1SECCOMP(2) Linux Programmer's Manual SECCOMP(2)
2
6 seccomp - operate on Secure Computing state of the process
7
9 #include <linux/seccomp.h>
10 #include <linux/filter.h>
11 #include <linux/audit.h>
12 #include <linux/signal.h>
13 #include <sys/ptrace.h>
14 int seccomp(unsigned int operation, unsigned int flags, void *args);
16
18 The seccomp() system call operates on the Secure Computing (seccomp)
19 state of the calling process.
20 Currently, Linux supports the following operation values:
22 SECCOMP_SET_MODE_STRICT
24 The only system calls that the calling thread is permitted to
25 make are read(2), write(2), _exit(2) (but not exit_group(2)),
26 and sigreturn(2). Other system calls result in the delivery of
27 a SIGKILL signal. Strict secure computing mode is useful for
28 number-crunching applications that may need to execute untrusted
29 byte code, perhaps obtained by reading from a pipe or socket.
30 Note that although the calling thread can no longer call sig‐
32 procmask(2), it can use sigreturn(2) to block all signals apart
33 from SIGKILL and SIGSTOP. This means that alarm(2) (for exam‐
34 ple) is not sufficient for restricting the process's execution
35 time. Instead, to reliably terminate the process, SIGKILL must
36 be used. This can be done by using timer_create(2) with
37 SIGEV_SIGNAL and sigev_signo set to SIGKILL, or by using setr‐
38 limit(2) to set the hard limit for RLIMIT_CPU.
39 This operation is available only if the kernel is configured
41 with CONFIG_SECCOMP enabled.
42 The value of flags must be 0, and args must be NULL.
44 This operation is functionally identical to the call:
46 prctl(PR_SET_SECCOMP, SECCOMP_MODE_STRICT);
48 SECCOMP_SET_MODE_FILTER
50 The system calls allowed are defined by a pointer to a Berkeley
51 Packet Filter (BPF) passed via args. This argument is a pointer
52 to a struct sock_fprog; it can be designed to filter arbitrary
53 system calls and system call arguments. If the filter is
54 invalid, seccomp() fails, returning EINVAL in errno.
55 If fork(2) or clone(2) is allowed by the filter, any child pro‐
57 cesses will be constrained to the same system call filters as
58 the parent. If execve(2) is allowed, the existing filters will
59 be preserved across a call to execve(2).
60 In order to use the SECCOMP_SET_MODE_FILTER operation, either
62 the caller must have the CAP_SYS_ADMIN capability in its user
63 namespace, or the thread must already have the no_new_privs bit
64 set. If that bit was not already set by an ancestor of this
65 thread, the thread must make the following call:
66 prctl(PR_SET_NO_NEW_PRIVS, 1);
68 Otherwise, the SECCOMP_SET_MODE_FILTER operation fails and
70 returns EACCES in errno. This requirement ensures that an
71 unprivileged process cannot apply a malicious filter and then
72 invoke a set-user-ID or other privileged program using
73 execve(2), thus potentially compromising that program. (Such a
74 malicious filter might, for example, cause an attempt to use
75 setuid(2) to set the caller's user IDs to nonzero values to
76 instead return 0 without actually making the system call. Thus,
77 the program might be tricked into retaining superuser privileges
78 in circumstances where it is possible to influence it to do dan‐
79 gerous things because it did not actually drop privileges.)
80 If prctl(2) or seccomp() is allowed by the attached filter, fur‐
82 ther filters may be added. This will increase evaluation time,
83 but allows for further reduction of the attack surface during
84 execution of a thread.
85 The SECCOMP_SET_MODE_FILTER operation is available only if the
87 kernel is configured with CONFIG_SECCOMP_FILTER enabled.
88 When flags is 0, this operation is functionally identical to the
90 call:
91 prctl(PR_SET_SECCOMP, SECCOMP_MODE_FILTER, args);
93 The recognized flags are:
95 SECCOMP_FILTER_FLAG_TSYNC
97 When adding a new filter, synchronize all other threads
98 of the calling process to the same seccomp filter tree.
99 A "filter tree" is the ordered list of filters attached
100 to a thread. (Attaching identical filters in separate
101 seccomp() calls results in different filters from this
102 perspective.)
103 If any thread cannot synchronize to the same filter tree,
105 the call will not attach the new seccomp filter, and will
106 fail, returning the first thread ID found that cannot
107 synchronize. Synchronization will fail if another thread
108 in the same process is in SECCOMP_MODE_STRICT or if it
109 has attached new seccomp filters to itself, diverging
110 from the calling thread's filter tree.
111 SECCOMP_FILTER_FLAG_LOG (since Linux 4.14)
113 All filter return actions except SECCOMP_RET_ALLOW should
114 be logged. An administrator may override this filter
115 flag by preventing specific actions from being logged via
116 the /proc/sys/kernel/seccomp/actions_logged file.
117 SECCOMP_GET_ACTION_AVAIL (since Linux 4.14)
119 Test to see if an action is supported by the kernel. This oper‐
120 ation is helpful to confirm that the kernel knows of a more
121 recently added filter return action since the kernel treats all
122 unknown actions as SECCOMP_RET_KILL_PROCESS.
123 The value of flags must be 0, and args must be a pointer to an
125 unsigned 32-bit filter return action.
126 Filters
128 When adding filters via SECCOMP_SET_MODE_FILTER, args points to a fil‐
129 ter program:
130 struct sock_fprog {
132 unsigned short len; /* Number of BPF instructions */
133 struct sock_filter *filter; /* Pointer to array of
134 BPF instructions */
135 };
136 Each program must contain one or more BPF instructions:
138 struct sock_filter { /* Filter block */
140 __u16 code; /* Actual filter code */
141 __u8 jt; /* Jump true */
142 __u8 jf; /* Jump false */
143 __u32 k; /* Generic multiuse field */
144 };
145 When executing the instructions, the BPF program operates on the system
147 call information made available (i.e., use the BPF_ABS addressing mode)
148 as a (read-only) buffer of the following form:
149 struct seccomp_data {
151 int nr; /* System call number */
152 __u32 arch; /* AUDIT_ARCH_* value
153 (see <linux/audit.h>) */
154 __u64 instruction_pointer; /* CPU instruction pointer */
155 __u64 args[6]; /* Up to 6 system call arguments */
156 };
157 Because numbering of system calls varies between architectures and some
159 architectures (e.g., x86-64) allow user-space code to use the calling
160 conventions of multiple architectures (and the convention being used
161 may vary over the life of a process that uses execve(2) to execute
162 binaries that employ the different conventions), it is usually neces‐
163 sary to verify the value of the arch field.
164 It is strongly recommended to use a whitelisting approach whenever pos‐
166 sible because such an approach is more robust and simple. A blacklist
167 will have to be updated whenever a potentially dangerous system call is
168 added (or a dangerous flag or option if those are blacklisted), and it
169 is often possible to alter the representation of a value without alter‐
170 ing its meaning, leading to a blacklist bypass. See also Caveats
171 below.
172 The arch field is not unique for all calling conventions. The x86-64
174 ABI and the x32 ABI both use AUDIT_ARCH_X86_64 as arch, and they run on
175 the same processors. Instead, the mask __X32_SYSCALL_BIT is used on
176 the system call number to tell the two ABIs apart.
177 This means that in order to create a seccomp-based blacklist for system
179 calls performed through the x86-64 ABI, it is necessary to not only
180 check that arch equals AUDIT_ARCH_X86_64, but also to explicitly reject
181 all system calls that contain __X32_SYSCALL_BIT in nr.
182 The instruction_pointer field provides the address of the machine-lan‐
184 guage instruction that performed the system call. This might be useful
185 in conjunction with the use of /proc/[pid]/maps to perform checks based
186 on which region (mapping) of the program made the system call. (Proba‐
187 bly, it is wise to lock down the mmap(2) and mprotect(2) system calls
188 to prevent the program from subverting such checks.)
189 When checking values from args against a blacklist, keep in mind that
191 arguments are often silently truncated before being processed, but
192 after the seccomp check. For example, this happens if the i386 ABI is
193 used on an x86-64 kernel: although the kernel will normally not look
194 beyond the 32 lowest bits of the arguments, the values of the full
195 64-bit registers will be present in the seccomp data. A less surpris‐
196 ing example is that if the x86-64 ABI is used to perform a system call
197 that takes an argument of type int, the more-significant half of the
198 argument register is ignored by the system call, but visible in the
199 seccomp data.
200 A seccomp filter returns a 32-bit value consisting of two parts: the
202 most significant 16 bits (corresponding to the mask defined by the con‐
203 stant SECCOMP_RET_ACTION_FULL) contain one of the "action" values
204 listed below; the least significant 16-bits (defined by the constant
205 SECCOMP_RET_DATA) are "data" to be associated with this return value.
206 If multiple filters exist, they are all executed, in reverse order of
208 their addition to the filter tree—that is, the most recently installed
209 filter is executed first. (Note that all filters will be called even
210 if one of the earlier filters returns SECCOMP_RET_KILL. This is done
211 to simplify the kernel code and to provide a tiny speed-up in the exe‐
212 cution of sets of filters by avoiding a check for this uncommon case.)
213 The return value for the evaluation of a given system call is the
214 first-seen action value of highest precedence (along with its accompa‐
215 nying data) returned by execution of all of the filters.
216 In decreasing order of precedence, the action values that may be
218 returned by a seccomp filter are:
219 SECCOMP_RET_KILL_PROCESS (since Linux 4.14)
221 This value results in immediate termination of the process, with
222 a core dump. The system call is not executed. By contrast with
223 SECCOMP_RET_KILL_THREAD below, all threads in the thread group
224 are terminated. (For a discussion of thread groups, see the
225 description of the CLONE_THREAD flag in clone(2).)
226 The process terminates as though killed by a SIGSYS signal.
228 Even if a signal handler has been registered for SIGSYS, the
229 handler will be ignored in this case and the process always ter‐
230 minates. To a parent process that is waiting on this process
231 (using waitpid(2) or similar), the returned wstatus will indi‐
232 cate that its child was terminated as though by a SIGSYS signal.
233 SECCOMP_RET_KILL_THREAD (or SECCOMP_RET_KILL)
235 This value results in immediate termination of the thread that
236 made the system call. The system call is not executed. Other
237 threads in the same thread group will continue to execute.
238 The thread terminates as though killed by a SIGSYS signal. See
240 SECCOMP_RET_KILL_PROCESS above.
241 Before Linux 4.11, any process terminated in this way would not
243 trigger a coredump (even though SIGSYS is documented in sig‐
244 nal(7) as having a default action of termination with a core
245 dump). Since Linux 4.11, a single-threaded process will dump
246 core if terminated in this way.
247 With the addition of SECCOMP_RET_KILL_PROCESS in Linux 4.14,
249 SECCOMP_RET_KILL_THREAD was added as a synonym for SEC‐
250 COMP_RET_KILL, in order to more clearly distinguish the two
251 actions.
252 SECCOMP_RET_TRAP
254 This value results in the kernel sending a thread-directed
255 SIGSYS signal to the triggering thread. (The system call is not
256 executed.) Various fields will be set in the siginfo_t struc‐
257 ture (see sigaction(2)) associated with signal:
258 * si_signo will contain SIGSYS.
260 * si_call_addr will show the address of the system call
262 instruction.
263 * si_syscall and si_arch will indicate which system call was
265 attempted.
266 * si_code will contain SYS_SECCOMP.
268 * si_errno will contain the SECCOMP_RET_DATA portion of the
270 filter return value.
271 The program counter will be as though the system call happened
273 (i.e., the program counter will not point to the system call
274 instruction). The return value register will contain an archi‐
275 tecture-dependent value; if resuming execution, set it to some‐
276 thing appropriate for the system call. (The architecture depen‐
277 dency is because replacing it with ENOSYS could overwrite some
278 useful information.)
279 SECCOMP_RET_ERRNO
281 This value results in the SECCOMP_RET_DATA portion of the fil‐
282 ter's return value being passed to user space as the errno value
283 without executing the system call.
284 SECCOMP_RET_TRACE
286 When returned, this value will cause the kernel to attempt to
287 notify a ptrace(2)-based tracer prior to executing the system
288 call. If there is no tracer present, the system call is not
289 executed and returns a failure status with errno set to ENOSYS.
290 A tracer will be notified if it requests PTRACE_O_TRACESECCOMP
292 using ptrace(PTRACE_SETOPTIONS). The tracer will be notified of
293 a PTRACE_EVENT_SECCOMP and the SECCOMP_RET_DATA portion of the
294 filter's return value will be available to the tracer via
295 PTRACE_GETEVENTMSG.
296 The tracer can skip the system call by changing the system call
298 number to -1. Alternatively, the tracer can change the system
299 call requested by changing the system call to a valid system
300 call number. If the tracer asks to skip the system call, then
301 the system call will appear to return the value that the tracer
302 puts in the return value register.
303 Before kernel 4.8, the seccomp check will not be run again after
305 the tracer is notified. (This means that, on older kernels,
306 seccomp-based sandboxes must not allow use of ptrace(2)—even of
307 other sandboxed processes—without extreme care; ptracers can use
308 this mechanism to escape from the seccomp sandbox.)
309 SECCOMP_RET_LOG (since Linux 4.14)
311 This value results in the system call being executed after the
312 filter return action is logged. An administrator may override
313 the logging of this action via the /proc/sys/kernel/sec‐
314 comp/actions_logged file.
315 SECCOMP_RET_ALLOW
317 This value results in the system call being executed.
318 If an action value other than one of the above is specified, then the
320 filter action is treated as either SECCOMP_RET_KILL_PROCESS (since
321 Linux 4.14) or SECCOMP_RET_KILL_THREAD (in Linux 4.13 and earlier).
322 /proc interfaces
324 The files in the directory /proc/sys/kernel/seccomp provide additional
325 seccomp information and configuration:
326 actions_avail (since Linux 4.14)
328 A read-only ordered list of seccomp filter return actions in
329 string form. The ordering, from left-to-right, is in decreasing
330 order of precedence. The list represents the set of seccomp
331 filter return actions supported by the kernel.
332 actions_logged (since Linux 4.14)
334 A read-write ordered list of seccomp filter return actions that
335 are allowed to be logged. Writes to the file do not need to be
336 in ordered form but reads from the file will be ordered in the
337 same way as the actions_avail file.
338 It is important to note that the value of actions_logged does
340 not prevent certain filter return actions from being logged when
341 the audit subsystem is configured to audit a task. If the
342 action is not found in the actions_logged file, the final deci‐
343 sion on whether to audit the action for that task is ultimately
344 left up to the audit subsystem to decide for all filter return
345 actions other than SECCOMP_RET_ALLOW.
346 The "allow" string is not accepted in the actions_logged file as
348 it is not possible to log SECCOMP_RET_ALLOW actions. Attempting
349 to write "allow" to the file will fail with the error EINVAL.
350 Audit logging of seccomp actions
352 Since Linux 4.14, the kernel provides the facility to log the actions
353 returned by seccomp filters in the audit log. The kernel makes the
354 decision to log an action based on the action type, whether or not the
355 action is present in the actions_logged file, and whether kernel audit‐
356 ing is enabled (e.g., via the kernel boot option audit=1). The rules
357 are as follows:
358 * If the action is SECCOMP_RET_ALLOW, the action is not logged.
360 * Otherwise, if the action is either SECCOMP_RET_KILL_PROCESS or SEC‐
362 COMP_RET_KILL_THREAD, and that action appears in the actions_logged
363 file, the action is logged.
364 * Otherwise, if the filter has requested logging (the SECCOMP_FIL‐
366 TER_FLAG_LOG flag) and the action appears in the actions_logged
367 file, the action is logged.
368 * Otherwise, if kernel auditing is enabled and the process is being
370 audited (autrace(8)), the action is logged.
371 * Otherwise, the action is not logged.
373
375 On success, seccomp() returns 0. On error, if SECCOMP_FIL‐
376 TER_FLAG_TSYNC was used, the return value is the ID of the thread that
377 caused the synchronization failure. (This ID is a kernel thread ID of
378 the type returned by clone(2) and gettid(2).) On other errors, -1 is
379 returned, and errno is set to indicate the cause of the error.
380
382 seccomp() can fail for the following reasons:
383 EACCESS
385 The caller did not have the CAP_SYS_ADMIN capability in its user
386 namespace, or had not set no_new_privs before using SEC‐
387 COMP_SET_MODE_FILTER.
388 EFAULT args was not a valid address.
390 EINVAL operation is unknown or is not supported by this kernel version
392 or configuration.
393 EINVAL The specified flags are invalid for the given operation.
395 EINVAL operation included BPF_ABS, but the specified offset was not
397 aligned to a 32-bit boundary or exceeded sizeof(struct sec‐
398 comp_data).
399 EINVAL A secure computing mode has already been set, and operation dif‐
401 fers from the existing setting.
402 EINVAL operation specified SECCOMP_SET_MODE_FILTER, but the filter pro‐
404 gram pointed to by args was not valid or the length of the fil‐
405 ter program was zero or exceeded BPF_MAXINSNS (4096) instruc‐
406 tions.
407 ENOMEM Out of memory.
409 ENOMEM The total length of all filter programs attached to the calling
411 thread would exceed MAX_INSNS_PER_PATH [22m(32768) instructions.
412 Note that for the purposes of calculating this limit, each
413 already existing filter program incurs an overhead penalty of 4
414 instructions.
415 EOPNOTSUPP
417 operation specified SECCOMP_GET_ACTION_AVAIL, but the kernel
418 does not support the filter return action specified by args.
419 ESRCH Another thread caused a failure during thread sync, but its ID
421 could not be determined.
422
424 The seccomp() system call first appeared in Linux 3.17.
425
427 The seccomp() system call is a nonstandard Linux extension.
428
430 Rather than hand-coding seccomp filters as shown in the example below,
431 you may prefer to employ the libseccomp library, which provides a
432 front-end for generating seccomp filters.
433 The Seccomp field of the /proc/[pid]/status file provides a method of
435 viewing the seccomp mode of a process; see proc(5).
436 seccomp() provides a superset of the functionality provided by the
438 prctl(2) PR_SET_SECCOMP operation (which does not support flags).
439 Since Linux 4.4, the ptrace(2) PTRACE_SECCOMP_GET_FILTER operation can
441 be used to dump a process's seccomp filters.
442 Architecture support for seccomp BPF
444 Architecture support for seccomp BPF filtering is available on the fol‐
445 lowing architectures:
446 * x86-64, i386, x32 (since Linux 3.5)
448 * ARM (since Linux 3.8)
449 * s390 (since Linux 3.8)
450 * MIPS (since Linux 3.16)
451 * ARM-64 (since Linux 3.19)
452 * PowerPC (since Linux 4.3)
453 * Tile (since Linux 4.3)
454 * PA-RISC (since Linux 4.6)
455 Caveats
457 There are various subtleties to consider when applying seccomp filters
458 to a program, including the following:
459 * Some traditional system calls have user-space implementations in the
461 vdso(7) on many architectures. Notable examples include clock_get‐
462 time(2), gettimeofday(2), and time(2). On such architectures, sec‐
463 comp filtering for these system calls will have no effect. (How‐
464 ever, there are cases where the vdso(7) implementations may fall
465 back to invoking the true system call, in which case seccomp filters
466 would see the system call.)
467 * Seccomp filtering is based on system call numbers. However, appli‐
469 cations typically do not directly invoke system calls, but instead
470 call wrapper functions in the C library which in turn invoke the
471 system calls. Consequently, one must be aware of the following:
472 · The glibc wrappers for some traditional system calls may actually
474 employ system calls with different names in the kernel. For
475 example, the exit(2) wrapper function actually employs the
476 exit_group(2) system call, and the fork(2) wrapper function actu‐
477 ally calls clone(2).
478 · The behavior of wrapper functions may vary across architectures,
480 according to the range of system calls provided on those archi‐
481 tectures. In other words, the same wrapper function may invoke
482 different system calls on different architectures.
483 · Finally, the behavior of wrapper functions can change across
485 glibc versions. For example, in older versions, the glibc wrap‐
486 per function for open(2) invoked the system call of the same
487 name, but starting in glibc 2.26, the implementation switched to
488 calling openat(2) on all architectures.
489 The consequence of the above points is that it may be necessary to fil‐
491 ter for a system call other than might be expected. Various manual
492 pages in Section 2 provide helpful details about the differences
493 between wrapper functions and the underlying system calls in subsec‐
494 tions entitled C library/kernel differences.
495 Furthermore, note that the application of seccomp filters even risks
497 causing bugs in an application, when the filters cause unexpected fail‐
498 ures for legitimate operations that the application might need to per‐
499 form. Such bugs may not easily be discovered when testing the seccomp
500 filters if the bugs occur in rarely used application code paths.
501 Seccomp-specific BPF details
503 Note the following BPF details specific to seccomp filters:
504 * The BPF_H and BPF_B size modifiers are not supported: all operations
506 must load and store (4-byte) words (BPF_W).
507 * To access the contents of the seccomp_data buffer, use the BPF_ABS
509 addressing mode modifier.
510 * The BPF_LEN addressing mode modifier yields an immediate mode oper‐
512 and whose value is the size of the seccomp_data buffer.
513
515 The program below accepts four or more arguments. The first three
516 arguments are a system call number, a numeric architecture identifier,
517 and an error number. The program uses these values to construct a BPF
518 filter that is used at run time to perform the following checks:
519 [1] If the program is not running on the specified architecture, the
521 BPF filter causes system calls to fail with the error ENOSYS.
522 [2] If the program attempts to execute the system call with the speci‐
524 fied number, the BPF filter causes the system call to fail, with
525 errno being set to the specified error number.
526 The remaining command-line arguments specify the pathname and addi‐
528 tional arguments of a program that the example program should attempt
529 to execute using execv(3) (a library function that employs the
530 execve(2) system call). Some example runs of the program are shown
531 below.
532 First, we display the architecture that we are running on (x86-64) and
534 then construct a shell function that looks up system call numbers on
535 this architecture:
536 $ uname -m
538 x86_64
539 $ syscall_nr() {
540 cat /usr/src/linux/arch/x86/syscalls/syscall_64.tbl | \
541 awk '$2 != "x32" && $3 == "'$1'" { print $1 }'
542 }
543 When the BPF filter rejects a system call (case [2] above), it causes
545 the system call to fail with the error number specified on the command
546 line. In the experiments shown here, we'll use error number 99:
547 $ errno 99
549 EADDRNOTAVAIL 99 Cannot assign requested address
550 In the following example, we attempt to run the command whoami(1), but
552 the BPF filter rejects the execve(2) system call, so that the command
553 is not even executed:
554 $ syscall_nr execve
556 59
557 $ ./a.out
558 Usage: ./a.out <syscall_nr> <arch> <errno> <prog> [<args>]
559 Hint for <arch>: AUDIT_ARCH_I386: 0x40000003
560 AUDIT_ARCH_X86_64: 0xC000003E
561 $ ./a.out 59 0xC000003E 99 /bin/whoami
562 execv: Cannot assign requested address
563 In the next example, the BPF filter rejects the write(2) system call,
565 so that, although it is successfully started, the whoami(1) command is
566 not able to write output:
567 $ syscall_nr write
569 1
570 $ ./a.out 1 0xC000003E 99 /bin/whoami
571 In the final example, the BPF filter rejects a system call that is not
573 used by the whoami(1) command, so it is able to successfully execute
574 and produce output:
575 $ syscall_nr preadv
577 295
578 $ ./a.out 295 0xC000003E 99 /bin/whoami
579 cecilia
580 Program source
582 #include <errno.h>
583 #include <stddef.h>
584 #include <stdio.h>
585 #include <stdlib.h>
586 #include <unistd.h>
587 #include <linux/audit.h>
588 #include <linux/filter.h>
589 #include <linux/seccomp.h>
590 #include <sys/prctl.h>
591 #define X32_SYSCALL_BIT 0x40000000
593 static int
595 install_filter(int syscall_nr, int t_arch, int f_errno)
596 {
597 unsigned int upper_nr_limit = 0xffffffff;
598 /* Assume that AUDIT_ARCH_X86_64 means the normal x86-64 ABI
600 (in the x32 ABI, all system calls have bit 30 set in the
601 'nr' field, meaning the numbers are >= X32_SYSCALL_BIT) */
602 if (t_arch == AUDIT_ARCH_X86_64)
603 upper_nr_limit = X32_SYSCALL_BIT - 1;
604 struct sock_filter filter[] = {
606 /* [0] Load architecture from 'seccomp_data' buffer into
607 accumulator */
608 BPF_STMT(BPF_LD | BPF_W | BPF_ABS,
609 (offsetof(struct seccomp_data, arch))),
610 /* [1] Jump forward 5 instructions if architecture does not
612 match 't_arch' */
613 BPF_JUMP(BPF_JMP | BPF_JEQ | BPF_K, t_arch, 0, 5),
614 /* [2] Load system call number from 'seccomp_data' buffer into
616 accumulator */
617 BPF_STMT(BPF_LD | BPF_W | BPF_ABS,
618 (offsetof(struct seccomp_data, nr))),
619 /* [3] Check ABI - only needed for x86-64 in blacklist use
621 cases. Use BPF_JGT instead of checking against the bit
622 mask to avoid having to reload the syscall number. */
623 BPF_JUMP(BPF_JMP | BPF_JGT | BPF_K, upper_nr_limit, 3, 0),
624 /* [4] Jump forward 1 instruction if system call number
626 does not match 'syscall_nr' */
627 BPF_JUMP(BPF_JMP | BPF_JEQ | BPF_K, syscall_nr, 0, 1),
628 /* [5] Matching architecture and system call: don't execute
630 the system call, and return 'f_errno' in 'errno' */
631 BPF_STMT(BPF_RET | BPF_K,
632 SECCOMP_RET_ERRNO | (f_errno & SECCOMP_RET_DATA)),
633 /* [6] Destination of system call number mismatch: allow other
635 system calls */
636 BPF_STMT(BPF_RET | BPF_K, SECCOMP_RET_ALLOW),
637 /* [7] Destination of architecture mismatch: kill task */
639 BPF_STMT(BPF_RET | BPF_K, SECCOMP_RET_KILL),
640 };
641 struct sock_fprog prog = {
643 .len = (unsigned short) (sizeof(filter) / sizeof(filter[0])),
644 .filter = filter,
645 };
646 if (seccomp(SECCOMP_SET_MODE_FILTER, 0, &prog)) {
648 perror("seccomp");
649 return 1;
650 }
651 return 0;
653 }
654 int
656 main(int argc, char **argv)
657 {
658 if (argc < 5) {
659 fprintf(stderr, "Usage: "
660 "%s <syscall_nr> <arch> <errno> <prog> [<args>]\n"
661 "Hint for <arch>: AUDIT_ARCH_I386: 0x%X\n"
662 " AUDIT_ARCH_X86_64: 0x%X\n"
663 "\n", argv[0], AUDIT_ARCH_I386, AUDIT_ARCH_X86_64);
664 exit(EXIT_FAILURE);
665 }
666 if (prctl(PR_SET_NO_NEW_PRIVS, 1, 0, 0, 0)) {
668 perror("prctl");
669 exit(EXIT_FAILURE);
670 }
671 if (install_filter(strtol(argv[1], NULL, 0),
673 strtol(argv[2], NULL, 0),
674 strtol(argv[3], NULL, 0)))
675 exit(EXIT_FAILURE);
676 execv(argv[4], &argv[4]);
678 perror("execv");
679 exit(EXIT_FAILURE);
680 }
681
683 strace(1), bpf(2), prctl(2), ptrace(2), sigaction(2), proc(5), sig‐
684 nal(7), socket(7)
685 Various pages from the libseccomp library, including:
687 scmp_sys_resolver(1), seccomp_init(3), seccomp_load(3), sec‐
688 comp_rule_add(3), and seccomp_export_bpf(3).
689 The kernel source files Documentation/networking/filter.txt and Docu‐
691 mentation/userspace-api/seccomp_filter.rst (or Documentation/prctl/sec‐
692 comp_filter.txt before Linux 4.13).
693 McCanne, S. and Jacobson, V. (1992) The BSD Packet Filter: A New Archi‐
695 tecture for User-level Packet Capture, Proceedings of the USENIX Winter
696 1993 Conference ⟨http://www.tcpdump.org/papers/bpf-usenix93.pdf⟩
697
699 This page is part of release 4.16 of the Linux man-pages project. A
700 description of the project, information about reporting bugs, and the
701 latest version of this page, can be found at
702 https://www.kernel.org/doc/man-pages/.
703Linux 2018-02-02 SECCOMP(2)