Add mount rules
Add the ability to control mounting and unmounting
The basic form of the rules are.
[audit] [deny] mount [conds]* [device] [ -> [conds] path],
[audit] [deny] remount [conds]* [path],
[audit] [deny] umount [conds]* [path],
[audit] [deny] pivotroot [oldroot=<value>] <path> -> <profile>
remount is just a short cut for mount options=remount
where [conds] can be
fstype=<expr>
options=<expr>
conds follow the extended conditional syntax of allowing either:
* a single value after the equals, which has the same character range as
regular IDS (ie most anything but it can't be terminated with a , (comma)
and if spaces or other characters are needed it can be quoted
eg.
options=foo
options = foo
options="foo bar"
* a list of values after the equals, the list of values is enclosed within
parenthesis () and its has a slightly reduced character set but again
elements can be quoted.
the separation between elements is whitespace and commas.
eg.
options=(foo bar)
options=(foo, bar)
options=(foo , bar)
options=(foo,bar)
The rules are flexible and follow a similar pattern as network, capability,
etc.
mount, # allow all mounts, but not umount or pivotroot
mount fstype=procfs, # allow mounting procfs anywhere
mount options=(bind, ro) /foo -> /bar, # readonly bind mount
mount /dev/sda -> /mnt,
mount /dev/sd** -> /mnt/**,
mount fstype=overlayfs options=(rw,upperdir=/tmp/upper/,lowerdir=/) overlay -> /mnt/
umount,
umount /m*,
Currently variables and regexs are are supported on the device and mount
point. ie.
mount <devince> -> <mount point>,
Regexes are supported in fstype and options. The options have a further
caveat that regexs only work if the option is fs specific option.
eg. options=(upperdir=/tmp/*,lowerdir=/)
regex's will not currently work against the standard options like ro, rw
nosuid
Conditionals (fstype) can only be applied to the device (source) at this
time and will be disregarded in situations where the mount is manipulating
an existing mount (bind, remount).
Options can be specified multiple times
mount option=rw option=(nosuid,upperdir=/foo),
and will be combined together into a single set of values
The ordering of the standard mount options (rw,ro, ...) does not matter
but the ordering of fs specific options does.
Specifying that the value of a particular option does not matter can be
acheived by providing both the positive and negative forms of and option
option=(rw,ro) options=(suid,nosuid)
For the fs specific options specifying that a particular value does not
matter is achieve using a regex with alternations.
Improvements to the syntax and order restrictions are planned for the
future.
Signed-off-by: John Johansen <john.johansen@canonical.com>
2012-02-24 04:19:38 -08:00
|
|
|
/*
|
|
|
|
* Copyright (c) 2010
|
|
|
|
* Canonical, Ltd. (All rights reserved)
|
|
|
|
*
|
|
|
|
* This program is free software; you can redistribute it and/or
|
|
|
|
* modify it under the terms of version 2 of the GNU General Public
|
|
|
|
* License published by the Free Software Foundation.
|
|
|
|
*
|
|
|
|
* This program is distributed in the hope that it will be useful,
|
|
|
|
* but WITHOUT ANY WARRANTY; without even the implied warranty of
|
|
|
|
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
|
|
|
|
* GNU General Public License for more details.
|
|
|
|
*
|
|
|
|
* You should have received a copy of the GNU General Public License
|
|
|
|
* along with this program; if not, contact Novell, Inc. or Canonical
|
|
|
|
* Ltd.
|
|
|
|
*/
|
|
|
|
|
|
|
|
/**
|
|
|
|
* The mount command, its mix of options and flags, its permissions and
|
|
|
|
* mapping are a mess.
|
|
|
|
* mount [-lhV]
|
|
|
|
*
|
|
|
|
* mount -a [-fFnrsvw] [-t vfstype] [-O optlist]
|
|
|
|
*
|
|
|
|
* mount [-fnrsvw] [-o option[,option]...] device|dir
|
|
|
|
*
|
|
|
|
* mount [-fnrsvw] [-t vfstype] [-o options] device dir
|
|
|
|
*
|
|
|
|
*----------------------------------------------------------------------
|
|
|
|
* Mount flags of no interest for apparmor mediation
|
|
|
|
* -a, --all
|
|
|
|
* -F fork for simultaneous mount
|
|
|
|
* -f fake, do everything except that actual system call
|
|
|
|
* -h --help
|
|
|
|
* -i, --internal-only
|
|
|
|
* -n mount without writing in /etc/mtab
|
|
|
|
* -O <optlist> limits what is auto mounted
|
|
|
|
* -p, --pass-fd num
|
|
|
|
* -s Tolerate sloppy mount options
|
|
|
|
* -U uuid
|
|
|
|
* -V --version
|
|
|
|
* --no-canonicalize
|
|
|
|
*
|
|
|
|
*----------------------------------------------------------------------
|
|
|
|
* what do we do with these
|
|
|
|
* -l list?
|
|
|
|
* -L <label> label
|
|
|
|
* -v --verbose deprecated
|
|
|
|
*
|
|
|
|
*----------------------------------------------------------------------
|
|
|
|
* Filesystem type
|
|
|
|
* -t <vfstype>
|
|
|
|
* vfstype=<vfstype>
|
|
|
|
*
|
|
|
|
*----------------------------------------------------------------------
|
|
|
|
* Mount Flags/options (-o --options)
|
|
|
|
* -o option[,option]
|
|
|
|
*
|
|
|
|
* The Linux kernel has 32 fs - independent mount flags, that mount command
|
|
|
|
* is responsible for stripping out and mapping to a 32 bit flags field.
|
|
|
|
* The mount commands mapping is documented below.
|
|
|
|
*
|
|
|
|
* Unfortunately we can not directly use this mapping as we need to be able
|
|
|
|
* represent, whether none, 1 or both options of a flag can be present for
|
|
|
|
* example
|
|
|
|
* ro, and rw information is stored in a single bit. But we need 2 bits
|
|
|
|
* of information.
|
|
|
|
* ro - the mount can only be readonly
|
|
|
|
* rw - the mount can only be rw
|
|
|
|
* ro/rw - the mount can be either ro/rw
|
|
|
|
* the fourth state of neither ro/rw does not exist, but still we need
|
|
|
|
* >1 bit to represent the possible choices
|
|
|
|
*
|
|
|
|
* The fs specific mount options are passed into the kernel as a string
|
|
|
|
* to be interpreted by the filesystem.
|
|
|
|
*
|
|
|
|
*
|
|
|
|
* #define MS_RDONLY 1 Mount read-only
|
|
|
|
* ro -r --read-only [source] dest
|
|
|
|
* rw -w
|
|
|
|
* #define MS_NOSUID 2 Ignore suid and sgid bits
|
|
|
|
* nosuid
|
|
|
|
* suid
|
|
|
|
* #define MS_NODEV 4 Disallow access to device special files
|
|
|
|
* nodev
|
|
|
|
* dev
|
|
|
|
* #define MS_NOEXEC 8 Disallow program execution
|
|
|
|
* noexec
|
|
|
|
* exec
|
|
|
|
* #define MS_SYNCHRONOUS 16 Writes are synced at once
|
|
|
|
* sync
|
|
|
|
* async
|
|
|
|
* #define MS_REMOUNT 32 Alter flags of a mounted FS
|
|
|
|
* remount source dest
|
|
|
|
* #define MS_MANDLOCK 64 Allow mandatory locks on an FS
|
|
|
|
* mand
|
|
|
|
* nomand
|
|
|
|
* #define MS_DIRSYNC 128 Directory modifications are synchronous
|
|
|
|
* dirsync
|
|
|
|
* #define MS_NOATIME 1024 Do not update access times
|
|
|
|
* noatime
|
|
|
|
* atime
|
|
|
|
* #define MS_NODIRATIME 2048 Do not update directory access times
|
|
|
|
* nodiratime
|
|
|
|
* diratime
|
|
|
|
* #define MS_BIND 4096
|
|
|
|
* --bind -B source dest
|
|
|
|
* #define MS_MOVE 8192
|
|
|
|
* --move -M source dest
|
|
|
|
* #define MS_REC 16384
|
|
|
|
* --rbind -R source dest
|
|
|
|
* --make-rshared dest
|
|
|
|
* --make-rslave dest
|
|
|
|
* --make-rprivate dest
|
|
|
|
* --make-runbindable dest
|
|
|
|
* #define MS_VERBOSE 32768 MS_VERBOSE is deprecated
|
|
|
|
* #define MS_SILENT 32768
|
|
|
|
* silent
|
|
|
|
* load
|
|
|
|
* #define MS_POSIXACL (1<<16) VFS does not apply the umask
|
|
|
|
* acl
|
|
|
|
* noacl
|
|
|
|
* #define MS_UNBINDABLE (1<<17) change to unbindable
|
|
|
|
* --make-unbindable dest
|
|
|
|
* #define MS_PRIVATE (1<<18) change to private
|
|
|
|
* --make-private dest
|
|
|
|
* #define MS_SLAVE (1<<19) change to slave
|
|
|
|
* --make-slave dest
|
|
|
|
* #define MS_SHARED (1<<20) change to shared
|
|
|
|
* --make-shared dest
|
|
|
|
* #define MS_RELATIME (1<<21) Update atime relative to mtime/ctime
|
|
|
|
* relatime
|
|
|
|
* norelatime
|
|
|
|
* #define MS_KERNMOUNT (1<<22) this is a kern_mount call
|
|
|
|
* #define MS_I_VERSION (1<<23) Update inode I_version field
|
|
|
|
* iversion
|
|
|
|
* noiversion
|
|
|
|
* #define MS_STRICTATIME (1<<24) Always perform atime updates
|
|
|
|
* strictatime
|
|
|
|
* nostrictatime
|
|
|
|
* #define MS_NOSEC (1<<28)
|
|
|
|
* #define MS_BORN (1<<29)
|
|
|
|
* #define MS_ACTIVE (1<<30)
|
|
|
|
* #define MS_NOUSER (1<<31)
|
|
|
|
* nouser
|
|
|
|
* user
|
|
|
|
*
|
|
|
|
* other mount options of interest
|
|
|
|
*
|
|
|
|
* selinux
|
|
|
|
* context=<context>
|
|
|
|
* fscontext=<context>
|
|
|
|
* defcontext=<context>,
|
|
|
|
* rootcontext=<context>
|
|
|
|
*
|
|
|
|
* defaults -> rw, suid, dev, exec, auto, nouser, async
|
|
|
|
* owner -> implies nosuid and nodev
|
|
|
|
* users -> implies noexec, nosuid, and nodev
|
|
|
|
*
|
|
|
|
*----------------------------------------------------------------------
|
|
|
|
* AppArmor mount rules
|
|
|
|
*
|
|
|
|
* AppArmor mount rules try to leverage mount syntax within apparmor syntax
|
|
|
|
* this can not be done entirely but it is largely covered.
|
|
|
|
*
|
|
|
|
* The general mount syntax is
|
|
|
|
* [audit] [deny] [owner] mount [conds]* [source] [ -> [conds] path],
|
|
|
|
* [audit] [deny] remount [conds]* [path],
|
|
|
|
* [audit] [deny] umount [conds]* [path],
|
|
|
|
*
|
|
|
|
* Note: leading owner option applies owner condition to both sours and dest
|
|
|
|
* path.
|
|
|
|
*
|
|
|
|
* where [conds] can be
|
|
|
|
* fstype=<expr>
|
|
|
|
* options=<expr>
|
|
|
|
* owner[=<expr>]
|
|
|
|
*
|
|
|
|
* <expr> := <re> | '(' (<re>[,])+ ')'
|
|
|
|
*
|
|
|
|
* If a condition is not specified then it is assumed to match all possible
|
|
|
|
* entries for it. ie. a missing fstype means all fstypes are matched.
|
|
|
|
* However if a condition is specified then the rule only grants permission
|
|
|
|
* for mounts matching the specified pattern.
|
|
|
|
*
|
|
|
|
* Examples.
|
|
|
|
* mount, # allow any mount
|
|
|
|
* mount /dev/foo, # allow mounting of /dev/foo anywhere
|
|
|
|
* mount options=ro /dev/foo, #allow mounting /dev/foo as read only
|
|
|
|
* mount options=(ro,foo) /dev/foo,
|
|
|
|
* mount options=ro options=foo /dev/foo,
|
|
|
|
* mount fstype=overlayfs options=(rw,upperdir=/tmp/upper/,lowerdir=/) overlay -> /mnt/
|
|
|
|
*
|
|
|
|
*----------------------------------------------------------------------
|
|
|
|
* pivotroot
|
|
|
|
* pivotroot [oldroot=<value>] <path> -> <profile>
|
|
|
|
* pivotroot <path> -> { }
|
|
|
|
*
|
|
|
|
*----------------------------------------------------------------------
|
|
|
|
* chroot
|
|
|
|
* chroot <path> -> <profile>
|
|
|
|
* chroot <path> -> { }
|
|
|
|
*
|
|
|
|
*----------------------------------------------------------------------
|
|
|
|
* AppArmor mount rule encoding
|
|
|
|
*
|
|
|
|
* TODO:
|
2020-11-19 12:30:04 -08:00
|
|
|
* add semantic checking of options against specified filesystem types
|
Add mount rules
Add the ability to control mounting and unmounting
The basic form of the rules are.
[audit] [deny] mount [conds]* [device] [ -> [conds] path],
[audit] [deny] remount [conds]* [path],
[audit] [deny] umount [conds]* [path],
[audit] [deny] pivotroot [oldroot=<value>] <path> -> <profile>
remount is just a short cut for mount options=remount
where [conds] can be
fstype=<expr>
options=<expr>
conds follow the extended conditional syntax of allowing either:
* a single value after the equals, which has the same character range as
regular IDS (ie most anything but it can't be terminated with a , (comma)
and if spaces or other characters are needed it can be quoted
eg.
options=foo
options = foo
options="foo bar"
* a list of values after the equals, the list of values is enclosed within
parenthesis () and its has a slightly reduced character set but again
elements can be quoted.
the separation between elements is whitespace and commas.
eg.
options=(foo bar)
options=(foo, bar)
options=(foo , bar)
options=(foo,bar)
The rules are flexible and follow a similar pattern as network, capability,
etc.
mount, # allow all mounts, but not umount or pivotroot
mount fstype=procfs, # allow mounting procfs anywhere
mount options=(bind, ro) /foo -> /bar, # readonly bind mount
mount /dev/sda -> /mnt,
mount /dev/sd** -> /mnt/**,
mount fstype=overlayfs options=(rw,upperdir=/tmp/upper/,lowerdir=/) overlay -> /mnt/
umount,
umount /m*,
Currently variables and regexs are are supported on the device and mount
point. ie.
mount <devince> -> <mount point>,
Regexes are supported in fstype and options. The options have a further
caveat that regexs only work if the option is fs specific option.
eg. options=(upperdir=/tmp/*,lowerdir=/)
regex's will not currently work against the standard options like ro, rw
nosuid
Conditionals (fstype) can only be applied to the device (source) at this
time and will be disregarded in situations where the mount is manipulating
an existing mount (bind, remount).
Options can be specified multiple times
mount option=rw option=(nosuid,upperdir=/foo),
and will be combined together into a single set of values
The ordering of the standard mount options (rw,ro, ...) does not matter
but the ordering of fs specific options does.
Specifying that the value of a particular option does not matter can be
acheived by providing both the positive and negative forms of and option
option=(rw,ro) options=(suid,nosuid)
For the fs specific options specifying that a particular value does not
matter is achieve using a regex with alternations.
Improvements to the syntax and order restrictions are planned for the
future.
Signed-off-by: John Johansen <john.johansen@canonical.com>
2012-02-24 04:19:38 -08:00
|
|
|
* to catch mount options that can't be covered.
|
|
|
|
*
|
|
|
|
*
|
|
|
|
*/
|
|
|
|
|
|
|
|
|
|
|
|
#include <stdlib.h>
|
|
|
|
#include <string.h>
|
2014-04-07 03:16:50 -07:00
|
|
|
#include <linux/limits.h>
|
Add mount rules
Add the ability to control mounting and unmounting
The basic form of the rules are.
[audit] [deny] mount [conds]* [device] [ -> [conds] path],
[audit] [deny] remount [conds]* [path],
[audit] [deny] umount [conds]* [path],
[audit] [deny] pivotroot [oldroot=<value>] <path> -> <profile>
remount is just a short cut for mount options=remount
where [conds] can be
fstype=<expr>
options=<expr>
conds follow the extended conditional syntax of allowing either:
* a single value after the equals, which has the same character range as
regular IDS (ie most anything but it can't be terminated with a , (comma)
and if spaces or other characters are needed it can be quoted
eg.
options=foo
options = foo
options="foo bar"
* a list of values after the equals, the list of values is enclosed within
parenthesis () and its has a slightly reduced character set but again
elements can be quoted.
the separation between elements is whitespace and commas.
eg.
options=(foo bar)
options=(foo, bar)
options=(foo , bar)
options=(foo,bar)
The rules are flexible and follow a similar pattern as network, capability,
etc.
mount, # allow all mounts, but not umount or pivotroot
mount fstype=procfs, # allow mounting procfs anywhere
mount options=(bind, ro) /foo -> /bar, # readonly bind mount
mount /dev/sda -> /mnt,
mount /dev/sd** -> /mnt/**,
mount fstype=overlayfs options=(rw,upperdir=/tmp/upper/,lowerdir=/) overlay -> /mnt/
umount,
umount /m*,
Currently variables and regexs are are supported on the device and mount
point. ie.
mount <devince> -> <mount point>,
Regexes are supported in fstype and options. The options have a further
caveat that regexs only work if the option is fs specific option.
eg. options=(upperdir=/tmp/*,lowerdir=/)
regex's will not currently work against the standard options like ro, rw
nosuid
Conditionals (fstype) can only be applied to the device (source) at this
time and will be disregarded in situations where the mount is manipulating
an existing mount (bind, remount).
Options can be specified multiple times
mount option=rw option=(nosuid,upperdir=/foo),
and will be combined together into a single set of values
The ordering of the standard mount options (rw,ro, ...) does not matter
but the ordering of fs specific options does.
Specifying that the value of a particular option does not matter can be
acheived by providing both the positive and negative forms of and option
option=(rw,ro) options=(suid,nosuid)
For the fs specific options specifying that a particular value does not
matter is achieve using a regex with alternations.
Improvements to the syntax and order restrictions are planned for the
future.
Signed-off-by: John Johansen <john.johansen@canonical.com>
2012-02-24 04:19:38 -08:00
|
|
|
|
|
|
|
#include "parser.h"
|
2014-04-07 03:16:50 -07:00
|
|
|
#include "policydb.h"
|
|
|
|
#include "profile.h"
|
Add mount rules
Add the ability to control mounting and unmounting
The basic form of the rules are.
[audit] [deny] mount [conds]* [device] [ -> [conds] path],
[audit] [deny] remount [conds]* [path],
[audit] [deny] umount [conds]* [path],
[audit] [deny] pivotroot [oldroot=<value>] <path> -> <profile>
remount is just a short cut for mount options=remount
where [conds] can be
fstype=<expr>
options=<expr>
conds follow the extended conditional syntax of allowing either:
* a single value after the equals, which has the same character range as
regular IDS (ie most anything but it can't be terminated with a , (comma)
and if spaces or other characters are needed it can be quoted
eg.
options=foo
options = foo
options="foo bar"
* a list of values after the equals, the list of values is enclosed within
parenthesis () and its has a slightly reduced character set but again
elements can be quoted.
the separation between elements is whitespace and commas.
eg.
options=(foo bar)
options=(foo, bar)
options=(foo , bar)
options=(foo,bar)
The rules are flexible and follow a similar pattern as network, capability,
etc.
mount, # allow all mounts, but not umount or pivotroot
mount fstype=procfs, # allow mounting procfs anywhere
mount options=(bind, ro) /foo -> /bar, # readonly bind mount
mount /dev/sda -> /mnt,
mount /dev/sd** -> /mnt/**,
mount fstype=overlayfs options=(rw,upperdir=/tmp/upper/,lowerdir=/) overlay -> /mnt/
umount,
umount /m*,
Currently variables and regexs are are supported on the device and mount
point. ie.
mount <devince> -> <mount point>,
Regexes are supported in fstype and options. The options have a further
caveat that regexs only work if the option is fs specific option.
eg. options=(upperdir=/tmp/*,lowerdir=/)
regex's will not currently work against the standard options like ro, rw
nosuid
Conditionals (fstype) can only be applied to the device (source) at this
time and will be disregarded in situations where the mount is manipulating
an existing mount (bind, remount).
Options can be specified multiple times
mount option=rw option=(nosuid,upperdir=/foo),
and will be combined together into a single set of values
The ordering of the standard mount options (rw,ro, ...) does not matter
but the ordering of fs specific options does.
Specifying that the value of a particular option does not matter can be
acheived by providing both the positive and negative forms of and option
option=(rw,ro) options=(suid,nosuid)
For the fs specific options specifying that a particular value does not
matter is achieve using a regex with alternations.
Improvements to the syntax and order restrictions are planned for the
future.
Signed-off-by: John Johansen <john.johansen@canonical.com>
2012-02-24 04:19:38 -08:00
|
|
|
#include "mount.h"
|
|
|
|
|
|
|
|
struct mnt_keyword_table {
|
2013-09-27 16:13:22 -07:00
|
|
|
const char *keyword;
|
Add mount rules
Add the ability to control mounting and unmounting
The basic form of the rules are.
[audit] [deny] mount [conds]* [device] [ -> [conds] path],
[audit] [deny] remount [conds]* [path],
[audit] [deny] umount [conds]* [path],
[audit] [deny] pivotroot [oldroot=<value>] <path> -> <profile>
remount is just a short cut for mount options=remount
where [conds] can be
fstype=<expr>
options=<expr>
conds follow the extended conditional syntax of allowing either:
* a single value after the equals, which has the same character range as
regular IDS (ie most anything but it can't be terminated with a , (comma)
and if spaces or other characters are needed it can be quoted
eg.
options=foo
options = foo
options="foo bar"
* a list of values after the equals, the list of values is enclosed within
parenthesis () and its has a slightly reduced character set but again
elements can be quoted.
the separation between elements is whitespace and commas.
eg.
options=(foo bar)
options=(foo, bar)
options=(foo , bar)
options=(foo,bar)
The rules are flexible and follow a similar pattern as network, capability,
etc.
mount, # allow all mounts, but not umount or pivotroot
mount fstype=procfs, # allow mounting procfs anywhere
mount options=(bind, ro) /foo -> /bar, # readonly bind mount
mount /dev/sda -> /mnt,
mount /dev/sd** -> /mnt/**,
mount fstype=overlayfs options=(rw,upperdir=/tmp/upper/,lowerdir=/) overlay -> /mnt/
umount,
umount /m*,
Currently variables and regexs are are supported on the device and mount
point. ie.
mount <devince> -> <mount point>,
Regexes are supported in fstype and options. The options have a further
caveat that regexs only work if the option is fs specific option.
eg. options=(upperdir=/tmp/*,lowerdir=/)
regex's will not currently work against the standard options like ro, rw
nosuid
Conditionals (fstype) can only be applied to the device (source) at this
time and will be disregarded in situations where the mount is manipulating
an existing mount (bind, remount).
Options can be specified multiple times
mount option=rw option=(nosuid,upperdir=/foo),
and will be combined together into a single set of values
The ordering of the standard mount options (rw,ro, ...) does not matter
but the ordering of fs specific options does.
Specifying that the value of a particular option does not matter can be
acheived by providing both the positive and negative forms of and option
option=(rw,ro) options=(suid,nosuid)
For the fs specific options specifying that a particular value does not
matter is achieve using a regex with alternations.
Improvements to the syntax and order restrictions are planned for the
future.
Signed-off-by: John Johansen <john.johansen@canonical.com>
2012-02-24 04:19:38 -08:00
|
|
|
unsigned int set;
|
|
|
|
unsigned int clear;
|
|
|
|
};
|
|
|
|
|
|
|
|
static struct mnt_keyword_table mnt_opts_table[] = {
|
|
|
|
{"ro", MS_RDONLY, 0},
|
|
|
|
{"r", MS_RDONLY, 0},
|
|
|
|
{"read-only", MS_RDONLY, 0},
|
|
|
|
{"rw", 0, MS_RDONLY},
|
|
|
|
{"w", 0, MS_RDONLY},
|
|
|
|
{"suid", 0, MS_NOSUID},
|
|
|
|
{"nosuid", MS_NOSUID, 0},
|
|
|
|
{"dev", 0, MS_NODEV},
|
|
|
|
{"nodev", MS_NODEV, 0},
|
|
|
|
{"exec", 0, MS_NOEXEC},
|
|
|
|
{"noexec", MS_NOEXEC, 0},
|
|
|
|
{"sync", MS_SYNC, 0},
|
|
|
|
{"async", 0, MS_SYNC},
|
|
|
|
{"remount", MS_REMOUNT, 0},
|
|
|
|
{"mand", MS_MAND, 0},
|
|
|
|
{"nomand", 0, MS_MAND},
|
|
|
|
{"dirsync", MS_DIRSYNC, 0},
|
|
|
|
{"atime", 0, MS_NOATIME},
|
|
|
|
{"noatime", MS_NOATIME, 0},
|
|
|
|
{"diratime", 0, MS_NODIRATIME},
|
|
|
|
{"nodiratime", MS_NODIRATIME, 0},
|
|
|
|
{"bind", MS_BIND, 0},
|
|
|
|
{"B", MS_BIND, 0},
|
|
|
|
{"move", MS_MOVE, 0},
|
|
|
|
{"M", MS_MOVE, 0},
|
|
|
|
{"rbind", MS_RBIND, 0},
|
|
|
|
{"R", MS_RBIND, 0},
|
|
|
|
{"verbose", MS_VERBOSE, 0},
|
|
|
|
{"silent", MS_SILENT, 0},
|
parser: Sync mount options parsing and documentation
There are a number of differences between what the apparmor.d(5) man
page lists as valid AppArmor mount rule options and what apparmor_parser
looks for when parsing mount rules. There are also typos in the man page
and parser around mount options. Here's the breakdown of problems and
fixes made in this patch:
* The apparmor.d(5) man page improperly documented a "nodirsync"
option.
- That mount option does not exist and the parser did not honor it.
Remove the mention from the apparmor.d(5) man page.
* The loud option was typoed as "load" in both the man page and parser
- There's no sense in preserving backwards compatibility. "load" is
simply wrong and should not be honored. The man page and parser are
updated to only use "loud".
* The rbind option wasn't listed in the man page.
- Add rbind to the man page. No change needed for the parser.
* The documented unbindable, private, slave, and shared options were
not correctly parsed. The parser expected
make-{unbindable,private,slave,shared}.
- The parser is updated to accept both the documented
{unbindable,private,slave,shared} options and their variants
prefixed with "make-". The man page will not document the "make-"
variants.
* The recursive {runbindable,rprivate,rslave,rshared} options were not
documented and were only recognized by the parser if they were
prefixed with "make-".
- The man page is updated to document the option strings that are not
prefixed with "make-". The parser still accepts the "make-"
variants.
* The man page documented a "rec" option but the parser didn't honor
it. The MS_REC macro is used by the mount utility to be bitwise OR'ed
with MS_{UNBINDABLE,PRIVATE,SLAVE,SHARED} to indicate the
corresponding recursive mount options.
- This is not an option that should be exposed in the AppArmor policy
since we already allow have the
{runbindable,rprivate,rslave,rshared} options.
* The man page typoed the {no,}relatime options as {no,}relative.
- The man page is updated to document the correct option strings. The
parser requires no change.
Bug: https://bugs.launchpad.net/bugs/1401619
Signed-off-by: Tyler Hicks <tyhicks@canonical.com>
Acked-by: John Johansen <john.johansen@canonical.com>
2014-12-12 08:20:31 -06:00
|
|
|
{"loud", 0, MS_SILENT},
|
Add mount rules
Add the ability to control mounting and unmounting
The basic form of the rules are.
[audit] [deny] mount [conds]* [device] [ -> [conds] path],
[audit] [deny] remount [conds]* [path],
[audit] [deny] umount [conds]* [path],
[audit] [deny] pivotroot [oldroot=<value>] <path> -> <profile>
remount is just a short cut for mount options=remount
where [conds] can be
fstype=<expr>
options=<expr>
conds follow the extended conditional syntax of allowing either:
* a single value after the equals, which has the same character range as
regular IDS (ie most anything but it can't be terminated with a , (comma)
and if spaces or other characters are needed it can be quoted
eg.
options=foo
options = foo
options="foo bar"
* a list of values after the equals, the list of values is enclosed within
parenthesis () and its has a slightly reduced character set but again
elements can be quoted.
the separation between elements is whitespace and commas.
eg.
options=(foo bar)
options=(foo, bar)
options=(foo , bar)
options=(foo,bar)
The rules are flexible and follow a similar pattern as network, capability,
etc.
mount, # allow all mounts, but not umount or pivotroot
mount fstype=procfs, # allow mounting procfs anywhere
mount options=(bind, ro) /foo -> /bar, # readonly bind mount
mount /dev/sda -> /mnt,
mount /dev/sd** -> /mnt/**,
mount fstype=overlayfs options=(rw,upperdir=/tmp/upper/,lowerdir=/) overlay -> /mnt/
umount,
umount /m*,
Currently variables and regexs are are supported on the device and mount
point. ie.
mount <devince> -> <mount point>,
Regexes are supported in fstype and options. The options have a further
caveat that regexs only work if the option is fs specific option.
eg. options=(upperdir=/tmp/*,lowerdir=/)
regex's will not currently work against the standard options like ro, rw
nosuid
Conditionals (fstype) can only be applied to the device (source) at this
time and will be disregarded in situations where the mount is manipulating
an existing mount (bind, remount).
Options can be specified multiple times
mount option=rw option=(nosuid,upperdir=/foo),
and will be combined together into a single set of values
The ordering of the standard mount options (rw,ro, ...) does not matter
but the ordering of fs specific options does.
Specifying that the value of a particular option does not matter can be
acheived by providing both the positive and negative forms of and option
option=(rw,ro) options=(suid,nosuid)
For the fs specific options specifying that a particular value does not
matter is achieve using a regex with alternations.
Improvements to the syntax and order restrictions are planned for the
future.
Signed-off-by: John Johansen <john.johansen@canonical.com>
2012-02-24 04:19:38 -08:00
|
|
|
{"acl", MS_ACL, 0},
|
|
|
|
{"noacl", 0, MS_ACL},
|
parser: Sync mount options parsing and documentation
There are a number of differences between what the apparmor.d(5) man
page lists as valid AppArmor mount rule options and what apparmor_parser
looks for when parsing mount rules. There are also typos in the man page
and parser around mount options. Here's the breakdown of problems and
fixes made in this patch:
* The apparmor.d(5) man page improperly documented a "nodirsync"
option.
- That mount option does not exist and the parser did not honor it.
Remove the mention from the apparmor.d(5) man page.
* The loud option was typoed as "load" in both the man page and parser
- There's no sense in preserving backwards compatibility. "load" is
simply wrong and should not be honored. The man page and parser are
updated to only use "loud".
* The rbind option wasn't listed in the man page.
- Add rbind to the man page. No change needed for the parser.
* The documented unbindable, private, slave, and shared options were
not correctly parsed. The parser expected
make-{unbindable,private,slave,shared}.
- The parser is updated to accept both the documented
{unbindable,private,slave,shared} options and their variants
prefixed with "make-". The man page will not document the "make-"
variants.
* The recursive {runbindable,rprivate,rslave,rshared} options were not
documented and were only recognized by the parser if they were
prefixed with "make-".
- The man page is updated to document the option strings that are not
prefixed with "make-". The parser still accepts the "make-"
variants.
* The man page documented a "rec" option but the parser didn't honor
it. The MS_REC macro is used by the mount utility to be bitwise OR'ed
with MS_{UNBINDABLE,PRIVATE,SLAVE,SHARED} to indicate the
corresponding recursive mount options.
- This is not an option that should be exposed in the AppArmor policy
since we already allow have the
{runbindable,rprivate,rslave,rshared} options.
* The man page typoed the {no,}relatime options as {no,}relative.
- The man page is updated to document the correct option strings. The
parser requires no change.
Bug: https://bugs.launchpad.net/bugs/1401619
Signed-off-by: Tyler Hicks <tyhicks@canonical.com>
Acked-by: John Johansen <john.johansen@canonical.com>
2014-12-12 08:20:31 -06:00
|
|
|
{"unbindable", MS_UNBINDABLE, 0},
|
Add mount rules
Add the ability to control mounting and unmounting
The basic form of the rules are.
[audit] [deny] mount [conds]* [device] [ -> [conds] path],
[audit] [deny] remount [conds]* [path],
[audit] [deny] umount [conds]* [path],
[audit] [deny] pivotroot [oldroot=<value>] <path> -> <profile>
remount is just a short cut for mount options=remount
where [conds] can be
fstype=<expr>
options=<expr>
conds follow the extended conditional syntax of allowing either:
* a single value after the equals, which has the same character range as
regular IDS (ie most anything but it can't be terminated with a , (comma)
and if spaces or other characters are needed it can be quoted
eg.
options=foo
options = foo
options="foo bar"
* a list of values after the equals, the list of values is enclosed within
parenthesis () and its has a slightly reduced character set but again
elements can be quoted.
the separation between elements is whitespace and commas.
eg.
options=(foo bar)
options=(foo, bar)
options=(foo , bar)
options=(foo,bar)
The rules are flexible and follow a similar pattern as network, capability,
etc.
mount, # allow all mounts, but not umount or pivotroot
mount fstype=procfs, # allow mounting procfs anywhere
mount options=(bind, ro) /foo -> /bar, # readonly bind mount
mount /dev/sda -> /mnt,
mount /dev/sd** -> /mnt/**,
mount fstype=overlayfs options=(rw,upperdir=/tmp/upper/,lowerdir=/) overlay -> /mnt/
umount,
umount /m*,
Currently variables and regexs are are supported on the device and mount
point. ie.
mount <devince> -> <mount point>,
Regexes are supported in fstype and options. The options have a further
caveat that regexs only work if the option is fs specific option.
eg. options=(upperdir=/tmp/*,lowerdir=/)
regex's will not currently work against the standard options like ro, rw
nosuid
Conditionals (fstype) can only be applied to the device (source) at this
time and will be disregarded in situations where the mount is manipulating
an existing mount (bind, remount).
Options can be specified multiple times
mount option=rw option=(nosuid,upperdir=/foo),
and will be combined together into a single set of values
The ordering of the standard mount options (rw,ro, ...) does not matter
but the ordering of fs specific options does.
Specifying that the value of a particular option does not matter can be
acheived by providing both the positive and negative forms of and option
option=(rw,ro) options=(suid,nosuid)
For the fs specific options specifying that a particular value does not
matter is achieve using a regex with alternations.
Improvements to the syntax and order restrictions are planned for the
future.
Signed-off-by: John Johansen <john.johansen@canonical.com>
2012-02-24 04:19:38 -08:00
|
|
|
{"make-unbindable", MS_UNBINDABLE, 0},
|
parser: Sync mount options parsing and documentation
There are a number of differences between what the apparmor.d(5) man
page lists as valid AppArmor mount rule options and what apparmor_parser
looks for when parsing mount rules. There are also typos in the man page
and parser around mount options. Here's the breakdown of problems and
fixes made in this patch:
* The apparmor.d(5) man page improperly documented a "nodirsync"
option.
- That mount option does not exist and the parser did not honor it.
Remove the mention from the apparmor.d(5) man page.
* The loud option was typoed as "load" in both the man page and parser
- There's no sense in preserving backwards compatibility. "load" is
simply wrong and should not be honored. The man page and parser are
updated to only use "loud".
* The rbind option wasn't listed in the man page.
- Add rbind to the man page. No change needed for the parser.
* The documented unbindable, private, slave, and shared options were
not correctly parsed. The parser expected
make-{unbindable,private,slave,shared}.
- The parser is updated to accept both the documented
{unbindable,private,slave,shared} options and their variants
prefixed with "make-". The man page will not document the "make-"
variants.
* The recursive {runbindable,rprivate,rslave,rshared} options were not
documented and were only recognized by the parser if they were
prefixed with "make-".
- The man page is updated to document the option strings that are not
prefixed with "make-". The parser still accepts the "make-"
variants.
* The man page documented a "rec" option but the parser didn't honor
it. The MS_REC macro is used by the mount utility to be bitwise OR'ed
with MS_{UNBINDABLE,PRIVATE,SLAVE,SHARED} to indicate the
corresponding recursive mount options.
- This is not an option that should be exposed in the AppArmor policy
since we already allow have the
{runbindable,rprivate,rslave,rshared} options.
* The man page typoed the {no,}relatime options as {no,}relative.
- The man page is updated to document the correct option strings. The
parser requires no change.
Bug: https://bugs.launchpad.net/bugs/1401619
Signed-off-by: Tyler Hicks <tyhicks@canonical.com>
Acked-by: John Johansen <john.johansen@canonical.com>
2014-12-12 08:20:31 -06:00
|
|
|
{"runbindable", MS_RUNBINDABLE, 0},
|
Add mount rules
Add the ability to control mounting and unmounting
The basic form of the rules are.
[audit] [deny] mount [conds]* [device] [ -> [conds] path],
[audit] [deny] remount [conds]* [path],
[audit] [deny] umount [conds]* [path],
[audit] [deny] pivotroot [oldroot=<value>] <path> -> <profile>
remount is just a short cut for mount options=remount
where [conds] can be
fstype=<expr>
options=<expr>
conds follow the extended conditional syntax of allowing either:
* a single value after the equals, which has the same character range as
regular IDS (ie most anything but it can't be terminated with a , (comma)
and if spaces or other characters are needed it can be quoted
eg.
options=foo
options = foo
options="foo bar"
* a list of values after the equals, the list of values is enclosed within
parenthesis () and its has a slightly reduced character set but again
elements can be quoted.
the separation between elements is whitespace and commas.
eg.
options=(foo bar)
options=(foo, bar)
options=(foo , bar)
options=(foo,bar)
The rules are flexible and follow a similar pattern as network, capability,
etc.
mount, # allow all mounts, but not umount or pivotroot
mount fstype=procfs, # allow mounting procfs anywhere
mount options=(bind, ro) /foo -> /bar, # readonly bind mount
mount /dev/sda -> /mnt,
mount /dev/sd** -> /mnt/**,
mount fstype=overlayfs options=(rw,upperdir=/tmp/upper/,lowerdir=/) overlay -> /mnt/
umount,
umount /m*,
Currently variables and regexs are are supported on the device and mount
point. ie.
mount <devince> -> <mount point>,
Regexes are supported in fstype and options. The options have a further
caveat that regexs only work if the option is fs specific option.
eg. options=(upperdir=/tmp/*,lowerdir=/)
regex's will not currently work against the standard options like ro, rw
nosuid
Conditionals (fstype) can only be applied to the device (source) at this
time and will be disregarded in situations where the mount is manipulating
an existing mount (bind, remount).
Options can be specified multiple times
mount option=rw option=(nosuid,upperdir=/foo),
and will be combined together into a single set of values
The ordering of the standard mount options (rw,ro, ...) does not matter
but the ordering of fs specific options does.
Specifying that the value of a particular option does not matter can be
acheived by providing both the positive and negative forms of and option
option=(rw,ro) options=(suid,nosuid)
For the fs specific options specifying that a particular value does not
matter is achieve using a regex with alternations.
Improvements to the syntax and order restrictions are planned for the
future.
Signed-off-by: John Johansen <john.johansen@canonical.com>
2012-02-24 04:19:38 -08:00
|
|
|
{"make-runbindable", MS_RUNBINDABLE, 0},
|
parser: Sync mount options parsing and documentation
There are a number of differences between what the apparmor.d(5) man
page lists as valid AppArmor mount rule options and what apparmor_parser
looks for when parsing mount rules. There are also typos in the man page
and parser around mount options. Here's the breakdown of problems and
fixes made in this patch:
* The apparmor.d(5) man page improperly documented a "nodirsync"
option.
- That mount option does not exist and the parser did not honor it.
Remove the mention from the apparmor.d(5) man page.
* The loud option was typoed as "load" in both the man page and parser
- There's no sense in preserving backwards compatibility. "load" is
simply wrong and should not be honored. The man page and parser are
updated to only use "loud".
* The rbind option wasn't listed in the man page.
- Add rbind to the man page. No change needed for the parser.
* The documented unbindable, private, slave, and shared options were
not correctly parsed. The parser expected
make-{unbindable,private,slave,shared}.
- The parser is updated to accept both the documented
{unbindable,private,slave,shared} options and their variants
prefixed with "make-". The man page will not document the "make-"
variants.
* The recursive {runbindable,rprivate,rslave,rshared} options were not
documented and were only recognized by the parser if they were
prefixed with "make-".
- The man page is updated to document the option strings that are not
prefixed with "make-". The parser still accepts the "make-"
variants.
* The man page documented a "rec" option but the parser didn't honor
it. The MS_REC macro is used by the mount utility to be bitwise OR'ed
with MS_{UNBINDABLE,PRIVATE,SLAVE,SHARED} to indicate the
corresponding recursive mount options.
- This is not an option that should be exposed in the AppArmor policy
since we already allow have the
{runbindable,rprivate,rslave,rshared} options.
* The man page typoed the {no,}relatime options as {no,}relative.
- The man page is updated to document the correct option strings. The
parser requires no change.
Bug: https://bugs.launchpad.net/bugs/1401619
Signed-off-by: Tyler Hicks <tyhicks@canonical.com>
Acked-by: John Johansen <john.johansen@canonical.com>
2014-12-12 08:20:31 -06:00
|
|
|
{"private", MS_PRIVATE, 0},
|
Add mount rules
Add the ability to control mounting and unmounting
The basic form of the rules are.
[audit] [deny] mount [conds]* [device] [ -> [conds] path],
[audit] [deny] remount [conds]* [path],
[audit] [deny] umount [conds]* [path],
[audit] [deny] pivotroot [oldroot=<value>] <path> -> <profile>
remount is just a short cut for mount options=remount
where [conds] can be
fstype=<expr>
options=<expr>
conds follow the extended conditional syntax of allowing either:
* a single value after the equals, which has the same character range as
regular IDS (ie most anything but it can't be terminated with a , (comma)
and if spaces or other characters are needed it can be quoted
eg.
options=foo
options = foo
options="foo bar"
* a list of values after the equals, the list of values is enclosed within
parenthesis () and its has a slightly reduced character set but again
elements can be quoted.
the separation between elements is whitespace and commas.
eg.
options=(foo bar)
options=(foo, bar)
options=(foo , bar)
options=(foo,bar)
The rules are flexible and follow a similar pattern as network, capability,
etc.
mount, # allow all mounts, but not umount or pivotroot
mount fstype=procfs, # allow mounting procfs anywhere
mount options=(bind, ro) /foo -> /bar, # readonly bind mount
mount /dev/sda -> /mnt,
mount /dev/sd** -> /mnt/**,
mount fstype=overlayfs options=(rw,upperdir=/tmp/upper/,lowerdir=/) overlay -> /mnt/
umount,
umount /m*,
Currently variables and regexs are are supported on the device and mount
point. ie.
mount <devince> -> <mount point>,
Regexes are supported in fstype and options. The options have a further
caveat that regexs only work if the option is fs specific option.
eg. options=(upperdir=/tmp/*,lowerdir=/)
regex's will not currently work against the standard options like ro, rw
nosuid
Conditionals (fstype) can only be applied to the device (source) at this
time and will be disregarded in situations where the mount is manipulating
an existing mount (bind, remount).
Options can be specified multiple times
mount option=rw option=(nosuid,upperdir=/foo),
and will be combined together into a single set of values
The ordering of the standard mount options (rw,ro, ...) does not matter
but the ordering of fs specific options does.
Specifying that the value of a particular option does not matter can be
acheived by providing both the positive and negative forms of and option
option=(rw,ro) options=(suid,nosuid)
For the fs specific options specifying that a particular value does not
matter is achieve using a regex with alternations.
Improvements to the syntax and order restrictions are planned for the
future.
Signed-off-by: John Johansen <john.johansen@canonical.com>
2012-02-24 04:19:38 -08:00
|
|
|
{"make-private", MS_PRIVATE, 0},
|
parser: Sync mount options parsing and documentation
There are a number of differences between what the apparmor.d(5) man
page lists as valid AppArmor mount rule options and what apparmor_parser
looks for when parsing mount rules. There are also typos in the man page
and parser around mount options. Here's the breakdown of problems and
fixes made in this patch:
* The apparmor.d(5) man page improperly documented a "nodirsync"
option.
- That mount option does not exist and the parser did not honor it.
Remove the mention from the apparmor.d(5) man page.
* The loud option was typoed as "load" in both the man page and parser
- There's no sense in preserving backwards compatibility. "load" is
simply wrong and should not be honored. The man page and parser are
updated to only use "loud".
* The rbind option wasn't listed in the man page.
- Add rbind to the man page. No change needed for the parser.
* The documented unbindable, private, slave, and shared options were
not correctly parsed. The parser expected
make-{unbindable,private,slave,shared}.
- The parser is updated to accept both the documented
{unbindable,private,slave,shared} options and their variants
prefixed with "make-". The man page will not document the "make-"
variants.
* The recursive {runbindable,rprivate,rslave,rshared} options were not
documented and were only recognized by the parser if they were
prefixed with "make-".
- The man page is updated to document the option strings that are not
prefixed with "make-". The parser still accepts the "make-"
variants.
* The man page documented a "rec" option but the parser didn't honor
it. The MS_REC macro is used by the mount utility to be bitwise OR'ed
with MS_{UNBINDABLE,PRIVATE,SLAVE,SHARED} to indicate the
corresponding recursive mount options.
- This is not an option that should be exposed in the AppArmor policy
since we already allow have the
{runbindable,rprivate,rslave,rshared} options.
* The man page typoed the {no,}relatime options as {no,}relative.
- The man page is updated to document the correct option strings. The
parser requires no change.
Bug: https://bugs.launchpad.net/bugs/1401619
Signed-off-by: Tyler Hicks <tyhicks@canonical.com>
Acked-by: John Johansen <john.johansen@canonical.com>
2014-12-12 08:20:31 -06:00
|
|
|
{"rprivate", MS_RPRIVATE, 0},
|
Add mount rules
Add the ability to control mounting and unmounting
The basic form of the rules are.
[audit] [deny] mount [conds]* [device] [ -> [conds] path],
[audit] [deny] remount [conds]* [path],
[audit] [deny] umount [conds]* [path],
[audit] [deny] pivotroot [oldroot=<value>] <path> -> <profile>
remount is just a short cut for mount options=remount
where [conds] can be
fstype=<expr>
options=<expr>
conds follow the extended conditional syntax of allowing either:
* a single value after the equals, which has the same character range as
regular IDS (ie most anything but it can't be terminated with a , (comma)
and if spaces or other characters are needed it can be quoted
eg.
options=foo
options = foo
options="foo bar"
* a list of values after the equals, the list of values is enclosed within
parenthesis () and its has a slightly reduced character set but again
elements can be quoted.
the separation between elements is whitespace and commas.
eg.
options=(foo bar)
options=(foo, bar)
options=(foo , bar)
options=(foo,bar)
The rules are flexible and follow a similar pattern as network, capability,
etc.
mount, # allow all mounts, but not umount or pivotroot
mount fstype=procfs, # allow mounting procfs anywhere
mount options=(bind, ro) /foo -> /bar, # readonly bind mount
mount /dev/sda -> /mnt,
mount /dev/sd** -> /mnt/**,
mount fstype=overlayfs options=(rw,upperdir=/tmp/upper/,lowerdir=/) overlay -> /mnt/
umount,
umount /m*,
Currently variables and regexs are are supported on the device and mount
point. ie.
mount <devince> -> <mount point>,
Regexes are supported in fstype and options. The options have a further
caveat that regexs only work if the option is fs specific option.
eg. options=(upperdir=/tmp/*,lowerdir=/)
regex's will not currently work against the standard options like ro, rw
nosuid
Conditionals (fstype) can only be applied to the device (source) at this
time and will be disregarded in situations where the mount is manipulating
an existing mount (bind, remount).
Options can be specified multiple times
mount option=rw option=(nosuid,upperdir=/foo),
and will be combined together into a single set of values
The ordering of the standard mount options (rw,ro, ...) does not matter
but the ordering of fs specific options does.
Specifying that the value of a particular option does not matter can be
acheived by providing both the positive and negative forms of and option
option=(rw,ro) options=(suid,nosuid)
For the fs specific options specifying that a particular value does not
matter is achieve using a regex with alternations.
Improvements to the syntax and order restrictions are planned for the
future.
Signed-off-by: John Johansen <john.johansen@canonical.com>
2012-02-24 04:19:38 -08:00
|
|
|
{"make-rprivate", MS_RPRIVATE, 0},
|
parser: Sync mount options parsing and documentation
There are a number of differences between what the apparmor.d(5) man
page lists as valid AppArmor mount rule options and what apparmor_parser
looks for when parsing mount rules. There are also typos in the man page
and parser around mount options. Here's the breakdown of problems and
fixes made in this patch:
* The apparmor.d(5) man page improperly documented a "nodirsync"
option.
- That mount option does not exist and the parser did not honor it.
Remove the mention from the apparmor.d(5) man page.
* The loud option was typoed as "load" in both the man page and parser
- There's no sense in preserving backwards compatibility. "load" is
simply wrong and should not be honored. The man page and parser are
updated to only use "loud".
* The rbind option wasn't listed in the man page.
- Add rbind to the man page. No change needed for the parser.
* The documented unbindable, private, slave, and shared options were
not correctly parsed. The parser expected
make-{unbindable,private,slave,shared}.
- The parser is updated to accept both the documented
{unbindable,private,slave,shared} options and their variants
prefixed with "make-". The man page will not document the "make-"
variants.
* The recursive {runbindable,rprivate,rslave,rshared} options were not
documented and were only recognized by the parser if they were
prefixed with "make-".
- The man page is updated to document the option strings that are not
prefixed with "make-". The parser still accepts the "make-"
variants.
* The man page documented a "rec" option but the parser didn't honor
it. The MS_REC macro is used by the mount utility to be bitwise OR'ed
with MS_{UNBINDABLE,PRIVATE,SLAVE,SHARED} to indicate the
corresponding recursive mount options.
- This is not an option that should be exposed in the AppArmor policy
since we already allow have the
{runbindable,rprivate,rslave,rshared} options.
* The man page typoed the {no,}relatime options as {no,}relative.
- The man page is updated to document the correct option strings. The
parser requires no change.
Bug: https://bugs.launchpad.net/bugs/1401619
Signed-off-by: Tyler Hicks <tyhicks@canonical.com>
Acked-by: John Johansen <john.johansen@canonical.com>
2014-12-12 08:20:31 -06:00
|
|
|
{"slave", MS_SLAVE, 0},
|
Add mount rules
Add the ability to control mounting and unmounting
The basic form of the rules are.
[audit] [deny] mount [conds]* [device] [ -> [conds] path],
[audit] [deny] remount [conds]* [path],
[audit] [deny] umount [conds]* [path],
[audit] [deny] pivotroot [oldroot=<value>] <path> -> <profile>
remount is just a short cut for mount options=remount
where [conds] can be
fstype=<expr>
options=<expr>
conds follow the extended conditional syntax of allowing either:
* a single value after the equals, which has the same character range as
regular IDS (ie most anything but it can't be terminated with a , (comma)
and if spaces or other characters are needed it can be quoted
eg.
options=foo
options = foo
options="foo bar"
* a list of values after the equals, the list of values is enclosed within
parenthesis () and its has a slightly reduced character set but again
elements can be quoted.
the separation between elements is whitespace and commas.
eg.
options=(foo bar)
options=(foo, bar)
options=(foo , bar)
options=(foo,bar)
The rules are flexible and follow a similar pattern as network, capability,
etc.
mount, # allow all mounts, but not umount or pivotroot
mount fstype=procfs, # allow mounting procfs anywhere
mount options=(bind, ro) /foo -> /bar, # readonly bind mount
mount /dev/sda -> /mnt,
mount /dev/sd** -> /mnt/**,
mount fstype=overlayfs options=(rw,upperdir=/tmp/upper/,lowerdir=/) overlay -> /mnt/
umount,
umount /m*,
Currently variables and regexs are are supported on the device and mount
point. ie.
mount <devince> -> <mount point>,
Regexes are supported in fstype and options. The options have a further
caveat that regexs only work if the option is fs specific option.
eg. options=(upperdir=/tmp/*,lowerdir=/)
regex's will not currently work against the standard options like ro, rw
nosuid
Conditionals (fstype) can only be applied to the device (source) at this
time and will be disregarded in situations where the mount is manipulating
an existing mount (bind, remount).
Options can be specified multiple times
mount option=rw option=(nosuid,upperdir=/foo),
and will be combined together into a single set of values
The ordering of the standard mount options (rw,ro, ...) does not matter
but the ordering of fs specific options does.
Specifying that the value of a particular option does not matter can be
acheived by providing both the positive and negative forms of and option
option=(rw,ro) options=(suid,nosuid)
For the fs specific options specifying that a particular value does not
matter is achieve using a regex with alternations.
Improvements to the syntax and order restrictions are planned for the
future.
Signed-off-by: John Johansen <john.johansen@canonical.com>
2012-02-24 04:19:38 -08:00
|
|
|
{"make-slave", MS_SLAVE, 0},
|
parser: Sync mount options parsing and documentation
There are a number of differences between what the apparmor.d(5) man
page lists as valid AppArmor mount rule options and what apparmor_parser
looks for when parsing mount rules. There are also typos in the man page
and parser around mount options. Here's the breakdown of problems and
fixes made in this patch:
* The apparmor.d(5) man page improperly documented a "nodirsync"
option.
- That mount option does not exist and the parser did not honor it.
Remove the mention from the apparmor.d(5) man page.
* The loud option was typoed as "load" in both the man page and parser
- There's no sense in preserving backwards compatibility. "load" is
simply wrong and should not be honored. The man page and parser are
updated to only use "loud".
* The rbind option wasn't listed in the man page.
- Add rbind to the man page. No change needed for the parser.
* The documented unbindable, private, slave, and shared options were
not correctly parsed. The parser expected
make-{unbindable,private,slave,shared}.
- The parser is updated to accept both the documented
{unbindable,private,slave,shared} options and their variants
prefixed with "make-". The man page will not document the "make-"
variants.
* The recursive {runbindable,rprivate,rslave,rshared} options were not
documented and were only recognized by the parser if they were
prefixed with "make-".
- The man page is updated to document the option strings that are not
prefixed with "make-". The parser still accepts the "make-"
variants.
* The man page documented a "rec" option but the parser didn't honor
it. The MS_REC macro is used by the mount utility to be bitwise OR'ed
with MS_{UNBINDABLE,PRIVATE,SLAVE,SHARED} to indicate the
corresponding recursive mount options.
- This is not an option that should be exposed in the AppArmor policy
since we already allow have the
{runbindable,rprivate,rslave,rshared} options.
* The man page typoed the {no,}relatime options as {no,}relative.
- The man page is updated to document the correct option strings. The
parser requires no change.
Bug: https://bugs.launchpad.net/bugs/1401619
Signed-off-by: Tyler Hicks <tyhicks@canonical.com>
Acked-by: John Johansen <john.johansen@canonical.com>
2014-12-12 08:20:31 -06:00
|
|
|
{"rslave", MS_RSLAVE, 0},
|
Add mount rules
Add the ability to control mounting and unmounting
The basic form of the rules are.
[audit] [deny] mount [conds]* [device] [ -> [conds] path],
[audit] [deny] remount [conds]* [path],
[audit] [deny] umount [conds]* [path],
[audit] [deny] pivotroot [oldroot=<value>] <path> -> <profile>
remount is just a short cut for mount options=remount
where [conds] can be
fstype=<expr>
options=<expr>
conds follow the extended conditional syntax of allowing either:
* a single value after the equals, which has the same character range as
regular IDS (ie most anything but it can't be terminated with a , (comma)
and if spaces or other characters are needed it can be quoted
eg.
options=foo
options = foo
options="foo bar"
* a list of values after the equals, the list of values is enclosed within
parenthesis () and its has a slightly reduced character set but again
elements can be quoted.
the separation between elements is whitespace and commas.
eg.
options=(foo bar)
options=(foo, bar)
options=(foo , bar)
options=(foo,bar)
The rules are flexible and follow a similar pattern as network, capability,
etc.
mount, # allow all mounts, but not umount or pivotroot
mount fstype=procfs, # allow mounting procfs anywhere
mount options=(bind, ro) /foo -> /bar, # readonly bind mount
mount /dev/sda -> /mnt,
mount /dev/sd** -> /mnt/**,
mount fstype=overlayfs options=(rw,upperdir=/tmp/upper/,lowerdir=/) overlay -> /mnt/
umount,
umount /m*,
Currently variables and regexs are are supported on the device and mount
point. ie.
mount <devince> -> <mount point>,
Regexes are supported in fstype and options. The options have a further
caveat that regexs only work if the option is fs specific option.
eg. options=(upperdir=/tmp/*,lowerdir=/)
regex's will not currently work against the standard options like ro, rw
nosuid
Conditionals (fstype) can only be applied to the device (source) at this
time and will be disregarded in situations where the mount is manipulating
an existing mount (bind, remount).
Options can be specified multiple times
mount option=rw option=(nosuid,upperdir=/foo),
and will be combined together into a single set of values
The ordering of the standard mount options (rw,ro, ...) does not matter
but the ordering of fs specific options does.
Specifying that the value of a particular option does not matter can be
acheived by providing both the positive and negative forms of and option
option=(rw,ro) options=(suid,nosuid)
For the fs specific options specifying that a particular value does not
matter is achieve using a regex with alternations.
Improvements to the syntax and order restrictions are planned for the
future.
Signed-off-by: John Johansen <john.johansen@canonical.com>
2012-02-24 04:19:38 -08:00
|
|
|
{"make-rslave", MS_RSLAVE, 0},
|
parser: Sync mount options parsing and documentation
There are a number of differences between what the apparmor.d(5) man
page lists as valid AppArmor mount rule options and what apparmor_parser
looks for when parsing mount rules. There are also typos in the man page
and parser around mount options. Here's the breakdown of problems and
fixes made in this patch:
* The apparmor.d(5) man page improperly documented a "nodirsync"
option.
- That mount option does not exist and the parser did not honor it.
Remove the mention from the apparmor.d(5) man page.
* The loud option was typoed as "load" in both the man page and parser
- There's no sense in preserving backwards compatibility. "load" is
simply wrong and should not be honored. The man page and parser are
updated to only use "loud".
* The rbind option wasn't listed in the man page.
- Add rbind to the man page. No change needed for the parser.
* The documented unbindable, private, slave, and shared options were
not correctly parsed. The parser expected
make-{unbindable,private,slave,shared}.
- The parser is updated to accept both the documented
{unbindable,private,slave,shared} options and their variants
prefixed with "make-". The man page will not document the "make-"
variants.
* The recursive {runbindable,rprivate,rslave,rshared} options were not
documented and were only recognized by the parser if they were
prefixed with "make-".
- The man page is updated to document the option strings that are not
prefixed with "make-". The parser still accepts the "make-"
variants.
* The man page documented a "rec" option but the parser didn't honor
it. The MS_REC macro is used by the mount utility to be bitwise OR'ed
with MS_{UNBINDABLE,PRIVATE,SLAVE,SHARED} to indicate the
corresponding recursive mount options.
- This is not an option that should be exposed in the AppArmor policy
since we already allow have the
{runbindable,rprivate,rslave,rshared} options.
* The man page typoed the {no,}relatime options as {no,}relative.
- The man page is updated to document the correct option strings. The
parser requires no change.
Bug: https://bugs.launchpad.net/bugs/1401619
Signed-off-by: Tyler Hicks <tyhicks@canonical.com>
Acked-by: John Johansen <john.johansen@canonical.com>
2014-12-12 08:20:31 -06:00
|
|
|
{"shared", MS_SHARED, 0},
|
Add mount rules
Add the ability to control mounting and unmounting
The basic form of the rules are.
[audit] [deny] mount [conds]* [device] [ -> [conds] path],
[audit] [deny] remount [conds]* [path],
[audit] [deny] umount [conds]* [path],
[audit] [deny] pivotroot [oldroot=<value>] <path> -> <profile>
remount is just a short cut for mount options=remount
where [conds] can be
fstype=<expr>
options=<expr>
conds follow the extended conditional syntax of allowing either:
* a single value after the equals, which has the same character range as
regular IDS (ie most anything but it can't be terminated with a , (comma)
and if spaces or other characters are needed it can be quoted
eg.
options=foo
options = foo
options="foo bar"
* a list of values after the equals, the list of values is enclosed within
parenthesis () and its has a slightly reduced character set but again
elements can be quoted.
the separation between elements is whitespace and commas.
eg.
options=(foo bar)
options=(foo, bar)
options=(foo , bar)
options=(foo,bar)
The rules are flexible and follow a similar pattern as network, capability,
etc.
mount, # allow all mounts, but not umount or pivotroot
mount fstype=procfs, # allow mounting procfs anywhere
mount options=(bind, ro) /foo -> /bar, # readonly bind mount
mount /dev/sda -> /mnt,
mount /dev/sd** -> /mnt/**,
mount fstype=overlayfs options=(rw,upperdir=/tmp/upper/,lowerdir=/) overlay -> /mnt/
umount,
umount /m*,
Currently variables and regexs are are supported on the device and mount
point. ie.
mount <devince> -> <mount point>,
Regexes are supported in fstype and options. The options have a further
caveat that regexs only work if the option is fs specific option.
eg. options=(upperdir=/tmp/*,lowerdir=/)
regex's will not currently work against the standard options like ro, rw
nosuid
Conditionals (fstype) can only be applied to the device (source) at this
time and will be disregarded in situations where the mount is manipulating
an existing mount (bind, remount).
Options can be specified multiple times
mount option=rw option=(nosuid,upperdir=/foo),
and will be combined together into a single set of values
The ordering of the standard mount options (rw,ro, ...) does not matter
but the ordering of fs specific options does.
Specifying that the value of a particular option does not matter can be
acheived by providing both the positive and negative forms of and option
option=(rw,ro) options=(suid,nosuid)
For the fs specific options specifying that a particular value does not
matter is achieve using a regex with alternations.
Improvements to the syntax and order restrictions are planned for the
future.
Signed-off-by: John Johansen <john.johansen@canonical.com>
2012-02-24 04:19:38 -08:00
|
|
|
{"make-shared", MS_SHARED, 0},
|
parser: Sync mount options parsing and documentation
There are a number of differences between what the apparmor.d(5) man
page lists as valid AppArmor mount rule options and what apparmor_parser
looks for when parsing mount rules. There are also typos in the man page
and parser around mount options. Here's the breakdown of problems and
fixes made in this patch:
* The apparmor.d(5) man page improperly documented a "nodirsync"
option.
- That mount option does not exist and the parser did not honor it.
Remove the mention from the apparmor.d(5) man page.
* The loud option was typoed as "load" in both the man page and parser
- There's no sense in preserving backwards compatibility. "load" is
simply wrong and should not be honored. The man page and parser are
updated to only use "loud".
* The rbind option wasn't listed in the man page.
- Add rbind to the man page. No change needed for the parser.
* The documented unbindable, private, slave, and shared options were
not correctly parsed. The parser expected
make-{unbindable,private,slave,shared}.
- The parser is updated to accept both the documented
{unbindable,private,slave,shared} options and their variants
prefixed with "make-". The man page will not document the "make-"
variants.
* The recursive {runbindable,rprivate,rslave,rshared} options were not
documented and were only recognized by the parser if they were
prefixed with "make-".
- The man page is updated to document the option strings that are not
prefixed with "make-". The parser still accepts the "make-"
variants.
* The man page documented a "rec" option but the parser didn't honor
it. The MS_REC macro is used by the mount utility to be bitwise OR'ed
with MS_{UNBINDABLE,PRIVATE,SLAVE,SHARED} to indicate the
corresponding recursive mount options.
- This is not an option that should be exposed in the AppArmor policy
since we already allow have the
{runbindable,rprivate,rslave,rshared} options.
* The man page typoed the {no,}relatime options as {no,}relative.
- The man page is updated to document the correct option strings. The
parser requires no change.
Bug: https://bugs.launchpad.net/bugs/1401619
Signed-off-by: Tyler Hicks <tyhicks@canonical.com>
Acked-by: John Johansen <john.johansen@canonical.com>
2014-12-12 08:20:31 -06:00
|
|
|
{"rshared", MS_RSHARED, 0},
|
Add mount rules
Add the ability to control mounting and unmounting
The basic form of the rules are.
[audit] [deny] mount [conds]* [device] [ -> [conds] path],
[audit] [deny] remount [conds]* [path],
[audit] [deny] umount [conds]* [path],
[audit] [deny] pivotroot [oldroot=<value>] <path> -> <profile>
remount is just a short cut for mount options=remount
where [conds] can be
fstype=<expr>
options=<expr>
conds follow the extended conditional syntax of allowing either:
* a single value after the equals, which has the same character range as
regular IDS (ie most anything but it can't be terminated with a , (comma)
and if spaces or other characters are needed it can be quoted
eg.
options=foo
options = foo
options="foo bar"
* a list of values after the equals, the list of values is enclosed within
parenthesis () and its has a slightly reduced character set but again
elements can be quoted.
the separation between elements is whitespace and commas.
eg.
options=(foo bar)
options=(foo, bar)
options=(foo , bar)
options=(foo,bar)
The rules are flexible and follow a similar pattern as network, capability,
etc.
mount, # allow all mounts, but not umount or pivotroot
mount fstype=procfs, # allow mounting procfs anywhere
mount options=(bind, ro) /foo -> /bar, # readonly bind mount
mount /dev/sda -> /mnt,
mount /dev/sd** -> /mnt/**,
mount fstype=overlayfs options=(rw,upperdir=/tmp/upper/,lowerdir=/) overlay -> /mnt/
umount,
umount /m*,
Currently variables and regexs are are supported on the device and mount
point. ie.
mount <devince> -> <mount point>,
Regexes are supported in fstype and options. The options have a further
caveat that regexs only work if the option is fs specific option.
eg. options=(upperdir=/tmp/*,lowerdir=/)
regex's will not currently work against the standard options like ro, rw
nosuid
Conditionals (fstype) can only be applied to the device (source) at this
time and will be disregarded in situations where the mount is manipulating
an existing mount (bind, remount).
Options can be specified multiple times
mount option=rw option=(nosuid,upperdir=/foo),
and will be combined together into a single set of values
The ordering of the standard mount options (rw,ro, ...) does not matter
but the ordering of fs specific options does.
Specifying that the value of a particular option does not matter can be
acheived by providing both the positive and negative forms of and option
option=(rw,ro) options=(suid,nosuid)
For the fs specific options specifying that a particular value does not
matter is achieve using a regex with alternations.
Improvements to the syntax and order restrictions are planned for the
future.
Signed-off-by: John Johansen <john.johansen@canonical.com>
2012-02-24 04:19:38 -08:00
|
|
|
{"make-rshared", MS_RSHARED, 0},
|
|
|
|
|
|
|
|
{"relatime", MS_RELATIME, 0},
|
|
|
|
{"norelatime", 0, MS_NORELATIME},
|
|
|
|
{"iversion", MS_IVERSION, 0},
|
|
|
|
{"noiversion", 0, MS_IVERSION},
|
|
|
|
{"strictatime", MS_STRICTATIME, 0},
|
2013-09-27 16:13:22 -07:00
|
|
|
{"user", 0, (unsigned int) MS_NOUSER},
|
|
|
|
{"nouser", (unsigned int) MS_NOUSER, 0},
|
Add mount rules
Add the ability to control mounting and unmounting
The basic form of the rules are.
[audit] [deny] mount [conds]* [device] [ -> [conds] path],
[audit] [deny] remount [conds]* [path],
[audit] [deny] umount [conds]* [path],
[audit] [deny] pivotroot [oldroot=<value>] <path> -> <profile>
remount is just a short cut for mount options=remount
where [conds] can be
fstype=<expr>
options=<expr>
conds follow the extended conditional syntax of allowing either:
* a single value after the equals, which has the same character range as
regular IDS (ie most anything but it can't be terminated with a , (comma)
and if spaces or other characters are needed it can be quoted
eg.
options=foo
options = foo
options="foo bar"
* a list of values after the equals, the list of values is enclosed within
parenthesis () and its has a slightly reduced character set but again
elements can be quoted.
the separation between elements is whitespace and commas.
eg.
options=(foo bar)
options=(foo, bar)
options=(foo , bar)
options=(foo,bar)
The rules are flexible and follow a similar pattern as network, capability,
etc.
mount, # allow all mounts, but not umount or pivotroot
mount fstype=procfs, # allow mounting procfs anywhere
mount options=(bind, ro) /foo -> /bar, # readonly bind mount
mount /dev/sda -> /mnt,
mount /dev/sd** -> /mnt/**,
mount fstype=overlayfs options=(rw,upperdir=/tmp/upper/,lowerdir=/) overlay -> /mnt/
umount,
umount /m*,
Currently variables and regexs are are supported on the device and mount
point. ie.
mount <devince> -> <mount point>,
Regexes are supported in fstype and options. The options have a further
caveat that regexs only work if the option is fs specific option.
eg. options=(upperdir=/tmp/*,lowerdir=/)
regex's will not currently work against the standard options like ro, rw
nosuid
Conditionals (fstype) can only be applied to the device (source) at this
time and will be disregarded in situations where the mount is manipulating
an existing mount (bind, remount).
Options can be specified multiple times
mount option=rw option=(nosuid,upperdir=/foo),
and will be combined together into a single set of values
The ordering of the standard mount options (rw,ro, ...) does not matter
but the ordering of fs specific options does.
Specifying that the value of a particular option does not matter can be
acheived by providing both the positive and negative forms of and option
option=(rw,ro) options=(suid,nosuid)
For the fs specific options specifying that a particular value does not
matter is achieve using a regex with alternations.
Improvements to the syntax and order restrictions are planned for the
future.
Signed-off-by: John Johansen <john.johansen@canonical.com>
2012-02-24 04:19:38 -08:00
|
|
|
|
2012-03-09 04:21:54 -08:00
|
|
|
{NULL, 0, 0}
|
Add mount rules
Add the ability to control mounting and unmounting
The basic form of the rules are.
[audit] [deny] mount [conds]* [device] [ -> [conds] path],
[audit] [deny] remount [conds]* [path],
[audit] [deny] umount [conds]* [path],
[audit] [deny] pivotroot [oldroot=<value>] <path> -> <profile>
remount is just a short cut for mount options=remount
where [conds] can be
fstype=<expr>
options=<expr>
conds follow the extended conditional syntax of allowing either:
* a single value after the equals, which has the same character range as
regular IDS (ie most anything but it can't be terminated with a , (comma)
and if spaces or other characters are needed it can be quoted
eg.
options=foo
options = foo
options="foo bar"
* a list of values after the equals, the list of values is enclosed within
parenthesis () and its has a slightly reduced character set but again
elements can be quoted.
the separation between elements is whitespace and commas.
eg.
options=(foo bar)
options=(foo, bar)
options=(foo , bar)
options=(foo,bar)
The rules are flexible and follow a similar pattern as network, capability,
etc.
mount, # allow all mounts, but not umount or pivotroot
mount fstype=procfs, # allow mounting procfs anywhere
mount options=(bind, ro) /foo -> /bar, # readonly bind mount
mount /dev/sda -> /mnt,
mount /dev/sd** -> /mnt/**,
mount fstype=overlayfs options=(rw,upperdir=/tmp/upper/,lowerdir=/) overlay -> /mnt/
umount,
umount /m*,
Currently variables and regexs are are supported on the device and mount
point. ie.
mount <devince> -> <mount point>,
Regexes are supported in fstype and options. The options have a further
caveat that regexs only work if the option is fs specific option.
eg. options=(upperdir=/tmp/*,lowerdir=/)
regex's will not currently work against the standard options like ro, rw
nosuid
Conditionals (fstype) can only be applied to the device (source) at this
time and will be disregarded in situations where the mount is manipulating
an existing mount (bind, remount).
Options can be specified multiple times
mount option=rw option=(nosuid,upperdir=/foo),
and will be combined together into a single set of values
The ordering of the standard mount options (rw,ro, ...) does not matter
but the ordering of fs specific options does.
Specifying that the value of a particular option does not matter can be
acheived by providing both the positive and negative forms of and option
option=(rw,ro) options=(suid,nosuid)
For the fs specific options specifying that a particular value does not
matter is achieve using a regex with alternations.
Improvements to the syntax and order restrictions are planned for the
future.
Signed-off-by: John Johansen <john.johansen@canonical.com>
2012-02-24 04:19:38 -08:00
|
|
|
};
|
|
|
|
|
|
|
|
static struct mnt_keyword_table mnt_conds_table[] = {
|
|
|
|
{"options", MNT_SRC_OPT, MNT_COND_OPTIONS},
|
|
|
|
{"option", MNT_SRC_OPT, MNT_COND_OPTIONS},
|
|
|
|
{"fstype", MNT_SRC_OPT | MNT_DST_OPT, MNT_COND_FSTYPE},
|
|
|
|
{"vfstype", MNT_SRC_OPT | MNT_DST_OPT, MNT_COND_FSTYPE},
|
|
|
|
|
2012-03-09 04:21:54 -08:00
|
|
|
{NULL, 0, 0}
|
Add mount rules
Add the ability to control mounting and unmounting
The basic form of the rules are.
[audit] [deny] mount [conds]* [device] [ -> [conds] path],
[audit] [deny] remount [conds]* [path],
[audit] [deny] umount [conds]* [path],
[audit] [deny] pivotroot [oldroot=<value>] <path> -> <profile>
remount is just a short cut for mount options=remount
where [conds] can be
fstype=<expr>
options=<expr>
conds follow the extended conditional syntax of allowing either:
* a single value after the equals, which has the same character range as
regular IDS (ie most anything but it can't be terminated with a , (comma)
and if spaces or other characters are needed it can be quoted
eg.
options=foo
options = foo
options="foo bar"
* a list of values after the equals, the list of values is enclosed within
parenthesis () and its has a slightly reduced character set but again
elements can be quoted.
the separation between elements is whitespace and commas.
eg.
options=(foo bar)
options=(foo, bar)
options=(foo , bar)
options=(foo,bar)
The rules are flexible and follow a similar pattern as network, capability,
etc.
mount, # allow all mounts, but not umount or pivotroot
mount fstype=procfs, # allow mounting procfs anywhere
mount options=(bind, ro) /foo -> /bar, # readonly bind mount
mount /dev/sda -> /mnt,
mount /dev/sd** -> /mnt/**,
mount fstype=overlayfs options=(rw,upperdir=/tmp/upper/,lowerdir=/) overlay -> /mnt/
umount,
umount /m*,
Currently variables and regexs are are supported on the device and mount
point. ie.
mount <devince> -> <mount point>,
Regexes are supported in fstype and options. The options have a further
caveat that regexs only work if the option is fs specific option.
eg. options=(upperdir=/tmp/*,lowerdir=/)
regex's will not currently work against the standard options like ro, rw
nosuid
Conditionals (fstype) can only be applied to the device (source) at this
time and will be disregarded in situations where the mount is manipulating
an existing mount (bind, remount).
Options can be specified multiple times
mount option=rw option=(nosuid,upperdir=/foo),
and will be combined together into a single set of values
The ordering of the standard mount options (rw,ro, ...) does not matter
but the ordering of fs specific options does.
Specifying that the value of a particular option does not matter can be
acheived by providing both the positive and negative forms of and option
option=(rw,ro) options=(suid,nosuid)
For the fs specific options specifying that a particular value does not
matter is achieve using a regex with alternations.
Improvements to the syntax and order restrictions are planned for the
future.
Signed-off-by: John Johansen <john.johansen@canonical.com>
2012-02-24 04:19:38 -08:00
|
|
|
};
|
|
|
|
|
2019-02-12 08:15:22 -08:00
|
|
|
static ostream &dump_flags(ostream &os,
|
|
|
|
pair <unsigned int, unsigned int> flags)
|
|
|
|
{
|
|
|
|
for (int i = 0; mnt_opts_table[i].keyword; i++) {
|
|
|
|
if ((flags.first & mnt_opts_table[i].set) ||
|
|
|
|
(flags.second & mnt_opts_table[i].clear))
|
|
|
|
os << mnt_opts_table[i].keyword;
|
|
|
|
}
|
|
|
|
return os;
|
|
|
|
}
|
|
|
|
|
|
|
|
ostream &operator<<(ostream &os, pair<unsigned int, unsigned int> flags)
|
|
|
|
{
|
|
|
|
return dump_flags(os, flags);
|
|
|
|
}
|
|
|
|
|
Add mount rules
Add the ability to control mounting and unmounting
The basic form of the rules are.
[audit] [deny] mount [conds]* [device] [ -> [conds] path],
[audit] [deny] remount [conds]* [path],
[audit] [deny] umount [conds]* [path],
[audit] [deny] pivotroot [oldroot=<value>] <path> -> <profile>
remount is just a short cut for mount options=remount
where [conds] can be
fstype=<expr>
options=<expr>
conds follow the extended conditional syntax of allowing either:
* a single value after the equals, which has the same character range as
regular IDS (ie most anything but it can't be terminated with a , (comma)
and if spaces or other characters are needed it can be quoted
eg.
options=foo
options = foo
options="foo bar"
* a list of values after the equals, the list of values is enclosed within
parenthesis () and its has a slightly reduced character set but again
elements can be quoted.
the separation between elements is whitespace and commas.
eg.
options=(foo bar)
options=(foo, bar)
options=(foo , bar)
options=(foo,bar)
The rules are flexible and follow a similar pattern as network, capability,
etc.
mount, # allow all mounts, but not umount or pivotroot
mount fstype=procfs, # allow mounting procfs anywhere
mount options=(bind, ro) /foo -> /bar, # readonly bind mount
mount /dev/sda -> /mnt,
mount /dev/sd** -> /mnt/**,
mount fstype=overlayfs options=(rw,upperdir=/tmp/upper/,lowerdir=/) overlay -> /mnt/
umount,
umount /m*,
Currently variables and regexs are are supported on the device and mount
point. ie.
mount <devince> -> <mount point>,
Regexes are supported in fstype and options. The options have a further
caveat that regexs only work if the option is fs specific option.
eg. options=(upperdir=/tmp/*,lowerdir=/)
regex's will not currently work against the standard options like ro, rw
nosuid
Conditionals (fstype) can only be applied to the device (source) at this
time and will be disregarded in situations where the mount is manipulating
an existing mount (bind, remount).
Options can be specified multiple times
mount option=rw option=(nosuid,upperdir=/foo),
and will be combined together into a single set of values
The ordering of the standard mount options (rw,ro, ...) does not matter
but the ordering of fs specific options does.
Specifying that the value of a particular option does not matter can be
acheived by providing both the positive and negative forms of and option
option=(rw,ro) options=(suid,nosuid)
For the fs specific options specifying that a particular value does not
matter is achieve using a regex with alternations.
Improvements to the syntax and order restrictions are planned for the
future.
Signed-off-by: John Johansen <john.johansen@canonical.com>
2012-02-24 04:19:38 -08:00
|
|
|
static int find_mnt_keyword(struct mnt_keyword_table *table, const char *name)
|
|
|
|
{
|
|
|
|
int i;
|
|
|
|
for (i = 0; table[i].keyword; i++) {
|
|
|
|
if (strcmp(name, table[i].keyword) == 0)
|
|
|
|
return i;
|
|
|
|
}
|
|
|
|
|
|
|
|
return -1;
|
|
|
|
}
|
|
|
|
|
|
|
|
int is_valid_mnt_cond(const char *name, int src)
|
|
|
|
{
|
|
|
|
int i;
|
|
|
|
i = find_mnt_keyword(mnt_conds_table, name);
|
|
|
|
if (i != -1)
|
|
|
|
return (mnt_conds_table[i].set & src);
|
|
|
|
return -1;
|
|
|
|
}
|
|
|
|
|
|
|
|
static unsigned int extract_flags(struct value_list **list, unsigned int *inv)
|
|
|
|
{
|
2019-02-12 08:15:22 -08:00
|
|
|
unsigned int flags = 0, invflags = 0;
|
Add mount rules
Add the ability to control mounting and unmounting
The basic form of the rules are.
[audit] [deny] mount [conds]* [device] [ -> [conds] path],
[audit] [deny] remount [conds]* [path],
[audit] [deny] umount [conds]* [path],
[audit] [deny] pivotroot [oldroot=<value>] <path> -> <profile>
remount is just a short cut for mount options=remount
where [conds] can be
fstype=<expr>
options=<expr>
conds follow the extended conditional syntax of allowing either:
* a single value after the equals, which has the same character range as
regular IDS (ie most anything but it can't be terminated with a , (comma)
and if spaces or other characters are needed it can be quoted
eg.
options=foo
options = foo
options="foo bar"
* a list of values after the equals, the list of values is enclosed within
parenthesis () and its has a slightly reduced character set but again
elements can be quoted.
the separation between elements is whitespace and commas.
eg.
options=(foo bar)
options=(foo, bar)
options=(foo , bar)
options=(foo,bar)
The rules are flexible and follow a similar pattern as network, capability,
etc.
mount, # allow all mounts, but not umount or pivotroot
mount fstype=procfs, # allow mounting procfs anywhere
mount options=(bind, ro) /foo -> /bar, # readonly bind mount
mount /dev/sda -> /mnt,
mount /dev/sd** -> /mnt/**,
mount fstype=overlayfs options=(rw,upperdir=/tmp/upper/,lowerdir=/) overlay -> /mnt/
umount,
umount /m*,
Currently variables and regexs are are supported on the device and mount
point. ie.
mount <devince> -> <mount point>,
Regexes are supported in fstype and options. The options have a further
caveat that regexs only work if the option is fs specific option.
eg. options=(upperdir=/tmp/*,lowerdir=/)
regex's will not currently work against the standard options like ro, rw
nosuid
Conditionals (fstype) can only be applied to the device (source) at this
time and will be disregarded in situations where the mount is manipulating
an existing mount (bind, remount).
Options can be specified multiple times
mount option=rw option=(nosuid,upperdir=/foo),
and will be combined together into a single set of values
The ordering of the standard mount options (rw,ro, ...) does not matter
but the ordering of fs specific options does.
Specifying that the value of a particular option does not matter can be
acheived by providing both the positive and negative forms of and option
option=(rw,ro) options=(suid,nosuid)
For the fs specific options specifying that a particular value does not
matter is achieve using a regex with alternations.
Improvements to the syntax and order restrictions are planned for the
future.
Signed-off-by: John Johansen <john.johansen@canonical.com>
2012-02-24 04:19:38 -08:00
|
|
|
*inv = 0;
|
|
|
|
|
|
|
|
struct value_list *entry, *tmp, *prev = NULL;
|
|
|
|
list_for_each_safe(*list, entry, tmp) {
|
|
|
|
int i;
|
|
|
|
i = find_mnt_keyword(mnt_opts_table, entry->value);
|
|
|
|
if (i != -1) {
|
|
|
|
flags |= mnt_opts_table[i].set;
|
2019-02-12 08:15:22 -08:00
|
|
|
invflags |= mnt_opts_table[i].clear;
|
Add mount rules
Add the ability to control mounting and unmounting
The basic form of the rules are.
[audit] [deny] mount [conds]* [device] [ -> [conds] path],
[audit] [deny] remount [conds]* [path],
[audit] [deny] umount [conds]* [path],
[audit] [deny] pivotroot [oldroot=<value>] <path> -> <profile>
remount is just a short cut for mount options=remount
where [conds] can be
fstype=<expr>
options=<expr>
conds follow the extended conditional syntax of allowing either:
* a single value after the equals, which has the same character range as
regular IDS (ie most anything but it can't be terminated with a , (comma)
and if spaces or other characters are needed it can be quoted
eg.
options=foo
options = foo
options="foo bar"
* a list of values after the equals, the list of values is enclosed within
parenthesis () and its has a slightly reduced character set but again
elements can be quoted.
the separation between elements is whitespace and commas.
eg.
options=(foo bar)
options=(foo, bar)
options=(foo , bar)
options=(foo,bar)
The rules are flexible and follow a similar pattern as network, capability,
etc.
mount, # allow all mounts, but not umount or pivotroot
mount fstype=procfs, # allow mounting procfs anywhere
mount options=(bind, ro) /foo -> /bar, # readonly bind mount
mount /dev/sda -> /mnt,
mount /dev/sd** -> /mnt/**,
mount fstype=overlayfs options=(rw,upperdir=/tmp/upper/,lowerdir=/) overlay -> /mnt/
umount,
umount /m*,
Currently variables and regexs are are supported on the device and mount
point. ie.
mount <devince> -> <mount point>,
Regexes are supported in fstype and options. The options have a further
caveat that regexs only work if the option is fs specific option.
eg. options=(upperdir=/tmp/*,lowerdir=/)
regex's will not currently work against the standard options like ro, rw
nosuid
Conditionals (fstype) can only be applied to the device (source) at this
time and will be disregarded in situations where the mount is manipulating
an existing mount (bind, remount).
Options can be specified multiple times
mount option=rw option=(nosuid,upperdir=/foo),
and will be combined together into a single set of values
The ordering of the standard mount options (rw,ro, ...) does not matter
but the ordering of fs specific options does.
Specifying that the value of a particular option does not matter can be
acheived by providing both the positive and negative forms of and option
option=(rw,ro) options=(suid,nosuid)
For the fs specific options specifying that a particular value does not
matter is achieve using a regex with alternations.
Improvements to the syntax and order restrictions are planned for the
future.
Signed-off-by: John Johansen <john.johansen@canonical.com>
2012-02-24 04:19:38 -08:00
|
|
|
PDEBUG(" extracting mount flag %s req: 0x%x inv: 0x%x"
|
|
|
|
" => req: 0x%x inv: 0x%x\n",
|
|
|
|
entry->value, mnt_opts_table[i].set,
|
2019-02-12 08:15:22 -08:00
|
|
|
mnt_opts_table[i].clear, flags, invflags);
|
Add mount rules
Add the ability to control mounting and unmounting
The basic form of the rules are.
[audit] [deny] mount [conds]* [device] [ -> [conds] path],
[audit] [deny] remount [conds]* [path],
[audit] [deny] umount [conds]* [path],
[audit] [deny] pivotroot [oldroot=<value>] <path> -> <profile>
remount is just a short cut for mount options=remount
where [conds] can be
fstype=<expr>
options=<expr>
conds follow the extended conditional syntax of allowing either:
* a single value after the equals, which has the same character range as
regular IDS (ie most anything but it can't be terminated with a , (comma)
and if spaces or other characters are needed it can be quoted
eg.
options=foo
options = foo
options="foo bar"
* a list of values after the equals, the list of values is enclosed within
parenthesis () and its has a slightly reduced character set but again
elements can be quoted.
the separation between elements is whitespace and commas.
eg.
options=(foo bar)
options=(foo, bar)
options=(foo , bar)
options=(foo,bar)
The rules are flexible and follow a similar pattern as network, capability,
etc.
mount, # allow all mounts, but not umount or pivotroot
mount fstype=procfs, # allow mounting procfs anywhere
mount options=(bind, ro) /foo -> /bar, # readonly bind mount
mount /dev/sda -> /mnt,
mount /dev/sd** -> /mnt/**,
mount fstype=overlayfs options=(rw,upperdir=/tmp/upper/,lowerdir=/) overlay -> /mnt/
umount,
umount /m*,
Currently variables and regexs are are supported on the device and mount
point. ie.
mount <devince> -> <mount point>,
Regexes are supported in fstype and options. The options have a further
caveat that regexs only work if the option is fs specific option.
eg. options=(upperdir=/tmp/*,lowerdir=/)
regex's will not currently work against the standard options like ro, rw
nosuid
Conditionals (fstype) can only be applied to the device (source) at this
time and will be disregarded in situations where the mount is manipulating
an existing mount (bind, remount).
Options can be specified multiple times
mount option=rw option=(nosuid,upperdir=/foo),
and will be combined together into a single set of values
The ordering of the standard mount options (rw,ro, ...) does not matter
but the ordering of fs specific options does.
Specifying that the value of a particular option does not matter can be
acheived by providing both the positive and negative forms of and option
option=(rw,ro) options=(suid,nosuid)
For the fs specific options specifying that a particular value does not
matter is achieve using a regex with alternations.
Improvements to the syntax and order restrictions are planned for the
future.
Signed-off-by: John Johansen <john.johansen@canonical.com>
2012-02-24 04:19:38 -08:00
|
|
|
if (prev)
|
|
|
|
prev->next = tmp;
|
|
|
|
if (entry == *list)
|
|
|
|
*list = tmp;
|
|
|
|
entry->next = NULL;
|
|
|
|
free_value_list(entry);
|
|
|
|
} else
|
|
|
|
prev = entry;
|
|
|
|
}
|
|
|
|
|
2019-02-12 08:15:22 -08:00
|
|
|
if (inv)
|
|
|
|
*inv = invflags;
|
|
|
|
|
Add mount rules
Add the ability to control mounting and unmounting
The basic form of the rules are.
[audit] [deny] mount [conds]* [device] [ -> [conds] path],
[audit] [deny] remount [conds]* [path],
[audit] [deny] umount [conds]* [path],
[audit] [deny] pivotroot [oldroot=<value>] <path> -> <profile>
remount is just a short cut for mount options=remount
where [conds] can be
fstype=<expr>
options=<expr>
conds follow the extended conditional syntax of allowing either:
* a single value after the equals, which has the same character range as
regular IDS (ie most anything but it can't be terminated with a , (comma)
and if spaces or other characters are needed it can be quoted
eg.
options=foo
options = foo
options="foo bar"
* a list of values after the equals, the list of values is enclosed within
parenthesis () and its has a slightly reduced character set but again
elements can be quoted.
the separation between elements is whitespace and commas.
eg.
options=(foo bar)
options=(foo, bar)
options=(foo , bar)
options=(foo,bar)
The rules are flexible and follow a similar pattern as network, capability,
etc.
mount, # allow all mounts, but not umount or pivotroot
mount fstype=procfs, # allow mounting procfs anywhere
mount options=(bind, ro) /foo -> /bar, # readonly bind mount
mount /dev/sda -> /mnt,
mount /dev/sd** -> /mnt/**,
mount fstype=overlayfs options=(rw,upperdir=/tmp/upper/,lowerdir=/) overlay -> /mnt/
umount,
umount /m*,
Currently variables and regexs are are supported on the device and mount
point. ie.
mount <devince> -> <mount point>,
Regexes are supported in fstype and options. The options have a further
caveat that regexs only work if the option is fs specific option.
eg. options=(upperdir=/tmp/*,lowerdir=/)
regex's will not currently work against the standard options like ro, rw
nosuid
Conditionals (fstype) can only be applied to the device (source) at this
time and will be disregarded in situations where the mount is manipulating
an existing mount (bind, remount).
Options can be specified multiple times
mount option=rw option=(nosuid,upperdir=/foo),
and will be combined together into a single set of values
The ordering of the standard mount options (rw,ro, ...) does not matter
but the ordering of fs specific options does.
Specifying that the value of a particular option does not matter can be
acheived by providing both the positive and negative forms of and option
option=(rw,ro) options=(suid,nosuid)
For the fs specific options specifying that a particular value does not
matter is achieve using a regex with alternations.
Improvements to the syntax and order restrictions are planned for the
future.
Signed-off-by: John Johansen <john.johansen@canonical.com>
2012-02-24 04:19:38 -08:00
|
|
|
return flags;
|
|
|
|
}
|
|
|
|
|
2019-02-12 08:15:22 -08:00
|
|
|
static bool conflicting_flags(unsigned int flags, unsigned int inv)
|
|
|
|
{
|
|
|
|
if (flags & inv) {
|
|
|
|
for (int i = 0; i < 31; i++) {
|
|
|
|
unsigned int mask = 1 << i;
|
|
|
|
if ((flags & inv) & mask) {
|
|
|
|
cerr << "conflicting flag value = ";
|
|
|
|
cerr << make_pair(flags, inv);
|
|
|
|
cerr << "\n";
|
|
|
|
}
|
|
|
|
}
|
|
|
|
return true;
|
|
|
|
}
|
|
|
|
return false;
|
|
|
|
}
|
|
|
|
|
Add mount rules
Add the ability to control mounting and unmounting
The basic form of the rules are.
[audit] [deny] mount [conds]* [device] [ -> [conds] path],
[audit] [deny] remount [conds]* [path],
[audit] [deny] umount [conds]* [path],
[audit] [deny] pivotroot [oldroot=<value>] <path> -> <profile>
remount is just a short cut for mount options=remount
where [conds] can be
fstype=<expr>
options=<expr>
conds follow the extended conditional syntax of allowing either:
* a single value after the equals, which has the same character range as
regular IDS (ie most anything but it can't be terminated with a , (comma)
and if spaces or other characters are needed it can be quoted
eg.
options=foo
options = foo
options="foo bar"
* a list of values after the equals, the list of values is enclosed within
parenthesis () and its has a slightly reduced character set but again
elements can be quoted.
the separation between elements is whitespace and commas.
eg.
options=(foo bar)
options=(foo, bar)
options=(foo , bar)
options=(foo,bar)
The rules are flexible and follow a similar pattern as network, capability,
etc.
mount, # allow all mounts, but not umount or pivotroot
mount fstype=procfs, # allow mounting procfs anywhere
mount options=(bind, ro) /foo -> /bar, # readonly bind mount
mount /dev/sda -> /mnt,
mount /dev/sd** -> /mnt/**,
mount fstype=overlayfs options=(rw,upperdir=/tmp/upper/,lowerdir=/) overlay -> /mnt/
umount,
umount /m*,
Currently variables and regexs are are supported on the device and mount
point. ie.
mount <devince> -> <mount point>,
Regexes are supported in fstype and options. The options have a further
caveat that regexs only work if the option is fs specific option.
eg. options=(upperdir=/tmp/*,lowerdir=/)
regex's will not currently work against the standard options like ro, rw
nosuid
Conditionals (fstype) can only be applied to the device (source) at this
time and will be disregarded in situations where the mount is manipulating
an existing mount (bind, remount).
Options can be specified multiple times
mount option=rw option=(nosuid,upperdir=/foo),
and will be combined together into a single set of values
The ordering of the standard mount options (rw,ro, ...) does not matter
but the ordering of fs specific options does.
Specifying that the value of a particular option does not matter can be
acheived by providing both the positive and negative forms of and option
option=(rw,ro) options=(suid,nosuid)
For the fs specific options specifying that a particular value does not
matter is achieve using a regex with alternations.
Improvements to the syntax and order restrictions are planned for the
future.
Signed-off-by: John Johansen <john.johansen@canonical.com>
2012-02-24 04:19:38 -08:00
|
|
|
static struct value_list *extract_fstype(struct cond_entry **conds)
|
|
|
|
{
|
|
|
|
struct value_list *list = NULL;
|
|
|
|
|
|
|
|
struct cond_entry *entry, *tmp, *prev = NULL;
|
|
|
|
|
|
|
|
list_for_each_safe(*conds, entry, tmp) {
|
|
|
|
if (strcmp(entry->name, "fstype") == 0 ||
|
|
|
|
strcmp(entry->name, "vfstype") == 0) {
|
|
|
|
PDEBUG(" extracting fstype\n");
|
2014-04-07 03:18:33 -07:00
|
|
|
list_remove_at(*conds, prev, entry);
|
Add mount rules
Add the ability to control mounting and unmounting
The basic form of the rules are.
[audit] [deny] mount [conds]* [device] [ -> [conds] path],
[audit] [deny] remount [conds]* [path],
[audit] [deny] umount [conds]* [path],
[audit] [deny] pivotroot [oldroot=<value>] <path> -> <profile>
remount is just a short cut for mount options=remount
where [conds] can be
fstype=<expr>
options=<expr>
conds follow the extended conditional syntax of allowing either:
* a single value after the equals, which has the same character range as
regular IDS (ie most anything but it can't be terminated with a , (comma)
and if spaces or other characters are needed it can be quoted
eg.
options=foo
options = foo
options="foo bar"
* a list of values after the equals, the list of values is enclosed within
parenthesis () and its has a slightly reduced character set but again
elements can be quoted.
the separation between elements is whitespace and commas.
eg.
options=(foo bar)
options=(foo, bar)
options=(foo , bar)
options=(foo,bar)
The rules are flexible and follow a similar pattern as network, capability,
etc.
mount, # allow all mounts, but not umount or pivotroot
mount fstype=procfs, # allow mounting procfs anywhere
mount options=(bind, ro) /foo -> /bar, # readonly bind mount
mount /dev/sda -> /mnt,
mount /dev/sd** -> /mnt/**,
mount fstype=overlayfs options=(rw,upperdir=/tmp/upper/,lowerdir=/) overlay -> /mnt/
umount,
umount /m*,
Currently variables and regexs are are supported on the device and mount
point. ie.
mount <devince> -> <mount point>,
Regexes are supported in fstype and options. The options have a further
caveat that regexs only work if the option is fs specific option.
eg. options=(upperdir=/tmp/*,lowerdir=/)
regex's will not currently work against the standard options like ro, rw
nosuid
Conditionals (fstype) can only be applied to the device (source) at this
time and will be disregarded in situations where the mount is manipulating
an existing mount (bind, remount).
Options can be specified multiple times
mount option=rw option=(nosuid,upperdir=/foo),
and will be combined together into a single set of values
The ordering of the standard mount options (rw,ro, ...) does not matter
but the ordering of fs specific options does.
Specifying that the value of a particular option does not matter can be
acheived by providing both the positive and negative forms of and option
option=(rw,ro) options=(suid,nosuid)
For the fs specific options specifying that a particular value does not
matter is achieve using a regex with alternations.
Improvements to the syntax and order restrictions are planned for the
future.
Signed-off-by: John Johansen <john.johansen@canonical.com>
2012-02-24 04:19:38 -08:00
|
|
|
list_append(entry->vals, list);
|
|
|
|
list = entry->vals;
|
|
|
|
entry->vals = NULL;
|
|
|
|
free_cond_entry(entry);
|
|
|
|
} else
|
|
|
|
prev = entry;
|
|
|
|
}
|
|
|
|
|
|
|
|
return list;
|
|
|
|
}
|
|
|
|
|
2023-03-28 04:23:20 -07:00
|
|
|
static struct cond_entry *extract_options(struct cond_entry **conds, int eq)
|
Add mount rules
Add the ability to control mounting and unmounting
The basic form of the rules are.
[audit] [deny] mount [conds]* [device] [ -> [conds] path],
[audit] [deny] remount [conds]* [path],
[audit] [deny] umount [conds]* [path],
[audit] [deny] pivotroot [oldroot=<value>] <path> -> <profile>
remount is just a short cut for mount options=remount
where [conds] can be
fstype=<expr>
options=<expr>
conds follow the extended conditional syntax of allowing either:
* a single value after the equals, which has the same character range as
regular IDS (ie most anything but it can't be terminated with a , (comma)
and if spaces or other characters are needed it can be quoted
eg.
options=foo
options = foo
options="foo bar"
* a list of values after the equals, the list of values is enclosed within
parenthesis () and its has a slightly reduced character set but again
elements can be quoted.
the separation between elements is whitespace and commas.
eg.
options=(foo bar)
options=(foo, bar)
options=(foo , bar)
options=(foo,bar)
The rules are flexible and follow a similar pattern as network, capability,
etc.
mount, # allow all mounts, but not umount or pivotroot
mount fstype=procfs, # allow mounting procfs anywhere
mount options=(bind, ro) /foo -> /bar, # readonly bind mount
mount /dev/sda -> /mnt,
mount /dev/sd** -> /mnt/**,
mount fstype=overlayfs options=(rw,upperdir=/tmp/upper/,lowerdir=/) overlay -> /mnt/
umount,
umount /m*,
Currently variables and regexs are are supported on the device and mount
point. ie.
mount <devince> -> <mount point>,
Regexes are supported in fstype and options. The options have a further
caveat that regexs only work if the option is fs specific option.
eg. options=(upperdir=/tmp/*,lowerdir=/)
regex's will not currently work against the standard options like ro, rw
nosuid
Conditionals (fstype) can only be applied to the device (source) at this
time and will be disregarded in situations where the mount is manipulating
an existing mount (bind, remount).
Options can be specified multiple times
mount option=rw option=(nosuid,upperdir=/foo),
and will be combined together into a single set of values
The ordering of the standard mount options (rw,ro, ...) does not matter
but the ordering of fs specific options does.
Specifying that the value of a particular option does not matter can be
acheived by providing both the positive and negative forms of and option
option=(rw,ro) options=(suid,nosuid)
For the fs specific options specifying that a particular value does not
matter is achieve using a regex with alternations.
Improvements to the syntax and order restrictions are planned for the
future.
Signed-off-by: John Johansen <john.johansen@canonical.com>
2012-02-24 04:19:38 -08:00
|
|
|
{
|
2023-03-28 04:23:20 -07:00
|
|
|
struct cond_entry *list = NULL, *entry, *tmp, *prev = NULL;
|
Add mount rules
Add the ability to control mounting and unmounting
The basic form of the rules are.
[audit] [deny] mount [conds]* [device] [ -> [conds] path],
[audit] [deny] remount [conds]* [path],
[audit] [deny] umount [conds]* [path],
[audit] [deny] pivotroot [oldroot=<value>] <path> -> <profile>
remount is just a short cut for mount options=remount
where [conds] can be
fstype=<expr>
options=<expr>
conds follow the extended conditional syntax of allowing either:
* a single value after the equals, which has the same character range as
regular IDS (ie most anything but it can't be terminated with a , (comma)
and if spaces or other characters are needed it can be quoted
eg.
options=foo
options = foo
options="foo bar"
* a list of values after the equals, the list of values is enclosed within
parenthesis () and its has a slightly reduced character set but again
elements can be quoted.
the separation between elements is whitespace and commas.
eg.
options=(foo bar)
options=(foo, bar)
options=(foo , bar)
options=(foo,bar)
The rules are flexible and follow a similar pattern as network, capability,
etc.
mount, # allow all mounts, but not umount or pivotroot
mount fstype=procfs, # allow mounting procfs anywhere
mount options=(bind, ro) /foo -> /bar, # readonly bind mount
mount /dev/sda -> /mnt,
mount /dev/sd** -> /mnt/**,
mount fstype=overlayfs options=(rw,upperdir=/tmp/upper/,lowerdir=/) overlay -> /mnt/
umount,
umount /m*,
Currently variables and regexs are are supported on the device and mount
point. ie.
mount <devince> -> <mount point>,
Regexes are supported in fstype and options. The options have a further
caveat that regexs only work if the option is fs specific option.
eg. options=(upperdir=/tmp/*,lowerdir=/)
regex's will not currently work against the standard options like ro, rw
nosuid
Conditionals (fstype) can only be applied to the device (source) at this
time and will be disregarded in situations where the mount is manipulating
an existing mount (bind, remount).
Options can be specified multiple times
mount option=rw option=(nosuid,upperdir=/foo),
and will be combined together into a single set of values
The ordering of the standard mount options (rw,ro, ...) does not matter
but the ordering of fs specific options does.
Specifying that the value of a particular option does not matter can be
acheived by providing both the positive and negative forms of and option
option=(rw,ro) options=(suid,nosuid)
For the fs specific options specifying that a particular value does not
matter is achieve using a regex with alternations.
Improvements to the syntax and order restrictions are planned for the
future.
Signed-off-by: John Johansen <john.johansen@canonical.com>
2012-02-24 04:19:38 -08:00
|
|
|
|
|
|
|
list_for_each_safe(*conds, entry, tmp) {
|
2012-03-26 06:19:21 -07:00
|
|
|
if ((strcmp(entry->name, "options") == 0 ||
|
|
|
|
strcmp(entry->name, "option") == 0) &&
|
|
|
|
entry->eq == eq) {
|
2014-04-07 03:18:33 -07:00
|
|
|
list_remove_at(*conds, prev, entry);
|
2023-03-28 14:35:24 -07:00
|
|
|
PDEBUG(" extracting %s %s\n", entry->name, entry->eq ?
|
|
|
|
"=" : "in");
|
2023-03-28 04:23:20 -07:00
|
|
|
list_append(entry, list);
|
|
|
|
list = entry;
|
Add mount rules
Add the ability to control mounting and unmounting
The basic form of the rules are.
[audit] [deny] mount [conds]* [device] [ -> [conds] path],
[audit] [deny] remount [conds]* [path],
[audit] [deny] umount [conds]* [path],
[audit] [deny] pivotroot [oldroot=<value>] <path> -> <profile>
remount is just a short cut for mount options=remount
where [conds] can be
fstype=<expr>
options=<expr>
conds follow the extended conditional syntax of allowing either:
* a single value after the equals, which has the same character range as
regular IDS (ie most anything but it can't be terminated with a , (comma)
and if spaces or other characters are needed it can be quoted
eg.
options=foo
options = foo
options="foo bar"
* a list of values after the equals, the list of values is enclosed within
parenthesis () and its has a slightly reduced character set but again
elements can be quoted.
the separation between elements is whitespace and commas.
eg.
options=(foo bar)
options=(foo, bar)
options=(foo , bar)
options=(foo,bar)
The rules are flexible and follow a similar pattern as network, capability,
etc.
mount, # allow all mounts, but not umount or pivotroot
mount fstype=procfs, # allow mounting procfs anywhere
mount options=(bind, ro) /foo -> /bar, # readonly bind mount
mount /dev/sda -> /mnt,
mount /dev/sd** -> /mnt/**,
mount fstype=overlayfs options=(rw,upperdir=/tmp/upper/,lowerdir=/) overlay -> /mnt/
umount,
umount /m*,
Currently variables and regexs are are supported on the device and mount
point. ie.
mount <devince> -> <mount point>,
Regexes are supported in fstype and options. The options have a further
caveat that regexs only work if the option is fs specific option.
eg. options=(upperdir=/tmp/*,lowerdir=/)
regex's will not currently work against the standard options like ro, rw
nosuid
Conditionals (fstype) can only be applied to the device (source) at this
time and will be disregarded in situations where the mount is manipulating
an existing mount (bind, remount).
Options can be specified multiple times
mount option=rw option=(nosuid,upperdir=/foo),
and will be combined together into a single set of values
The ordering of the standard mount options (rw,ro, ...) does not matter
but the ordering of fs specific options does.
Specifying that the value of a particular option does not matter can be
acheived by providing both the positive and negative forms of and option
option=(rw,ro) options=(suid,nosuid)
For the fs specific options specifying that a particular value does not
matter is achieve using a regex with alternations.
Improvements to the syntax and order restrictions are planned for the
future.
Signed-off-by: John Johansen <john.johansen@canonical.com>
2012-02-24 04:19:38 -08:00
|
|
|
} else
|
|
|
|
prev = entry;
|
|
|
|
}
|
|
|
|
|
|
|
|
return list;
|
|
|
|
}
|
|
|
|
|
2023-03-28 04:23:20 -07:00
|
|
|
static void perror_conds(const char *rule, struct cond_entry *conds)
|
|
|
|
{
|
|
|
|
struct cond_entry *entry;
|
|
|
|
|
|
|
|
list_for_each(conds, entry) {
|
|
|
|
PERROR( "unsupported %s condition '%s%s(...)'\n", rule, entry->name, entry->eq ? "=" : " in ");
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
static void perror_vals(const char *rule, struct value_list *vals)
|
|
|
|
{
|
|
|
|
struct value_list *entry;
|
|
|
|
|
|
|
|
list_for_each(vals, entry) {
|
|
|
|
PERROR( "unsupported %s value '%s'\n", rule, entry->value);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
static void process_one_option(struct cond_entry *&opts, unsigned int &flags,
|
|
|
|
unsigned int &inv_flags)
|
|
|
|
{
|
|
|
|
struct cond_entry *entry;
|
|
|
|
struct value_list *vals;
|
|
|
|
|
parser: support multiple mount conditionals in a single rule
Now that flag processing for mount rules with single option
conditionals are fixed e-enable multiple mount conditionals on a
single mount rule. The mount conditionals are equivalent to specifying
multiple rules.
mount options=(a,b,c) options=(c,d),
is the same as
mount options=(a,b,c),
mount options=(c,d),
and
mount options in (a,b,c) options in (c,d),
is the same as
mount options in (a,b,c),
mount options in (c,d),
when multiple options= and options in are combined in a single rule
it is the same as the cross product of the options.
where
mount options=(a,b,c) options in (d,e),
is a single rule.
mount options=(a,b,c) options=(d,e) options in (f),
is equivalent to
mount options=(a,b,c) options in (f),
mount options=(d,e) options in (f),
and while it is not recommended that multiple options= and options in
conditions be used in a single rule.
mount options=(a,b,c) options=(d,e) options in (f) options in (g),
is equivalent to
mount options=(a,b,c) options in (f),
mount options=(a,b,c) options in (g),
mount options=(d,e) options in (f),
mount options=(d,e) options in (g),
Bug Link: https://bugs.launchpad.net/apparmor/+bug/1597017
Signed-off-by: John Johansen <john.johansen@canonical.com>
- rebased to bba1a023bf
- fixed infinite loop in mnt_rule::gen_policy_re
Signed-off-by: Alexander Mikhalitsyn <aleksandr.mikhalitsyn@canonical.com>
Acked-by: John Johansen <john.johansen@canonical.com>
2023-03-28 12:43:18 -07:00
|
|
|
entry = list_pop(opts);
|
2023-03-28 04:23:20 -07:00
|
|
|
vals = entry->vals;
|
|
|
|
entry->vals = NULL;
|
|
|
|
/* fail if there are any unknown optional flags */
|
|
|
|
if (opts) {
|
|
|
|
PERROR(" unsupported multiple 'mount options %s(...)'\n", entry->eq ? "=" : " in ");
|
|
|
|
exit(1);
|
|
|
|
}
|
|
|
|
free_cond_entry(entry);
|
|
|
|
|
|
|
|
flags = extract_flags(&vals, &inv_flags);
|
|
|
|
if (vals) {
|
|
|
|
perror_vals("mount option", vals);
|
|
|
|
exit(1);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2014-04-07 03:16:50 -07:00
|
|
|
mnt_rule::mnt_rule(struct cond_entry *src_conds, char *device_p,
|
2014-10-02 12:58:54 -07:00
|
|
|
struct cond_entry *dst_conds unused, char *mnt_point_p,
|
2021-06-11 03:20:18 -07:00
|
|
|
perms_t perms_p):
|
2021-09-10 13:37:54 -07:00
|
|
|
perms_rule_t(AA_CLASS_MOUNT),
|
2014-04-07 03:16:50 -07:00
|
|
|
mnt_point(mnt_point_p), device(device_p), trans(NULL), opts(NULL),
|
2021-09-04 03:28:18 -07:00
|
|
|
flagsv(0), opt_flagsv(0)
|
Add mount rules
Add the ability to control mounting and unmounting
The basic form of the rules are.
[audit] [deny] mount [conds]* [device] [ -> [conds] path],
[audit] [deny] remount [conds]* [path],
[audit] [deny] umount [conds]* [path],
[audit] [deny] pivotroot [oldroot=<value>] <path> -> <profile>
remount is just a short cut for mount options=remount
where [conds] can be
fstype=<expr>
options=<expr>
conds follow the extended conditional syntax of allowing either:
* a single value after the equals, which has the same character range as
regular IDS (ie most anything but it can't be terminated with a , (comma)
and if spaces or other characters are needed it can be quoted
eg.
options=foo
options = foo
options="foo bar"
* a list of values after the equals, the list of values is enclosed within
parenthesis () and its has a slightly reduced character set but again
elements can be quoted.
the separation between elements is whitespace and commas.
eg.
options=(foo bar)
options=(foo, bar)
options=(foo , bar)
options=(foo,bar)
The rules are flexible and follow a similar pattern as network, capability,
etc.
mount, # allow all mounts, but not umount or pivotroot
mount fstype=procfs, # allow mounting procfs anywhere
mount options=(bind, ro) /foo -> /bar, # readonly bind mount
mount /dev/sda -> /mnt,
mount /dev/sd** -> /mnt/**,
mount fstype=overlayfs options=(rw,upperdir=/tmp/upper/,lowerdir=/) overlay -> /mnt/
umount,
umount /m*,
Currently variables and regexs are are supported on the device and mount
point. ie.
mount <devince> -> <mount point>,
Regexes are supported in fstype and options. The options have a further
caveat that regexs only work if the option is fs specific option.
eg. options=(upperdir=/tmp/*,lowerdir=/)
regex's will not currently work against the standard options like ro, rw
nosuid
Conditionals (fstype) can only be applied to the device (source) at this
time and will be disregarded in situations where the mount is manipulating
an existing mount (bind, remount).
Options can be specified multiple times
mount option=rw option=(nosuid,upperdir=/foo),
and will be combined together into a single set of values
The ordering of the standard mount options (rw,ro, ...) does not matter
but the ordering of fs specific options does.
Specifying that the value of a particular option does not matter can be
acheived by providing both the positive and negative forms of and option
option=(rw,ro) options=(suid,nosuid)
For the fs specific options specifying that a particular value does not
matter is achieve using a regex with alternations.
Improvements to the syntax and order restrictions are planned for the
future.
Signed-off-by: John Johansen <john.johansen@canonical.com>
2012-02-24 04:19:38 -08:00
|
|
|
{
|
|
|
|
/* FIXME: dst_conds are ignored atm */
|
2014-04-07 03:16:50 -07:00
|
|
|
dev_type = extract_fstype(&src_conds);
|
Add mount rules
Add the ability to control mounting and unmounting
The basic form of the rules are.
[audit] [deny] mount [conds]* [device] [ -> [conds] path],
[audit] [deny] remount [conds]* [path],
[audit] [deny] umount [conds]* [path],
[audit] [deny] pivotroot [oldroot=<value>] <path> -> <profile>
remount is just a short cut for mount options=remount
where [conds] can be
fstype=<expr>
options=<expr>
conds follow the extended conditional syntax of allowing either:
* a single value after the equals, which has the same character range as
regular IDS (ie most anything but it can't be terminated with a , (comma)
and if spaces or other characters are needed it can be quoted
eg.
options=foo
options = foo
options="foo bar"
* a list of values after the equals, the list of values is enclosed within
parenthesis () and its has a slightly reduced character set but again
elements can be quoted.
the separation between elements is whitespace and commas.
eg.
options=(foo bar)
options=(foo, bar)
options=(foo , bar)
options=(foo,bar)
The rules are flexible and follow a similar pattern as network, capability,
etc.
mount, # allow all mounts, but not umount or pivotroot
mount fstype=procfs, # allow mounting procfs anywhere
mount options=(bind, ro) /foo -> /bar, # readonly bind mount
mount /dev/sda -> /mnt,
mount /dev/sd** -> /mnt/**,
mount fstype=overlayfs options=(rw,upperdir=/tmp/upper/,lowerdir=/) overlay -> /mnt/
umount,
umount /m*,
Currently variables and regexs are are supported on the device and mount
point. ie.
mount <devince> -> <mount point>,
Regexes are supported in fstype and options. The options have a further
caveat that regexs only work if the option is fs specific option.
eg. options=(upperdir=/tmp/*,lowerdir=/)
regex's will not currently work against the standard options like ro, rw
nosuid
Conditionals (fstype) can only be applied to the device (source) at this
time and will be disregarded in situations where the mount is manipulating
an existing mount (bind, remount).
Options can be specified multiple times
mount option=rw option=(nosuid,upperdir=/foo),
and will be combined together into a single set of values
The ordering of the standard mount options (rw,ro, ...) does not matter
but the ordering of fs specific options does.
Specifying that the value of a particular option does not matter can be
acheived by providing both the positive and negative forms of and option
option=(rw,ro) options=(suid,nosuid)
For the fs specific options specifying that a particular value does not
matter is achieve using a regex with alternations.
Improvements to the syntax and order restrictions are planned for the
future.
Signed-off-by: John Johansen <john.johansen@canonical.com>
2012-02-24 04:19:38 -08:00
|
|
|
|
2014-04-07 03:16:50 -07:00
|
|
|
if (src_conds) {
|
2023-03-28 04:23:20 -07:00
|
|
|
/* move options in () to local list */
|
|
|
|
struct cond_entry *opts_in = extract_options(&src_conds, 0);
|
Add mount rules
Add the ability to control mounting and unmounting
The basic form of the rules are.
[audit] [deny] mount [conds]* [device] [ -> [conds] path],
[audit] [deny] remount [conds]* [path],
[audit] [deny] umount [conds]* [path],
[audit] [deny] pivotroot [oldroot=<value>] <path> -> <profile>
remount is just a short cut for mount options=remount
where [conds] can be
fstype=<expr>
options=<expr>
conds follow the extended conditional syntax of allowing either:
* a single value after the equals, which has the same character range as
regular IDS (ie most anything but it can't be terminated with a , (comma)
and if spaces or other characters are needed it can be quoted
eg.
options=foo
options = foo
options="foo bar"
* a list of values after the equals, the list of values is enclosed within
parenthesis () and its has a slightly reduced character set but again
elements can be quoted.
the separation between elements is whitespace and commas.
eg.
options=(foo bar)
options=(foo, bar)
options=(foo , bar)
options=(foo,bar)
The rules are flexible and follow a similar pattern as network, capability,
etc.
mount, # allow all mounts, but not umount or pivotroot
mount fstype=procfs, # allow mounting procfs anywhere
mount options=(bind, ro) /foo -> /bar, # readonly bind mount
mount /dev/sda -> /mnt,
mount /dev/sd** -> /mnt/**,
mount fstype=overlayfs options=(rw,upperdir=/tmp/upper/,lowerdir=/) overlay -> /mnt/
umount,
umount /m*,
Currently variables and regexs are are supported on the device and mount
point. ie.
mount <devince> -> <mount point>,
Regexes are supported in fstype and options. The options have a further
caveat that regexs only work if the option is fs specific option.
eg. options=(upperdir=/tmp/*,lowerdir=/)
regex's will not currently work against the standard options like ro, rw
nosuid
Conditionals (fstype) can only be applied to the device (source) at this
time and will be disregarded in situations where the mount is manipulating
an existing mount (bind, remount).
Options can be specified multiple times
mount option=rw option=(nosuid,upperdir=/foo),
and will be combined together into a single set of values
The ordering of the standard mount options (rw,ro, ...) does not matter
but the ordering of fs specific options does.
Specifying that the value of a particular option does not matter can be
acheived by providing both the positive and negative forms of and option
option=(rw,ro) options=(suid,nosuid)
For the fs specific options specifying that a particular value does not
matter is achieve using a regex with alternations.
Improvements to the syntax and order restrictions are planned for the
future.
Signed-off-by: John Johansen <john.johansen@canonical.com>
2012-02-24 04:19:38 -08:00
|
|
|
|
2023-03-28 04:23:20 -07:00
|
|
|
if (opts_in) {
|
2019-02-12 08:15:22 -08:00
|
|
|
unsigned int tmpflags = 0, tmpinv_flags = 0;
|
parser: support multiple mount conditionals in a single rule
Now that flag processing for mount rules with single option
conditionals are fixed e-enable multiple mount conditionals on a
single mount rule. The mount conditionals are equivalent to specifying
multiple rules.
mount options=(a,b,c) options=(c,d),
is the same as
mount options=(a,b,c),
mount options=(c,d),
and
mount options in (a,b,c) options in (c,d),
is the same as
mount options in (a,b,c),
mount options in (c,d),
when multiple options= and options in are combined in a single rule
it is the same as the cross product of the options.
where
mount options=(a,b,c) options in (d,e),
is a single rule.
mount options=(a,b,c) options=(d,e) options in (f),
is equivalent to
mount options=(a,b,c) options in (f),
mount options=(d,e) options in (f),
and while it is not recommended that multiple options= and options in
conditions be used in a single rule.
mount options=(a,b,c) options=(d,e) options in (f) options in (g),
is equivalent to
mount options=(a,b,c) options in (f),
mount options=(a,b,c) options in (g),
mount options=(d,e) options in (f),
mount options=(d,e) options in (g),
Bug Link: https://bugs.launchpad.net/apparmor/+bug/1597017
Signed-off-by: John Johansen <john.johansen@canonical.com>
- rebased to bba1a023bf
- fixed infinite loop in mnt_rule::gen_policy_re
Signed-off-by: Alexander Mikhalitsyn <aleksandr.mikhalitsyn@canonical.com>
Acked-by: John Johansen <john.johansen@canonical.com>
2023-03-28 12:43:18 -07:00
|
|
|
struct cond_entry *entry;
|
|
|
|
|
|
|
|
while ((entry = list_pop(opts_in))) {
|
|
|
|
process_one_option(entry, tmpflags,
|
|
|
|
tmpinv_flags);
|
|
|
|
/* optional flags if set/clear mean the same
|
|
|
|
* thing and can be represented by a single
|
|
|
|
* bitset, also there is no need to check for
|
|
|
|
* conflicting flags when they are optional
|
|
|
|
*/
|
|
|
|
opt_flagsv.push_back(tmpflags | tmpinv_flags);
|
|
|
|
}
|
2023-03-28 04:23:20 -07:00
|
|
|
}
|
|
|
|
|
|
|
|
/* move options=() to opts list */
|
|
|
|
struct cond_entry *opts_eq = extract_options(&src_conds, 1);
|
|
|
|
if (opts_eq) {
|
2019-02-12 08:15:22 -08:00
|
|
|
unsigned int tmpflags = 0, tmpinv_flags = 0;
|
parser: support multiple mount conditionals in a single rule
Now that flag processing for mount rules with single option
conditionals are fixed e-enable multiple mount conditionals on a
single mount rule. The mount conditionals are equivalent to specifying
multiple rules.
mount options=(a,b,c) options=(c,d),
is the same as
mount options=(a,b,c),
mount options=(c,d),
and
mount options in (a,b,c) options in (c,d),
is the same as
mount options in (a,b,c),
mount options in (c,d),
when multiple options= and options in are combined in a single rule
it is the same as the cross product of the options.
where
mount options=(a,b,c) options in (d,e),
is a single rule.
mount options=(a,b,c) options=(d,e) options in (f),
is equivalent to
mount options=(a,b,c) options in (f),
mount options=(d,e) options in (f),
and while it is not recommended that multiple options= and options in
conditions be used in a single rule.
mount options=(a,b,c) options=(d,e) options in (f) options in (g),
is equivalent to
mount options=(a,b,c) options in (f),
mount options=(a,b,c) options in (g),
mount options=(d,e) options in (f),
mount options=(d,e) options in (g),
Bug Link: https://bugs.launchpad.net/apparmor/+bug/1597017
Signed-off-by: John Johansen <john.johansen@canonical.com>
- rebased to bba1a023bf
- fixed infinite loop in mnt_rule::gen_policy_re
Signed-off-by: Alexander Mikhalitsyn <aleksandr.mikhalitsyn@canonical.com>
Acked-by: John Johansen <john.johansen@canonical.com>
2023-03-28 12:43:18 -07:00
|
|
|
struct cond_entry *entry;
|
|
|
|
|
|
|
|
while ((entry = list_pop(opts_eq))) {
|
|
|
|
process_one_option(entry, tmpflags,
|
|
|
|
tmpinv_flags);
|
|
|
|
/* throw away tmpinv_flags, only needed in
|
|
|
|
* consistancy check
|
|
|
|
*/
|
2021-06-11 03:20:18 -07:00
|
|
|
if (perms_p & AA_DUMMY_REMOUNT)
|
parser: support multiple mount conditionals in a single rule
Now that flag processing for mount rules with single option
conditionals are fixed e-enable multiple mount conditionals on a
single mount rule. The mount conditionals are equivalent to specifying
multiple rules.
mount options=(a,b,c) options=(c,d),
is the same as
mount options=(a,b,c),
mount options=(c,d),
and
mount options in (a,b,c) options in (c,d),
is the same as
mount options in (a,b,c),
mount options in (c,d),
when multiple options= and options in are combined in a single rule
it is the same as the cross product of the options.
where
mount options=(a,b,c) options in (d,e),
is a single rule.
mount options=(a,b,c) options=(d,e) options in (f),
is equivalent to
mount options=(a,b,c) options in (f),
mount options=(d,e) options in (f),
and while it is not recommended that multiple options= and options in
conditions be used in a single rule.
mount options=(a,b,c) options=(d,e) options in (f) options in (g),
is equivalent to
mount options=(a,b,c) options in (f),
mount options=(a,b,c) options in (g),
mount options=(d,e) options in (f),
mount options=(d,e) options in (g),
Bug Link: https://bugs.launchpad.net/apparmor/+bug/1597017
Signed-off-by: John Johansen <john.johansen@canonical.com>
- rebased to bba1a023bf
- fixed infinite loop in mnt_rule::gen_policy_re
Signed-off-by: Alexander Mikhalitsyn <aleksandr.mikhalitsyn@canonical.com>
Acked-by: John Johansen <john.johansen@canonical.com>
2023-03-28 12:43:18 -07:00
|
|
|
tmpflags |= MS_REMOUNT;
|
|
|
|
|
|
|
|
if (conflicting_flags(tmpflags, tmpinv_flags)) {
|
|
|
|
PERROR("conflicting flags in the rule\n");
|
|
|
|
exit(1);
|
|
|
|
}
|
|
|
|
|
|
|
|
flagsv.push_back(tmpflags);
|
2019-02-12 08:15:22 -08:00
|
|
|
}
|
2023-03-28 04:23:20 -07:00
|
|
|
}
|
2014-04-07 03:16:50 -07:00
|
|
|
|
2023-03-28 04:23:20 -07:00
|
|
|
if (src_conds) {
|
|
|
|
perror_conds("mount", src_conds);
|
|
|
|
exit(1);
|
|
|
|
}
|
|
|
|
}
|
2014-04-07 03:16:50 -07:00
|
|
|
|
parser: support multiple mount conditionals in a single rule
Now that flag processing for mount rules with single option
conditionals are fixed e-enable multiple mount conditionals on a
single mount rule. The mount conditionals are equivalent to specifying
multiple rules.
mount options=(a,b,c) options=(c,d),
is the same as
mount options=(a,b,c),
mount options=(c,d),
and
mount options in (a,b,c) options in (c,d),
is the same as
mount options in (a,b,c),
mount options in (c,d),
when multiple options= and options in are combined in a single rule
it is the same as the cross product of the options.
where
mount options=(a,b,c) options in (d,e),
is a single rule.
mount options=(a,b,c) options=(d,e) options in (f),
is equivalent to
mount options=(a,b,c) options in (f),
mount options=(d,e) options in (f),
and while it is not recommended that multiple options= and options in
conditions be used in a single rule.
mount options=(a,b,c) options=(d,e) options in (f) options in (g),
is equivalent to
mount options=(a,b,c) options in (f),
mount options=(a,b,c) options in (g),
mount options=(d,e) options in (f),
mount options=(d,e) options in (g),
Bug Link: https://bugs.launchpad.net/apparmor/+bug/1597017
Signed-off-by: John Johansen <john.johansen@canonical.com>
- rebased to bba1a023bf
- fixed infinite loop in mnt_rule::gen_policy_re
Signed-off-by: Alexander Mikhalitsyn <aleksandr.mikhalitsyn@canonical.com>
Acked-by: John Johansen <john.johansen@canonical.com>
2023-03-28 12:43:18 -07:00
|
|
|
if (!(flagsv.size() + opt_flagsv.size())) {
|
2023-03-28 04:23:20 -07:00
|
|
|
/* no flag options, and not remount, allow everything */
|
2021-06-11 03:20:18 -07:00
|
|
|
if (perms_p & AA_DUMMY_REMOUNT) {
|
parser: support multiple mount conditionals in a single rule
Now that flag processing for mount rules with single option
conditionals are fixed e-enable multiple mount conditionals on a
single mount rule. The mount conditionals are equivalent to specifying
multiple rules.
mount options=(a,b,c) options=(c,d),
is the same as
mount options=(a,b,c),
mount options=(c,d),
and
mount options in (a,b,c) options in (c,d),
is the same as
mount options in (a,b,c),
mount options in (c,d),
when multiple options= and options in are combined in a single rule
it is the same as the cross product of the options.
where
mount options=(a,b,c) options in (d,e),
is a single rule.
mount options=(a,b,c) options=(d,e) options in (f),
is equivalent to
mount options=(a,b,c) options in (f),
mount options=(d,e) options in (f),
and while it is not recommended that multiple options= and options in
conditions be used in a single rule.
mount options=(a,b,c) options=(d,e) options in (f) options in (g),
is equivalent to
mount options=(a,b,c) options in (f),
mount options=(a,b,c) options in (g),
mount options=(d,e) options in (f),
mount options=(d,e) options in (g),
Bug Link: https://bugs.launchpad.net/apparmor/+bug/1597017
Signed-off-by: John Johansen <john.johansen@canonical.com>
- rebased to bba1a023bf
- fixed infinite loop in mnt_rule::gen_policy_re
Signed-off-by: Alexander Mikhalitsyn <aleksandr.mikhalitsyn@canonical.com>
Acked-by: John Johansen <john.johansen@canonical.com>
2023-03-28 12:43:18 -07:00
|
|
|
flagsv.push_back(MS_REMOUNT);
|
|
|
|
opt_flagsv.push_back(MS_REMOUNT_FLAGS & ~MS_REMOUNT);
|
2023-03-28 04:23:20 -07:00
|
|
|
} else {
|
parser: support multiple mount conditionals in a single rule
Now that flag processing for mount rules with single option
conditionals are fixed e-enable multiple mount conditionals on a
single mount rule. The mount conditionals are equivalent to specifying
multiple rules.
mount options=(a,b,c) options=(c,d),
is the same as
mount options=(a,b,c),
mount options=(c,d),
and
mount options in (a,b,c) options in (c,d),
is the same as
mount options in (a,b,c),
mount options in (c,d),
when multiple options= and options in are combined in a single rule
it is the same as the cross product of the options.
where
mount options=(a,b,c) options in (d,e),
is a single rule.
mount options=(a,b,c) options=(d,e) options in (f),
is equivalent to
mount options=(a,b,c) options in (f),
mount options=(d,e) options in (f),
and while it is not recommended that multiple options= and options in
conditions be used in a single rule.
mount options=(a,b,c) options=(d,e) options in (f) options in (g),
is equivalent to
mount options=(a,b,c) options in (f),
mount options=(a,b,c) options in (g),
mount options=(d,e) options in (f),
mount options=(d,e) options in (g),
Bug Link: https://bugs.launchpad.net/apparmor/+bug/1597017
Signed-off-by: John Johansen <john.johansen@canonical.com>
- rebased to bba1a023bf
- fixed infinite loop in mnt_rule::gen_policy_re
Signed-off-by: Alexander Mikhalitsyn <aleksandr.mikhalitsyn@canonical.com>
Acked-by: John Johansen <john.johansen@canonical.com>
2023-03-28 12:43:18 -07:00
|
|
|
flagsv.push_back(MS_ALL_FLAGS);
|
|
|
|
opt_flagsv.push_back(MS_ALL_FLAGS);
|
Add mount rules
Add the ability to control mounting and unmounting
The basic form of the rules are.
[audit] [deny] mount [conds]* [device] [ -> [conds] path],
[audit] [deny] remount [conds]* [path],
[audit] [deny] umount [conds]* [path],
[audit] [deny] pivotroot [oldroot=<value>] <path> -> <profile>
remount is just a short cut for mount options=remount
where [conds] can be
fstype=<expr>
options=<expr>
conds follow the extended conditional syntax of allowing either:
* a single value after the equals, which has the same character range as
regular IDS (ie most anything but it can't be terminated with a , (comma)
and if spaces or other characters are needed it can be quoted
eg.
options=foo
options = foo
options="foo bar"
* a list of values after the equals, the list of values is enclosed within
parenthesis () and its has a slightly reduced character set but again
elements can be quoted.
the separation between elements is whitespace and commas.
eg.
options=(foo bar)
options=(foo, bar)
options=(foo , bar)
options=(foo,bar)
The rules are flexible and follow a similar pattern as network, capability,
etc.
mount, # allow all mounts, but not umount or pivotroot
mount fstype=procfs, # allow mounting procfs anywhere
mount options=(bind, ro) /foo -> /bar, # readonly bind mount
mount /dev/sda -> /mnt,
mount /dev/sd** -> /mnt/**,
mount fstype=overlayfs options=(rw,upperdir=/tmp/upper/,lowerdir=/) overlay -> /mnt/
umount,
umount /m*,
Currently variables and regexs are are supported on the device and mount
point. ie.
mount <devince> -> <mount point>,
Regexes are supported in fstype and options. The options have a further
caveat that regexs only work if the option is fs specific option.
eg. options=(upperdir=/tmp/*,lowerdir=/)
regex's will not currently work against the standard options like ro, rw
nosuid
Conditionals (fstype) can only be applied to the device (source) at this
time and will be disregarded in situations where the mount is manipulating
an existing mount (bind, remount).
Options can be specified multiple times
mount option=rw option=(nosuid,upperdir=/foo),
and will be combined together into a single set of values
The ordering of the standard mount options (rw,ro, ...) does not matter
but the ordering of fs specific options does.
Specifying that the value of a particular option does not matter can be
acheived by providing both the positive and negative forms of and option
option=(rw,ro) options=(suid,nosuid)
For the fs specific options specifying that a particular value does not
matter is achieve using a regex with alternations.
Improvements to the syntax and order restrictions are planned for the
future.
Signed-off-by: John Johansen <john.johansen@canonical.com>
2012-02-24 04:19:38 -08:00
|
|
|
}
|
parser: support multiple mount conditionals in a single rule
Now that flag processing for mount rules with single option
conditionals are fixed e-enable multiple mount conditionals on a
single mount rule. The mount conditionals are equivalent to specifying
multiple rules.
mount options=(a,b,c) options=(c,d),
is the same as
mount options=(a,b,c),
mount options=(c,d),
and
mount options in (a,b,c) options in (c,d),
is the same as
mount options in (a,b,c),
mount options in (c,d),
when multiple options= and options in are combined in a single rule
it is the same as the cross product of the options.
where
mount options=(a,b,c) options in (d,e),
is a single rule.
mount options=(a,b,c) options=(d,e) options in (f),
is equivalent to
mount options=(a,b,c) options in (f),
mount options=(d,e) options in (f),
and while it is not recommended that multiple options= and options in
conditions be used in a single rule.
mount options=(a,b,c) options=(d,e) options in (f) options in (g),
is equivalent to
mount options=(a,b,c) options in (f),
mount options=(a,b,c) options in (g),
mount options=(d,e) options in (f),
mount options=(d,e) options in (g),
Bug Link: https://bugs.launchpad.net/apparmor/+bug/1597017
Signed-off-by: John Johansen <john.johansen@canonical.com>
- rebased to bba1a023bf
- fixed infinite loop in mnt_rule::gen_policy_re
Signed-off-by: Alexander Mikhalitsyn <aleksandr.mikhalitsyn@canonical.com>
Acked-by: John Johansen <john.johansen@canonical.com>
2023-03-28 12:43:18 -07:00
|
|
|
} else if (!(flagsv.size())) {
|
2023-03-28 04:23:20 -07:00
|
|
|
/* no flags but opts set */
|
2021-06-11 03:20:18 -07:00
|
|
|
if (perms_p & AA_DUMMY_REMOUNT)
|
parser: support multiple mount conditionals in a single rule
Now that flag processing for mount rules with single option
conditionals are fixed e-enable multiple mount conditionals on a
single mount rule. The mount conditionals are equivalent to specifying
multiple rules.
mount options=(a,b,c) options=(c,d),
is the same as
mount options=(a,b,c),
mount options=(c,d),
and
mount options in (a,b,c) options in (c,d),
is the same as
mount options in (a,b,c),
mount options in (c,d),
when multiple options= and options in are combined in a single rule
it is the same as the cross product of the options.
where
mount options=(a,b,c) options in (d,e),
is a single rule.
mount options=(a,b,c) options=(d,e) options in (f),
is equivalent to
mount options=(a,b,c) options in (f),
mount options=(d,e) options in (f),
and while it is not recommended that multiple options= and options in
conditions be used in a single rule.
mount options=(a,b,c) options=(d,e) options in (f) options in (g),
is equivalent to
mount options=(a,b,c) options in (f),
mount options=(a,b,c) options in (g),
mount options=(d,e) options in (f),
mount options=(d,e) options in (g),
Bug Link: https://bugs.launchpad.net/apparmor/+bug/1597017
Signed-off-by: John Johansen <john.johansen@canonical.com>
- rebased to bba1a023bf
- fixed infinite loop in mnt_rule::gen_policy_re
Signed-off-by: Alexander Mikhalitsyn <aleksandr.mikhalitsyn@canonical.com>
Acked-by: John Johansen <john.johansen@canonical.com>
2023-03-28 12:43:18 -07:00
|
|
|
flagsv.push_back(MS_REMOUNT);
|
|
|
|
else
|
|
|
|
flagsv.push_back(0);
|
|
|
|
} else if (!(opt_flagsv.size())) {
|
|
|
|
opt_flagsv.push_back(0);
|
Add mount rules
Add the ability to control mounting and unmounting
The basic form of the rules are.
[audit] [deny] mount [conds]* [device] [ -> [conds] path],
[audit] [deny] remount [conds]* [path],
[audit] [deny] umount [conds]* [path],
[audit] [deny] pivotroot [oldroot=<value>] <path> -> <profile>
remount is just a short cut for mount options=remount
where [conds] can be
fstype=<expr>
options=<expr>
conds follow the extended conditional syntax of allowing either:
* a single value after the equals, which has the same character range as
regular IDS (ie most anything but it can't be terminated with a , (comma)
and if spaces or other characters are needed it can be quoted
eg.
options=foo
options = foo
options="foo bar"
* a list of values after the equals, the list of values is enclosed within
parenthesis () and its has a slightly reduced character set but again
elements can be quoted.
the separation between elements is whitespace and commas.
eg.
options=(foo bar)
options=(foo, bar)
options=(foo , bar)
options=(foo,bar)
The rules are flexible and follow a similar pattern as network, capability,
etc.
mount, # allow all mounts, but not umount or pivotroot
mount fstype=procfs, # allow mounting procfs anywhere
mount options=(bind, ro) /foo -> /bar, # readonly bind mount
mount /dev/sda -> /mnt,
mount /dev/sd** -> /mnt/**,
mount fstype=overlayfs options=(rw,upperdir=/tmp/upper/,lowerdir=/) overlay -> /mnt/
umount,
umount /m*,
Currently variables and regexs are are supported on the device and mount
point. ie.
mount <devince> -> <mount point>,
Regexes are supported in fstype and options. The options have a further
caveat that regexs only work if the option is fs specific option.
eg. options=(upperdir=/tmp/*,lowerdir=/)
regex's will not currently work against the standard options like ro, rw
nosuid
Conditionals (fstype) can only be applied to the device (source) at this
time and will be disregarded in situations where the mount is manipulating
an existing mount (bind, remount).
Options can be specified multiple times
mount option=rw option=(nosuid,upperdir=/foo),
and will be combined together into a single set of values
The ordering of the standard mount options (rw,ro, ...) does not matter
but the ordering of fs specific options does.
Specifying that the value of a particular option does not matter can be
acheived by providing both the positive and negative forms of and option
option=(rw,ro) options=(suid,nosuid)
For the fs specific options specifying that a particular value does not
matter is achieve using a regex with alternations.
Improvements to the syntax and order restrictions are planned for the
future.
Signed-off-by: John Johansen <john.johansen@canonical.com>
2012-02-24 04:19:38 -08:00
|
|
|
}
|
|
|
|
|
2021-06-11 03:20:18 -07:00
|
|
|
if (perms_p & AA_DUMMY_REMOUNT) {
|
|
|
|
perms_p = AA_MAY_MOUNT;
|
2014-04-07 03:16:50 -07:00
|
|
|
}
|
2021-06-11 03:20:18 -07:00
|
|
|
perms = perms_p;
|
Add mount rules
Add the ability to control mounting and unmounting
The basic form of the rules are.
[audit] [deny] mount [conds]* [device] [ -> [conds] path],
[audit] [deny] remount [conds]* [path],
[audit] [deny] umount [conds]* [path],
[audit] [deny] pivotroot [oldroot=<value>] <path> -> <profile>
remount is just a short cut for mount options=remount
where [conds] can be
fstype=<expr>
options=<expr>
conds follow the extended conditional syntax of allowing either:
* a single value after the equals, which has the same character range as
regular IDS (ie most anything but it can't be terminated with a , (comma)
and if spaces or other characters are needed it can be quoted
eg.
options=foo
options = foo
options="foo bar"
* a list of values after the equals, the list of values is enclosed within
parenthesis () and its has a slightly reduced character set but again
elements can be quoted.
the separation between elements is whitespace and commas.
eg.
options=(foo bar)
options=(foo, bar)
options=(foo , bar)
options=(foo,bar)
The rules are flexible and follow a similar pattern as network, capability,
etc.
mount, # allow all mounts, but not umount or pivotroot
mount fstype=procfs, # allow mounting procfs anywhere
mount options=(bind, ro) /foo -> /bar, # readonly bind mount
mount /dev/sda -> /mnt,
mount /dev/sd** -> /mnt/**,
mount fstype=overlayfs options=(rw,upperdir=/tmp/upper/,lowerdir=/) overlay -> /mnt/
umount,
umount /m*,
Currently variables and regexs are are supported on the device and mount
point. ie.
mount <devince> -> <mount point>,
Regexes are supported in fstype and options. The options have a further
caveat that regexs only work if the option is fs specific option.
eg. options=(upperdir=/tmp/*,lowerdir=/)
regex's will not currently work against the standard options like ro, rw
nosuid
Conditionals (fstype) can only be applied to the device (source) at this
time and will be disregarded in situations where the mount is manipulating
an existing mount (bind, remount).
Options can be specified multiple times
mount option=rw option=(nosuid,upperdir=/foo),
and will be combined together into a single set of values
The ordering of the standard mount options (rw,ro, ...) does not matter
but the ordering of fs specific options does.
Specifying that the value of a particular option does not matter can be
acheived by providing both the positive and negative forms of and option
option=(rw,ro) options=(suid,nosuid)
For the fs specific options specifying that a particular value does not
matter is achieve using a regex with alternations.
Improvements to the syntax and order restrictions are planned for the
future.
Signed-off-by: John Johansen <john.johansen@canonical.com>
2012-02-24 04:19:38 -08:00
|
|
|
}
|
|
|
|
|
2014-04-07 03:16:50 -07:00
|
|
|
ostream &mnt_rule::dump(ostream &os)
|
Add mount rules
Add the ability to control mounting and unmounting
The basic form of the rules are.
[audit] [deny] mount [conds]* [device] [ -> [conds] path],
[audit] [deny] remount [conds]* [path],
[audit] [deny] umount [conds]* [path],
[audit] [deny] pivotroot [oldroot=<value>] <path> -> <profile>
remount is just a short cut for mount options=remount
where [conds] can be
fstype=<expr>
options=<expr>
conds follow the extended conditional syntax of allowing either:
* a single value after the equals, which has the same character range as
regular IDS (ie most anything but it can't be terminated with a , (comma)
and if spaces or other characters are needed it can be quoted
eg.
options=foo
options = foo
options="foo bar"
* a list of values after the equals, the list of values is enclosed within
parenthesis () and its has a slightly reduced character set but again
elements can be quoted.
the separation between elements is whitespace and commas.
eg.
options=(foo bar)
options=(foo, bar)
options=(foo , bar)
options=(foo,bar)
The rules are flexible and follow a similar pattern as network, capability,
etc.
mount, # allow all mounts, but not umount or pivotroot
mount fstype=procfs, # allow mounting procfs anywhere
mount options=(bind, ro) /foo -> /bar, # readonly bind mount
mount /dev/sda -> /mnt,
mount /dev/sd** -> /mnt/**,
mount fstype=overlayfs options=(rw,upperdir=/tmp/upper/,lowerdir=/) overlay -> /mnt/
umount,
umount /m*,
Currently variables and regexs are are supported on the device and mount
point. ie.
mount <devince> -> <mount point>,
Regexes are supported in fstype and options. The options have a further
caveat that regexs only work if the option is fs specific option.
eg. options=(upperdir=/tmp/*,lowerdir=/)
regex's will not currently work against the standard options like ro, rw
nosuid
Conditionals (fstype) can only be applied to the device (source) at this
time and will be disregarded in situations where the mount is manipulating
an existing mount (bind, remount).
Options can be specified multiple times
mount option=rw option=(nosuid,upperdir=/foo),
and will be combined together into a single set of values
The ordering of the standard mount options (rw,ro, ...) does not matter
but the ordering of fs specific options does.
Specifying that the value of a particular option does not matter can be
acheived by providing both the positive and negative forms of and option
option=(rw,ro) options=(suid,nosuid)
For the fs specific options specifying that a particular value does not
matter is achieve using a regex with alternations.
Improvements to the syntax and order restrictions are planned for the
future.
Signed-off-by: John Johansen <john.johansen@canonical.com>
2012-02-24 04:19:38 -08:00
|
|
|
{
|
2021-09-04 03:28:18 -07:00
|
|
|
prefix_rule_t::dump(os);
|
|
|
|
|
2021-06-11 03:20:18 -07:00
|
|
|
if (perms & AA_MAY_MOUNT)
|
2014-04-07 03:16:50 -07:00
|
|
|
os << "mount";
|
2021-06-11 03:20:18 -07:00
|
|
|
else if (perms & AA_MAY_UMOUNT)
|
2014-04-07 03:16:50 -07:00
|
|
|
os << "umount";
|
2021-06-11 03:20:18 -07:00
|
|
|
else if (perms & AA_MAY_PIVOTROOT)
|
2014-04-07 03:16:50 -07:00
|
|
|
os << "pivotroot";
|
|
|
|
else
|
2020-11-19 12:32:36 -08:00
|
|
|
os << "error: unknown mount perm";
|
Add mount rules
Add the ability to control mounting and unmounting
The basic form of the rules are.
[audit] [deny] mount [conds]* [device] [ -> [conds] path],
[audit] [deny] remount [conds]* [path],
[audit] [deny] umount [conds]* [path],
[audit] [deny] pivotroot [oldroot=<value>] <path> -> <profile>
remount is just a short cut for mount options=remount
where [conds] can be
fstype=<expr>
options=<expr>
conds follow the extended conditional syntax of allowing either:
* a single value after the equals, which has the same character range as
regular IDS (ie most anything but it can't be terminated with a , (comma)
and if spaces or other characters are needed it can be quoted
eg.
options=foo
options = foo
options="foo bar"
* a list of values after the equals, the list of values is enclosed within
parenthesis () and its has a slightly reduced character set but again
elements can be quoted.
the separation between elements is whitespace and commas.
eg.
options=(foo bar)
options=(foo, bar)
options=(foo , bar)
options=(foo,bar)
The rules are flexible and follow a similar pattern as network, capability,
etc.
mount, # allow all mounts, but not umount or pivotroot
mount fstype=procfs, # allow mounting procfs anywhere
mount options=(bind, ro) /foo -> /bar, # readonly bind mount
mount /dev/sda -> /mnt,
mount /dev/sd** -> /mnt/**,
mount fstype=overlayfs options=(rw,upperdir=/tmp/upper/,lowerdir=/) overlay -> /mnt/
umount,
umount /m*,
Currently variables and regexs are are supported on the device and mount
point. ie.
mount <devince> -> <mount point>,
Regexes are supported in fstype and options. The options have a further
caveat that regexs only work if the option is fs specific option.
eg. options=(upperdir=/tmp/*,lowerdir=/)
regex's will not currently work against the standard options like ro, rw
nosuid
Conditionals (fstype) can only be applied to the device (source) at this
time and will be disregarded in situations where the mount is manipulating
an existing mount (bind, remount).
Options can be specified multiple times
mount option=rw option=(nosuid,upperdir=/foo),
and will be combined together into a single set of values
The ordering of the standard mount options (rw,ro, ...) does not matter
but the ordering of fs specific options does.
Specifying that the value of a particular option does not matter can be
acheived by providing both the positive and negative forms of and option
option=(rw,ro) options=(suid,nosuid)
For the fs specific options specifying that a particular value does not
matter is achieve using a regex with alternations.
Improvements to the syntax and order restrictions are planned for the
future.
Signed-off-by: John Johansen <john.johansen@canonical.com>
2012-02-24 04:19:38 -08:00
|
|
|
|
parser: support multiple mount conditionals in a single rule
Now that flag processing for mount rules with single option
conditionals are fixed e-enable multiple mount conditionals on a
single mount rule. The mount conditionals are equivalent to specifying
multiple rules.
mount options=(a,b,c) options=(c,d),
is the same as
mount options=(a,b,c),
mount options=(c,d),
and
mount options in (a,b,c) options in (c,d),
is the same as
mount options in (a,b,c),
mount options in (c,d),
when multiple options= and options in are combined in a single rule
it is the same as the cross product of the options.
where
mount options=(a,b,c) options in (d,e),
is a single rule.
mount options=(a,b,c) options=(d,e) options in (f),
is equivalent to
mount options=(a,b,c) options in (f),
mount options=(d,e) options in (f),
and while it is not recommended that multiple options= and options in
conditions be used in a single rule.
mount options=(a,b,c) options=(d,e) options in (f) options in (g),
is equivalent to
mount options=(a,b,c) options in (f),
mount options=(a,b,c) options in (g),
mount options=(d,e) options in (f),
mount options=(d,e) options in (g),
Bug Link: https://bugs.launchpad.net/apparmor/+bug/1597017
Signed-off-by: John Johansen <john.johansen@canonical.com>
- rebased to bba1a023bf
- fixed infinite loop in mnt_rule::gen_policy_re
Signed-off-by: Alexander Mikhalitsyn <aleksandr.mikhalitsyn@canonical.com>
Acked-by: John Johansen <john.johansen@canonical.com>
2023-03-28 12:43:18 -07:00
|
|
|
for (unsigned int i = 0; i < flagsv.size(); i++)
|
|
|
|
os << " flags=(0x" << hex << flagsv[i] << ")";
|
|
|
|
for (unsigned int i = 0; i < opt_flagsv.size(); i++)
|
|
|
|
os << " flags in (0x" << hex << opt_flagsv[i] << ")";
|
|
|
|
|
2014-04-07 03:16:50 -07:00
|
|
|
if (dev_type) {
|
|
|
|
os << " type=";
|
|
|
|
print_value_list(dev_type);
|
|
|
|
}
|
|
|
|
if (opts) {
|
|
|
|
os << " options=";
|
|
|
|
print_value_list(opts);
|
|
|
|
}
|
|
|
|
if (device)
|
|
|
|
os << " " << device;
|
|
|
|
if (mnt_point)
|
|
|
|
os << " -> " << mnt_point;
|
|
|
|
if (trans)
|
|
|
|
os << " -> " << trans;
|
|
|
|
|
2021-09-04 03:28:18 -07:00
|
|
|
|
|
|
|
os << " " << "(0x" << hex << perms << "/0x" << (audit != AUDIT_UNSPECIFIED ? perms : 0) << ")";
|
2014-04-07 03:16:50 -07:00
|
|
|
os << ",\n";
|
|
|
|
|
|
|
|
return os;
|
|
|
|
}
|
Add mount rules
Add the ability to control mounting and unmounting
The basic form of the rules are.
[audit] [deny] mount [conds]* [device] [ -> [conds] path],
[audit] [deny] remount [conds]* [path],
[audit] [deny] umount [conds]* [path],
[audit] [deny] pivotroot [oldroot=<value>] <path> -> <profile>
remount is just a short cut for mount options=remount
where [conds] can be
fstype=<expr>
options=<expr>
conds follow the extended conditional syntax of allowing either:
* a single value after the equals, which has the same character range as
regular IDS (ie most anything but it can't be terminated with a , (comma)
and if spaces or other characters are needed it can be quoted
eg.
options=foo
options = foo
options="foo bar"
* a list of values after the equals, the list of values is enclosed within
parenthesis () and its has a slightly reduced character set but again
elements can be quoted.
the separation between elements is whitespace and commas.
eg.
options=(foo bar)
options=(foo, bar)
options=(foo , bar)
options=(foo,bar)
The rules are flexible and follow a similar pattern as network, capability,
etc.
mount, # allow all mounts, but not umount or pivotroot
mount fstype=procfs, # allow mounting procfs anywhere
mount options=(bind, ro) /foo -> /bar, # readonly bind mount
mount /dev/sda -> /mnt,
mount /dev/sd** -> /mnt/**,
mount fstype=overlayfs options=(rw,upperdir=/tmp/upper/,lowerdir=/) overlay -> /mnt/
umount,
umount /m*,
Currently variables and regexs are are supported on the device and mount
point. ie.
mount <devince> -> <mount point>,
Regexes are supported in fstype and options. The options have a further
caveat that regexs only work if the option is fs specific option.
eg. options=(upperdir=/tmp/*,lowerdir=/)
regex's will not currently work against the standard options like ro, rw
nosuid
Conditionals (fstype) can only be applied to the device (source) at this
time and will be disregarded in situations where the mount is manipulating
an existing mount (bind, remount).
Options can be specified multiple times
mount option=rw option=(nosuid,upperdir=/foo),
and will be combined together into a single set of values
The ordering of the standard mount options (rw,ro, ...) does not matter
but the ordering of fs specific options does.
Specifying that the value of a particular option does not matter can be
acheived by providing both the positive and negative forms of and option
option=(rw,ro) options=(suid,nosuid)
For the fs specific options specifying that a particular value does not
matter is achieve using a regex with alternations.
Improvements to the syntax and order restrictions are planned for the
future.
Signed-off-by: John Johansen <john.johansen@canonical.com>
2012-02-24 04:19:38 -08:00
|
|
|
|
2014-04-07 03:16:50 -07:00
|
|
|
/* does not currently support expansion of vars in options */
|
|
|
|
int mnt_rule::expand_variables(void)
|
|
|
|
{
|
2020-09-01 03:18:58 -07:00
|
|
|
struct value_list *ent;
|
2014-04-07 03:16:50 -07:00
|
|
|
int error = 0;
|
|
|
|
|
|
|
|
error = expand_entry_variables(&mnt_point);
|
|
|
|
if (error)
|
|
|
|
return error;
|
2020-09-01 03:10:01 -07:00
|
|
|
filter_slashes(mnt_point);
|
2014-04-07 03:16:50 -07:00
|
|
|
error = expand_entry_variables(&device);
|
|
|
|
if (error)
|
|
|
|
return error;
|
2020-09-01 03:10:01 -07:00
|
|
|
filter_slashes(device);
|
2014-04-07 03:16:50 -07:00
|
|
|
error = expand_entry_variables(&trans);
|
|
|
|
if (error)
|
|
|
|
return error;
|
|
|
|
|
2020-09-01 03:18:58 -07:00
|
|
|
list_for_each(dev_type, ent) {
|
|
|
|
error = expand_entry_variables(&ent->value);
|
|
|
|
if (error)
|
|
|
|
return error;
|
|
|
|
}
|
|
|
|
list_for_each(opts, ent) {
|
|
|
|
error = expand_entry_variables(&ent->value);
|
|
|
|
if (error)
|
|
|
|
return error;
|
|
|
|
}
|
|
|
|
|
2014-04-07 03:16:50 -07:00
|
|
|
return 0;
|
Add mount rules
Add the ability to control mounting and unmounting
The basic form of the rules are.
[audit] [deny] mount [conds]* [device] [ -> [conds] path],
[audit] [deny] remount [conds]* [path],
[audit] [deny] umount [conds]* [path],
[audit] [deny] pivotroot [oldroot=<value>] <path> -> <profile>
remount is just a short cut for mount options=remount
where [conds] can be
fstype=<expr>
options=<expr>
conds follow the extended conditional syntax of allowing either:
* a single value after the equals, which has the same character range as
regular IDS (ie most anything but it can't be terminated with a , (comma)
and if spaces or other characters are needed it can be quoted
eg.
options=foo
options = foo
options="foo bar"
* a list of values after the equals, the list of values is enclosed within
parenthesis () and its has a slightly reduced character set but again
elements can be quoted.
the separation between elements is whitespace and commas.
eg.
options=(foo bar)
options=(foo, bar)
options=(foo , bar)
options=(foo,bar)
The rules are flexible and follow a similar pattern as network, capability,
etc.
mount, # allow all mounts, but not umount or pivotroot
mount fstype=procfs, # allow mounting procfs anywhere
mount options=(bind, ro) /foo -> /bar, # readonly bind mount
mount /dev/sda -> /mnt,
mount /dev/sd** -> /mnt/**,
mount fstype=overlayfs options=(rw,upperdir=/tmp/upper/,lowerdir=/) overlay -> /mnt/
umount,
umount /m*,
Currently variables and regexs are are supported on the device and mount
point. ie.
mount <devince> -> <mount point>,
Regexes are supported in fstype and options. The options have a further
caveat that regexs only work if the option is fs specific option.
eg. options=(upperdir=/tmp/*,lowerdir=/)
regex's will not currently work against the standard options like ro, rw
nosuid
Conditionals (fstype) can only be applied to the device (source) at this
time and will be disregarded in situations where the mount is manipulating
an existing mount (bind, remount).
Options can be specified multiple times
mount option=rw option=(nosuid,upperdir=/foo),
and will be combined together into a single set of values
The ordering of the standard mount options (rw,ro, ...) does not matter
but the ordering of fs specific options does.
Specifying that the value of a particular option does not matter can be
acheived by providing both the positive and negative forms of and option
option=(rw,ro) options=(suid,nosuid)
For the fs specific options specifying that a particular value does not
matter is achieve using a regex with alternations.
Improvements to the syntax and order restrictions are planned for the
future.
Signed-off-by: John Johansen <john.johansen@canonical.com>
2012-02-24 04:19:38 -08:00
|
|
|
}
|
|
|
|
|
2014-04-07 03:16:50 -07:00
|
|
|
static int build_mnt_flags(char *buffer, int size, unsigned int flags,
|
2023-03-28 04:23:20 -07:00
|
|
|
unsigned int opt_flags)
|
Add mount rules
Add the ability to control mounting and unmounting
The basic form of the rules are.
[audit] [deny] mount [conds]* [device] [ -> [conds] path],
[audit] [deny] remount [conds]* [path],
[audit] [deny] umount [conds]* [path],
[audit] [deny] pivotroot [oldroot=<value>] <path> -> <profile>
remount is just a short cut for mount options=remount
where [conds] can be
fstype=<expr>
options=<expr>
conds follow the extended conditional syntax of allowing either:
* a single value after the equals, which has the same character range as
regular IDS (ie most anything but it can't be terminated with a , (comma)
and if spaces or other characters are needed it can be quoted
eg.
options=foo
options = foo
options="foo bar"
* a list of values after the equals, the list of values is enclosed within
parenthesis () and its has a slightly reduced character set but again
elements can be quoted.
the separation between elements is whitespace and commas.
eg.
options=(foo bar)
options=(foo, bar)
options=(foo , bar)
options=(foo,bar)
The rules are flexible and follow a similar pattern as network, capability,
etc.
mount, # allow all mounts, but not umount or pivotroot
mount fstype=procfs, # allow mounting procfs anywhere
mount options=(bind, ro) /foo -> /bar, # readonly bind mount
mount /dev/sda -> /mnt,
mount /dev/sd** -> /mnt/**,
mount fstype=overlayfs options=(rw,upperdir=/tmp/upper/,lowerdir=/) overlay -> /mnt/
umount,
umount /m*,
Currently variables and regexs are are supported on the device and mount
point. ie.
mount <devince> -> <mount point>,
Regexes are supported in fstype and options. The options have a further
caveat that regexs only work if the option is fs specific option.
eg. options=(upperdir=/tmp/*,lowerdir=/)
regex's will not currently work against the standard options like ro, rw
nosuid
Conditionals (fstype) can only be applied to the device (source) at this
time and will be disregarded in situations where the mount is manipulating
an existing mount (bind, remount).
Options can be specified multiple times
mount option=rw option=(nosuid,upperdir=/foo),
and will be combined together into a single set of values
The ordering of the standard mount options (rw,ro, ...) does not matter
but the ordering of fs specific options does.
Specifying that the value of a particular option does not matter can be
acheived by providing both the positive and negative forms of and option
option=(rw,ro) options=(suid,nosuid)
For the fs specific options specifying that a particular value does not
matter is achieve using a regex with alternations.
Improvements to the syntax and order restrictions are planned for the
future.
Signed-off-by: John Johansen <john.johansen@canonical.com>
2012-02-24 04:19:38 -08:00
|
|
|
{
|
2014-04-07 03:16:50 -07:00
|
|
|
char *p = buffer;
|
|
|
|
int i, len = 0;
|
|
|
|
|
|
|
|
if (flags == MS_ALL_FLAGS) {
|
|
|
|
/* all flags are optional */
|
|
|
|
len = snprintf(p, size, "%s", default_match_pattern);
|
|
|
|
if (len < 0 || len >= size)
|
|
|
|
return FALSE;
|
|
|
|
return TRUE;
|
|
|
|
}
|
|
|
|
for (i = 0; i <= 31; ++i) {
|
2023-03-28 04:23:20 -07:00
|
|
|
if ((opt_flags) & (1 << i))
|
2014-04-07 03:16:50 -07:00
|
|
|
len = snprintf(p, size, "(\\x%02x|)", i + 1);
|
|
|
|
else if (flags & (1 << i))
|
|
|
|
len = snprintf(p, size, "\\x%02x", i + 1);
|
|
|
|
else /* no entry = not set */
|
|
|
|
continue;
|
|
|
|
|
|
|
|
if (len < 0 || len >= size)
|
|
|
|
return FALSE;
|
|
|
|
p += len;
|
|
|
|
size -= len;
|
|
|
|
}
|
Add mount rules
Add the ability to control mounting and unmounting
The basic form of the rules are.
[audit] [deny] mount [conds]* [device] [ -> [conds] path],
[audit] [deny] remount [conds]* [path],
[audit] [deny] umount [conds]* [path],
[audit] [deny] pivotroot [oldroot=<value>] <path> -> <profile>
remount is just a short cut for mount options=remount
where [conds] can be
fstype=<expr>
options=<expr>
conds follow the extended conditional syntax of allowing either:
* a single value after the equals, which has the same character range as
regular IDS (ie most anything but it can't be terminated with a , (comma)
and if spaces or other characters are needed it can be quoted
eg.
options=foo
options = foo
options="foo bar"
* a list of values after the equals, the list of values is enclosed within
parenthesis () and its has a slightly reduced character set but again
elements can be quoted.
the separation between elements is whitespace and commas.
eg.
options=(foo bar)
options=(foo, bar)
options=(foo , bar)
options=(foo,bar)
The rules are flexible and follow a similar pattern as network, capability,
etc.
mount, # allow all mounts, but not umount or pivotroot
mount fstype=procfs, # allow mounting procfs anywhere
mount options=(bind, ro) /foo -> /bar, # readonly bind mount
mount /dev/sda -> /mnt,
mount /dev/sd** -> /mnt/**,
mount fstype=overlayfs options=(rw,upperdir=/tmp/upper/,lowerdir=/) overlay -> /mnt/
umount,
umount /m*,
Currently variables and regexs are are supported on the device and mount
point. ie.
mount <devince> -> <mount point>,
Regexes are supported in fstype and options. The options have a further
caveat that regexs only work if the option is fs specific option.
eg. options=(upperdir=/tmp/*,lowerdir=/)
regex's will not currently work against the standard options like ro, rw
nosuid
Conditionals (fstype) can only be applied to the device (source) at this
time and will be disregarded in situations where the mount is manipulating
an existing mount (bind, remount).
Options can be specified multiple times
mount option=rw option=(nosuid,upperdir=/foo),
and will be combined together into a single set of values
The ordering of the standard mount options (rw,ro, ...) does not matter
but the ordering of fs specific options does.
Specifying that the value of a particular option does not matter can be
acheived by providing both the positive and negative forms of and option
option=(rw,ro) options=(suid,nosuid)
For the fs specific options specifying that a particular value does not
matter is achieve using a regex with alternations.
Improvements to the syntax and order restrictions are planned for the
future.
Signed-off-by: John Johansen <john.johansen@canonical.com>
2012-02-24 04:19:38 -08:00
|
|
|
|
2014-04-07 03:16:50 -07:00
|
|
|
/* this needs to go once the backend is updated. */
|
|
|
|
if (buffer == p) {
|
|
|
|
/* match nothing - use impossible 254 as regex parser doesn't
|
|
|
|
* like the empty string
|
|
|
|
*/
|
|
|
|
if (size < 9)
|
|
|
|
return FALSE;
|
Add mount rules
Add the ability to control mounting and unmounting
The basic form of the rules are.
[audit] [deny] mount [conds]* [device] [ -> [conds] path],
[audit] [deny] remount [conds]* [path],
[audit] [deny] umount [conds]* [path],
[audit] [deny] pivotroot [oldroot=<value>] <path> -> <profile>
remount is just a short cut for mount options=remount
where [conds] can be
fstype=<expr>
options=<expr>
conds follow the extended conditional syntax of allowing either:
* a single value after the equals, which has the same character range as
regular IDS (ie most anything but it can't be terminated with a , (comma)
and if spaces or other characters are needed it can be quoted
eg.
options=foo
options = foo
options="foo bar"
* a list of values after the equals, the list of values is enclosed within
parenthesis () and its has a slightly reduced character set but again
elements can be quoted.
the separation between elements is whitespace and commas.
eg.
options=(foo bar)
options=(foo, bar)
options=(foo , bar)
options=(foo,bar)
The rules are flexible and follow a similar pattern as network, capability,
etc.
mount, # allow all mounts, but not umount or pivotroot
mount fstype=procfs, # allow mounting procfs anywhere
mount options=(bind, ro) /foo -> /bar, # readonly bind mount
mount /dev/sda -> /mnt,
mount /dev/sd** -> /mnt/**,
mount fstype=overlayfs options=(rw,upperdir=/tmp/upper/,lowerdir=/) overlay -> /mnt/
umount,
umount /m*,
Currently variables and regexs are are supported on the device and mount
point. ie.
mount <devince> -> <mount point>,
Regexes are supported in fstype and options. The options have a further
caveat that regexs only work if the option is fs specific option.
eg. options=(upperdir=/tmp/*,lowerdir=/)
regex's will not currently work against the standard options like ro, rw
nosuid
Conditionals (fstype) can only be applied to the device (source) at this
time and will be disregarded in situations where the mount is manipulating
an existing mount (bind, remount).
Options can be specified multiple times
mount option=rw option=(nosuid,upperdir=/foo),
and will be combined together into a single set of values
The ordering of the standard mount options (rw,ro, ...) does not matter
but the ordering of fs specific options does.
Specifying that the value of a particular option does not matter can be
acheived by providing both the positive and negative forms of and option
option=(rw,ro) options=(suid,nosuid)
For the fs specific options specifying that a particular value does not
matter is achieve using a regex with alternations.
Improvements to the syntax and order restrictions are planned for the
future.
Signed-off-by: John Johansen <john.johansen@canonical.com>
2012-02-24 04:19:38 -08:00
|
|
|
|
2014-04-07 03:16:50 -07:00
|
|
|
strcpy(p, "(\\xfe|)");
|
|
|
|
}
|
Add mount rules
Add the ability to control mounting and unmounting
The basic form of the rules are.
[audit] [deny] mount [conds]* [device] [ -> [conds] path],
[audit] [deny] remount [conds]* [path],
[audit] [deny] umount [conds]* [path],
[audit] [deny] pivotroot [oldroot=<value>] <path> -> <profile>
remount is just a short cut for mount options=remount
where [conds] can be
fstype=<expr>
options=<expr>
conds follow the extended conditional syntax of allowing either:
* a single value after the equals, which has the same character range as
regular IDS (ie most anything but it can't be terminated with a , (comma)
and if spaces or other characters are needed it can be quoted
eg.
options=foo
options = foo
options="foo bar"
* a list of values after the equals, the list of values is enclosed within
parenthesis () and its has a slightly reduced character set but again
elements can be quoted.
the separation between elements is whitespace and commas.
eg.
options=(foo bar)
options=(foo, bar)
options=(foo , bar)
options=(foo,bar)
The rules are flexible and follow a similar pattern as network, capability,
etc.
mount, # allow all mounts, but not umount or pivotroot
mount fstype=procfs, # allow mounting procfs anywhere
mount options=(bind, ro) /foo -> /bar, # readonly bind mount
mount /dev/sda -> /mnt,
mount /dev/sd** -> /mnt/**,
mount fstype=overlayfs options=(rw,upperdir=/tmp/upper/,lowerdir=/) overlay -> /mnt/
umount,
umount /m*,
Currently variables and regexs are are supported on the device and mount
point. ie.
mount <devince> -> <mount point>,
Regexes are supported in fstype and options. The options have a further
caveat that regexs only work if the option is fs specific option.
eg. options=(upperdir=/tmp/*,lowerdir=/)
regex's will not currently work against the standard options like ro, rw
nosuid
Conditionals (fstype) can only be applied to the device (source) at this
time and will be disregarded in situations where the mount is manipulating
an existing mount (bind, remount).
Options can be specified multiple times
mount option=rw option=(nosuid,upperdir=/foo),
and will be combined together into a single set of values
The ordering of the standard mount options (rw,ro, ...) does not matter
but the ordering of fs specific options does.
Specifying that the value of a particular option does not matter can be
acheived by providing both the positive and negative forms of and option
option=(rw,ro) options=(suid,nosuid)
For the fs specific options specifying that a particular value does not
matter is achieve using a regex with alternations.
Improvements to the syntax and order restrictions are planned for the
future.
Signed-off-by: John Johansen <john.johansen@canonical.com>
2012-02-24 04:19:38 -08:00
|
|
|
|
2014-04-07 03:16:50 -07:00
|
|
|
return TRUE;
|
|
|
|
}
|
2013-09-06 13:39:41 -07:00
|
|
|
|
2014-04-07 03:16:50 -07:00
|
|
|
static int build_mnt_opts(std::string& buffer, struct value_list *opts)
|
|
|
|
{
|
|
|
|
struct value_list *ent;
|
|
|
|
pattern_t ptype;
|
|
|
|
int pos;
|
Add mount rules
Add the ability to control mounting and unmounting
The basic form of the rules are.
[audit] [deny] mount [conds]* [device] [ -> [conds] path],
[audit] [deny] remount [conds]* [path],
[audit] [deny] umount [conds]* [path],
[audit] [deny] pivotroot [oldroot=<value>] <path> -> <profile>
remount is just a short cut for mount options=remount
where [conds] can be
fstype=<expr>
options=<expr>
conds follow the extended conditional syntax of allowing either:
* a single value after the equals, which has the same character range as
regular IDS (ie most anything but it can't be terminated with a , (comma)
and if spaces or other characters are needed it can be quoted
eg.
options=foo
options = foo
options="foo bar"
* a list of values after the equals, the list of values is enclosed within
parenthesis () and its has a slightly reduced character set but again
elements can be quoted.
the separation between elements is whitespace and commas.
eg.
options=(foo bar)
options=(foo, bar)
options=(foo , bar)
options=(foo,bar)
The rules are flexible and follow a similar pattern as network, capability,
etc.
mount, # allow all mounts, but not umount or pivotroot
mount fstype=procfs, # allow mounting procfs anywhere
mount options=(bind, ro) /foo -> /bar, # readonly bind mount
mount /dev/sda -> /mnt,
mount /dev/sd** -> /mnt/**,
mount fstype=overlayfs options=(rw,upperdir=/tmp/upper/,lowerdir=/) overlay -> /mnt/
umount,
umount /m*,
Currently variables and regexs are are supported on the device and mount
point. ie.
mount <devince> -> <mount point>,
Regexes are supported in fstype and options. The options have a further
caveat that regexs only work if the option is fs specific option.
eg. options=(upperdir=/tmp/*,lowerdir=/)
regex's will not currently work against the standard options like ro, rw
nosuid
Conditionals (fstype) can only be applied to the device (source) at this
time and will be disregarded in situations where the mount is manipulating
an existing mount (bind, remount).
Options can be specified multiple times
mount option=rw option=(nosuid,upperdir=/foo),
and will be combined together into a single set of values
The ordering of the standard mount options (rw,ro, ...) does not matter
but the ordering of fs specific options does.
Specifying that the value of a particular option does not matter can be
acheived by providing both the positive and negative forms of and option
option=(rw,ro) options=(suid,nosuid)
For the fs specific options specifying that a particular value does not
matter is achieve using a regex with alternations.
Improvements to the syntax and order restrictions are planned for the
future.
Signed-off-by: John Johansen <john.johansen@canonical.com>
2012-02-24 04:19:38 -08:00
|
|
|
|
2014-04-07 03:16:50 -07:00
|
|
|
if (!opts) {
|
|
|
|
buffer.append(default_match_pattern);
|
|
|
|
return TRUE;
|
|
|
|
}
|
Add mount rules
Add the ability to control mounting and unmounting
The basic form of the rules are.
[audit] [deny] mount [conds]* [device] [ -> [conds] path],
[audit] [deny] remount [conds]* [path],
[audit] [deny] umount [conds]* [path],
[audit] [deny] pivotroot [oldroot=<value>] <path> -> <profile>
remount is just a short cut for mount options=remount
where [conds] can be
fstype=<expr>
options=<expr>
conds follow the extended conditional syntax of allowing either:
* a single value after the equals, which has the same character range as
regular IDS (ie most anything but it can't be terminated with a , (comma)
and if spaces or other characters are needed it can be quoted
eg.
options=foo
options = foo
options="foo bar"
* a list of values after the equals, the list of values is enclosed within
parenthesis () and its has a slightly reduced character set but again
elements can be quoted.
the separation between elements is whitespace and commas.
eg.
options=(foo bar)
options=(foo, bar)
options=(foo , bar)
options=(foo,bar)
The rules are flexible and follow a similar pattern as network, capability,
etc.
mount, # allow all mounts, but not umount or pivotroot
mount fstype=procfs, # allow mounting procfs anywhere
mount options=(bind, ro) /foo -> /bar, # readonly bind mount
mount /dev/sda -> /mnt,
mount /dev/sd** -> /mnt/**,
mount fstype=overlayfs options=(rw,upperdir=/tmp/upper/,lowerdir=/) overlay -> /mnt/
umount,
umount /m*,
Currently variables and regexs are are supported on the device and mount
point. ie.
mount <devince> -> <mount point>,
Regexes are supported in fstype and options. The options have a further
caveat that regexs only work if the option is fs specific option.
eg. options=(upperdir=/tmp/*,lowerdir=/)
regex's will not currently work against the standard options like ro, rw
nosuid
Conditionals (fstype) can only be applied to the device (source) at this
time and will be disregarded in situations where the mount is manipulating
an existing mount (bind, remount).
Options can be specified multiple times
mount option=rw option=(nosuid,upperdir=/foo),
and will be combined together into a single set of values
The ordering of the standard mount options (rw,ro, ...) does not matter
but the ordering of fs specific options does.
Specifying that the value of a particular option does not matter can be
acheived by providing both the positive and negative forms of and option
option=(rw,ro) options=(suid,nosuid)
For the fs specific options specifying that a particular value does not
matter is achieve using a regex with alternations.
Improvements to the syntax and order restrictions are planned for the
future.
Signed-off-by: John Johansen <john.johansen@canonical.com>
2012-02-24 04:19:38 -08:00
|
|
|
|
2014-04-07 03:16:50 -07:00
|
|
|
list_for_each(opts, ent) {
|
2015-02-12 10:19:16 -08:00
|
|
|
ptype = convert_aaregex_to_pcre(ent->value, 0, glob_default, buffer, &pos);
|
2014-04-07 03:16:50 -07:00
|
|
|
if (ptype == ePatternInvalid)
|
|
|
|
return FALSE;
|
Add mount rules
Add the ability to control mounting and unmounting
The basic form of the rules are.
[audit] [deny] mount [conds]* [device] [ -> [conds] path],
[audit] [deny] remount [conds]* [path],
[audit] [deny] umount [conds]* [path],
[audit] [deny] pivotroot [oldroot=<value>] <path> -> <profile>
remount is just a short cut for mount options=remount
where [conds] can be
fstype=<expr>
options=<expr>
conds follow the extended conditional syntax of allowing either:
* a single value after the equals, which has the same character range as
regular IDS (ie most anything but it can't be terminated with a , (comma)
and if spaces or other characters are needed it can be quoted
eg.
options=foo
options = foo
options="foo bar"
* a list of values after the equals, the list of values is enclosed within
parenthesis () and its has a slightly reduced character set but again
elements can be quoted.
the separation between elements is whitespace and commas.
eg.
options=(foo bar)
options=(foo, bar)
options=(foo , bar)
options=(foo,bar)
The rules are flexible and follow a similar pattern as network, capability,
etc.
mount, # allow all mounts, but not umount or pivotroot
mount fstype=procfs, # allow mounting procfs anywhere
mount options=(bind, ro) /foo -> /bar, # readonly bind mount
mount /dev/sda -> /mnt,
mount /dev/sd** -> /mnt/**,
mount fstype=overlayfs options=(rw,upperdir=/tmp/upper/,lowerdir=/) overlay -> /mnt/
umount,
umount /m*,
Currently variables and regexs are are supported on the device and mount
point. ie.
mount <devince> -> <mount point>,
Regexes are supported in fstype and options. The options have a further
caveat that regexs only work if the option is fs specific option.
eg. options=(upperdir=/tmp/*,lowerdir=/)
regex's will not currently work against the standard options like ro, rw
nosuid
Conditionals (fstype) can only be applied to the device (source) at this
time and will be disregarded in situations where the mount is manipulating
an existing mount (bind, remount).
Options can be specified multiple times
mount option=rw option=(nosuid,upperdir=/foo),
and will be combined together into a single set of values
The ordering of the standard mount options (rw,ro, ...) does not matter
but the ordering of fs specific options does.
Specifying that the value of a particular option does not matter can be
acheived by providing both the positive and negative forms of and option
option=(rw,ro) options=(suid,nosuid)
For the fs specific options specifying that a particular value does not
matter is achieve using a regex with alternations.
Improvements to the syntax and order restrictions are planned for the
future.
Signed-off-by: John Johansen <john.johansen@canonical.com>
2012-02-24 04:19:38 -08:00
|
|
|
|
2014-04-07 03:16:50 -07:00
|
|
|
if (ent->next)
|
|
|
|
buffer.append(",");
|
|
|
|
}
|
Add mount rules
Add the ability to control mounting and unmounting
The basic form of the rules are.
[audit] [deny] mount [conds]* [device] [ -> [conds] path],
[audit] [deny] remount [conds]* [path],
[audit] [deny] umount [conds]* [path],
[audit] [deny] pivotroot [oldroot=<value>] <path> -> <profile>
remount is just a short cut for mount options=remount
where [conds] can be
fstype=<expr>
options=<expr>
conds follow the extended conditional syntax of allowing either:
* a single value after the equals, which has the same character range as
regular IDS (ie most anything but it can't be terminated with a , (comma)
and if spaces or other characters are needed it can be quoted
eg.
options=foo
options = foo
options="foo bar"
* a list of values after the equals, the list of values is enclosed within
parenthesis () and its has a slightly reduced character set but again
elements can be quoted.
the separation between elements is whitespace and commas.
eg.
options=(foo bar)
options=(foo, bar)
options=(foo , bar)
options=(foo,bar)
The rules are flexible and follow a similar pattern as network, capability,
etc.
mount, # allow all mounts, but not umount or pivotroot
mount fstype=procfs, # allow mounting procfs anywhere
mount options=(bind, ro) /foo -> /bar, # readonly bind mount
mount /dev/sda -> /mnt,
mount /dev/sd** -> /mnt/**,
mount fstype=overlayfs options=(rw,upperdir=/tmp/upper/,lowerdir=/) overlay -> /mnt/
umount,
umount /m*,
Currently variables and regexs are are supported on the device and mount
point. ie.
mount <devince> -> <mount point>,
Regexes are supported in fstype and options. The options have a further
caveat that regexs only work if the option is fs specific option.
eg. options=(upperdir=/tmp/*,lowerdir=/)
regex's will not currently work against the standard options like ro, rw
nosuid
Conditionals (fstype) can only be applied to the device (source) at this
time and will be disregarded in situations where the mount is manipulating
an existing mount (bind, remount).
Options can be specified multiple times
mount option=rw option=(nosuid,upperdir=/foo),
and will be combined together into a single set of values
The ordering of the standard mount options (rw,ro, ...) does not matter
but the ordering of fs specific options does.
Specifying that the value of a particular option does not matter can be
acheived by providing both the positive and negative forms of and option
option=(rw,ro) options=(suid,nosuid)
For the fs specific options specifying that a particular value does not
matter is achieve using a regex with alternations.
Improvements to the syntax and order restrictions are planned for the
future.
Signed-off-by: John Johansen <john.johansen@canonical.com>
2012-02-24 04:19:38 -08:00
|
|
|
|
2014-04-07 03:16:50 -07:00
|
|
|
return TRUE;
|
|
|
|
}
|
2013-09-06 13:39:41 -07:00
|
|
|
|
2020-08-09 14:51:55 -04:00
|
|
|
void mnt_rule::warn_once(const char *name)
|
2014-04-07 03:16:50 -07:00
|
|
|
{
|
2020-08-09 14:51:55 -04:00
|
|
|
rule_t::warn_once(name, "mount rules not enforce");
|
Add mount rules
Add the ability to control mounting and unmounting
The basic form of the rules are.
[audit] [deny] mount [conds]* [device] [ -> [conds] path],
[audit] [deny] remount [conds]* [path],
[audit] [deny] umount [conds]* [path],
[audit] [deny] pivotroot [oldroot=<value>] <path> -> <profile>
remount is just a short cut for mount options=remount
where [conds] can be
fstype=<expr>
options=<expr>
conds follow the extended conditional syntax of allowing either:
* a single value after the equals, which has the same character range as
regular IDS (ie most anything but it can't be terminated with a , (comma)
and if spaces or other characters are needed it can be quoted
eg.
options=foo
options = foo
options="foo bar"
* a list of values after the equals, the list of values is enclosed within
parenthesis () and its has a slightly reduced character set but again
elements can be quoted.
the separation between elements is whitespace and commas.
eg.
options=(foo bar)
options=(foo, bar)
options=(foo , bar)
options=(foo,bar)
The rules are flexible and follow a similar pattern as network, capability,
etc.
mount, # allow all mounts, but not umount or pivotroot
mount fstype=procfs, # allow mounting procfs anywhere
mount options=(bind, ro) /foo -> /bar, # readonly bind mount
mount /dev/sda -> /mnt,
mount /dev/sd** -> /mnt/**,
mount fstype=overlayfs options=(rw,upperdir=/tmp/upper/,lowerdir=/) overlay -> /mnt/
umount,
umount /m*,
Currently variables and regexs are are supported on the device and mount
point. ie.
mount <devince> -> <mount point>,
Regexes are supported in fstype and options. The options have a further
caveat that regexs only work if the option is fs specific option.
eg. options=(upperdir=/tmp/*,lowerdir=/)
regex's will not currently work against the standard options like ro, rw
nosuid
Conditionals (fstype) can only be applied to the device (source) at this
time and will be disregarded in situations where the mount is manipulating
an existing mount (bind, remount).
Options can be specified multiple times
mount option=rw option=(nosuid,upperdir=/foo),
and will be combined together into a single set of values
The ordering of the standard mount options (rw,ro, ...) does not matter
but the ordering of fs specific options does.
Specifying that the value of a particular option does not matter can be
acheived by providing both the positive and negative forms of and option
option=(rw,ro) options=(suid,nosuid)
For the fs specific options specifying that a particular value does not
matter is achieve using a regex with alternations.
Improvements to the syntax and order restrictions are planned for the
future.
Signed-off-by: John Johansen <john.johansen@canonical.com>
2012-02-24 04:19:38 -08:00
|
|
|
}
|
|
|
|
|
2016-07-02 00:49:24 -07:00
|
|
|
|
parser: support multiple mount conditionals in a single rule
Now that flag processing for mount rules with single option
conditionals are fixed e-enable multiple mount conditionals on a
single mount rule. The mount conditionals are equivalent to specifying
multiple rules.
mount options=(a,b,c) options=(c,d),
is the same as
mount options=(a,b,c),
mount options=(c,d),
and
mount options in (a,b,c) options in (c,d),
is the same as
mount options in (a,b,c),
mount options in (c,d),
when multiple options= and options in are combined in a single rule
it is the same as the cross product of the options.
where
mount options=(a,b,c) options in (d,e),
is a single rule.
mount options=(a,b,c) options=(d,e) options in (f),
is equivalent to
mount options=(a,b,c) options in (f),
mount options=(d,e) options in (f),
and while it is not recommended that multiple options= and options in
conditions be used in a single rule.
mount options=(a,b,c) options=(d,e) options in (f) options in (g),
is equivalent to
mount options=(a,b,c) options in (f),
mount options=(a,b,c) options in (g),
mount options=(d,e) options in (f),
mount options=(d,e) options in (g),
Bug Link: https://bugs.launchpad.net/apparmor/+bug/1597017
Signed-off-by: John Johansen <john.johansen@canonical.com>
- rebased to bba1a023bf
- fixed infinite loop in mnt_rule::gen_policy_re
Signed-off-by: Alexander Mikhalitsyn <aleksandr.mikhalitsyn@canonical.com>
Acked-by: John Johansen <john.johansen@canonical.com>
2023-03-28 12:43:18 -07:00
|
|
|
int mnt_rule::gen_policy_remount(Profile &prof, int &count,
|
|
|
|
unsigned int flags, unsigned int opt_flags)
|
2016-07-02 00:49:24 -07:00
|
|
|
{
|
|
|
|
std::string mntbuf;
|
|
|
|
std::string devbuf;
|
|
|
|
std::string typebuf;
|
|
|
|
char flagsbuf[PATH_MAX + 3];
|
|
|
|
std::string optsbuf;
|
|
|
|
char class_mount_hdr[64];
|
|
|
|
const char *vec[5];
|
|
|
|
|
|
|
|
sprintf(class_mount_hdr, "\\x%02x", AA_CLASS_MOUNT);
|
|
|
|
|
|
|
|
/* remount can't be conditional on device and type */
|
|
|
|
/* rule class single byte header */
|
|
|
|
mntbuf.assign(class_mount_hdr);
|
|
|
|
if (mnt_point) {
|
|
|
|
/* both device && mnt_point or just mnt_point */
|
|
|
|
if (!convert_entry(mntbuf, mnt_point))
|
|
|
|
goto fail;
|
|
|
|
vec[0] = mntbuf.c_str();
|
|
|
|
} else {
|
|
|
|
if (!convert_entry(mntbuf, device))
|
|
|
|
goto fail;
|
|
|
|
vec[0] = mntbuf.c_str();
|
|
|
|
}
|
|
|
|
/* skip device */
|
|
|
|
vec[1] = default_match_pattern;
|
|
|
|
/* skip type */
|
|
|
|
vec[2] = default_match_pattern;
|
|
|
|
|
2016-07-02 02:23:18 -07:00
|
|
|
if (!build_mnt_flags(flagsbuf, PATH_MAX, flags & MS_REMOUNT_FLAGS,
|
2023-03-28 04:23:20 -07:00
|
|
|
opt_flags & MS_REMOUNT_FLAGS))
|
2016-07-02 00:49:24 -07:00
|
|
|
goto fail;
|
|
|
|
|
|
|
|
vec[3] = flagsbuf;
|
|
|
|
|
2021-06-11 03:07:54 -07:00
|
|
|
perms_t tmpperms, tmpaudit;
|
|
|
|
if (opts) {
|
2021-06-11 03:20:18 -07:00
|
|
|
tmpperms = AA_MATCH_CONT;
|
2021-06-11 03:07:54 -07:00
|
|
|
tmpaudit = 0;
|
|
|
|
} else {
|
|
|
|
/* dependent on full expansion of any data match perms */
|
2021-06-11 03:20:18 -07:00
|
|
|
tmpperms = perms;
|
2021-08-30 14:31:03 -07:00
|
|
|
tmpaudit = audit == AUDIT_FORCE ? perms : 0;
|
2021-06-11 03:07:54 -07:00
|
|
|
}
|
|
|
|
/* match for up to but not including data
|
|
|
|
* if a data match is required this only has AA_MATCH_CONT perms
|
|
|
|
* else it has full perms
|
|
|
|
*/
|
2021-09-09 01:42:51 -07:00
|
|
|
if (!prof.policy.rules->add_rule_vec(rule_mode == RULE_DENY, tmpperms, tmpaudit, 4,
|
2016-07-02 00:49:24 -07:00
|
|
|
vec, dfaflags, false))
|
|
|
|
goto fail;
|
|
|
|
count++;
|
|
|
|
|
|
|
|
if (opts) {
|
|
|
|
/* rule with data match required */
|
|
|
|
optsbuf.clear();
|
|
|
|
if (!build_mnt_opts(optsbuf, opts))
|
|
|
|
goto fail;
|
|
|
|
vec[4] = optsbuf.c_str();
|
2021-09-09 01:42:51 -07:00
|
|
|
if (!prof.policy.rules->add_rule_vec(rule_mode == RULE_DENY, perms,
|
2021-08-30 14:31:03 -07:00
|
|
|
(audit == AUDIT_FORCE ? perms : 0),
|
2016-07-02 00:49:24 -07:00
|
|
|
5, vec, dfaflags, false))
|
|
|
|
goto fail;
|
|
|
|
count++;
|
|
|
|
}
|
|
|
|
|
|
|
|
return RULE_OK;
|
|
|
|
|
|
|
|
fail:
|
|
|
|
return RULE_ERROR;
|
|
|
|
}
|
|
|
|
|
parser: support multiple mount conditionals in a single rule
Now that flag processing for mount rules with single option
conditionals are fixed e-enable multiple mount conditionals on a
single mount rule. The mount conditionals are equivalent to specifying
multiple rules.
mount options=(a,b,c) options=(c,d),
is the same as
mount options=(a,b,c),
mount options=(c,d),
and
mount options in (a,b,c) options in (c,d),
is the same as
mount options in (a,b,c),
mount options in (c,d),
when multiple options= and options in are combined in a single rule
it is the same as the cross product of the options.
where
mount options=(a,b,c) options in (d,e),
is a single rule.
mount options=(a,b,c) options=(d,e) options in (f),
is equivalent to
mount options=(a,b,c) options in (f),
mount options=(d,e) options in (f),
and while it is not recommended that multiple options= and options in
conditions be used in a single rule.
mount options=(a,b,c) options=(d,e) options in (f) options in (g),
is equivalent to
mount options=(a,b,c) options in (f),
mount options=(a,b,c) options in (g),
mount options=(d,e) options in (f),
mount options=(d,e) options in (g),
Bug Link: https://bugs.launchpad.net/apparmor/+bug/1597017
Signed-off-by: John Johansen <john.johansen@canonical.com>
- rebased to bba1a023bf
- fixed infinite loop in mnt_rule::gen_policy_re
Signed-off-by: Alexander Mikhalitsyn <aleksandr.mikhalitsyn@canonical.com>
Acked-by: John Johansen <john.johansen@canonical.com>
2023-03-28 12:43:18 -07:00
|
|
|
int mnt_rule::gen_policy_bind_mount(Profile &prof, int &count,
|
|
|
|
unsigned int flags, unsigned int opt_flags)
|
2016-07-02 00:49:24 -07:00
|
|
|
{
|
|
|
|
std::string mntbuf;
|
|
|
|
std::string devbuf;
|
|
|
|
std::string typebuf;
|
|
|
|
char flagsbuf[PATH_MAX + 3];
|
|
|
|
std::string optsbuf;
|
|
|
|
char class_mount_hdr[64];
|
|
|
|
const char *vec[5];
|
|
|
|
|
|
|
|
sprintf(class_mount_hdr, "\\x%02x", AA_CLASS_MOUNT);
|
|
|
|
|
|
|
|
/* bind mount rules can't be conditional on dev_type or data */
|
|
|
|
/* rule class single byte header */
|
|
|
|
mntbuf.assign(class_mount_hdr);
|
|
|
|
if (!convert_entry(mntbuf, mnt_point))
|
|
|
|
goto fail;
|
|
|
|
vec[0] = mntbuf.c_str();
|
|
|
|
if (!clear_and_convert_entry(devbuf, device))
|
|
|
|
goto fail;
|
|
|
|
vec[1] = devbuf.c_str();
|
|
|
|
/* skip type */
|
|
|
|
vec[2] = default_match_pattern;
|
|
|
|
|
2016-07-02 02:23:18 -07:00
|
|
|
if (!build_mnt_flags(flagsbuf, PATH_MAX, flags & MS_BIND_FLAGS,
|
2023-03-28 04:23:20 -07:00
|
|
|
opt_flags & MS_BIND_FLAGS))
|
2016-07-02 00:49:24 -07:00
|
|
|
goto fail;
|
|
|
|
vec[3] = flagsbuf;
|
2021-09-09 01:42:51 -07:00
|
|
|
if (!prof.policy.rules->add_rule_vec(rule_mode == RULE_DENY, perms, audit == AUDIT_FORCE ? perms : 0,
|
2021-06-11 03:49:06 -07:00
|
|
|
4, vec,
|
2016-07-02 00:49:24 -07:00
|
|
|
dfaflags, false))
|
|
|
|
goto fail;
|
|
|
|
count++;
|
|
|
|
|
|
|
|
return RULE_OK;
|
|
|
|
|
|
|
|
fail:
|
|
|
|
return RULE_ERROR;
|
|
|
|
}
|
|
|
|
|
parser: support multiple mount conditionals in a single rule
Now that flag processing for mount rules with single option
conditionals are fixed e-enable multiple mount conditionals on a
single mount rule. The mount conditionals are equivalent to specifying
multiple rules.
mount options=(a,b,c) options=(c,d),
is the same as
mount options=(a,b,c),
mount options=(c,d),
and
mount options in (a,b,c) options in (c,d),
is the same as
mount options in (a,b,c),
mount options in (c,d),
when multiple options= and options in are combined in a single rule
it is the same as the cross product of the options.
where
mount options=(a,b,c) options in (d,e),
is a single rule.
mount options=(a,b,c) options=(d,e) options in (f),
is equivalent to
mount options=(a,b,c) options in (f),
mount options=(d,e) options in (f),
and while it is not recommended that multiple options= and options in
conditions be used in a single rule.
mount options=(a,b,c) options=(d,e) options in (f) options in (g),
is equivalent to
mount options=(a,b,c) options in (f),
mount options=(a,b,c) options in (g),
mount options=(d,e) options in (f),
mount options=(d,e) options in (g),
Bug Link: https://bugs.launchpad.net/apparmor/+bug/1597017
Signed-off-by: John Johansen <john.johansen@canonical.com>
- rebased to bba1a023bf
- fixed infinite loop in mnt_rule::gen_policy_re
Signed-off-by: Alexander Mikhalitsyn <aleksandr.mikhalitsyn@canonical.com>
Acked-by: John Johansen <john.johansen@canonical.com>
2023-03-28 12:43:18 -07:00
|
|
|
int mnt_rule::gen_policy_change_mount_type(Profile &prof, int &count,
|
|
|
|
unsigned int flags,
|
|
|
|
unsigned int opt_flags)
|
2016-07-02 00:49:24 -07:00
|
|
|
{
|
|
|
|
std::string mntbuf;
|
|
|
|
std::string devbuf;
|
|
|
|
std::string typebuf;
|
|
|
|
char flagsbuf[PATH_MAX + 3];
|
|
|
|
std::string optsbuf;
|
|
|
|
char class_mount_hdr[64];
|
|
|
|
const char *vec[5];
|
|
|
|
|
|
|
|
sprintf(class_mount_hdr, "\\x%02x", AA_CLASS_MOUNT);
|
|
|
|
|
|
|
|
/* change type base rules can not be conditional on device,
|
|
|
|
* device type or data
|
|
|
|
*/
|
|
|
|
/* rule class single byte header */
|
|
|
|
mntbuf.assign(class_mount_hdr);
|
|
|
|
if (!convert_entry(mntbuf, mnt_point))
|
|
|
|
goto fail;
|
|
|
|
vec[0] = mntbuf.c_str();
|
|
|
|
/* skip device and type */
|
|
|
|
vec[1] = default_match_pattern;
|
|
|
|
vec[2] = default_match_pattern;
|
|
|
|
|
2016-07-02 02:23:18 -07:00
|
|
|
if (!build_mnt_flags(flagsbuf, PATH_MAX, flags & MS_MAKE_FLAGS,
|
2023-03-28 04:23:20 -07:00
|
|
|
opt_flags & MS_MAKE_FLAGS))
|
2016-07-02 00:49:24 -07:00
|
|
|
goto fail;
|
|
|
|
vec[3] = flagsbuf;
|
2021-09-09 01:42:51 -07:00
|
|
|
if (!prof.policy.rules->add_rule_vec(rule_mode == RULE_DENY, perms, audit == AUDIT_FORCE ? perms : 0,
|
2021-06-11 03:49:06 -07:00
|
|
|
4, vec,
|
2016-07-02 00:49:24 -07:00
|
|
|
dfaflags, false))
|
|
|
|
goto fail;
|
|
|
|
count++;
|
|
|
|
|
|
|
|
return RULE_OK;
|
|
|
|
|
|
|
|
fail:
|
|
|
|
return RULE_ERROR;
|
|
|
|
}
|
|
|
|
|
parser: support multiple mount conditionals in a single rule
Now that flag processing for mount rules with single option
conditionals are fixed e-enable multiple mount conditionals on a
single mount rule. The mount conditionals are equivalent to specifying
multiple rules.
mount options=(a,b,c) options=(c,d),
is the same as
mount options=(a,b,c),
mount options=(c,d),
and
mount options in (a,b,c) options in (c,d),
is the same as
mount options in (a,b,c),
mount options in (c,d),
when multiple options= and options in are combined in a single rule
it is the same as the cross product of the options.
where
mount options=(a,b,c) options in (d,e),
is a single rule.
mount options=(a,b,c) options=(d,e) options in (f),
is equivalent to
mount options=(a,b,c) options in (f),
mount options=(d,e) options in (f),
and while it is not recommended that multiple options= and options in
conditions be used in a single rule.
mount options=(a,b,c) options=(d,e) options in (f) options in (g),
is equivalent to
mount options=(a,b,c) options in (f),
mount options=(a,b,c) options in (g),
mount options=(d,e) options in (f),
mount options=(d,e) options in (g),
Bug Link: https://bugs.launchpad.net/apparmor/+bug/1597017
Signed-off-by: John Johansen <john.johansen@canonical.com>
- rebased to bba1a023bf
- fixed infinite loop in mnt_rule::gen_policy_re
Signed-off-by: Alexander Mikhalitsyn <aleksandr.mikhalitsyn@canonical.com>
Acked-by: John Johansen <john.johansen@canonical.com>
2023-03-28 12:43:18 -07:00
|
|
|
int mnt_rule::gen_policy_move_mount(Profile &prof, int &count,
|
|
|
|
unsigned int flags, unsigned int opt_flags)
|
2016-07-02 00:49:24 -07:00
|
|
|
{
|
|
|
|
std::string mntbuf;
|
|
|
|
std::string devbuf;
|
|
|
|
std::string typebuf;
|
|
|
|
char flagsbuf[PATH_MAX + 3];
|
|
|
|
std::string optsbuf;
|
|
|
|
char class_mount_hdr[64];
|
|
|
|
const char *vec[5];
|
|
|
|
|
|
|
|
sprintf(class_mount_hdr, "\\x%02x", AA_CLASS_MOUNT);
|
|
|
|
|
|
|
|
/* mount move rules can not be conditional on dev_type,
|
|
|
|
* or data
|
|
|
|
*/
|
|
|
|
/* rule class single byte header */
|
|
|
|
mntbuf.assign(class_mount_hdr);
|
|
|
|
if (!convert_entry(mntbuf, mnt_point))
|
|
|
|
goto fail;
|
|
|
|
vec[0] = mntbuf.c_str();
|
|
|
|
if (!clear_and_convert_entry(devbuf, device))
|
|
|
|
goto fail;
|
|
|
|
vec[1] = devbuf.c_str();
|
|
|
|
/* skip type */
|
|
|
|
vec[2] = default_match_pattern;
|
|
|
|
|
2016-07-02 02:23:18 -07:00
|
|
|
if (!build_mnt_flags(flagsbuf, PATH_MAX, flags & MS_MOVE_FLAGS,
|
2023-03-28 04:23:20 -07:00
|
|
|
opt_flags & MS_MOVE_FLAGS))
|
2016-07-02 00:49:24 -07:00
|
|
|
goto fail;
|
|
|
|
vec[3] = flagsbuf;
|
2021-09-09 01:42:51 -07:00
|
|
|
if (!prof.policy.rules->add_rule_vec(rule_mode == RULE_DENY, perms, audit == AUDIT_FORCE ? perms : 0,
|
2021-06-11 03:49:06 -07:00
|
|
|
4, vec,
|
2016-07-02 00:49:24 -07:00
|
|
|
dfaflags, false))
|
|
|
|
goto fail;
|
|
|
|
count++;
|
|
|
|
|
|
|
|
return RULE_OK;
|
|
|
|
|
|
|
|
fail:
|
|
|
|
return RULE_ERROR;
|
|
|
|
}
|
|
|
|
|
parser: support multiple mount conditionals in a single rule
Now that flag processing for mount rules with single option
conditionals are fixed e-enable multiple mount conditionals on a
single mount rule. The mount conditionals are equivalent to specifying
multiple rules.
mount options=(a,b,c) options=(c,d),
is the same as
mount options=(a,b,c),
mount options=(c,d),
and
mount options in (a,b,c) options in (c,d),
is the same as
mount options in (a,b,c),
mount options in (c,d),
when multiple options= and options in are combined in a single rule
it is the same as the cross product of the options.
where
mount options=(a,b,c) options in (d,e),
is a single rule.
mount options=(a,b,c) options=(d,e) options in (f),
is equivalent to
mount options=(a,b,c) options in (f),
mount options=(d,e) options in (f),
and while it is not recommended that multiple options= and options in
conditions be used in a single rule.
mount options=(a,b,c) options=(d,e) options in (f) options in (g),
is equivalent to
mount options=(a,b,c) options in (f),
mount options=(a,b,c) options in (g),
mount options=(d,e) options in (f),
mount options=(d,e) options in (g),
Bug Link: https://bugs.launchpad.net/apparmor/+bug/1597017
Signed-off-by: John Johansen <john.johansen@canonical.com>
- rebased to bba1a023bf
- fixed infinite loop in mnt_rule::gen_policy_re
Signed-off-by: Alexander Mikhalitsyn <aleksandr.mikhalitsyn@canonical.com>
Acked-by: John Johansen <john.johansen@canonical.com>
2023-03-28 12:43:18 -07:00
|
|
|
int mnt_rule::gen_policy_new_mount(Profile &prof, int &count,
|
|
|
|
unsigned int flags, unsigned int opt_flags)
|
2016-07-02 00:49:24 -07:00
|
|
|
{
|
|
|
|
std::string mntbuf;
|
|
|
|
std::string devbuf;
|
|
|
|
std::string typebuf;
|
|
|
|
char flagsbuf[PATH_MAX + 3];
|
|
|
|
std::string optsbuf;
|
|
|
|
char class_mount_hdr[64];
|
|
|
|
const char *vec[5];
|
|
|
|
|
|
|
|
sprintf(class_mount_hdr, "\\x%02x", AA_CLASS_MOUNT);
|
|
|
|
|
|
|
|
/* rule class single byte header */
|
|
|
|
mntbuf.assign(class_mount_hdr);
|
|
|
|
if (!convert_entry(mntbuf, mnt_point))
|
|
|
|
goto fail;
|
|
|
|
vec[0] = mntbuf.c_str();
|
|
|
|
if (!clear_and_convert_entry(devbuf, device))
|
|
|
|
goto fail;
|
|
|
|
vec[1] = devbuf.c_str();
|
|
|
|
typebuf.clear();
|
|
|
|
if (!build_list_val_expr(typebuf, dev_type))
|
|
|
|
goto fail;
|
|
|
|
vec[2] = typebuf.c_str();
|
|
|
|
|
2016-07-02 02:23:18 -07:00
|
|
|
if (!build_mnt_flags(flagsbuf, PATH_MAX, flags & MS_NEW_FLAGS,
|
2023-03-28 04:23:20 -07:00
|
|
|
opt_flags & MS_NEW_FLAGS))
|
2016-07-02 00:49:24 -07:00
|
|
|
goto fail;
|
|
|
|
vec[3] = flagsbuf;
|
|
|
|
|
2021-06-11 03:07:54 -07:00
|
|
|
perms_t tmpperms, tmpaudit;
|
|
|
|
if (opts) {
|
2021-06-11 03:20:18 -07:00
|
|
|
tmpperms = AA_MATCH_CONT;
|
2021-06-11 03:07:54 -07:00
|
|
|
tmpaudit = 0;
|
|
|
|
} else {
|
2021-06-11 03:20:18 -07:00
|
|
|
tmpperms = perms;
|
2021-08-30 14:31:03 -07:00
|
|
|
tmpaudit = audit == AUDIT_FORCE ? perms : 0;
|
2021-06-11 03:07:54 -07:00
|
|
|
}
|
2016-07-02 00:49:24 -07:00
|
|
|
/* rule for match without required data || data MATCH_CONT */
|
2021-09-09 01:42:51 -07:00
|
|
|
if (!prof.policy.rules->add_rule_vec(rule_mode == RULE_DENY, tmpperms, tmpaudit, 4,
|
2016-07-02 00:49:24 -07:00
|
|
|
vec, dfaflags, false))
|
|
|
|
goto fail;
|
|
|
|
count++;
|
|
|
|
|
|
|
|
if (opts) {
|
|
|
|
/* rule with data match required */
|
|
|
|
optsbuf.clear();
|
|
|
|
if (!build_mnt_opts(optsbuf, opts))
|
|
|
|
goto fail;
|
|
|
|
vec[4] = optsbuf.c_str();
|
2021-09-09 01:42:51 -07:00
|
|
|
if (!prof.policy.rules->add_rule_vec(rule_mode == RULE_DENY, perms,
|
2021-08-30 14:31:03 -07:00
|
|
|
audit == AUDIT_FORCE ? perms : 0,
|
2016-07-02 00:49:24 -07:00
|
|
|
5, vec, dfaflags, false))
|
|
|
|
goto fail;
|
|
|
|
count++;
|
|
|
|
}
|
|
|
|
|
|
|
|
return RULE_OK;
|
|
|
|
|
|
|
|
fail:
|
|
|
|
return RULE_ERROR;
|
|
|
|
}
|
|
|
|
|
parser: support multiple mount conditionals in a single rule
Now that flag processing for mount rules with single option
conditionals are fixed e-enable multiple mount conditionals on a
single mount rule. The mount conditionals are equivalent to specifying
multiple rules.
mount options=(a,b,c) options=(c,d),
is the same as
mount options=(a,b,c),
mount options=(c,d),
and
mount options in (a,b,c) options in (c,d),
is the same as
mount options in (a,b,c),
mount options in (c,d),
when multiple options= and options in are combined in a single rule
it is the same as the cross product of the options.
where
mount options=(a,b,c) options in (d,e),
is a single rule.
mount options=(a,b,c) options=(d,e) options in (f),
is equivalent to
mount options=(a,b,c) options in (f),
mount options=(d,e) options in (f),
and while it is not recommended that multiple options= and options in
conditions be used in a single rule.
mount options=(a,b,c) options=(d,e) options in (f) options in (g),
is equivalent to
mount options=(a,b,c) options in (f),
mount options=(a,b,c) options in (g),
mount options=(d,e) options in (f),
mount options=(d,e) options in (g),
Bug Link: https://bugs.launchpad.net/apparmor/+bug/1597017
Signed-off-by: John Johansen <john.johansen@canonical.com>
- rebased to bba1a023bf
- fixed infinite loop in mnt_rule::gen_policy_re
Signed-off-by: Alexander Mikhalitsyn <aleksandr.mikhalitsyn@canonical.com>
Acked-by: John Johansen <john.johansen@canonical.com>
2023-03-28 12:43:18 -07:00
|
|
|
int mnt_rule::gen_flag_rules(Profile &prof, int &count, unsigned int flags,
|
|
|
|
unsigned int opt_flags)
|
Add mount rules
Add the ability to control mounting and unmounting
The basic form of the rules are.
[audit] [deny] mount [conds]* [device] [ -> [conds] path],
[audit] [deny] remount [conds]* [path],
[audit] [deny] umount [conds]* [path],
[audit] [deny] pivotroot [oldroot=<value>] <path> -> <profile>
remount is just a short cut for mount options=remount
where [conds] can be
fstype=<expr>
options=<expr>
conds follow the extended conditional syntax of allowing either:
* a single value after the equals, which has the same character range as
regular IDS (ie most anything but it can't be terminated with a , (comma)
and if spaces or other characters are needed it can be quoted
eg.
options=foo
options = foo
options="foo bar"
* a list of values after the equals, the list of values is enclosed within
parenthesis () and its has a slightly reduced character set but again
elements can be quoted.
the separation between elements is whitespace and commas.
eg.
options=(foo bar)
options=(foo, bar)
options=(foo , bar)
options=(foo,bar)
The rules are flexible and follow a similar pattern as network, capability,
etc.
mount, # allow all mounts, but not umount or pivotroot
mount fstype=procfs, # allow mounting procfs anywhere
mount options=(bind, ro) /foo -> /bar, # readonly bind mount
mount /dev/sda -> /mnt,
mount /dev/sd** -> /mnt/**,
mount fstype=overlayfs options=(rw,upperdir=/tmp/upper/,lowerdir=/) overlay -> /mnt/
umount,
umount /m*,
Currently variables and regexs are are supported on the device and mount
point. ie.
mount <devince> -> <mount point>,
Regexes are supported in fstype and options. The options have a further
caveat that regexs only work if the option is fs specific option.
eg. options=(upperdir=/tmp/*,lowerdir=/)
regex's will not currently work against the standard options like ro, rw
nosuid
Conditionals (fstype) can only be applied to the device (source) at this
time and will be disregarded in situations where the mount is manipulating
an existing mount (bind, remount).
Options can be specified multiple times
mount option=rw option=(nosuid,upperdir=/foo),
and will be combined together into a single set of values
The ordering of the standard mount options (rw,ro, ...) does not matter
but the ordering of fs specific options does.
Specifying that the value of a particular option does not matter can be
acheived by providing both the positive and negative forms of and option
option=(rw,ro) options=(suid,nosuid)
For the fs specific options specifying that a particular value does not
matter is achieve using a regex with alternations.
Improvements to the syntax and order restrictions are planned for the
future.
Signed-off-by: John Johansen <john.johansen@canonical.com>
2012-02-24 04:19:38 -08:00
|
|
|
{
|
2023-03-28 04:23:20 -07:00
|
|
|
/*
|
|
|
|
* XXX: added !flags to cover cases like:
|
|
|
|
* mount options in (bind) /d -> /4,
|
|
|
|
*/
|
2021-06-11 03:20:18 -07:00
|
|
|
if ((perms & AA_MAY_MOUNT) && (!flags || flags == MS_ALL_FLAGS)) {
|
2016-07-02 01:56:51 -07:00
|
|
|
/* no mount flags specified, generate multiple rules */
|
|
|
|
if (!device && !dev_type &&
|
parser: support multiple mount conditionals in a single rule
Now that flag processing for mount rules with single option
conditionals are fixed e-enable multiple mount conditionals on a
single mount rule. The mount conditionals are equivalent to specifying
multiple rules.
mount options=(a,b,c) options=(c,d),
is the same as
mount options=(a,b,c),
mount options=(c,d),
and
mount options in (a,b,c) options in (c,d),
is the same as
mount options in (a,b,c),
mount options in (c,d),
when multiple options= and options in are combined in a single rule
it is the same as the cross product of the options.
where
mount options=(a,b,c) options in (d,e),
is a single rule.
mount options=(a,b,c) options=(d,e) options in (f),
is equivalent to
mount options=(a,b,c) options in (f),
mount options=(d,e) options in (f),
and while it is not recommended that multiple options= and options in
conditions be used in a single rule.
mount options=(a,b,c) options=(d,e) options in (f) options in (g),
is equivalent to
mount options=(a,b,c) options in (f),
mount options=(a,b,c) options in (g),
mount options=(d,e) options in (f),
mount options=(d,e) options in (g),
Bug Link: https://bugs.launchpad.net/apparmor/+bug/1597017
Signed-off-by: John Johansen <john.johansen@canonical.com>
- rebased to bba1a023bf
- fixed infinite loop in mnt_rule::gen_policy_re
Signed-off-by: Alexander Mikhalitsyn <aleksandr.mikhalitsyn@canonical.com>
Acked-by: John Johansen <john.johansen@canonical.com>
2023-03-28 12:43:18 -07:00
|
|
|
gen_policy_remount(prof, count, flags, opt_flags) == RULE_ERROR)
|
|
|
|
return RULE_ERROR;
|
2016-07-02 01:56:51 -07:00
|
|
|
if (!dev_type && !opts &&
|
parser: support multiple mount conditionals in a single rule
Now that flag processing for mount rules with single option
conditionals are fixed e-enable multiple mount conditionals on a
single mount rule. The mount conditionals are equivalent to specifying
multiple rules.
mount options=(a,b,c) options=(c,d),
is the same as
mount options=(a,b,c),
mount options=(c,d),
and
mount options in (a,b,c) options in (c,d),
is the same as
mount options in (a,b,c),
mount options in (c,d),
when multiple options= and options in are combined in a single rule
it is the same as the cross product of the options.
where
mount options=(a,b,c) options in (d,e),
is a single rule.
mount options=(a,b,c) options=(d,e) options in (f),
is equivalent to
mount options=(a,b,c) options in (f),
mount options=(d,e) options in (f),
and while it is not recommended that multiple options= and options in
conditions be used in a single rule.
mount options=(a,b,c) options=(d,e) options in (f) options in (g),
is equivalent to
mount options=(a,b,c) options in (f),
mount options=(a,b,c) options in (g),
mount options=(d,e) options in (f),
mount options=(d,e) options in (g),
Bug Link: https://bugs.launchpad.net/apparmor/+bug/1597017
Signed-off-by: John Johansen <john.johansen@canonical.com>
- rebased to bba1a023bf
- fixed infinite loop in mnt_rule::gen_policy_re
Signed-off-by: Alexander Mikhalitsyn <aleksandr.mikhalitsyn@canonical.com>
Acked-by: John Johansen <john.johansen@canonical.com>
2023-03-28 12:43:18 -07:00
|
|
|
gen_policy_bind_mount(prof, count, flags, opt_flags) == RULE_ERROR)
|
|
|
|
return RULE_ERROR;
|
2016-07-02 01:56:51 -07:00
|
|
|
if (!device && !dev_type && !opts &&
|
parser: support multiple mount conditionals in a single rule
Now that flag processing for mount rules with single option
conditionals are fixed e-enable multiple mount conditionals on a
single mount rule. The mount conditionals are equivalent to specifying
multiple rules.
mount options=(a,b,c) options=(c,d),
is the same as
mount options=(a,b,c),
mount options=(c,d),
and
mount options in (a,b,c) options in (c,d),
is the same as
mount options in (a,b,c),
mount options in (c,d),
when multiple options= and options in are combined in a single rule
it is the same as the cross product of the options.
where
mount options=(a,b,c) options in (d,e),
is a single rule.
mount options=(a,b,c) options=(d,e) options in (f),
is equivalent to
mount options=(a,b,c) options in (f),
mount options=(d,e) options in (f),
and while it is not recommended that multiple options= and options in
conditions be used in a single rule.
mount options=(a,b,c) options=(d,e) options in (f) options in (g),
is equivalent to
mount options=(a,b,c) options in (f),
mount options=(a,b,c) options in (g),
mount options=(d,e) options in (f),
mount options=(d,e) options in (g),
Bug Link: https://bugs.launchpad.net/apparmor/+bug/1597017
Signed-off-by: John Johansen <john.johansen@canonical.com>
- rebased to bba1a023bf
- fixed infinite loop in mnt_rule::gen_policy_re
Signed-off-by: Alexander Mikhalitsyn <aleksandr.mikhalitsyn@canonical.com>
Acked-by: John Johansen <john.johansen@canonical.com>
2023-03-28 12:43:18 -07:00
|
|
|
gen_policy_change_mount_type(prof, count, flags, opt_flags) == RULE_ERROR)
|
|
|
|
return RULE_ERROR;
|
2016-07-02 01:56:51 -07:00
|
|
|
if (!dev_type && !opts &&
|
parser: support multiple mount conditionals in a single rule
Now that flag processing for mount rules with single option
conditionals are fixed e-enable multiple mount conditionals on a
single mount rule. The mount conditionals are equivalent to specifying
multiple rules.
mount options=(a,b,c) options=(c,d),
is the same as
mount options=(a,b,c),
mount options=(c,d),
and
mount options in (a,b,c) options in (c,d),
is the same as
mount options in (a,b,c),
mount options in (c,d),
when multiple options= and options in are combined in a single rule
it is the same as the cross product of the options.
where
mount options=(a,b,c) options in (d,e),
is a single rule.
mount options=(a,b,c) options=(d,e) options in (f),
is equivalent to
mount options=(a,b,c) options in (f),
mount options=(d,e) options in (f),
and while it is not recommended that multiple options= and options in
conditions be used in a single rule.
mount options=(a,b,c) options=(d,e) options in (f) options in (g),
is equivalent to
mount options=(a,b,c) options in (f),
mount options=(a,b,c) options in (g),
mount options=(d,e) options in (f),
mount options=(d,e) options in (g),
Bug Link: https://bugs.launchpad.net/apparmor/+bug/1597017
Signed-off-by: John Johansen <john.johansen@canonical.com>
- rebased to bba1a023bf
- fixed infinite loop in mnt_rule::gen_policy_re
Signed-off-by: Alexander Mikhalitsyn <aleksandr.mikhalitsyn@canonical.com>
Acked-by: John Johansen <john.johansen@canonical.com>
2023-03-28 12:43:18 -07:00
|
|
|
gen_policy_move_mount(prof, count, flags, opt_flags) == RULE_ERROR)
|
|
|
|
return RULE_ERROR;
|
|
|
|
|
|
|
|
return gen_policy_new_mount(prof, count, flags, opt_flags);
|
2021-06-11 03:20:18 -07:00
|
|
|
} else if ((perms & AA_MAY_MOUNT) && (flags & MS_REMOUNT)
|
2016-07-02 01:56:51 -07:00
|
|
|
&& !device && !dev_type) {
|
parser: support multiple mount conditionals in a single rule
Now that flag processing for mount rules with single option
conditionals are fixed e-enable multiple mount conditionals on a
single mount rule. The mount conditionals are equivalent to specifying
multiple rules.
mount options=(a,b,c) options=(c,d),
is the same as
mount options=(a,b,c),
mount options=(c,d),
and
mount options in (a,b,c) options in (c,d),
is the same as
mount options in (a,b,c),
mount options in (c,d),
when multiple options= and options in are combined in a single rule
it is the same as the cross product of the options.
where
mount options=(a,b,c) options in (d,e),
is a single rule.
mount options=(a,b,c) options=(d,e) options in (f),
is equivalent to
mount options=(a,b,c) options in (f),
mount options=(d,e) options in (f),
and while it is not recommended that multiple options= and options in
conditions be used in a single rule.
mount options=(a,b,c) options=(d,e) options in (f) options in (g),
is equivalent to
mount options=(a,b,c) options in (f),
mount options=(a,b,c) options in (g),
mount options=(d,e) options in (f),
mount options=(d,e) options in (g),
Bug Link: https://bugs.launchpad.net/apparmor/+bug/1597017
Signed-off-by: John Johansen <john.johansen@canonical.com>
- rebased to bba1a023bf
- fixed infinite loop in mnt_rule::gen_policy_re
Signed-off-by: Alexander Mikhalitsyn <aleksandr.mikhalitsyn@canonical.com>
Acked-by: John Johansen <john.johansen@canonical.com>
2023-03-28 12:43:18 -07:00
|
|
|
return gen_policy_remount(prof, count, flags, opt_flags);
|
2021-06-11 03:20:18 -07:00
|
|
|
} else if ((perms & AA_MAY_MOUNT) && (flags & MS_BIND)
|
2016-07-02 01:56:51 -07:00
|
|
|
&& !dev_type && !opts) {
|
parser: support multiple mount conditionals in a single rule
Now that flag processing for mount rules with single option
conditionals are fixed e-enable multiple mount conditionals on a
single mount rule. The mount conditionals are equivalent to specifying
multiple rules.
mount options=(a,b,c) options=(c,d),
is the same as
mount options=(a,b,c),
mount options=(c,d),
and
mount options in (a,b,c) options in (c,d),
is the same as
mount options in (a,b,c),
mount options in (c,d),
when multiple options= and options in are combined in a single rule
it is the same as the cross product of the options.
where
mount options=(a,b,c) options in (d,e),
is a single rule.
mount options=(a,b,c) options=(d,e) options in (f),
is equivalent to
mount options=(a,b,c) options in (f),
mount options=(d,e) options in (f),
and while it is not recommended that multiple options= and options in
conditions be used in a single rule.
mount options=(a,b,c) options=(d,e) options in (f) options in (g),
is equivalent to
mount options=(a,b,c) options in (f),
mount options=(a,b,c) options in (g),
mount options=(d,e) options in (f),
mount options=(d,e) options in (g),
Bug Link: https://bugs.launchpad.net/apparmor/+bug/1597017
Signed-off-by: John Johansen <john.johansen@canonical.com>
- rebased to bba1a023bf
- fixed infinite loop in mnt_rule::gen_policy_re
Signed-off-by: Alexander Mikhalitsyn <aleksandr.mikhalitsyn@canonical.com>
Acked-by: John Johansen <john.johansen@canonical.com>
2023-03-28 12:43:18 -07:00
|
|
|
return gen_policy_bind_mount(prof, count, flags, opt_flags);
|
2021-06-11 03:20:18 -07:00
|
|
|
} else if ((perms & AA_MAY_MOUNT) &&
|
2016-07-02 02:23:18 -07:00
|
|
|
(flags & (MS_MAKE_CMDS))
|
2016-07-02 01:56:51 -07:00
|
|
|
&& !device && !dev_type && !opts) {
|
parser: support multiple mount conditionals in a single rule
Now that flag processing for mount rules with single option
conditionals are fixed e-enable multiple mount conditionals on a
single mount rule. The mount conditionals are equivalent to specifying
multiple rules.
mount options=(a,b,c) options=(c,d),
is the same as
mount options=(a,b,c),
mount options=(c,d),
and
mount options in (a,b,c) options in (c,d),
is the same as
mount options in (a,b,c),
mount options in (c,d),
when multiple options= and options in are combined in a single rule
it is the same as the cross product of the options.
where
mount options=(a,b,c) options in (d,e),
is a single rule.
mount options=(a,b,c) options=(d,e) options in (f),
is equivalent to
mount options=(a,b,c) options in (f),
mount options=(d,e) options in (f),
and while it is not recommended that multiple options= and options in
conditions be used in a single rule.
mount options=(a,b,c) options=(d,e) options in (f) options in (g),
is equivalent to
mount options=(a,b,c) options in (f),
mount options=(a,b,c) options in (g),
mount options=(d,e) options in (f),
mount options=(d,e) options in (g),
Bug Link: https://bugs.launchpad.net/apparmor/+bug/1597017
Signed-off-by: John Johansen <john.johansen@canonical.com>
- rebased to bba1a023bf
- fixed infinite loop in mnt_rule::gen_policy_re
Signed-off-by: Alexander Mikhalitsyn <aleksandr.mikhalitsyn@canonical.com>
Acked-by: John Johansen <john.johansen@canonical.com>
2023-03-28 12:43:18 -07:00
|
|
|
return gen_policy_change_mount_type(prof, count, flags, opt_flags);
|
2021-06-11 03:20:18 -07:00
|
|
|
} else if ((perms & AA_MAY_MOUNT) && (flags & MS_MOVE)
|
2016-07-02 01:56:51 -07:00
|
|
|
&& !dev_type && !opts) {
|
parser: support multiple mount conditionals in a single rule
Now that flag processing for mount rules with single option
conditionals are fixed e-enable multiple mount conditionals on a
single mount rule. The mount conditionals are equivalent to specifying
multiple rules.
mount options=(a,b,c) options=(c,d),
is the same as
mount options=(a,b,c),
mount options=(c,d),
and
mount options in (a,b,c) options in (c,d),
is the same as
mount options in (a,b,c),
mount options in (c,d),
when multiple options= and options in are combined in a single rule
it is the same as the cross product of the options.
where
mount options=(a,b,c) options in (d,e),
is a single rule.
mount options=(a,b,c) options=(d,e) options in (f),
is equivalent to
mount options=(a,b,c) options in (f),
mount options=(d,e) options in (f),
and while it is not recommended that multiple options= and options in
conditions be used in a single rule.
mount options=(a,b,c) options=(d,e) options in (f) options in (g),
is equivalent to
mount options=(a,b,c) options in (f),
mount options=(a,b,c) options in (g),
mount options=(d,e) options in (f),
mount options=(d,e) options in (g),
Bug Link: https://bugs.launchpad.net/apparmor/+bug/1597017
Signed-off-by: John Johansen <john.johansen@canonical.com>
- rebased to bba1a023bf
- fixed infinite loop in mnt_rule::gen_policy_re
Signed-off-by: Alexander Mikhalitsyn <aleksandr.mikhalitsyn@canonical.com>
Acked-by: John Johansen <john.johansen@canonical.com>
2023-03-28 12:43:18 -07:00
|
|
|
return gen_policy_move_mount(prof, count, flags, opt_flags);
|
2021-06-11 03:20:18 -07:00
|
|
|
} else if ((perms & AA_MAY_MOUNT) &&
|
2023-03-28 14:35:24 -07:00
|
|
|
((flags | opt_flags) & ~MS_CMDS)) {
|
2014-04-07 03:16:50 -07:00
|
|
|
/* generic mount if flags are set that are not covered by
|
|
|
|
* above commands
|
|
|
|
*/
|
parser: support multiple mount conditionals in a single rule
Now that flag processing for mount rules with single option
conditionals are fixed e-enable multiple mount conditionals on a
single mount rule. The mount conditionals are equivalent to specifying
multiple rules.
mount options=(a,b,c) options=(c,d),
is the same as
mount options=(a,b,c),
mount options=(c,d),
and
mount options in (a,b,c) options in (c,d),
is the same as
mount options in (a,b,c),
mount options in (c,d),
when multiple options= and options in are combined in a single rule
it is the same as the cross product of the options.
where
mount options=(a,b,c) options in (d,e),
is a single rule.
mount options=(a,b,c) options=(d,e) options in (f),
is equivalent to
mount options=(a,b,c) options in (f),
mount options=(d,e) options in (f),
and while it is not recommended that multiple options= and options in
conditions be used in a single rule.
mount options=(a,b,c) options=(d,e) options in (f) options in (g),
is equivalent to
mount options=(a,b,c) options in (f),
mount options=(a,b,c) options in (g),
mount options=(d,e) options in (f),
mount options=(d,e) options in (g),
Bug Link: https://bugs.launchpad.net/apparmor/+bug/1597017
Signed-off-by: John Johansen <john.johansen@canonical.com>
- rebased to bba1a023bf
- fixed infinite loop in mnt_rule::gen_policy_re
Signed-off-by: Alexander Mikhalitsyn <aleksandr.mikhalitsyn@canonical.com>
Acked-by: John Johansen <john.johansen@canonical.com>
2023-03-28 12:43:18 -07:00
|
|
|
return gen_policy_new_mount(prof, count, flags, opt_flags);
|
2023-03-28 14:35:24 -07:00
|
|
|
} /* else must be RULE_OK for some rules */
|
parser: support multiple mount conditionals in a single rule
Now that flag processing for mount rules with single option
conditionals are fixed e-enable multiple mount conditionals on a
single mount rule. The mount conditionals are equivalent to specifying
multiple rules.
mount options=(a,b,c) options=(c,d),
is the same as
mount options=(a,b,c),
mount options=(c,d),
and
mount options in (a,b,c) options in (c,d),
is the same as
mount options in (a,b,c),
mount options in (c,d),
when multiple options= and options in are combined in a single rule
it is the same as the cross product of the options.
where
mount options=(a,b,c) options in (d,e),
is a single rule.
mount options=(a,b,c) options=(d,e) options in (f),
is equivalent to
mount options=(a,b,c) options in (f),
mount options=(d,e) options in (f),
and while it is not recommended that multiple options= and options in
conditions be used in a single rule.
mount options=(a,b,c) options=(d,e) options in (f) options in (g),
is equivalent to
mount options=(a,b,c) options in (f),
mount options=(a,b,c) options in (g),
mount options=(d,e) options in (f),
mount options=(d,e) options in (g),
Bug Link: https://bugs.launchpad.net/apparmor/+bug/1597017
Signed-off-by: John Johansen <john.johansen@canonical.com>
- rebased to bba1a023bf
- fixed infinite loop in mnt_rule::gen_policy_re
Signed-off-by: Alexander Mikhalitsyn <aleksandr.mikhalitsyn@canonical.com>
Acked-by: John Johansen <john.johansen@canonical.com>
2023-03-28 12:43:18 -07:00
|
|
|
|
|
|
|
return RULE_OK;
|
|
|
|
}
|
|
|
|
|
|
|
|
int mnt_rule::gen_policy_re(Profile &prof)
|
|
|
|
{
|
|
|
|
std::string mntbuf;
|
|
|
|
std::string devbuf;
|
|
|
|
std::string typebuf;
|
|
|
|
std::string optsbuf;
|
|
|
|
char class_mount_hdr[64];
|
|
|
|
const char *vec[5];
|
|
|
|
int count = 0;
|
|
|
|
|
|
|
|
if (!features_supports_mount) {
|
|
|
|
warn_once(prof.name);
|
|
|
|
return RULE_NOT_SUPPORTED;
|
|
|
|
}
|
|
|
|
|
|
|
|
sprintf(class_mount_hdr, "\\x%02x", AA_CLASS_MOUNT);
|
|
|
|
|
|
|
|
/* a single mount rule may result in multiple matching rules being
|
|
|
|
* created in the backend to cover all the possible choices
|
|
|
|
*/
|
|
|
|
for (size_t i = 0; i < flagsv.size(); i++) {
|
|
|
|
for (size_t j = 0; j < opt_flagsv.size(); j++) {
|
|
|
|
if (gen_flag_rules(prof, count, flagsv[i], opt_flagsv[j]) == RULE_ERROR)
|
|
|
|
goto fail;
|
|
|
|
}
|
2014-04-07 03:16:50 -07:00
|
|
|
}
|
2021-06-11 03:20:18 -07:00
|
|
|
if (perms & AA_MAY_UMOUNT) {
|
2014-04-07 03:16:50 -07:00
|
|
|
/* rule class single byte header */
|
|
|
|
mntbuf.assign(class_mount_hdr);
|
|
|
|
if (!convert_entry(mntbuf, mnt_point))
|
|
|
|
goto fail;
|
|
|
|
vec[0] = mntbuf.c_str();
|
2021-09-09 01:42:51 -07:00
|
|
|
if (!prof.policy.rules->add_rule_vec(rule_mode == RULE_DENY, perms,
|
2021-08-30 14:31:03 -07:00
|
|
|
(audit == AUDIT_FORCE ? perms : 0), 1, vec,
|
2021-06-11 03:07:54 -07:00
|
|
|
dfaflags, false))
|
2014-04-07 03:16:50 -07:00
|
|
|
goto fail;
|
|
|
|
count++;
|
|
|
|
}
|
2021-06-11 03:20:18 -07:00
|
|
|
if (perms & AA_MAY_PIVOTROOT) {
|
2014-04-07 03:16:50 -07:00
|
|
|
/* rule class single byte header */
|
|
|
|
mntbuf.assign(class_mount_hdr);
|
|
|
|
if (!convert_entry(mntbuf, mnt_point))
|
|
|
|
goto fail;
|
|
|
|
vec[0] = mntbuf.c_str();
|
|
|
|
if (!clear_and_convert_entry(devbuf, device))
|
|
|
|
goto fail;
|
|
|
|
vec[1] = devbuf.c_str();
|
2021-09-09 01:42:51 -07:00
|
|
|
if (!prof.policy.rules->add_rule_vec(rule_mode == RULE_DENY, perms,
|
2021-08-30 14:31:03 -07:00
|
|
|
(audit == AUDIT_FORCE ? perms : 0), 2, vec,
|
2021-06-11 03:07:54 -07:00
|
|
|
dfaflags, false))
|
2014-04-07 03:16:50 -07:00
|
|
|
goto fail;
|
|
|
|
count++;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (!count)
|
|
|
|
/* didn't actually encode anything */
|
|
|
|
goto fail;
|
|
|
|
|
|
|
|
return RULE_OK;
|
|
|
|
|
|
|
|
fail:
|
2014-12-12 08:21:31 -06:00
|
|
|
PERROR("Encoding of mount rule failed\n");
|
2014-04-07 03:16:50 -07:00
|
|
|
return RULE_ERROR;
|
Add mount rules
Add the ability to control mounting and unmounting
The basic form of the rules are.
[audit] [deny] mount [conds]* [device] [ -> [conds] path],
[audit] [deny] remount [conds]* [path],
[audit] [deny] umount [conds]* [path],
[audit] [deny] pivotroot [oldroot=<value>] <path> -> <profile>
remount is just a short cut for mount options=remount
where [conds] can be
fstype=<expr>
options=<expr>
conds follow the extended conditional syntax of allowing either:
* a single value after the equals, which has the same character range as
regular IDS (ie most anything but it can't be terminated with a , (comma)
and if spaces or other characters are needed it can be quoted
eg.
options=foo
options = foo
options="foo bar"
* a list of values after the equals, the list of values is enclosed within
parenthesis () and its has a slightly reduced character set but again
elements can be quoted.
the separation between elements is whitespace and commas.
eg.
options=(foo bar)
options=(foo, bar)
options=(foo , bar)
options=(foo,bar)
The rules are flexible and follow a similar pattern as network, capability,
etc.
mount, # allow all mounts, but not umount or pivotroot
mount fstype=procfs, # allow mounting procfs anywhere
mount options=(bind, ro) /foo -> /bar, # readonly bind mount
mount /dev/sda -> /mnt,
mount /dev/sd** -> /mnt/**,
mount fstype=overlayfs options=(rw,upperdir=/tmp/upper/,lowerdir=/) overlay -> /mnt/
umount,
umount /m*,
Currently variables and regexs are are supported on the device and mount
point. ie.
mount <devince> -> <mount point>,
Regexes are supported in fstype and options. The options have a further
caveat that regexs only work if the option is fs specific option.
eg. options=(upperdir=/tmp/*,lowerdir=/)
regex's will not currently work against the standard options like ro, rw
nosuid
Conditionals (fstype) can only be applied to the device (source) at this
time and will be disregarded in situations where the mount is manipulating
an existing mount (bind, remount).
Options can be specified multiple times
mount option=rw option=(nosuid,upperdir=/foo),
and will be combined together into a single set of values
The ordering of the standard mount options (rw,ro, ...) does not matter
but the ordering of fs specific options does.
Specifying that the value of a particular option does not matter can be
acheived by providing both the positive and negative forms of and option
option=(rw,ro) options=(suid,nosuid)
For the fs specific options specifying that a particular value does not
matter is achieve using a regex with alternations.
Improvements to the syntax and order restrictions are planned for the
future.
Signed-off-by: John Johansen <john.johansen@canonical.com>
2012-02-24 04:19:38 -08:00
|
|
|
}
|
2014-04-07 03:16:50 -07:00
|
|
|
|
2021-09-08 00:25:28 -07:00
|
|
|
void mnt_rule::post_parse_profile(Profile &prof)
|
2014-04-07 03:16:50 -07:00
|
|
|
{
|
|
|
|
if (trans) {
|
2021-06-09 00:56:59 -07:00
|
|
|
perms_t perms = 0;
|
2014-04-07 03:16:50 -07:00
|
|
|
int n = add_entry_to_x_table(&prof, trans);
|
|
|
|
if (!n) {
|
|
|
|
PERROR("Profile %s has too many specified profile transitions.\n", prof.name);
|
|
|
|
exit(1);
|
|
|
|
}
|
|
|
|
|
2021-06-11 03:20:18 -07:00
|
|
|
if (perms & AA_USER_EXEC)
|
2021-06-09 00:56:59 -07:00
|
|
|
perms |= SHIFT_PERMS(n << 10, AA_USER_SHIFT);
|
2021-06-11 03:20:18 -07:00
|
|
|
if (perms & AA_OTHER_EXEC)
|
2021-06-09 00:56:59 -07:00
|
|
|
perms |= SHIFT_PERMS(n << 10, AA_OTHER_SHIFT);
|
2021-06-11 03:20:18 -07:00
|
|
|
perms = ((perms & ~AA_ALL_EXEC_MODIFIERS) |
|
2021-06-09 00:56:59 -07:00
|
|
|
(perms & AA_ALL_EXEC_MODIFIERS));
|
2014-04-07 03:16:50 -07:00
|
|
|
|
|
|
|
trans = NULL;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
|