411 lines · plain
1.. SPDX-License-Identifier: GPL-2.02 3====================================================================4Miscellaneous Device control operations for the autofs kernel module5====================================================================6 7The problem8===========9 10There is a problem with active restarts in autofs (that is to say11restarting autofs when there are busy mounts).12 13During normal operation autofs uses a file descriptor opened on the14directory that is being managed in order to be able to issue control15operations. Using a file descriptor gives ioctl operations access to16autofs specific information stored in the super block. The operations17are things such as setting an autofs mount catatonic, setting the18expire timeout and requesting expire checks. As is explained below,19certain types of autofs triggered mounts can end up covering an autofs20mount itself which prevents us being able to use open(2) to obtain a21file descriptor for these operations if we don't already have one open.22 23Currently autofs uses "umount -l" (lazy umount) to clear active mounts24at restart. While using lazy umount works for most cases, anything that25needs to walk back up the mount tree to construct a path, such as26getcwd(2) and the proc file system /proc/<pid>/cwd, no longer works27because the point from which the path is constructed has been detached28from the mount tree.29 30The actual problem with autofs is that it can't reconnect to existing31mounts. Immediately one thinks of just adding the ability to remount32autofs file systems would solve it, but alas, that can't work. This is33because autofs direct mounts and the implementation of "on demand mount34and expire" of nested mount trees have the file system mounted directly35on top of the mount trigger directory dentry.36 37For example, there are two types of automount maps, direct (in the kernel38module source you will see a third type called an offset, which is just39a direct mount in disguise) and indirect.40 41Here is a master map with direct and indirect map entries::42 43 /- /etc/auto.direct44 /test /etc/auto.indirect45 46and the corresponding map files::47 48 /etc/auto.direct:49 50 /automount/dparse/g6 budgie:/autofs/export151 /automount/dparse/g1 shark:/autofs/export152 and so on.53 54/etc/auto.indirect::55 56 g1 shark:/autofs/export157 g6 budgie:/autofs/export158 and so on.59 60For the above indirect map an autofs file system is mounted on /test and61mounts are triggered for each sub-directory key by the inode lookup62operation. So we see a mount of shark:/autofs/export1 on /test/g1, for63example.64 65The way that direct mounts are handled is by making an autofs mount on66each full path, such as /automount/dparse/g1, and using it as a mount67trigger. So when we walk on the path we mount shark:/autofs/export1 "on68top of this mount point". Since these are always directories we can69use the follow_link inode operation to trigger the mount.70 71But, each entry in direct and indirect maps can have offsets (making72them multi-mount map entries).73 74For example, an indirect mount map entry could also be::75 76 g1 \77 / shark:/autofs/export5/testing/test \78 /s1 shark:/autofs/export/testing/test/s1 \79 /s2 shark:/autofs/export5/testing/test/s2 \80 /s1/ss1 shark:/autofs/export1 \81 /s2/ss2 shark:/autofs/export282 83and a similarly a direct mount map entry could also be::84 85 /automount/dparse/g1 \86 / shark:/autofs/export5/testing/test \87 /s1 shark:/autofs/export/testing/test/s1 \88 /s2 shark:/autofs/export5/testing/test/s2 \89 /s1/ss1 shark:/autofs/export2 \90 /s2/ss2 shark:/autofs/export291 92One of the issues with version 4 of autofs was that, when mounting an93entry with a large number of offsets, possibly with nesting, we needed94to mount and umount all of the offsets as a single unit. Not really a95problem, except for people with a large number of offsets in map entries.96This mechanism is used for the well known "hosts" map and we have seen97cases (in 2.4) where the available number of mounts are exhausted or98where the number of privileged ports available is exhausted.99 100In version 5 we mount only as we go down the tree of offsets and101similarly for expiring them which resolves the above problem. There is102somewhat more detail to the implementation but it isn't needed for the103sake of the problem explanation. The one important detail is that these104offsets are implemented using the same mechanism as the direct mounts105above and so the mount points can be covered by a mount.106 107The current autofs implementation uses an ioctl file descriptor opened108on the mount point for control operations. The references held by the109descriptor are accounted for in checks made to determine if a mount is110in use and is also used to access autofs file system information held111in the mount super block. So the use of a file handle needs to be112retained.113 114 115The Solution116============117 118To be able to restart autofs leaving existing direct, indirect and119offset mounts in place we need to be able to obtain a file handle120for these potentially covered autofs mount points. Rather than just121implement an isolated operation it was decided to re-implement the122existing ioctl interface and add new operations to provide this123functionality.124 125In addition, to be able to reconstruct a mount tree that has busy mounts,126the uid and gid of the last user that triggered the mount needs to be127available because these can be used as macro substitution variables in128autofs maps. They are recorded at mount request time and an operation129has been added to retrieve them.130 131Since we're re-implementing the control interface, a couple of other132problems with the existing interface have been addressed. First, when133a mount or expire operation completes a status is returned to the134kernel by either a "send ready" or a "send fail" operation. The135"send fail" operation of the ioctl interface could only ever send136ENOENT so the re-implementation allows user space to send an actual137status. Another expensive operation in user space, for those using138very large maps, is discovering if a mount is present. Usually this139involves scanning /proc/mounts and since it needs to be done quite140often it can introduce significant overhead when there are many entries141in the mount table. An operation to lookup the mount status of a mount142point dentry (covered or not) has also been added.143 144Current kernel development policy recommends avoiding the use of the145ioctl mechanism in favor of systems such as Netlink. An implementation146using this system was attempted to evaluate its suitability and it was147found to be inadequate, in this case. The Generic Netlink system was148used for this as raw Netlink would lead to a significant increase in149complexity. There's no question that the Generic Netlink system is an150elegant solution for common case ioctl functions but it's not a complete151replacement probably because its primary purpose in life is to be a152message bus implementation rather than specifically an ioctl replacement.153While it would be possible to work around this there is one concern154that lead to the decision to not use it. This is that the autofs155expire in the daemon has become far to complex because umount156candidates are enumerated, almost for no other reason than to "count"157the number of times to call the expire ioctl. This involves scanning158the mount table which has proved to be a big overhead for users with159large maps. The best way to improve this is try and get back to the160way the expire was done long ago. That is, when an expire request is161issued for a mount (file handle) we should continually call back to162the daemon until we can't umount any more mounts, then return the163appropriate status to the daemon. At the moment we just expire one164mount at a time. A Generic Netlink implementation would exclude this165possibility for future development due to the requirements of the166message bus architecture.167 168 169autofs Miscellaneous Device mount control interface170====================================================171 172The control interface is opening a device node, typically /dev/autofs.173 174All the ioctls use a common structure to pass the needed parameter175information and return operation results::176 177 struct autofs_dev_ioctl {178 __u32 ver_major;179 __u32 ver_minor;180 __u32 size; /* total size of data passed in181 * including this struct */182 __s32 ioctlfd; /* automount command fd */183 184 /* Command parameters */185 union {186 struct args_protover protover;187 struct args_protosubver protosubver;188 struct args_openmount openmount;189 struct args_ready ready;190 struct args_fail fail;191 struct args_setpipefd setpipefd;192 struct args_timeout timeout;193 struct args_requester requester;194 struct args_expire expire;195 struct args_askumount askumount;196 struct args_ismountpoint ismountpoint;197 };198 199 char path[];200 };201 202The ioctlfd field is a mount point file descriptor of an autofs mount203point. It is returned by the open call and is used by all calls except204the check for whether a given path is a mount point, where it may205optionally be used to check a specific mount corresponding to a given206mount point file descriptor, and when requesting the uid and gid of the207last successful mount on a directory within the autofs file system.208 209The union is used to communicate parameters and results of calls made210as described below.211 212The path field is used to pass a path where it is needed and the size field213is used account for the increased structure length when translating the214structure sent from user space.215 216This structure can be initialized before setting specific fields by using217the void function call init_autofs_dev_ioctl(``struct autofs_dev_ioctl *``).218 219All of the ioctls perform a copy of this structure from user space to220kernel space and return -EINVAL if the size parameter is smaller than221the structure size itself, -ENOMEM if the kernel memory allocation fails222or -EFAULT if the copy itself fails. Other checks include a version check223of the compiled in user space version against the module version and a224mismatch results in a -EINVAL return. If the size field is greater than225the structure size then a path is assumed to be present and is checked to226ensure it begins with a "/" and is NULL terminated, otherwise -EINVAL is227returned. Following these checks, for all ioctl commands except228AUTOFS_DEV_IOCTL_VERSION_CMD, AUTOFS_DEV_IOCTL_OPENMOUNT_CMD and229AUTOFS_DEV_IOCTL_CLOSEMOUNT_CMD the ioctlfd is validated and if it is230not a valid descriptor or doesn't correspond to an autofs mount point231an error of -EBADF, -ENOTTY or -EINVAL (not an autofs descriptor) is232returned.233 234 235The ioctls236==========237 238An example of an implementation which uses this interface can be seen239in autofs version 5.0.4 and later in file lib/dev-ioctl-lib.c of the240distribution tar available for download from kernel.org in directory241/pub/linux/daemons/autofs/v5.242 243The device node ioctl operations implemented by this interface are:244 245 246AUTOFS_DEV_IOCTL_VERSION247------------------------248 249Get the major and minor version of the autofs device ioctl kernel module250implementation. It requires an initialized struct autofs_dev_ioctl as an251input parameter and sets the version information in the passed in structure.252It returns 0 on success or the error -EINVAL if a version mismatch is253detected.254 255 256AUTOFS_DEV_IOCTL_PROTOVER_CMD and AUTOFS_DEV_IOCTL_PROTOSUBVER_CMD257------------------------------------------------------------------258 259Get the major and minor version of the autofs protocol version understood260by loaded module. This call requires an initialized struct autofs_dev_ioctl261with the ioctlfd field set to a valid autofs mount point descriptor262and sets the requested version number in version field of struct args_protover263or sub_version field of struct args_protosubver. These commands return2640 on success or one of the negative error codes if validation fails.265 266 267AUTOFS_DEV_IOCTL_OPENMOUNT and AUTOFS_DEV_IOCTL_CLOSEMOUNT268----------------------------------------------------------269 270Obtain and release a file descriptor for an autofs managed mount point271path. The open call requires an initialized struct autofs_dev_ioctl with272the path field set and the size field adjusted appropriately as well273as the devid field of struct args_openmount set to the device number of274the autofs mount. The device number can be obtained from the mount options275shown in /proc/mounts. The close call requires an initialized struct276autofs_dev_ioct with the ioctlfd field set to the descriptor obtained277from the open call. The release of the file descriptor can also be done278with close(2) so any open descriptors will also be closed at process exit.279The close call is included in the implemented operations largely for280completeness and to provide for a consistent user space implementation.281 282 283AUTOFS_DEV_IOCTL_READY_CMD and AUTOFS_DEV_IOCTL_FAIL_CMD284--------------------------------------------------------285 286Return mount and expire result status from user space to the kernel.287Both of these calls require an initialized struct autofs_dev_ioctl288with the ioctlfd field set to the descriptor obtained from the open289call and the token field of struct args_ready or struct args_fail set290to the wait queue token number, received by user space in the foregoing291mount or expire request. The status field of struct args_fail is set to292the errno of the operation. It is set to 0 on success.293 294 295AUTOFS_DEV_IOCTL_SETPIPEFD_CMD296------------------------------297 298Set the pipe file descriptor used for kernel communication to the daemon.299Normally this is set at mount time using an option but when reconnecting300to a existing mount we need to use this to tell the autofs mount about301the new kernel pipe descriptor. In order to protect mounts against302incorrectly setting the pipe descriptor we also require that the autofs303mount be catatonic (see next call).304 305The call requires an initialized struct autofs_dev_ioctl with the306ioctlfd field set to the descriptor obtained from the open call and307the pipefd field of struct args_setpipefd set to descriptor of the pipe.308On success the call also sets the process group id used to identify the309controlling process (eg. the owning automount(8) daemon) to the process310group of the caller.311 312 313AUTOFS_DEV_IOCTL_CATATONIC_CMD314------------------------------315 316Make the autofs mount point catatonic. The autofs mount will no longer317issue mount requests, the kernel communication pipe descriptor is released318and any remaining waits in the queue released.319 320The call requires an initialized struct autofs_dev_ioctl with the321ioctlfd field set to the descriptor obtained from the open call.322 323 324AUTOFS_DEV_IOCTL_TIMEOUT_CMD325----------------------------326 327Set the expire timeout for mounts within an autofs mount point.328 329The call requires an initialized struct autofs_dev_ioctl with the330ioctlfd field set to the descriptor obtained from the open call.331 332 333AUTOFS_DEV_IOCTL_REQUESTER_CMD334------------------------------335 336Return the uid and gid of the last process to successfully trigger a the337mount on the given path dentry.338 339The call requires an initialized struct autofs_dev_ioctl with the path340field set to the mount point in question and the size field adjusted341appropriately. Upon return the uid field of struct args_requester contains342the uid and gid field the gid.343 344When reconstructing an autofs mount tree with active mounts we need to345re-connect to mounts that may have used the original process uid and346gid (or string variations of them) for mount lookups within the map entry.347This call provides the ability to obtain this uid and gid so they may be348used by user space for the mount map lookups.349 350 351AUTOFS_DEV_IOCTL_EXPIRE_CMD352---------------------------353 354Issue an expire request to the kernel for an autofs mount. Typically355this ioctl is called until no further expire candidates are found.356 357The call requires an initialized struct autofs_dev_ioctl with the358ioctlfd field set to the descriptor obtained from the open call. In359addition an immediate expire that's independent of the mount timeout,360and a forced expire that's independent of whether the mount is busy,361can be requested by setting the how field of struct args_expire to362AUTOFS_EXP_IMMEDIATE or AUTOFS_EXP_FORCED, respectively . If no363expire candidates can be found the ioctl returns -1 with errno set to364EAGAIN.365 366This call causes the kernel module to check the mount corresponding367to the given ioctlfd for mounts that can be expired, issues an expire368request back to the daemon and waits for completion.369 370AUTOFS_DEV_IOCTL_ASKUMOUNT_CMD371------------------------------372 373Checks if an autofs mount point is in use.374 375The call requires an initialized struct autofs_dev_ioctl with the376ioctlfd field set to the descriptor obtained from the open call and377it returns the result in the may_umount field of struct args_askumount,3781 for busy and 0 otherwise.379 380 381AUTOFS_DEV_IOCTL_ISMOUNTPOINT_CMD382---------------------------------383 384Check if the given path is a mountpoint.385 386The call requires an initialized struct autofs_dev_ioctl. There are two387possible variations. Both use the path field set to the path of the mount388point to check and the size field adjusted appropriately. One uses the389ioctlfd field to identify a specific mount point to check while the other390variation uses the path and optionally in.type field of struct args_ismountpoint391set to an autofs mount type. The call returns 1 if this is a mount point392and sets out.devid field to the device number of the mount and out.magic393field to the relevant super block magic number (described below) or 0 if394it isn't a mountpoint. In both cases the device number (as returned395by new_encode_dev()) is returned in out.devid field.396 397If supplied with a file descriptor we're looking for a specific mount,398not necessarily at the top of the mounted stack. In this case the path399the descriptor corresponds to is considered a mountpoint if it is itself400a mountpoint or contains a mount, such as a multi-mount without a root401mount. In this case we return 1 if the descriptor corresponds to a mount402point and also returns the super magic of the covering mount if there403is one or 0 if it isn't a mountpoint.404 405If a path is supplied (and the ioctlfd field is set to -1) then the path406is looked up and is checked to see if it is the root of a mount. If a407type is also given we are looking for a particular autofs mount and if408a match isn't found a fail is returned. If the located path is the409root of a mount 1 is returned along with the super magic of the mount410or 0 otherwise.411