githubmirror/audit-userspace

docs/auditctl.8
374 строки20 KB
Steve Grubb
Update documentation for watches
24 сен 2023, 21:19
24 сен 2023, 21:19dd846b5
.TH AUDITCTL "8" "Sep 2023" "Red Hat" "System Administration Utilities"
.SH NAME
auditctl \- a utility to assist controlling the kernel's audit system
.SH SYNOPSIS
\fBauditctl\fP [\fIoptions\fP]
.SH DESCRIPTION
The \fBauditctl\fP program is used to configure kernel options related to auditing, to see status of the configuration, and to load discretionary audit rules.
.SH CONFIGURATION OPTIONS
.TP
.BI \-b\  backlog
Set max number (limit) of outstanding audit buffers allowed (Kernel Default=64) If all buffers are full, the failure flag is consulted by the kernel for action.
.TP
.BI \-\-backlog_wait_time \ \fIwait_time\fP
Set the time for the kernel to wait (Kernel Default 60*HZ) when the backlog limit is reached before queuing more audit events to be transferred to auditd. The number must be greater than or equal to zero and less than 10 times the default value.
.TP
.BI \-\-reset_backlog_wait_time_actual
Reset the actual backlog wait time counter shown by the status command.
.TP
.B \-c
Continue loading rules in spite of an error. This summarizes the results of loading the rules. The exit code will not be success if any rule fails to load.
.TP
.B \-D
Delete all rules and watches. This can take a key option (\-k), too.
.TP
\fB\-e\fP [\fB0\fP..\fB2\fP]
Set enabled flag. When \fB0\fP is passed, this can be used to temporarily disable auditing. When \fB1\fP is passed as an argument, it will enable auditing. To lock the audit configuration so that it can't be changed, pass a \fB2\fP as the argument. Locking the configuration is intended to be the last command in audit.rules for anyone wishing this feature to be active. Any attempt to change the configuration in this mode will be audited and denied. The configuration can only be changed by rebooting the machine.
.TP
\fB\-f\fP [\fB0\fP..\fB2\fP]
Set failure mode
\fB0\fP=silent \fB1\fP=printk \fB2\fP=panic. This option lets you determine how you want the kernel to handle critical errors. Example conditions where this mode may have an effect includes: transmission errors to userspace audit daemon, backlog limit exceeded, out of kernel memory, and rate limit exceeded. The default value is \fB1\fP. Secure environments will probably want to set this to \fB2\fP.
.TP
.B \-h
Help
.TP
.B \-i
When given by itself, ignore errors when reading rules from a file. This causes auditctl to always return a success exit code. If passed as an argument to
.B \-s
then it gives an interpretation of the numbers to human readable words if possible.
.TP
.BI \-\-loginuid-immutable
This option tells the kernel to make loginuids unchangeable once they are set. Changing loginuids requires CAP_AUDIT_CONTROL. So, its not something that can be done by unprivileged users. Setting this makes loginuid tamper-proof, but can cause some problems in certain kinds of containers.
.TP
.BI \-q\  mount-point,subtree
If you have an existing directory watch and bind or move mount another subtree in the watched subtree, you need to tell the kernel to make the subtree being mounted equivalent to the directory being watched. If the subtree is already mounted at the time the directory watch is issued, the subtree is automatically tagged for watching. Please note the comma separating the two values. Omitting it will cause errors.
.TP
.BI \-r\  rate
Set limit in messages/sec (\fB0\fP=none). If this \fIrate\fP is non-zero and is exceeded, the failure flag is consulted by the kernel for action. The default value is \fB0\fP.
.TP
.BI \-\-reset-lost
Reset the lost record counter shown by the status command.
.TP
.BI \-R\  file
Read and execute auditctl commands from a \fIfile\fP. The commands are executed line-by-line, in the order that they appear in the file. The file must be owned by root and not readable by other users, or else it will be rejected. Empty lines are skipped. Lines starting with the '#' character are treated as comment lines. Each line is executed as if it was provided to auditctl as command line arguments. Since auditctl is the one reading the file and not a shell such as bash, do not escape special shell characters. See the EXAMPLES section for an example.
.TP
.BI \-\-signal\  signal
Send a signal to the audit daemon. You must have privileges to do this. Supported signals are
.I TERM, HUP, USR1, USR2, CONT
 and user friendly versions
.I stop, reload, rotate, resume, state.
.TP
.BI \-t
Trim the subtrees after a mount command.
.SH STATUS OPTIONS
.TP
.B \-l
List all rules 1 per line. Two more options may be given to this command. You can give either a key option (\-k) to list rules that match a key or a (\-i) to have a0 through a3 interpreted to help determine the syscall argument values are correct .
.TP
.BI \-m\  text
Send a user space message into the audit system. This can only be done if you have CAP_AUDIT_WRITE capability (normally the root user has this). The resulting event will be the USER type.
.TP
.B \-s
Report the kernel's audit subsystem status. It will tell you the in-kernel values that can be set by \fB-e\fP, \fB-f\fP, \fB-r\fP, and \fB-b\fP options. The pid value is the process number of the audit daemon. Note that a pid of 0 indicates that the audit daemon is not running. The lost entry will tell you how many event records that have been discarded due to the kernel audit queue overflowing. The backlog field tells how many event records are currently queued waiting for auditd to read them. This option can be followed by the \fB-i\fP to get a couple fields interpreted.
.TP
.BI \-v
Print the version of auditctl.

.SH RULE OPTIONS
.TP
.BI \-a\  [ list,action | action,list ]
Append rule to the end of \fIlist\fP with \fIaction\fP. Please note the comma separating the two values. Omitting it will cause errors. The fields may be in either order. It could be list,action or action,list. The following describes the valid \fIlist\fP names:
.RS
.TP 12
.B task
Add a rule to the per task list. This rule list is used only at the time a task is created -- when fork() or clone() are called by the parent task. When using this list, you should only use fields that are known at task creation time, such as the uid, gid, etc.
.TP
.B exit
Add a rule to the syscall exit list. This list is used upon exit from a system call to determine if an audit event should be created.
.TP
.B user
Add a rule to the user message filter list. This list is used by the kernel to filter events originating in user space before relaying them to the audit daemon. It should be noted that the only fields that are valid are: uid, auid, gid, pid, subj_user, subj_role, subj_type, subj_sen, subj_clr, msgtype, and executable name. All other fields will be treated as non-matching. It should be understood that any event originating from user space from a process that has CAP_AUDIT_WRITE will be recorded into the audit trail. This means that the most likely use for this filter is with rules that have an action of never since nothing has to be done to allow events to be recorded.
.TP
.B exclude
Add a rule to the event type exclusion filter list. This list is used to filter events that you do not want to see. For example, if you do not want to see any avc messages, you would using this list to record that. Events can be excluded by process ID, user ID, group ID, login user ID, message type, subject context, or executable name. The action is ignored and uses its default of "never".
.TP
.B filesystem
Add a rule that will be applied to a whole filesystem. The filesystem must be identified with a fstype field. Normally this filter is used to exclude any events for a whole filesystem such as tracefs or debugfs.
.TP
.B io_uring
Add a rule to the io_uring syscall filter. Rules against this filter specify the syscall operation using the -S syscall notion explained below. You can add a key field to the rule so that it may be grouped with other rules watching the same underlying syscall.
.RE

The following describes the valid \fIactions\fP for the rule:
.RS
.TP 12
.B never
No audit records will be generated. This can be used to suppress event generation. In general, you want suppressions at the top of the list instead of the bottom. This is because the event triggers on the first matching rule.
.TP
.B always
Allocate an audit context, always fill it in at syscall entry time, and always write out a record at syscall exit time.
.RE
.TP
.BI \-A\  list , action
Add rule to the beginning \fIlist\fP with \fIaction\fP.
.TP
\fB\-C\fP [\fIf\fP\fB=\fP\fIf\fP | \fIf\fP\fB!=\fP\fIf\fP]
Build an inter-field comparison rule: field, operation, field. You may pass multiple comparisons on a single command line. Each one must start with \fB\-C\fP. Each inter-field equation is anded with each other as well as equations starting with \fB\-F\fP to trigger an audit record. There are 2 operators supported - equal, and not equal. Valid fields are:
.RS
.TP 12
.B auid, uid, euid, suid, fsuid, obj_uid; and gid, egid, sgid, fsgid, obj_gid
.RE

.RS
The two groups of uid and gid cannot be mixed. But any comparison within the group can be made. The obj_uid/gid fields are collected from the object of the event such as a file or directory.
.RE

.TP
.BI \-d\  list , action
Delete rule from \fIlist\fP with \fIaction\fP. The rule is deleted only if it exactly matches syscall name(s) and every field name and value.
.TP
\fB\-F\fP [\fIn\fP\fB=\fP\fIv\fP | \fIn\fP\fB!=\fP\fIv\fP | \fIn\fP\fB<\fP\fIv\fP | \fIn\fP\fB>\fP\fIv\fP | \fIn\fP\fB<=\fP\fIv\fP | \fIn\fP\fB>=\fP\fIv\fP | \fIn\fP\fB&\fP\fIv\fP | \fIn\fP\fB&=\fP\fIv\fP]
Build a rule field: name, operation, value. You may have up to 64 fields passed on a single command line. Each one must start with \fB\-F\fP. Each field equation is anded with each other (as well as equations starting with \fB\-C\fP) to trigger an audit record. There are 8 operators supported - equal, not equal, less than, greater than, less than or equal, and greater than or equal, bit mask, and bit test respectively. Bit test will "and" the values and check that they are equal, bit mask just "ands" the values. Fields that take a user ID may instead have the user's name; the program will convert the name to user ID. The same is true of group names. Valid fields are:
.RS
.TP 12
.B a0, a1, a2, a3
Respectively, the first 4 arguments to a syscall. Note that string arguments are not supported. This is because the kernel is passed a pointer to the string. Triggering on a pointer address value is not likely to work. So, when using this, you should only use on numeric values. This is most likely to be used on platforms that multiplex socket or IPC operations.
.TP
.B arch
The CPU architecture of the syscall. The arch can be found doing 'uname \-m'. If you do not know the arch of your machine but you want to use the 32 bit syscall table and your machine supports 32 bit, you can also use
.B b32
for the arch. The same applies to the 64 bit syscall table, you can use
.B b64.
In this way, you can write rules that are somewhat arch independent because the family type will be auto detected. However, syscalls can be arch specific and what is available on x86_64, may not be available on ppc. The arch directive should precede the \-S option so that auditctl knows which internal table to use to look up the syscall numbers.
.TP
.B auid
The original ID the user logged in with. Its an abbreviation of audit uid. Sometimes its referred to as loginuid. Either the user account text or number may be used.
.TP
.B devmajor
Device Major Number
.TP
.B devminor
Device Minor Number
.TP
.B dir
Full Path of Directory to watch. This will place a recursive watch on the directory and its whole subtree. It can only be used on exit list. See "\fB\-w\fP".
.TP
.B egid
Effective Group ID. May be numeric or the groups name.
.TP
.B euid
Effective User ID. May be numeric or the user account name.
.TP
.B exe
Absolute path to application that while executing this rule will apply to. It supports = and != operators. Note that you can only use this once for each rule.
.TP
.B exit
Exit value from a syscall. If the exit code is an errno, you may use the text representation, too.
.TP
.B fsgid
Filesystem Group ID. May be numeric or the groups name.
.TP
.B fstype
File system type. This is used with the filesystem rule list. The only values supported are debugfs and tracefs.
.TP
.B fsuid
Filesystem User ID. May be numeric or the user account name.
.TP
.B filetype
The target file's type. Can be either file, dir, socket, link, character, block, or fifo.
.TP
.B gid
Group ID. May be numeric or the groups name.
.TP
.B inode
Inode Number
.TP
.B key
Set a filter key on an audit rule. The filter key is an arbitrary string of text that can be up to 31 bytes long. It can uniquely identify the audit records produced by a rule. Typical use is for when you have several rules that together satisfy a security requirement. The key value can be searched on with ausearch so that no matter which rule triggered the event, you can find its results. The key can also be used on delete all (\-D) and list rules (\-l) to select rules with a specific key. You may have more than one key on a rule if you want to be able to search logged events in multiple ways or if you have an auditd plugin that uses a key to aid its analysis.
.TP
.B msgtype
This is used to match the event's record type. It should only be used on the exclude or user filter lists.
.TP
.B obj_uid
Object's UID
.TP
.B obj_gid
Object's GID
.TP
.B obj_user
Resource's SE Linux User
.TP
.B obj_role
Resource's SE Linux Role
.TP
.B obj_type
Resource's SE Linux Type
.TP
.B obj_lev_low
Resource's SE Linux Low Level
.TP
.B obj_lev_high
Resource's SE Linux High Level
.TP
.B path
Insert a watch for the file system object at \fIpath\fP. You cannot insert a watch to the top level directory. This is prohibited by the kernel. Wildcards are not supported either and will generate a warning. The way that watches work is by tracking the inode internally. This can only be used on exit list.
.TP
.B perm
Permission filter for file operations. Supply the access type that a file system watch will trigger on. \fBr\fP=read, \fBw\fP=write, \fBx\fP=execute, \fBa\fP=attribute change. These permissions are not the standard file permissions, but rather the kind of syscall that would do this kind of thing. The read & write syscalls are omitted from this set since they would overwhelm the logs. But rather for reads or writes, the open flags are looked at to see what permission was requested. The perm field can only be used on exit list. You can use this without specifying a syscall and the kernel will select the syscalls that satisfy the access permissions being requested. This also requires supplying an arch parameter before the perm field. This way the kernel can better determine what syscalls are needed. Not supplying an arch will result in
.B all
system calls being subject to audit. This will lower system performance.

.TP
.B pers
OS Personality Number
.TP
.B pid
Process ID
.TP
.B ppid
Parent's Process ID
.TP
.B saddr_fam
Address family number as found in /usr/include/bits/socket.h. For example, IPv4 would be 2 and IPv6 would be 10.
.TP
.B sessionid
User's login session ID
.TP
.B subj_user
Program's SE Linux User
.TP
.B subj_role
Program's SE Linux Role
.TP
.B subj_type
Program's SE Linux Type
.TP
.B subj_sen
Program's SE Linux Sensitivity
.TP
.B subj_clr
Program's SE Linux Clearance
.TP
.B sgid
Saved Group ID. See getresgid(2) man page.
.TP
.B success
If the exit value is >= 0 this is true/yes otherwise its false/no. When writing a rule, use a 1 for true/yes and a 0 for false/no
.TP
.B suid
Saved User ID. See getresuid(2) man page.
.TP
.B uid
User ID. May be numeric or the user account name.
.RE
.TP
.BI \-k\  key
Set a filter key on an audit rule. This is deprecated when used with watches. Convert any watches to the syscall form of rules. It is still valid for use with deleting or listing rules.
.TP
\fB\-p\fP [\fBr\fP|\fBw\fP|\fBx\fP|\fBa\fP]
Describe the permission access type that a file system watch will trigger on. This is deprecated. Convert watches to the syscall form.
.TP
\fB\-S\fP [\fISyscall name or number\fP|\fBall\fP]
Any \fIsyscall name\fP or \fInumber\fP may be used. The word '\fBall\fP' may also be used.  If the given syscall is made by a program, then start an audit record. If a field rule is given and no syscall is specified, it will default to all syscalls. You may also specify multiple syscalls in the same rule by using multiple \-S options in the same rule. Doing so improves performance since fewer rules need to be evaluated. Alternatively, you may pass a comma separated list of syscall names. If you are on a bi-arch system, like x86_64, you should be aware that auditctl simply takes the text, looks it up for the native arch (in this case b64) and sends that rule to the kernel. If there are no additional arch directives, IT WILL APPLY TO BOTH 32 & 64 BIT SYSCALLS. This can have undesirable effects since there is no guarantee that any syscall has the same number on both 32 and 64 bit interfaces. You will likely want to control this and write 2 rules, one with arch equal to b32 and one with b64 to make sure the kernel finds the events that you intend. See the arch field discussion for more info.
.TP
.BI \-w\  path
Place a watch on path. If the path is a file, it's almost the same as using the \-F path option on a syscall rule. If the watch is on a directory, it's almost the same as using the \-F dir option on a syscall rule. The \-w form of writing watches is for backwards compatibility and is
.B deprecated due to poor system performance.
Convert watches of this form to the syscall based form. The only valid options when using a watch are the \-p and \-k.
.TP
.BI \-W\  path
Remove a watch for the file system object at \fIpath\fP. The rule must match exactly. See \fB-d\fP discussion for more info.
.SH "PERFORMANCE TIPS"
Syscall rules get evaluated for each syscall for every program. If you have 10 syscall rules, every program on your system will delay during a syscall while the audit system evaluates each rule. Too many syscall rules will hurt performance. Try to combine as many as you can whenever the filter, action, key, and fields are identical. For example:

.nf
.B auditctl \-a always,exit \-F arch=b64 \-S openat \-F success=0
.fi
.nf
.B auditctl \-a always,exit \-F arch=b64 \-S truncate \-F success=0
.fi

could be re-written as one rule:

.nf
.B auditctl \-a always,exit \-F arch=b64 \-S openat \-S truncate \-F success=0
.fi

Also, try to use file system auditing wherever practical. This improves performance. For example, if you were wanting to capture all failed opens & truncates like above, but were only concerned about files in /etc and didn't care about /usr or /sbin, its possible to use this rule:

.nf
.B auditctl \-a always,exit \-F arch=b64 \-S openat,truncate \-F dir=/etc \-F success=0
.fi

This will be higher performance since the kernel will not evaluate it each and every syscall. It will be handled by the filesystem auditing code and only checked on filesystem related syscalls.
.SH "EXAMPLES"
To see all syscalls made by a specific program:

.nf
# By pid:
.B auditctl \-a always,exit \-S all \-F pid=1005
# By executable path
.B auditctl \-a always,exit \-S all \-F exe=/usr/bin/ls
.fi

To see files opened by a specific user:

.nf
.B auditctl \-a always,exit \-S openat \-F auid=510
.fi

To see unsuccessful openat calls:

.nf
.B auditctl \-a always,exit \-S openat \-F success=0
.fi

To watch a file for changes (2 ways to express):

.nf
.B auditctl \-w /etc/shadow \-p wa  # Note this slows the system
.B auditctl \-a always,exit \-F arch=b64 \-F path=/etc/shadow \-F perm=wa
.fi

To recursively watch a directory for changes (2 ways to express):

.nf
.B auditctl \-w /etc/ \-p wa  # Note this slows the system
.B auditctl \-a always,exit \-F arch=b64 \-F dir=/etc/ \-F perm=wa
.fi

To see if an admin is accessing other user's files:

.nf
.B auditctl \-a always,exit \-F dir=/home/ \-F uid=0 \-C auid!=obj_uid
.fi

This is an example rules file:

.nf
# Remove all existing rules
\-D
# Never record sudo invocations
\-A exclude,always \-F exe=/usr/bin/sudo
.fi

.SH DISABLED BY DEFAULT

On many systems auditd is configured to install an 
.B -a never,task
rule by default. This rule causes every new process to skip all audit rule processing. This is usually done to avoid a small performance overhead imposed by syscall auditing. If you want to use auditd, you need to remove that rule by deleting 10-no-audit.rules and adding 10-base-config.rules to the audit rules directory.

If you have defined audit rules that are not matching when they should, check auditctl -l to make sure there is no never,task rule there.

.SH FILES
.TP
.I /etc/audit/audit.rules /etc/audit/audit-stop.rules

.SH "SEE ALSO"
.BR audit.rules (7),
.BR ausearch(8),
.BR aureport(8),
.BR auditd (8).

.SH AUTHOR
Steve Grubb