The audit system should collect information about usage of privileged commands for all users.
These are commands with suid or sgid bits on and they are specially risky in local block
device partitions not mounted with noexec and nosuid options. Therefore, these partitions
should be first identified by the following command:
For all partitions listed by the previous command, it is necessary to search for
setuid / setgid programs using the following command:
$ sudo find PARTITION -xdev -perm /6000 -type f 2>/dev/null
For each setuid / setgid program identified by the previous command, an audit rule must be
present in the appropriate place using the following line structure:
If the auditd daemon is configured to use the augenrules program to read
audit rules during daemon startup, add the line to a file with suffix .rules in the
/etc/audit/rules.d directory, replacing the PROG_PATH part with the full path
of that setuid / setgid identified program.
If the auditd daemon is configured to use the auditctl utility instead, add
the line to the /etc/audit/audit.rules file, also replacing the PROG_PATH part
with the full path of that setuid / setgid identified program.
warning alert:
Warning
This rule checks for multiple syscalls related to privileged commands. If needed to check
specific privileged commands, other more specific rules should be considered. For example:
audit_rules_privileged_commands_su
audit_rules_privileged_commands_umount
audit_rules_privileged_commands_passwd
warning alert:
Warning
Note that OVAL check and Bash / Ansible remediation of this rule
explicitly excludes file systems mounted at /proc directory
and its subdirectories. It is a virtual file system and it doesn't
contain executable applications. At the same time, interacting with this
file system during check or remediation caused undesirable errors.
Rationale
Misuse of privileged functions, either intentionally or unintentionally by authorized users,
or by unauthorized external entities that have compromised system accounts, is a serious and
ongoing concern that can have significant adverse impacts on organizations.
Auditing the use of privileged functions is one way to detect such misuse and identify the
risk from insider and advanced persistent threats.
Privileged programs are subject to escalation-of-privilege attacks, which attempt to subvert
their normal role of providing some necessary but limited capability. As such, motivation
exists to monitor these programs for unusual activity.
ISA-62443-2-1-2009, Security for Industrial Automation and Control Systems Part 2-1: Establishing an Industrial Automation and Control Systems Security Program
- name: Gather the package facts
package_facts:
manager: auto
tags:
- CCE-91251-9
- CJIS-5.4.1.1
- NIST-800-171-3.1.7
- NIST-800-53-AC-2(4)
- NIST-800-53-AC-6(9)
- NIST-800-53-AU-12(c)
- NIST-800-53-AU-2(d)
- NIST-800-53-CM-6(a)
- PCI-DSS-Req-10.2.2
- audit_rules_privileged_commands
- configure_strategy
- low_complexity
- low_disruption
- medium_severity
- no_reboot_needed
- name: Ensure auditd Collects Information on the Use of Privileged Commands - Set
List of Mount Points Which Permits Execution of Privileged Commands
ansible.builtin.set_fact:
privileged_mount_points: '{{(ansible_facts.mounts | rejectattr(''options'', ''search'',
''noexec|nosuid'') | rejectattr(''mount'', ''match'', ''/proc($|/.*$)'') | map(attribute=''mount'')
| list ) }}'
when:
- '"audit" in ansible_facts.packages'
- ansible_virtualization_type not in ["docker", "lxc", "openvz", "podman", "container"]
tags:
- CCE-91251-9
- CJIS-5.4.1.1
- NIST-800-171-3.1.7
- NIST-800-53-AC-2(4)
- NIST-800-53-AC-6(9)
- NIST-800-53-AU-12(c)
- NIST-800-53-AU-2(d)
- NIST-800-53-CM-6(a)
- PCI-DSS-Req-10.2.2
- audit_rules_privileged_commands
- configure_strategy
- low_complexity
- low_disruption
- medium_severity
- no_reboot_needed
- name: Ensure auditd Collects Information on the Use of Privileged Commands - Search
for Privileged Commands in Eligible Mount Points
ansible.builtin.shell:
cmd: find {{ item }} -xdev -perm /6000 -type f 2>/dev/null
register: result_privileged_commands_search
changed_when: false
failed_when: false
with_items: '{{ privileged_mount_points }}'
when:
- '"audit" in ansible_facts.packages'
- ansible_virtualization_type not in ["docker", "lxc", "openvz", "podman", "container"]
tags:
- CCE-91251-9
- CJIS-5.4.1.1
- NIST-800-171-3.1.7
- NIST-800-53-AC-2(4)
- NIST-800-53-AC-6(9)
- NIST-800-53-AU-12(c)
- NIST-800-53-AU-2(d)
- NIST-800-53-CM-6(a)
- PCI-DSS-Req-10.2.2
- audit_rules_privileged_commands
- configure_strategy
- low_complexity
- low_disruption
- medium_severity
- no_reboot_needed
- name: Ensure auditd Collects Information on the Use of Privileged Commands - Set
List of Privileged Commands Found in Eligible Mount Points
ansible.builtin.set_fact:
privileged_commands: '{{ privileged_commands | default([]) + item.stdout_lines
}}'
loop: '{{ result_privileged_commands_search.results }}'
when:
- '"audit" in ansible_facts.packages'
- ansible_virtualization_type not in ["docker", "lxc", "openvz", "podman", "container"]
- item is not skipped
tags:
- CCE-91251-9
- CJIS-5.4.1.1
- NIST-800-171-3.1.7
- NIST-800-53-AC-2(4)
- NIST-800-53-AC-6(9)
- NIST-800-53-AU-12(c)
- NIST-800-53-AU-2(d)
- NIST-800-53-CM-6(a)
- PCI-DSS-Req-10.2.2
- audit_rules_privileged_commands
- configure_strategy
- low_complexity
- low_disruption
- medium_severity
- no_reboot_needed
- name: Ensure auditd Collects Information on the Use of Privileged Commands - Privileged
Commands are Present in the System
block:
- name: Ensure auditd Collects Information on the Use of Privileged Commands - Ensure
Rules for All Privileged Commands in augenrules Format
ansible.builtin.lineinfile:
path: /etc/audit/rules.d/privileged.rules
line: -a always,exit -F path={{ item }} -F perm=x -F auid>=1000 -F auid!=unset
-F key=privileged
regexp: ^.*path={{ item | regex_escape() }} .*$
create: true
with_items:
- '{{ privileged_commands }}'
- name: Ensure auditd Collects Information on the Use of Privileged Commands - Ensure
Rules for All Privileged Commands in auditctl Format
ansible.builtin.lineinfile:
path: /etc/audit/audit.rules
line: -a always,exit -F path={{ item }} -F perm=x -F auid>=1000 -F auid!=unset
-F key=privileged
regexp: ^.*path={{ item | regex_escape() }} .*$
create: true
with_items:
- '{{ privileged_commands }}'
- name: Ensure auditd Collects Information on the Use of Privileged Commands - Search
for Duplicated Rules in Other Files
ansible.builtin.find:
paths: /etc/audit/rules.d
recurse: false
contains: ^-a always,exit -F path={{ item }} .*$
patterns: '*.rules'
with_items:
- '{{ privileged_commands }}'
register: result_augenrules_files
- name: Ensure auditd Collects Information on the Use of Privileged Commands - Ensure
Rules for Privileged Commands are Defined Only in One File
ansible.builtin.lineinfile:
path: '{{ item.1.path }}'
regexp: ^-a always,exit -F path={{ item.0.item }} .*$
state: absent
with_subelements:
- '{{ result_augenrules_files.results }}'
- files
when:
- item.1.path != '/etc/audit/rules.d/privileged.rules'
when:
- '"audit" in ansible_facts.packages'
- ansible_virtualization_type not in ["docker", "lxc", "openvz", "podman", "container"]
- privileged_commands is defined
tags:
- CCE-91251-9
- CJIS-5.4.1.1
- NIST-800-171-3.1.7
- NIST-800-53-AC-2(4)
- NIST-800-53-AC-6(9)
- NIST-800-53-AU-12(c)
- NIST-800-53-AU-2(d)
- NIST-800-53-CM-6(a)
- PCI-DSS-Req-10.2.2
- audit_rules_privileged_commands
- configure_strategy
- low_complexity
- low_disruption
- medium_severity
- no_reboot_needed
Remediation - Shell Script
# Remediation is applicable only in certain platforms
if [ ! -f /.dockerenv ] && [ ! -f /run/.containerenv ] && rpm --quiet -q audit; then
ACTION_ARCH_FILTERS="-a always,exit"
AUID_FILTERS="-F auid>=1000 -F auid!=unset"
SYSCALL=""
KEY="privileged"
SYSCALL_GROUPING=""
FILTER_NODEV=$(awk '/nodev/ { print $2 }' /proc/filesystems | paste -sd,)
PARTITIONS=$(findmnt -n -l -k -it $FILTER_NODEV | grep -Pv "noexec|nosuid|/proc($|/.*$)" | awk '{ print $1 }')
for PARTITION in $PARTITIONS; do
PRIV_CMDS=$(find "${PARTITION}" -xdev -perm /6000 -type f 2>/dev/null)
for PRIV_CMD in $PRIV_CMDS; do
OTHER_FILTERS="-F path=$PRIV_CMD -F perm=x"
# Perform the remediation for both possible tools: 'auditctl' and 'augenrules'
unset syscall_a
unset syscall_grouping
unset syscall_string
unset syscall
unset file_to_edit
unset rule_to_edit
unset rule_syscalls_to_edit
unset other_string
unset auid_string
unset full_rule
# Load macro arguments into arrays
read -a syscall_a <<< $SYSCALL
read -a syscall_grouping <<< $SYSCALL_GROUPING
# Create a list of audit *.rules files that should be inspected for presence and correctness
# of a particular audit rule. The scheme is as follows:
#
# -----------------------------------------------------------------------------------------
# Tool used to load audit rules | Rule already defined | Audit rules file to inspect |
# -----------------------------------------------------------------------------------------
# auditctl | Doesn't matter | /etc/audit/audit.rules |
# -----------------------------------------------------------------------------------------
# augenrules | Yes | /etc/audit/rules.d/*.rules |
# augenrules | No | /etc/audit/rules.d/$key.rules |
# -----------------------------------------------------------------------------------------
#
files_to_inspect=()
# If audit tool is 'augenrules', then check if the audit rule is defined
# If rule is defined, add '/etc/audit/rules.d/*.rules' to the list for inspection
# If rule isn't defined yet, add '/etc/audit/rules.d/$key.rules' to the list for inspection
default_file="/etc/audit/rules.d/$KEY.rules"
# As other_filters may include paths, lets use a different delimiter for it
# The "F" script expression tells sed to print the filenames where the expressions matched
readarray -t files_to_inspect < <(sed -s -n -e "/^$ACTION_ARCH_FILTERS/!d" -e "\#$OTHER_FILTERS#!d" -e "/$AUID_FILTERS/!d" -e "F" /etc/audit/rules.d/*.rules)
# Case when particular rule isn't defined in /etc/audit/rules.d/*.rules yet
if [ ${#files_to_inspect[@]} -eq "0" ]
then
file_to_inspect="/etc/audit/rules.d/$KEY.rules"
files_to_inspect=("$file_to_inspect")
if [ ! -e "$file_to_inspect" ]
then
touch "$file_to_inspect"
chmod 0640 "$file_to_inspect"
fi
fi
# After converting to jinja, we cannot return; therefore we skip the rest of the macro if needed instead
skip=1
for audit_file in "${files_to_inspect[@]}"
do
# Filter existing $audit_file rules' definitions to select those that satisfy the rule pattern,
# i.e, collect rules that match:
# * the action, list and arch, (2-nd argument)
# * the other filters, (3-rd argument)
# * the auid filters, (4-rd argument)
readarray -t similar_rules < <(sed -e "/^$ACTION_ARCH_FILTERS/!d" -e "\#$OTHER_FILTERS#!d" -e "/$AUID_FILTERS/!d" "$audit_file")
candidate_rules=()
# Filter out rules that have more fields then required. This will remove rules more specific than the required scope
for s_rule in "${similar_rules[@]}"
do
# Strip all the options and fields we know of,
# than check if there was any field left over
extra_fields=$(sed -E -e "s/^$ACTION_ARCH_FILTERS//" -e "s#$OTHER_FILTERS##" -e "s/$AUID_FILTERS//" -e "s/((:?-S [[:alnum:],]+)+)//g" -e "s/-F key=\w+|-k \w+//"<<< "$s_rule")
grep -q -- "-F" <<< "$extra_fields" || candidate_rules+=("$s_rule")
done
if [[ ${#syscall_a[@]} -ge 1 ]]
then
# Check if the syscall we want is present in any of the similar existing rules
for rule in "${candidate_rules[@]}"
do
rule_syscalls=$(echo "$rule" | grep -o -P '(-S [\w,]+)+' | xargs)
all_syscalls_found=0
for syscall in "${syscall_a[@]}"
do
grep -q -- "\b${syscall}\b" <<< "$rule_syscalls" || {
# A syscall was not found in the candidate rule
all_syscalls_found=1
}
done
if [[ $all_syscalls_found -eq 0 ]]
then
# We found a rule with all the syscall(s) we want; skip rest of macro
skip=0
break
fi
# Check if this rule can be grouped with our target syscall and keep track of it
for syscall_g in "${syscall_grouping[@]}"
do
if grep -q -- "\b${syscall_g}\b" <<< "$rule_syscalls"
then
file_to_edit=${audit_file}
rule_to_edit=${rule}
rule_syscalls_to_edit=${rule_syscalls}
fi
done
done
else
# If there is any candidate rule, it is compliant; skip rest of macro
if [ "${#candidate_rules[@]}" -gt 0 ]
then
skip=0
fi
fi
if [ "$skip" -eq 0 ]; then
break
fi
done
if [ "$skip" -ne 0 ]; then
# We checked all rules that matched the expected resemblance pattern (action, arch & auid)
# At this point we know if we need to either append the $full_rule or group
# the syscall together with an exsiting rule
# Append the full_rule if it cannot be grouped to any other rule
if [ -z ${rule_to_edit+x} ]
then
# Build full_rule while avoid adding double spaces when other_filters is empty
if [ "${#syscall_a[@]}" -gt 0 ]
then
syscall_string=""
for syscall in "${syscall_a[@]}"
do
syscall_string+=" -S $syscall"
done
fi
other_string=$([[ $OTHER_FILTERS ]] && echo " $OTHER_FILTERS") || /bin/true
auid_string=$([[ $AUID_FILTERS ]] && echo " $AUID_FILTERS") || /bin/true
full_rule="$ACTION_ARCH_FILTERS${syscall_string}${other_string}${auid_string} -F key=$KEY" || /bin/true
echo "$full_rule" >> "$default_file"
chmod o-rwx ${default_file}
else
# Check if the syscalls are declared as a comma separated list or
# as multiple -S parameters
if grep -q -- "," <<< "${rule_syscalls_to_edit}"
then
delimiter=","
else
delimiter=" -S "
fi
new_grouped_syscalls="${rule_syscalls_to_edit}"
for syscall in "${syscall_a[@]}"
do
grep -q -- "\b${syscall}\b" <<< "${rule_syscalls_to_edit}" || {
# A syscall was not found in the candidate rule
new_grouped_syscalls+="${delimiter}${syscall}"
}
done
# Group the syscall in the rule
sed -i -e "\#${rule_to_edit}#s#${rule_syscalls_to_edit}#${new_grouped_syscalls}#" "$file_to_edit"
fi
fi
unset syscall_a
unset syscall_grouping
unset syscall_string
unset syscall
unset file_to_edit
unset rule_to_edit
unset rule_syscalls_to_edit
unset other_string
unset auid_string
unset full_rule
# Load macro arguments into arrays
read -a syscall_a <<< $SYSCALL
read -a syscall_grouping <<< $SYSCALL_GROUPING
# Create a list of audit *.rules files that should be inspected for presence and correctness
# of a particular audit rule. The scheme is as follows:
#
# -----------------------------------------------------------------------------------------
# Tool used to load audit rules | Rule already defined | Audit rules file to inspect |
# -----------------------------------------------------------------------------------------
# auditctl | Doesn't matter | /etc/audit/audit.rules |
# -----------------------------------------------------------------------------------------
# augenrules | Yes | /etc/audit/rules.d/*.rules |
# augenrules | No | /etc/audit/rules.d/$key.rules |
# -----------------------------------------------------------------------------------------
#
files_to_inspect=()
# If audit tool is 'auditctl', then add '/etc/audit/audit.rules'
# file to the list of files to be inspected
default_file="/etc/audit/audit.rules"
files_to_inspect+=('/etc/audit/audit.rules' )
# After converting to jinja, we cannot return; therefore we skip the rest of the macro if needed instead
skip=1
for audit_file in "${files_to_inspect[@]}"
do
# Filter existing $audit_file rules' definitions to select those that satisfy the rule pattern,
# i.e, collect rules that match:
# * the action, list and arch, (2-nd argument)
# * the other filters, (3-rd argument)
# * the auid filters, (4-rd argument)
readarray -t similar_rules < <(sed -e "/^$ACTION_ARCH_FILTERS/!d" -e "\#$OTHER_FILTERS#!d" -e "/$AUID_FILTERS/!d" "$audit_file")
candidate_rules=()
# Filter out rules that have more fields then required. This will remove rules more specific than the required scope
for s_rule in "${similar_rules[@]}"
do
# Strip all the options and fields we know of,
# than check if there was any field left over
extra_fields=$(sed -E -e "s/^$ACTION_ARCH_FILTERS//" -e "s#$OTHER_FILTERS##" -e "s/$AUID_FILTERS//" -e "s/((:?-S [[:alnum:],]+)+)//g" -e "s/-F key=\w+|-k \w+//"<<< "$s_rule")
grep -q -- "-F" <<< "$extra_fields" || candidate_rules+=("$s_rule")
done
if [[ ${#syscall_a[@]} -ge 1 ]]
then
# Check if the syscall we want is present in any of the similar existing rules
for rule in "${candidate_rules[@]}"
do
rule_syscalls=$(echo "$rule" | grep -o -P '(-S [\w,]+)+' | xargs)
all_syscalls_found=0
for syscall in "${syscall_a[@]}"
do
grep -q -- "\b${syscall}\b" <<< "$rule_syscalls" || {
# A syscall was not found in the candidate rule
all_syscalls_found=1
}
done
if [[ $all_syscalls_found -eq 0 ]]
then
# We found a rule with all the syscall(s) we want; skip rest of macro
skip=0
break
fi
# Check if this rule can be grouped with our target syscall and keep track of it
for syscall_g in "${syscall_grouping[@]}"
do
if grep -q -- "\b${syscall_g}\b" <<< "$rule_syscalls"
then
file_to_edit=${audit_file}
rule_to_edit=${rule}
rule_syscalls_to_edit=${rule_syscalls}
fi
done
done
else
# If there is any candidate rule, it is compliant; skip rest of macro
if [ "${#candidate_rules[@]}" -gt 0 ]
then
skip=0
fi
fi
if [ "$skip" -eq 0 ]; then
break
fi
done
if [ "$skip" -ne 0 ]; then
# We checked all rules that matched the expected resemblance pattern (action, arch & auid)
# At this point we know if we need to either append the $full_rule or group
# the syscall together with an exsiting rule
# Append the full_rule if it cannot be grouped to any other rule
if [ -z ${rule_to_edit+x} ]
then
# Build full_rule while avoid adding double spaces when other_filters is empty
if [ "${#syscall_a[@]}" -gt 0 ]
then
syscall_string=""
for syscall in "${syscall_a[@]}"
do
syscall_string+=" -S $syscall"
done
fi
other_string=$([[ $OTHER_FILTERS ]] && echo " $OTHER_FILTERS") || /bin/true
auid_string=$([[ $AUID_FILTERS ]] && echo " $AUID_FILTERS") || /bin/true
full_rule="$ACTION_ARCH_FILTERS${syscall_string}${other_string}${auid_string} -F key=$KEY" || /bin/true
echo "$full_rule" >> "$default_file"
chmod o-rwx ${default_file}
else
# Check if the syscalls are declared as a comma separated list or
# as multiple -S parameters
if grep -q -- "," <<< "${rule_syscalls_to_edit}"
then
delimiter=","
else
delimiter=" -S "
fi
new_grouped_syscalls="${rule_syscalls_to_edit}"
for syscall in "${syscall_a[@]}"
do
grep -q -- "\b${syscall}\b" <<< "${rule_syscalls_to_edit}" || {
# A syscall was not found in the candidate rule
new_grouped_syscalls+="${delimiter}${syscall}"
}
done
# Group the syscall in the rule
sed -i -e "\#${rule_to_edit}#s#${rule_syscalls_to_edit}#${new_grouped_syscalls}#" "$file_to_edit"
fi
fi
done
done
else
>&2 echo 'Remediation is not applicable, nothing was done'
fi