linux/security/landlock/audit.c

491 lines
14 KiB

landlock: Add AUDIT_LANDLOCK_ACCESS and log ptrace denials Add a new AUDIT_LANDLOCK_ACCESS record type dedicated to an access request denied by a Landlock domain. AUDIT_LANDLOCK_ACCESS indicates that something unexpected happened. For now, only denied access are logged, which means that any AUDIT_LANDLOCK_ACCESS record is always followed by a SYSCALL record with "success=no". However, log parsers should check this syscall property because this is the only sign that a request was denied. Indeed, we could have "success=yes" if Landlock would support a "permissive" mode. We could also add a new field to AUDIT_LANDLOCK_DOMAIN for this mode (see following commit). By default, the only logged access requests are those coming from the same executed program that enforced the Landlock restriction on itself. In other words, no audit record are created for a task after it called execve(2). This is required to avoid log spam because programs may only be aware of their own restrictions, but not the inherited ones. Following commits will allow to conditionally generate AUDIT_LANDLOCK_ACCESS records according to dedicated landlock_restrict_self(2)'s flags. The AUDIT_LANDLOCK_ACCESS message contains: - the "domain" ID restricting the action on an object, - the "blockers" that are missing to allow the requested access, - a set of fields identifying the related object (e.g. task identified with "opid" and "ocomm"). The blockers are implicit restrictions (e.g. ptrace), or explicit access rights (e.g. filesystem), or explicit scopes (e.g. signal). This field contains a list of at least one element, each separated with a comma. The initial blocker is "ptrace", which describe all implicit Landlock restrictions related to ptrace (e.g. deny tracing of tasks outside a sandbox). Add audit support to ptrace_access_check and ptrace_traceme hooks. For the ptrace_access_check case, we log the current/parent domain and the child task. For the ptrace_traceme case, we log the parent domain and the current/child task. Indeed, the requester and the target are the current task, but the action would be performed by the parent task. Audit event sample: type=LANDLOCK_ACCESS msg=audit(1729738800.349:44): domain=195ba459b blockers=ptrace opid=1 ocomm="systemd" type=SYSCALL msg=audit(1729738800.349:44): arch=c000003e syscall=101 success=no [...] pid=300 auid=0 A following commit adds user documentation. Add KUnit tests to check reading of domain ID relative to layer level. The quick return for non-landlocked tasks is moved from task_ptrace() to each LSM hooks. It is not useful to inline the audit_enabled check because other computation are performed by landlock_log_denial(). Use scoped guards for RCU read-side critical sections. Cc: Günther Noack <gnoack@google.com> Acked-by: Paul Moore <paul@paul-moore.com> Link: https://lore.kernel.org/r/20250320190717.2287696-10-mic@digikod.net Signed-off-by: Mickaël Salaün <mic@digikod.net>
2025-03-20 20:06:58 +01:00
landlock: Add AUDIT_LANDLOCK_ACCESS and log ptrace denials Add a new AUDIT_LANDLOCK_ACCESS record type dedicated to an access request denied by a Landlock domain. AUDIT_LANDLOCK_ACCESS indicates that something unexpected happened. For now, only denied access are logged, which means that any AUDIT_LANDLOCK_ACCESS record is always followed by a SYSCALL record with "success=no". However, log parsers should check this syscall property because this is the only sign that a request was denied. Indeed, we could have "success=yes" if Landlock would support a "permissive" mode. We could also add a new field to AUDIT_LANDLOCK_DOMAIN for this mode (see following commit). By default, the only logged access requests are those coming from the same executed program that enforced the Landlock restriction on itself. In other words, no audit record are created for a task after it called execve(2). This is required to avoid log spam because programs may only be aware of their own restrictions, but not the inherited ones. Following commits will allow to conditionally generate AUDIT_LANDLOCK_ACCESS records according to dedicated landlock_restrict_self(2)'s flags. The AUDIT_LANDLOCK_ACCESS message contains: - the "domain" ID restricting the action on an object, - the "blockers" that are missing to allow the requested access, - a set of fields identifying the related object (e.g. task identified with "opid" and "ocomm"). The blockers are implicit restrictions (e.g. ptrace), or explicit access rights (e.g. filesystem), or explicit scopes (e.g. signal). This field contains a list of at least one element, each separated with a comma. The initial blocker is "ptrace", which describe all implicit Landlock restrictions related to ptrace (e.g. deny tracing of tasks outside a sandbox). Add audit support to ptrace_access_check and ptrace_traceme hooks. For the ptrace_access_check case, we log the current/parent domain and the child task. For the ptrace_traceme case, we log the parent domain and the current/child task. Indeed, the requester and the target are the current task, but the action would be performed by the parent task. Audit event sample: type=LANDLOCK_ACCESS msg=audit(1729738800.349:44): domain=195ba459b blockers=ptrace opid=1 ocomm="systemd" type=SYSCALL msg=audit(1729738800.349:44): arch=c000003e syscall=101 success=no [...] pid=300 auid=0 A following commit adds user documentation. Add KUnit tests to check reading of domain ID relative to layer level. The quick return for non-landlocked tasks is moved from task_ptrace() to each LSM hooks. It is not useful to inline the audit_enabled check because other computation are performed by landlock_log_denial(). Use scoped guards for RCU read-side critical sections. Cc: Günther Noack <gnoack@google.com> Acked-by: Paul Moore <paul@paul-moore.com> Link: https://lore.kernel.org/r/20250320190717.2287696-10-mic@digikod.net Signed-off-by: Mickaël Salaün <mic@digikod.net>
2025-03-20 20:06:58 +01:00
landlock: Add AUDIT_LANDLOCK_DOMAIN and log domain status Asynchronously log domain information when it first denies an access. This minimize the amount of generated logs, which makes it possible to always log denials for the current execution since they should not happen. These records are identified with the new AUDIT_LANDLOCK_DOMAIN type. The AUDIT_LANDLOCK_DOMAIN message contains: - the "domain" ID which is described; - the "status" which can either be "allocated" or "deallocated"; - the "mode" which is for now only "enforcing"; - for the "allocated" status, a minimal set of properties to easily identify the task that loaded the domain's policy with landlock_restrict_self(2): "pid", "uid", executable path ("exe"), and command line ("comm"); - for the "deallocated" state, the number of "denials" accounted to this domain, which is at least 1. This requires each domain to save these task properties at creation time in the new struct landlock_details. A reference to the PID is kept for the lifetime of the domain to avoid race conditions when investigating the related task. The executable path is resolved and stored to not keep a reference to the filesystem and block related actions. All these metadata are stored for the lifetime of the related domain and should then be minimal. The required memory is not accounted to the task calling landlock_restrict_self(2) contrary to most other Landlock allocations (see related comment). The AUDIT_LANDLOCK_DOMAIN record follows the first AUDIT_LANDLOCK_ACCESS record for the same domain, which is always followed by AUDIT_SYSCALL and AUDIT_PROCTITLE. This is in line with the audit logic to first record the cause of an event, and then add context with other types of record. Audit event sample for a first denial: type=LANDLOCK_ACCESS msg=audit(1732186800.349:44): domain=195ba459b blockers=ptrace opid=1 ocomm="systemd" type=LANDLOCK_DOMAIN msg=audit(1732186800.349:44): domain=195ba459b status=allocated mode=enforcing pid=300 uid=0 exe="/root/sandboxer" comm="sandboxer" type=SYSCALL msg=audit(1732186800.349:44): arch=c000003e syscall=101 success=no [...] pid=300 auid=0 Audit event sample for a following denial: type=LANDLOCK_ACCESS msg=audit(1732186800.372:45): domain=195ba459b blockers=ptrace opid=1 ocomm="systemd" type=SYSCALL msg=audit(1732186800.372:45): arch=c000003e syscall=101 success=no [...] pid=300 auid=0 Log domain deletion with the "deallocated" state when a domain was previously logged. This makes it possible for log parsers to free potential resources when a domain ID will never show again. The number of denied access requests is useful to easily check how many access requests a domain blocked and potentially if some of them are missing in logs because of audit rate limiting, audit rules, or Landlock log configuration flags (see following commit). Audit event sample for a deletion of a domain that denied something: type=LANDLOCK_DOMAIN msg=audit(1732186800.393:46): domain=195ba459b status=deallocated denials=2 Cc: Günther Noack <gnoack@google.com> Acked-by: Paul Moore <paul@paul-moore.com> Link: https://lore.kernel.org/r/20250320190717.2287696-11-mic@digikod.net [mic: Update comment and GFP flag for landlock_log_drop_domain()] Signed-off-by: Mickaël Salaün <mic@digikod.net>
2025-03-20 20:06:59 +01:00
landlock: Add AUDIT_LANDLOCK_ACCESS and log ptrace denials Add a new AUDIT_LANDLOCK_ACCESS record type dedicated to an access request denied by a Landlock domain. AUDIT_LANDLOCK_ACCESS indicates that something unexpected happened. For now, only denied access are logged, which means that any AUDIT_LANDLOCK_ACCESS record is always followed by a SYSCALL record with "success=no". However, log parsers should check this syscall property because this is the only sign that a request was denied. Indeed, we could have "success=yes" if Landlock would support a "permissive" mode. We could also add a new field to AUDIT_LANDLOCK_DOMAIN for this mode (see following commit). By default, the only logged access requests are those coming from the same executed program that enforced the Landlock restriction on itself. In other words, no audit record are created for a task after it called execve(2). This is required to avoid log spam because programs may only be aware of their own restrictions, but not the inherited ones. Following commits will allow to conditionally generate AUDIT_LANDLOCK_ACCESS records according to dedicated landlock_restrict_self(2)'s flags. The AUDIT_LANDLOCK_ACCESS message contains: - the "domain" ID restricting the action on an object, - the "blockers" that are missing to allow the requested access, - a set of fields identifying the related object (e.g. task identified with "opid" and "ocomm"). The blockers are implicit restrictions (e.g. ptrace), or explicit access rights (e.g. filesystem), or explicit scopes (e.g. signal). This field contains a list of at least one element, each separated with a comma. The initial blocker is "ptrace", which describe all implicit Landlock restrictions related to ptrace (e.g. deny tracing of tasks outside a sandbox). Add audit support to ptrace_access_check and ptrace_traceme hooks. For the ptrace_access_check case, we log the current/parent domain and the child task. For the ptrace_traceme case, we log the parent domain and the current/child task. Indeed, the requester and the target are the current task, but the action would be performed by the parent task. Audit event sample: type=LANDLOCK_ACCESS msg=audit(1729738800.349:44): domain=195ba459b blockers=ptrace opid=1 ocomm="systemd" type=SYSCALL msg=audit(1729738800.349:44): arch=c000003e syscall=101 success=no [...] pid=300 auid=0 A following commit adds user documentation. Add KUnit tests to check reading of domain ID relative to layer level. The quick return for non-landlocked tasks is moved from task_ptrace() to each LSM hooks. It is not useful to inline the audit_enabled check because other computation are performed by landlock_log_denial(). Use scoped guards for RCU read-side critical sections. Cc: Günther Noack <gnoack@google.com> Acked-by: Paul Moore <paul@paul-moore.com> Link: https://lore.kernel.org/r/20250320190717.2287696-10-mic@digikod.net Signed-off-by: Mickaël Salaün <mic@digikod.net>
2025-03-20 20:06:58 +01:00
landlock: Add AUDIT_LANDLOCK_ACCESS and log ptrace denials Add a new AUDIT_LANDLOCK_ACCESS record type dedicated to an access request denied by a Landlock domain. AUDIT_LANDLOCK_ACCESS indicates that something unexpected happened. For now, only denied access are logged, which means that any AUDIT_LANDLOCK_ACCESS record is always followed by a SYSCALL record with "success=no". However, log parsers should check this syscall property because this is the only sign that a request was denied. Indeed, we could have "success=yes" if Landlock would support a "permissive" mode. We could also add a new field to AUDIT_LANDLOCK_DOMAIN for this mode (see following commit). By default, the only logged access requests are those coming from the same executed program that enforced the Landlock restriction on itself. In other words, no audit record are created for a task after it called execve(2). This is required to avoid log spam because programs may only be aware of their own restrictions, but not the inherited ones. Following commits will allow to conditionally generate AUDIT_LANDLOCK_ACCESS records according to dedicated landlock_restrict_self(2)'s flags. The AUDIT_LANDLOCK_ACCESS message contains: - the "domain" ID restricting the action on an object, - the "blockers" that are missing to allow the requested access, - a set of fields identifying the related object (e.g. task identified with "opid" and "ocomm"). The blockers are implicit restrictions (e.g. ptrace), or explicit access rights (e.g. filesystem), or explicit scopes (e.g. signal). This field contains a list of at least one element, each separated with a comma. The initial blocker is "ptrace", which describe all implicit Landlock restrictions related to ptrace (e.g. deny tracing of tasks outside a sandbox). Add audit support to ptrace_access_check and ptrace_traceme hooks. For the ptrace_access_check case, we log the current/parent domain and the child task. For the ptrace_traceme case, we log the parent domain and the current/child task. Indeed, the requester and the target are the current task, but the action would be performed by the parent task. Audit event sample: type=LANDLOCK_ACCESS msg=audit(1729738800.349:44): domain=195ba459b blockers=ptrace opid=1 ocomm="systemd" type=SYSCALL msg=audit(1729738800.349:44): arch=c000003e syscall=101 success=no [...] pid=300 auid=0 A following commit adds user documentation. Add KUnit tests to check reading of domain ID relative to layer level. The quick return for non-landlocked tasks is moved from task_ptrace() to each LSM hooks. It is not useful to inline the audit_enabled check because other computation are performed by landlock_log_denial(). Use scoped guards for RCU read-side critical sections. Cc: Günther Noack <gnoack@google.com> Acked-by: Paul Moore <paul@paul-moore.com> Link: https://lore.kernel.org/r/20250320190717.2287696-10-mic@digikod.net Signed-off-by: Mickaël Salaün <mic@digikod.net>
2025-03-20 20:06:58 +01:00
landlock: Add AUDIT_LANDLOCK_ACCESS and log ptrace denials Add a new AUDIT_LANDLOCK_ACCESS record type dedicated to an access request denied by a Landlock domain. AUDIT_LANDLOCK_ACCESS indicates that something unexpected happened. For now, only denied access are logged, which means that any AUDIT_LANDLOCK_ACCESS record is always followed by a SYSCALL record with "success=no". However, log parsers should check this syscall property because this is the only sign that a request was denied. Indeed, we could have "success=yes" if Landlock would support a "permissive" mode. We could also add a new field to AUDIT_LANDLOCK_DOMAIN for this mode (see following commit). By default, the only logged access requests are those coming from the same executed program that enforced the Landlock restriction on itself. In other words, no audit record are created for a task after it called execve(2). This is required to avoid log spam because programs may only be aware of their own restrictions, but not the inherited ones. Following commits will allow to conditionally generate AUDIT_LANDLOCK_ACCESS records according to dedicated landlock_restrict_self(2)'s flags. The AUDIT_LANDLOCK_ACCESS message contains: - the "domain" ID restricting the action on an object, - the "blockers" that are missing to allow the requested access, - a set of fields identifying the related object (e.g. task identified with "opid" and "ocomm"). The blockers are implicit restrictions (e.g. ptrace), or explicit access rights (e.g. filesystem), or explicit scopes (e.g. signal). This field contains a list of at least one element, each separated with a comma. The initial blocker is "ptrace", which describe all implicit Landlock restrictions related to ptrace (e.g. deny tracing of tasks outside a sandbox). Add audit support to ptrace_access_check and ptrace_traceme hooks. For the ptrace_access_check case, we log the current/parent domain and the child task. For the ptrace_traceme case, we log the parent domain and the current/child task. Indeed, the requester and the target are the current task, but the action would be performed by the parent task. Audit event sample: type=LANDLOCK_ACCESS msg=audit(1729738800.349:44): domain=195ba459b blockers=ptrace opid=1 ocomm="systemd" type=SYSCALL msg=audit(1729738800.349:44): arch=c000003e syscall=101 success=no [...] pid=300 auid=0 A following commit adds user documentation. Add KUnit tests to check reading of domain ID relative to layer level. The quick return for non-landlocked tasks is moved from task_ptrace() to each LSM hooks. It is not useful to inline the audit_enabled check because other computation are performed by landlock_log_denial(). Use scoped guards for RCU read-side critical sections. Cc: Günther Noack <gnoack@google.com> Acked-by: Paul Moore <paul@paul-moore.com> Link: https://lore.kernel.org/r/20250320190717.2287696-10-mic@digikod.net Signed-off-by: Mickaël Salaün <mic@digikod.net>
2025-03-20 20:06:58 +01:00
landlock: Add AUDIT_LANDLOCK_ACCESS and log ptrace denials Add a new AUDIT_LANDLOCK_ACCESS record type dedicated to an access request denied by a Landlock domain. AUDIT_LANDLOCK_ACCESS indicates that something unexpected happened. For now, only denied access are logged, which means that any AUDIT_LANDLOCK_ACCESS record is always followed by a SYSCALL record with "success=no". However, log parsers should check this syscall property because this is the only sign that a request was denied. Indeed, we could have "success=yes" if Landlock would support a "permissive" mode. We could also add a new field to AUDIT_LANDLOCK_DOMAIN for this mode (see following commit). By default, the only logged access requests are those coming from the same executed program that enforced the Landlock restriction on itself. In other words, no audit record are created for a task after it called execve(2). This is required to avoid log spam because programs may only be aware of their own restrictions, but not the inherited ones. Following commits will allow to conditionally generate AUDIT_LANDLOCK_ACCESS records according to dedicated landlock_restrict_self(2)'s flags. The AUDIT_LANDLOCK_ACCESS message contains: - the "domain" ID restricting the action on an object, - the "blockers" that are missing to allow the requested access, - a set of fields identifying the related object (e.g. task identified with "opid" and "ocomm"). The blockers are implicit restrictions (e.g. ptrace), or explicit access rights (e.g. filesystem), or explicit scopes (e.g. signal). This field contains a list of at least one element, each separated with a comma. The initial blocker is "ptrace", which describe all implicit Landlock restrictions related to ptrace (e.g. deny tracing of tasks outside a sandbox). Add audit support to ptrace_access_check and ptrace_traceme hooks. For the ptrace_access_check case, we log the current/parent domain and the child task. For the ptrace_traceme case, we log the parent domain and the current/child task. Indeed, the requester and the target are the current task, but the action would be performed by the parent task. Audit event sample: type=LANDLOCK_ACCESS msg=audit(1729738800.349:44): domain=195ba459b blockers=ptrace opid=1 ocomm="systemd" type=SYSCALL msg=audit(1729738800.349:44): arch=c000003e syscall=101 success=no [...] pid=300 auid=0 A following commit adds user documentation. Add KUnit tests to check reading of domain ID relative to layer level. The quick return for non-landlocked tasks is moved from task_ptrace() to each LSM hooks. It is not useful to inline the audit_enabled check because other computation are performed by landlock_log_denial(). Use scoped guards for RCU read-side critical sections. Cc: Günther Noack <gnoack@google.com> Acked-by: Paul Moore <paul@paul-moore.com> Link: https://lore.kernel.org/r/20250320190717.2287696-10-mic@digikod.net Signed-off-by: Mickaël Salaün <mic@digikod.net>
2025-03-20 20:06:58 +01:00
landlock: Add AUDIT_LANDLOCK_ACCESS and log ptrace denials Add a new AUDIT_LANDLOCK_ACCESS record type dedicated to an access request denied by a Landlock domain. AUDIT_LANDLOCK_ACCESS indicates that something unexpected happened. For now, only denied access are logged, which means that any AUDIT_LANDLOCK_ACCESS record is always followed by a SYSCALL record with "success=no". However, log parsers should check this syscall property because this is the only sign that a request was denied. Indeed, we could have "success=yes" if Landlock would support a "permissive" mode. We could also add a new field to AUDIT_LANDLOCK_DOMAIN for this mode (see following commit). By default, the only logged access requests are those coming from the same executed program that enforced the Landlock restriction on itself. In other words, no audit record are created for a task after it called execve(2). This is required to avoid log spam because programs may only be aware of their own restrictions, but not the inherited ones. Following commits will allow to conditionally generate AUDIT_LANDLOCK_ACCESS records according to dedicated landlock_restrict_self(2)'s flags. The AUDIT_LANDLOCK_ACCESS message contains: - the "domain" ID restricting the action on an object, - the "blockers" that are missing to allow the requested access, - a set of fields identifying the related object (e.g. task identified with "opid" and "ocomm"). The blockers are implicit restrictions (e.g. ptrace), or explicit access rights (e.g. filesystem), or explicit scopes (e.g. signal). This field contains a list of at least one element, each separated with a comma. The initial blocker is "ptrace", which describe all implicit Landlock restrictions related to ptrace (e.g. deny tracing of tasks outside a sandbox). Add audit support to ptrace_access_check and ptrace_traceme hooks. For the ptrace_access_check case, we log the current/parent domain and the child task. For the ptrace_traceme case, we log the parent domain and the current/child task. Indeed, the requester and the target are the current task, but the action would be performed by the parent task. Audit event sample: type=LANDLOCK_ACCESS msg=audit(1729738800.349:44): domain=195ba459b blockers=ptrace opid=1 ocomm="systemd" type=SYSCALL msg=audit(1729738800.349:44): arch=c000003e syscall=101 success=no [...] pid=300 auid=0 A following commit adds user documentation. Add KUnit tests to check reading of domain ID relative to layer level. The quick return for non-landlocked tasks is moved from task_ptrace() to each LSM hooks. It is not useful to inline the audit_enabled check because other computation are performed by landlock_log_denial(). Use scoped guards for RCU read-side critical sections. Cc: Günther Noack <gnoack@google.com> Acked-by: Paul Moore <paul@paul-moore.com> Link: https://lore.kernel.org/r/20250320190717.2287696-10-mic@digikod.net Signed-off-by: Mickaël Salaün <mic@digikod.net>
2025-03-20 20:06:58 +01:00
landlock: Add AUDIT_LANDLOCK_ACCESS and log ptrace denials Add a new AUDIT_LANDLOCK_ACCESS record type dedicated to an access request denied by a Landlock domain. AUDIT_LANDLOCK_ACCESS indicates that something unexpected happened. For now, only denied access are logged, which means that any AUDIT_LANDLOCK_ACCESS record is always followed by a SYSCALL record with "success=no". However, log parsers should check this syscall property because this is the only sign that a request was denied. Indeed, we could have "success=yes" if Landlock would support a "permissive" mode. We could also add a new field to AUDIT_LANDLOCK_DOMAIN for this mode (see following commit). By default, the only logged access requests are those coming from the same executed program that enforced the Landlock restriction on itself. In other words, no audit record are created for a task after it called execve(2). This is required to avoid log spam because programs may only be aware of their own restrictions, but not the inherited ones. Following commits will allow to conditionally generate AUDIT_LANDLOCK_ACCESS records according to dedicated landlock_restrict_self(2)'s flags. The AUDIT_LANDLOCK_ACCESS message contains: - the "domain" ID restricting the action on an object, - the "blockers" that are missing to allow the requested access, - a set of fields identifying the related object (e.g. task identified with "opid" and "ocomm"). The blockers are implicit restrictions (e.g. ptrace), or explicit access rights (e.g. filesystem), or explicit scopes (e.g. signal). This field contains a list of at least one element, each separated with a comma. The initial blocker is "ptrace", which describe all implicit Landlock restrictions related to ptrace (e.g. deny tracing of tasks outside a sandbox). Add audit support to ptrace_access_check and ptrace_traceme hooks. For the ptrace_access_check case, we log the current/parent domain and the child task. For the ptrace_traceme case, we log the parent domain and the current/child task. Indeed, the requester and the target are the current task, but the action would be performed by the parent task. Audit event sample: type=LANDLOCK_ACCESS msg=audit(1729738800.349:44): domain=195ba459b blockers=ptrace opid=1 ocomm="systemd" type=SYSCALL msg=audit(1729738800.349:44): arch=c000003e syscall=101 success=no [...] pid=300 auid=0 A following commit adds user documentation. Add KUnit tests to check reading of domain ID relative to layer level. The quick return for non-landlocked tasks is moved from task_ptrace() to each LSM hooks. It is not useful to inline the audit_enabled check because other computation are performed by landlock_log_denial(). Use scoped guards for RCU read-side critical sections. Cc: Günther Noack <gnoack@google.com> Acked-by: Paul Moore <paul@paul-moore.com> Link: https://lore.kernel.org/r/20250320190717.2287696-10-mic@digikod.net Signed-off-by: Mickaël Salaün <mic@digikod.net>
2025-03-20 20:06:58 +01:00
landlock: Add AUDIT_LANDLOCK_ACCESS and log ptrace denials Add a new AUDIT_LANDLOCK_ACCESS record type dedicated to an access request denied by a Landlock domain. AUDIT_LANDLOCK_ACCESS indicates that something unexpected happened. For now, only denied access are logged, which means that any AUDIT_LANDLOCK_ACCESS record is always followed by a SYSCALL record with "success=no". However, log parsers should check this syscall property because this is the only sign that a request was denied. Indeed, we could have "success=yes" if Landlock would support a "permissive" mode. We could also add a new field to AUDIT_LANDLOCK_DOMAIN for this mode (see following commit). By default, the only logged access requests are those coming from the same executed program that enforced the Landlock restriction on itself. In other words, no audit record are created for a task after it called execve(2). This is required to avoid log spam because programs may only be aware of their own restrictions, but not the inherited ones. Following commits will allow to conditionally generate AUDIT_LANDLOCK_ACCESS records according to dedicated landlock_restrict_self(2)'s flags. The AUDIT_LANDLOCK_ACCESS message contains: - the "domain" ID restricting the action on an object, - the "blockers" that are missing to allow the requested access, - a set of fields identifying the related object (e.g. task identified with "opid" and "ocomm"). The blockers are implicit restrictions (e.g. ptrace), or explicit access rights (e.g. filesystem), or explicit scopes (e.g. signal). This field contains a list of at least one element, each separated with a comma. The initial blocker is "ptrace", which describe all implicit Landlock restrictions related to ptrace (e.g. deny tracing of tasks outside a sandbox). Add audit support to ptrace_access_check and ptrace_traceme hooks. For the ptrace_access_check case, we log the current/parent domain and the child task. For the ptrace_traceme case, we log the parent domain and the current/child task. Indeed, the requester and the target are the current task, but the action would be performed by the parent task. Audit event sample: type=LANDLOCK_ACCESS msg=audit(1729738800.349:44): domain=195ba459b blockers=ptrace opid=1 ocomm="systemd" type=SYSCALL msg=audit(1729738800.349:44): arch=c000003e syscall=101 success=no [...] pid=300 auid=0 A following commit adds user documentation. Add KUnit tests to check reading of domain ID relative to layer level. The quick return for non-landlocked tasks is moved from task_ptrace() to each LSM hooks. It is not useful to inline the audit_enabled check because other computation are performed by landlock_log_denial(). Use scoped guards for RCU read-side critical sections. Cc: Günther Noack <gnoack@google.com> Acked-by: Paul Moore <paul@paul-moore.com> Link: https://lore.kernel.org/r/20250320190717.2287696-10-mic@digikod.net Signed-off-by: Mickaël Salaün <mic@digikod.net>
2025-03-20 20:06:58 +01:00
landlock: Add AUDIT_LANDLOCK_ACCESS and log ptrace denials Add a new AUDIT_LANDLOCK_ACCESS record type dedicated to an access request denied by a Landlock domain. AUDIT_LANDLOCK_ACCESS indicates that something unexpected happened. For now, only denied access are logged, which means that any AUDIT_LANDLOCK_ACCESS record is always followed by a SYSCALL record with "success=no". However, log parsers should check this syscall property because this is the only sign that a request was denied. Indeed, we could have "success=yes" if Landlock would support a "permissive" mode. We could also add a new field to AUDIT_LANDLOCK_DOMAIN for this mode (see following commit). By default, the only logged access requests are those coming from the same executed program that enforced the Landlock restriction on itself. In other words, no audit record are created for a task after it called execve(2). This is required to avoid log spam because programs may only be aware of their own restrictions, but not the inherited ones. Following commits will allow to conditionally generate AUDIT_LANDLOCK_ACCESS records according to dedicated landlock_restrict_self(2)'s flags. The AUDIT_LANDLOCK_ACCESS message contains: - the "domain" ID restricting the action on an object, - the "blockers" that are missing to allow the requested access, - a set of fields identifying the related object (e.g. task identified with "opid" and "ocomm"). The blockers are implicit restrictions (e.g. ptrace), or explicit access rights (e.g. filesystem), or explicit scopes (e.g. signal). This field contains a list of at least one element, each separated with a comma. The initial blocker is "ptrace", which describe all implicit Landlock restrictions related to ptrace (e.g. deny tracing of tasks outside a sandbox). Add audit support to ptrace_access_check and ptrace_traceme hooks. For the ptrace_access_check case, we log the current/parent domain and the child task. For the ptrace_traceme case, we log the parent domain and the current/child task. Indeed, the requester and the target are the current task, but the action would be performed by the parent task. Audit event sample: type=LANDLOCK_ACCESS msg=audit(1729738800.349:44): domain=195ba459b blockers=ptrace opid=1 ocomm="systemd" type=SYSCALL msg=audit(1729738800.349:44): arch=c000003e syscall=101 success=no [...] pid=300 auid=0 A following commit adds user documentation. Add KUnit tests to check reading of domain ID relative to layer level. The quick return for non-landlocked tasks is moved from task_ptrace() to each LSM hooks. It is not useful to inline the audit_enabled check because other computation are performed by landlock_log_denial(). Use scoped guards for RCU read-side critical sections. Cc: Günther Noack <gnoack@google.com> Acked-by: Paul Moore <paul@paul-moore.com> Link: https://lore.kernel.org/r/20250320190717.2287696-10-mic@digikod.net Signed-off-by: Mickaël Salaün <mic@digikod.net>
2025-03-20 20:06:58 +01:00
landlock: Add AUDIT_LANDLOCK_DOMAIN and log domain status Asynchronously log domain information when it first denies an access. This minimize the amount of generated logs, which makes it possible to always log denials for the current execution since they should not happen. These records are identified with the new AUDIT_LANDLOCK_DOMAIN type. The AUDIT_LANDLOCK_DOMAIN message contains: - the "domain" ID which is described; - the "status" which can either be "allocated" or "deallocated"; - the "mode" which is for now only "enforcing"; - for the "allocated" status, a minimal set of properties to easily identify the task that loaded the domain's policy with landlock_restrict_self(2): "pid", "uid", executable path ("exe"), and command line ("comm"); - for the "deallocated" state, the number of "denials" accounted to this domain, which is at least 1. This requires each domain to save these task properties at creation time in the new struct landlock_details. A reference to the PID is kept for the lifetime of the domain to avoid race conditions when investigating the related task. The executable path is resolved and stored to not keep a reference to the filesystem and block related actions. All these metadata are stored for the lifetime of the related domain and should then be minimal. The required memory is not accounted to the task calling landlock_restrict_self(2) contrary to most other Landlock allocations (see related comment). The AUDIT_LANDLOCK_DOMAIN record follows the first AUDIT_LANDLOCK_ACCESS record for the same domain, which is always followed by AUDIT_SYSCALL and AUDIT_PROCTITLE. This is in line with the audit logic to first record the cause of an event, and then add context with other types of record. Audit event sample for a first denial: type=LANDLOCK_ACCESS msg=audit(1732186800.349:44): domain=195ba459b blockers=ptrace opid=1 ocomm="systemd" type=LANDLOCK_DOMAIN msg=audit(1732186800.349:44): domain=195ba459b status=allocated mode=enforcing pid=300 uid=0 exe="/root/sandboxer" comm="sandboxer" type=SYSCALL msg=audit(1732186800.349:44): arch=c000003e syscall=101 success=no [...] pid=300 auid=0 Audit event sample for a following denial: type=LANDLOCK_ACCESS msg=audit(1732186800.372:45): domain=195ba459b blockers=ptrace opid=1 ocomm="systemd" type=SYSCALL msg=audit(1732186800.372:45): arch=c000003e syscall=101 success=no [...] pid=300 auid=0 Log domain deletion with the "deallocated" state when a domain was previously logged. This makes it possible for log parsers to free potential resources when a domain ID will never show again. The number of denied access requests is useful to easily check how many access requests a domain blocked and potentially if some of them are missing in logs because of audit rate limiting, audit rules, or Landlock log configuration flags (see following commit). Audit event sample for a deletion of a domain that denied something: type=LANDLOCK_DOMAIN msg=audit(1732186800.393:46): domain=195ba459b status=deallocated denials=2 Cc: Günther Noack <gnoack@google.com> Acked-by: Paul Moore <paul@paul-moore.com> Link: https://lore.kernel.org/r/20250320190717.2287696-11-mic@digikod.net [mic: Update comment and GFP flag for landlock_log_drop_domain()] Signed-off-by: Mickaël Salaün <mic@digikod.net>
2025-03-20 20:06:59 +01:00
landlock: Add AUDIT_LANDLOCK_ACCESS and log ptrace denials Add a new AUDIT_LANDLOCK_ACCESS record type dedicated to an access request denied by a Landlock domain. AUDIT_LANDLOCK_ACCESS indicates that something unexpected happened. For now, only denied access are logged, which means that any AUDIT_LANDLOCK_ACCESS record is always followed by a SYSCALL record with "success=no". However, log parsers should check this syscall property because this is the only sign that a request was denied. Indeed, we could have "success=yes" if Landlock would support a "permissive" mode. We could also add a new field to AUDIT_LANDLOCK_DOMAIN for this mode (see following commit). By default, the only logged access requests are those coming from the same executed program that enforced the Landlock restriction on itself. In other words, no audit record are created for a task after it called execve(2). This is required to avoid log spam because programs may only be aware of their own restrictions, but not the inherited ones. Following commits will allow to conditionally generate AUDIT_LANDLOCK_ACCESS records according to dedicated landlock_restrict_self(2)'s flags. The AUDIT_LANDLOCK_ACCESS message contains: - the "domain" ID restricting the action on an object, - the "blockers" that are missing to allow the requested access, - a set of fields identifying the related object (e.g. task identified with "opid" and "ocomm"). The blockers are implicit restrictions (e.g. ptrace), or explicit access rights (e.g. filesystem), or explicit scopes (e.g. signal). This field contains a list of at least one element, each separated with a comma. The initial blocker is "ptrace", which describe all implicit Landlock restrictions related to ptrace (e.g. deny tracing of tasks outside a sandbox). Add audit support to ptrace_access_check and ptrace_traceme hooks. For the ptrace_access_check case, we log the current/parent domain and the child task. For the ptrace_traceme case, we log the parent domain and the current/child task. Indeed, the requester and the target are the current task, but the action would be performed by the parent task. Audit event sample: type=LANDLOCK_ACCESS msg=audit(1729738800.349:44): domain=195ba459b blockers=ptrace opid=1 ocomm="systemd" type=SYSCALL msg=audit(1729738800.349:44): arch=c000003e syscall=101 success=no [...] pid=300 auid=0 A following commit adds user documentation. Add KUnit tests to check reading of domain ID relative to layer level. The quick return for non-landlocked tasks is moved from task_ptrace() to each LSM hooks. It is not useful to inline the audit_enabled check because other computation are performed by landlock_log_denial(). Use scoped guards for RCU read-side critical sections. Cc: Günther Noack <gnoack@google.com> Acked-by: Paul Moore <paul@paul-moore.com> Link: https://lore.kernel.org/r/20250320190717.2287696-10-mic@digikod.net Signed-off-by: Mickaël Salaün <mic@digikod.net>
2025-03-20 20:06:58 +01:00
landlock: Add AUDIT_LANDLOCK_ACCESS and log ptrace denials Add a new AUDIT_LANDLOCK_ACCESS record type dedicated to an access request denied by a Landlock domain. AUDIT_LANDLOCK_ACCESS indicates that something unexpected happened. For now, only denied access are logged, which means that any AUDIT_LANDLOCK_ACCESS record is always followed by a SYSCALL record with "success=no". However, log parsers should check this syscall property because this is the only sign that a request was denied. Indeed, we could have "success=yes" if Landlock would support a "permissive" mode. We could also add a new field to AUDIT_LANDLOCK_DOMAIN for this mode (see following commit). By default, the only logged access requests are those coming from the same executed program that enforced the Landlock restriction on itself. In other words, no audit record are created for a task after it called execve(2). This is required to avoid log spam because programs may only be aware of their own restrictions, but not the inherited ones. Following commits will allow to conditionally generate AUDIT_LANDLOCK_ACCESS records according to dedicated landlock_restrict_self(2)'s flags. The AUDIT_LANDLOCK_ACCESS message contains: - the "domain" ID restricting the action on an object, - the "blockers" that are missing to allow the requested access, - a set of fields identifying the related object (e.g. task identified with "opid" and "ocomm"). The blockers are implicit restrictions (e.g. ptrace), or explicit access rights (e.g. filesystem), or explicit scopes (e.g. signal). This field contains a list of at least one element, each separated with a comma. The initial blocker is "ptrace", which describe all implicit Landlock restrictions related to ptrace (e.g. deny tracing of tasks outside a sandbox). Add audit support to ptrace_access_check and ptrace_traceme hooks. For the ptrace_access_check case, we log the current/parent domain and the child task. For the ptrace_traceme case, we log the parent domain and the current/child task. Indeed, the requester and the target are the current task, but the action would be performed by the parent task. Audit event sample: type=LANDLOCK_ACCESS msg=audit(1729738800.349:44): domain=195ba459b blockers=ptrace opid=1 ocomm="systemd" type=SYSCALL msg=audit(1729738800.349:44): arch=c000003e syscall=101 success=no [...] pid=300 auid=0 A following commit adds user documentation. Add KUnit tests to check reading of domain ID relative to layer level. The quick return for non-landlocked tasks is moved from task_ptrace() to each LSM hooks. It is not useful to inline the audit_enabled check because other computation are performed by landlock_log_denial(). Use scoped guards for RCU read-side critical sections. Cc: Günther Noack <gnoack@google.com> Acked-by: Paul Moore <paul@paul-moore.com> Link: https://lore.kernel.org/r/20250320190717.2287696-10-mic@digikod.net Signed-off-by: Mickaël Salaün <mic@digikod.net>
2025-03-20 20:06:58 +01:00
landlock: Transpose the layer masks data structure The layer masks data structure tracks the requested but unfulfilled access rights during an operation's security check. It stores one bit for each combination of access right and layer index. If the bit is set, that access right is not granted (yet) in the given layer and we have to traverse the path further upwards to grant it. Previously, the layer masks were stored as arrays mapping from access right indices to layer_mask_t. The layer_mask_t value then indicates all layers in which the given access right is still (tentatively) denied. This patch introduces struct layer_access_masks instead: This struct contains an array with the access_mask_t of each (tentatively) denied access right in that layer. The hypothesis of this patch is that this simplifies the code enough so that the resulting code will run faster: * We can use bitwise operations in multiple places where we previously looped over bits individually with macros. (Should require less branch speculation and lends itself to better loop unrolling.) * Code is ~75 lines smaller. Other noteworthy changes: * In no_more_access(), call a new helper function may_refer(), which only solves the asymmetric case. Previously, the code interleaved the checks for the two symmetric cases in RENAME_EXCHANGE. It feels that the code is clearer when renames without RENAME_EXCHANGE are more obviously the normal case. Tradeoffs: This change improves performance, at a slight size increase to the layer masks data structure. This fixes the size of the data structure at 32 bytes for all types of access rights. (64, once we introduce a 17th filesystem access right). For filesystem access rights, at the moment, the data structure has the same size as before, but once we introduce the 17th filesystem access right, it will double in size (from 32 to 64 bytes), as access_mask_t grows from 16 to 32 bit [1]. Link: https://lore.kernel.org/all/20260120.haeCh4li9Vae@digikod.net/ [1] Signed-off-by: Günther Noack <gnoack3000@gmail.com> Link: https://lore.kernel.org/r/20260206151154.97915-5-gnoack3000@gmail.com [mic: Cosmetic fixes, moved struct layer_access_masks definition] Signed-off-by: Mickaël Salaün <mic@digikod.net>
2026-02-06 16:11:55 +01:00
landlock: Transpose the layer masks data structure The layer masks data structure tracks the requested but unfulfilled access rights during an operation's security check. It stores one bit for each combination of access right and layer index. If the bit is set, that access right is not granted (yet) in the given layer and we have to traverse the path further upwards to grant it. Previously, the layer masks were stored as arrays mapping from access right indices to layer_mask_t. The layer_mask_t value then indicates all layers in which the given access right is still (tentatively) denied. This patch introduces struct layer_access_masks instead: This struct contains an array with the access_mask_t of each (tentatively) denied access right in that layer. The hypothesis of this patch is that this simplifies the code enough so that the resulting code will run faster: * We can use bitwise operations in multiple places where we previously looped over bits individually with macros. (Should require less branch speculation and lends itself to better loop unrolling.) * Code is ~75 lines smaller. Other noteworthy changes: * In no_more_access(), call a new helper function may_refer(), which only solves the asymmetric case. Previously, the code interleaved the checks for the two symmetric cases in RENAME_EXCHANGE. It feels that the code is clearer when renames without RENAME_EXCHANGE are more obviously the normal case. Tradeoffs: This change improves performance, at a slight size increase to the layer masks data structure. This fixes the size of the data structure at 32 bytes for all types of access rights. (64, once we introduce a 17th filesystem access right). For filesystem access rights, at the moment, the data structure has the same size as before, but once we introduce the 17th filesystem access right, it will double in size (from 32 to 64 bytes), as access_mask_t grows from 16 to 32 bit [1]. Link: https://lore.kernel.org/all/20260120.haeCh4li9Vae@digikod.net/ [1] Signed-off-by: Günther Noack <gnoack3000@gmail.com> Link: https://lore.kernel.org/r/20260206151154.97915-5-gnoack3000@gmail.com [mic: Cosmetic fixes, moved struct layer_access_masks definition] Signed-off-by: Mickaël Salaün <mic@digikod.net>
2026-02-06 16:11:55 +01:00
landlock: Transpose the layer masks data structure The layer masks data structure tracks the requested but unfulfilled access rights during an operation's security check. It stores one bit for each combination of access right and layer index. If the bit is set, that access right is not granted (yet) in the given layer and we have to traverse the path further upwards to grant it. Previously, the layer masks were stored as arrays mapping from access right indices to layer_mask_t. The layer_mask_t value then indicates all layers in which the given access right is still (tentatively) denied. This patch introduces struct layer_access_masks instead: This struct contains an array with the access_mask_t of each (tentatively) denied access right in that layer. The hypothesis of this patch is that this simplifies the code enough so that the resulting code will run faster: * We can use bitwise operations in multiple places where we previously looped over bits individually with macros. (Should require less branch speculation and lends itself to better loop unrolling.) * Code is ~75 lines smaller. Other noteworthy changes: * In no_more_access(), call a new helper function may_refer(), which only solves the asymmetric case. Previously, the code interleaved the checks for the two symmetric cases in RENAME_EXCHANGE. It feels that the code is clearer when renames without RENAME_EXCHANGE are more obviously the normal case. Tradeoffs: This change improves performance, at a slight size increase to the layer masks data structure. This fixes the size of the data structure at 32 bytes for all types of access rights. (64, once we introduce a 17th filesystem access right). For filesystem access rights, at the moment, the data structure has the same size as before, but once we introduce the 17th filesystem access right, it will double in size (from 32 to 64 bytes), as access_mask_t grows from 16 to 32 bit [1]. Link: https://lore.kernel.org/all/20260120.haeCh4li9Vae@digikod.net/ [1] Signed-off-by: Günther Noack <gnoack3000@gmail.com> Link: https://lore.kernel.org/r/20260206151154.97915-5-gnoack3000@gmail.com [mic: Cosmetic fixes, moved struct layer_access_masks definition] Signed-off-by: Mickaël Salaün <mic@digikod.net>
2026-02-06 16:11:55 +01:00
landlock: Transpose the layer masks data structure The layer masks data structure tracks the requested but unfulfilled access rights during an operation's security check. It stores one bit for each combination of access right and layer index. If the bit is set, that access right is not granted (yet) in the given layer and we have to traverse the path further upwards to grant it. Previously, the layer masks were stored as arrays mapping from access right indices to layer_mask_t. The layer_mask_t value then indicates all layers in which the given access right is still (tentatively) denied. This patch introduces struct layer_access_masks instead: This struct contains an array with the access_mask_t of each (tentatively) denied access right in that layer. The hypothesis of this patch is that this simplifies the code enough so that the resulting code will run faster: * We can use bitwise operations in multiple places where we previously looped over bits individually with macros. (Should require less branch speculation and lends itself to better loop unrolling.) * Code is ~75 lines smaller. Other noteworthy changes: * In no_more_access(), call a new helper function may_refer(), which only solves the asymmetric case. Previously, the code interleaved the checks for the two symmetric cases in RENAME_EXCHANGE. It feels that the code is clearer when renames without RENAME_EXCHANGE are more obviously the normal case. Tradeoffs: This change improves performance, at a slight size increase to the layer masks data structure. This fixes the size of the data structure at 32 bytes for all types of access rights. (64, once we introduce a 17th filesystem access right). For filesystem access rights, at the moment, the data structure has the same size as before, but once we introduce the 17th filesystem access right, it will double in size (from 32 to 64 bytes), as access_mask_t grows from 16 to 32 bit [1]. Link: https://lore.kernel.org/all/20260120.haeCh4li9Vae@digikod.net/ [1] Signed-off-by: Günther Noack <gnoack3000@gmail.com> Link: https://lore.kernel.org/r/20260206151154.97915-5-gnoack3000@gmail.com [mic: Cosmetic fixes, moved struct layer_access_masks definition] Signed-off-by: Mickaël Salaün <mic@digikod.net>
2026-02-06 16:11:55 +01:00
landlock: Transpose the layer masks data structure The layer masks data structure tracks the requested but unfulfilled access rights during an operation's security check. It stores one bit for each combination of access right and layer index. If the bit is set, that access right is not granted (yet) in the given layer and we have to traverse the path further upwards to grant it. Previously, the layer masks were stored as arrays mapping from access right indices to layer_mask_t. The layer_mask_t value then indicates all layers in which the given access right is still (tentatively) denied. This patch introduces struct layer_access_masks instead: This struct contains an array with the access_mask_t of each (tentatively) denied access right in that layer. The hypothesis of this patch is that this simplifies the code enough so that the resulting code will run faster: * We can use bitwise operations in multiple places where we previously looped over bits individually with macros. (Should require less branch speculation and lends itself to better loop unrolling.) * Code is ~75 lines smaller. Other noteworthy changes: * In no_more_access(), call a new helper function may_refer(), which only solves the asymmetric case. Previously, the code interleaved the checks for the two symmetric cases in RENAME_EXCHANGE. It feels that the code is clearer when renames without RENAME_EXCHANGE are more obviously the normal case. Tradeoffs: This change improves performance, at a slight size increase to the layer masks data structure. This fixes the size of the data structure at 32 bytes for all types of access rights. (64, once we introduce a 17th filesystem access right). For filesystem access rights, at the moment, the data structure has the same size as before, but once we introduce the 17th filesystem access right, it will double in size (from 32 to 64 bytes), as access_mask_t grows from 16 to 32 bit [1]. Link: https://lore.kernel.org/all/20260120.haeCh4li9Vae@digikod.net/ [1] Signed-off-by: Günther Noack <gnoack3000@gmail.com> Link: https://lore.kernel.org/r/20260206151154.97915-5-gnoack3000@gmail.com [mic: Cosmetic fixes, moved struct layer_access_masks definition] Signed-off-by: Mickaël Salaün <mic@digikod.net>
2026-02-06 16:11:55 +01:00
landlock: Transpose the layer masks data structure The layer masks data structure tracks the requested but unfulfilled access rights during an operation's security check. It stores one bit for each combination of access right and layer index. If the bit is set, that access right is not granted (yet) in the given layer and we have to traverse the path further upwards to grant it. Previously, the layer masks were stored as arrays mapping from access right indices to layer_mask_t. The layer_mask_t value then indicates all layers in which the given access right is still (tentatively) denied. This patch introduces struct layer_access_masks instead: This struct contains an array with the access_mask_t of each (tentatively) denied access right in that layer. The hypothesis of this patch is that this simplifies the code enough so that the resulting code will run faster: * We can use bitwise operations in multiple places where we previously looped over bits individually with macros. (Should require less branch speculation and lends itself to better loop unrolling.) * Code is ~75 lines smaller. Other noteworthy changes: * In no_more_access(), call a new helper function may_refer(), which only solves the asymmetric case. Previously, the code interleaved the checks for the two symmetric cases in RENAME_EXCHANGE. It feels that the code is clearer when renames without RENAME_EXCHANGE are more obviously the normal case. Tradeoffs: This change improves performance, at a slight size increase to the layer masks data structure. This fixes the size of the data structure at 32 bytes for all types of access rights. (64, once we introduce a 17th filesystem access right). For filesystem access rights, at the moment, the data structure has the same size as before, but once we introduce the 17th filesystem access right, it will double in size (from 32 to 64 bytes), as access_mask_t grows from 16 to 32 bit [1]. Link: https://lore.kernel.org/all/20260120.haeCh4li9Vae@digikod.net/ [1] Signed-off-by: Günther Noack <gnoack3000@gmail.com> Link: https://lore.kernel.org/r/20260206151154.97915-5-gnoack3000@gmail.com [mic: Cosmetic fixes, moved struct layer_access_masks definition] Signed-off-by: Mickaël Salaün <mic@digikod.net>
2026-02-06 16:11:55 +01:00
landlock: Transpose the layer masks data structure The layer masks data structure tracks the requested but unfulfilled access rights during an operation's security check. It stores one bit for each combination of access right and layer index. If the bit is set, that access right is not granted (yet) in the given layer and we have to traverse the path further upwards to grant it. Previously, the layer masks were stored as arrays mapping from access right indices to layer_mask_t. The layer_mask_t value then indicates all layers in which the given access right is still (tentatively) denied. This patch introduces struct layer_access_masks instead: This struct contains an array with the access_mask_t of each (tentatively) denied access right in that layer. The hypothesis of this patch is that this simplifies the code enough so that the resulting code will run faster: * We can use bitwise operations in multiple places where we previously looped over bits individually with macros. (Should require less branch speculation and lends itself to better loop unrolling.) * Code is ~75 lines smaller. Other noteworthy changes: * In no_more_access(), call a new helper function may_refer(), which only solves the asymmetric case. Previously, the code interleaved the checks for the two symmetric cases in RENAME_EXCHANGE. It feels that the code is clearer when renames without RENAME_EXCHANGE are more obviously the normal case. Tradeoffs: This change improves performance, at a slight size increase to the layer masks data structure. This fixes the size of the data structure at 32 bytes for all types of access rights. (64, once we introduce a 17th filesystem access right). For filesystem access rights, at the moment, the data structure has the same size as before, but once we introduce the 17th filesystem access right, it will double in size (from 32 to 64 bytes), as access_mask_t grows from 16 to 32 bit [1]. Link: https://lore.kernel.org/all/20260120.haeCh4li9Vae@digikod.net/ [1] Signed-off-by: Günther Noack <gnoack3000@gmail.com> Link: https://lore.kernel.org/r/20260206151154.97915-5-gnoack3000@gmail.com [mic: Cosmetic fixes, moved struct layer_access_masks definition] Signed-off-by: Mickaël Salaün <mic@digikod.net>
2026-02-06 16:11:55 +01:00
landlock: Transpose the layer masks data structure The layer masks data structure tracks the requested but unfulfilled access rights during an operation's security check. It stores one bit for each combination of access right and layer index. If the bit is set, that access right is not granted (yet) in the given layer and we have to traverse the path further upwards to grant it. Previously, the layer masks were stored as arrays mapping from access right indices to layer_mask_t. The layer_mask_t value then indicates all layers in which the given access right is still (tentatively) denied. This patch introduces struct layer_access_masks instead: This struct contains an array with the access_mask_t of each (tentatively) denied access right in that layer. The hypothesis of this patch is that this simplifies the code enough so that the resulting code will run faster: * We can use bitwise operations in multiple places where we previously looped over bits individually with macros. (Should require less branch speculation and lends itself to better loop unrolling.) * Code is ~75 lines smaller. Other noteworthy changes: * In no_more_access(), call a new helper function may_refer(), which only solves the asymmetric case. Previously, the code interleaved the checks for the two symmetric cases in RENAME_EXCHANGE. It feels that the code is clearer when renames without RENAME_EXCHANGE are more obviously the normal case. Tradeoffs: This change improves performance, at a slight size increase to the layer masks data structure. This fixes the size of the data structure at 32 bytes for all types of access rights. (64, once we introduce a 17th filesystem access right). For filesystem access rights, at the moment, the data structure has the same size as before, but once we introduce the 17th filesystem access right, it will double in size (from 32 to 64 bytes), as access_mask_t grows from 16 to 32 bit [1]. Link: https://lore.kernel.org/all/20260120.haeCh4li9Vae@digikod.net/ [1] Signed-off-by: Günther Noack <gnoack3000@gmail.com> Link: https://lore.kernel.org/r/20260206151154.97915-5-gnoack3000@gmail.com [mic: Cosmetic fixes, moved struct layer_access_masks definition] Signed-off-by: Mickaël Salaün <mic@digikod.net>
2026-02-06 16:11:55 +01:00
landlock: Transpose the layer masks data structure The layer masks data structure tracks the requested but unfulfilled access rights during an operation's security check. It stores one bit for each combination of access right and layer index. If the bit is set, that access right is not granted (yet) in the given layer and we have to traverse the path further upwards to grant it. Previously, the layer masks were stored as arrays mapping from access right indices to layer_mask_t. The layer_mask_t value then indicates all layers in which the given access right is still (tentatively) denied. This patch introduces struct layer_access_masks instead: This struct contains an array with the access_mask_t of each (tentatively) denied access right in that layer. The hypothesis of this patch is that this simplifies the code enough so that the resulting code will run faster: * We can use bitwise operations in multiple places where we previously looped over bits individually with macros. (Should require less branch speculation and lends itself to better loop unrolling.) * Code is ~75 lines smaller. Other noteworthy changes: * In no_more_access(), call a new helper function may_refer(), which only solves the asymmetric case. Previously, the code interleaved the checks for the two symmetric cases in RENAME_EXCHANGE. It feels that the code is clearer when renames without RENAME_EXCHANGE are more obviously the normal case. Tradeoffs: This change improves performance, at a slight size increase to the layer masks data structure. This fixes the size of the data structure at 32 bytes for all types of access rights. (64, once we introduce a 17th filesystem access right). For filesystem access rights, at the moment, the data structure has the same size as before, but once we introduce the 17th filesystem access right, it will double in size (from 32 to 64 bytes), as access_mask_t grows from 16 to 32 bit [1]. Link: https://lore.kernel.org/all/20260120.haeCh4li9Vae@digikod.net/ [1] Signed-off-by: Günther Noack <gnoack3000@gmail.com> Link: https://lore.kernel.org/r/20260206151154.97915-5-gnoack3000@gmail.com [mic: Cosmetic fixes, moved struct layer_access_masks definition] Signed-off-by: Mickaël Salaün <mic@digikod.net>
2026-02-06 16:11:55 +01:00
landlock: Transpose the layer masks data structure The layer masks data structure tracks the requested but unfulfilled access rights during an operation's security check. It stores one bit for each combination of access right and layer index. If the bit is set, that access right is not granted (yet) in the given layer and we have to traverse the path further upwards to grant it. Previously, the layer masks were stored as arrays mapping from access right indices to layer_mask_t. The layer_mask_t value then indicates all layers in which the given access right is still (tentatively) denied. This patch introduces struct layer_access_masks instead: This struct contains an array with the access_mask_t of each (tentatively) denied access right in that layer. The hypothesis of this patch is that this simplifies the code enough so that the resulting code will run faster: * We can use bitwise operations in multiple places where we previously looped over bits individually with macros. (Should require less branch speculation and lends itself to better loop unrolling.) * Code is ~75 lines smaller. Other noteworthy changes: * In no_more_access(), call a new helper function may_refer(), which only solves the asymmetric case. Previously, the code interleaved the checks for the two symmetric cases in RENAME_EXCHANGE. It feels that the code is clearer when renames without RENAME_EXCHANGE are more obviously the normal case. Tradeoffs: This change improves performance, at a slight size increase to the layer masks data structure. This fixes the size of the data structure at 32 bytes for all types of access rights. (64, once we introduce a 17th filesystem access right). For filesystem access rights, at the moment, the data structure has the same size as before, but once we introduce the 17th filesystem access right, it will double in size (from 32 to 64 bytes), as access_mask_t grows from 16 to 32 bit [1]. Link: https://lore.kernel.org/all/20260120.haeCh4li9Vae@digikod.net/ [1] Signed-off-by: Günther Noack <gnoack3000@gmail.com> Link: https://lore.kernel.org/r/20260206151154.97915-5-gnoack3000@gmail.com [mic: Cosmetic fixes, moved struct layer_access_masks definition] Signed-off-by: Mickaël Salaün <mic@digikod.net>
2026-02-06 16:11:55 +01:00
landlock: Transpose the layer masks data structure The layer masks data structure tracks the requested but unfulfilled access rights during an operation's security check. It stores one bit for each combination of access right and layer index. If the bit is set, that access right is not granted (yet) in the given layer and we have to traverse the path further upwards to grant it. Previously, the layer masks were stored as arrays mapping from access right indices to layer_mask_t. The layer_mask_t value then indicates all layers in which the given access right is still (tentatively) denied. This patch introduces struct layer_access_masks instead: This struct contains an array with the access_mask_t of each (tentatively) denied access right in that layer. The hypothesis of this patch is that this simplifies the code enough so that the resulting code will run faster: * We can use bitwise operations in multiple places where we previously looped over bits individually with macros. (Should require less branch speculation and lends itself to better loop unrolling.) * Code is ~75 lines smaller. Other noteworthy changes: * In no_more_access(), call a new helper function may_refer(), which only solves the asymmetric case. Previously, the code interleaved the checks for the two symmetric cases in RENAME_EXCHANGE. It feels that the code is clearer when renames without RENAME_EXCHANGE are more obviously the normal case. Tradeoffs: This change improves performance, at a slight size increase to the layer masks data structure. This fixes the size of the data structure at 32 bytes for all types of access rights. (64, once we introduce a 17th filesystem access right). For filesystem access rights, at the moment, the data structure has the same size as before, but once we introduce the 17th filesystem access right, it will double in size (from 32 to 64 bytes), as access_mask_t grows from 16 to 32 bit [1]. Link: https://lore.kernel.org/all/20260120.haeCh4li9Vae@digikod.net/ [1] Signed-off-by: Günther Noack <gnoack3000@gmail.com> Link: https://lore.kernel.org/r/20260206151154.97915-5-gnoack3000@gmail.com [mic: Cosmetic fixes, moved struct layer_access_masks definition] Signed-off-by: Mickaël Salaün <mic@digikod.net>
2026-02-06 16:11:55 +01:00
landlock: Add AUDIT_LANDLOCK_ACCESS and log ptrace denials Add a new AUDIT_LANDLOCK_ACCESS record type dedicated to an access request denied by a Landlock domain. AUDIT_LANDLOCK_ACCESS indicates that something unexpected happened. For now, only denied access are logged, which means that any AUDIT_LANDLOCK_ACCESS record is always followed by a SYSCALL record with "success=no". However, log parsers should check this syscall property because this is the only sign that a request was denied. Indeed, we could have "success=yes" if Landlock would support a "permissive" mode. We could also add a new field to AUDIT_LANDLOCK_DOMAIN for this mode (see following commit). By default, the only logged access requests are those coming from the same executed program that enforced the Landlock restriction on itself. In other words, no audit record are created for a task after it called execve(2). This is required to avoid log spam because programs may only be aware of their own restrictions, but not the inherited ones. Following commits will allow to conditionally generate AUDIT_LANDLOCK_ACCESS records according to dedicated landlock_restrict_self(2)'s flags. The AUDIT_LANDLOCK_ACCESS message contains: - the "domain" ID restricting the action on an object, - the "blockers" that are missing to allow the requested access, - a set of fields identifying the related object (e.g. task identified with "opid" and "ocomm"). The blockers are implicit restrictions (e.g. ptrace), or explicit access rights (e.g. filesystem), or explicit scopes (e.g. signal). This field contains a list of at least one element, each separated with a comma. The initial blocker is "ptrace", which describe all implicit Landlock restrictions related to ptrace (e.g. deny tracing of tasks outside a sandbox). Add audit support to ptrace_access_check and ptrace_traceme hooks. For the ptrace_access_check case, we log the current/parent domain and the child task. For the ptrace_traceme case, we log the parent domain and the current/child task. Indeed, the requester and the target are the current task, but the action would be performed by the parent task. Audit event sample: type=LANDLOCK_ACCESS msg=audit(1729738800.349:44): domain=195ba459b blockers=ptrace opid=1 ocomm="systemd" type=SYSCALL msg=audit(1729738800.349:44): arch=c000003e syscall=101 success=no [...] pid=300 auid=0 A following commit adds user documentation. Add KUnit tests to check reading of domain ID relative to layer level. The quick return for non-landlocked tasks is moved from task_ptrace() to each LSM hooks. It is not useful to inline the audit_enabled check because other computation are performed by landlock_log_denial(). Use scoped guards for RCU read-side critical sections. Cc: Günther Noack <gnoack@google.com> Acked-by: Paul Moore <paul@paul-moore.com> Link: https://lore.kernel.org/r/20250320190717.2287696-10-mic@digikod.net Signed-off-by: Mickaël Salaün <mic@digikod.net>
2025-03-20 20:06:58 +01:00
landlock: Add AUDIT_LANDLOCK_ACCESS and log ptrace denials Add a new AUDIT_LANDLOCK_ACCESS record type dedicated to an access request denied by a Landlock domain. AUDIT_LANDLOCK_ACCESS indicates that something unexpected happened. For now, only denied access are logged, which means that any AUDIT_LANDLOCK_ACCESS record is always followed by a SYSCALL record with "success=no". However, log parsers should check this syscall property because this is the only sign that a request was denied. Indeed, we could have "success=yes" if Landlock would support a "permissive" mode. We could also add a new field to AUDIT_LANDLOCK_DOMAIN for this mode (see following commit). By default, the only logged access requests are those coming from the same executed program that enforced the Landlock restriction on itself. In other words, no audit record are created for a task after it called execve(2). This is required to avoid log spam because programs may only be aware of their own restrictions, but not the inherited ones. Following commits will allow to conditionally generate AUDIT_LANDLOCK_ACCESS records according to dedicated landlock_restrict_self(2)'s flags. The AUDIT_LANDLOCK_ACCESS message contains: - the "domain" ID restricting the action on an object, - the "blockers" that are missing to allow the requested access, - a set of fields identifying the related object (e.g. task identified with "opid" and "ocomm"). The blockers are implicit restrictions (e.g. ptrace), or explicit access rights (e.g. filesystem), or explicit scopes (e.g. signal). This field contains a list of at least one element, each separated with a comma. The initial blocker is "ptrace", which describe all implicit Landlock restrictions related to ptrace (e.g. deny tracing of tasks outside a sandbox). Add audit support to ptrace_access_check and ptrace_traceme hooks. For the ptrace_access_check case, we log the current/parent domain and the child task. For the ptrace_traceme case, we log the parent domain and the current/child task. Indeed, the requester and the target are the current task, but the action would be performed by the parent task. Audit event sample: type=LANDLOCK_ACCESS msg=audit(1729738800.349:44): domain=195ba459b blockers=ptrace opid=1 ocomm="systemd" type=SYSCALL msg=audit(1729738800.349:44): arch=c000003e syscall=101 success=no [...] pid=300 auid=0 A following commit adds user documentation. Add KUnit tests to check reading of domain ID relative to layer level. The quick return for non-landlocked tasks is moved from task_ptrace() to each LSM hooks. It is not useful to inline the audit_enabled check because other computation are performed by landlock_log_denial(). Use scoped guards for RCU read-side critical sections. Cc: Günther Noack <gnoack@google.com> Acked-by: Paul Moore <paul@paul-moore.com> Link: https://lore.kernel.org/r/20250320190717.2287696-10-mic@digikod.net Signed-off-by: Mickaël Salaün <mic@digikod.net>
2025-03-20 20:06:58 +01:00
landlock: Add AUDIT_LANDLOCK_ACCESS and log ptrace denials Add a new AUDIT_LANDLOCK_ACCESS record type dedicated to an access request denied by a Landlock domain. AUDIT_LANDLOCK_ACCESS indicates that something unexpected happened. For now, only denied access are logged, which means that any AUDIT_LANDLOCK_ACCESS record is always followed by a SYSCALL record with "success=no". However, log parsers should check this syscall property because this is the only sign that a request was denied. Indeed, we could have "success=yes" if Landlock would support a "permissive" mode. We could also add a new field to AUDIT_LANDLOCK_DOMAIN for this mode (see following commit). By default, the only logged access requests are those coming from the same executed program that enforced the Landlock restriction on itself. In other words, no audit record are created for a task after it called execve(2). This is required to avoid log spam because programs may only be aware of their own restrictions, but not the inherited ones. Following commits will allow to conditionally generate AUDIT_LANDLOCK_ACCESS records according to dedicated landlock_restrict_self(2)'s flags. The AUDIT_LANDLOCK_ACCESS message contains: - the "domain" ID restricting the action on an object, - the "blockers" that are missing to allow the requested access, - a set of fields identifying the related object (e.g. task identified with "opid" and "ocomm"). The blockers are implicit restrictions (e.g. ptrace), or explicit access rights (e.g. filesystem), or explicit scopes (e.g. signal). This field contains a list of at least one element, each separated with a comma. The initial blocker is "ptrace", which describe all implicit Landlock restrictions related to ptrace (e.g. deny tracing of tasks outside a sandbox). Add audit support to ptrace_access_check and ptrace_traceme hooks. For the ptrace_access_check case, we log the current/parent domain and the child task. For the ptrace_traceme case, we log the parent domain and the current/child task. Indeed, the requester and the target are the current task, but the action would be performed by the parent task. Audit event sample: type=LANDLOCK_ACCESS msg=audit(1729738800.349:44): domain=195ba459b blockers=ptrace opid=1 ocomm="systemd" type=SYSCALL msg=audit(1729738800.349:44): arch=c000003e syscall=101 success=no [...] pid=300 auid=0 A following commit adds user documentation. Add KUnit tests to check reading of domain ID relative to layer level. The quick return for non-landlocked tasks is moved from task_ptrace() to each LSM hooks. It is not useful to inline the audit_enabled check because other computation are performed by landlock_log_denial(). Use scoped guards for RCU read-side critical sections. Cc: Günther Noack <gnoack@google.com> Acked-by: Paul Moore <paul@paul-moore.com> Link: https://lore.kernel.org/r/20250320190717.2287696-10-mic@digikod.net Signed-off-by: Mickaël Salaün <mic@digikod.net>
2025-03-20 20:06:58 +01:00
landlock: Transpose the layer masks data structure The layer masks data structure tracks the requested but unfulfilled access rights during an operation's security check. It stores one bit for each combination of access right and layer index. If the bit is set, that access right is not granted (yet) in the given layer and we have to traverse the path further upwards to grant it. Previously, the layer masks were stored as arrays mapping from access right indices to layer_mask_t. The layer_mask_t value then indicates all layers in which the given access right is still (tentatively) denied. This patch introduces struct layer_access_masks instead: This struct contains an array with the access_mask_t of each (tentatively) denied access right in that layer. The hypothesis of this patch is that this simplifies the code enough so that the resulting code will run faster: * We can use bitwise operations in multiple places where we previously looped over bits individually with macros. (Should require less branch speculation and lends itself to better loop unrolling.) * Code is ~75 lines smaller. Other noteworthy changes: * In no_more_access(), call a new helper function may_refer(), which only solves the asymmetric case. Previously, the code interleaved the checks for the two symmetric cases in RENAME_EXCHANGE. It feels that the code is clearer when renames without RENAME_EXCHANGE are more obviously the normal case. Tradeoffs: This change improves performance, at a slight size increase to the layer masks data structure. This fixes the size of the data structure at 32 bytes for all types of access rights. (64, once we introduce a 17th filesystem access right). For filesystem access rights, at the moment, the data structure has the same size as before, but once we introduce the 17th filesystem access right, it will double in size (from 32 to 64 bytes), as access_mask_t grows from 16 to 32 bit [1]. Link: https://lore.kernel.org/all/20260120.haeCh4li9Vae@digikod.net/ [1] Signed-off-by: Günther Noack <gnoack3000@gmail.com> Link: https://lore.kernel.org/r/20260206151154.97915-5-gnoack3000@gmail.com [mic: Cosmetic fixes, moved struct layer_access_masks definition] Signed-off-by: Mickaël Salaün <mic@digikod.net>
2026-02-06 16:11:55 +01:00
landlock: Transpose the layer masks data structure The layer masks data structure tracks the requested but unfulfilled access rights during an operation's security check. It stores one bit for each combination of access right and layer index. If the bit is set, that access right is not granted (yet) in the given layer and we have to traverse the path further upwards to grant it. Previously, the layer masks were stored as arrays mapping from access right indices to layer_mask_t. The layer_mask_t value then indicates all layers in which the given access right is still (tentatively) denied. This patch introduces struct layer_access_masks instead: This struct contains an array with the access_mask_t of each (tentatively) denied access right in that layer. The hypothesis of this patch is that this simplifies the code enough so that the resulting code will run faster: * We can use bitwise operations in multiple places where we previously looped over bits individually with macros. (Should require less branch speculation and lends itself to better loop unrolling.) * Code is ~75 lines smaller. Other noteworthy changes: * In no_more_access(), call a new helper function may_refer(), which only solves the asymmetric case. Previously, the code interleaved the checks for the two symmetric cases in RENAME_EXCHANGE. It feels that the code is clearer when renames without RENAME_EXCHANGE are more obviously the normal case. Tradeoffs: This change improves performance, at a slight size increase to the layer masks data structure. This fixes the size of the data structure at 32 bytes for all types of access rights. (64, once we introduce a 17th filesystem access right). For filesystem access rights, at the moment, the data structure has the same size as before, but once we introduce the 17th filesystem access right, it will double in size (from 32 to 64 bytes), as access_mask_t grows from 16 to 32 bit [1]. Link: https://lore.kernel.org/all/20260120.haeCh4li9Vae@digikod.net/ [1] Signed-off-by: Günther Noack <gnoack3000@gmail.com> Link: https://lore.kernel.org/r/20260206151154.97915-5-gnoack3000@gmail.com [mic: Cosmetic fixes, moved struct layer_access_masks definition] Signed-off-by: Mickaël Salaün <mic@digikod.net>
2026-02-06 16:11:55 +01:00
landlock: Add AUDIT_LANDLOCK_ACCESS and log ptrace denials Add a new AUDIT_LANDLOCK_ACCESS record type dedicated to an access request denied by a Landlock domain. AUDIT_LANDLOCK_ACCESS indicates that something unexpected happened. For now, only denied access are logged, which means that any AUDIT_LANDLOCK_ACCESS record is always followed by a SYSCALL record with "success=no". However, log parsers should check this syscall property because this is the only sign that a request was denied. Indeed, we could have "success=yes" if Landlock would support a "permissive" mode. We could also add a new field to AUDIT_LANDLOCK_DOMAIN for this mode (see following commit). By default, the only logged access requests are those coming from the same executed program that enforced the Landlock restriction on itself. In other words, no audit record are created for a task after it called execve(2). This is required to avoid log spam because programs may only be aware of their own restrictions, but not the inherited ones. Following commits will allow to conditionally generate AUDIT_LANDLOCK_ACCESS records according to dedicated landlock_restrict_self(2)'s flags. The AUDIT_LANDLOCK_ACCESS message contains: - the "domain" ID restricting the action on an object, - the "blockers" that are missing to allow the requested access, - a set of fields identifying the related object (e.g. task identified with "opid" and "ocomm"). The blockers are implicit restrictions (e.g. ptrace), or explicit access rights (e.g. filesystem), or explicit scopes (e.g. signal). This field contains a list of at least one element, each separated with a comma. The initial blocker is "ptrace", which describe all implicit Landlock restrictions related to ptrace (e.g. deny tracing of tasks outside a sandbox). Add audit support to ptrace_access_check and ptrace_traceme hooks. For the ptrace_access_check case, we log the current/parent domain and the child task. For the ptrace_traceme case, we log the parent domain and the current/child task. Indeed, the requester and the target are the current task, but the action would be performed by the parent task. Audit event sample: type=LANDLOCK_ACCESS msg=audit(1729738800.349:44): domain=195ba459b blockers=ptrace opid=1 ocomm="systemd" type=SYSCALL msg=audit(1729738800.349:44): arch=c000003e syscall=101 success=no [...] pid=300 auid=0 A following commit adds user documentation. Add KUnit tests to check reading of domain ID relative to layer level. The quick return for non-landlocked tasks is moved from task_ptrace() to each LSM hooks. It is not useful to inline the audit_enabled check because other computation are performed by landlock_log_denial(). Use scoped guards for RCU read-side critical sections. Cc: Günther Noack <gnoack@google.com> Acked-by: Paul Moore <paul@paul-moore.com> Link: https://lore.kernel.org/r/20250320190717.2287696-10-mic@digikod.net Signed-off-by: Mickaël Salaün <mic@digikod.net>
2025-03-20 20:06:58 +01:00
landlock: Add LANDLOCK_RESTRICT_SELF_LOG_*_EXEC_* flags Most of the time we want to log denied access because they should not happen and such information helps diagnose issues. However, when sandboxing processes that we know will try to access denied resources (e.g. unknown, bogus, or malicious binary), we might want to not log related access requests that might fill up logs. By default, denied requests are logged until the task call execve(2). If the LANDLOCK_RESTRICT_SELF_LOG_SAME_EXEC_OFF flag is set, denied requests will not be logged for the same executed file. If the LANDLOCK_RESTRICT_SELF_LOG_NEW_EXEC_ON flag is set, denied requests from after an execve(2) call will be logged. The rationale is that a program should know its own behavior, but not necessarily the behavior of other programs. Because LANDLOCK_RESTRICT_SELF_LOG_SAME_EXEC_OFF is set for a specific Landlock domain, it makes it possible to selectively mask some access requests that would be logged by a parent domain, which might be handy for unprivileged processes to limit logs. However, system administrators should still use the audit filtering mechanism. There is intentionally no audit nor sysctl configuration to re-enable these logs. This is delegated to the user space program. Increment the Landlock ABI version to reflect this interface change. Cc: Günther Noack <gnoack@google.com> Cc: Paul Moore <paul@paul-moore.com> Link: https://lore.kernel.org/r/20250320190717.2287696-18-mic@digikod.net [mic: Rename variables and fix __maybe_unused] Signed-off-by: Mickaël Salaün <mic@digikod.net>
2025-03-20 20:07:06 +01:00
landlock: Add AUDIT_LANDLOCK_DOMAIN and log domain status Asynchronously log domain information when it first denies an access. This minimize the amount of generated logs, which makes it possible to always log denials for the current execution since they should not happen. These records are identified with the new AUDIT_LANDLOCK_DOMAIN type. The AUDIT_LANDLOCK_DOMAIN message contains: - the "domain" ID which is described; - the "status" which can either be "allocated" or "deallocated"; - the "mode" which is for now only "enforcing"; - for the "allocated" status, a minimal set of properties to easily identify the task that loaded the domain's policy with landlock_restrict_self(2): "pid", "uid", executable path ("exe"), and command line ("comm"); - for the "deallocated" state, the number of "denials" accounted to this domain, which is at least 1. This requires each domain to save these task properties at creation time in the new struct landlock_details. A reference to the PID is kept for the lifetime of the domain to avoid race conditions when investigating the related task. The executable path is resolved and stored to not keep a reference to the filesystem and block related actions. All these metadata are stored for the lifetime of the related domain and should then be minimal. The required memory is not accounted to the task calling landlock_restrict_self(2) contrary to most other Landlock allocations (see related comment). The AUDIT_LANDLOCK_DOMAIN record follows the first AUDIT_LANDLOCK_ACCESS record for the same domain, which is always followed by AUDIT_SYSCALL and AUDIT_PROCTITLE. This is in line with the audit logic to first record the cause of an event, and then add context with other types of record. Audit event sample for a first denial: type=LANDLOCK_ACCESS msg=audit(1732186800.349:44): domain=195ba459b blockers=ptrace opid=1 ocomm="systemd" type=LANDLOCK_DOMAIN msg=audit(1732186800.349:44): domain=195ba459b status=allocated mode=enforcing pid=300 uid=0 exe="/root/sandboxer" comm="sandboxer" type=SYSCALL msg=audit(1732186800.349:44): arch=c000003e syscall=101 success=no [...] pid=300 auid=0 Audit event sample for a following denial: type=LANDLOCK_ACCESS msg=audit(1732186800.372:45): domain=195ba459b blockers=ptrace opid=1 ocomm="systemd" type=SYSCALL msg=audit(1732186800.372:45): arch=c000003e syscall=101 success=no [...] pid=300 auid=0 Log domain deletion with the "deallocated" state when a domain was previously logged. This makes it possible for log parsers to free potential resources when a domain ID will never show again. The number of denied access requests is useful to easily check how many access requests a domain blocked and potentially if some of them are missing in logs because of audit rate limiting, audit rules, or Landlock log configuration flags (see following commit). Audit event sample for a deletion of a domain that denied something: type=LANDLOCK_DOMAIN msg=audit(1732186800.393:46): domain=195ba459b status=deallocated denials=2 Cc: Günther Noack <gnoack@google.com> Acked-by: Paul Moore <paul@paul-moore.com> Link: https://lore.kernel.org/r/20250320190717.2287696-11-mic@digikod.net [mic: Update comment and GFP flag for landlock_log_drop_domain()] Signed-off-by: Mickaël Salaün <mic@digikod.net>
2025-03-20 20:06:59 +01:00
landlock: Add LANDLOCK_RESTRICT_SELF_LOG_*_EXEC_* flags Most of the time we want to log denied access because they should not happen and such information helps diagnose issues. However, when sandboxing processes that we know will try to access denied resources (e.g. unknown, bogus, or malicious binary), we might want to not log related access requests that might fill up logs. By default, denied requests are logged until the task call execve(2). If the LANDLOCK_RESTRICT_SELF_LOG_SAME_EXEC_OFF flag is set, denied requests will not be logged for the same executed file. If the LANDLOCK_RESTRICT_SELF_LOG_NEW_EXEC_ON flag is set, denied requests from after an execve(2) call will be logged. The rationale is that a program should know its own behavior, but not necessarily the behavior of other programs. Because LANDLOCK_RESTRICT_SELF_LOG_SAME_EXEC_OFF is set for a specific Landlock domain, it makes it possible to selectively mask some access requests that would be logged by a parent domain, which might be handy for unprivileged processes to limit logs. However, system administrators should still use the audit filtering mechanism. There is intentionally no audit nor sysctl configuration to re-enable these logs. This is delegated to the user space program. Increment the Landlock ABI version to reflect this interface change. Cc: Günther Noack <gnoack@google.com> Cc: Paul Moore <paul@paul-moore.com> Link: https://lore.kernel.org/r/20250320190717.2287696-18-mic@digikod.net [mic: Rename variables and fix __maybe_unused] Signed-off-by: Mickaël Salaün <mic@digikod.net>
2025-03-20 20:07:06 +01:00
landlock: Add LANDLOCK_RESTRICT_SELF_LOG_*_EXEC_* flags Most of the time we want to log denied access because they should not happen and such information helps diagnose issues. However, when sandboxing processes that we know will try to access denied resources (e.g. unknown, bogus, or malicious binary), we might want to not log related access requests that might fill up logs. By default, denied requests are logged until the task call execve(2). If the LANDLOCK_RESTRICT_SELF_LOG_SAME_EXEC_OFF flag is set, denied requests will not be logged for the same executed file. If the LANDLOCK_RESTRICT_SELF_LOG_NEW_EXEC_ON flag is set, denied requests from after an execve(2) call will be logged. The rationale is that a program should know its own behavior, but not necessarily the behavior of other programs. Because LANDLOCK_RESTRICT_SELF_LOG_SAME_EXEC_OFF is set for a specific Landlock domain, it makes it possible to selectively mask some access requests that would be logged by a parent domain, which might be handy for unprivileged processes to limit logs. However, system administrators should still use the audit filtering mechanism. There is intentionally no audit nor sysctl configuration to re-enable these logs. This is delegated to the user space program. Increment the Landlock ABI version to reflect this interface change. Cc: Günther Noack <gnoack@google.com> Cc: Paul Moore <paul@paul-moore.com> Link: https://lore.kernel.org/r/20250320190717.2287696-18-mic@digikod.net [mic: Rename variables and fix __maybe_unused] Signed-off-by: Mickaël Salaün <mic@digikod.net>
2025-03-20 20:07:06 +01:00
landlock: Add AUDIT_LANDLOCK_ACCESS and log ptrace denials Add a new AUDIT_LANDLOCK_ACCESS record type dedicated to an access request denied by a Landlock domain. AUDIT_LANDLOCK_ACCESS indicates that something unexpected happened. For now, only denied access are logged, which means that any AUDIT_LANDLOCK_ACCESS record is always followed by a SYSCALL record with "success=no". However, log parsers should check this syscall property because this is the only sign that a request was denied. Indeed, we could have "success=yes" if Landlock would support a "permissive" mode. We could also add a new field to AUDIT_LANDLOCK_DOMAIN for this mode (see following commit). By default, the only logged access requests are those coming from the same executed program that enforced the Landlock restriction on itself. In other words, no audit record are created for a task after it called execve(2). This is required to avoid log spam because programs may only be aware of their own restrictions, but not the inherited ones. Following commits will allow to conditionally generate AUDIT_LANDLOCK_ACCESS records according to dedicated landlock_restrict_self(2)'s flags. The AUDIT_LANDLOCK_ACCESS message contains: - the "domain" ID restricting the action on an object, - the "blockers" that are missing to allow the requested access, - a set of fields identifying the related object (e.g. task identified with "opid" and "ocomm"). The blockers are implicit restrictions (e.g. ptrace), or explicit access rights (e.g. filesystem), or explicit scopes (e.g. signal). This field contains a list of at least one element, each separated with a comma. The initial blocker is "ptrace", which describe all implicit Landlock restrictions related to ptrace (e.g. deny tracing of tasks outside a sandbox). Add audit support to ptrace_access_check and ptrace_traceme hooks. For the ptrace_access_check case, we log the current/parent domain and the child task. For the ptrace_traceme case, we log the parent domain and the current/child task. Indeed, the requester and the target are the current task, but the action would be performed by the parent task. Audit event sample: type=LANDLOCK_ACCESS msg=audit(1729738800.349:44): domain=195ba459b blockers=ptrace opid=1 ocomm="systemd" type=SYSCALL msg=audit(1729738800.349:44): arch=c000003e syscall=101 success=no [...] pid=300 auid=0 A following commit adds user documentation. Add KUnit tests to check reading of domain ID relative to layer level. The quick return for non-landlocked tasks is moved from task_ptrace() to each LSM hooks. It is not useful to inline the audit_enabled check because other computation are performed by landlock_log_denial(). Use scoped guards for RCU read-side critical sections. Cc: Günther Noack <gnoack@google.com> Acked-by: Paul Moore <paul@paul-moore.com> Link: https://lore.kernel.org/r/20250320190717.2287696-10-mic@digikod.net Signed-off-by: Mickaël Salaün <mic@digikod.net>
2025-03-20 20:06:58 +01:00
landlock: Add AUDIT_LANDLOCK_ACCESS and log ptrace denials Add a new AUDIT_LANDLOCK_ACCESS record type dedicated to an access request denied by a Landlock domain. AUDIT_LANDLOCK_ACCESS indicates that something unexpected happened. For now, only denied access are logged, which means that any AUDIT_LANDLOCK_ACCESS record is always followed by a SYSCALL record with "success=no". However, log parsers should check this syscall property because this is the only sign that a request was denied. Indeed, we could have "success=yes" if Landlock would support a "permissive" mode. We could also add a new field to AUDIT_LANDLOCK_DOMAIN for this mode (see following commit). By default, the only logged access requests are those coming from the same executed program that enforced the Landlock restriction on itself. In other words, no audit record are created for a task after it called execve(2). This is required to avoid log spam because programs may only be aware of their own restrictions, but not the inherited ones. Following commits will allow to conditionally generate AUDIT_LANDLOCK_ACCESS records according to dedicated landlock_restrict_self(2)'s flags. The AUDIT_LANDLOCK_ACCESS message contains: - the "domain" ID restricting the action on an object, - the "blockers" that are missing to allow the requested access, - a set of fields identifying the related object (e.g. task identified with "opid" and "ocomm"). The blockers are implicit restrictions (e.g. ptrace), or explicit access rights (e.g. filesystem), or explicit scopes (e.g. signal). This field contains a list of at least one element, each separated with a comma. The initial blocker is "ptrace", which describe all implicit Landlock restrictions related to ptrace (e.g. deny tracing of tasks outside a sandbox). Add audit support to ptrace_access_check and ptrace_traceme hooks. For the ptrace_access_check case, we log the current/parent domain and the child task. For the ptrace_traceme case, we log the parent domain and the current/child task. Indeed, the requester and the target are the current task, but the action would be performed by the parent task. Audit event sample: type=LANDLOCK_ACCESS msg=audit(1729738800.349:44): domain=195ba459b blockers=ptrace opid=1 ocomm="systemd" type=SYSCALL msg=audit(1729738800.349:44): arch=c000003e syscall=101 success=no [...] pid=300 auid=0 A following commit adds user documentation. Add KUnit tests to check reading of domain ID relative to layer level. The quick return for non-landlocked tasks is moved from task_ptrace() to each LSM hooks. It is not useful to inline the audit_enabled check because other computation are performed by landlock_log_denial(). Use scoped guards for RCU read-side critical sections. Cc: Günther Noack <gnoack@google.com> Acked-by: Paul Moore <paul@paul-moore.com> Link: https://lore.kernel.org/r/20250320190717.2287696-10-mic@digikod.net Signed-off-by: Mickaël Salaün <mic@digikod.net>
2025-03-20 20:06:58 +01:00
landlock: Add AUDIT_LANDLOCK_DOMAIN and log domain status Asynchronously log domain information when it first denies an access. This minimize the amount of generated logs, which makes it possible to always log denials for the current execution since they should not happen. These records are identified with the new AUDIT_LANDLOCK_DOMAIN type. The AUDIT_LANDLOCK_DOMAIN message contains: - the "domain" ID which is described; - the "status" which can either be "allocated" or "deallocated"; - the "mode" which is for now only "enforcing"; - for the "allocated" status, a minimal set of properties to easily identify the task that loaded the domain's policy with landlock_restrict_self(2): "pid", "uid", executable path ("exe"), and command line ("comm"); - for the "deallocated" state, the number of "denials" accounted to this domain, which is at least 1. This requires each domain to save these task properties at creation time in the new struct landlock_details. A reference to the PID is kept for the lifetime of the domain to avoid race conditions when investigating the related task. The executable path is resolved and stored to not keep a reference to the filesystem and block related actions. All these metadata are stored for the lifetime of the related domain and should then be minimal. The required memory is not accounted to the task calling landlock_restrict_self(2) contrary to most other Landlock allocations (see related comment). The AUDIT_LANDLOCK_DOMAIN record follows the first AUDIT_LANDLOCK_ACCESS record for the same domain, which is always followed by AUDIT_SYSCALL and AUDIT_PROCTITLE. This is in line with the audit logic to first record the cause of an event, and then add context with other types of record. Audit event sample for a first denial: type=LANDLOCK_ACCESS msg=audit(1732186800.349:44): domain=195ba459b blockers=ptrace opid=1 ocomm="systemd" type=LANDLOCK_DOMAIN msg=audit(1732186800.349:44): domain=195ba459b status=allocated mode=enforcing pid=300 uid=0 exe="/root/sandboxer" comm="sandboxer" type=SYSCALL msg=audit(1732186800.349:44): arch=c000003e syscall=101 success=no [...] pid=300 auid=0 Audit event sample for a following denial: type=LANDLOCK_ACCESS msg=audit(1732186800.372:45): domain=195ba459b blockers=ptrace opid=1 ocomm="systemd" type=SYSCALL msg=audit(1732186800.372:45): arch=c000003e syscall=101 success=no [...] pid=300 auid=0 Log domain deletion with the "deallocated" state when a domain was previously logged. This makes it possible for log parsers to free potential resources when a domain ID will never show again. The number of denied access requests is useful to easily check how many access requests a domain blocked and potentially if some of them are missing in logs because of audit rate limiting, audit rules, or Landlock log configuration flags (see following commit). Audit event sample for a deletion of a domain that denied something: type=LANDLOCK_DOMAIN msg=audit(1732186800.393:46): domain=195ba459b status=deallocated denials=2 Cc: Günther Noack <gnoack@google.com> Acked-by: Paul Moore <paul@paul-moore.com> Link: https://lore.kernel.org/r/20250320190717.2287696-11-mic@digikod.net [mic: Update comment and GFP flag for landlock_log_drop_domain()] Signed-off-by: Mickaël Salaün <mic@digikod.net>
2025-03-20 20:06:59 +01:00
landlock: Add AUDIT_LANDLOCK_ACCESS and log ptrace denials Add a new AUDIT_LANDLOCK_ACCESS record type dedicated to an access request denied by a Landlock domain. AUDIT_LANDLOCK_ACCESS indicates that something unexpected happened. For now, only denied access are logged, which means that any AUDIT_LANDLOCK_ACCESS record is always followed by a SYSCALL record with "success=no". However, log parsers should check this syscall property because this is the only sign that a request was denied. Indeed, we could have "success=yes" if Landlock would support a "permissive" mode. We could also add a new field to AUDIT_LANDLOCK_DOMAIN for this mode (see following commit). By default, the only logged access requests are those coming from the same executed program that enforced the Landlock restriction on itself. In other words, no audit record are created for a task after it called execve(2). This is required to avoid log spam because programs may only be aware of their own restrictions, but not the inherited ones. Following commits will allow to conditionally generate AUDIT_LANDLOCK_ACCESS records according to dedicated landlock_restrict_self(2)'s flags. The AUDIT_LANDLOCK_ACCESS message contains: - the "domain" ID restricting the action on an object, - the "blockers" that are missing to allow the requested access, - a set of fields identifying the related object (e.g. task identified with "opid" and "ocomm"). The blockers are implicit restrictions (e.g. ptrace), or explicit access rights (e.g. filesystem), or explicit scopes (e.g. signal). This field contains a list of at least one element, each separated with a comma. The initial blocker is "ptrace", which describe all implicit Landlock restrictions related to ptrace (e.g. deny tracing of tasks outside a sandbox). Add audit support to ptrace_access_check and ptrace_traceme hooks. For the ptrace_access_check case, we log the current/parent domain and the child task. For the ptrace_traceme case, we log the parent domain and the current/child task. Indeed, the requester and the target are the current task, but the action would be performed by the parent task. Audit event sample: type=LANDLOCK_ACCESS msg=audit(1729738800.349:44): domain=195ba459b blockers=ptrace opid=1 ocomm="systemd" type=SYSCALL msg=audit(1729738800.349:44): arch=c000003e syscall=101 success=no [...] pid=300 auid=0 A following commit adds user documentation. Add KUnit tests to check reading of domain ID relative to layer level. The quick return for non-landlocked tasks is moved from task_ptrace() to each LSM hooks. It is not useful to inline the audit_enabled check because other computation are performed by landlock_log_denial(). Use scoped guards for RCU read-side critical sections. Cc: Günther Noack <gnoack@google.com> Acked-by: Paul Moore <paul@paul-moore.com> Link: https://lore.kernel.org/r/20250320190717.2287696-10-mic@digikod.net Signed-off-by: Mickaël Salaün <mic@digikod.net>
2025-03-20 20:06:58 +01:00
landlock: Add AUDIT_LANDLOCK_ACCESS and log ptrace denials Add a new AUDIT_LANDLOCK_ACCESS record type dedicated to an access request denied by a Landlock domain. AUDIT_LANDLOCK_ACCESS indicates that something unexpected happened. For now, only denied access are logged, which means that any AUDIT_LANDLOCK_ACCESS record is always followed by a SYSCALL record with "success=no". However, log parsers should check this syscall property because this is the only sign that a request was denied. Indeed, we could have "success=yes" if Landlock would support a "permissive" mode. We could also add a new field to AUDIT_LANDLOCK_DOMAIN for this mode (see following commit). By default, the only logged access requests are those coming from the same executed program that enforced the Landlock restriction on itself. In other words, no audit record are created for a task after it called execve(2). This is required to avoid log spam because programs may only be aware of their own restrictions, but not the inherited ones. Following commits will allow to conditionally generate AUDIT_LANDLOCK_ACCESS records according to dedicated landlock_restrict_self(2)'s flags. The AUDIT_LANDLOCK_ACCESS message contains: - the "domain" ID restricting the action on an object, - the "blockers" that are missing to allow the requested access, - a set of fields identifying the related object (e.g. task identified with "opid" and "ocomm"). The blockers are implicit restrictions (e.g. ptrace), or explicit access rights (e.g. filesystem), or explicit scopes (e.g. signal). This field contains a list of at least one element, each separated with a comma. The initial blocker is "ptrace", which describe all implicit Landlock restrictions related to ptrace (e.g. deny tracing of tasks outside a sandbox). Add audit support to ptrace_access_check and ptrace_traceme hooks. For the ptrace_access_check case, we log the current/parent domain and the child task. For the ptrace_traceme case, we log the parent domain and the current/child task. Indeed, the requester and the target are the current task, but the action would be performed by the parent task. Audit event sample: type=LANDLOCK_ACCESS msg=audit(1729738800.349:44): domain=195ba459b blockers=ptrace opid=1 ocomm="systemd" type=SYSCALL msg=audit(1729738800.349:44): arch=c000003e syscall=101 success=no [...] pid=300 auid=0 A following commit adds user documentation. Add KUnit tests to check reading of domain ID relative to layer level. The quick return for non-landlocked tasks is moved from task_ptrace() to each LSM hooks. It is not useful to inline the audit_enabled check because other computation are performed by landlock_log_denial(). Use scoped guards for RCU read-side critical sections. Cc: Günther Noack <gnoack@google.com> Acked-by: Paul Moore <paul@paul-moore.com> Link: https://lore.kernel.org/r/20250320190717.2287696-10-mic@digikod.net Signed-off-by: Mickaël Salaün <mic@digikod.net>
2025-03-20 20:06:58 +01:00