg          337
>> +#define __NR_xstat             338
>> +#define __NR_fxstat            339
>>
>>  #ifdef __KERNEL__
>>
>> -#define NR_syscalls 338
>> +#define NR_syscalls 340
>>
>>  #define __ARCH_WANT_IPC_PARSE_VERSION
>>  #define __ARCH_WANT_OLD_READDIR
>> diff --git a/arch/x86/include/asm/unistd_64.h b/arch/x86/include/asm/unistd_64.h
>> index ff4307b..c90d240 100644
>> --- a/arch/x86/include/asm/unistd_64.h
>> +++ b/arch/x86/include/asm/unistd_64.h
>> @@ -663,6 +663,10 @@ __SYSCALL(__NR_rt_tgsigqueueinfo, sys_rt_tgsigqueueinfo)
>>  __SYSCALL(__NR_perf_event_open, sys_perf_event_open)
>>  #define __NR_recvmmsg                          299
>>  __SYSCALL(__NR_recvmmsg, sys_recvmmsg)
>> +#define __NR_xstat                             300
>> +__SYSCALL(__NR_xstat, sys_xstat)
>> +#define __NR_fxstat                            301
>> +__SYSCALL(__NR_fxstat, sys_fxstat)
>>
>>  #ifndef __NO_STUBS
>>  #define __ARCH_WANT_OLD_READDIR
>> diff --git a/fs/afs/inode.c b/fs/afs/inode.c
>> index ee3190a..f624c5a 100644
>> --- a/fs/afs/inode.c
>> +++ b/fs/afs/inode.c
>> @@ -300,16 +300,17 @@ error_unlock:
>>  /*
>>  * read the attributes of an inode
>>  */
>> -int afs_getattr(struct vfsmount *mnt, struct dentry *dentry,
>> -                     struct kstat *stat)
>> +int afs_getattr(struct vfsmount *mnt, struct dentry *dentry, struct kstat *stat)
>>  {
>> -       struct inode *inode;
>> -
>> -       inode = dentry->d_inode;
>> +       struct inode *inode = dentry->d_inode;
>>
>>        _enter("{ ino=%lu v=%u }", inode->i_ino, inode->i_generation);
>>
>>        generic_fillattr(inode, stat);
>> +
>> +       stat->result_mask |= XSTAT_REQUEST_GEN | XSTAT_REQUEST_DATA_VERSION;
>> +       stat->gen = inode->i_generation;
>> +       stat->data_version = inode->i_version;
>>        return 0;
>>  }
>>
>> diff --git a/fs/ecryptfs/inode.c b/fs/ecryptfs/inode.c
>> index 31ef525..0b02272 100644
>> --- a/fs/ecryptfs/inode.c
>> +++ b/fs/ecryptfs/inode.c
>> @@ -994,6 +994,8 @@ int ecryptfs_getattr(struct vfsmount *mnt, struct dentry *dentry,
>>        struct kstat lower_stat;
>>        int rc;
>>
>> +       lower_stat.query_flags = stat->query_flags;
>> +       lower_stat.request_mask = stat->request_mask | XSTAT_REQUEST_BLOCKS;
>>        rc = vfs_getattr(ecryptfs_dentry_to_lower_mnt(dentry),
>>                         ecryptfs_dentry_to_lower(dentry), &lower_stat);
>>        if (!rc) {
>> diff --git a/fs/ext4/ext4.h b/fs/ext4/ext4.h
>> index 19a4de5..96823f3 100644
>> --- a/fs/ext4/ext4.h
>> +++ b/fs/ext4/ext4.h
>> @@ -1571,6 +1571,8 @@ extern int  ext4_write_inode(struct inode *, struct writeback_control *);
>>  extern int  ext4_setattr(struct dentry *, struct iattr *);
>>  extern int  ext4_getattr(struct vfsmount *mnt, struct dentry *dentry,
>>                                struct kstat *stat);
>> +extern int  ext4_file_getattr(struct vfsmount *mnt, struct dentry *dentry,
>> +                               struct kstat *stat);
>>  extern void ext4_delete_inode(struct inode *);
>>  extern int  ext4_sync_inode(handle_t *, struct inode *);
>>  extern void ext4_dirty_inode(struct inode *);
>> diff --git a/fs/ext4/file.c b/fs/ext4/file.c
>> index 5313ae4..18c29ab 100644
>> --- a/fs/ext4/file.c
>> +++ b/fs/ext4/file.c
>> @@ -150,7 +150,7 @@ const struct file_operations ext4_file_operations = {
>>  const struct inode_operations ext4_file_inode_operations = {
>>        .truncate       = ext4_truncate,
>>        .setattr        = ext4_setattr,
>> -       .getattr        = ext4_getattr,
>> +       .getattr        = ext4_file_getattr,
>>  #ifdef CONFIG_EXT4_FS_XATTR
>>        .setxattr       = generic_setxattr,
>>        .getxattr       = generic_getxattr,
>> diff --git a/fs/ext4/inode.c b/fs/ext4/inode.c
>> index 42272d6..f9a730a 100644
>> --- a/fs/ext4/inode.c
>> +++ b/fs/ext4/inode.c
>> @@ -5550,12 +5550,33 @@ err_out:
>>  int ext4_getattr(struct vfsmount *mnt, struct dentry *dentry,
>>                 struct kstat *stat)
>>  {
>> -       struct inode *inode;
>> -       unsigned long delalloc_blocks;
>> +       struct inode *inode = dentry->d_inode;
>>
>> -       inode = dentry->d_inode;
>>        generic_fillattr(inode, stat);
>>
>> +       stat->result_mask |= XSTAT_REQUEST_BTIME;
>> +       stat->btime.tv_sec = EXT4_I(inode)->i_crtime.tv_sec;
>> +       stat->btime.tv_nsec = EXT4_I(inode)->i_crtime.tv_nsec;
>> +
>> +       if (inode->i_ino != EXT4_ROOT_INO) {
>> +               stat->result_mask |= XSTAT_REQUEST_GEN;
>> +               stat->gen = inode->i_generation;
>> +       }
>> +       if (S_ISDIR(inode->i_mode) || test_opt(inode->i_sb, I_VERSION)) {
>> +               stat->result_mask |= XSTAT_REQUEST_DATA_VERSION;
>> +               stat->data_version = inode->i_version;
>> +       }
>> +       return 0;
>> +}
>> +
>> +int ext4_file_getattr(struct vfsmount *mnt, struct dentry *dentry,
>> +                     struct kstat *stat)
>> +{
>> +       struct inode *inode = dentry->d_inode;
>> +       unsigned long delalloc_blocks;
>> +
>> +       ext4_getattr(mnt, dentry, stat);
>> +
>>        /*
>>         * We can't update i_blocks if the block allocation is delayed
>>         * otherwise in the case of system crash before the real block
>> diff --git a/fs/ext4/namei.c b/fs/ext4/namei.c
>> index a43e661..0f776c7 100644
>> --- a/fs/ext4/namei.c
>> +++ b/fs/ext4/namei.c
>> @@ -2542,6 +2542,7 @@ const struct inode_operations ext4_dir_inode_operations = {
>>        .mknod          = ext4_mknod,
>>        .rename         = ext4_rename,
>>        .setattr        = ext4_setattr,
>> +       .getattr        = ext4_getattr,
>>  #ifdef CONFIG_EXT4_FS_XATTR
>>        .setxattr       = generic_setxattr,
>>        .getxattr       = generic_getxattr,
>> @@ -2554,6 +2555,7 @@ const struct inode_operations ext4_dir_inode_operations = {
>>
>>  const struct inode_operations ext4_special_inode_operations = {
>>        .setattr        = ext4_setattr,
>> +       .getattr        = ext4_getattr,
>>  #ifdef CONFIG_EXT4_FS_XATTR
>>        .setxattr       = generic_setxattr,
>>        .getxattr       = generic_getxattr,
>> diff --git a/fs/ext4/symlink.c b/fs/ext4/symlink.c
>> index ed9354a..d8fe7fb 100644
>> --- a/fs/ext4/symlink.c
>> +++ b/fs/ext4/symlink.c
>> @@ -35,6 +35,7 @@ const struct inode_operations ext4_symlink_inode_operations = {
>>        .follow_link    = page_follow_link_light,
>>        .put_link       = page_put_link,
>>        .setattr        = ext4_setattr,
>> +       .getattr        = ext4_getattr,
>>  #ifdef CONFIG_EXT4_FS_XATTR
>>        .setxattr       = generic_setxattr,
>>        .getxattr       = generic_getxattr,
>> @@ -47,6 +48,7 @@ const struct inode_operations ext4_fast_symlink_inode_operations = {
>>        .readlink       = generic_readlink,
>>        .follow_link    = ext4_follow_link,
>>        .setattr        = ext4_setattr,
>> +       .getattr        = ext4_getattr,
>>  #ifdef CONFIG_EXT4_FS_XATTR
>>        .setxattr       = generic_setxattr,
>>        .getxattr       = generic_getxattr,
>> diff --git a/fs/nfs/inode.c b/fs/nfs/inode.c
>> index 099b351..8c6de96 100644
>> --- a/fs/nfs/inode.c
>> +++ b/fs/nfs/inode.c
>> @@ -495,11 +495,21 @@ void nfs_setattr_update_inode(struct inode *inode, struct iattr *attr)
>>  int nfs_getattr(struct vfsmount *mnt, struct dentry *dentry, struct kstat *stat)
>>  {
>>        struct inode *inode = dentry->d_inode;
>> +       unsigned force = stat->query_flags & AT_FORCE_ATTR_SYNC;
>>        int need_atime = NFS_I(inode)->cache_validity & NFS_INO_INVALID_ATIME;
>>        int err;
>>
>> -       /* Flush out writes to the server in order to update c/mtime.  */
>> -       if (S_ISREG(inode->i_mode)) {
>> +       if (NFS_SERVER(inode)->nfs_client->rpc_ops->version < 4)
>> +               stat->request_mask &= ~XSTAT_REQUEST_DATA_VERSION;
>> +
>> +       /* Flush out writes to the server in order to update c/mtime
>> +        * or data version if the user wants them */
>> +       if ((force || stat->request_mask & (XSTAT_REQUEST_MTIME |
>> +                                           XSTAT_REQUEST_CTIME |
>> +                                           XSTAT_REQUEST_DATA_VERSION
>> +                                           )) &&
>> +           S_ISREG(inode->i_mode)
>> +           ) {
>>                err = filemap_write_and_wait(inode->i_mapping);
>>                if (err)
>>                        goto out;
>> @@ -514,18 +524,30 @@ int nfs_getattr(struct vfsmount *mnt, struct dentry *dentry, struct kstat *stat)
>>         *  - NFS never sets MS_NOATIME or MS_NODIRATIME so there is
>>         *    no point in checking those.
>>         */
>> -       if ((mnt->mnt_flags & MNT_NOATIME) ||
>> -           ((mnt->mnt_flags & MNT_NODIRATIME) && S_ISDIR(inode->i_mode)))
>> +       if (!(stat->request_mask & XSTAT_REQUEST_ATIME) ||
>> +           (mnt->mnt_flags & MNT_NOATIME) ||
>> +           ((mnt->mnt_flags & MNT_NODIRATIME) && S_ISDIR(inode->i_mode)))
>>                need_atime = 0;
>>
>> -       if (need_atime)
>> -               err = __nfs_revalidate_inode(NFS_SERVER(inode), inode);
>> -       else
>> -               err = nfs_revalidate_inode(NFS_SERVER(inode), inode);
>> -       if (!err) {
>> -               generic_fillattr(inode, stat);
>> -               stat->ino = nfs_compat_user_ino64(NFS_FILEID(inode));
>> +       if (force || stat->request_mask & (XSTAT_REQUEST__BASIC_STATS |
>> +                                          XSTAT_REQUEST_DATA_VERSION)
>> +           ) {
>> +               if (force || need_atime)
>> +                       err = __nfs_revalidate_inode(NFS_SERVER(inode), inode);
>> +               else
>> +                       err = nfs_revalidate_inode(NFS_SERVER(inode), inode);
>> +               if (err)
>> +                       goto out;
>>        }
>> +
>> +       generic_fillattr(inode, stat);
>> +       stat->ino = nfs_compat_user_ino64(NFS_FILEID(inode));
>> +
>> +       if (stat->request_mask & XSTAT_REQUEST_DATA_VERSION) {
>> +               stat->data_version = NFS_I(inode)->change_attr;
>> +               stat->result_mask |= XSTAT_REQUEST_DATA_VERSION;
>> +       }
>> +
>>  out:
>>        return err;
>>  }
>> @@ -770,7 +792,7 @@ int nfs_revalidate_inode(struct nfs_server *server, struct inode *inode)
>>  static int nfs_invalidate_mapping(struct inode *inode, struct address_space *mapping)
>>  {
>>        struct nfs_inode *nfsi = NFS_I(inode);
>> -
>> +
>>        if (mapping->nrpages != 0) {
>>                int ret = invalidate_inode_pages2(mapping);
>>                if (ret < 0)
>> diff --git a/fs/nfsd/nfs3proc.c b/fs/nfsd/nfs3proc.c
>> index 3d68f45..310ff05 100644
>> --- a/fs/nfsd/nfs3proc.c
>> +++ b/fs/nfsd/nfs3proc.c
>> @@ -55,6 +55,8 @@ nfsd3_proc_getattr(struct svc_rqst *rqstp, struct nfsd_fhandle  *argp,
>>        if (nfserr)
>>                RETURN_STATUS(nfserr);
>>
>> +       resp->stat.query_flags = 0;
>> +       resp->stat.request_mask = XSTAT_REQUEST__EXTENDED_STATS;
>>        err = vfs_getattr(resp->fh.fh_export->ex_path.mnt,
>>                          resp->fh.fh_dentry, &resp->stat);
>>        nfserr = nfserrno(err);
>> diff --git a/fs/nfsd/nfs3xdr.c b/fs/nfsd/nfs3xdr.c
>> index 2a533a0..eaa3c3b 100644
>> --- a/fs/nfsd/nfs3xdr.c
>> +++ b/fs/nfsd/nfs3xdr.c
>> @@ -205,6 +205,8 @@ encode_post_op_attr(struct svc_rqst *rqstp, __be32 *p, struct svc_fh *fhp)
>>                int err;
>>                struct kstat stat;
>>
>> +               stat.query_flags = 0;
>> +               stat.request_mask = XSTAT_REQUEST__EXTENDED_STATS;
>>                err = vfs_getattr(fhp->fh_export->ex_path.mnt, dentry, &stat);
>>                if (!err) {
>>                        *p++ = xdr_one;         /* attributes follow */
>> @@ -257,6 +259,8 @@ void fill_post_wcc(struct svc_fh *fhp)
>>        if (fhp->fh_post_saved)
>>                printk("nfsd: inode locked twice during operation.\n");
>>
>> +       fhp->fh_post_attr.query_flags = 0;
>> +       fhp->fh_post_attr.request_mask = XSTAT_REQUEST__EXTENDED_STATS;
>>        err = vfs_getattr(fhp->fh_export->ex_path.mnt, fhp->fh_dentry,
>>                        &fhp->fh_post_attr);
>>        fhp->fh_post_change = fhp->fh_dentry->d_inode->i_version;
>> diff --git a/fs/nfsd/nfs4xdr.c b/fs/nfsd/nfs4xdr.c
>> index ac17a70..e9d1b59 100644
>> --- a/fs/nfsd/nfs4xdr.c
>> +++ b/fs/nfsd/nfs4xdr.c
>> @@ -1769,6 +1769,8 @@ nfsd4_encode_fattr(struct svc_fh *fhp, struct svc_export *exp,
>>                        goto out;
>>        }
>>
>> +       stat.query_flags = 0;
>> +       stat.request_mask = XSTAT_REQUEST__EXTENDED_STATS;
>>        err = vfs_getattr(exp->ex_path.mnt, dentry, &stat);
>>        if (err)
>>                goto out_nfserr;
>> @@ -2139,6 +2141,8 @@ out_acl:
>>                                if (path.dentry != path.mnt->mnt_root)
>>                                        break;
>>                        }
>> +                       stat.query_flags = 0;
>> +                       stat.request_mask = XSTAT_REQUEST__EXTENDED_STATS;
>>                        err = vfs_getattr(path.mnt, path.dentry, &stat);
>>                        path_put(&path);
>>                        if (err)
>> diff --git a/fs/nfsd/nfsproc.c b/fs/nfsd/nfsproc.c
>> index a047ad6..7c0e74b 100644
>> --- a/fs/nfsd/nfsproc.c
>> +++ b/fs/nfsd/nfsproc.c
>> @@ -26,6 +26,8 @@ static __be32
>>  nfsd_return_attrs(__be32 err, struct nfsd_attrstat *resp)
>>  {
>>        if (err) return err;
>> +       resp->stat.query_flags = 0;
>> +       resp->stat.request_mask = XSTAT_REQUEST__EXTENDED_STATS;
>>        return nfserrno(vfs_getattr(resp->fh.fh_export->ex_path.mnt,
>>                                    resp->fh.fh_dentry,
>>                                    &resp->stat));
>> @@ -34,6 +36,8 @@ static __be32
>>  nfsd_return_dirop(__be32 err, struct nfsd_diropres *resp)
>>  {
>>        if (err) return err;
>> +       resp->stat.query_flags = 0;
>> +       resp->stat.request_mask = XSTAT_REQUEST__EXTENDED_STATS;
>>        return nfserrno(vfs_getattr(resp->fh.fh_export->ex_path.mnt,
>>                                    resp->fh.fh_dentry,
>>                                    &resp->stat));
>> @@ -150,6 +154,8 @@ nfsd_proc_read(struct svc_rqst *rqstp, struct nfsd_readargs *argp,
>>                                  &resp->count);
>>
>>        if (nfserr) return nfserr;
>> +       resp->stat.query_flags = 0;
>> +       resp->stat.request_mask = XSTAT_REQUEST__EXTENDED_STATS;
>>        return nfserrno(vfs_getattr(resp->fh.fh_export->ex_path.mnt,
>>                                    resp->fh.fh_dentry,
>>                                    &resp->stat));
>> diff --git a/fs/nfsd/nfsxdr.c b/fs/nfsd/nfsxdr.c
>> index 4ce005d..a595fb6 100644
>> --- a/fs/nfsd/nfsxdr.c
>> +++ b/fs/nfsd/nfsxdr.c
>> @@ -197,6 +197,8 @@ encode_fattr(struct svc_rqst *rqstp, __be32 *p, struct svc_fh *fhp,
>>  __be32 *nfs2svc_encode_fattr(struct svc_rqst *rqstp, __be32 *p, struct svc_fh *fhp)
>>  {
>>        struct kstat stat;
>> +       stat.query_flags = 0;
>> +       stat.request_mask = XSTAT_REQUEST__EXTENDED_STATS;
>>        vfs_getattr(fhp->fh_export->ex_path.mnt, fhp->fh_dentry, &stat);
>>        return encode_fattr(rqstp, p, fhp, &stat);
>>  }
>> diff --git a/fs/stat.c b/fs/stat.c
>> index 12e90e2..2fb1527 100644
>> --- a/fs/stat.c
>> +++ b/fs/stat.c
>> @@ -33,6 +33,9 @@ void generic_fillattr(struct inode *inode, struct kstat *stat)
>>        stat->size = i_size_read(inode);
>>        stat->blocks = inode->i_blocks;
>>        stat->blksize = (1 << inode->i_blkbits);
>> +       stat->result_mask |= XSTAT_REQUEST__BASIC_STATS & ~XSTAT_REQUEST_RDEV;
>> +       if (unlikely(S_ISBLK(stat->mode) || S_ISCHR(stat->mode)))
>> +               stat->result_mask |= XSTAT_REQUEST_RDEV;
>>  }
>>
>>  EXPORT_SYMBOL(generic_fillattr);
>> @@ -42,6 +45,8 @@ int vfs_getattr(struct vfsmount *mnt, struct dentry *dentry, struct kstat *stat)
>>        struct inode *inode = dentry->d_inode;
>>        int retval;
>>
>> +       stat->result_mask = 0;
>> +
>>        retval = security_inode_getattr(mnt, dentry);
>>        if (retval)
>>                return retval;
>> @@ -55,41 +60,64 @@ int vfs_getattr(struct vfsmount *mnt, struct dentry *dentry, struct kstat *stat)
>>
>>  EXPORT_SYMBOL(vfs_getattr);
>>
>> -int vfs_fstat(unsigned int fd, struct kstat *stat)
>> +/*
>> + * VFS entrypoint to get extended stats by file descriptor
>> + */
>> +int vfs_fxstat(unsigned int fd, int flags, struct kstat *stat)
>>  {
>>        struct file *f = fget(fd);
>>        int error = -EBADF;
>>
>> +       if (flags & ~KSTAT_QUERY_FLAGS)
>> +               return -EINVAL;
>> +       stat->query_flags = flags;
>> +
>>        if (f) {
>>                error = vfs_getattr(f->f_path.mnt, f->f_path.dentry, stat);
>>                fput(f);
>>        }
>>        return error;
>>  }
>> +EXPORT_SYMBOL(vfs_fxstat);
>> +
>> +int vfs_fstat(unsigned int fd, struct kstat *stat)
>> +{
>> +       stat->request_mask = XSTAT_REQUEST__BASIC_STATS;
>> +       return vfs_fxstat(fd, 0, stat);
>> +}
>>  EXPORT_SYMBOL(vfs_fstat);
>>
>> -int vfs_fstatat(int dfd, const char __user *filename, struct kstat *stat,
>> -               int flag)
>> +/*
>> + * VFS entrypoint to get extended stats by filename
>> + */
>> +int vfs_xstat(int dfd, const char __user *filename, int flags,
>> +             struct kstat *stat)
>>  {
>>        struct path path;
>> -       int error = -EINVAL;
>> -       int lookup_flags = 0;
>> +       int error, lookup_flags;
>>
>> -       if ((flag & ~AT_SYMLINK_NOFOLLOW) != 0)
>> -               goto out;
>> +       if (flags & ~(AT_SYMLINK_NOFOLLOW | KSTAT_QUERY_FLAGS))
>> +               return -EINVAL;
>>
>> -       if (!(flag & AT_SYMLINK_NOFOLLOW))
>> -               lookup_flags |= LOOKUP_FOLLOW;
>> +       stat->query_flags = flags;
>> +       lookup_flags = (flags & AT_SYMLINK_NOFOLLOW) ? 0 : LOOKUP_FOLLOW;
>>
>>        error = user_path_at(dfd, filename, lookup_flags, &path);
>> -       if (error)
>> -               goto out;
>> -
>> -       error = vfs_getattr(path.mnt, path.dentry, stat);
>> -       path_put(&path);
>> -out:
>> +       if (!error) {
>> +               error = vfs_getattr(path.mnt, path.dentry, stat);
>> +               path_put(&path);
>> +       }
>>        return error;
>>  }
>> +EXPORT_SYMBOL(vfs_xstat);
>> +
>> +int vfs_fstatat(int dfd, const char __user *filename, struct kstat *stat,
>> +               int flags)
>> +{
>> +       stat->request_mask = XSTAT_REQUEST__BASIC_STATS;
>> +       stat->query_flags = 0;
>> +       return vfs_xstat(dfd, filename, flags, stat);
>> +}
>>  EXPORT_SYMBOL(vfs_fstatat);
>>
>>  int vfs_stat(const char __user *name, struct kstat *stat)
>> @@ -115,7 +143,7 @@ static int cp_old_stat(struct kstat *stat, struct __old_kernel_stat __user * sta
>>  {
>>        static int warncount = 5;
>>        struct __old_kernel_stat tmp;
>> -
>> +
>>        if (warncount > 0) {
>>                warncount--;
>>                printk(KERN_WARNING "VFS: Warning: %s using old stat() call. Recompile your binary.\n",
>> @@ -140,7 +168,7 @@ static int cp_old_stat(struct kstat *stat, struct __old_kernel_stat __user * sta
>>  #if BITS_PER_LONG == 32
>>        if (stat->size > MAX_NON_LFS)
>>                return -EOVERFLOW;
>> -#endif
>> +#endif
>>        tmp.st_size = stat->size;
>>        tmp.st_atime = stat->atime.tv_sec;
>>        tmp.st_mtime = stat->mtime.tv_sec;
>> @@ -222,7 +250,7 @@ static int cp_new_stat(struct kstat *stat, struct stat __user *statbuf)
>>  #if BITS_PER_LONG == 32
>>        if (stat->size > MAX_NON_LFS)
>>                return -EOVERFLOW;
>> -#endif
>> +#endif
>>        tmp.st_size = stat->size;
>>        tmp.st_atime = stat->atime.tv_sec;
>>        tmp.st_mtime = stat->mtime.tv_sec;
>> @@ -408,6 +436,117 @@ SYSCALL_DEFINE4(fstatat64, int, dfd, const char __user *, filename,
>>  }
>>  #endif /* __ARCH_WANT_STAT64 */
>>
>> +/*
>> + * Get the xstat parameters if supplied
>> + */
>> +static int xstat_get_params(struct xstat_parameters __user *_params,
>> +                           struct kstat *stat)
>> +{
>> +       struct xstat_parameters params;
>> +
>> +       memset(stat, 0xde, sizeof(*stat));      // DEBUGGING
>> +
>> +       if (_params) {
>> +               if (copy_from_user(&params, _params, sizeof(params)) != 0)
>> +                       return -EFAULT;
>> +               stat->request_mask =
>> +                       params.request_mask & XSTAT_REQUEST__ALL_STATS;
>> +       } else {
>> +               stat->request_mask = XSTAT_REQUEST__EXTENDED_STATS;
>> +       }
>> +       stat->result_mask = 0;
>> +       return 0;
>> +}
>> +
>> +/*
>> + * copy the extended stats to userspace and return the amount of data written
>> + * into the buffer
>> + */
>> +static long xstat_set_result(struct kstat *stat,
>> +                            struct xstat __user *buffer, size_t bufsize)
>> +{
>> +       struct xstat tmp;
>> +       size_t copy;
>> +
>> +       /* transfer the fixed results */
>> +       memset(&tmp, 0, sizeof(tmp));
>> +       tmp.st_result_mask      = stat->result_mask;
>> +       tmp.st_mode             = stat->mode;
>> +       tmp.st_nlink            = stat->nlink;
>> +       tmp.st_uid              = stat->uid;
>> +       tmp.st_gid              = stat->gid;
>> +       tmp.st_blksize          = stat->blksize;
>> +       tmp.st_rdev.major       = MAJOR(stat->rdev);
>> +       tmp.st_rdev.minor       = MINOR(stat->rdev);
>> +       tmp.st_dev.major        = MAJOR(stat->dev);
>> +       tmp.st_dev.minor        = MINOR(stat->dev);
>> +       tmp.st_atime.tv_sec     = stat->atime.tv_sec;
>> +       tmp.st_atime.tv_nsec    = stat->atime.tv_nsec;
>> +       tmp.st_mtime.tv_sec     = stat->mtime.tv_sec;
>> +       tmp.st_mtime.tv_nsec    = stat->mtime.tv_nsec;
>> +       tmp.st_ctime.tv_sec     = stat->ctime.tv_sec;
>> +       tmp.st_ctime.tv_nsec    = stat->ctime.tv_nsec;
>> +       tmp.st_ino              = stat->ino;
>> +       tmp.st_size             = stat->size;
>> +       tmp.st_blocks           = stat->blocks;
>> +
>> +       if (tmp.st_result_mask & XSTAT_REQUEST_BTIME) {
>> +               tmp.st_btime.tv_sec     = stat->btime.tv_sec;
>> +               tmp.st_btime.tv_nsec    = stat->btime.tv_nsec;
>> +       }
>> +       if (tmp.st_result_mask & XSTAT_REQUEST_GEN)
>> +               tmp.st_gen              = stat->gen;
>> +       if (tmp.st_result_mask & XSTAT_REQUEST_DATA_VERSION)
>> +               tmp.st_data_version     = stat->data_version;
>> +
>> +       copy = sizeof(tmp);
>> +       if (copy > bufsize)
>> +               copy = bufsize;
>> +       if (copy_to_user(buffer, &tmp, copy) != 0)
>> +               return -EFAULT;
>> +       return sizeof(tmp);
>> +}
>> +
>> +/*
>> + * System call to get extended stats by path
>> + */
>> +SYSCALL_DEFINE6(xstat,
>> +               int, dfd, const char __user *, filename, unsigned, atflag,
>> +               struct xstat_parameters __user *, params,
>> +               struct xstat __user *, buffer, size_t, bufsize)
>> +{
>> +       struct kstat stat;
>> +       int error;
>> +
>> +       error = xstat_get_params(params, &stat);
>> +       if (error != 0)
>> +               return error;
>> +       error = vfs_xstat(dfd, filename, atflag, &stat);
>> +       if (error)
>> +               return error;
>> +       return xstat_set_result(&stat, buffer, bufsize);
>> +}
>> +
>> +/*
>> + * System call to get extended stats by file descriptor
>> + */
>> +SYSCALL_DEFINE5(fxstat, unsigned int, fd, unsigned int, flags,
>> +               struct xstat_parameters __user *, params,
>> +               struct xstat __user *, buffer, size_t, bufsize)
>> +{
>> +       struct kstat stat;
>> +       int error;
>> +
>> +       error = xstat_get_params(params, &stat);
>> +       if (error < 0)
>> +               return error;
>> +       error = vfs_fxstat(fd, flags, &stat);
>> +       if (error)
>> +               return error;
>> +
>> +       return xstat_set_result(&stat, buffer, bufsize);
>> +}
>> +
>>  /* Caller is here responsible for sufficient locking (ie. inode->i_lock) */
>>  void __inode_add_bytes(struct inode *inode, loff_t bytes)
>>  {
>> diff --git a/include/linux/fcntl.h b/include/linux/fcntl.h
>> index afc00af..bcf8083 100644
>> --- a/include/linux/fcntl.h
>> +++ b/include/linux/fcntl.h
>> @@ -45,6 +45,7 @@
>>  #define AT_REMOVEDIR           0x200   /* Remove directory instead of
>>                                            unlinking file.  */
>>  #define AT_SYMLINK_FOLLOW      0x400   /* Follow symbolic links.  */
>> +#define AT_FORCE_ATTR_SYNC     0x800   /* Force the attributes to be sync'd with the server */
>>
>>  #ifdef __KERNEL__
>>
>> diff --git a/include/linux/fs.h b/include/linux/fs.h
>> index a18bcea..9ce2119 100644
>> --- a/include/linux/fs.h
>> +++ b/include/linux/fs.h
>> @@ -2343,6 +2343,8 @@ extern int vfs_stat(const char __user *, struct kstat *);
>>  extern int vfs_lstat(const char __user *, struct kstat *);
>>  extern int vfs_fstat(unsigned int, struct kstat *);
>>  extern int vfs_fstatat(int , const char __user *, struct kstat *, int);
>> +extern int vfs_xstat(int, const char __user *, int, struct kstat *);
>> +extern int vfs_xfstat(unsigned int, struct kstat *);
>>
>>  extern int do_vfs_ioctl(struct file *filp, unsigned int fd, unsigned int cmd,
>>                    unsigned long arg);
>> diff --git a/include/linux/stat.h b/include/linux/stat.h
>> index 611c398..e0b89e4 100644
>> --- a/include/linux/stat.h
>> +++ b/include/linux/stat.h
>> @@ -46,6 +46,99 @@
>>
>>  #endif
>>
>> +/*
>> + * Extended stat structures
>> + */
>> +struct xstat_parameters {
>> +       /* Query request/result mask
>> +        *
>> +        * Bits should be set in request_mask to request particular items
>> +        * before calling xstat() or fxstat().
>> +        *
>> +        * For each item in the set XSTAT_REQUEST__EXTENDED_STATS:
>> +        *
>> +        * - if not available at all, the bit will be cleared before returning
>> +        *   and the field will be cleared; otherwise,
>> +        *
>> +        * - if AT_FORCE_ATTR_SYNC is set, then the datum will be synchronised
>> +        *   to the server and the bit will be set on return; otherwise,
>> +        *
>> +        * - if requested, the datum will be synchronised to a server or other
>> +        *   hardware if out of date before being returned, and the bit will be
>> +        *   set on return; otherwise,
>> +        *
>> +        * - if not requested, but available in approximate form without any
>> +        *   effort, it will be filled in anyway, and the bit will be set upon
>> +        *   return (it might not be up to date, however, and no attempt will
>> +        *   be made to synchronise the internal state first); otherwise,
>> +        *
>> +        * - the bit will be cleared before returning, and the field will be
>> +         *   cleared.
>> +        *
>> +        * For each item not in the set XSTAT_REQUEST__EXTENDED_STATS
>> +        *
>> +        * - if not available at all, the bit will be cleared, and no result
>> +         *   data will be returned; otherwise,
>> +        *
>> +        * - if requested, the datum will be synchronised to a server or other
>> +        *   hardware before being appended if necessary, and the bit will be
>> +        *   set on return; otherwise,
>> +        *
>> +        * - the bit will be cleared, and no result data will be returned.
>> +        *
>> +        * Items in XSTAT_REQUEST__BASIC_STATS may be marked unavailable on
>> +        * return, but they will have a value installed for compatibility
>> +        * purposes.
>> +        */
>> +       unsigned long long      request_mask;
>> +#define XSTAT_REQUEST_MODE             0x00000001ULL   /* want/got st_mode */
>> +#define XSTAT_REQUEST_NLINK            0x00000002ULL   /* want/got st_nlink */
>> +#define XSTAT_REQUEST_UID              0x00000004ULL   /* want/got st_uid */
>> +#define XSTAT_REQUEST_GID              0x00000008ULL   /* want/got st_gid */
>> +#define XSTAT_REQUEST_RDEV             0x00000010ULL   /* want/got st_rdev */
>> +#define XSTAT_REQUEST_ATIME            0x00000020ULL   /* want/got st_atime */
>> +#define XSTAT_REQUEST_MTIME            0x00000040ULL   /* want/got st_mtime */
>> +#define XSTAT_REQUEST_CTIME            0x00000080ULL   /* want/got st_ctime */
>> +#define XSTAT_REQUEST_INO              0x00000100ULL   /* want/got st_ino */
>> +#define XSTAT_REQUEST_SIZE             0x00000200ULL   /* want/got st_size */
>> +#define XSTAT_REQUEST_BLOCKS           0x00000400ULL   /* want/got st_blocks */
>> +#define XSTAT_REQUEST__BASIC_STATS     0x000007ffULL   /* the stuff in the normal stat struct */
>> +#define XSTAT_REQUEST_BTIME            0x00000800ULL   /* want/got st_btime */
>> +#define XSTAT_REQUEST_GEN              0x00001000ULL   /* want/got st_gen */
>> +#define XSTAT_REQUEST_DATA_VERSION     0x00002000ULL   /* want/got st_data_version */
>> +#define XSTAT_REQUEST__EXTENDED_STATS  0x00003fffULL   /* the stuff in the xstat struct */
>> +#define XSTAT_REQUEST__ALL_STATS       0x00003fffULL   /* the defined set of requestables */
>> +};
>> +
>> +struct xstat_dev {
>> +       unsigned int            major, minor;
>> +};
>> +
>> +struct xstat_time {
>> +       unsigned long long      tv_sec, tv_nsec;
>> +};
>> +
>> +struct xstat {
>> +       unsigned int            st_mode;        /* file mode */
>> +       unsigned int            st_nlink;       /* number of hard links */
>> +       unsigned int            st_uid;         /* user ID of owner */
>> +       unsigned int            st_gid;         /* group ID of owner */
>> +       struct xstat_dev        st_rdev;        /* device ID of special file */
>> +       struct xstat_dev        st_dev;         /* ID of device containing file */
>> +       struct xstat_time       st_atime;       /* last access time */
>> +       struct xstat_time       st_mtime;       /* last data modification time */
>> +       struct xstat_time       st_ctime;       /* last attribute change time */
>> +       struct xstat_time       st_btime;       /* file creation time */
>> +       unsigned long long      st_ino;         /* inode number */
>> +       unsigned long long      st_size;        /* file size */
>> +       unsigned long long      st_blksize;     /* block size for filesystem I/O */
>> +       unsigned long long      st_blocks;      /* number of 512-byte blocks allocated */
>> +       unsigned long long      st_gen;         /* inode generation number */
>> +       unsigned long long      st_data_version; /* data version number */
>> +       unsigned long long      st_result_mask; /* what requests were written */
>> +       unsigned long long      st_extra_results[0]; /* extra requested results */
>> +};
>> +
>>  #ifdef __KERNEL__
>>  #define S_IRWXUGO      (S_IRWXU|S_IRWXG|S_IRWXO)
>>  #define S_IALLUGO      (S_ISUID|S_ISGID|S_ISVTX|S_IRWXUGO)
>> @@ -67,14 +160,20 @@ struct kstat {
>>        uid_t           uid;
>>        gid_t           gid;
>>        dev_t           rdev;
>> +       unsigned int    query_flags;            /* operational flags */
>> +#define KSTAT_QUERY_FLAGS (AT_FORCE_ATTR_SYNC)
>>        loff_t          size;
>> -       struct timespec  atime;
>> +       struct timespec atime;
>>        struct timespec mtime;
>>        struct timespec ctime;
>> +       struct timespec btime;                  /* file creation time */
>>        unsigned long   blksize;
>>        unsigned long long      blocks;
>> +       u64             request_mask;           /* what fields the user asked for */
>> +       u64             result_mask;            /* what fields the user got */
>> +       u64             gen;                    /* inode generation */
>> +       u64             data_version;
>>  };
>>
>>  #endif
>> -
>>  #endif
>> diff --git a/include/linux/syscalls.h b/include/linux/syscalls.h
>> index 8812a63..5d68b4c 100644
>> --- a/include/linux/syscalls.h
>> +++ b/include/linux/syscalls.h
>> @@ -44,6 +44,8 @@ struct shmid_ds;
>>  struct sockaddr;
>>  struct stat;
>>  struct stat64;
>> +struct xstat_parameters;
>> +struct xstat;
>>  struct statfs;
>>  struct statfs64;
>>  struct __sysctl_args;
>> @@ -824,4 +826,11 @@ asmlinkage long sys_mmap_pgoff(unsigned long addr, unsigned long len,
>>                        unsigned long fd, unsigned long pgoff);
>>  asmlinkage long sys_old_mmap(struct mmap_arg_struct __user *arg);
>>
>> +asmlinkage long sys_xstat(int, const char __user *, unsigned,
>> +                         struct xstat_parameters __user *,
>> +                         struct xstat __user *, size_t);
>> +asmlinkage long sys_fxstat(unsigned, unsigned,
>> +                          struct xstat_parameters __user *,
>> +                          struct xstat __user *, size_t);
>> +
>>  #endif
>>
>> --
>> To unsubscribe from this list: send the line "unsubscribe linux-fsdevel" in
>> the body of a message to majordomo@vger.kernel.org
>> More majordomo info at  http://vger.kernel.org/majordomo-info.html
>>
>
>
>
> --
> Michael Kerrisk Linux man-pages maintainer;
> http://www.kernel.org/doc/man-pages/
> Author of "The Linux Programming Interface", http://blog.man7.org/
> --
> To unsubscribe from this list: send the line "unsubscribe linux-api" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
>



-- 
Michael Kerrisk
Linux man-pages maintainer; http://www.kernel.org/doc/man-pages/
Author of "The Linux Programming Interface" http://blog.man7.org/
--
To unsubscribe from this list: send the line "unsubscribe linux-ext4" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: [PATCH] Add a pair of system calls to make extended file stats available [ver #3]

[David Howells]

Michael Kerrisk <mtk.manpages@gmail.com> wrote:

> * Include information from the "inode_info" structure, most notably
> i_flags, but perhaps other info as well.

This thought has occurred to me, but are the contents of i_flags identical for
all filesystems?  Certainly, i_flags doesn't seem to match the FS_IOC_GETFLAGS
mask.  For example:

	#define	FS_SECRM_FL			0x00000001

vs:

	#define S_SYNC		1	/* Writes are synced at once */

I've also been asked to provide st_flags as for BSD, which aren't compatible
either:-/.

Some questions:

 (1) Does it make sense to rearrange the S_xxxx flags to match the numbers for
     FS_xxxx_FL?

 (2) Does it make sense to do the BSD st_flags compatibility in userspace?

 (3) Can we add a couple more flags to make Samba's life easier?  Say an
     archived bit and a hidden bit?

 (4) Do I actually need to provide a mask saying what st_flags are applicable
     to the file you've just queried?

 (5) How often are these flags required?  E.g. does it make more sense to keep
     them as an additional result, or does it make sense to stick them in the
     kstat and xstat structs, knowing that these are allocated on the kernel
     stack maybe as three times if an ecryptfs file?

> * Return a bit mask indicating the presence of additional information
> associated with the i-node. Here, I am thinking of flags that indicate
> that the file has any of the following: capabilities, an ACL, and
> extended attributes (obviously a superset of the previous). I could
> imagine some apps that, having got the xstat info, would be interested
> to obtain some of this other info.
> 
> Obviously, the above only make sense if the overhead of providing the
> extra information is low.

That might make sense as an 'additional result'.  These things may have to be
probed for on disk or on a server, so you might not want to return them by
default, and you may want to indicate what the filesystem can support vs what
the file actually has:

	u64	st_fs_additional_info;	/* what the filesystem supports */
	u64	st_file_additional_info; /* what the file actually has */

	#define XST_ADDINFO_CAPABILITY_MASK
	#define XST_ADDINFO_ACL
	#define XST_ADDINFO_XATTRS
	#define XST_ADDINFO_SECLABEL

David

Re: [PATCH] Add a pair of system calls to make extended file stats available [ver #3]

[Andreas Dilger]

On 2010-07-05, at 08:59, David Howells wrote:
> Michael Kerrisk <mtk.manpages-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org> wrote:
> 
>> * Include information from the "inode_info" structure, most notably
>> i_flags, but perhaps other info as well.
> 
> This thought has occurred to me, but are the contents of i_flags identical for
> all filesystems?  Certainly, i_flags doesn't seem to match the FS_IOC_GETFLAGS
> mask.  For example:
> 
> 	#define	FS_SECRM_FL			0x00000001
> 
> vs:
> 
> 	#define S_SYNC		1	/* Writes are synced at once */
> 
> (1) Does it make sense to rearrange the S_xxxx flags to match the numbers for
>   FS_xxxx_FL?

I saw your patch to that effect.  I'm of mixed feelings about this, since the S_* flags have traditionally been changed on an ad-hoc basis and I don't necessarily want to let this leak into the on-disk format of these flags for ext*.

One way to ensure that this holds true is to have a compile-time assertion that the respective S_* flags match FS_*_FL and EXT_*_FL.

We use a macro in Lustre for compile-time assertions that depends on the detection of duplicate values in a switch() statement:

/** Compile-time assertion.
* Check an invariant described by a constant expression at compile time by
* forcing a compiler error if it does not hold.  @cond must be a constant
* expression as defined by the ISO C Standard:
*
*       6.8.4.2  The switch statement
*       ....
*       [#3] The expression of each case label shall be  an  integer
*       constant   expression  and  no  two  of  the  case  constant
*       expressions in the same switch statement shall have the same
*       value after conversion...
*
*/
#define CLASSERT(cond) do {switch(42) {case (cond): case 0: break;}} while (0)

> (2) Does it make sense to do the BSD st_flags compatibility in userspace?
> 
> (3) Can we add a couple more flags to make Samba's life easier?  Say an
>   archived bit and a hidden bit?

I wouldn't object to that.  The BSD flags indicate that the hidden bit should only affect GUI display managers, not "ls".

> (4) Do I actually need to provide a mask saying what st_flags are applicable
>   to the file you've just queried?

Hmm, good question.

> (5) How often are these flags required?  E.g. does it make more sense to keep
>   them as an additional result, or does it make sense to stick them in the
>   kstat and xstat structs, knowing that these are allocated on the kernel
>   stack maybe as three times if an ecryptfs file?

If they aren't requested by userspace, the cost is mostly irrelevant.  I think on OSX these flags are returned for every "ls" call, to mark the inodes with xattrs every time.

Cheers, Andreas

Re: [PATCH] Add a pair of system calls to make extended file stats available [ver #3]

[David Howells]

Andreas Dilger <adilger-m1MBpc4rdrD3fQ9qLvQP4Q@public.gmane.org> wrote:

> I saw your patch to that effect.  I'm of mixed feelings about this, since
> the S_* flags have traditionally been changed on an ad-hoc basis and I don't
> necessarily want to let this leak into the on-disk format of these flags for
> ext*.
> 
> One way to ensure that this holds true is to have a compile-time assertion
> that the respective S_* flags match FS_*_FL and EXT_*_FL.

Something like this:

	#if S_SYNC != FS_SYNC_FL               ||      \
	       S_IMMUTABLE != FS_IMMUTABLE_FL  ||      \
	       S_APPEND != FS_APPEND_FL        ||      \
	       S_NOATIME != FS_NOATIME_FL      ||      \
	       S_DIRSYNC != FS_DIRSYNC_FL
	#error Ext2/3/4 assumes these equivalences
	#endif

would do the trick.

> > (5) How often are these flags required?  E.g. does it make more sense to
> >   keep them as an additional result, or does it make sense to stick them
> >   in the kstat and xstat structs, knowing that these are allocated on the
> >   kernel stack maybe as three times if an ecryptfs file?
> 
> If they aren't requested by userspace, the cost is mostly irrelevant.  I
> think on OSX these flags are returned for every "ls" call, to mark the
> inodes with xattrs every time.

I suppose.  I was thinking that they may have to be retrieved from the server.

More concerning to me is the effect of adding more fields to the kstat struct.

Nonetheless, having these flags around may be useful to CIFS, Samba, NFS and
NFSD as various of them may appear in those protocols.  Certainly, SMB passes
a bit indicating compression around (ATTR_COMPRESSED).

David

Re: [PATCH] Add a pair of system calls to make extended file stats available [ver #3]

[Michael Kerrisk]

Hi David,

A couple of comments below.

On Thu, Jul 1, 2010 at 1:36 AM, David Howells <dhowells@redhat.com> wrote:
> Add a pair of system calls to make extended file stats available, including
> file creation time, inode version and data version where available through the
> underlying filesystem.
>
> [This depends on the previously posted pair of patches to (a) constify a number
>  of syscall string and buffer arguments and (b) rearrange AFS's use of
>  i_version and i_generation].
>
> The following structures are defined for their use:
>
>        struct xstat_parameters {
>                unsigned long long      request_mask;

Poor name, since it's a value-result arg? Better maybe something like
"field_mask"?

>        };
>
>        struct xstat_dev {
>                unsigned int            major, minor;
>        };
>
>        struct xstat_time {
>                unsigned long long      tv_sec, tv_nsec;
>        };
>
>        struct xstat {
>                unsigned int            st_mode;
>                unsigned int            st_nlink;
>                unsigned int            st_uid;
>                unsigned int            st_gid;
>                struct xstat_dev        st_rdev;
>                struct xstat_dev        st_dev;
>                struct xstat_time       st_atime;
>                struct xstat_time       st_mtime;
>                struct xstat_time       st_ctime;
>                struct xstat_time       st_btime;
>                unsigned long long      st_ino;
>                unsigned long long      st_size;
>                unsigned long long      st_blksize;
>                unsigned long long      st_blocks;
>                unsigned long long      st_gen;
>                unsigned long long      st_data_version;
>                unsigned long long      st_result_mask;
>                unsigned long long      st_extra_results[0];
>        };
>
> where st_btime is the file creation time, st_gen is the inode generation
> (i_generation), st_data_version is the data version number (i_version),
> request_mask and st_result_mask are bitmasks of data desired/provided and
> st_extra_results[] is where as-yet undefined fields are appended.
>
> The defined bits in request_mask and st_result_mask are:
>
>        XSTAT_REQUEST_MODE              Want/got st_mode
>        XSTAT_REQUEST_NLINK             Want/got st_nlink
>        XSTAT_REQUEST_UID               Want/got st_uid
>        XSTAT_REQUEST_GID               Want/got st_gid
>        XSTAT_REQUEST_RDEV              Want/got st_rdev
>        XSTAT_REQUEST_ATIME             Want/got st_atime
>        XSTAT_REQUEST_MTIME             Want/got st_mtime
>        XSTAT_REQUEST_CTIME             Want/got st_ctime
>        XSTAT_REQUEST_INO               Want/got st_ino
>        XSTAT_REQUEST_SIZE              Want/got st_size
>        XSTAT_REQUEST_BLOCKS            Want/got st_blocks
>        XSTAT_REQUEST__BASIC_STATS      The stuff in the normal stat struct
>        XSTAT_REQUEST_BTIME             Want/got st_btime
>        XSTAT_REQUEST_GEN               Want/got st_gen
>        XSTAT_REQUEST_DATA_VERSION      Want/got st_data_version
>        XSTAT_REQUEST__EXTENDED_STATS   The stuff in the xstat struct
>        XSTAT_REQUEST__ALL_STATS        The defined set of requestables
>
> The system calls are:
>
>        ssize_t ret = xstat(int dfd,
>                            const char *filename,
>                            unsigned flags,
>                            const struct xstat_parameters *params,
>                            struct xstat *buffer,
>                            size_t buflen);
>
>        ssize_t ret = fxstat(unsigned fd,
>                             unsigned flags,
>                             const struct xstat_parameters *params,
>                             struct xstat *buffer,
>                             size_t buflen);
>
>
> The dfd, filename, flags and fd parameters indicate the file to query.  There
> is no equivalent of lstat() as that can be emulated with xstat() by passing
> AT_SYMLINK_NOFOLLOW in flags.
>
> AT_FORCE_ATTR_SYNC can also be set in flags.  This will require a network
> filesystem to synchronise its attributes with the server.
>
> When the system call is executed, the request_mask bitmask is read from the
> parameter block to work out what the user is requesting.  If params is NULL,
> then request_mask will be assumed to be XSTAT_REQUEST__GET_ANYWAY.

There is no XSTAT_REQUEST__GET_ANYWAY, AFAICS. I guess here you meant
XSTAT_REQUEST__EXTENDED_STATS? Or?


> The request_mask should be set by the caller to specify extra results that the
> caller may desire.  These come in a number of classes:
>
>  (0) dev, blksize.
>
>     These are local data and are always available.
>
>  (1) mode, nlinks, uid, gid, [amc]time, ino, size, blocks.
>
>     These will be returned whether the caller asks for them or not.  The
>     corresponding bits in result_mask will be set to indicate their presence.
>
>     If the caller didn't ask for them, then they may be approximated.  For
>     example, NFS won't waste any time updating them from the server, unless as
>     a byproduct of updating something requested.
>
>  (2) rdev.
>
>     As for class (1), but this won't be returned if the file is not a blockdev
>     or chardev.  The bit will be cleared if the value is not returned.
>
>  (3) File creation time, inode generation and data version.
>
>     These will be returned if available whether the caller asked for them or
>     not.  The corresponding bits in result_mask will be set or cleared as
>     appropriate to indicate their presence.
>
>     If the caller didn't ask for them, then they may be approximated.  For
>     example, NFS won't waste any time updating them from the server, unless
>     as a byproduct of updating something requested.
>
>  (4) Extra results.
>
>     These will only be returned if the caller asked for them by setting their
>     bits in request_mask.  They will be placed in the buffer after the xstat
>     struct in ascending result_mask bit order.  Any bit set in request_mask
>     mask will be left set in result_mask if the result is available and
>     cleared otherwise.
>
>     The pointer into the results list will be rounded up to the nearest 8-byte
>     boundary after each result is written in.  The size of each extra result
>     is specific to the definition for that result.
>
>     No extra results are currently defined.
>
> If the buffer is insufficiently big, the syscall returns the amount of space it
> will need to write the complete result set and returns a partial result in the
> buffer.

This case is almost certainly a user error, so why not simply return
an error (-1 and ERANGE or E2BIG)? The above approach invites
userspace errors of the form:

if (xtat(...) < 0) { /* How users often check for error */
    /* I'll handle the error */
} else {
    /* The call succeeded; I'm fine */
}

Instead, more complex error-handling is required for *every* call:

ret = xstat(..., buflen);
if (ret < 0 || ret > buflen)
    /* I'll handle the error */
} else {
    /* The call succeeded; I'm fine */
}

If you are looking for a way to inform the user about the required
buffer size, I think it would be better to take a leaf from the
getxattr(2) book: if 'buflen' is zero, then do nothing with the output
arg, but return the size that would be required.

Cheers,

Michael
--
To unsubscribe from this list: send the line "unsubscribe linux-ext4" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: [PATCH] Add a pair of system calls to make extended file stats available [ver #3]

[David Howells]

Michael Kerrisk <mtk.manpages@gmail.com> wrote:

> >        struct xstat_parameters {
> >                unsigned long long      request_mask;
> 
> Poor name, since it's a value-result arg? Better maybe something like
> "field_mask"?

No.  The contents of xstat_parameters aren't changed.  request_mask is what
you're asking for, result_mask in the xstat struct is what you actually got.

result_mask may be more or less than request_mask as the filesystem isn't
obliged to supply anything you didn't ask for, and may not be able to supply
something you did ask for, and may give you stuff anyway that you didn't ask
for if it's trivial to do so.

> There is no XSTAT_REQUEST__GET_ANYWAY, AFAICS. I guess here you meant
> XSTAT_REQUEST__EXTENDED_STATS? Or?

Yep.  I forgot to change that in the patch description.

> This case is almost certainly a user error, so why not simply return
> an error (-1 and ERANGE or E2BIG)? The above approach invites
> userspace errors of the form:
> 
> if (xtat(...) < 0) { /* How users often check for error */
>     /* I'll handle the error */
> } else {
>     /* The call succeeded; I'm fine */
> }

I suppose.

> If you are looking for a way to inform the user about the required
> buffer size, I think it would be better to take a leaf from the
> getxattr(2) book: if 'buflen' is zero, then do nothing with the output
> arg, but return the size that would be required.

That's reasonable.

David
--
To unsubscribe from this list: send the line "unsubscribe linux-ext4" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html