NetBSD/share/man/man4/systrace.4

339 lines
8.8 KiB
Groff

.\" $NetBSD: systrace.4,v 1.15 2007/03/07 00:41:16 dogcow Exp $
.\" $OpenBSD: systrace.4,v 1.2 2002/06/03 15:44:17 mpech Exp $
.\"
.\" Copyright (c) 2002 CubeSoft Communications, Inc.
.\" All rights reserved.
.\"
.\" Redistribution and use in source and binary forms, with or without
.\" modification, are permitted provided that the following conditions
.\" are met:
.\" 1. Redistribution of source code must retain the above copyright
.\" notice, this list of conditions and the following disclaimer.
.\" 2. Neither the name of CubeSoft Communications, nor the names of its
.\" contributors may be used to endorse or promote products derived from
.\" this software without specific prior written permission.
.\"
.\" THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR
.\" IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
.\" WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
.\" ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT,
.\" INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
.\" (INCLUDING BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
.\" SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
.\" HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT,
.\" STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING
.\" IN ANY WAY OUT OF THE USE OF THIS SOFTWARE EVEN IF ADVISED OF THE
.\" POSSIBILITY OF SUCH DAMAGE.
.\"
.Dd June 30, 2005
.Dt SYSTRACE 4
.Os
.Sh NAME
.Nm systrace
.Nd enforce and generate policies for system calls
.Sh SYNOPSIS
.Cd "options SYSTRACE"
.Sh DESCRIPTION
The
.Nm
kernel facility provides a mechanism to manipulate and enforce access
policies for system calls.
Using the
.Nm
facility it is possible to monitor and control a process's access to
the kernel through system calls.
.Pp
Access to the
.Nm
facility is provided to userland processes, such as
.Xr systrace 1 ,
through an
.Xr ioctl 2
interface on the pseudo-device
.Pa /dev/systrace .
This interface allows messages to be sent from the kernel to the
userland process to request confirmation of an access policy.
.Pp
The
.Nm
facility can assign the following policies to system calls for a
specific process:
.Bl -tag -offset indent -width "XXXXXX"
.It Dv SYSTR_POLICY_ASK
Send a message on
.Pa /dev/systrace
requesting the access policy for the system call.
.It Dv SYSTR_POLICY_PERMIT
Immediately allow the system call.
.It Dv SYSTR_POLICY_NEVER
Immediately deny the system call and return an error code.
.El
.Sh SYSTRACE MESSAGES
A
.Xr read 2
operation on the
.Nm
pseudo-device will block if there are no pending messages, or
return the following structure:
.Bd -literal
struct str_message {
int32_t msg_type;
#define SYSTR_MSG_ASK 1
#define SYSTR_MSG_RES 2
#define SYSTR_MSG_EMUL 3
#define SYSTR_MSG_CHILD 4
#define SYSTR_MSG_UGID 5
#define SYSTR_MSG_POLICYFREE 6
#define SYSTR_MSG_EXECVE 7
#define SYSTR_MSG_SCRIPTNAME 8
pid_t msg_pid;
uint16_t msg_seqnr; /* answer has to match seqnr */
int16_t msg_policy;
union {
struct str_msg_emul msg_emul;
struct str_msg_ugid msg_ugid;
struct str_msg_ask msg_ask;
struct str_msg_child msg_child;
struct str_msg_execve msg_execve;
} msg_data;
};
struct str_msg_emul {
char emul[SYSTR_EMULEN];
};
struct str_msg_ugid {
uid_t uid;
gid_t gid;
};
struct str_msg_execve {
char path[MAXPATHLEN];
};
struct str_msg_ask {
int32_t code;
int32_t argsize;
register_t args[SYSTR_MAXARGS];
register_t rval[2];
int32_t result;
};
struct str_msg_child {
pid_t new_pid;
};
.Ed
.Pp
These messages are all sent to the userland control process.
.Bl -tag -width SYSTR_MSG_XXXXXXXXXX
.It Dv SYSTR_MSG_ASK
This message is sent whenever the kernel does not have a cached simple
policy for system call number
.Ar code
within the currently set emulation.
.It Dv SYSTR_MSG_RES
This message is sent whenever a system call is flagged with
.Dv SYSTR_FLAGS_RESULT .
.It Dv SYSTR_MSG_EMUL
This message is sent whenever a process gains a child.
.It Dv SYSTR_MSG_UGID
This message is sent whenever the effective UID or GID has changed
during the execution of a system call.
.It Dv SYSTR_MSG_POLICYFREE
This message is sent whenever the kernel frees the policy identified by
.Va msg_policy .
.It Dv SYSTR_MSG_EXECVE
This message is sent whenever, before a call to
.Xr execve 2
a process is privileged (technically, the process has the
,Dv P_SUGID
flag set),
but after the call these privileges have been dropped.
The new image name is specified in the
.Ar path
argument.
.El
.Sh IOCTL INTERFACE
The
.Nm
facility supports the following
.Xr ioctl 2
operations:
.Bl -tag -width "XXXXXX"
.It Dv STRIOCATTACH Fa "pid_t"
Attach to the process with the specified process ID.
This operation will fail under the following conditions:
.Pp
.Bl -enum -offset indent -compact -width 2n
.It
The process is trying to attach to itself.
.It
The process is a system process.
.It
The process is being traced already.
.It
You do not own the process and you are not root.
.It
The process is
.Xr init 8 ,
and the
kernel was not compiled with
.Cd option INSECURE .
.El
.It Dv STRIOCDETACH Fa "pid_t"
Wake up the process if it is waiting for an answer, and detach from
it.
.It Dv STRIOCANSWER Fa "struct systrace_answer"
Notify the
.Nm
facility in response to a
.Dv SYSTR_MSG_ASK
message what to do with a system call that was assigned a policy of
.Dv SYSTR_POLICY_ASK .
.Bd -literal
struct systrace_answer {
pid_t stra_pid; /* PID of process being traced */
uint16_t stra_seqnr;
int16_t reserved;
uid_t stra_seteuid;
gid_t stra_setegid;
int32_t stra_policy; /* Policy to assign */
int32_t stra_error; /* Return value of denied syscall
(will return EPERM if zero) */
int32_t stra_flags;
#define SYSTR_FLAGS_RESULT 0x0001 /* Report syscall result */
#define SYSTR_FLAGS_SETEUID 0x002
#define SYSTR_FLAGS_SETEGID 0x004
};
.Ed
.Pp
Valid return values for
.Em stra_policy
are
.Dv SYSTR_POLICY_PERMIT ,
.Dv SYSTR_POLICY_ASK ,
and
.Dv SYSTR_POLICY_NEVER .
.It Dv STRIOCIO Fa "struct systrace_io"
Copy data in/out of the process being traced.
.Bd -literal
struct systrace_io {
pid_t strio_pid; /* PID of process being traced */
int32_t strio_ops;
#define SYSTR_READ 1
#define SYSTR_WRITE 2
void *strio_offs;
void *strio_addr;
size_t strio_len;
};
.Ed
.It Dv STRIOCPOLICY Fa "struct systrace_policy"
Manipulate the set of policies.
.Bd -literal
struct systrace_policy {
int strp_op;
#define SYSTR_POLICY_NEW 1 /* Allocate a new policy */
#define SYSTR_POLICY_ASSIGN 2 /* Assign policy to process */
#define SYSTR_POLICY_MODIFY 3 /* Modify an entry */
int32_t strp_num;
union {
struct {
int16_t code;
#define SYSTR_POLICY_ASK 0
#define SYSTR_POLICY_PERMIT 1
#define SYSTR_POLICY_NEVER 2
int16_t policy;
} assign;
pid_t pid;
int32_t maxents;
} strp_data;
#define strp_pid strp_data.pid
#define strp_maxents strp_data.maxents
#define strp_code strp_data.assign.code
#define strp_policy strp_data.assign.policy
};
.Ed
.Pp
The
.Dv SYSTR_POLICY_NEW
operation allocates a new policy with all entries initialized to
.Dv SYSTR_POLICY_ASK ,
and returns the new policy number into
.Va strp_num .
The
.Dv SYSTR_POLICY_ASSIGN
operation attaches the policy identified by
.Va strp_num
to
.Va strp_pid ,
with a maximum of
.Va strp_maxents
entries.
The
.Dv SYSTR_POLICY_MODIFY
operation changes the entry indexed by
.Va strp_code
to
.Va strp_policy .
.It Dv STRIOCGETCWD Fa "pid_t"
Set the working directory of the current process to that of the
specified process.
.It Dv STRIOCRESCWD
Restore the working directory of the current process.
.It Dv STRIOCREPORT Fa "pid_t *"
Report the current emulation a process is using inside the
.Vt msg_emul
structure.
.It Dv STRIOCREPLACE Fa "struct systrace_replace *"
Arrange for system call arguments to be replaced by arguments supplied
by the monitoring process.
.Bd -literal
struct systrace_replace {
pid_t strr_pid;
uint16_t strr_seqnr;
int16_t reserved;
int32_t strr_nrepl; /* # of arguments to replace */
void *strr_base; /* Base user memory */
size_t strr_len; /* Length of memory */
int32_t strr_argind[SYSTR_MAXARGS]; /* Argument indexes */
size_t strr_off[SYSTR_MAXARGS]; /* Argumen offsets */
size_t strr_offlen[SYSTR_MAXARGS]; /* Argument sizes */
int32_t strr_flags[SYSTR_MAXARGS];
};
.Ed
.It Dv STRIOCSCRIPTNAME Fa "struct systrace_scriptname *"
Set the path of executed scripts to
.Va sn_scriptname .
.Bd -literal
struct systrace_scriptname {
pid_t sn_pid;
char sn_scriptname[MAXPATHLEN];
};
.Ed
.El
.Sh FILES
.Bl -tag -width "/dev/systrace" -compact
.It Pa /dev/systrace
system call tracing facility
.El
.Sh SEE ALSO
.Xr systrace 1 ,
.Xr ioctl 2 ,
.Xr read 2 ,
.Xr options 4 ,
.Xr init 8
.Sh HISTORY
The
.Nm
facility first appeared in
.Ox 3.2 ,
and then in
.Nx 2.0 .
.\" .Sh BUGS
.Sh CAVEATS
When creating new policies, if
.Va strp_maxents
is not large enough to accommodate any system calls needed for
fundamental process operations, the traced process will block
forever.