Skip to contentSkip to conversation

new mount API

GIT PULL In review Al Viro 8 participants 32 messages Aug 24, 2018 – Aug 29, 2018 VFS

lore
Share & tools
lore

Al Viro

author maintainer
#0 lore raw

new mount API series from Dave Howells

To quote his cover letter,
Here are a set of patches to create a filesystem context prior to setting up a new mount, populating it with the parsed options/binary data, creating the superblock and then effecting the mount. This is also used for remount since much of the parsing stuff is common in many filesystems.

This allows namespaces and other information to be conveyed through the mount procedure.

This also allows Mikl�s Szeredi's idea of doing:

fd = fsopen("nfs");
fsconfig(fd, FSCONFIG_SET_STRING, "option", "val", 0); fsconfig(fd, FSCONFIG_CMD_CREATE, NULL, NULL, 0); mfd = fsmount(fd, MS_NODEV);
move_mount(mfd, "", AT_FDCWD, "/mnt", MOVE_MOUNT_F_EMPTY_PATH);

that he presented at LSF-2017 to be implemented (see the relevant patches in the series).

I didn't use netlink as that would make the core kernel depend on CONFIG_NET and CONFIG_NETLINK and would introduce network namespacing issues.

I've implemented filesystem context handling for procfs, nfs, mqueue, cpuset, kernfs, sysfs, cgroup and afs filesystems.

Unconverted filesystems are handled by a legacy filesystem wrapper.

One trivial conflict in fs/file_table.c:__fput(); resolved as

	if (unlikely(mode & FMODE_NEED_UNMOUNT))
		dissolve_on_fput(mnt);
	dput(dentry);
	mntput(mnt);
out:
	file_free(file);

The following changes since commit 32206ab36553be8714d9253ce33f4085681d369c:

x86/intel_rdt: Provide pseudo-locking hooks within rdt_mount (2018-06-23 12:53:19 +0200)

are available in the git repository at:

git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs.git work.mount

for you to fetch changes up to 294fb3407bf29abf653fbe169af4bbb38a146db5:

proc: Set correct userns for new proc super created by a new pid_namespace (2018-08-21 09:28:43 +0100)

----------------------------------------------------------------
Al Viro (2):
vfs: syscall: Add open_tree(2) to reference or clone a mount teach move_mount(2) to work with OPEN_TREE_CLONE

Andrei Vagin (1):
proc: Set correct userns for new proc super created by a new pid_namespace

David Howells (42):
vfs: Require specification of size of mount data for internal mounts
vfs: syscall: Add move_mount(2) to move mounts around
vfs: Suppress MS_* flag defs within the kernel unless explicitly enabled
vfs: Introduce the basic header for the new mount API's filesystem context
vfs: Introduce logging functions
vfs: Add configuration parser helpers
vfs: Add LSM hooks for the new mount API
selinux: Implement the new mount API LSM hooks
smack: Implement filesystem context security hooks
apparmor: Implement security hooks for the new mount API
tomoyo: Implement security hooks for the new mount API
vfs: Separate changing mount flags full remount
vfs: Implement a filesystem superblock creation/configuration context
vfs: Remove unused code after filesystem context changes
procfs: Move proc_fill_super() to fs/proc/root.c
proc: Add fs_context support to procfs
ipc: Convert mqueue fs to fs_context
cpuset: Use fs_context
kernfs, sysfs, cgroup, intel_rdt: Support fs_context
hugetlbfs: Convert to fs_context
vfs: Remove kern_mount_data()
vfs: Provide documentation for new mount API
Make anon_inodes unconditional
vfs: syscall: Add fsopen() to prepare for superblock creation
vfs: Implement logging through fs_context
vfs: Add some logging to the core users of the fs_context log
vfs: syscall: Add fsconfig() for configuring and managing a context
vfs: syscall: Add fsmount() to create a mount for a superblock
vfs: syscall: Add fspick() to select a superblock for reconfiguration
afs: Add fs_context support
afs: Use fs_context to pass parameters over automount
vfs: Add a sample program for the new mount API
vfs: syscall: Add fsinfo() to query filesystem information
afs: Add fsinfo support
vfs: Allow fsinfo() to query what's in an fs_context
vfs: Allow fsinfo() to be used to query an fs parameter description
vfs: Implement parameter value retrieval with fsinfo()
vfs: Fix vfs_dup_fs_context()
vfs: Fix fs_context logging when there's no log
afs: Move the source fs parameter to the first position
vfs: Pass path info fsinfo and rename get_fsinfo sb op to fsinfo
vfs: Adjust fsinfo sample output

 Documentation/filesystems/mount_api.txt   | 706 ++++++++++++++++++++++++
 arch/arc/kernel/setup.c                   |   1 +
 arch/arm/kernel/atags_parse.c             |   1 +
 arch/arm/kvm/Kconfig                      |   1 -
 arch/arm64/kvm/Kconfig                    |   1 -
 arch/ia64/kernel/perfmon.c                |   3 +-
 arch/mips/kvm/Kconfig                     |   1 -
 arch/powerpc/kvm/Kconfig                  |   1 -
 arch/powerpc/platforms/cell/spufs/inode.c |   6 +-
 arch/s390/hypfs/inode.c                   |   7 +-
 arch/s390/kvm/Kconfig                     |   1 -
 arch/sh/kernel/setup.c                    |   1 +
 arch/sparc/kernel/setup_32.c              |   1 +
 arch/sparc/kernel/setup_64.c              |   1 +
 arch/x86/Kconfig                          |   1 -
 arch/x86/entry/syscalls/syscall_32.tbl    |   7 +
 arch/x86/entry/syscalls/syscall_64.tbl    |   7 +
 arch/x86/kernel/cpu/intel_rdt.h           |  15 +
 arch/x86/kernel/cpu/intel_rdt_rdtgroup.c  | 184 ++++---
 arch/x86/kernel/setup.c                   |   1 +
 arch/x86/kvm/Kconfig                      |   1 -
 drivers/base/Kconfig                      |   1 -
 drivers/base/devtmpfs.c                   |   7 +-
 drivers/char/tpm/Kconfig                  |   1 -
 drivers/dax/super.c                       |   2 +-
 drivers/dma-buf/Kconfig                   |   1 -
 drivers/gpio/Kconfig                      |   1 -
 drivers/gpu/drm/drm_drv.c                 |   3 +-
 drivers/gpu/drm/i915/i915_gemfs.c         |   2 +-
 drivers/iio/Kconfig                       |   1 -
 drivers/infiniband/Kconfig                |   1 -
 drivers/infiniband/hw/qib/qib_fs.c        |   7 +-
 drivers/misc/cxl/api.c                    |   3 +-
 drivers/misc/ibmasm/ibmasmfs.c            |  11 +-
 drivers/mtd/mtdsuper.c                    |  26 +-
 drivers/oprofile/oprofilefs.c             |   8 +-
 drivers/scsi/cxlflash/ocxl_hw.c           |   2 +-
 drivers/usb/gadget/function/f_fs.c        |   7 +-
 drivers/usb/gadget/legacy/inode.c         |   7 +-
 drivers/vfio/Kconfig                      |   1 -
 drivers/virtio/virtio_balloon.c           |   2 +-
 drivers/xen/xenfs/super.c                 |   7 +-
 fs/9p/vfs_super.c                         |   2 +-
 fs/Kconfig                                |   7 +
 fs/Makefile                               |   5 +-
 fs/adfs/super.c                           |   9 +-
 fs/affs/super.c                           |  13 +-
 fs/afs/internal.h                         |  10 +-
 fs/afs/mntpt.c                            | 147 ++---
 fs/afs/super.c                            | 634 +++++++++++++--------
 fs/afs/volume.c                           |   4 +-
 fs/aio.c                                  |   3 +-
 fs/anon_inodes.c                          |   3 +-
 fs/autofs/autofs_i.h                      |   2 +-
 fs/autofs/init.c                          |   4 +-
 fs/autofs/inode.c                         |   3 +-
 fs/befs/linuxvfs.c                        |  11 +-
 fs/bfs/inode.c                            |   8 +-
 fs/binfmt_misc.c                          |   7 +-
 fs/block_dev.c                            |   2 +-
 fs/btrfs/super.c                          |  30 +-
 fs/btrfs/tests/btrfs-tests.c              |   2 +-
 fs/ceph/super.c                           |   3 +-
 fs/cifs/cifs_dfs_ref.c                    |   3 +-
 fs/cifs/cifsfs.c                          |  18 +-
 fs/coda/inode.c                           |  11 +-
 fs/configfs/mount.c                       |   7 +-
 fs/cramfs/inode.c                         |  17 +-
 fs/debugfs/inode.c                        |  14 +-
 fs/devpts/inode.c                         |  10 +-
 fs/ecryptfs/main.c                        |   2 +-
 fs/efivarfs/super.c                       |   9 +-
 fs/efs/super.c                            |  14 +-
 fs/exofs/super.c                          |   7 +-
 fs/ext2/super.c                           |  14 +-
 fs/ext4/super.c                           |  16 +-
 fs/f2fs/super.c                           |  13 +-
 fs/fat/inode.c                            |   3 +-
 fs/fat/namei_msdos.c                      |   8 +-
 fs/fat/namei_vfat.c                       |   8 +-
 fs/file_table.c                           |   9 +-
 fs/filesystems.c                          |   4 +
 fs/freevxfs/vxfs_super.c                  |  12 +-
 fs/fs_context.c                           | 779 ++++++++++++++++++++++++++
 fs/fs_parser.c                            | 483 ++++++++++++++++
 fs/fsopen.c                               | 491 +++++++++++++++++
 fs/fuse/control.c                         |   9 +-
 fs/fuse/inode.c                           |  16 +-
 fs/gfs2/ops_fstype.c                      |   6 +-
 fs/gfs2/super.c                           |   4 +-
 fs/hfs/super.c                            |  12 +-
 fs/hfsplus/super.c                        |  12 +-
 fs/hostfs/hostfs_kern.c                   |   7 +-
 fs/hpfs/super.c                           |  11 +-
 fs/hugetlbfs/inode.c                      | 455 ++++++++++------
 fs/internal.h                             |  13 +-
 fs/isofs/inode.c                          |  11 +-
 fs/jffs2/super.c                          |  10 +-
 fs/jfs/super.c                            |  11 +-
 fs/kernfs/mount.c                         | 102 ++--
 fs/libfs.c                                |  19 +-
 fs/minix/inode.c                          |  14 +-
 fs/namei.c                                |   4 +-
 fs/namespace.c                            | 879 ++++++++++++++++++++++--------
 fs/nfs/internal.h                         |   4 +-
 fs/nfs/namespace.c                        |   3 +-
 fs/nfs/nfs4namespace.c                    |   3 +-
 fs/nfs/nfs4super.c                        |  27 +-
 fs/nfs/super.c                            |  22 +-
 fs/nfsd/nfsctl.c                          |   8 +-
 fs/nilfs2/super.c                         |  10 +-
 fs/notify/fanotify/Kconfig                |   1 -
 fs/notify/inotify/Kconfig                 |   1 -
 fs/nsfs.c                                 |   3 +-
 fs/ntfs/super.c                           |  13 +-
 fs/ocfs2/dlmfs/dlmfs.c                    |   5 +-
 fs/ocfs2/super.c                          |  14 +-
 fs/omfs/inode.c                           |   9 +-
 fs/openpromfs/inode.c                     |  11 +-
 fs/orangefs/orangefs-kernel.h             |   2 +-
 fs/orangefs/super.c                       |   5 +-
 fs/overlayfs/super.c                      |  11 +-
 fs/pipe.c                                 |   3 +-
 fs/pnode.c                                |   1 +
 fs/proc/inode.c                           |  50 +-
 fs/proc/internal.h                        |   6 +-
 fs/proc/root.c                            | 247 ++++++---
 fs/pstore/inode.c                         |  10 +-
 fs/qnx4/inode.c                           |  14 +-
 fs/qnx6/inode.c                           |  14 +-
 fs/ramfs/inode.c                          |   6 +-
 fs/reiserfs/super.c                       |  14 +-
 fs/romfs/super.c                          |  13 +-
 fs/squashfs/super.c                       |  12 +-
 fs/statfs.c                               | 571 +++++++++++++++++++
 fs/super.c                                | 397 +++++++++++---
 fs/sysfs/mount.c                          |  67 ++-
 fs/sysv/inode.c                           |   3 +-
 fs/sysv/super.c                           |  16 +-
 fs/tracefs/inode.c                        |  10 +-
 fs/ubifs/super.c                          |   5 +-
 fs/udf/super.c                            |  16 +-
 fs/ufs/super.c                            |  11 +-
 fs/xfs/xfs_super.c                        |  10 +-
 include/linux/cgroup.h                    |   3 +-
 include/linux/debugfs.h                   |   8 +-
 include/linux/fs.h                        |  50 +-
 include/linux/fs_context.h                | 208 +++++++
 include/linux/fs_parser.h                 | 117 ++++
 include/linux/fsinfo.h                    |  41 ++
 include/linux/kernfs.h                    |  41 +-
 include/linux/lsm_hooks.h                 |  79 ++-
 include/linux/module.h                    |   6 +
 include/linux/mount.h                     |  10 +-
 include/linux/mtd/super.h                 |   4 +-
 include/linux/ramfs.h                     |   4 +-
 include/linux/security.h                  |  72 ++-
 include/linux/shmem_fs.h                  |   3 +-
 include/linux/syscalls.h                  |  13 +
 include/uapi/linux/fcntl.h                |   2 +
 include/uapi/linux/fs.h                   |  82 ++-
 include/uapi/linux/fsinfo.h               | 302 ++++++++++
 include/uapi/linux/mount.h                |  75 +++
 init/Kconfig                              |  10 -
 init/do_mounts.c                          |   5 +-
 init/do_mounts_initrd.c                   |   1 +
 ipc/mqueue.c                              | 120 +++-
 kernel/bpf/inode.c                        |   7 +-
 kernel/cgroup/cgroup-internal.h           |  50 +-
 kernel/cgroup/cgroup-v1.c                 | 415 ++++++++------
 kernel/cgroup/cgroup.c                    | 285 +++++++---
 kernel/cgroup/cpuset.c                    |  67 ++-
 kernel/trace/trace.c                      |   7 +-
 mm/shmem.c                                |  10 +-
 mm/zsmalloc.c                             |   3 +-
 net/socket.c                              |   3 +-
 net/sunrpc/rpc_pipe.c                     |   7 +-
 samples/Kconfig                           |   6 +
 samples/Makefile                          |   2 +-
 samples/mount_api/Makefile                |   7 +
 samples/mount_api/test-fsmount.c          | 118 ++++
 samples/statx/Makefile                    |   7 +-
 samples/statx/test-fs-query.c             | 137 +++++
 samples/statx/test-fsinfo.c               | 584 ++++++++++++++++++++
 security/apparmor/apparmorfs.c            |   8 +-
 security/apparmor/include/mount.h         |  11 +-
 security/apparmor/lsm.c                   | 111 +++-
 security/apparmor/mount.c                 |  47 ++
 security/inode.c                          |   7 +-
 security/security.c                       |  65 ++-
 security/selinux/hooks.c                  | 320 ++++++++++-
 security/selinux/selinuxfs.c              |   8 +-
 security/smack/smack.h                    |  11 +-
 security/smack/smack_lsm.c                | 373 +++++++++++--
 security/smack/smackfs.c                  |   9 +-
 security/tomoyo/common.h                  |   3 +
 security/tomoyo/mount.c                   |  46 ++
 security/tomoyo/tomoyo.c                  |  19 +-
 198 files changed, 9288 insertions(+), 1874 deletions(-)
 create mode 100644 Documentation/filesystems/mount_api.txt
 create mode 100644 fs/fs_context.c
 create mode 100644 fs/fs_parser.c
 create mode 100644 fs/fsopen.c
 create mode 100644 include/linux/fs_context.h
 create mode 100644 include/linux/fs_parser.h
 create mode 100644 include/linux/fsinfo.h
 create mode 100644 include/uapi/linux/fsinfo.h
 create mode 100644 include/uapi/linux/mount.h
 create mode 100644 samples/mount_api/Makefile
 create mode 100644 samples/mount_api/test-fsmount.c
 create mode 100644 samples/statx/test-fs-query.c
 create mode 100644 samples/statx/test-fsinfo.c
#1 lore raw
Quoted 31 lines · Al Viro

new mount API series from Dave Howells

To quote his cover letter,
Here are a set of patches to create a filesystem context prior to setting up a new mount, populating it with the parsed options/binary data, creating the superblock and then effecting the mount. This is also used for remount since much of the parsing stuff is common in many filesystems.

This allows namespaces and other information to be conveyed through the mount procedure.

This also allows Miklós Szeredi's idea of doing:

fd = fsopen("nfs");
fsconfig(fd, FSCONFIG_SET_STRING, "option", "val", 0); fsconfig(fd, FSCONFIG_CMD_CREATE, NULL, NULL, 0); mfd = fsmount(fd, MS_NODEV);
move_mount(mfd, "", AT_FDCWD, "/mnt", MOVE_MOUNT_F_EMPTY_PATH);

that he presented at LSF-2017 to be implemented (see the relevant patches in the series).

I didn't use netlink as that would make the core kernel depend on CONFIG_NET and CONFIG_NETLINK and would introduce network namespacing issues.

I've implemented filesystem context handling for procfs, nfs, mqueue, cpuset, kernfs, sysfs, cgroup and afs filesystems.

Unconverted filesystems are handled by a legacy filesystem wrapper.

Quoted 8 lines

One trivial conflict in fs/file_table.c:__fput(); resolved as

        if (unlikely(mode & FMODE_NEED_UNMOUNT))
                dissolve_on_fput(mnt);
        dput(dentry);
        mntput(mnt);
out:
        file_free(file);

Has anything been done to ensure that the behavior when doing FSCONFIG_CMD_CREATE against an already-mounted block device is reasonable?

#2 lore raw
Quoted 3 lines · Andy Lutomirski

Has anything been done to ensure that the behavior when doing FSCONFIG_CMD_CREATE against an already-mounted block device is reasonable?

For the moment, I've left it as the same behaviour as for mount(2) since mount(2) now uses the same mechanism internally and we aren't permitted to break userspace.

I would like to add at least one flag to stipulate that, in the case of an incompatible collision, you can get a failure - but defining what is meant by incompatible isn't necessarily trivial, and would vary by filesystem *and* the LSM.

However, I don't want to start reengineering everything this close to the merge window and we don't really need it immediately.

David

#3 lore raw
Quoted 10 lines · David Howells
Quoted 3 lines · Andy Lutomirski

Has anything been done to ensure that the behavior when doing FSCONFIG_CMD_CREATE against an already-mounted block device is reasonable?

For the moment, I've left it as the same behaviour as for mount(2) since mount(2) now uses the same mechanism internally and we aren't permitted to break userspace.

Quoted 8 lines

I would like to add at least one flag to stipulate that, in the case of an incompatible collision, you can get a failure - but defining what is meant by incompatible isn't necessarily trivial, and would vary by filesystem *and* the LSM.

However, I don't want to start reengineering everything this close to the merge window and we don't really need it immediately.

The problem is that, once this ends up merged, then we're kind of stuck with it, too. It would be a bit sad if your better proposal for handling nfs and instantiation of filesystems in general were added in the next release and then we end up with the current FSCONFIG_CMD_CREATE as a permanently supported but non-preferred option.

author maintainer
#4 lore raw
Quoted 20 lines · Andy Lutomirski
Quoted 10 lines · David Howells
Quoted 3 lines · Andy Lutomirski

Has anything been done to ensure that the behavior when doing FSCONFIG_CMD_CREATE against an already-mounted block device is reasonable?

For the moment, I've left it as the same behaviour as for mount(2) since mount(2) now uses the same mechanism internally and we aren't permitted to break userspace.

Quoted 8 lines

I would like to add at least one flag to stipulate that, in the case of an incompatible collision, you can get a failure - but defining what is meant by incompatible isn't necessarily trivial, and would vary by filesystem *and* the LSM.

However, I don't want to start reengineering everything this close to the merge window and we don't really need it immediately.

Quoted 6 lines

The problem is that, once this ends up merged, then we're kind of stuck with it, too. It would be a bit sad if your better proposal for handling nfs and instantiation of filesystems in general were added in the next release and then we end up with the current FSCONFIG_CMD_CREATE as a permanently supported but non-preferred option.

For fuck sake, mount(2) is a permanently supported option! Folks, get over it - you are mixing entirely different issues. You know, I know and everyone even remotely sane knows that mount(2) *IS* *NOT* *GOING* *AWAY*. And it's not changing semantics either. So bemoaning the "permanently supported non-preferred options" is utter lunacy. It's there and it will remain there, period.

We do not break userland. And "their local scripts would break terribly inside userns container" does *NOT* render those second-class in any sense. So you can curse the current behaviour of mount(2) and I even agree with some of that, but we are not going to be able to remove that. Yes, it would've been nice if, etc., but it's not going to happen. Not now, not for many years.

#5 lore raw
Quoted 24 lines · Al Viro
Quoted 22 lines · Andy Lutomirski
Quoted 19 lines · David Howells
Quoted 3 lines · Andy Lutomirski

Has anything been done to ensure that the behavior when doing FSCONFIG_CMD_CREATE against an already-mounted block device is reasonable?

For the moment, I've left it as the same behaviour as for mount(2) since mount(2) now uses the same mechanism internally and we aren't permitted to break userspace.

I would like to add at least one flag to stipulate that, in the case of an incompatible collision, you can get a failure - but defining what is meant by incompatible isn't necessarily trivial, and would vary by filesystem *and* the LSM.

However, I don't want to start reengineering everything this close to the merge window and we don't really need it immediately.

Quoted 8 lines
Quoted 6 lines

The problem is that, once this ends up merged, then we're kind of stuck with it, too. It would be a bit sad if your better proposal for handling nfs and instantiation of filesystems in general were added in the next release and then we end up with the current FSCONFIG_CMD_CREATE as a permanently supported but non-preferred option.

For fuck sake, mount(2) is a permanently supported option!

Exactly. mount(2) has broken semantics and it’s permanently supported.

If this merge request gets pulled, then FSCONFIG_CMD_CREATE will *also* be a permanently supported API with broken semantics.

author maintainer
#6 lore raw
Quoted 5 lines · Andy Lutomirski
Quoted 1 line

For fuck sake, mount(2) is a permanently supported option!

Exactly. mount(2) has broken semantics and it’s permanently supported.

If this merge request gets pulled, then FSCONFIG_CMD_CREATE will *also* be a permanently supported API with broken semantics.

Oh no - mount(2) behaviour can be expressed that way! The horror...

Andy, this is bullshit. You are saying that dealing with mount(2) mess of ABI (badly unorthogonal set of operations, overloading from hell, etc.) must be tied with massive rework of fs drivers. Why?

It's a simple enough question and pardon me, but your "it's broken" doesn't inspire any confidence that you even understand what the current behaviour is and what its problems (they are real enough) are. I have seen a lot of handwaving from you in these threads, combined with "surely, it must work thus", followed with "it's a special case/class/whatnot" or "surely, it must be broken" every time you find something that doesn't "work thus".

Bugger it; explain why we must combine untangling the existing ABI (and we *will* have to keep the existing semantics possible to express for sys_mount() sake) with this, or with anything else, for that matter.

#7 lore raw
Quoted 6 lines · Al Viro
Quoted 4 lines · Andy Lutomirski
Quoted 1 line

For fuck sake, mount(2) is a permanently supported option!

Exactly. mount(2) has broken semantics and it’s permanently supported.

Quoted 8 lines
Quoted 1 line

If this merge request gets pulled, then FSCONFIG_CMD_CREATE will *also* be a permanently supported API with broken semantics.

Oh no - mount(2) behaviour can be expressed that way! The horror...

Andy, this is bullshit. You are saying that dealing with mount(2) mess of ABI (badly unorthogonal set of operations, overloading from hell, etc.) must be tied with massive rework of fs drivers. Why?

When this was reviewed earlier, a problem was identified. I asked if it had been addressed. I did *not* say that it was mandatory to address it, nor did I say anything about reworking fs drivers.

A reasonable answer might have been "avoiding this pitfall in the new API would involve a large amount of reworking of existing filesystem drivers. I think that the new API, as is, has enough benefits that it makes sense to merge it even with this pitfall, and, if needed, we can introduce an improved version down the road."

I had at least hoped for a better answer than "bugger it."

--Andy

#8 lore raw
Quoted 9 lines · Andy Lutomirski

When this was reviewed earlier, a problem was identified. I asked if it had been addressed. I did *not* say that it was mandatory to address it, nor did I say anything about reworking fs drivers.

A reasonable answer might have been "avoiding this pitfall in the new API would involve a large amount of reworking of existing filesystem drivers. I think that the new API, as is, has enough benefits that it makes sense to merge it even with this pitfall, and, if needed, we can introduce an improved version down the road."

It's also not clear that an API that you think is "cleaner" would actually be more usable. In fact, I believe it's going to be a sh*t show for userspace, because it won't be obvious what will work, and what will cause an error of the form, "sorry we can't do this cleaner thing that some people think is better". Which means a huge amount of special casing in the program, or a lot of very surprising failures that will then get exposed to the system administrator, many of whom haven't really had much of a problem with the existing mount(8) user interface.

It may very well be that the "cleaner" approach will be hellacious from a human usability perspective. Figuring all of this out could take months and months (stalling the new mount API for a long time), and we may never know for sure until we implement a full prototype of kernel changes, massive rewrite of the file systems, and userspace programs --- and then see what is the best way to expose the radically different semantics depending on what individual file systems can and can not do to the poor human system administrator who is just trying to mount the d*mned file system.

- Ted

#9 lore raw
Quoted 11 lines · Theodore Y. Ts'o
Quoted 9 lines · Andy Lutomirski

When this was reviewed earlier, a problem was identified. I asked if it had been addressed. I did *not* say that it was mandatory to address it, nor did I say anything about reworking fs drivers.

A reasonable answer might have been "avoiding this pitfall in the new API would involve a large amount of reworking of existing filesystem drivers. I think that the new API, as is, has enough benefits that it makes sense to merge it even with this pitfall, and, if needed, we can introduce an improved version down the road."

Quoted 9 lines

It's also not clear that an API that you think is "cleaner" would actually be more usable. In fact, I believe it's going to be a sh*t show for userspace, because it won't be obvious what will work, and what will cause an error of the form, "sorry we can't do this cleaner thing that some people think is better". Which means a huge amount of special casing in the program, or a lot of very surprising failures that will then get exposed to the system administrator, many of whom haven't really had much of a problem with the existing mount(8) user interface.

In what way is the kernel better suited to read the mind of the poor sysadmin, than a userspace helper program?

No, I don't think this argument is about the mount(8) interface, which is what the sysadmin is using and can continue to use for the foreseeable future, with basically the same feature set

It's about new uses of the mount(2) API that *would* be possible if
it was a saner interface. And so doing the new mount API radically
different than what current mount(2) is doing is quite all right in my opinion.

Quoted 9 lines

It may very well be that the "cleaner" approach will be hellacious from a human usability perspective. Figuring all of this out could take months and months (stalling the new mount API for a long time), and we may never know for sure until we implement a full prototype of kernel changes, massive rewrite of the file systems, and userspace programs --- and then see what is the best way to expose the radically different semantics depending on what individual file systems can and can not do to the poor human system administrator who is just trying to mount the d*mned file system.

I'm not against merging this patchset (have nits about resuing the MS_* constants for the new API that I've complained about, but that's really easy to fix before -final), but it may make sense to differentiate the legacy behavior from a saner one from the very start. I.e. rename FSCONFIG_CMD_CREATE to something implying it's actually *not* a create in exclusive sense that one would first imply (and yeah, we have creat() and O_CREAT, which don't imply exclusivity, yet they at least have clear semantics that current super block creation does definitely not).

Also, we don't have a massive conversion of filesystems yet to the new internal API, so we could require anything converted to support the exclusive create semantics.

None of this requires months of figuring out, just a bit of tweaking and acknowledging that the current semantics are weird, to say the least.

Thanks,
Miklos

#10 lore raw
Quoted 22 lines · Miklos Szeredi
Quoted 11 lines · Theodore Y. Ts'o
Quoted 9 lines · Andy Lutomirski

When this was reviewed earlier, a problem was identified. I asked if it had been addressed. I did *not* say that it was mandatory to address it, nor did I say anything about reworking fs drivers.

A reasonable answer might have been "avoiding this pitfall in the new API would involve a large amount of reworking of existing filesystem drivers. I think that the new API, as is, has enough benefits that it makes sense to merge it even with this pitfall, and, if needed, we can introduce an improved version down the road."

Quoted 9 lines

It's also not clear that an API that you think is "cleaner" would actually be more usable. In fact, I believe it's going to be a sh*t show for userspace, because it won't be obvious what will work, and what will cause an error of the form, "sorry we can't do this cleaner thing that some people think is better". Which means a huge amount of special casing in the program, or a lot of very surprising failures that will then get exposed to the system administrator, many of whom haven't really had much of a problem with the existing mount(8) user interface.

Quoted 2 lines

In what way is the kernel better suited to read the mind of the poor sysadmin, than a userspace helper program?

I have a concrete example: mount -oloop. You can leave it to mount(8) to automagically find an existing loop device or setup a new one, or you can do the low level thing and set up your own loop device and mount it. Same story as NFS and friends, except it's not the kernel that does the magic "same source -> same sb" policy but the
mount(8) utility. Ever notice the difference? See? And yeah, there
are races involved, and userspace is perfectly suited to deal with them.

Thanks,
Miklos

#11 lore raw
Quoted 26 lines · Miklos Szeredi
Quoted 22 lines · Miklos Szeredi
Quoted 11 lines · Theodore Y. Ts'o
Quoted 9 lines · Andy Lutomirski

When this was reviewed earlier, a problem was identified. I asked if it had been addressed. I did *not* say that it was mandatory to address it, nor did I say anything about reworking fs drivers.

A reasonable answer might have been "avoiding this pitfall in the new API would involve a large amount of reworking of existing filesystem drivers. I think that the new API, as is, has enough benefits that it makes sense to merge it even with this pitfall, and, if needed, we can introduce an improved version down the road."

Quoted 9 lines

It's also not clear that an API that you think is "cleaner" would actually be more usable. In fact, I believe it's going to be a sh*t show for userspace, because it won't be obvious what will work, and what will cause an error of the form, "sorry we can't do this cleaner thing that some people think is better". Which means a huge amount of special casing in the program, or a lot of very surprising failures that will then get exposed to the system administrator, many of whom haven't really had much of a problem with the existing mount(8) user interface.

Quoted 2 lines

In what way is the kernel better suited to read the mind of the poor sysadmin, than a userspace helper program?

Quoted 8 lines

I have a concrete example: mount -oloop. You can leave it to mount(8) to automagically find an existing loop device or setup a new one, or you can do the low level thing and set up your own loop device and mount it. Same story as NFS and friends, except it's not the kernel that does the magic "same source -> same sb" policy but the
mount(8) utility. Ever notice the difference? See? And yeah, there
are races involved, and userspace is perfectly suited to deal with them.

And to continue from that thought, a namespace for filesystem instances could, for example, live under /dev/fs/$FSTYPE/$INSTANCE

Naming the $INSTANCE could be done by the one creating that instance, or in case of legacy mounts could have a fixed prefix + sequence number, or something similar.

All this could come later, let's just not exclude the possibility of using the new API in this way.

Thanks,
Miklos

#14 lore raw
Quoted 3 lines · Miklos Szeredi

I'm not against merging this patchset (have nits about resuing the MS_* constants for the new API that I've complained about, but that's really easy to fix before -final),

Can you send me a patch that does what you want here? They're only used by fsmount() for creating a vfsmount and not used for the superblock creation or reconfiguration.

Quoted 2 lines

but it may make sense to differentiate the legacy behavior from a saner one from the very start. I.e. rename FSCONFIG_CMD_CREATE

I would suggest leaving it as-is and add an FSCONFIG_CMD_CREATE_EXCLUSIVE.

Quoted 4 lines

to something implying it's actually *not* a create in exclusive sense that one would first imply (and yeah, we have creat() and O_CREAT, which don't imply exclusivity, yet they at least have clear semantics that current super block creation does definitely not).

The problem is that "exclusivity" isn't necessarily an easy thing to define. Take nfs4 and btrfs for example. They creating a backing superblock that the actual node is derived from (though in different ways). How do you define what "exclusive" means in their case?

David

#15 lore raw
Quoted 20 lines · David Howells
Quoted 3 lines · Miklos Szeredi

I'm not against merging this patchset (have nits about resuing the MS_* constants for the new API that I've complained about, but that's really easy to fix before -final),

Can you send me a patch that does what you want here? They're only used by fsmount() for creating a vfsmount and not used for the superblock creation or reconfiguration.

Quoted 2 lines

but it may make sense to differentiate the legacy behavior from a saner one from the very start. I.e. rename FSCONFIG_CMD_CREATE

I would suggest leaving it as-is and add an FSCONFIG_CMD_CREATE_EXCLUSIVE.

Quoted 4 lines

to something implying it's actually *not* a create in exclusive sense that one would first imply (and yeah, we have creat() and O_CREAT, which don't imply exclusivity, yet they at least have clear semantics that current super block creation does definitely not).

Quoted 4 lines

The problem is that "exclusivity" isn't necessarily an easy thing to define. Take nfs4 and btrfs for example. They creating a backing superblock that the actual node is derived from (though in different ways). How do you define what "exclusive" means in their case?

Exclusive creation of a filesystem instance is just that: we promise that no one else is using that filesystem instance, and we can set options as we like, guaranteed that no bother to any other uses.

Now, with block filesystems that can only be done once, obviously, so if someone already created a btrfs instance using /dev/foobar then trying to create another instance will fail with EEXIST, EBUSY or whaterver.

For NFS it's a different story, it *is* possible to create several instances of the filesystem even if connecting to the same server, except then they will not share caches, will not have local coherency, etc..

Regardless, I think a "filesystem instance" is a pretty clear concept, that we can base a future API on and it's not just an internal implementation detail that the current mount(2) API makes look like. It *is* in fact the only concept that makes FSCONFIG_CMD_RECONFIGURE have sane meaning (i.e. I want to change *that* instance, and am fully aware what affect that will have on mounts of that instance).

Compare that to "mount -o remount path"; that one would imply we are changing something for a mount that is at "path", but if there are more than one users of that instance (which is based on magic algorithm inside the filesystem) than that might have unwanted side effects. It's what Eric's complaint is about, and it's I think the root of the issue at hand.

Thanks,
Miklos

#16 lore raw
In reply to David Howells
Quoted 9 lines · David Howells
Quoted 3 lines · Miklos Szeredi

I'm not against merging this patchset (have nits about resuing the MS_* constants for the new API that I've complained about, but that's really easy to fix before -final),

Can you send me a patch that does what you want here? They're only used by fsmount() for creating a vfsmount and not used for the superblock creation or reconfiguration.

I'm bothered by the fact that we use the same MS_ prefix in the old mount(2) api and the new fsmount(2) api. What happens if we introduce new flags for fsmount(2) and are already out of flags for mount(2)? I see a big mess that way.

Also notice, how the old code just totally ignored MS_RELATIME? Bit of rot in the new interface already.

So here's a patch. The MNT_ prefix is already used by libmount, and we don't want any confusion arising from that. So how about M_*? Short and sweet, just like O_* for open(2).

Reusing the same values as MNT_ isn't important, but it does reduce the code size a bit. Especially if we'd switch to MNT_STRICTATIME internally as well.

Thanks,
Miklos
---

fs/namespace.c +12 -18
index e34e3fd064b0..a4b8503f80b8 100644
@@ -3351,7 +3351,7 @@ EXPORT_SYMBOL_GPL(kern_mount);
  * Create a kernel mount representation for a new, prepared superblock
  * (specified by fs_fd) and attach to an open_tree-like file descriptor.
  */
-SYSCALL_DEFINE3(fsmount, int, fs_fd, unsigned int, flags, unsigned int, ms_flags)
+SYSCALL_DEFINE3(fsmount, int, fs_fd, unsigned int, flags, unsigned int, m_flags)
 {
 	struct fs_context *fc;
 	struct file *file;
@@ -3366,27 +3366,21 @@ SYSCALL_DEFINE3(fsmount, int, fs_fd, unsigned int, flags, unsigned int, ms_flags
 	if ((flags & ~(FSMOUNT_CLOEXEC)) != 0)
 		return -EINVAL;
 
-	if (ms_flags & ~(MS_RDONLY | MS_NOSUID | MS_NODEV | MS_NOEXEC |
-			 MS_NOATIME | MS_NODIRATIME | MS_RELATIME |
-			 MS_STRICTATIME))
+	if (m_flags & ~(M_NOSUID | M_NODEV | M_NOEXEC |	M_NOATIME |
+			M_NODIRATIME | M_STRICTATIME | M_RDONLY))
 		return -EINVAL;
 
-	if (ms_flags & MS_RDONLY)
-		mnt_flags |= MNT_READONLY;
-	if (ms_flags & MS_NOSUID)
-		mnt_flags |= MNT_NOSUID;
-	if (ms_flags & MS_NODEV)
-		mnt_flags |= MNT_NODEV;
-	if (ms_flags & MS_NOEXEC)
-		mnt_flags |= MNT_NOEXEC;
-	if (ms_flags & MS_NODIRATIME)
-		mnt_flags |= MNT_NODIRATIME;
+	BUILD_BUG_ON(M_NOSUID != MNT_NOSUID || M_NODEV != MNT_NODEV ||
+		     M_NOEXEC != MNT_NOEXEC || M_NOATIME != MNT_NOATIME ||
+		     M_NODIRATIME != MNT_NODIRATIME ||
+		     M_STRICTATIME != MNT_RELATIME || M_RDONLY != MNT_READONLY);
+
+	/* Relatime is the default, but internally that's what we flag */
+	mnt_flags = m_flags & ~M_STRICTATIME;
 
-	if (ms_flags & MS_STRICTATIME) {
-		if (ms_flags & MS_NOATIME)
+	if (m_flags & M_STRICTATIME) {
+		if (m_flags & M_NOATIME)
 			return -EINVAL;
-	} else if (ms_flags & MS_NOATIME) {
-		mnt_flags |= MNT_NOATIME;
 	} else {
 		mnt_flags |= MNT_RELATIME;
 	}
include/uapi/linux/mount.h +9 -0
index 3634e065836c..331e69f5a7bf 100644
@@ -1,6 +1,15 @@
 #ifndef _UAPI_LINUX_MOUNT_H
 #define _UAPI_LINUX_MOUNT_H
 
+/* Mount flags passed to fsmount(2) */
+#define M_NOSUID	0x01
+#define M_NODEV		0x02
+#define M_NOEXEC	0x04
+#define M_NOATIME	0x08
+#define M_NODIRATIME	0x10
+#define M_STRICTATIME	0x20
+#define M_RDONLY	0x40
+
 /*
  * These are the fs-independent mount-flags: up to 32 flags are supported
  *
#17 lore raw

include/uapi/linux/mount.h line 11

Quoted 8 lines · Miklos Szeredi
include/uapi/linux/mount.h +8 -0
+/* Mount flags passed to fsmount(2) */
+#define M_NOSUID	0x01
+#define M_NODEV		0x02
+#define M_NOEXEC	0x04
+#define M_NOATIME	0x08
+#define M_NODIRATIME	0x10
+#define M_STRICTATIME	0x20
+#define M_RDONLY	0x40

If we're going to do this, I would suggest a longer prefix than just 'M' and renumber them to put *_RDONLY first.

fs/namespace.c SYSCALL_DEFINE3(fsmount, int, fs_fd, unsigned int, flags, unsigned int, ms_flags · line 3376

Quoted 4 lines
fs/namespace.c +4 -0
+	BUILD_BUG_ON(M_NOSUID != MNT_NOSUID || M_NODEV != MNT_NODEV ||
+		     M_NOEXEC != MNT_NOEXEC || M_NOATIME != MNT_NOATIME ||
+		     M_NODIRATIME != MNT_NODIRATIME ||
+		     M_STRICTATIME != MNT_RELATIME || M_RDONLY != MNT_READONLY);

Please don't, please do:

	if (ms_flags & M_RDONLY)
		mnt_flags |= MNT_READONLY;

Yes, and at some point I'd also like to compress the numbering on the SB_* constants and break the identity as Christoph suggested.

David

#19 lore raw
In reply to Miklos Szeredi
Quoted 11 lines · Miklos Szeredi

I'm bothered by the fact that we use the same MS_ prefix in the old mount(2) api and the new fsmount(2) api. What happens if we introduce new flags for fsmount(2) and are already out of flags for mount(2)? I see a big mess that way.

Also notice, how the old code just totally ignored MS_RELATIME? Bit of rot in the new interface already.

So here's a patch. The MNT_ prefix is already used by libmount, and we don't want any confusion arising from that. So how about M_*? Short and sweet, just like O_* for open(2).

+1 Please.

Signature · 2 lines
 Karel Zak  <[email protected]>
 http://karelzak.blogspot.com
#18 lore raw
In reply to David Howells
Quoted 12 lines · David Howells

include/uapi/linux/mount.h line 11

Quoted 8 lines · Miklos Szeredi
include/uapi/linux/mount.h +8 -0
+/* Mount flags passed to fsmount(2) */
+#define M_NOSUID     0x01
+#define M_NODEV              0x02
+#define M_NOEXEC     0x04
+#define M_NOATIME    0x08
+#define M_NODIRATIME 0x10
+#define M_STRICTATIME        0x20
+#define M_RDONLY     0x40

If we're going to do this, I would suggest a longer prefix than just 'M'

Any suggestions?

We could do "MOUNT_", but that sounds tad too long. But hey, that
doesn't matter much either. Let me know your preference.

Quoted 8 lines

and
renumber them to put *_RDONLY first.

fs/namespace.c SYSCALL_DEFINE3(fsmount, int, fs_fd, unsigned int, flags, unsigned int, ms_flags · line 3376

Quoted 4 lines
fs/namespace.c +4 -0
+     BUILD_BUG_ON(M_NOSUID != MNT_NOSUID || M_NODEV != MNT_NODEV ||
+                  M_NOEXEC != MNT_NOEXEC || M_NOATIME != MNT_NOATIME ||
+                  M_NODIRATIME != MNT_NODIRATIME ||
+                  M_STRICTATIME != MNT_RELATIME || M_RDONLY != MNT_READONLY);
Quoted 4 lines

Please don't, please do:

if (ms_flags & M_RDONLY)
mnt_flags |= MNT_READONLY;

As I said, I don't really care either way.

Thanks,
Miklos

#20 lore raw
In reply to David Howells
Quoted 11 lines · David Howells

The problem is that "exclusivity" isn't necessarily an easy thing to define. Take nfs4 and btrfs for example. They creating a backing superblock that the actual node is derived from (though in different ways). How do you define what "exclusive" means in their case?

I would argue that “exclusive” means “semantically equivalent to getting a fully independent instance.” This means, at least, that all options need to be respected as specified (or fail if they can’t be) and future reconfigure/remount actions only affect *this* instance and not others.

NFS might already meet these requirements, although I’m not sure about the latter one.

But it might be that adding CREATE_EXCLUSIVE is not all that useful and it would be better to have an API like you sketched where you first explicitly instantiate the driver and then you create however many trees (a la fsmount()) you want, where each tree has its own set of per-tree options like “subvol” for btrfs.

#21 lore raw
Quoted 6 lines · Andy Lutomirski
Quoted 6 lines · David Howells
Quoted 7 lines
Quoted 5 lines

The problem is that "exclusivity" isn't necessarily an easy thing to define. Take nfs4 and btrfs for example. They creating a backing superblock that the actual node is derived from (though in different ways). How do you define what "exclusive" means in their case?

I would argue that “exclusive” means “semantically equivalent to getting a fully independent instance.”

The only way to get semantically equivalent instance is to create a fully independent instance. See reconfigure (nee remount).

Thanks,
Miklos

#22 lore raw
Quoted 16 lines · Miklos Szeredi
Quoted 14 lines · Andy Lutomirski
Quoted 11 lines
Quoted 3 lines · David Howells

The problem is that "exclusivity" isn't necessarily an easy thing to define. Take nfs4 and btrfs for example. They creating a backing superblock that the actual node is derived from (though in different ways). How do you define what "exclusive" means in their case?

Quoted 6 lines
Quoted 1 line

I would argue that “exclusive” means “semantically equivalent to getting a fully independent instance.”

The only way to get semantically equivalent instance is to create a fully independent instance. See reconfigure (nee remount).

Hmm. Is it that case in the current patchset that you can do CMD_CREATE and reconfigure the result and some *other* existing mount will change? If so, that’s rather unfriendly to users.

#23 lore raw
Quoted 16 lines · Andy Lutomirski
Quoted 16 lines · Miklos Szeredi
Quoted 14 lines · Andy Lutomirski
Quoted 11 lines
Quoted 3 lines · David Howells

The problem is that "exclusivity" isn't necessarily an easy thing to define. Take nfs4 and btrfs for example. They creating a backing superblock that the actual node is derived from (though in different ways). How do you define what "exclusive" means in their case?

Quoted 8 lines
Quoted 6 lines
Quoted 1 line

I would argue that “exclusive” means “semantically equivalent to getting a fully independent instance.”

The only way to get semantically equivalent instance is to create a fully independent instance. See reconfigure (nee remount).

Hmm. Is it that case in the current patchset that you can do CMD_CREATE and reconfigure the result and some *other* existing mount will change? If so, that’s rather unfriendly to users.

Exactly.

Thanks,
Miklos

#24 lore raw
In reply to Andy Lutomirski
Quoted 3 lines · Andy Lutomirski

Hmm. Is it that case in the current patchset that you can do CMD_CREATE and reconfigure the result and some *other* existing mount will change? If so, that’s rather unfriendly to users.

The default behaviour has to be the same as mount(2).

David

#25 lore raw
Quoted 8 lines · David Howells
Quoted 3 lines · Andy Lutomirski

Hmm. Is it that case in the current patchset that you can do CMD_CREATE and reconfigure the result and some *other* existing mount will change? If so, that’s rather unfriendly to users.

The default behaviour has to be the same as mount(2).

Why?

Obviously mount(2) needs to keep working the way it does now. Most likely mount(8) also needs to retain its current behavior. But I don't see why the new API needs be bug-compatible with mount(2).

--Andy

#26 lore raw
Quoted 8 lines · Andy Lutomirski
Quoted 7 lines · David Howells
Quoted 3 lines · Andy Lutomirski

Hmm. Is it that case in the current patchset that you can do CMD_CREATE and reconfigure the result and some *other* existing mount will change? If so, that’s rather unfriendly to users.

The default behaviour has to be the same as mount(2).

This is rubbish. Anyone wanting the mount(2) behavior can use mount(2).

About exclusive create: can't we just look at the active reference count of the superblock returned by ->get_tree() (if it's one, we are the only users, i.e. the create was exclusive)?

Thanks,
Miklos

author maintainer
#27 lore raw
Quoted 8 lines · Miklos Szeredi
Quoted 7 lines · Andy Lutomirski
Quoted 6 lines · David Howells
Quoted 3 lines · Andy Lutomirski

Hmm. Is it that case in the current patchset that you can do CMD_CREATE and reconfigure the result and some *other* existing mount will change? If so, that’s rather unfriendly to users.

Quoted 7 lines
Quoted 1 line
Quoted 1 line

The default behaviour has to be the same as mount(2).

This is rubbish. Anyone wanting the mount(2) behavior can use mount(2).

About exclusive create: can't we just look at the active reference count of the superblock returned by ->get_tree() (if it's one, we are the only users, i.e. the create was exclusive)?

Let me get it straight - your default behaviour would routinely refuse NFS mount simply because somebody has already mounted from the same server?

The main reason for keeping existing semantics is very, very simple: nobody has offered a sane replacement. For all warts (and I admit that policy re sharing turned out to be rather bad for any kind of situation with non-cooperative admins - in partial defense, back in 2000/2001 when it was done anybody talking about something like userns would've gotten laughed at, for a lot of good reasons) mount(2) semantics is defined *and* needs to be supported anyway.

All suggested "better replacements" were bloody problematic. Sure, we'll need to sort that out, but again, why the hell tie that to untangling the sodding mount(2) ABI?

#29 lore raw

Sigh.

I just noticed that this tree reintroduces bugs that were fixed with cfb2f6f6e0ba ("Revert "mqueue: switch to on-demand creation of internal mount"")

So in addition to ongoing discussion about what the new mount api should look like we have bugs in the implementation. This really does not look like this branch is ready to be merged yet.

Eric

#30 lore raw
Quoted 6 lines · Eric W. Biederman

I just noticed that this tree reintroduces bugs that were fixed with cfb2f6f6e0ba ("Revert "mqueue: switch to on-demand creation of internal mount"")

So in addition to ongoing discussion about what the new mount api should look like ...

Which don't need to be solved *before* the current patches go in. No strong reason has been given why the desired changes cannot wait one cycle.

The core needs to be upstream before we can start porting most of the filesystems to it.

Quoted 1 line

... we have bugs in the implementation.

Allegedly. It would be useful if you could've pointed out what it was that you think you see. I don't see offhand what it reintroduces.

And how come this hasn't been noticed since the patches have been sat in linux-next for a while?

Quoted 1 line

This really does not look like this branch is ready to be merged yet.

I disagree (probably obviously).

David

#31 lore raw
Quoted 4 lines · David Howells
Quoted 1 line

... we have bugs in the implementation.

Allegedly. It would be useful if you could've pointed out what it was that you think you see. I don't see offhand what it reintroduces.

This?

- ns->mq_mnt = kern_mount_data(&mqueue_fs_type, ns, 0);
- if (IS_ERR(ns->mq_mnt)) {
- int err = PTR_ERR(ns->mq_mnt);
- ns->mq_mnt = NULL;
- return err;
- }
+ m = mq_create_mount(&init_ipc_ns);
+ if (IS_ERR(m))
+ return PTR_ERR(m);
+ ns->mq_mnt = m;

Should be passing ns into mq_create_mount() rather than init_ipc_ns maybe?

David

#12 lore raw

PATCH 1/1 mqueue: Fix bug from mount API conversion

In reply to Al Viro

Hi Al, Linus,

I *think* the attached patch is needed with the new mount API.

David
---
mqueue: Fix bug from mount API conversion

Fix a possible bug from the mount API conversion of the mqueue filesystem whereby mq_init_ns() passes the root IPC namespace to mq_create_mount() rather than the namespace passed in.

I don't know if this is the bug Eric is reporting since he didn't give any details.

---

ipc/mqueue.c +1 -1
index 0f102210f89e..4fbbd6f5b588 100644
@@ -1618,7 +1618,7 @@ int mq_init_ns(struct ipc_namespace *ns)
 	ns->mq_msg_default   = DFLT_MSG;
 	ns->mq_msgsize_default  = DFLT_MSGSIZE;
 
-	m = mq_create_mount(&init_ipc_ns);
+	m = mq_create_mount(ns);
 	if (IS_ERR(m))
 		return PTR_ERR(m);
 	ns->mq_mnt = m;
author maintainer
#13 lore raw

Re: [PATCH] mqueue: Fix bug from mount API conversion

Quoted 8 lines · David Howells

Hi Al, Linus,

I *think* the attached patch is needed with the new mount API.

David
---
mqueue: Fix bug from mount API conversion

Quoted 6 lines

Fix a possible bug from the mount API conversion of the mqueue filesystem whereby mq_init_ns() passes the root IPC namespace to mq_create_mount() rather than the namespace passed in.

I don't know if this is the bug Eric is reporting since he didn't give any details.

Applied and pushed.

#28 lore raw
In reply to Al Viro
Quoted 11 lines · Al Viro
Quoted 10 lines · Miklos Szeredi
Quoted 8 lines · Andy Lutomirski
Quoted 7 lines · David Howells
Quoted 3 lines · Andy Lutomirski

Hmm. Is it that case in the current patchset that you can do CMD_CREATE and reconfigure the result and some *other* existing mount will change? If so, that’s rather unfriendly to users.

The default behaviour has to be the same as mount(2).

Quoted 8 lines
Quoted 5 lines

This is rubbish. Anyone wanting the mount(2) behavior can use mount(2).

About exclusive create: can't we just look at the active reference count of the superblock returned by ->get_tree() (if it's one, we are the only users, i.e. the create was exclusive)?

Let me get it straight - your default behaviour would routinely refuse NFS mount simply because somebody has already mounted from the same server?

Yeah, bad idea.

Quoted 11 lines

The main reason for keeping existing semantics is very, very simple: nobody has offered a sane replacement. For all warts (and I admit that policy re sharing turned out to be rather bad for any kind of situation with non-cooperative admins - in partial defense, back in 2000/2001 when it was done anybody talking about something like userns would've gotten laughed at, for a lot of good reasons) mount(2) semantics is defined *and* needs to be supported anyway.

All suggested "better replacements" were bloody problematic. Sure, we'll need to sort that out, but again, why the hell tie that to untangling the sodding mount(2) ABI?

What I was primarily suggesting is pretty simple, yet nobody seems to get it: name the current behavior what it is: legacy. Call this op e.g. CMD_LEGACY_FIND_OR_CREATE. It means "you're using the shiny new API, but be warned: we'll still use the old and stinky way to get the super block".

We'll hopefully untangle it, sooner rather than later, and then have ops with less scary names.

Thanks,
Miklos