Debian Codegraph @

upstream/debian-unstable/man7/bpf-helpers.7

raw
1 .\" Man page generated from reStructuredText
2 .\" by the Docutils 0.22.4 manpage writer.
3 .
4 .
5 .nr rst2man-indent-level 0
6 .
7 .de1 rstReportMargin
8 \\$1 \\n[an-margin]
9 level \\n[rst2man-indent-level]
10 level margin: \\n[rst2man-indent\\n[rst2man-indent-level]]
11 -
12 \\n[rst2man-indent0]
13 \\n[rst2man-indent1]
14 \\n[rst2man-indent2]
15 ..
16 .de1 INDENT
17 .\" .rstReportMargin pre:
18 . RS \\$1
19 . nr rst2man-indent\\n[rst2man-indent-level] \\n[an-margin]
20 . nr rst2man-indent-level +1
21 .\" .rstReportMargin post:
22 ..
23 .de UNINDENT
24 . RE
25 .\" indent \\n[an-margin]
26 .\" old: \\n[rst2man-indent\\n[rst2man-indent-level]]
27 .nr rst2man-indent-level -1
28 .\" new: \\n[rst2man-indent\\n[rst2man-indent-level]]
29 .in \\n[rst2man-indent\\n[rst2man-indent-level]]u
30 ..
31 .TH "BPF-HELPERS" "7" "2025-10-27" "Linux v7.0"
32 .SH Name
33 BPF-HELPERS \- list of eBPF helper functions
34 .\" Copyright (C) All BPF authors and contributors from 2014 to present.
35 .
36 .\" See git log include/uapi/linux/bpf.h in kernel tree for details.
37 .
38 .\"
39 .
40 .\" SPDX-License-Identifier: Linux-man-pages-copyleft
41 .
42 .\"
43 .
44 .\" Please do not edit this file. It was generated from the documentation
45 .
46 .\" located in file include/uapi/linux/bpf.h of the Linux kernel sources
47 .
48 .\" (helpers description), and from scripts/bpf_doc.py in the same
49 .
50 .\" repository (header and footer).
51 .
52 .SH DESCRIPTION
53 .sp
54 The extended Berkeley Packet Filter (eBPF) subsystem consists in programs
55 written in a pseudo\-assembly language, then attached to one of the several
56 kernel hooks and run in reaction of specific events. This framework differs
57 from the older, \(dqclassic\(dq BPF (or \(dqcBPF\(dq) in several aspects, one of them being
58 the ability to call special functions (or \(dqhelpers\(dq) from within a program.
59 These functions are restricted to a white\-list of helpers defined in the
60 kernel.
61 .sp
62 These helpers are used by eBPF programs to interact with the system, or with
63 the context in which they work. For instance, they can be used to print
64 debugging messages, to get the time since the system was booted, to interact
65 with eBPF maps, or to manipulate network packets. Since there are several eBPF
66 program types, and that they do not run in the same context, each program type
67 can only call a subset of those helpers.
68 .sp
69 Due to eBPF conventions, a helper can not have more than five arguments.
70 .sp
71 Internally, eBPF programs call directly into the compiled helper functions
72 without requiring any foreign\-function interface. As a result, calling helpers
73 introduces no overhead, thus offering excellent performance.
74 .sp
75 This document is an attempt to list and document the helpers available to eBPF
76 developers. They are sorted by chronological order (the oldest helpers in the
77 kernel at the top).
78 .SH HELPERS
79 .INDENT 0.0
80 .TP
81 .B \fBvoid *bpf_map_lookup_elem(struct bpf_map *\fP\fImap\fP\fB, const void *\fP\fIkey\fP\fB)\fP
82 .INDENT 7.0
83 .TP
84 .B Description
85 Perform a lookup in \fImap\fP for an entry associated to \fIkey\fP\&.
86 .TP
87 .B Return
88 Map value associated to \fIkey\fP, or \fBNULL\fP if no entry was
89 found.
90 .UNINDENT
91 .TP
92 .B \fBlong bpf_map_update_elem(struct bpf_map *\fP\fImap\fP\fB, const void *\fP\fIkey\fP\fB, const void *\fP\fIvalue\fP\fB, u64\fP \fIflags\fP\fB)\fP
93 .INDENT 7.0
94 .TP
95 .B Description
96 Add or update the value of the entry associated to \fIkey\fP in
97 \fImap\fP with \fIvalue\fP\&. \fIflags\fP is one of:
98 .INDENT 7.0
99 .TP
100 .B \fBBPF_NOEXIST\fP
101 The entry for \fIkey\fP must not exist in the map.
102 .TP
103 .B \fBBPF_EXIST\fP
104 The entry for \fIkey\fP must already exist in the map.
105 .TP
106 .B \fBBPF_ANY\fP
107 No condition on the existence of the entry for \fIkey\fP\&.
108 .UNINDENT
109 .sp
110 Flag value \fBBPF_NOEXIST\fP cannot be used for maps of types
111 \fBBPF_MAP_TYPE_ARRAY\fP or \fBBPF_MAP_TYPE_PERCPU_ARRAY\fP (all
112 elements always exist), the helper would return an error.
113 .TP
114 .B Return
115 0 on success, or a negative error in case of failure.
116 .UNINDENT
117 .TP
118 .B \fBlong bpf_map_delete_elem(struct bpf_map *\fP\fImap\fP\fB, const void *\fP\fIkey\fP\fB)\fP
119 .INDENT 7.0
120 .TP
121 .B Description
122 Delete entry with \fIkey\fP from \fImap\fP\&.
123 .TP
124 .B Return
125 0 on success, or a negative error in case of failure.
126 .UNINDENT
127 .TP
128 .B \fBlong bpf_probe_read(void *\fP\fIdst\fP\fB, u32\fP \fIsize\fP\fB, const void *\fP\fIunsafe_ptr\fP\fB)\fP
129 .INDENT 7.0
130 .TP
131 .B Description
132 For tracing programs, safely attempt to read \fIsize\fP bytes from
133 kernel space address \fIunsafe_ptr\fP and store the data in \fIdst\fP\&.
134 .sp
135 Generally, use \fBbpf_probe_read_user\fP() or
136 \fBbpf_probe_read_kernel\fP() instead.
137 .TP
138 .B Return
139 0 on success, or a negative error in case of failure.
140 .UNINDENT
141 .TP
142 .B \fBu64 bpf_ktime_get_ns(void)\fP
143 .INDENT 7.0
144 .TP
145 .B Description
146 Return the time elapsed since system boot, in nanoseconds.
147 Does not include time the system was suspended.
148 See: \fBclock_gettime\fP(\fBCLOCK_MONOTONIC\fP)
149 .TP
150 .B Return
151 Current \fIktime\fP\&.
152 .UNINDENT
153 .TP
154 .B \fBlong bpf_trace_printk(const char *\fP\fIfmt\fP\fB, u32\fP \fIfmt_size\fP\fB, ...)\fP
155 .INDENT 7.0
156 .TP
157 .B Description
158 This helper is a \(dqprintk()\-like\(dq facility for debugging. It
159 prints a message defined by format \fIfmt\fP (of size \fIfmt_size\fP)
160 to file \fI/sys/kernel/tracing/trace\fP from TraceFS, if
161 available. It can take up to three additional \fBu64\fP
162 arguments (as an eBPF helpers, the total number of arguments is
163 limited to five).
164 .sp
165 Each time the helper is called, it appends a line to the trace.
166 Lines are discarded while \fI/sys/kernel/tracing/trace\fP is
167 open, use \fI/sys/kernel/tracing/trace_pipe\fP to avoid this.
168 The format of the trace is customizable, and the exact output
169 one will get depends on the options set in
170 \fI/sys/kernel/tracing/trace_options\fP (see also the
171 \fIREADME\fP file under the same directory). However, it usually
172 defaults to something like:
173 .INDENT 7.0
174 .INDENT 3.5
175 .sp
176 .EX
177 telnet\-470 [001] .N.. 419421.045894: 0x00000001: <formatted msg>
178 .EE
179 .UNINDENT
180 .UNINDENT
181 .sp
182 In the above:
183 .INDENT 7.0
184 .INDENT 3.5
185 .INDENT 0.0
186 .IP \(bu 2
187 \fBtelnet\fP is the name of the current task.
188 .IP \(bu 2
189 \fB470\fP is the PID of the current task.
190 .IP \(bu 2
191 \fB001\fP is the CPU number on which the task is
192 running.
193 .IP \(bu 2
194 In \fB\&.N..\fP, each character refers to a set of
195 options (whether irqs are enabled, scheduling
196 options, whether hard/softirqs are running, level of
197 preempt_disabled respectively). \fBN\fP means that
198 \fBTIF_NEED_RESCHED\fP and \fBPREEMPT_NEED_RESCHED\fP
199 are set.
200 .IP \(bu 2
201 \fB419421.045894\fP is a timestamp.
202 .IP \(bu 2
203 \fB0x00000001\fP is a fake value used by BPF for the
204 instruction pointer register.
205 .IP \(bu 2
206 \fB<formatted msg>\fP is the message formatted with
207 \fIfmt\fP\&.
208 .UNINDENT
209 .UNINDENT
210 .UNINDENT
211 .sp
212 The conversion specifiers supported by \fIfmt\fP are similar, but
213 more limited than for printk(). They are \fB%d\fP, \fB%i\fP,
214 \fB%u\fP, \fB%x\fP, \fB%ld\fP, \fB%li\fP, \fB%lu\fP, \fB%lx\fP, \fB%lld\fP,
215 \fB%lli\fP, \fB%llu\fP, \fB%llx\fP, \fB%p\fP, \fB%s\fP\&. No modifier (size
216 of field, padding with zeroes, etc.) is available, and the
217 helper will return \fB\-EINVAL\fP (but print nothing) if it
218 encounters an unknown specifier.
219 .sp
220 Also, note that \fBbpf_trace_printk\fP() is slow, and should
221 only be used for debugging purposes. For this reason, a notice
222 block (spanning several lines) is printed to kernel logs and
223 states that the helper should not be used \(dqfor production use\(dq
224 the first time this helper is used (or more precisely, when
225 \fBtrace_printk\fP() buffers are allocated). For passing values
226 to user space, perf events should be preferred.
227 .TP
228 .B Return
229 The number of bytes written to the buffer, or a negative error
230 in case of failure.
231 .UNINDENT
232 .TP
233 .B \fBu32 bpf_get_prandom_u32(void)\fP
234 .INDENT 7.0
235 .TP
236 .B Description
237 Get a pseudo\-random number.
238 .sp
239 From a security point of view, this helper uses its own
240 pseudo\-random internal state, and cannot be used to infer the
241 seed of other random functions in the kernel. However, it is
242 essential to note that the generator used by the helper is not
243 cryptographically secure.
244 .TP
245 .B Return
246 A random 32\-bit unsigned value.
247 .UNINDENT
248 .TP
249 .B \fBu32 bpf_get_smp_processor_id(void)\fP
250 .INDENT 7.0
251 .TP
252 .B Description
253 Get the SMP (symmetric multiprocessing) processor id. Note that
254 all programs run with migration disabled, which means that the
255 SMP processor id is stable during all the execution of the
256 program.
257 .TP
258 .B Return
259 The SMP id of the processor running the program.
260 .UNINDENT
261 .TP
262 .B \fBlong bpf_skb_store_bytes(struct sk_buff *\fP\fIskb\fP\fB, u32\fP \fIoffset\fP\fB, const void *\fP\fIfrom\fP\fB, u32\fP \fIlen\fP\fB, u64\fP \fIflags\fP\fB)\fP
263 .INDENT 7.0
264 .TP
265 .B Description
266 Store \fIlen\fP bytes from address \fIfrom\fP into the packet
267 associated to \fIskb\fP, at \fIoffset\fP\&. The \fIflags\fP are a combination
268 of the following values:
269 .INDENT 7.0
270 .TP
271 .B \fBBPF_F_RECOMPUTE_CSUM\fP
272 Automatically update \fIskb\fP\fB\->csum\fP after storing the
273 bytes.
274 .TP
275 .B \fBBPF_F_INVALIDATE_HASH\fP
276 Set \fIskb\fP\fB\->hash\fP, \fIskb\fP\fB\->swhash\fP and \fIskb\fP\fB\->l4hash\fP to 0.
277 .UNINDENT
278 .sp
279 A call to this helper is susceptible to change the underlying
280 packet buffer. Therefore, at load time, all checks on pointers
281 previously done by the verifier are invalidated and must be
282 performed again, if the helper is used in combination with
283 direct packet access.
284 .TP
285 .B Return
286 0 on success, or a negative error in case of failure.
287 .UNINDENT
288 .TP
289 .B \fBlong bpf_l3_csum_replace(struct sk_buff *\fP\fIskb\fP\fB, u32\fP \fIoffset\fP\fB, u64\fP \fIfrom\fP\fB, u64\fP \fIto\fP\fB, u64\fP \fIsize\fP\fB)\fP
290 .INDENT 7.0
291 .TP
292 .B Description
293 Recompute the layer 3 (e.g. IP) checksum for the packet
294 associated to \fIskb\fP\&. Computation is incremental, so the helper
295 must know the former value of the header field that was
296 modified (\fIfrom\fP), the new value of this field (\fIto\fP), and the
297 number of bytes (2 or 4) for this field, stored in \fIsize\fP\&.
298 Alternatively, it is possible to store the difference between
299 the previous and the new values of the header field in \fIto\fP, by
300 setting \fIfrom\fP and \fIsize\fP to 0. For both methods, \fIoffset\fP
301 indicates the location of the IP checksum within the packet.
302 .sp
303 This helper works in combination with \fBbpf_csum_diff\fP(),
304 which does not update the checksum in\-place, but offers more
305 flexibility and can handle sizes larger than 2 or 4 for the
306 checksum to update.
307 .sp
308 A call to this helper is susceptible to change the underlying
309 packet buffer. Therefore, at load time, all checks on pointers
310 previously done by the verifier are invalidated and must be
311 performed again, if the helper is used in combination with
312 direct packet access.
313 .TP
314 .B Return
315 0 on success, or a negative error in case of failure.
316 .UNINDENT
317 .TP
318 .B \fBlong bpf_l4_csum_replace(struct sk_buff *\fP\fIskb\fP\fB, u32\fP \fIoffset\fP\fB, u64\fP \fIfrom\fP\fB, u64\fP \fIto\fP\fB, u64\fP \fIflags\fP\fB)\fP
319 .INDENT 7.0
320 .TP
321 .B Description
322 Recompute the layer 4 (e.g. TCP, UDP or ICMP) checksum for the
323 packet associated to \fIskb\fP\&. Computation is incremental, so the
324 helper must know the former value of the header field that was
325 modified (\fIfrom\fP), the new value of this field (\fIto\fP), and the
326 number of bytes (2 or 4) for this field, stored on the lowest
327 four bits of \fIflags\fP\&. Alternatively, it is possible to store
328 the difference between the previous and the new values of the
329 header field in \fIto\fP, by setting \fIfrom\fP and the four lowest
330 bits of \fIflags\fP to 0. For both methods, \fIoffset\fP indicates the
331 location of the IP checksum within the packet. In addition to
332 the size of the field, \fIflags\fP can be added (bitwise OR) actual
333 flags. With \fBBPF_F_MARK_MANGLED_0\fP, a null checksum is left
334 untouched (unless \fBBPF_F_MARK_ENFORCE\fP is added as well), and
335 for updates resulting in a null checksum the value is set to
336 \fBCSUM_MANGLED_0\fP instead. Flag \fBBPF_F_PSEUDO_HDR\fP indicates
337 that the modified header field is part of the pseudo\-header.
338 Flag \fBBPF_F_IPV6\fP should be set for IPv6 packets.
339 .sp
340 This helper works in combination with \fBbpf_csum_diff\fP(),
341 which does not update the checksum in\-place, but offers more
342 flexibility and can handle sizes larger than 2 or 4 for the
343 checksum to update.
344 .sp
345 A call to this helper is susceptible to change the underlying
346 packet buffer. Therefore, at load time, all checks on pointers
347 previously done by the verifier are invalidated and must be
348 performed again, if the helper is used in combination with
349 direct packet access.
350 .TP
351 .B Return
352 0 on success, or a negative error in case of failure.
353 .UNINDENT
354 .TP
355 .B \fBlong bpf_tail_call(void *\fP\fIctx\fP\fB, struct bpf_map *\fP\fIprog_array_map\fP\fB, u32\fP \fIindex\fP\fB)\fP
356 .INDENT 7.0
357 .TP
358 .B Description
359 This special helper is used to trigger a \(dqtail call\(dq, or in
360 other words, to jump into another eBPF program. The same stack
361 frame is used (but values on stack and in registers for the
362 caller are not accessible to the callee). This mechanism allows
363 for program chaining, either for raising the maximum number of
364 available eBPF instructions, or to execute given programs in
365 conditional blocks. For security reasons, there is an upper
366 limit to the number of successive tail calls that can be
367 performed.
368 .sp
369 Upon call of this helper, the program attempts to jump into a
370 program referenced at index \fIindex\fP in \fIprog_array_map\fP, a
371 special map of type \fBBPF_MAP_TYPE_PROG_ARRAY\fP, and passes
372 \fIctx\fP, a pointer to the context.
373 .sp
374 If the call succeeds, the kernel immediately runs the first
375 instruction of the new program. This is not a function call,
376 and it never returns to the previous program. If the call
377 fails, then the helper has no effect, and the caller continues
378 to run its subsequent instructions. A call can fail if the
379 destination program for the jump does not exist (i.e. \fIindex\fP
380 is superior to the number of entries in \fIprog_array_map\fP), or
381 if the maximum number of tail calls has been reached for this
382 chain of programs. This limit is defined in the kernel by the
383 macro \fBMAX_TAIL_CALL_CNT\fP (not accessible to user space),
384 which is currently set to 33.
385 .TP
386 .B Return
387 0 on success, or a negative error in case of failure.
388 .UNINDENT
389 .TP
390 .B \fBlong bpf_clone_redirect(struct sk_buff *\fP\fIskb\fP\fB, u32\fP \fIifindex\fP\fB, u64\fP \fIflags\fP\fB)\fP
391 .INDENT 7.0
392 .TP
393 .B Description
394 Clone and redirect the packet associated to \fIskb\fP to another
395 net device of index \fIifindex\fP\&. Both ingress and egress
396 interfaces can be used for redirection. The \fBBPF_F_INGRESS\fP
397 value in \fIflags\fP is used to make the distinction (ingress path
398 is selected if the flag is present, egress path otherwise).
399 This is the only flag supported for now.
400 .sp
401 In comparison with \fBbpf_redirect\fP() helper,
402 \fBbpf_clone_redirect\fP() has the associated cost of
403 duplicating the packet buffer, but this can be executed out of
404 the eBPF program. Conversely, \fBbpf_redirect\fP() is more
405 efficient, but it is handled through an action code where the
406 redirection happens only after the eBPF program has returned.
407 .sp
408 A call to this helper is susceptible to change the underlying
409 packet buffer. Therefore, at load time, all checks on pointers
410 previously done by the verifier are invalidated and must be
411 performed again, if the helper is used in combination with
412 direct packet access.
413 .TP
414 .B Return
415 0 on success, or a negative error in case of failure. Positive
416 error indicates a potential drop or congestion in the target
417 device. The particular positive error codes are not defined.
418 .UNINDENT
419 .TP
420 .B \fBu64 bpf_get_current_pid_tgid(void)\fP
421 .INDENT 7.0
422 .TP
423 .B Description
424 Get the current pid and tgid.
425 .TP
426 .B Return
427 A 64\-bit integer containing the current tgid and pid, and
428 created as such:
429 \fIcurrent_task\fP\fB\->tgid << 32 |\fP
430 \fIcurrent_task\fP\fB\->pid\fP\&.
431 .UNINDENT
432 .TP
433 .B \fBu64 bpf_get_current_uid_gid(void)\fP
434 .INDENT 7.0
435 .TP
436 .B Description
437 Get the current uid and gid.
438 .TP
439 .B Return
440 A 64\-bit integer containing the current GID and UID, and
441 created as such: \fIcurrent_gid\fP \fB<< 32 |\fP \fIcurrent_uid\fP\&.
442 .UNINDENT
443 .TP
444 .B \fBlong bpf_get_current_comm(void *\fP\fIbuf\fP\fB, u32\fP \fIsize_of_buf\fP\fB)\fP
445 .INDENT 7.0
446 .TP
447 .B Description
448 Copy the \fBcomm\fP attribute of the current task into \fIbuf\fP of
449 \fIsize_of_buf\fP\&. The \fBcomm\fP attribute contains the name of
450 the executable (excluding the path) for the current task. The
451 \fIsize_of_buf\fP must be strictly positive. On success, the
452 helper makes sure that the \fIbuf\fP is NUL\-terminated. On failure,
453 it is filled with zeroes.
454 .TP
455 .B Return
456 0 on success, or a negative error in case of failure.
457 .UNINDENT
458 .TP
459 .B \fBu32 bpf_get_cgroup_classid(struct sk_buff *\fP\fIskb\fP\fB)\fP
460 .INDENT 7.0
461 .TP
462 .B Description
463 Retrieve the classid for the current task, i.e. for the net_cls
464 cgroup to which \fIskb\fP belongs.
465 .sp
466 This helper can be used on TC egress path, but not on ingress.
467 .sp
468 The net_cls cgroup provides an interface to tag network packets
469 based on a user\-provided identifier for all traffic coming from
470 the tasks belonging to the related cgroup. See also the related
471 kernel documentation, available from the Linux sources in file
472 \fIDocumentation/admin\-guide/cgroup\-v1/net_cls.rst\fP\&.
473 .sp
474 The Linux kernel has two versions for cgroups: there are
475 cgroups v1 and cgroups v2. Both are available to users, who can
476 use a mixture of them, but note that the net_cls cgroup is for
477 cgroup v1 only. This makes it incompatible with BPF programs
478 run on cgroups, which is a cgroup\-v2\-only feature (a socket can
479 only hold data for one version of cgroups at a time).
480 .sp
481 This helper is only available is the kernel was compiled with
482 the \fBCONFIG_CGROUP_NET_CLASSID\fP configuration option set to
483 \(dq\fBy\fP\(dq or to \(dq\fBm\fP\(dq.
484 .TP
485 .B Return
486 The classid, or 0 for the default unconfigured classid.
487 .UNINDENT
488 .TP
489 .B \fBlong bpf_skb_vlan_push(struct sk_buff *\fP\fIskb\fP\fB, __be16\fP \fIvlan_proto\fP\fB, u16\fP \fIvlan_tci\fP\fB)\fP
490 .INDENT 7.0
491 .TP
492 .B Description
493 Push a \fIvlan_tci\fP (VLAN tag control information) of protocol
494 \fIvlan_proto\fP to the packet associated to \fIskb\fP, then update
495 the checksum. Note that if \fIvlan_proto\fP is different from
496 \fBETH_P_8021Q\fP and \fBETH_P_8021AD\fP, it is considered to
497 be \fBETH_P_8021Q\fP\&.
498 .sp
499 A call to this helper is susceptible to change the underlying
500 packet buffer. Therefore, at load time, all checks on pointers
501 previously done by the verifier are invalidated and must be
502 performed again, if the helper is used in combination with
503 direct packet access.
504 .TP
505 .B Return
506 0 on success, or a negative error in case of failure.
507 .UNINDENT
508 .TP
509 .B \fBlong bpf_skb_vlan_pop(struct sk_buff *\fP\fIskb\fP\fB)\fP
510 .INDENT 7.0
511 .TP
512 .B Description
513 Pop a VLAN header from the packet associated to \fIskb\fP\&.
514 .sp
515 A call to this helper is susceptible to change the underlying
516 packet buffer. Therefore, at load time, all checks on pointers
517 previously done by the verifier are invalidated and must be
518 performed again, if the helper is used in combination with
519 direct packet access.
520 .TP
521 .B Return
522 0 on success, or a negative error in case of failure.
523 .UNINDENT
524 .TP
525 .B \fBlong bpf_skb_get_tunnel_key(struct sk_buff *\fP\fIskb\fP\fB, struct bpf_tunnel_key *\fP\fIkey\fP\fB, u32\fP \fIsize\fP\fB, u64\fP \fIflags\fP\fB)\fP
526 .INDENT 7.0
527 .TP
528 .B Description
529 Get tunnel metadata. This helper takes a pointer \fIkey\fP to an
530 empty \fBstruct bpf_tunnel_key\fP of \fBsize\fP, that will be
531 filled with tunnel metadata for the packet associated to \fIskb\fP\&.
532 The \fIflags\fP can be set to \fBBPF_F_TUNINFO_IPV6\fP, which
533 indicates that the tunnel is based on IPv6 protocol instead of
534 IPv4.
535 .sp
536 The \fBstruct bpf_tunnel_key\fP is an object that generalizes the
537 principal parameters used by various tunneling protocols into a
538 single struct. This way, it can be used to easily make a
539 decision based on the contents of the encapsulation header,
540 \(dqsummarized\(dq in this struct. In particular, it holds the IP
541 address of the remote end (IPv4 or IPv6, depending on the case)
542 in \fIkey\fP\fB\->remote_ipv4\fP or \fIkey\fP\fB\->remote_ipv6\fP\&. Also,
543 this struct exposes the \fIkey\fP\fB\->tunnel_id\fP, which is
544 generally mapped to a VNI (Virtual Network Identifier), making
545 it programmable together with the \fBbpf_skb_set_tunnel_key\fP() helper.
546 .sp
547 Let\(aqs imagine that the following code is part of a program
548 attached to the TC ingress interface, on one end of a GRE
549 tunnel, and is supposed to filter out all messages coming from
550 remote ends with IPv4 address other than 10.0.0.1:
551 .INDENT 7.0
552 .INDENT 3.5
553 .sp
554 .EX
555 int ret;
556 struct bpf_tunnel_key key = {};
557
558 ret = bpf_skb_get_tunnel_key(skb, &key, sizeof(key), 0);
559 if (ret < 0)
560 return TC_ACT_SHOT; // drop packet
561
562 if (key.remote_ipv4 != 0x0a000001)
563 return TC_ACT_SHOT; // drop packet
564
565 return TC_ACT_OK; // accept packet
566 .EE
567 .UNINDENT
568 .UNINDENT
569 .sp
570 This interface can also be used with all encapsulation devices
571 that can operate in \(dqcollect metadata\(dq mode: instead of having
572 one network device per specific configuration, the \(dqcollect
573 metadata\(dq mode only requires a single device where the
574 configuration can be extracted from this helper.
575 .sp
576 This can be used together with various tunnels such as VXLan,
577 Geneve, GRE or IP in IP (IPIP).
578 .TP
579 .B Return
580 0 on success, or a negative error in case of failure.
581 .UNINDENT
582 .TP
583 .B \fBlong bpf_skb_set_tunnel_key(struct sk_buff *\fP\fIskb\fP\fB, struct bpf_tunnel_key *\fP\fIkey\fP\fB, u32\fP \fIsize\fP\fB, u64\fP \fIflags\fP\fB)\fP
584 .INDENT 7.0
585 .TP
586 .B Description
587 Populate tunnel metadata for packet associated to \fIskb.\fP The
588 tunnel metadata is set to the contents of \fIkey\fP, of \fIsize\fP\&. The
589 \fIflags\fP can be set to a combination of the following values:
590 .INDENT 7.0
591 .TP
592 .B \fBBPF_F_TUNINFO_IPV6\fP
593 Indicate that the tunnel is based on IPv6 protocol
594 instead of IPv4.
595 .TP
596 .B \fBBPF_F_ZERO_CSUM_TX\fP
597 For IPv4 packets, add a flag to tunnel metadata
598 indicating that checksum computation should be skipped
599 and checksum set to zeroes.
600 .TP
601 .B \fBBPF_F_DONT_FRAGMENT\fP
602 Add a flag to tunnel metadata indicating that the
603 packet should not be fragmented.
604 .TP
605 .B \fBBPF_F_SEQ_NUMBER\fP
606 Add a flag to tunnel metadata indicating that a
607 sequence number should be added to tunnel header before
608 sending the packet. This flag was added for GRE
609 encapsulation, but might be used with other protocols
610 as well in the future.
611 .TP
612 .B \fBBPF_F_NO_TUNNEL_KEY\fP
613 Add a flag to tunnel metadata indicating that no tunnel
614 key should be set in the resulting tunnel header.
615 .UNINDENT
616 .sp
617 Here is a typical usage on the transmit path:
618 .INDENT 7.0
619 .INDENT 3.5
620 .sp
621 .EX
622 struct bpf_tunnel_key key;
623 populate key ...
624 bpf_skb_set_tunnel_key(skb, &key, sizeof(key), 0);
625 bpf_clone_redirect(skb, vxlan_dev_ifindex, 0);
626 .EE
627 .UNINDENT
628 .UNINDENT
629 .sp
630 See also the description of the \fBbpf_skb_get_tunnel_key\fP()
631 helper for additional information.
632 .TP
633 .B Return
634 0 on success, or a negative error in case of failure.
635 .UNINDENT
636 .TP
637 .B \fBu64 bpf_perf_event_read(struct bpf_map *\fP\fImap\fP\fB, u64\fP \fIflags\fP\fB)\fP
638 .INDENT 7.0
639 .TP
640 .B Description
641 Read the value of a perf event counter. This helper relies on a
642 \fImap\fP of type \fBBPF_MAP_TYPE_PERF_EVENT_ARRAY\fP\&. The nature of
643 the perf event counter is selected when \fImap\fP is updated with
644 perf event file descriptors. The \fImap\fP is an array whose size
645 is the number of available CPUs, and each cell contains a value
646 relative to one CPU. The value to retrieve is indicated by
647 \fIflags\fP, that contains the index of the CPU to look up, masked
648 with \fBBPF_F_INDEX_MASK\fP\&. Alternatively, \fIflags\fP can be set to
649 \fBBPF_F_CURRENT_CPU\fP to indicate that the value for the
650 current CPU should be retrieved.
651 .sp
652 Note that before Linux 4.13, only hardware perf event can be
653 retrieved.
654 .sp
655 Also, be aware that the newer helper
656 \fBbpf_perf_event_read_value\fP() is recommended over
657 \fBbpf_perf_event_read\fP() in general. The latter has some ABI
658 quirks where error and counter value are used as a return code
659 (which is wrong to do since ranges may overlap). This issue is
660 fixed with \fBbpf_perf_event_read_value\fP(), which at the same
661 time provides more features over the \fBbpf_perf_event_read\fP() interface. Please refer to the description of
662 \fBbpf_perf_event_read_value\fP() for details.
663 .TP
664 .B Return
665 The value of the perf event counter read from the map, or a
666 negative error code in case of failure.
667 .UNINDENT
668 .TP
669 .B \fBlong bpf_redirect(u32\fP \fIifindex\fP\fB, u64\fP \fIflags\fP\fB)\fP
670 .INDENT 7.0
671 .TP
672 .B Description
673 Redirect the packet to another net device of index \fIifindex\fP\&.
674 This helper is somewhat similar to \fBbpf_clone_redirect\fP(), except that the packet is not cloned, which provides
675 increased performance.
676 .sp
677 Except for XDP, both ingress and egress interfaces can be used
678 for redirection. The \fBBPF_F_INGRESS\fP value in \fIflags\fP is used
679 to make the distinction (ingress path is selected if the flag
680 is present, egress path otherwise). Currently, XDP only
681 supports redirection to the egress interface, and accepts no
682 flag at all.
683 .sp
684 The same effect can also be attained with the more generic
685 \fBbpf_redirect_map\fP(), which uses a BPF map to store the
686 redirect target instead of providing it directly to the helper.
687 .TP
688 .B Return
689 For XDP, the helper returns \fBXDP_REDIRECT\fP on success or
690 \fBXDP_ABORTED\fP on error. For other program types, the values
691 are \fBTC_ACT_REDIRECT\fP on success or \fBTC_ACT_SHOT\fP on
692 error.
693 .UNINDENT
694 .TP
695 .B \fBu32 bpf_get_route_realm(struct sk_buff *\fP\fIskb\fP\fB)\fP
696 .INDENT 7.0
697 .TP
698 .B Description
699 Retrieve the realm or the route, that is to say the
700 \fBtclassid\fP field of the destination for the \fIskb\fP\&. The
701 identifier retrieved is a user\-provided tag, similar to the
702 one used with the net_cls cgroup (see description for
703 \fBbpf_get_cgroup_classid\fP() helper), but here this tag is
704 held by a route (a destination entry), not by a task.
705 .sp
706 Retrieving this identifier works with the clsact TC egress hook
707 (see also \fBtc\-bpf(8)\fP), or alternatively on conventional
708 classful egress qdiscs, but not on TC ingress path. In case of
709 clsact TC egress hook, this has the advantage that, internally,
710 the destination entry has not been dropped yet in the transmit
711 path. Therefore, the destination entry does not need to be
712 artificially held via \fBnetif_keep_dst\fP() for a classful
713 qdisc until the \fIskb\fP is freed.
714 .sp
715 This helper is available only if the kernel was compiled with
716 \fBCONFIG_IP_ROUTE_CLASSID\fP configuration option.
717 .TP
718 .B Return
719 The realm of the route for the packet associated to \fIskb\fP, or 0
720 if none was found.
721 .UNINDENT
722 .TP
723 .B \fBlong bpf_perf_event_output(void *\fP\fIctx\fP\fB, struct bpf_map *\fP\fImap\fP\fB, u64\fP \fIflags\fP\fB, void *\fP\fIdata\fP\fB, u64\fP \fIsize\fP\fB)\fP
724 .INDENT 7.0
725 .TP
726 .B Description
727 Write raw \fIdata\fP blob into a special BPF perf event held by
728 \fImap\fP of type \fBBPF_MAP_TYPE_PERF_EVENT_ARRAY\fP\&. This perf
729 event must have the following attributes: \fBPERF_SAMPLE_RAW\fP
730 as \fBsample_type\fP, \fBPERF_TYPE_SOFTWARE\fP as \fBtype\fP, and
731 \fBPERF_COUNT_SW_BPF_OUTPUT\fP as \fBconfig\fP\&.
732 .sp
733 The \fIflags\fP are used to indicate the index in \fImap\fP for which
734 the value must be put, masked with \fBBPF_F_INDEX_MASK\fP\&.
735 Alternatively, \fIflags\fP can be set to \fBBPF_F_CURRENT_CPU\fP
736 to indicate that the index of the current CPU core should be
737 used.
738 .sp
739 The value to write, of \fIsize\fP, is passed through eBPF stack and
740 pointed by \fIdata\fP\&.
741 .sp
742 The context of the program \fIctx\fP needs also be passed to the
743 helper.
744 .sp
745 On user space, a program willing to read the values needs to
746 call \fBperf_event_open\fP() on the perf event (either for
747 one or for all CPUs) and to store the file descriptor into the
748 \fImap\fP\&. This must be done before the eBPF program can send data
749 into it. An example is available in file
750 \fIsamples/bpf/trace_output_user.c\fP in the Linux kernel source
751 tree (the eBPF program counterpart is in
752 \fIsamples/bpf/trace_output.bpf.c\fP).
753 .sp
754 \fBbpf_perf_event_output\fP() achieves better performance
755 than \fBbpf_trace_printk\fP() for sharing data with user
756 space, and is much better suitable for streaming data from eBPF
757 programs.
758 .sp
759 Note that this helper is not restricted to tracing use cases
760 and can be used with programs attached to TC or XDP as well,
761 where it allows for passing data to user space listeners. Data
762 can be:
763 .INDENT 7.0
764 .IP \(bu 2
765 Only custom structs,
766 .IP \(bu 2
767 Only the packet payload, or
768 .IP \(bu 2
769 A combination of both.
770 .UNINDENT
771 .TP
772 .B Return
773 0 on success, or a negative error in case of failure.
774 .UNINDENT
775 .TP
776 .B \fBlong bpf_skb_load_bytes(const void *\fP\fIskb\fP\fB, u32\fP \fIoffset\fP\fB, void *\fP\fIto\fP\fB, u32\fP \fIlen\fP\fB)\fP
777 .INDENT 7.0
778 .TP
779 .B Description
780 This helper was provided as an easy way to load data from a
781 packet. It can be used to load \fIlen\fP bytes from \fIoffset\fP from
782 the packet associated to \fIskb\fP, into the buffer pointed by
783 \fIto\fP\&.
784 .sp
785 Since Linux 4.7, usage of this helper has mostly been replaced
786 by \(dqdirect packet access\(dq, enabling packet data to be
787 manipulated with \fIskb\fP\fB\->data\fP and \fIskb\fP\fB\->data_end\fP
788 pointing respectively to the first byte of packet data and to
789 the byte after the last byte of packet data. However, it
790 remains useful if one wishes to read large quantities of data
791 at once from a packet into the eBPF stack.
792 .TP
793 .B Return
794 0 on success, or a negative error in case of failure.
795 .UNINDENT
796 .TP
797 .B \fBlong bpf_get_stackid(void *\fP\fIctx\fP\fB, struct bpf_map *\fP\fImap\fP\fB, u64\fP \fIflags\fP\fB)\fP
798 .INDENT 7.0
799 .TP
800 .B Description
801 Walk a user or a kernel stack and return its id. To achieve
802 this, the helper needs \fIctx\fP, which is a pointer to the context
803 on which the tracing program is executed, and a pointer to a
804 \fImap\fP of type \fBBPF_MAP_TYPE_STACK_TRACE\fP\&.
805 .sp
806 The last argument, \fIflags\fP, holds the number of stack frames to
807 skip (from 0 to 255), masked with
808 \fBBPF_F_SKIP_FIELD_MASK\fP\&. The next bits can be used to set
809 a combination of the following flags:
810 .INDENT 7.0
811 .TP
812 .B \fBBPF_F_USER_STACK\fP
813 Collect a user space stack instead of a kernel stack.
814 .TP
815 .B \fBBPF_F_FAST_STACK_CMP\fP
816 Compare stacks by hash only.
817 .TP
818 .B \fBBPF_F_REUSE_STACKID\fP
819 If two different stacks hash into the same \fIstackid\fP,
820 discard the old one.
821 .UNINDENT
822 .sp
823 The stack id retrieved is a 32 bit long integer handle which
824 can be further combined with other data (including other stack
825 ids) and used as a key into maps. This can be useful for
826 generating a variety of graphs (such as flame graphs or off\-cpu
827 graphs).
828 .sp
829 For walking a stack, this helper is an improvement over
830 \fBbpf_probe_read\fP(), which can be used with unrolled loops
831 but is not efficient and consumes a lot of eBPF instructions.
832 Instead, \fBbpf_get_stackid\fP() can collect up to
833 \fBPERF_MAX_STACK_DEPTH\fP both kernel and user frames. Note that
834 this limit can be controlled with the \fBsysctl\fP program, and
835 that it should be manually increased in order to profile long
836 user stacks (such as stacks for Java programs). To do so, use:
837 .INDENT 7.0
838 .INDENT 3.5
839 .sp
840 .EX
841 # sysctl kernel.perf_event_max_stack=<new value>
842 .EE
843 .UNINDENT
844 .UNINDENT
845 .TP
846 .B Return
847 The positive or null stack id on success, or a negative error
848 in case of failure.
849 .UNINDENT
850 .TP
851 .B \fBs64 bpf_csum_diff(__be32 *\fP\fIfrom\fP\fB, u32\fP \fIfrom_size\fP\fB, __be32 *\fP\fIto\fP\fB, u32\fP \fIto_size\fP\fB, __wsum\fP \fIseed\fP\fB)\fP
852 .INDENT 7.0
853 .TP
854 .B Description
855 Compute a checksum difference, from the raw buffer pointed by
856 \fIfrom\fP, of length \fIfrom_size\fP (that must be a multiple of 4),
857 towards the raw buffer pointed by \fIto\fP, of size \fIto_size\fP
858 (same remark). An optional \fIseed\fP can be added to the value
859 (this can be cascaded, the seed may come from a previous call
860 to the helper).
861 .sp
862 This is flexible enough to be used in several ways:
863 .INDENT 7.0
864 .IP \(bu 2
865 With \fIfrom_size\fP == 0, \fIto_size\fP > 0 and \fIseed\fP set to
866 checksum, it can be used when pushing new data.
867 .IP \(bu 2
868 With \fIfrom_size\fP > 0, \fIto_size\fP == 0 and \fIseed\fP set to
869 checksum, it can be used when removing data from a packet.
870 .IP \(bu 2
871 With \fIfrom_size\fP > 0, \fIto_size\fP > 0 and \fIseed\fP set to 0, it
872 can be used to compute a diff. Note that \fIfrom_size\fP and
873 \fIto_size\fP do not need to be equal.
874 .UNINDENT
875 .sp
876 This helper can be used in combination with
877 \fBbpf_l3_csum_replace\fP() and \fBbpf_l4_csum_replace\fP(), to
878 which one can feed in the difference computed with
879 \fBbpf_csum_diff\fP().
880 .TP
881 .B Return
882 The checksum result, or a negative error code in case of
883 failure.
884 .UNINDENT
885 .TP
886 .B \fBlong bpf_skb_get_tunnel_opt(struct sk_buff *\fP\fIskb\fP\fB, void *\fP\fIopt\fP\fB, u32\fP \fIsize\fP\fB)\fP
887 .INDENT 7.0
888 .TP
889 .B Description
890 Retrieve tunnel options metadata for the packet associated to
891 \fIskb\fP, and store the raw tunnel option data to the buffer \fIopt\fP
892 of \fIsize\fP\&.
893 .sp
894 This helper can be used with encapsulation devices that can
895 operate in \(dqcollect metadata\(dq mode (please refer to the related
896 note in the description of \fBbpf_skb_get_tunnel_key\fP() for
897 more details). A particular example where this can be used is
898 in combination with the Geneve encapsulation protocol, where it
899 allows for pushing (with \fBbpf_skb_get_tunnel_opt\fP() helper)
900 and retrieving arbitrary TLVs (Type\-Length\-Value headers) from
901 the eBPF program. This allows for full customization of these
902 headers.
903 .TP
904 .B Return
905 The size of the option data retrieved.
906 .UNINDENT
907 .TP
908 .B \fBlong bpf_skb_set_tunnel_opt(struct sk_buff *\fP\fIskb\fP\fB, void *\fP\fIopt\fP\fB, u32\fP \fIsize\fP\fB)\fP
909 .INDENT 7.0
910 .TP
911 .B Description
912 Set tunnel options metadata for the packet associated to \fIskb\fP
913 to the option data contained in the raw buffer \fIopt\fP of \fIsize\fP\&.
914 .sp
915 See also the description of the \fBbpf_skb_get_tunnel_opt\fP()
916 helper for additional information.
917 .TP
918 .B Return
919 0 on success, or a negative error in case of failure.
920 .UNINDENT
921 .TP
922 .B \fBlong bpf_skb_change_proto(struct sk_buff *\fP\fIskb\fP\fB, __be16\fP \fIproto\fP\fB, u64\fP \fIflags\fP\fB)\fP
923 .INDENT 7.0
924 .TP
925 .B Description
926 Change the protocol of the \fIskb\fP to \fIproto\fP\&. Currently
927 supported are transition from IPv4 to IPv6, and from IPv6 to
928 IPv4. The helper takes care of the groundwork for the
929 transition, including resizing the socket buffer. The eBPF
930 program is expected to fill the new headers, if any, via
931 \fBskb_store_bytes\fP() and to recompute the checksums with
932 \fBbpf_l3_csum_replace\fP() and \fBbpf_l4_csum_replace\fP(). The main case for this helper is to perform NAT64
933 operations out of an eBPF program.
934 .sp
935 Internally, the GSO type is marked as dodgy so that headers are
936 checked and segments are recalculated by the GSO/GRO engine.
937 The size for GSO target is adapted as well.
938 .sp
939 All values for \fIflags\fP are reserved for future usage, and must
940 be left at zero.
941 .sp
942 A call to this helper is susceptible to change the underlying
943 packet buffer. Therefore, at load time, all checks on pointers
944 previously done by the verifier are invalidated and must be
945 performed again, if the helper is used in combination with
946 direct packet access.
947 .TP
948 .B Return
949 0 on success, or a negative error in case of failure.
950 .UNINDENT
951 .TP
952 .B \fBlong bpf_skb_change_type(struct sk_buff *\fP\fIskb\fP\fB, u32\fP \fItype\fP\fB)\fP
953 .INDENT 7.0
954 .TP
955 .B Description
956 Change the packet type for the packet associated to \fIskb\fP\&. This
957 comes down to setting \fIskb\fP\fB\->pkt_type\fP to \fItype\fP, except
958 the eBPF program does not have a write access to \fIskb\fP\fB\->pkt_type\fP beside this helper. Using a helper here allows
959 for graceful handling of errors.
960 .sp
961 The major use case is to change incoming \fIskb*s to
962 **PACKET_HOST*\fP in a programmatic way instead of having to
963 recirculate via \fBredirect\fP(..., \fBBPF_F_INGRESS\fP), for
964 example.
965 .sp
966 Note that \fItype\fP only allows certain values. At this time, they
967 are:
968 .INDENT 7.0
969 .TP
970 .B \fBPACKET_HOST\fP
971 Packet is for us.
972 .TP
973 .B \fBPACKET_BROADCAST\fP
974 Send packet to all.
975 .TP
976 .B \fBPACKET_MULTICAST\fP
977 Send packet to group.
978 .TP
979 .B \fBPACKET_OTHERHOST\fP
980 Send packet to someone else.
981 .UNINDENT
982 .TP
983 .B Return
984 0 on success, or a negative error in case of failure.
985 .UNINDENT
986 .TP
987 .B \fBlong bpf_skb_under_cgroup(struct sk_buff *\fP\fIskb\fP\fB, struct bpf_map *\fP\fImap\fP\fB, u32\fP \fIindex\fP\fB)\fP
988 .INDENT 7.0
989 .TP
990 .B Description
991 Check whether \fIskb\fP is a descendant of the cgroup2 held by
992 \fImap\fP of type \fBBPF_MAP_TYPE_CGROUP_ARRAY\fP, at \fIindex\fP\&.
993 .TP
994 .B Return
995 The return value depends on the result of the test, and can be:
996 .INDENT 7.0
997 .IP \(bu 2
998 0, if the \fIskb\fP failed the cgroup2 descendant test.
999 .IP \(bu 2
1000 1, if the \fIskb\fP succeeded the cgroup2 descendant test.
1001 .IP \(bu 2
1002 A negative error code, if an error occurred.
1003 .UNINDENT
1004 .UNINDENT
1005 .TP
1006 .B \fBu32 bpf_get_hash_recalc(struct sk_buff *\fP\fIskb\fP\fB)\fP
1007 .INDENT 7.0
1008 .TP
1009 .B Description
1010 Retrieve the hash of the packet, \fIskb\fP\fB\->hash\fP\&. If it is
1011 not set, in particular if the hash was cleared due to mangling,
1012 recompute this hash. Later accesses to the hash can be done
1013 directly with \fIskb\fP\fB\->hash\fP\&.
1014 .sp
1015 Calling \fBbpf_set_hash_invalid\fP(), changing a packet
1016 prototype with \fBbpf_skb_change_proto\fP(), or calling
1017 \fBbpf_skb_store_bytes\fP() with the
1018 \fBBPF_F_INVALIDATE_HASH\fP are actions susceptible to clear
1019 the hash and to trigger a new computation for the next call to
1020 \fBbpf_get_hash_recalc\fP().
1021 .TP
1022 .B Return
1023 The 32\-bit hash.
1024 .UNINDENT
1025 .TP
1026 .B \fBu64 bpf_get_current_task(void)\fP
1027 .INDENT 7.0
1028 .TP
1029 .B Description
1030 Get the current task.
1031 .TP
1032 .B Return
1033 A pointer to the current task struct.
1034 .UNINDENT
1035 .TP
1036 .B \fBlong bpf_probe_write_user(void *\fP\fIdst\fP\fB, const void *\fP\fIsrc\fP\fB, u32\fP \fIlen\fP\fB)\fP
1037 .INDENT 7.0
1038 .TP
1039 .B Description
1040 Attempt in a safe way to write \fIlen\fP bytes from the buffer
1041 \fIsrc\fP to \fIdst\fP in memory. It only works for threads that are in
1042 user context, and \fIdst\fP must be a valid user space address.
1043 .sp
1044 This helper should not be used to implement any kind of
1045 security mechanism because of TOC\-TOU attacks, but rather to
1046 debug, divert, and manipulate execution of semi\-cooperative
1047 processes.
1048 .sp
1049 Keep in mind that this feature is meant for experiments, and it
1050 has a risk of crashing the system and running programs.
1051 Therefore, when an eBPF program using this helper is attached,
1052 a warning including PID and process name is printed to kernel
1053 logs.
1054 .TP
1055 .B Return
1056 0 on success, or a negative error in case of failure.
1057 .UNINDENT
1058 .TP
1059 .B \fBlong bpf_current_task_under_cgroup(struct bpf_map *\fP\fImap\fP\fB, u32\fP \fIindex\fP\fB)\fP
1060 .INDENT 7.0
1061 .TP
1062 .B Description
1063 Check whether the probe is being run is the context of a given
1064 subset of the cgroup2 hierarchy. The cgroup2 to test is held by
1065 \fImap\fP of type \fBBPF_MAP_TYPE_CGROUP_ARRAY\fP, at \fIindex\fP\&.
1066 .TP
1067 .B Return
1068 The return value depends on the result of the test, and can be:
1069 .INDENT 7.0
1070 .IP \(bu 2
1071 1, if current task belongs to the cgroup2.
1072 .IP \(bu 2
1073 0, if current task does not belong to the cgroup2.
1074 .IP \(bu 2
1075 A negative error code, if an error occurred.
1076 .UNINDENT
1077 .UNINDENT
1078 .TP
1079 .B \fBlong bpf_skb_change_tail(struct sk_buff *\fP\fIskb\fP\fB, u32\fP \fIlen\fP\fB, u64\fP \fIflags\fP\fB)\fP
1080 .INDENT 7.0
1081 .TP
1082 .B Description
1083 Resize (trim or grow) the packet associated to \fIskb\fP to the
1084 new \fIlen\fP\&. The \fIflags\fP are reserved for future usage, and must
1085 be left at zero.
1086 .sp
1087 The basic idea is that the helper performs the needed work to
1088 change the size of the packet, then the eBPF program rewrites
1089 the rest via helpers like \fBbpf_skb_store_bytes\fP(),
1090 \fBbpf_l3_csum_replace\fP(), \fBbpf_l3_csum_replace\fP()
1091 and others. This helper is a slow path utility intended for
1092 replies with control messages. And because it is targeted for
1093 slow path, the helper itself can afford to be slow: it
1094 implicitly linearizes, unclones and drops offloads from the
1095 \fIskb\fP\&.
1096 .sp
1097 A call to this helper is susceptible to change the underlying
1098 packet buffer. Therefore, at load time, all checks on pointers
1099 previously done by the verifier are invalidated and must be
1100 performed again, if the helper is used in combination with
1101 direct packet access.
1102 .TP
1103 .B Return
1104 0 on success, or a negative error in case of failure.
1105 .UNINDENT
1106 .TP
1107 .B \fBlong bpf_skb_pull_data(struct sk_buff *\fP\fIskb\fP\fB, u32\fP \fIlen\fP\fB)\fP
1108 .INDENT 7.0
1109 .TP
1110 .B Description
1111 Pull in non\-linear data in case the \fIskb\fP is non\-linear and not
1112 all of \fIlen\fP are part of the linear section. Make \fIlen\fP bytes
1113 from \fIskb\fP readable and writable. If a zero value is passed for
1114 \fIlen\fP, then all bytes in the linear part of \fIskb\fP will be made
1115 readable and writable.
1116 .sp
1117 This helper is only needed for reading and writing with direct
1118 packet access.
1119 .sp
1120 For direct packet access, testing that offsets to access
1121 are within packet boundaries (test on \fIskb\fP\fB\->data_end\fP) is
1122 susceptible to fail if offsets are invalid, or if the requested
1123 data is in non\-linear parts of the \fIskb\fP\&. On failure the
1124 program can just bail out, or in the case of a non\-linear
1125 buffer, use a helper to make the data available. The
1126 \fBbpf_skb_load_bytes\fP() helper is a first solution to access
1127 the data. Another one consists in using \fBbpf_skb_pull_data\fP
1128 to pull in once the non\-linear parts, then retesting and
1129 eventually access the data.
1130 .sp
1131 At the same time, this also makes sure the \fIskb\fP is uncloned,
1132 which is a necessary condition for direct write. As this needs
1133 to be an invariant for the write part only, the verifier
1134 detects writes and adds a prologue that is calling
1135 \fBbpf_skb_pull_data()\fP to effectively unclone the \fIskb\fP from
1136 the very beginning in case it is indeed cloned.
1137 .sp
1138 A call to this helper is susceptible to change the underlying
1139 packet buffer. Therefore, at load time, all checks on pointers
1140 previously done by the verifier are invalidated and must be
1141 performed again, if the helper is used in combination with
1142 direct packet access.
1143 .TP
1144 .B Return
1145 0 on success, or a negative error in case of failure.
1146 .UNINDENT
1147 .TP
1148 .B \fBs64 bpf_csum_update(struct sk_buff *\fP\fIskb\fP\fB, __wsum\fP \fIcsum\fP\fB)\fP
1149 .INDENT 7.0
1150 .TP
1151 .B Description
1152 Add the checksum \fIcsum\fP into \fIskb\fP\fB\->csum\fP in case the
1153 driver has supplied a checksum for the entire packet into that
1154 field. Return an error otherwise. This helper is intended to be
1155 used in combination with \fBbpf_csum_diff\fP(), in particular
1156 when the checksum needs to be updated after data has been
1157 written into the packet through direct packet access.
1158 .TP
1159 .B Return
1160 The checksum on success, or a negative error code in case of
1161 failure.
1162 .UNINDENT
1163 .TP
1164 .B \fBvoid bpf_set_hash_invalid(struct sk_buff *\fP\fIskb\fP\fB)\fP
1165 .INDENT 7.0
1166 .TP
1167 .B Description
1168 Invalidate the current \fIskb\fP\fB\->hash\fP\&. It can be used after
1169 mangling on headers through direct packet access, in order to
1170 indicate that the hash is outdated and to trigger a
1171 recalculation the next time the kernel tries to access this
1172 hash or when the \fBbpf_get_hash_recalc\fP() helper is called.
1173 .TP
1174 .B Return
1175 void.
1176 .UNINDENT
1177 .TP
1178 .B \fBlong bpf_get_numa_node_id(void)\fP
1179 .INDENT 7.0
1180 .TP
1181 .B Description
1182 Return the id of the current NUMA node. The primary use case
1183 for this helper is the selection of sockets for the local NUMA
1184 node, when the program is attached to sockets using the
1185 \fBSO_ATTACH_REUSEPORT_EBPF\fP option (see also \fBsocket(7)\fP),
1186 but the helper is also available to other eBPF program types,
1187 similarly to \fBbpf_get_smp_processor_id\fP().
1188 .TP
1189 .B Return
1190 The id of current NUMA node.
1191 .UNINDENT
1192 .TP
1193 .B \fBlong bpf_skb_change_head(struct sk_buff *\fP\fIskb\fP\fB, u32\fP \fIlen\fP\fB, u64\fP \fIflags\fP\fB)\fP
1194 .INDENT 7.0
1195 .TP
1196 .B Description
1197 Grows headroom of packet associated to \fIskb\fP and adjusts the
1198 offset of the MAC header accordingly, adding \fIlen\fP bytes of
1199 space. It automatically extends and reallocates memory as
1200 required.
1201 .sp
1202 This helper can be used on a layer 3 \fIskb\fP to push a MAC header
1203 for redirection into a layer 2 device.
1204 .sp
1205 All values for \fIflags\fP are reserved for future usage, and must
1206 be left at zero.
1207 .sp
1208 A call to this helper is susceptible to change the underlying
1209 packet buffer. Therefore, at load time, all checks on pointers
1210 previously done by the verifier are invalidated and must be
1211 performed again, if the helper is used in combination with
1212 direct packet access.
1213 .TP
1214 .B Return
1215 0 on success, or a negative error in case of failure.
1216 .UNINDENT
1217 .TP
1218 .B \fBlong bpf_xdp_adjust_head(struct xdp_buff *\fP\fIxdp_md\fP\fB, int\fP \fIdelta\fP\fB)\fP
1219 .INDENT 7.0
1220 .TP
1221 .B Description
1222 Adjust (move) \fIxdp_md\fP\fB\->data\fP by \fIdelta\fP bytes. Note that
1223 it is possible to use a negative value for \fIdelta\fP\&. This helper
1224 can be used to prepare the packet for pushing or popping
1225 headers.
1226 .sp
1227 A call to this helper is susceptible to change the underlying
1228 packet buffer. Therefore, at load time, all checks on pointers
1229 previously done by the verifier are invalidated and must be
1230 performed again, if the helper is used in combination with
1231 direct packet access.
1232 .TP
1233 .B Return
1234 0 on success, or a negative error in case of failure.
1235 .UNINDENT
1236 .TP
1237 .B \fBlong bpf_probe_read_str(void *\fP\fIdst\fP\fB, u32\fP \fIsize\fP\fB, const void *\fP\fIunsafe_ptr\fP\fB)\fP
1238 .INDENT 7.0
1239 .TP
1240 .B Description
1241 Copy a NUL terminated string from an unsafe kernel address
1242 \fIunsafe_ptr\fP to \fIdst\fP\&. See \fBbpf_probe_read_kernel_str\fP() for
1243 more details.
1244 .sp
1245 Generally, use \fBbpf_probe_read_user_str\fP() or
1246 \fBbpf_probe_read_kernel_str\fP() instead.
1247 .TP
1248 .B Return
1249 On success, the strictly positive length of the string,
1250 including the trailing NUL character. On error, a negative
1251 value.
1252 .UNINDENT
1253 .TP
1254 .B \fBu64 bpf_get_socket_cookie(struct sk_buff *\fP\fIskb\fP\fB)\fP
1255 .INDENT 7.0
1256 .TP
1257 .B Description
1258 If the \fBstruct sk_buff\fP pointed by \fIskb\fP has a known socket,
1259 retrieve the cookie (generated by the kernel) of this socket.
1260 If no cookie has been set yet, generate a new cookie. Once
1261 generated, the socket cookie remains stable for the life of the
1262 socket. This helper can be useful for monitoring per socket
1263 networking traffic statistics as it provides a global socket
1264 identifier that can be assumed unique.
1265 .TP
1266 .B Return
1267 A 8\-byte long unique number on success, or 0 if the socket
1268 field is missing inside \fIskb\fP\&.
1269 .UNINDENT
1270 .TP
1271 .B \fBu64 bpf_get_socket_cookie(struct bpf_sock_addr *\fP\fIctx\fP\fB)\fP
1272 .INDENT 7.0
1273 .TP
1274 .B Description
1275 Equivalent to bpf_get_socket_cookie() helper that accepts
1276 \fIskb\fP, but gets socket from \fBstruct bpf_sock_addr\fP context.
1277 .TP
1278 .B Return
1279 A 8\-byte long unique number.
1280 .UNINDENT
1281 .TP
1282 .B \fBu64 bpf_get_socket_cookie(struct bpf_sock_ops *\fP\fIctx\fP\fB)\fP
1283 .INDENT 7.0
1284 .TP
1285 .B Description
1286 Equivalent to \fBbpf_get_socket_cookie\fP() helper that accepts
1287 \fIskb\fP, but gets socket from \fBstruct bpf_sock_ops\fP context.
1288 .TP
1289 .B Return
1290 A 8\-byte long unique number.
1291 .UNINDENT
1292 .TP
1293 .B \fBu64 bpf_get_socket_cookie(struct sock *\fP\fIsk\fP\fB)\fP
1294 .INDENT 7.0
1295 .TP
1296 .B Description
1297 Equivalent to \fBbpf_get_socket_cookie\fP() helper that accepts
1298 \fIsk\fP, but gets socket from a BTF \fBstruct sock\fP\&. This helper
1299 also works for sleepable programs.
1300 .TP
1301 .B Return
1302 A 8\-byte long unique number or 0 if \fIsk\fP is NULL.
1303 .UNINDENT
1304 .TP
1305 .B \fBu32 bpf_get_socket_uid(struct sk_buff *\fP\fIskb\fP\fB)\fP
1306 .INDENT 7.0
1307 .TP
1308 .B Description
1309 Get the owner UID of the socked associated to \fIskb\fP\&.
1310 .TP
1311 .B Return
1312 The owner UID of the socket associated to \fIskb\fP\&. If the socket
1313 is \fBNULL\fP, or if it is not a full socket (i.e. if it is a
1314 time\-wait or a request socket instead), \fBoverflowuid\fP value
1315 is returned (note that \fBoverflowuid\fP might also be the actual
1316 UID value for the socket).
1317 .UNINDENT
1318 .TP
1319 .B \fBlong bpf_set_hash(struct sk_buff *\fP\fIskb\fP\fB, u32\fP \fIhash\fP\fB)\fP
1320 .INDENT 7.0
1321 .TP
1322 .B Description
1323 Set the full hash for \fIskb\fP (set the field \fIskb\fP\fB\->hash\fP)
1324 to value \fIhash\fP\&.
1325 .TP
1326 .B Return
1327 0
1328 .UNINDENT
1329 .TP
1330 .B \fBlong bpf_setsockopt(void *\fP\fIbpf_socket\fP\fB, int\fP \fIlevel\fP\fB, int\fP \fIoptname\fP\fB, void *\fP\fIoptval\fP\fB, int\fP \fIoptlen\fP\fB)\fP
1331 .INDENT 7.0
1332 .TP
1333 .B Description
1334 Emulate a call to \fBsetsockopt()\fP on the socket associated to
1335 \fIbpf_socket\fP, which must be a full socket. The \fIlevel\fP at
1336 which the option resides and the name \fIoptname\fP of the option
1337 must be specified, see \fBsetsockopt(2)\fP for more information.
1338 The option value of length \fIoptlen\fP is pointed by \fIoptval\fP\&.
1339 .sp
1340 \fIbpf_socket\fP should be one of the following:
1341 .INDENT 7.0
1342 .IP \(bu 2
1343 \fBstruct bpf_sock_ops\fP for \fBBPF_PROG_TYPE_SOCK_OPS\fP\&.
1344 .IP \(bu 2
1345 \fBstruct bpf_sock_addr\fP for \fBBPF_CGROUP_INET4_CONNECT\fP,
1346 \fBBPF_CGROUP_INET6_CONNECT\fP and \fBBPF_CGROUP_UNIX_CONNECT\fP\&.
1347 .UNINDENT
1348 .sp
1349 This helper actually implements a subset of \fBsetsockopt()\fP\&.
1350 It supports the following \fIlevel\fPs:
1351 .INDENT 7.0
1352 .IP \(bu 2
1353 \fBSOL_SOCKET\fP, which supports the following \fIoptname\fPs:
1354 \fBSO_RCVBUF\fP, \fBSO_SNDBUF\fP, \fBSO_MAX_PACING_RATE\fP,
1355 \fBSO_PRIORITY\fP, \fBSO_RCVLOWAT\fP, \fBSO_MARK\fP,
1356 \fBSO_BINDTODEVICE\fP, \fBSO_KEEPALIVE\fP, \fBSO_REUSEADDR\fP,
1357 \fBSO_REUSEPORT\fP, \fBSO_BINDTOIFINDEX\fP, \fBSO_TXREHASH\fP\&.
1358 .IP \(bu 2
1359 \fBIPPROTO_TCP\fP, which supports the following \fIoptname\fPs:
1360 \fBTCP_CONGESTION\fP, \fBTCP_BPF_IW\fP,
1361 \fBTCP_BPF_SNDCWND_CLAMP\fP, \fBTCP_SAVE_SYN\fP,
1362 \fBTCP_KEEPIDLE\fP, \fBTCP_KEEPINTVL\fP, \fBTCP_KEEPCNT\fP,
1363 \fBTCP_SYNCNT\fP, \fBTCP_USER_TIMEOUT\fP, \fBTCP_NOTSENT_LOWAT\fP,
1364 \fBTCP_NODELAY\fP, \fBTCP_MAXSEG\fP, \fBTCP_WINDOW_CLAMP\fP,
1365 \fBTCP_THIN_LINEAR_TIMEOUTS\fP, \fBTCP_BPF_DELACK_MAX\fP,
1366 \fBTCP_BPF_RTO_MIN\fP, \fBTCP_BPF_SOCK_OPS_CB_FLAGS\fP\&.
1367 .IP \(bu 2
1368 \fBIPPROTO_IP\fP, which supports \fIoptname\fP \fBIP_TOS\fP\&.
1369 .IP \(bu 2
1370 \fBIPPROTO_IPV6\fP, which supports the following \fIoptname\fPs:
1371 \fBIPV6_TCLASS\fP, \fBIPV6_AUTOFLOWLABEL\fP\&.
1372 .UNINDENT
1373 .TP
1374 .B Return
1375 0 on success, or a negative error in case of failure.
1376 .UNINDENT
1377 .TP
1378 .B \fBlong bpf_skb_adjust_room(struct sk_buff *\fP\fIskb\fP\fB, s32\fP \fIlen_diff\fP\fB, u32\fP \fImode\fP\fB, u64\fP \fIflags\fP\fB)\fP
1379 .INDENT 7.0
1380 .TP
1381 .B Description
1382 Grow or shrink the room for data in the packet associated to
1383 \fIskb\fP by \fIlen_diff\fP, and according to the selected \fImode\fP\&.
1384 .sp
1385 By default, the helper will reset any offloaded checksum
1386 indicator of the skb to CHECKSUM_NONE. This can be avoided
1387 by the following flag:
1388 .INDENT 7.0
1389 .IP \(bu 2
1390 \fBBPF_F_ADJ_ROOM_NO_CSUM_RESET\fP: Do not reset offloaded
1391 checksum data of the skb to CHECKSUM_NONE.
1392 .UNINDENT
1393 .sp
1394 There are two supported modes at this time:
1395 .INDENT 7.0
1396 .IP \(bu 2
1397 \fBBPF_ADJ_ROOM_MAC\fP: Adjust room at the mac layer
1398 (room space is added or removed between the layer 2 and
1399 layer 3 headers).
1400 .IP \(bu 2
1401 \fBBPF_ADJ_ROOM_NET\fP: Adjust room at the network layer
1402 (room space is added or removed between the layer 3 and
1403 layer 4 headers).
1404 .UNINDENT
1405 .sp
1406 The following flags are supported at this time:
1407 .INDENT 7.0
1408 .IP \(bu 2
1409 \fBBPF_F_ADJ_ROOM_FIXED_GSO\fP: Do not adjust gso_size.
1410 Adjusting mss in this way is not allowed for datagrams.
1411 .IP \(bu 2
1412 \fBBPF_F_ADJ_ROOM_ENCAP_L3_IPV4\fP,
1413 \fBBPF_F_ADJ_ROOM_ENCAP_L3_IPV6\fP:
1414 Any new space is reserved to hold a tunnel header.
1415 Configure skb offsets and other fields accordingly.
1416 .IP \(bu 2
1417 \fBBPF_F_ADJ_ROOM_ENCAP_L4_GRE\fP,
1418 \fBBPF_F_ADJ_ROOM_ENCAP_L4_UDP\fP:
1419 Use with ENCAP_L3 flags to further specify the tunnel type.
1420 .IP \(bu 2
1421 \fBBPF_F_ADJ_ROOM_ENCAP_L2\fP(\fIlen\fP):
1422 Use with ENCAP_L3/L4 flags to further specify the tunnel
1423 type; \fIlen\fP is the length of the inner MAC header.
1424 .IP \(bu 2
1425 \fBBPF_F_ADJ_ROOM_ENCAP_L2_ETH\fP:
1426 Use with BPF_F_ADJ_ROOM_ENCAP_L2 flag to further specify the
1427 L2 type as Ethernet.
1428 .IP \(bu 2
1429 \fBBPF_F_ADJ_ROOM_DECAP_L3_IPV4\fP,
1430 \fBBPF_F_ADJ_ROOM_DECAP_L3_IPV6\fP:
1431 Indicate the new IP header version after decapsulating the outer
1432 IP header. Used when the inner and outer IP versions are different.
1433 .UNINDENT
1434 .sp
1435 A call to this helper is susceptible to change the underlying
1436 packet buffer. Therefore, at load time, all checks on pointers
1437 previously done by the verifier are invalidated and must be
1438 performed again, if the helper is used in combination with
1439 direct packet access.
1440 .TP
1441 .B Return
1442 0 on success, or a negative error in case of failure.
1443 .UNINDENT
1444 .TP
1445 .B \fBlong bpf_redirect_map(struct bpf_map *\fP\fImap\fP\fB, u64\fP \fIkey\fP\fB, u64\fP \fIflags\fP\fB)\fP
1446 .INDENT 7.0
1447 .TP
1448 .B Description
1449 Redirect the packet to the endpoint referenced by \fImap\fP at
1450 index \fIkey\fP\&. Depending on its type, this \fImap\fP can contain
1451 references to net devices (for forwarding packets through other
1452 ports), or to CPUs (for redirecting XDP frames to another CPU;
1453 but this is only implemented for native XDP (with driver
1454 support) as of this writing).
1455 .sp
1456 The lower two bits of \fIflags\fP are used as the return code if
1457 the map lookup fails. This is so that the return value can be
1458 one of the XDP program return codes up to \fBXDP_TX\fP, as chosen
1459 by the caller. The higher bits of \fIflags\fP can be set to
1460 BPF_F_BROADCAST or BPF_F_EXCLUDE_INGRESS as defined below.
1461 .sp
1462 With BPF_F_BROADCAST the packet will be broadcasted to all the
1463 interfaces in the map, with BPF_F_EXCLUDE_INGRESS the ingress
1464 interface will be excluded when do broadcasting.
1465 .sp
1466 See also \fBbpf_redirect\fP(), which only supports redirecting
1467 to an ifindex, but doesn\(aqt require a map to do so.
1468 .TP
1469 .B Return
1470 \fBXDP_REDIRECT\fP on success, or the value of the two lower bits
1471 of the \fIflags\fP argument on error.
1472 .UNINDENT
1473 .TP
1474 .B \fBlong bpf_sk_redirect_map(struct sk_buff *\fP\fIskb\fP\fB, struct bpf_map *\fP\fImap\fP\fB, u32\fP \fIkey\fP\fB, u64\fP \fIflags\fP\fB)\fP
1475 .INDENT 7.0
1476 .TP
1477 .B Description
1478 Redirect the packet to the socket referenced by \fImap\fP (of type
1479 \fBBPF_MAP_TYPE_SOCKMAP\fP) at index \fIkey\fP\&. Both ingress and
1480 egress interfaces can be used for redirection. The
1481 \fBBPF_F_INGRESS\fP value in \fIflags\fP is used to make the
1482 distinction (ingress path is selected if the flag is present,
1483 egress path otherwise). This is the only flag supported for now.
1484 .TP
1485 .B Return
1486 \fBSK_PASS\fP on success, or \fBSK_DROP\fP on error.
1487 .UNINDENT
1488 .TP
1489 .B \fBlong bpf_sock_map_update(struct bpf_sock_ops *\fP\fIskops\fP\fB, struct bpf_map *\fP\fImap\fP\fB, void *\fP\fIkey\fP\fB, u64\fP \fIflags\fP\fB)\fP
1490 .INDENT 7.0
1491 .TP
1492 .B Description
1493 Add an entry to, or update a \fImap\fP referencing sockets. The
1494 \fIskops\fP is used as a new value for the entry associated to
1495 \fIkey\fP\&. \fIflags\fP is one of:
1496 .INDENT 7.0
1497 .TP
1498 .B \fBBPF_NOEXIST\fP
1499 The entry for \fIkey\fP must not exist in the map.
1500 .TP
1501 .B \fBBPF_EXIST\fP
1502 The entry for \fIkey\fP must already exist in the map.
1503 .TP
1504 .B \fBBPF_ANY\fP
1505 No condition on the existence of the entry for \fIkey\fP\&.
1506 .UNINDENT
1507 .sp
1508 If the \fImap\fP has eBPF programs (parser and verdict), those will
1509 be inherited by the socket being added. If the socket is
1510 already attached to eBPF programs, this results in an error.
1511 .TP
1512 .B Return
1513 0 on success, or a negative error in case of failure.
1514 .UNINDENT
1515 .TP
1516 .B \fBlong bpf_xdp_adjust_meta(struct xdp_buff *\fP\fIxdp_md\fP\fB, int\fP \fIdelta\fP\fB)\fP
1517 .INDENT 7.0
1518 .TP
1519 .B Description
1520 Adjust the address pointed by \fIxdp_md\fP\fB\->data_meta\fP by
1521 \fIdelta\fP (which can be positive or negative). Note that this
1522 operation modifies the address stored in \fIxdp_md\fP\fB\->data\fP,
1523 so the latter must be loaded only after the helper has been
1524 called.
1525 .sp
1526 The use of \fIxdp_md\fP\fB\->data_meta\fP is optional and programs
1527 are not required to use it. The rationale is that when the
1528 packet is processed with XDP (e.g. as DoS filter), it is
1529 possible to push further meta data along with it before passing
1530 to the stack, and to give the guarantee that an ingress eBPF
1531 program attached as a TC classifier on the same device can pick
1532 this up for further post\-processing. Since TC works with socket
1533 buffers, it remains possible to set from XDP the \fBmark\fP or
1534 \fBpriority\fP pointers, or other pointers for the socket buffer.
1535 Having this scratch space generic and programmable allows for
1536 more flexibility as the user is free to store whatever meta
1537 data they need.
1538 .sp
1539 A call to this helper is susceptible to change the underlying
1540 packet buffer. Therefore, at load time, all checks on pointers
1541 previously done by the verifier are invalidated and must be
1542 performed again, if the helper is used in combination with
1543 direct packet access.
1544 .TP
1545 .B Return
1546 0 on success, or a negative error in case of failure.
1547 .UNINDENT
1548 .TP
1549 .B \fBlong bpf_perf_event_read_value(struct bpf_map *\fP\fImap\fP\fB, u64\fP \fIflags\fP\fB, struct bpf_perf_event_value *\fP\fIbuf\fP\fB, u32\fP \fIbuf_size\fP\fB)\fP
1550 .INDENT 7.0
1551 .TP
1552 .B Description
1553 Read the value of a perf event counter, and store it into \fIbuf\fP
1554 of size \fIbuf_size\fP\&. This helper relies on a \fImap\fP of type
1555 \fBBPF_MAP_TYPE_PERF_EVENT_ARRAY\fP\&. The nature of the perf event
1556 counter is selected when \fImap\fP is updated with perf event file
1557 descriptors. The \fImap\fP is an array whose size is the number of
1558 available CPUs, and each cell contains a value relative to one
1559 CPU. The value to retrieve is indicated by \fIflags\fP, that
1560 contains the index of the CPU to look up, masked with
1561 \fBBPF_F_INDEX_MASK\fP\&. Alternatively, \fIflags\fP can be set to
1562 \fBBPF_F_CURRENT_CPU\fP to indicate that the value for the
1563 current CPU should be retrieved.
1564 .sp
1565 This helper behaves in a way close to
1566 \fBbpf_perf_event_read\fP() helper, save that instead of
1567 just returning the value observed, it fills the \fIbuf\fP
1568 structure. This allows for additional data to be retrieved: in
1569 particular, the enabled and running times (in \fIbuf\fP\fB\->enabled\fP and \fIbuf\fP\fB\->running\fP, respectively) are
1570 copied. In general, \fBbpf_perf_event_read_value\fP() is
1571 recommended over \fBbpf_perf_event_read\fP(), which has some
1572 ABI issues and provides fewer functionalities.
1573 .sp
1574 These values are interesting, because hardware PMU (Performance
1575 Monitoring Unit) counters are limited resources. When there are
1576 more PMU based perf events opened than available counters,
1577 kernel will multiplex these events so each event gets certain
1578 percentage (but not all) of the PMU time. In case that
1579 multiplexing happens, the number of samples or counter value
1580 will not reflect the case compared to when no multiplexing
1581 occurs. This makes comparison between different runs difficult.
1582 Typically, the counter value should be normalized before
1583 comparing to other experiments. The usual normalization is done
1584 as follows.
1585 .INDENT 7.0
1586 .INDENT 3.5
1587 .sp
1588 .EX
1589 normalized_counter = counter * t_enabled / t_running
1590 .EE
1591 .UNINDENT
1592 .UNINDENT
1593 .sp
1594 Where t_enabled is the time enabled for event and t_running is
1595 the time running for event since last normalization. The
1596 enabled and running times are accumulated since the perf event
1597 open. To achieve scaling factor between two invocations of an
1598 eBPF program, users can use CPU id as the key (which is
1599 typical for perf array usage model) to remember the previous
1600 value and do the calculation inside the eBPF program.
1601 .TP
1602 .B Return
1603 0 on success, or a negative error in case of failure.
1604 .UNINDENT
1605 .TP
1606 .B \fBlong bpf_perf_prog_read_value(struct bpf_perf_event_data *\fP\fIctx\fP\fB, struct bpf_perf_event_value *\fP\fIbuf\fP\fB, u32\fP \fIbuf_size\fP\fB)\fP
1607 .INDENT 7.0
1608 .TP
1609 .B Description
1610 For an eBPF program attached to a perf event, retrieve the
1611 value of the event counter associated to \fIctx\fP and store it in
1612 the structure pointed by \fIbuf\fP and of size \fIbuf_size\fP\&. Enabled
1613 and running times are also stored in the structure (see
1614 description of helper \fBbpf_perf_event_read_value\fP() for
1615 more details).
1616 .TP
1617 .B Return
1618 0 on success, or a negative error in case of failure.
1619 .UNINDENT
1620 .TP
1621 .B \fBlong bpf_getsockopt(void *\fP\fIbpf_socket\fP\fB, int\fP \fIlevel\fP\fB, int\fP \fIoptname\fP\fB, void *\fP\fIoptval\fP\fB, int\fP \fIoptlen\fP\fB)\fP
1622 .INDENT 7.0
1623 .TP
1624 .B Description
1625 Emulate a call to \fBgetsockopt()\fP on the socket associated to
1626 \fIbpf_socket\fP, which must be a full socket. The \fIlevel\fP at
1627 which the option resides and the name \fIoptname\fP of the option
1628 must be specified, see \fBgetsockopt(2)\fP for more information.
1629 The retrieved value is stored in the structure pointed by
1630 \fIopval\fP and of length \fIoptlen\fP\&.
1631 .sp
1632 \fIbpf_socket\fP should be one of the following:
1633 .INDENT 7.0
1634 .IP \(bu 2
1635 \fBstruct bpf_sock_ops\fP for \fBBPF_PROG_TYPE_SOCK_OPS\fP\&.
1636 .IP \(bu 2
1637 \fBstruct bpf_sock_addr\fP for \fBBPF_CGROUP_INET4_CONNECT\fP,
1638 \fBBPF_CGROUP_INET6_CONNECT\fP and \fBBPF_CGROUP_UNIX_CONNECT\fP\&.
1639 .UNINDENT
1640 .sp
1641 This helper actually implements a subset of \fBgetsockopt()\fP\&.
1642 It supports the same set of \fIoptname\fPs that is supported by
1643 the \fBbpf_setsockopt\fP() helper. The exceptions are
1644 \fBTCP_BPF_*\fP is \fBbpf_setsockopt\fP() only and
1645 \fBTCP_SAVED_SYN\fP is \fBbpf_getsockopt\fP() only.
1646 .TP
1647 .B Return
1648 0 on success, or a negative error in case of failure.
1649 .UNINDENT
1650 .TP
1651 .B \fBlong bpf_override_return(struct pt_regs *\fP\fIregs\fP\fB, u64\fP \fIrc\fP\fB)\fP
1652 .INDENT 7.0
1653 .TP
1654 .B Description
1655 Used for error injection, this helper uses kprobes to override
1656 the return value of the probed function, and to set it to \fIrc\fP\&.
1657 The first argument is the context \fIregs\fP on which the kprobe
1658 works.
1659 .sp
1660 This helper works by setting the PC (program counter)
1661 to an override function which is run in place of the original
1662 probed function. This means the probed function is not run at
1663 all. The replacement function just returns with the required
1664 value.
1665 .sp
1666 This helper has security implications, and thus is subject to
1667 restrictions. It is only available if the kernel was compiled
1668 with the \fBCONFIG_BPF_KPROBE_OVERRIDE\fP configuration
1669 option, and in this case it only works on functions tagged with
1670 \fBALLOW_ERROR_INJECTION\fP in the kernel code.
1671 .TP
1672 .B Return
1673 0
1674 .UNINDENT
1675 .TP
1676 .B \fBlong bpf_sock_ops_cb_flags_set(struct bpf_sock_ops *\fP\fIbpf_sock\fP\fB, int\fP \fIargval\fP\fB)\fP
1677 .INDENT 7.0
1678 .TP
1679 .B Description
1680 Attempt to set the value of the \fBbpf_sock_ops_cb_flags\fP field
1681 for the full TCP socket associated to \fIbpf_sock_ops\fP to
1682 \fIargval\fP\&.
1683 .sp
1684 The primary use of this field is to determine if there should
1685 be calls to eBPF programs of type
1686 \fBBPF_PROG_TYPE_SOCK_OPS\fP at various points in the TCP
1687 code. A program of the same type can change its value, per
1688 connection and as necessary, when the connection is
1689 established. This field is directly accessible for reading, but
1690 this helper must be used for updates in order to return an
1691 error if an eBPF program tries to set a callback that is not
1692 supported in the current kernel.
1693 .sp
1694 \fIargval\fP is a flag array which can combine these flags:
1695 .INDENT 7.0
1696 .IP \(bu 2
1697 \fBBPF_SOCK_OPS_RTO_CB_FLAG\fP (retransmission time out)
1698 .IP \(bu 2
1699 \fBBPF_SOCK_OPS_RETRANS_CB_FLAG\fP (retransmission)
1700 .IP \(bu 2
1701 \fBBPF_SOCK_OPS_STATE_CB_FLAG\fP (TCP state change)
1702 .IP \(bu 2
1703 \fBBPF_SOCK_OPS_RTT_CB_FLAG\fP (every RTT)
1704 .UNINDENT
1705 .sp
1706 Therefore, this function can be used to clear a callback flag by
1707 setting the appropriate bit to zero. e.g. to disable the RTO
1708 callback:
1709 .INDENT 7.0
1710 .TP
1711 .B \fBbpf_sock_ops_cb_flags_set(bpf_sock,\fP
1712 \fBbpf_sock\->bpf_sock_ops_cb_flags & ~BPF_SOCK_OPS_RTO_CB_FLAG)\fP
1713 .UNINDENT
1714 .sp
1715 Here are some examples of where one could call such eBPF
1716 program:
1717 .INDENT 7.0
1718 .IP \(bu 2
1719 When RTO fires.
1720 .IP \(bu 2
1721 When a packet is retransmitted.
1722 .IP \(bu 2
1723 When the connection terminates.
1724 .IP \(bu 2
1725 When a packet is sent.
1726 .IP \(bu 2
1727 When a packet is received.
1728 .UNINDENT
1729 .TP
1730 .B Return
1731 Code \fB\-EINVAL\fP if the socket is not a full TCP socket;
1732 otherwise, a positive number containing the bits that could not
1733 be set is returned (which comes down to 0 if all bits were set
1734 as required).
1735 .UNINDENT
1736 .TP
1737 .B \fBlong bpf_msg_redirect_map(struct sk_msg_buff *\fP\fImsg\fP\fB, struct bpf_map *\fP\fImap\fP\fB, u32\fP \fIkey\fP\fB, u64\fP \fIflags\fP\fB)\fP
1738 .INDENT 7.0
1739 .TP
1740 .B Description
1741 This helper is used in programs implementing policies at the
1742 socket level. If the message \fImsg\fP is allowed to pass (i.e. if
1743 the verdict eBPF program returns \fBSK_PASS\fP), redirect it to
1744 the socket referenced by \fImap\fP (of type
1745 \fBBPF_MAP_TYPE_SOCKMAP\fP) at index \fIkey\fP\&. Both ingress and
1746 egress interfaces can be used for redirection. The
1747 \fBBPF_F_INGRESS\fP value in \fIflags\fP is used to make the
1748 distinction (ingress path is selected if the flag is present,
1749 egress path otherwise). This is the only flag supported for now.
1750 .TP
1751 .B Return
1752 \fBSK_PASS\fP on success, or \fBSK_DROP\fP on error.
1753 .UNINDENT
1754 .TP
1755 .B \fBlong bpf_msg_apply_bytes(struct sk_msg_buff *\fP\fImsg\fP\fB, u32\fP \fIbytes\fP\fB)\fP
1756 .INDENT 7.0
1757 .TP
1758 .B Description
1759 For socket policies, apply the verdict of the eBPF program to
1760 the next \fIbytes\fP (number of bytes) of message \fImsg\fP\&.
1761 .sp
1762 For example, this helper can be used in the following cases:
1763 .INDENT 7.0
1764 .IP \(bu 2
1765 A single \fBsendmsg\fP() or \fBsendfile\fP() system call
1766 contains multiple logical messages that the eBPF program is
1767 supposed to read and for which it should apply a verdict.
1768 .IP \(bu 2
1769 An eBPF program only cares to read the first \fIbytes\fP of a
1770 \fImsg\fP\&. If the message has a large payload, then setting up
1771 and calling the eBPF program repeatedly for all bytes, even
1772 though the verdict is already known, would create unnecessary
1773 overhead.
1774 .UNINDENT
1775 .sp
1776 When called from within an eBPF program, the helper sets a
1777 counter internal to the BPF infrastructure, that is used to
1778 apply the last verdict to the next \fIbytes\fP\&. If \fIbytes\fP is
1779 smaller than the current data being processed from a
1780 \fBsendmsg\fP() or \fBsendfile\fP() system call, the first
1781 \fIbytes\fP will be sent and the eBPF program will be re\-run with
1782 the pointer for start of data pointing to byte number \fIbytes\fP
1783 \fB+ 1\fP\&. If \fIbytes\fP is larger than the current data being
1784 processed, then the eBPF verdict will be applied to multiple
1785 \fBsendmsg\fP() or \fBsendfile\fP() calls until \fIbytes\fP are
1786 consumed.
1787 .sp
1788 Note that if a socket closes with the internal counter holding
1789 a non\-zero value, this is not a problem because data is not
1790 being buffered for \fIbytes\fP and is sent as it is received.
1791 .TP
1792 .B Return
1793 0
1794 .UNINDENT
1795 .TP
1796 .B \fBlong bpf_msg_cork_bytes(struct sk_msg_buff *\fP\fImsg\fP\fB, u32\fP \fIbytes\fP\fB)\fP
1797 .INDENT 7.0
1798 .TP
1799 .B Description
1800 For socket policies, prevent the execution of the verdict eBPF
1801 program for message \fImsg\fP until \fIbytes\fP (byte number) have been
1802 accumulated.
1803 .sp
1804 This can be used when one needs a specific number of bytes
1805 before a verdict can be assigned, even if the data spans
1806 multiple \fBsendmsg\fP() or \fBsendfile\fP() calls. The extreme
1807 case would be a user calling \fBsendmsg\fP() repeatedly with
1808 1\-byte long message segments. Obviously, this is bad for
1809 performance, but it is still valid. If the eBPF program needs
1810 \fIbytes\fP bytes to validate a header, this helper can be used to
1811 prevent the eBPF program to be called again until \fIbytes\fP have
1812 been accumulated.
1813 .TP
1814 .B Return
1815 0
1816 .UNINDENT
1817 .TP
1818 .B \fBlong bpf_msg_pull_data(struct sk_msg_buff *\fP\fImsg\fP\fB, u32\fP \fIstart\fP\fB, u32\fP \fIend\fP\fB, u64\fP \fIflags\fP\fB)\fP
1819 .INDENT 7.0
1820 .TP
1821 .B Description
1822 For socket policies, pull in non\-linear data from user space
1823 for \fImsg\fP and set pointers \fImsg\fP\fB\->data\fP and \fImsg\fP\fB\->data_end\fP to \fIstart\fP and \fIend\fP bytes offsets into \fImsg\fP,
1824 respectively.
1825 .sp
1826 If a program of type \fBBPF_PROG_TYPE_SK_MSG\fP is run on a
1827 \fImsg\fP it can only parse data that the (\fBdata\fP, \fBdata_end\fP)
1828 pointers have already consumed. For \fBsendmsg\fP() hooks this
1829 is likely the first scatterlist element. But for calls relying
1830 on the \fBsendpage\fP handler (e.g. \fBsendfile\fP()) this will
1831 be the range (\fB0\fP, \fB0\fP) because the data is shared with
1832 user space and by default the objective is to avoid allowing
1833 user space to modify data while (or after) eBPF verdict is
1834 being decided. This helper can be used to pull in data and to
1835 set the start and end pointer to given values. Data will be
1836 copied if necessary (i.e. if data was not linear and if start
1837 and end pointers do not point to the same chunk).
1838 .sp
1839 A call to this helper is susceptible to change the underlying
1840 packet buffer. Therefore, at load time, all checks on pointers
1841 previously done by the verifier are invalidated and must be
1842 performed again, if the helper is used in combination with
1843 direct packet access.
1844 .sp
1845 All values for \fIflags\fP are reserved for future usage, and must
1846 be left at zero.
1847 .TP
1848 .B Return
1849 0 on success, or a negative error in case of failure.
1850 .UNINDENT
1851 .TP
1852 .B \fBlong bpf_bind(struct bpf_sock_addr *\fP\fIctx\fP\fB, struct sockaddr *\fP\fIaddr\fP\fB, int\fP \fIaddr_len\fP\fB)\fP
1853 .INDENT 7.0
1854 .TP
1855 .B Description
1856 Bind the socket associated to \fIctx\fP to the address pointed by
1857 \fIaddr\fP, of length \fIaddr_len\fP\&. This allows for making outgoing
1858 connection from the desired IP address, which can be useful for
1859 example when all processes inside a cgroup should use one
1860 single IP address on a host that has multiple IP configured.
1861 .sp
1862 This helper works for IPv4 and IPv6, TCP and UDP sockets. The
1863 domain (\fIaddr\fP\fB\->sa_family\fP) must be \fBAF_INET\fP (or
1864 \fBAF_INET6\fP). It\(aqs advised to pass zero port (\fBsin_port\fP
1865 or \fBsin6_port\fP) which triggers IP_BIND_ADDRESS_NO_PORT\-like
1866 behavior and lets the kernel efficiently pick up an unused
1867 port as long as 4\-tuple is unique. Passing non\-zero port might
1868 lead to degraded performance.
1869 .TP
1870 .B Return
1871 0 on success, or a negative error in case of failure.
1872 .UNINDENT
1873 .TP
1874 .B \fBlong bpf_xdp_adjust_tail(struct xdp_buff *\fP\fIxdp_md\fP\fB, int\fP \fIdelta\fP\fB)\fP
1875 .INDENT 7.0
1876 .TP
1877 .B Description
1878 Adjust (move) \fIxdp_md\fP\fB\->data_end\fP by \fIdelta\fP bytes. It is
1879 possible to both shrink and grow the packet tail.
1880 Shrink done via \fIdelta\fP being a negative integer.
1881 .sp
1882 A call to this helper is susceptible to change the underlying
1883 packet buffer. Therefore, at load time, all checks on pointers
1884 previously done by the verifier are invalidated and must be
1885 performed again, if the helper is used in combination with
1886 direct packet access.
1887 .TP
1888 .B Return
1889 0 on success, or a negative error in case of failure.
1890 .UNINDENT
1891 .TP
1892 .B \fBlong bpf_skb_get_xfrm_state(struct sk_buff *\fP\fIskb\fP\fB, u32\fP \fIindex\fP\fB, struct bpf_xfrm_state *\fP\fIxfrm_state\fP\fB, u32\fP \fIsize\fP\fB, u64\fP \fIflags\fP\fB)\fP
1893 .INDENT 7.0
1894 .TP
1895 .B Description
1896 Retrieve the XFRM state (IP transform framework, see also
1897 \fBip\-xfrm(8)\fP) at \fIindex\fP in XFRM \(dqsecurity path\(dq for \fIskb\fP\&.
1898 .sp
1899 The retrieved value is stored in the \fBstruct bpf_xfrm_state\fP
1900 pointed by \fIxfrm_state\fP and of length \fIsize\fP\&.
1901 .sp
1902 All values for \fIflags\fP are reserved for future usage, and must
1903 be left at zero.
1904 .sp
1905 This helper is available only if the kernel was compiled with
1906 \fBCONFIG_XFRM\fP configuration option.
1907 .TP
1908 .B Return
1909 0 on success, or a negative error in case of failure.
1910 .UNINDENT
1911 .TP
1912 .B \fBlong bpf_get_stack(void *\fP\fIctx\fP\fB, void *\fP\fIbuf\fP\fB, u32\fP \fIsize\fP\fB, u64\fP \fIflags\fP\fB)\fP
1913 .INDENT 7.0
1914 .TP
1915 .B Description
1916 Return a user or a kernel stack in bpf program provided buffer.
1917 To achieve this, the helper needs \fIctx\fP, which is a pointer
1918 to the context on which the tracing program is executed.
1919 To store the stacktrace, the bpf program provides \fIbuf\fP with
1920 a nonnegative \fIsize\fP\&.
1921 .sp
1922 The last argument, \fIflags\fP, holds the number of stack frames to
1923 skip (from 0 to 255), masked with
1924 \fBBPF_F_SKIP_FIELD_MASK\fP\&. The next bits can be used to set
1925 the following flags:
1926 .INDENT 7.0
1927 .TP
1928 .B \fBBPF_F_USER_STACK\fP
1929 Collect a user space stack instead of a kernel stack.
1930 .TP
1931 .B \fBBPF_F_USER_BUILD_ID\fP
1932 Collect (build_id, file_offset) instead of ips for user
1933 stack, only valid if \fBBPF_F_USER_STACK\fP is also
1934 specified.
1935 .sp
1936 \fIfile_offset\fP is an offset relative to the beginning
1937 of the executable or shared object file backing the vma
1938 which the \fIip\fP falls in. It is \fInot\fP an offset relative
1939 to that object\(aqs base address. Accordingly, it must be
1940 adjusted by adding (sh_addr \- sh_offset), where
1941 sh_{addr,offset} correspond to the executable section
1942 containing \fIfile_offset\fP in the object, for comparisons
1943 to symbols\(aq st_value to be valid.
1944 .UNINDENT
1945 .sp
1946 \fBbpf_get_stack\fP() can collect up to
1947 \fBPERF_MAX_STACK_DEPTH\fP both kernel and user frames, subject
1948 to sufficient large buffer size. Note that
1949 this limit can be controlled with the \fBsysctl\fP program, and
1950 that it should be manually increased in order to profile long
1951 user stacks (such as stacks for Java programs). To do so, use:
1952 .INDENT 7.0
1953 .INDENT 3.5
1954 .sp
1955 .EX
1956 # sysctl kernel.perf_event_max_stack=<new value>
1957 .EE
1958 .UNINDENT
1959 .UNINDENT
1960 .TP
1961 .B Return
1962 The non\-negative copied \fIbuf\fP length equal to or less than
1963 \fIsize\fP on success, or a negative error in case of failure.
1964 .UNINDENT
1965 .TP
1966 .B \fBlong bpf_skb_load_bytes_relative(const void *\fP\fIskb\fP\fB, u32\fP \fIoffset\fP\fB, void *\fP\fIto\fP\fB, u32\fP \fIlen\fP\fB, u32\fP \fIstart_header\fP\fB)\fP
1967 .INDENT 7.0
1968 .TP
1969 .B Description
1970 This helper is similar to \fBbpf_skb_load_bytes\fP() in that
1971 it provides an easy way to load \fIlen\fP bytes from \fIoffset\fP
1972 from the packet associated to \fIskb\fP, into the buffer pointed
1973 by \fIto\fP\&. The difference to \fBbpf_skb_load_bytes\fP() is that
1974 a fifth argument \fIstart_header\fP exists in order to select a
1975 base offset to start from. \fIstart_header\fP can be one of:
1976 .INDENT 7.0
1977 .TP
1978 .B \fBBPF_HDR_START_MAC\fP
1979 Base offset to load data from is \fIskb\fP\(aqs mac header.
1980 .TP
1981 .B \fBBPF_HDR_START_NET\fP
1982 Base offset to load data from is \fIskb\fP\(aqs network header.
1983 .UNINDENT
1984 .sp
1985 In general, \(dqdirect packet access\(dq is the preferred method to
1986 access packet data, however, this helper is in particular useful
1987 in socket filters where \fIskb\fP\fB\->data\fP does not always point
1988 to the start of the mac header and where \(dqdirect packet access\(dq
1989 is not available.
1990 .TP
1991 .B Return
1992 0 on success, or a negative error in case of failure.
1993 .UNINDENT
1994 .TP
1995 .B \fBlong bpf_fib_lookup(void *\fP\fIctx\fP\fB, struct bpf_fib_lookup *\fP\fIparams\fP\fB, int\fP \fIplen\fP\fB, u32\fP \fIflags\fP\fB)\fP
1996 .INDENT 7.0
1997 .TP
1998 .B Description
1999 Do FIB lookup in kernel tables using parameters in \fIparams\fP\&.
2000 If lookup is successful and result shows packet is to be
2001 forwarded, the neighbor tables are searched for the nexthop.
2002 If successful (ie., FIB lookup shows forwarding and nexthop
2003 is resolved), the nexthop address is returned in ipv4_dst
2004 or ipv6_dst based on family, smac is set to mac address of
2005 egress device, dmac is set to nexthop mac address, rt_metric
2006 is set to metric from route (IPv4/IPv6 only), and ifindex
2007 is set to the device index of the nexthop from the FIB lookup.
2008 .sp
2009 \fIplen\fP argument is the size of the passed in struct.
2010 \fIflags\fP argument can be a combination of one or more of the
2011 following values:
2012 .INDENT 7.0
2013 .TP
2014 .B \fBBPF_FIB_LOOKUP_DIRECT\fP
2015 Do a direct table lookup vs full lookup using FIB
2016 rules.
2017 .TP
2018 .B \fBBPF_FIB_LOOKUP_TBID\fP
2019 Used with BPF_FIB_LOOKUP_DIRECT.
2020 Use the routing table ID present in \fIparams\fP\->tbid
2021 for the fib lookup.
2022 .TP
2023 .B \fBBPF_FIB_LOOKUP_OUTPUT\fP
2024 Perform lookup from an egress perspective (default is
2025 ingress).
2026 .TP
2027 .B \fBBPF_FIB_LOOKUP_SKIP_NEIGH\fP
2028 Skip the neighbour table lookup. \fIparams\fP\->dmac
2029 and \fIparams\fP\->smac will not be set as output. A common
2030 use case is to call \fBbpf_redirect_neigh\fP() after
2031 doing \fBbpf_fib_lookup\fP().
2032 .TP
2033 .B \fBBPF_FIB_LOOKUP_SRC\fP
2034 Derive and set source IP addr in \fIparams\fP\->ipv{4,6}_src
2035 for the nexthop. If the src addr cannot be derived,
2036 \fBBPF_FIB_LKUP_RET_NO_SRC_ADDR\fP is returned. In this
2037 case, \fIparams\fP\->dmac and \fIparams\fP\->smac are not set either.
2038 .TP
2039 .B \fBBPF_FIB_LOOKUP_MARK\fP
2040 Use the mark present in \fIparams\fP\->mark for the fib lookup.
2041 This option should not be used with BPF_FIB_LOOKUP_DIRECT,
2042 as it only has meaning for full lookups.
2043 .UNINDENT
2044 .sp
2045 \fIctx\fP is either \fBstruct xdp_md\fP for XDP programs or
2046 \fBstruct sk_buff\fP tc cls_act programs.
2047 .TP
2048 .B Return
2049 .INDENT 7.0
2050 .IP \(bu 2
2051 < 0 if any input argument is invalid
2052 .IP \(bu 2
2053 0 on success (packet is forwarded, nexthop neighbor exists)
2054 .IP \(bu 2
2055 > 0 one of \fBBPF_FIB_LKUP_RET_\fP codes explaining why the
2056 packet is not forwarded or needs assist from full stack
2057 .UNINDENT
2058 .sp
2059 If lookup fails with BPF_FIB_LKUP_RET_FRAG_NEEDED, then the MTU
2060 was exceeded and output params\->mtu_result contains the MTU.
2061 .UNINDENT
2062 .TP
2063 .B \fBlong bpf_sock_hash_update(struct bpf_sock_ops *\fP\fIskops\fP\fB, struct bpf_map *\fP\fImap\fP\fB, void *\fP\fIkey\fP\fB, u64\fP \fIflags\fP\fB)\fP
2064 .INDENT 7.0
2065 .TP
2066 .B Description
2067 Add an entry to, or update a sockhash \fImap\fP referencing sockets.
2068 The \fIskops\fP is used as a new value for the entry associated to
2069 \fIkey\fP\&. \fIflags\fP is one of:
2070 .INDENT 7.0
2071 .TP
2072 .B \fBBPF_NOEXIST\fP
2073 The entry for \fIkey\fP must not exist in the map.
2074 .TP
2075 .B \fBBPF_EXIST\fP
2076 The entry for \fIkey\fP must already exist in the map.
2077 .TP
2078 .B \fBBPF_ANY\fP
2079 No condition on the existence of the entry for \fIkey\fP\&.
2080 .UNINDENT
2081 .sp
2082 If the \fImap\fP has eBPF programs (parser and verdict), those will
2083 be inherited by the socket being added. If the socket is
2084 already attached to eBPF programs, this results in an error.
2085 .TP
2086 .B Return
2087 0 on success, or a negative error in case of failure.
2088 .UNINDENT
2089 .TP
2090 .B \fBlong bpf_msg_redirect_hash(struct sk_msg_buff *\fP\fImsg\fP\fB, struct bpf_map *\fP\fImap\fP\fB, void *\fP\fIkey\fP\fB, u64\fP \fIflags\fP\fB)\fP
2091 .INDENT 7.0
2092 .TP
2093 .B Description
2094 This helper is used in programs implementing policies at the
2095 socket level. If the message \fImsg\fP is allowed to pass (i.e. if
2096 the verdict eBPF program returns \fBSK_PASS\fP), redirect it to
2097 the socket referenced by \fImap\fP (of type
2098 \fBBPF_MAP_TYPE_SOCKHASH\fP) using hash \fIkey\fP\&. Both ingress and
2099 egress interfaces can be used for redirection. The
2100 \fBBPF_F_INGRESS\fP value in \fIflags\fP is used to make the
2101 distinction (ingress path is selected if the flag is present,
2102 egress path otherwise). This is the only flag supported for now.
2103 .TP
2104 .B Return
2105 \fBSK_PASS\fP on success, or \fBSK_DROP\fP on error.
2106 .UNINDENT
2107 .TP
2108 .B \fBlong bpf_sk_redirect_hash(struct sk_buff *\fP\fIskb\fP\fB, struct bpf_map *\fP\fImap\fP\fB, void *\fP\fIkey\fP\fB, u64\fP \fIflags\fP\fB)\fP
2109 .INDENT 7.0
2110 .TP
2111 .B Description
2112 This helper is used in programs implementing policies at the
2113 skb socket level. If the sk_buff \fIskb\fP is allowed to pass (i.e.
2114 if the verdict eBPF program returns \fBSK_PASS\fP), redirect it
2115 to the socket referenced by \fImap\fP (of type
2116 \fBBPF_MAP_TYPE_SOCKHASH\fP) using hash \fIkey\fP\&. Both ingress and
2117 egress interfaces can be used for redirection. The
2118 \fBBPF_F_INGRESS\fP value in \fIflags\fP is used to make the
2119 distinction (ingress path is selected if the flag is present,
2120 egress otherwise). This is the only flag supported for now.
2121 .TP
2122 .B Return
2123 \fBSK_PASS\fP on success, or \fBSK_DROP\fP on error.
2124 .UNINDENT
2125 .TP
2126 .B \fBlong bpf_lwt_push_encap(struct sk_buff *\fP\fIskb\fP\fB, u32\fP \fItype\fP\fB, void *\fP\fIhdr\fP\fB, u32\fP \fIlen\fP\fB)\fP
2127 .INDENT 7.0
2128 .TP
2129 .B Description
2130 Encapsulate the packet associated to \fIskb\fP within a Layer 3
2131 protocol header. This header is provided in the buffer at
2132 address \fIhdr\fP, with \fIlen\fP its size in bytes. \fItype\fP indicates
2133 the protocol of the header and can be one of:
2134 .INDENT 7.0
2135 .TP
2136 .B \fBBPF_LWT_ENCAP_SEG6\fP
2137 IPv6 encapsulation with Segment Routing Header
2138 (\fBstruct ipv6_sr_hdr\fP). \fIhdr\fP only contains the SRH,
2139 the IPv6 header is computed by the kernel.
2140 .TP
2141 .B \fBBPF_LWT_ENCAP_SEG6_INLINE\fP
2142 Only works if \fIskb\fP contains an IPv6 packet. Insert a
2143 Segment Routing Header (\fBstruct ipv6_sr_hdr\fP) inside
2144 the IPv6 header.
2145 .TP
2146 .B \fBBPF_LWT_ENCAP_IP\fP
2147 IP encapsulation (GRE/GUE/IPIP/etc). The outer header
2148 must be IPv4 or IPv6, followed by zero or more
2149 additional headers, up to \fBLWT_BPF_MAX_HEADROOM\fP
2150 total bytes in all prepended headers. Please note that
2151 if \fBskb_is_gso\fP(\fIskb\fP) is true, no more than two
2152 headers can be prepended, and the inner header, if
2153 present, should be either GRE or UDP/GUE.
2154 .UNINDENT
2155 .sp
2156 \fBBPF_LWT_ENCAP_SEG6\fP* types can be called by BPF programs
2157 of type \fBBPF_PROG_TYPE_LWT_IN\fP; \fBBPF_LWT_ENCAP_IP\fP type can
2158 be called by bpf programs of types \fBBPF_PROG_TYPE_LWT_IN\fP and
2159 \fBBPF_PROG_TYPE_LWT_XMIT\fP\&.
2160 .sp
2161 A call to this helper is susceptible to change the underlying
2162 packet buffer. Therefore, at load time, all checks on pointers
2163 previously done by the verifier are invalidated and must be
2164 performed again, if the helper is used in combination with
2165 direct packet access.
2166 .TP
2167 .B Return
2168 0 on success, or a negative error in case of failure.
2169 .UNINDENT
2170 .TP
2171 .B \fBlong bpf_lwt_seg6_store_bytes(struct sk_buff *\fP\fIskb\fP\fB, u32\fP \fIoffset\fP\fB, const void *\fP\fIfrom\fP\fB, u32\fP \fIlen\fP\fB)\fP
2172 .INDENT 7.0
2173 .TP
2174 .B Description
2175 Store \fIlen\fP bytes from address \fIfrom\fP into the packet
2176 associated to \fIskb\fP, at \fIoffset\fP\&. Only the flags, tag and TLVs
2177 inside the outermost IPv6 Segment Routing Header can be
2178 modified through this helper.
2179 .sp
2180 A call to this helper is susceptible to change the underlying
2181 packet buffer. Therefore, at load time, all checks on pointers
2182 previously done by the verifier are invalidated and must be
2183 performed again, if the helper is used in combination with
2184 direct packet access.
2185 .TP
2186 .B Return
2187 0 on success, or a negative error in case of failure.
2188 .UNINDENT
2189 .TP
2190 .B \fBlong bpf_lwt_seg6_adjust_srh(struct sk_buff *\fP\fIskb\fP\fB, u32\fP \fIoffset\fP\fB, s32\fP \fIdelta\fP\fB)\fP
2191 .INDENT 7.0
2192 .TP
2193 .B Description
2194 Adjust the size allocated to TLVs in the outermost IPv6
2195 Segment Routing Header contained in the packet associated to
2196 \fIskb\fP, at position \fIoffset\fP by \fIdelta\fP bytes. Only offsets
2197 after the segments are accepted. \fIdelta\fP can be as well
2198 positive (growing) as negative (shrinking).
2199 .sp
2200 A call to this helper is susceptible to change the underlying
2201 packet buffer. Therefore, at load time, all checks on pointers
2202 previously done by the verifier are invalidated and must be
2203 performed again, if the helper is used in combination with
2204 direct packet access.
2205 .TP
2206 .B Return
2207 0 on success, or a negative error in case of failure.
2208 .UNINDENT
2209 .TP
2210 .B \fBlong bpf_lwt_seg6_action(struct sk_buff *\fP\fIskb\fP\fB, u32\fP \fIaction\fP\fB, void *\fP\fIparam\fP\fB, u32\fP \fIparam_len\fP\fB)\fP
2211 .INDENT 7.0
2212 .TP
2213 .B Description
2214 Apply an IPv6 Segment Routing action of type \fIaction\fP to the
2215 packet associated to \fIskb\fP\&. Each action takes a parameter
2216 contained at address \fIparam\fP, and of length \fIparam_len\fP bytes.
2217 \fIaction\fP can be one of:
2218 .INDENT 7.0
2219 .TP
2220 .B \fBSEG6_LOCAL_ACTION_END_X\fP
2221 End.X action: Endpoint with Layer\-3 cross\-connect.
2222 Type of \fIparam\fP: \fBstruct in6_addr\fP\&.
2223 .TP
2224 .B \fBSEG6_LOCAL_ACTION_END_T\fP
2225 End.T action: Endpoint with specific IPv6 table lookup.
2226 Type of \fIparam\fP: \fBint\fP\&.
2227 .TP
2228 .B \fBSEG6_LOCAL_ACTION_END_B6\fP
2229 End.B6 action: Endpoint bound to an SRv6 policy.
2230 Type of \fIparam\fP: \fBstruct ipv6_sr_hdr\fP\&.
2231 .TP
2232 .B \fBSEG6_LOCAL_ACTION_END_B6_ENCAP\fP
2233 End.B6.Encap action: Endpoint bound to an SRv6
2234 encapsulation policy.
2235 Type of \fIparam\fP: \fBstruct ipv6_sr_hdr\fP\&.
2236 .UNINDENT
2237 .sp
2238 A call to this helper is susceptible to change the underlying
2239 packet buffer. Therefore, at load time, all checks on pointers
2240 previously done by the verifier are invalidated and must be
2241 performed again, if the helper is used in combination with
2242 direct packet access.
2243 .TP
2244 .B Return
2245 0 on success, or a negative error in case of failure.
2246 .UNINDENT
2247 .TP
2248 .B \fBlong bpf_rc_repeat(void *\fP\fIctx\fP\fB)\fP
2249 .INDENT 7.0
2250 .TP
2251 .B Description
2252 This helper is used in programs implementing IR decoding, to
2253 report a successfully decoded repeat key message. This delays
2254 the generation of a key up event for previously generated
2255 key down event.
2256 .sp
2257 Some IR protocols like NEC have a special IR message for
2258 repeating last button, for when a button is held down.
2259 .sp
2260 The \fIctx\fP should point to the lirc sample as passed into
2261 the program.
2262 .sp
2263 This helper is only available is the kernel was compiled with
2264 the \fBCONFIG_BPF_LIRC_MODE2\fP configuration option set to
2265 \(dq\fBy\fP\(dq.
2266 .TP
2267 .B Return
2268 0
2269 .UNINDENT
2270 .TP
2271 .B \fBlong bpf_rc_keydown(void *\fP\fIctx\fP\fB, u32\fP \fIprotocol\fP\fB, u64\fP \fIscancode\fP\fB, u32\fP \fItoggle\fP\fB)\fP
2272 .INDENT 7.0
2273 .TP
2274 .B Description
2275 This helper is used in programs implementing IR decoding, to
2276 report a successfully decoded key press with \fIscancode\fP,
2277 \fItoggle\fP value in the given \fIprotocol\fP\&. The scancode will be
2278 translated to a keycode using the rc keymap, and reported as
2279 an input key down event. After a period a key up event is
2280 generated. This period can be extended by calling either
2281 \fBbpf_rc_keydown\fP() again with the same values, or calling
2282 \fBbpf_rc_repeat\fP().
2283 .sp
2284 Some protocols include a toggle bit, in case the button was
2285 released and pressed again between consecutive scancodes.
2286 .sp
2287 The \fIctx\fP should point to the lirc sample as passed into
2288 the program.
2289 .sp
2290 The \fIprotocol\fP is the decoded protocol number (see
2291 \fBenum rc_proto\fP for some predefined values).
2292 .sp
2293 This helper is only available is the kernel was compiled with
2294 the \fBCONFIG_BPF_LIRC_MODE2\fP configuration option set to
2295 \(dq\fBy\fP\(dq.
2296 .TP
2297 .B Return
2298 0
2299 .UNINDENT
2300 .TP
2301 .B \fBu64 bpf_skb_cgroup_id(struct sk_buff *\fP\fIskb\fP\fB)\fP
2302 .INDENT 7.0
2303 .TP
2304 .B Description
2305 Return the cgroup v2 id of the socket associated with the \fIskb\fP\&.
2306 This is roughly similar to the \fBbpf_get_cgroup_classid\fP()
2307 helper for cgroup v1 by providing a tag resp. identifier that
2308 can be matched on or used for map lookups e.g. to implement
2309 policy. The cgroup v2 id of a given path in the hierarchy is
2310 exposed in user space through the f_handle API in order to get
2311 to the same 64\-bit id.
2312 .sp
2313 This helper can be used on TC egress path, but not on ingress,
2314 and is available only if the kernel was compiled with the
2315 \fBCONFIG_SOCK_CGROUP_DATA\fP configuration option.
2316 .TP
2317 .B Return
2318 The id is returned or 0 in case the id could not be retrieved.
2319 .UNINDENT
2320 .TP
2321 .B \fBu64 bpf_get_current_cgroup_id(void)\fP
2322 .INDENT 7.0
2323 .TP
2324 .B Description
2325 Get the current cgroup id based on the cgroup within which
2326 the current task is running.
2327 .TP
2328 .B Return
2329 A 64\-bit integer containing the current cgroup id based
2330 on the cgroup within which the current task is running.
2331 .UNINDENT
2332 .TP
2333 .B \fBvoid *bpf_get_local_storage(void *\fP\fImap\fP\fB, u64\fP \fIflags\fP\fB)\fP
2334 .INDENT 7.0
2335 .TP
2336 .B Description
2337 Get the pointer to the local storage area.
2338 The type and the size of the local storage is defined
2339 by the \fImap\fP argument.
2340 The \fIflags\fP meaning is specific for each map type,
2341 and has to be 0 for cgroup local storage.
2342 .sp
2343 Depending on the BPF program type, a local storage area
2344 can be shared between multiple instances of the BPF program,
2345 running simultaneously.
2346 .sp
2347 A user should care about the synchronization by himself.
2348 For example, by using the \fBBPF_ATOMIC\fP instructions to alter
2349 the shared data.
2350 .TP
2351 .B Return
2352 A pointer to the local storage area.
2353 .UNINDENT
2354 .TP
2355 .B \fBlong bpf_sk_select_reuseport(struct sk_reuseport_md *\fP\fIreuse\fP\fB, struct bpf_map *\fP\fImap\fP\fB, void *\fP\fIkey\fP\fB, u64\fP \fIflags\fP\fB)\fP
2356 .INDENT 7.0
2357 .TP
2358 .B Description
2359 Select a \fBSO_REUSEPORT\fP socket from a
2360 \fBBPF_MAP_TYPE_REUSEPORT_SOCKARRAY\fP \fImap\fP\&.
2361 It checks the selected socket is matching the incoming
2362 request in the socket buffer.
2363 .TP
2364 .B Return
2365 0 on success, or a negative error in case of failure.
2366 .UNINDENT
2367 .TP
2368 .B \fBu64 bpf_skb_ancestor_cgroup_id(struct sk_buff *\fP\fIskb\fP\fB, int\fP \fIancestor_level\fP\fB)\fP
2369 .INDENT 7.0
2370 .TP
2371 .B Description
2372 Return id of cgroup v2 that is ancestor of cgroup associated
2373 with the \fIskb\fP at the \fIancestor_level\fP\&. The root cgroup is at
2374 \fIancestor_level\fP zero and each step down the hierarchy
2375 increments the level. If \fIancestor_level\fP == level of cgroup
2376 associated with \fIskb\fP, then return value will be same as that
2377 of \fBbpf_skb_cgroup_id\fP().
2378 .sp
2379 The helper is useful to implement policies based on cgroups
2380 that are upper in hierarchy than immediate cgroup associated
2381 with \fIskb\fP\&.
2382 .sp
2383 The format of returned id and helper limitations are same as in
2384 \fBbpf_skb_cgroup_id\fP().
2385 .TP
2386 .B Return
2387 The id is returned or 0 in case the id could not be retrieved.
2388 .UNINDENT
2389 .TP
2390 .B \fBstruct bpf_sock *bpf_sk_lookup_tcp(void *\fP\fIctx\fP\fB, struct bpf_sock_tuple *\fP\fItuple\fP\fB, u32\fP \fItuple_size\fP\fB, u64\fP \fInetns\fP\fB, u64\fP \fIflags\fP\fB)\fP
2391 .INDENT 7.0
2392 .TP
2393 .B Description
2394 Look for TCP socket matching \fItuple\fP, optionally in a child
2395 network namespace \fInetns\fP\&. The return value must be checked,
2396 and if non\-\fBNULL\fP, released via \fBbpf_sk_release\fP().
2397 .sp
2398 The \fIctx\fP should point to the context of the program, such as
2399 the skb or socket (depending on the hook in use). This is used
2400 to determine the base network namespace for the lookup.
2401 .sp
2402 \fItuple_size\fP must be one of:
2403 .INDENT 7.0
2404 .TP
2405 .B \fBsizeof\fP(\fItuple\fP\fB\->ipv4\fP)
2406 Look for an IPv4 socket.
2407 .TP
2408 .B \fBsizeof\fP(\fItuple\fP\fB\->ipv6\fP)
2409 Look for an IPv6 socket.
2410 .UNINDENT
2411 .sp
2412 If the \fInetns\fP is a negative signed 32\-bit integer, then the
2413 socket lookup table in the netns associated with the \fIctx\fP
2414 will be used. For the TC hooks, this is the netns of the device
2415 in the skb. For socket hooks, this is the netns of the socket.
2416 If \fInetns\fP is any other signed 32\-bit value greater than or
2417 equal to zero then it specifies the ID of the netns relative to
2418 the netns associated with the \fIctx\fP\&. \fInetns\fP values beyond the
2419 range of 32\-bit integers are reserved for future use.
2420 .sp
2421 All values for \fIflags\fP are reserved for future usage, and must
2422 be left at zero.
2423 .sp
2424 This helper is available only if the kernel was compiled with
2425 \fBCONFIG_NET\fP configuration option.
2426 .TP
2427 .B Return
2428 Pointer to \fBstruct bpf_sock\fP, or \fBNULL\fP in case of failure.
2429 For sockets with reuseport option, the \fBstruct bpf_sock\fP
2430 result is from \fIreuse\fP\fB\->socks\fP[] using the hash of the
2431 tuple.
2432 .UNINDENT
2433 .TP
2434 .B \fBstruct bpf_sock *bpf_sk_lookup_udp(void *\fP\fIctx\fP\fB, struct bpf_sock_tuple *\fP\fItuple\fP\fB, u32\fP \fItuple_size\fP\fB, u64\fP \fInetns\fP\fB, u64\fP \fIflags\fP\fB)\fP
2435 .INDENT 7.0
2436 .TP
2437 .B Description
2438 Look for UDP socket matching \fItuple\fP, optionally in a child
2439 network namespace \fInetns\fP\&. The return value must be checked,
2440 and if non\-\fBNULL\fP, released via \fBbpf_sk_release\fP().
2441 .sp
2442 The \fIctx\fP should point to the context of the program, such as
2443 the skb or socket (depending on the hook in use). This is used
2444 to determine the base network namespace for the lookup.
2445 .sp
2446 \fItuple_size\fP must be one of:
2447 .INDENT 7.0
2448 .TP
2449 .B \fBsizeof\fP(\fItuple\fP\fB\->ipv4\fP)
2450 Look for an IPv4 socket.
2451 .TP
2452 .B \fBsizeof\fP(\fItuple\fP\fB\->ipv6\fP)
2453 Look for an IPv6 socket.
2454 .UNINDENT
2455 .sp
2456 If the \fInetns\fP is a negative signed 32\-bit integer, then the
2457 socket lookup table in the netns associated with the \fIctx\fP
2458 will be used. For the TC hooks, this is the netns of the device
2459 in the skb. For socket hooks, this is the netns of the socket.
2460 If \fInetns\fP is any other signed 32\-bit value greater than or
2461 equal to zero then it specifies the ID of the netns relative to
2462 the netns associated with the \fIctx\fP\&. \fInetns\fP values beyond the
2463 range of 32\-bit integers are reserved for future use.
2464 .sp
2465 All values for \fIflags\fP are reserved for future usage, and must
2466 be left at zero.
2467 .sp
2468 This helper is available only if the kernel was compiled with
2469 \fBCONFIG_NET\fP configuration option.
2470 .TP
2471 .B Return
2472 Pointer to \fBstruct bpf_sock\fP, or \fBNULL\fP in case of failure.
2473 For sockets with reuseport option, the \fBstruct bpf_sock\fP
2474 result is from \fIreuse\fP\fB\->socks\fP[] using the hash of the
2475 tuple.
2476 .UNINDENT
2477 .TP
2478 .B \fBlong bpf_sk_release(void *\fP\fIsock\fP\fB)\fP
2479 .INDENT 7.0
2480 .TP
2481 .B Description
2482 Release the reference held by \fIsock\fP\&. \fIsock\fP must be a
2483 non\-\fBNULL\fP pointer that was returned from
2484 \fBbpf_sk_lookup_xxx\fP().
2485 .TP
2486 .B Return
2487 0 on success, or a negative error in case of failure.
2488 .UNINDENT
2489 .TP
2490 .B \fBlong bpf_map_push_elem(struct bpf_map *\fP\fImap\fP\fB, const void *\fP\fIvalue\fP\fB, u64\fP \fIflags\fP\fB)\fP
2491 .INDENT 7.0
2492 .TP
2493 .B Description
2494 Push an element \fIvalue\fP in \fImap\fP\&. \fIflags\fP is one of:
2495 .INDENT 7.0
2496 .TP
2497 .B \fBBPF_EXIST\fP
2498 If the queue/stack is full, the oldest element is
2499 removed to make room for this.
2500 .UNINDENT
2501 .TP
2502 .B Return
2503 0 on success, or a negative error in case of failure.
2504 .UNINDENT
2505 .TP
2506 .B \fBlong bpf_map_pop_elem(struct bpf_map *\fP\fImap\fP\fB, void *\fP\fIvalue\fP\fB)\fP
2507 .INDENT 7.0
2508 .TP
2509 .B Description
2510 Pop an element from \fImap\fP\&.
2511 .TP
2512 .B Return
2513 0 on success, or a negative error in case of failure.
2514 .UNINDENT
2515 .TP
2516 .B \fBlong bpf_map_peek_elem(struct bpf_map *\fP\fImap\fP\fB, void *\fP\fIvalue\fP\fB)\fP
2517 .INDENT 7.0
2518 .TP
2519 .B Description
2520 Get an element from \fImap\fP without removing it.
2521 .TP
2522 .B Return
2523 0 on success, or a negative error in case of failure.
2524 .UNINDENT
2525 .TP
2526 .B \fBlong bpf_msg_push_data(struct sk_msg_buff *\fP\fImsg\fP\fB, u32\fP \fIstart\fP\fB, u32\fP \fIlen\fP\fB, u64\fP \fIflags\fP\fB)\fP
2527 .INDENT 7.0
2528 .TP
2529 .B Description
2530 For socket policies, insert \fIlen\fP bytes into \fImsg\fP at offset
2531 \fIstart\fP\&.
2532 .sp
2533 If a program of type \fBBPF_PROG_TYPE_SK_MSG\fP is run on a
2534 \fImsg\fP it may want to insert metadata or options into the \fImsg\fP\&.
2535 This can later be read and used by any of the lower layer BPF
2536 hooks.
2537 .sp
2538 This helper may fail if under memory pressure (a malloc
2539 fails) in these cases BPF programs will get an appropriate
2540 error and BPF programs will need to handle them.
2541 .TP
2542 .B Return
2543 0 on success, or a negative error in case of failure.
2544 .UNINDENT
2545 .TP
2546 .B \fBlong bpf_msg_pop_data(struct sk_msg_buff *\fP\fImsg\fP\fB, u32\fP \fIstart\fP\fB, u32\fP \fIlen\fP\fB, u64\fP \fIflags\fP\fB)\fP
2547 .INDENT 7.0
2548 .TP
2549 .B Description
2550 Will remove \fIlen\fP bytes from a \fImsg\fP starting at byte \fIstart\fP\&.
2551 This may result in \fBENOMEM\fP errors under certain situations if
2552 an allocation and copy are required due to a full ring buffer.
2553 However, the helper will try to avoid doing the allocation
2554 if possible. Other errors can occur if input parameters are
2555 invalid either due to \fIstart\fP byte not being valid part of \fImsg\fP
2556 payload and/or \fIpop\fP value being to large.
2557 .TP
2558 .B Return
2559 0 on success, or a negative error in case of failure.
2560 .UNINDENT
2561 .TP
2562 .B \fBlong bpf_rc_pointer_rel(void *\fP\fIctx\fP\fB, s32\fP \fIrel_x\fP\fB, s32\fP \fIrel_y\fP\fB)\fP
2563 .INDENT 7.0
2564 .TP
2565 .B Description
2566 This helper is used in programs implementing IR decoding, to
2567 report a successfully decoded pointer movement.
2568 .sp
2569 The \fIctx\fP should point to the lirc sample as passed into
2570 the program.
2571 .sp
2572 This helper is only available is the kernel was compiled with
2573 the \fBCONFIG_BPF_LIRC_MODE2\fP configuration option set to
2574 \(dq\fBy\fP\(dq.
2575 .TP
2576 .B Return
2577 0
2578 .UNINDENT
2579 .TP
2580 .B \fBlong bpf_spin_lock(struct bpf_spin_lock *\fP\fIlock\fP\fB)\fP
2581 .INDENT 7.0
2582 .TP
2583 .B Description
2584 Acquire a spinlock represented by the pointer \fIlock\fP, which is
2585 stored as part of a value of a map. Taking the lock allows to
2586 safely update the rest of the fields in that value. The
2587 spinlock can (and must) later be released with a call to
2588 \fBbpf_spin_unlock\fP(\fIlock\fP).
2589 .sp
2590 Spinlocks in BPF programs come with a number of restrictions
2591 and constraints:
2592 .INDENT 7.0
2593 .IP \(bu 2
2594 \fBbpf_spin_lock\fP objects are only allowed inside maps of
2595 types \fBBPF_MAP_TYPE_HASH\fP and \fBBPF_MAP_TYPE_ARRAY\fP (this
2596 list could be extended in the future).
2597 .IP \(bu 2
2598 BTF description of the map is mandatory.
2599 .IP \(bu 2
2600 The BPF program can take ONE lock at a time, since taking two
2601 or more could cause dead locks.
2602 .IP \(bu 2
2603 Only one \fBstruct bpf_spin_lock\fP is allowed per map element.
2604 .IP \(bu 2
2605 When the lock is taken, calls (either BPF to BPF or helpers)
2606 are not allowed.
2607 .IP \(bu 2
2608 The \fBBPF_LD_ABS\fP and \fBBPF_LD_IND\fP instructions are not
2609 allowed inside a spinlock\-ed region.
2610 .IP \(bu 2
2611 The BPF program MUST call \fBbpf_spin_unlock\fP() to release
2612 the lock, on all execution paths, before it returns.
2613 .IP \(bu 2
2614 The BPF program can access \fBstruct bpf_spin_lock\fP only via
2615 the \fBbpf_spin_lock\fP() and \fBbpf_spin_unlock\fP()
2616 helpers. Loading or storing data into the \fBstruct
2617 bpf_spin_lock\fP \fIlock\fP\fB;\fP field of a map is not allowed.
2618 .IP \(bu 2
2619 To use the \fBbpf_spin_lock\fP() helper, the BTF description
2620 of the map value must be a struct and have \fBstruct
2621 bpf_spin_lock\fP \fIanyname\fP\fB;\fP field at the top level.
2622 Nested lock inside another struct is not allowed.
2623 .IP \(bu 2
2624 The \fBstruct bpf_spin_lock\fP \fIlock\fP field in a map value must
2625 be aligned on a multiple of 4 bytes in that value.
2626 .IP \(bu 2
2627 Syscall with command \fBBPF_MAP_LOOKUP_ELEM\fP does not copy
2628 the \fBbpf_spin_lock\fP field to user space.
2629 .IP \(bu 2
2630 Syscall with command \fBBPF_MAP_UPDATE_ELEM\fP, or update from
2631 a BPF program, do not update the \fBbpf_spin_lock\fP field.
2632 .IP \(bu 2
2633 \fBbpf_spin_lock\fP cannot be on the stack or inside a
2634 networking packet (it can only be inside of a map values).
2635 .IP \(bu 2
2636 \fBbpf_spin_lock\fP is available to root only.
2637 .IP \(bu 2
2638 Tracing programs and socket filter programs cannot use
2639 \fBbpf_spin_lock\fP() due to insufficient preemption checks
2640 (but this may change in the future).
2641 .IP \(bu 2
2642 \fBbpf_spin_lock\fP is not allowed in inner maps of map\-in\-map.
2643 .UNINDENT
2644 .TP
2645 .B Return
2646 0
2647 .UNINDENT
2648 .TP
2649 .B \fBlong bpf_spin_unlock(struct bpf_spin_lock *\fP\fIlock\fP\fB)\fP
2650 .INDENT 7.0
2651 .TP
2652 .B Description
2653 Release the \fIlock\fP previously locked by a call to
2654 \fBbpf_spin_lock\fP(\fIlock\fP).
2655 .TP
2656 .B Return
2657 0
2658 .UNINDENT
2659 .TP
2660 .B \fBstruct bpf_sock *bpf_sk_fullsock(struct bpf_sock *\fP\fIsk\fP\fB)\fP
2661 .INDENT 7.0
2662 .TP
2663 .B Description
2664 This helper gets a \fBstruct bpf_sock\fP pointer such
2665 that all the fields in this \fBbpf_sock\fP can be accessed.
2666 .TP
2667 .B Return
2668 A \fBstruct bpf_sock\fP pointer on success, or \fBNULL\fP in
2669 case of failure.
2670 .UNINDENT
2671 .TP
2672 .B \fBstruct bpf_tcp_sock *bpf_tcp_sock(struct bpf_sock *\fP\fIsk\fP\fB)\fP
2673 .INDENT 7.0
2674 .TP
2675 .B Description
2676 This helper gets a \fBstruct bpf_tcp_sock\fP pointer from a
2677 \fBstruct bpf_sock\fP pointer.
2678 .TP
2679 .B Return
2680 A \fBstruct bpf_tcp_sock\fP pointer on success, or \fBNULL\fP in
2681 case of failure.
2682 .UNINDENT
2683 .TP
2684 .B \fBlong bpf_skb_ecn_set_ce(struct sk_buff *\fP\fIskb\fP\fB)\fP
2685 .INDENT 7.0
2686 .TP
2687 .B Description
2688 Set ECN (Explicit Congestion Notification) field of IP header
2689 to \fBCE\fP (Congestion Encountered) if current value is \fBECT\fP
2690 (ECN Capable Transport). Otherwise, do nothing. Works with IPv6
2691 and IPv4.
2692 .TP
2693 .B Return
2694 1 if the \fBCE\fP flag is set (either by the current helper call
2695 or because it was already present), 0 if it is not set.
2696 .UNINDENT
2697 .TP
2698 .B \fBstruct bpf_sock *bpf_get_listener_sock(struct bpf_sock *\fP\fIsk\fP\fB)\fP
2699 .INDENT 7.0
2700 .TP
2701 .B Description
2702 Return a \fBstruct bpf_sock\fP pointer in \fBTCP_LISTEN\fP state.
2703 \fBbpf_sk_release\fP() is unnecessary and not allowed.
2704 .TP
2705 .B Return
2706 A \fBstruct bpf_sock\fP pointer on success, or \fBNULL\fP in
2707 case of failure.
2708 .UNINDENT
2709 .TP
2710 .B \fBstruct bpf_sock *bpf_skc_lookup_tcp(void *\fP\fIctx\fP\fB, struct bpf_sock_tuple *\fP\fItuple\fP\fB, u32\fP \fItuple_size\fP\fB, u64\fP \fInetns\fP\fB, u64\fP \fIflags\fP\fB)\fP
2711 .INDENT 7.0
2712 .TP
2713 .B Description
2714 Look for TCP socket matching \fItuple\fP, optionally in a child
2715 network namespace \fInetns\fP\&. The return value must be checked,
2716 and if non\-\fBNULL\fP, released via \fBbpf_sk_release\fP().
2717 .sp
2718 This function is identical to \fBbpf_sk_lookup_tcp\fP(), except
2719 that it also returns timewait or request sockets. Use
2720 \fBbpf_sk_fullsock\fP() or \fBbpf_tcp_sock\fP() to access the
2721 full structure.
2722 .sp
2723 This helper is available only if the kernel was compiled with
2724 \fBCONFIG_NET\fP configuration option.
2725 .TP
2726 .B Return
2727 Pointer to \fBstruct bpf_sock\fP, or \fBNULL\fP in case of failure.
2728 For sockets with reuseport option, the \fBstruct bpf_sock\fP
2729 result is from \fIreuse\fP\fB\->socks\fP[] using the hash of the
2730 tuple.
2731 .UNINDENT
2732 .TP
2733 .B \fBlong bpf_tcp_check_syncookie(void *\fP\fIsk\fP\fB, void *\fP\fIiph\fP\fB, u32\fP \fIiph_len\fP\fB, struct tcphdr *\fP\fIth\fP\fB, u32\fP \fIth_len\fP\fB)\fP
2734 .INDENT 7.0
2735 .TP
2736 .B Description
2737 Check whether \fIiph\fP and \fIth\fP contain a valid SYN cookie ACK for
2738 the listening socket in \fIsk\fP\&.
2739 .sp
2740 \fIiph\fP points to the start of the IPv4 or IPv6 header, while
2741 \fIiph_len\fP contains \fBsizeof\fP(\fBstruct iphdr\fP) or
2742 \fBsizeof\fP(\fBstruct ipv6hdr\fP).
2743 .sp
2744 \fIth\fP points to the start of the TCP header, while \fIth_len\fP
2745 contains the length of the TCP header (at least
2746 \fBsizeof\fP(\fBstruct tcphdr\fP)).
2747 .TP
2748 .B Return
2749 0 if \fIiph\fP and \fIth\fP are a valid SYN cookie ACK, or a negative
2750 error otherwise.
2751 .UNINDENT
2752 .TP
2753 .B \fBlong bpf_sysctl_get_name(struct bpf_sysctl *\fP\fIctx\fP\fB, char *\fP\fIbuf\fP\fB, size_t\fP \fIbuf_len\fP\fB, u64\fP \fIflags\fP\fB)\fP
2754 .INDENT 7.0
2755 .TP
2756 .B Description
2757 Get name of sysctl in /proc/sys/ and copy it into provided by
2758 program buffer \fIbuf\fP of size \fIbuf_len\fP\&.
2759 .sp
2760 The buffer is always NUL terminated, unless it\(aqs zero\-sized.
2761 .sp
2762 If \fIflags\fP is zero, full name (e.g. \(dqnet/ipv4/tcp_mem\(dq) is
2763 copied. Use \fBBPF_F_SYSCTL_BASE_NAME\fP flag to copy base name
2764 only (e.g. \(dqtcp_mem\(dq).
2765 .TP
2766 .B Return
2767 Number of character copied (not including the trailing NUL).
2768 .sp
2769 \fB\-E2BIG\fP if the buffer wasn\(aqt big enough (\fIbuf\fP will contain
2770 truncated name in this case).
2771 .UNINDENT
2772 .TP
2773 .B \fBlong bpf_sysctl_get_current_value(struct bpf_sysctl *\fP\fIctx\fP\fB, char *\fP\fIbuf\fP\fB, size_t\fP \fIbuf_len\fP\fB)\fP
2774 .INDENT 7.0
2775 .TP
2776 .B Description
2777 Get current value of sysctl as it is presented in /proc/sys
2778 (incl. newline, etc), and copy it as a string into provided
2779 by program buffer \fIbuf\fP of size \fIbuf_len\fP\&.
2780 .sp
2781 The whole value is copied, no matter what file position user
2782 space issued e.g. sys_read at.
2783 .sp
2784 The buffer is always NUL terminated, unless it\(aqs zero\-sized.
2785 .TP
2786 .B Return
2787 Number of character copied (not including the trailing NUL).
2788 .sp
2789 \fB\-E2BIG\fP if the buffer wasn\(aqt big enough (\fIbuf\fP will contain
2790 truncated name in this case).
2791 .sp
2792 \fB\-EINVAL\fP if current value was unavailable, e.g. because
2793 sysctl is uninitialized and read returns \-EIO for it.
2794 .UNINDENT
2795 .TP
2796 .B \fBlong bpf_sysctl_get_new_value(struct bpf_sysctl *\fP\fIctx\fP\fB, char *\fP\fIbuf\fP\fB, size_t\fP \fIbuf_len\fP\fB)\fP
2797 .INDENT 7.0
2798 .TP
2799 .B Description
2800 Get new value being written by user space to sysctl (before
2801 the actual write happens) and copy it as a string into
2802 provided by program buffer \fIbuf\fP of size \fIbuf_len\fP\&.
2803 .sp
2804 User space may write new value at file position > 0.
2805 .sp
2806 The buffer is always NUL terminated, unless it\(aqs zero\-sized.
2807 .TP
2808 .B Return
2809 Number of character copied (not including the trailing NUL).
2810 .sp
2811 \fB\-E2BIG\fP if the buffer wasn\(aqt big enough (\fIbuf\fP will contain
2812 truncated name in this case).
2813 .sp
2814 \fB\-EINVAL\fP if sysctl is being read.
2815 .UNINDENT
2816 .TP
2817 .B \fBlong bpf_sysctl_set_new_value(struct bpf_sysctl *\fP\fIctx\fP\fB, const char *\fP\fIbuf\fP\fB, size_t\fP \fIbuf_len\fP\fB)\fP
2818 .INDENT 7.0
2819 .TP
2820 .B Description
2821 Override new value being written by user space to sysctl with
2822 value provided by program in buffer \fIbuf\fP of size \fIbuf_len\fP\&.
2823 .sp
2824 \fIbuf\fP should contain a string in same form as provided by user
2825 space on sysctl write.
2826 .sp
2827 User space may write new value at file position > 0. To override
2828 the whole sysctl value file position should be set to zero.
2829 .TP
2830 .B Return
2831 0 on success.
2832 .sp
2833 \fB\-E2BIG\fP if the \fIbuf_len\fP is too big.
2834 .sp
2835 \fB\-EINVAL\fP if sysctl is being read.
2836 .UNINDENT
2837 .TP
2838 .B \fBlong bpf_strtol(const char *\fP\fIbuf\fP\fB, size_t\fP \fIbuf_len\fP\fB, u64\fP \fIflags\fP\fB, long *\fP\fIres\fP\fB)\fP
2839 .INDENT 7.0
2840 .TP
2841 .B Description
2842 Convert the initial part of the string from buffer \fIbuf\fP of
2843 size \fIbuf_len\fP to a long integer according to the given base
2844 and save the result in \fIres\fP\&.
2845 .sp
2846 The string may begin with an arbitrary amount of white space
2847 (as determined by \fBisspace\fP(3)) followed by a single
2848 optional \(aq\fB\-\fP\(aq sign.
2849 .sp
2850 Five least significant bits of \fIflags\fP encode base, other bits
2851 are currently unused.
2852 .sp
2853 Base must be either 8, 10, 16 or 0 to detect it automatically
2854 similar to user space \fBstrtol\fP(3).
2855 .TP
2856 .B Return
2857 Number of characters consumed on success. Must be positive but
2858 no more than \fIbuf_len\fP\&.
2859 .sp
2860 \fB\-EINVAL\fP if no valid digits were found or unsupported base
2861 was provided.
2862 .sp
2863 \fB\-ERANGE\fP if resulting value was out of range.
2864 .UNINDENT
2865 .TP
2866 .B \fBlong bpf_strtoul(const char *\fP\fIbuf\fP\fB, size_t\fP \fIbuf_len\fP\fB, u64\fP \fIflags\fP\fB, unsigned long *\fP\fIres\fP\fB)\fP
2867 .INDENT 7.0
2868 .TP
2869 .B Description
2870 Convert the initial part of the string from buffer \fIbuf\fP of
2871 size \fIbuf_len\fP to an unsigned long integer according to the
2872 given base and save the result in \fIres\fP\&.
2873 .sp
2874 The string may begin with an arbitrary amount of white space
2875 (as determined by \fBisspace\fP(3)).
2876 .sp
2877 Five least significant bits of \fIflags\fP encode base, other bits
2878 are currently unused.
2879 .sp
2880 Base must be either 8, 10, 16 or 0 to detect it automatically
2881 similar to user space \fBstrtoul\fP(3).
2882 .TP
2883 .B Return
2884 Number of characters consumed on success. Must be positive but
2885 no more than \fIbuf_len\fP\&.
2886 .sp
2887 \fB\-EINVAL\fP if no valid digits were found or unsupported base
2888 was provided.
2889 .sp
2890 \fB\-ERANGE\fP if resulting value was out of range.
2891 .UNINDENT
2892 .TP
2893 .B \fBvoid *bpf_sk_storage_get(struct bpf_map *\fP\fImap\fP\fB, void *\fP\fIsk\fP\fB, void *\fP\fIvalue\fP\fB, u64\fP \fIflags\fP\fB)\fP
2894 .INDENT 7.0
2895 .TP
2896 .B Description
2897 Get a bpf\-local\-storage from a \fIsk\fP\&.
2898 .sp
2899 Logically, it could be thought of getting the value from
2900 a \fImap\fP with \fIsk\fP as the \fBkey\fP\&. From this
2901 perspective, the usage is not much different from
2902 \fBbpf_map_lookup_elem\fP(\fImap\fP, \fB&\fP\fIsk\fP) except this
2903 helper enforces the key must be a full socket and the map must
2904 be a \fBBPF_MAP_TYPE_SK_STORAGE\fP also.
2905 .sp
2906 Underneath, the value is stored locally at \fIsk\fP instead of
2907 the \fImap\fP\&. The \fImap\fP is used as the bpf\-local\-storage
2908 \(dqtype\(dq. The bpf\-local\-storage \(dqtype\(dq (i.e. the \fImap\fP) is
2909 searched against all bpf\-local\-storages residing at \fIsk\fP\&.
2910 .sp
2911 \fIsk\fP is a kernel \fBstruct sock\fP pointer for LSM program.
2912 \fIsk\fP is a \fBstruct bpf_sock\fP pointer for other program types.
2913 .sp
2914 An optional \fIflags\fP (\fBBPF_SK_STORAGE_GET_F_CREATE\fP) can be
2915 used such that a new bpf\-local\-storage will be
2916 created if one does not exist. \fIvalue\fP can be used
2917 together with \fBBPF_SK_STORAGE_GET_F_CREATE\fP to specify
2918 the initial value of a bpf\-local\-storage. If \fIvalue\fP is
2919 \fBNULL\fP, the new bpf\-local\-storage will be zero initialized.
2920 .TP
2921 .B Return
2922 A bpf\-local\-storage pointer is returned on success.
2923 .sp
2924 \fBNULL\fP if not found or there was an error in adding
2925 a new bpf\-local\-storage.
2926 .UNINDENT
2927 .TP
2928 .B \fBlong bpf_sk_storage_delete(struct bpf_map *\fP\fImap\fP\fB, void *\fP\fIsk\fP\fB)\fP
2929 .INDENT 7.0
2930 .TP
2931 .B Description
2932 Delete a bpf\-local\-storage from a \fIsk\fP\&.
2933 .TP
2934 .B Return
2935 0 on success.
2936 .sp
2937 \fB\-ENOENT\fP if the bpf\-local\-storage cannot be found.
2938 \fB\-EINVAL\fP if sk is not a fullsock (e.g. a request_sock).
2939 .UNINDENT
2940 .TP
2941 .B \fBlong bpf_send_signal(u32\fP \fIsig\fP\fB)\fP
2942 .INDENT 7.0
2943 .TP
2944 .B Description
2945 Send signal \fIsig\fP to the process of the current task.
2946 The signal may be delivered to any of this process\(aqs threads.
2947 .TP
2948 .B Return
2949 0 on success or successfully queued.
2950 .sp
2951 \fB\-EBUSY\fP if work queue under nmi is full.
2952 .sp
2953 \fB\-EINVAL\fP if \fIsig\fP is invalid.
2954 .sp
2955 \fB\-EPERM\fP if no permission to send the \fIsig\fP\&.
2956 .sp
2957 \fB\-EAGAIN\fP if bpf program can try again.
2958 .UNINDENT
2959 .TP
2960 .B \fBs64 bpf_tcp_gen_syncookie(void *\fP\fIsk\fP\fB, void *\fP\fIiph\fP\fB, u32\fP \fIiph_len\fP\fB, struct tcphdr *\fP\fIth\fP\fB, u32\fP \fIth_len\fP\fB)\fP
2961 .INDENT 7.0
2962 .TP
2963 .B Description
2964 Try to issue a SYN cookie for the packet with corresponding
2965 IP/TCP headers, \fIiph\fP and \fIth\fP, on the listening socket in \fIsk\fP\&.
2966 .sp
2967 \fIiph\fP points to the start of the IPv4 or IPv6 header, while
2968 \fIiph_len\fP contains \fBsizeof\fP(\fBstruct iphdr\fP) or
2969 \fBsizeof\fP(\fBstruct ipv6hdr\fP).
2970 .sp
2971 \fIth\fP points to the start of the TCP header, while \fIth_len\fP
2972 contains the length of the TCP header with options (at least
2973 \fBsizeof\fP(\fBstruct tcphdr\fP)).
2974 .TP
2975 .B Return
2976 On success, lower 32 bits hold the generated SYN cookie in
2977 followed by 16 bits which hold the MSS value for that cookie,
2978 and the top 16 bits are unused.
2979 .sp
2980 On failure, the returned value is one of the following:
2981 .sp
2982 \fB\-EINVAL\fP SYN cookie cannot be issued due to error
2983 .sp
2984 \fB\-ENOENT\fP SYN cookie should not be issued (no SYN flood)
2985 .sp
2986 \fB\-EOPNOTSUPP\fP kernel configuration does not enable SYN cookies
2987 .sp
2988 \fB\-EPROTONOSUPPORT\fP IP packet version is not 4 or 6
2989 .UNINDENT
2990 .TP
2991 .B \fBlong bpf_skb_output(void *\fP\fIctx\fP\fB, struct bpf_map *\fP\fImap\fP\fB, u64\fP \fIflags\fP\fB, void *\fP\fIdata\fP\fB, u64\fP \fIsize\fP\fB)\fP
2992 .INDENT 7.0
2993 .TP
2994 .B Description
2995 Write raw \fIdata\fP blob into a special BPF perf event held by
2996 \fImap\fP of type \fBBPF_MAP_TYPE_PERF_EVENT_ARRAY\fP\&. This perf
2997 event must have the following attributes: \fBPERF_SAMPLE_RAW\fP
2998 as \fBsample_type\fP, \fBPERF_TYPE_SOFTWARE\fP as \fBtype\fP, and
2999 \fBPERF_COUNT_SW_BPF_OUTPUT\fP as \fBconfig\fP\&.
3000 .sp
3001 The \fIflags\fP are used to indicate the index in \fImap\fP for which
3002 the value must be put, masked with \fBBPF_F_INDEX_MASK\fP\&.
3003 Alternatively, \fIflags\fP can be set to \fBBPF_F_CURRENT_CPU\fP
3004 to indicate that the index of the current CPU core should be
3005 used.
3006 .sp
3007 The value to write, of \fIsize\fP, is passed through eBPF stack and
3008 pointed by \fIdata\fP\&.
3009 .sp
3010 \fIctx\fP is a pointer to in\-kernel struct sk_buff.
3011 .sp
3012 This helper is similar to \fBbpf_perf_event_output\fP() but
3013 restricted to raw_tracepoint bpf programs.
3014 .TP
3015 .B Return
3016 0 on success, or a negative error in case of failure.
3017 .UNINDENT
3018 .TP
3019 .B \fBlong bpf_probe_read_user(void *\fP\fIdst\fP\fB, u32\fP \fIsize\fP\fB, const void *\fP\fIunsafe_ptr\fP\fB)\fP
3020 .INDENT 7.0
3021 .TP
3022 .B Description
3023 Safely attempt to read \fIsize\fP bytes from user space address
3024 \fIunsafe_ptr\fP and store the data in \fIdst\fP\&.
3025 .TP
3026 .B Return
3027 0 on success, or a negative error in case of failure.
3028 .UNINDENT
3029 .TP
3030 .B \fBlong bpf_probe_read_kernel(void *\fP\fIdst\fP\fB, u32\fP \fIsize\fP\fB, const void *\fP\fIunsafe_ptr\fP\fB)\fP
3031 .INDENT 7.0
3032 .TP
3033 .B Description
3034 Safely attempt to read \fIsize\fP bytes from kernel space address
3035 \fIunsafe_ptr\fP and store the data in \fIdst\fP\&.
3036 .TP
3037 .B Return
3038 0 on success, or a negative error in case of failure.
3039 .UNINDENT
3040 .TP
3041 .B \fBlong bpf_probe_read_user_str(void *\fP\fIdst\fP\fB, u32\fP \fIsize\fP\fB, const void *\fP\fIunsafe_ptr\fP\fB)\fP
3042 .INDENT 7.0
3043 .TP
3044 .B Description
3045 Copy a NUL terminated string from an unsafe user address
3046 \fIunsafe_ptr\fP to \fIdst\fP\&. The \fIsize\fP should include the
3047 terminating NUL byte. In case the string length is smaller than
3048 \fIsize\fP, the target is not padded with further NUL bytes. If the
3049 string length is larger than \fIsize\fP, just \fIsize\fP\-1 bytes are
3050 copied and the last byte is set to NUL.
3051 .sp
3052 On success, returns the number of bytes that were written,
3053 including the terminal NUL. This makes this helper useful in
3054 tracing programs for reading strings, and more importantly to
3055 get its length at runtime. See the following snippet:
3056 .INDENT 7.0
3057 .INDENT 3.5
3058 .sp
3059 .EX
3060 SEC(\(dqkprobe/sys_open\(dq)
3061 void bpf_sys_open(struct pt_regs *ctx)
3062 {
3063 char buf[PATHLEN]; // PATHLEN is defined to 256
3064 int res = bpf_probe_read_user_str(buf, sizeof(buf),
3065 ctx\->di);
3066
3067 // Consume buf, for example push it to
3068 // userspace via bpf_perf_event_output(); we
3069 // can use res (the string length) as event
3070 // size, after checking its boundaries.
3071 }
3072 .EE
3073 .UNINDENT
3074 .UNINDENT
3075 .sp
3076 In comparison, using \fBbpf_probe_read_user\fP() helper here
3077 instead to read the string would require to estimate the length
3078 at compile time, and would often result in copying more memory
3079 than necessary.
3080 .sp
3081 Another useful use case is when parsing individual process
3082 arguments or individual environment variables navigating
3083 \fIcurrent\fP\fB\->mm\->arg_start\fP and \fIcurrent\fP\fB\->mm\->env_start\fP: using this helper and the return value,
3084 one can quickly iterate at the right offset of the memory area.
3085 .TP
3086 .B Return
3087 On success, the strictly positive length of the output string,
3088 including the trailing NUL character. On error, a negative
3089 value.
3090 .UNINDENT
3091 .TP
3092 .B \fBlong bpf_probe_read_kernel_str(void *\fP\fIdst\fP\fB, u32\fP \fIsize\fP\fB, const void *\fP\fIunsafe_ptr\fP\fB)\fP
3093 .INDENT 7.0
3094 .TP
3095 .B Description
3096 Copy a NUL terminated string from an unsafe kernel address \fIunsafe_ptr\fP
3097 to \fIdst\fP\&. Same semantics as with \fBbpf_probe_read_user_str\fP() apply.
3098 .TP
3099 .B Return
3100 On success, the strictly positive length of the string, including
3101 the trailing NUL character. On error, a negative value.
3102 .UNINDENT
3103 .TP
3104 .B \fBlong bpf_tcp_send_ack(void *\fP\fItp\fP\fB, u32\fP \fIrcv_nxt\fP\fB)\fP
3105 .INDENT 7.0
3106 .TP
3107 .B Description
3108 Send out a tcp\-ack. \fItp\fP is the in\-kernel struct \fBtcp_sock\fP\&.
3109 \fIrcv_nxt\fP is the ack_seq to be sent out.
3110 .TP
3111 .B Return
3112 0 on success, or a negative error in case of failure.
3113 .UNINDENT
3114 .TP
3115 .B \fBlong bpf_send_signal_thread(u32\fP \fIsig\fP\fB)\fP
3116 .INDENT 7.0
3117 .TP
3118 .B Description
3119 Send signal \fIsig\fP to the thread corresponding to the current task.
3120 .TP
3121 .B Return
3122 0 on success or successfully queued.
3123 .sp
3124 \fB\-EBUSY\fP if work queue under nmi is full.
3125 .sp
3126 \fB\-EINVAL\fP if \fIsig\fP is invalid.
3127 .sp
3128 \fB\-EPERM\fP if no permission to send the \fIsig\fP\&.
3129 .sp
3130 \fB\-EAGAIN\fP if bpf program can try again.
3131 .UNINDENT
3132 .TP
3133 .B \fBu64 bpf_jiffies64(void)\fP
3134 .INDENT 7.0
3135 .TP
3136 .B Description
3137 Obtain the 64bit jiffies
3138 .TP
3139 .B Return
3140 The 64 bit jiffies
3141 .UNINDENT
3142 .TP
3143 .B \fBlong bpf_read_branch_records(struct bpf_perf_event_data *\fP\fIctx\fP\fB, void *\fP\fIbuf\fP\fB, u32\fP \fIsize\fP\fB, u64\fP \fIflags\fP\fB)\fP
3144 .INDENT 7.0
3145 .TP
3146 .B Description
3147 For an eBPF program attached to a perf event, retrieve the
3148 branch records (\fBstruct perf_branch_entry\fP) associated to \fIctx\fP
3149 and store it in the buffer pointed by \fIbuf\fP up to size
3150 \fIsize\fP bytes.
3151 .TP
3152 .B Return
3153 On success, number of bytes written to \fIbuf\fP\&. On error, a
3154 negative value.
3155 .sp
3156 The \fIflags\fP can be set to \fBBPF_F_GET_BRANCH_RECORDS_SIZE\fP to
3157 instead return the number of bytes required to store all the
3158 branch entries. If this flag is set, \fIbuf\fP may be NULL.
3159 .sp
3160 \fB\-EINVAL\fP if arguments invalid or \fBsize\fP not a multiple
3161 of \fBsizeof\fP(\fBstruct perf_branch_entry\fP).
3162 .sp
3163 \fB\-ENOENT\fP if architecture does not support branch records.
3164 .UNINDENT
3165 .TP
3166 .B \fBlong bpf_get_ns_current_pid_tgid(u64\fP \fIdev\fP\fB, u64\fP \fIino\fP\fB, struct bpf_pidns_info *\fP\fInsdata\fP\fB, u32\fP \fIsize\fP\fB)\fP
3167 .INDENT 7.0
3168 .TP
3169 .B Description
3170 Returns 0 on success, values for \fIpid\fP and \fItgid\fP as seen from the current
3171 \fInamespace\fP will be returned in \fInsdata\fP\&.
3172 .TP
3173 .B Return
3174 0 on success, or one of the following in case of failure:
3175 .sp
3176 \fB\-EINVAL\fP if dev and inum supplied don\(aqt match dev_t and inode number
3177 with nsfs of current task, or if dev conversion to dev_t lost high bits.
3178 .sp
3179 \fB\-ENOENT\fP if pidns does not exists for the current task.
3180 .UNINDENT
3181 .TP
3182 .B \fBlong bpf_xdp_output(void *\fP\fIctx\fP\fB, struct bpf_map *\fP\fImap\fP\fB, u64\fP \fIflags\fP\fB, void *\fP\fIdata\fP\fB, u64\fP \fIsize\fP\fB)\fP
3183 .INDENT 7.0
3184 .TP
3185 .B Description
3186 Write raw \fIdata\fP blob into a special BPF perf event held by
3187 \fImap\fP of type \fBBPF_MAP_TYPE_PERF_EVENT_ARRAY\fP\&. This perf
3188 event must have the following attributes: \fBPERF_SAMPLE_RAW\fP
3189 as \fBsample_type\fP, \fBPERF_TYPE_SOFTWARE\fP as \fBtype\fP, and
3190 \fBPERF_COUNT_SW_BPF_OUTPUT\fP as \fBconfig\fP\&.
3191 .sp
3192 The \fIflags\fP are used to indicate the index in \fImap\fP for which
3193 the value must be put, masked with \fBBPF_F_INDEX_MASK\fP\&.
3194 Alternatively, \fIflags\fP can be set to \fBBPF_F_CURRENT_CPU\fP
3195 to indicate that the index of the current CPU core should be
3196 used.
3197 .sp
3198 The value to write, of \fIsize\fP, is passed through eBPF stack and
3199 pointed by \fIdata\fP\&.
3200 .sp
3201 \fIctx\fP is a pointer to in\-kernel struct xdp_buff.
3202 .sp
3203 This helper is similar to \fBbpf_perf_eventoutput\fP() but
3204 restricted to raw_tracepoint bpf programs.
3205 .TP
3206 .B Return
3207 0 on success, or a negative error in case of failure.
3208 .UNINDENT
3209 .TP
3210 .B \fBu64 bpf_get_netns_cookie(void *\fP\fIctx\fP\fB)\fP
3211 .INDENT 7.0
3212 .TP
3213 .B Description
3214 Retrieve the cookie (generated by the kernel) of the network
3215 namespace the input \fIctx\fP is associated with. The network
3216 namespace cookie remains stable for its lifetime and provides
3217 a global identifier that can be assumed unique. If \fIctx\fP is
3218 NULL, then the helper returns the cookie for the initial
3219 network namespace. The cookie itself is very similar to that
3220 of \fBbpf_get_socket_cookie\fP() helper, but for network
3221 namespaces instead of sockets.
3222 .TP
3223 .B Return
3224 A 8\-byte long opaque number.
3225 .UNINDENT
3226 .TP
3227 .B \fBu64 bpf_get_current_ancestor_cgroup_id(int\fP \fIancestor_level\fP\fB)\fP
3228 .INDENT 7.0
3229 .TP
3230 .B Description
3231 Return id of cgroup v2 that is ancestor of the cgroup associated
3232 with the current task at the \fIancestor_level\fP\&. The root cgroup
3233 is at \fIancestor_level\fP zero and each step down the hierarchy
3234 increments the level. If \fIancestor_level\fP == level of cgroup
3235 associated with the current task, then return value will be the
3236 same as that of \fBbpf_get_current_cgroup_id\fP().
3237 .sp
3238 The helper is useful to implement policies based on cgroups
3239 that are upper in hierarchy than immediate cgroup associated
3240 with the current task.
3241 .sp
3242 The format of returned id and helper limitations are same as in
3243 \fBbpf_get_current_cgroup_id\fP().
3244 .TP
3245 .B Return
3246 The id is returned or 0 in case the id could not be retrieved.
3247 .UNINDENT
3248 .TP
3249 .B \fBlong bpf_sk_assign(struct sk_buff *\fP\fIskb\fP\fB, void *\fP\fIsk\fP\fB, u64\fP \fIflags\fP\fB)\fP
3250 .INDENT 7.0
3251 .TP
3252 .B Description
3253 Helper is overloaded depending on BPF program type. This
3254 description applies to \fBBPF_PROG_TYPE_SCHED_CLS\fP and
3255 \fBBPF_PROG_TYPE_SCHED_ACT\fP programs.
3256 .sp
3257 Assign the \fIsk\fP to the \fIskb\fP\&. When combined with appropriate
3258 routing configuration to receive the packet towards the socket,
3259 will cause \fIskb\fP to be delivered to the specified socket.
3260 Subsequent redirection of \fIskb\fP via \fBbpf_redirect\fP(),
3261 \fBbpf_clone_redirect\fP() or other methods outside of BPF may
3262 interfere with successful delivery to the socket.
3263 .sp
3264 This operation is only valid from TC ingress path.
3265 .sp
3266 The \fIflags\fP argument must be zero.
3267 .TP
3268 .B Return
3269 0 on success, or a negative error in case of failure:
3270 .sp
3271 \fB\-EINVAL\fP if specified \fIflags\fP are not supported.
3272 .sp
3273 \fB\-ENOENT\fP if the socket is unavailable for assignment.
3274 .sp
3275 \fB\-ENETUNREACH\fP if the socket is unreachable (wrong netns).
3276 .sp
3277 \fB\-EOPNOTSUPP\fP if the operation is not supported, for example
3278 a call from outside of TC ingress.
3279 .UNINDENT
3280 .TP
3281 .B \fBlong bpf_sk_assign(struct bpf_sk_lookup *\fP\fIctx\fP\fB, struct bpf_sock *\fP\fIsk\fP\fB, u64\fP \fIflags\fP\fB)\fP
3282 .INDENT 7.0
3283 .TP
3284 .B Description
3285 Helper is overloaded depending on BPF program type. This
3286 description applies to \fBBPF_PROG_TYPE_SK_LOOKUP\fP programs.
3287 .sp
3288 Select the \fIsk\fP as a result of a socket lookup.
3289 .sp
3290 For the operation to succeed passed socket must be compatible
3291 with the packet description provided by the \fIctx\fP object.
3292 .sp
3293 L4 protocol (\fBIPPROTO_TCP\fP or \fBIPPROTO_UDP\fP) must
3294 be an exact match. While IP family (\fBAF_INET\fP or
3295 \fBAF_INET6\fP) must be compatible, that is IPv6 sockets
3296 that are not v6\-only can be selected for IPv4 packets.
3297 .sp
3298 Only TCP listeners and UDP unconnected sockets can be
3299 selected. \fIsk\fP can also be NULL to reset any previous
3300 selection.
3301 .sp
3302 \fIflags\fP argument can combination of following values:
3303 .INDENT 7.0
3304 .IP \(bu 2
3305 \fBBPF_SK_LOOKUP_F_REPLACE\fP to override the previous
3306 socket selection, potentially done by a BPF program
3307 that ran before us.
3308 .IP \(bu 2
3309 \fBBPF_SK_LOOKUP_F_NO_REUSEPORT\fP to skip
3310 load\-balancing within reuseport group for the socket
3311 being selected.
3312 .UNINDENT
3313 .sp
3314 On success \fIctx\->sk\fP will point to the selected socket.
3315 .TP
3316 .B Return
3317 0 on success, or a negative errno in case of failure.
3318 .INDENT 7.0
3319 .IP \(bu 2
3320 \fB\-EAFNOSUPPORT\fP if socket family (\fIsk\->family\fP) is
3321 not compatible with packet family (\fIctx\->family\fP).
3322 .IP \(bu 2
3323 \fB\-EEXIST\fP if socket has been already selected,
3324 potentially by another program, and
3325 \fBBPF_SK_LOOKUP_F_REPLACE\fP flag was not specified.
3326 .IP \(bu 2
3327 \fB\-EINVAL\fP if unsupported flags were specified.
3328 .IP \(bu 2
3329 \fB\-EPROTOTYPE\fP if socket L4 protocol
3330 (\fIsk\->protocol\fP) doesn\(aqt match packet protocol
3331 (\fIctx\->protocol\fP).
3332 .IP \(bu 2
3333 \fB\-ESOCKTNOSUPPORT\fP if socket is not in allowed
3334 state (TCP listening or UDP unconnected).
3335 .UNINDENT
3336 .UNINDENT
3337 .TP
3338 .B \fBu64 bpf_ktime_get_boot_ns(void)\fP
3339 .INDENT 7.0
3340 .TP
3341 .B Description
3342 Return the time elapsed since system boot, in nanoseconds.
3343 Does include the time the system was suspended.
3344 See: \fBclock_gettime\fP(\fBCLOCK_BOOTTIME\fP)
3345 .TP
3346 .B Return
3347 Current \fIktime\fP\&.
3348 .UNINDENT
3349 .TP
3350 .B \fBlong bpf_seq_printf(struct seq_file *\fP\fIm\fP\fB, const char *\fP\fIfmt\fP\fB, u32\fP \fIfmt_size\fP\fB, const void *\fP\fIdata\fP\fB, u32\fP \fIdata_len\fP\fB)\fP
3351 .INDENT 7.0
3352 .TP
3353 .B Description
3354 \fBbpf_seq_printf\fP() uses seq_file \fBseq_printf\fP() to print
3355 out the format string.
3356 The \fIm\fP represents the seq_file. The \fIfmt\fP and \fIfmt_size\fP are for
3357 the format string itself. The \fIdata\fP and \fIdata_len\fP are format string
3358 arguments. The \fIdata\fP are a \fBu64\fP array and corresponding format string
3359 values are stored in the array. For strings and pointers where pointees
3360 are accessed, only the pointer values are stored in the \fIdata\fP array.
3361 The \fIdata_len\fP is the size of \fIdata\fP in bytes \- must be a multiple of 8.
3362 .sp
3363 Formats \fB%s\fP, \fB%p{i,I}{4,6}\fP requires to read kernel memory.
3364 Reading kernel memory may fail due to either invalid address or
3365 valid address but requiring a major memory fault. If reading kernel memory
3366 fails, the string for \fB%s\fP will be an empty string, and the ip
3367 address for \fB%p{i,I}{4,6}\fP will be 0. Not returning error to
3368 bpf program is consistent with what \fBbpf_trace_printk\fP() does for now.
3369 .TP
3370 .B Return
3371 0 on success, or a negative error in case of failure:
3372 .sp
3373 \fB\-EBUSY\fP if per\-CPU memory copy buffer is busy, can try again
3374 by returning 1 from bpf program.
3375 .sp
3376 \fB\-EINVAL\fP if arguments are invalid, or if \fIfmt\fP is invalid/unsupported.
3377 .sp
3378 \fB\-E2BIG\fP if \fIfmt\fP contains too many format specifiers.
3379 .sp
3380 \fB\-EOVERFLOW\fP if an overflow happened: The same object will be tried again.
3381 .UNINDENT
3382 .TP
3383 .B \fBlong bpf_seq_write(struct seq_file *\fP\fIm\fP\fB, const void *\fP\fIdata\fP\fB, u32\fP \fIlen\fP\fB)\fP
3384 .INDENT 7.0
3385 .TP
3386 .B Description
3387 \fBbpf_seq_write\fP() uses seq_file \fBseq_write\fP() to write the data.
3388 The \fIm\fP represents the seq_file. The \fIdata\fP and \fIlen\fP represent the
3389 data to write in bytes.
3390 .TP
3391 .B Return
3392 0 on success, or a negative error in case of failure:
3393 .sp
3394 \fB\-EOVERFLOW\fP if an overflow happened: The same object will be tried again.
3395 .UNINDENT
3396 .TP
3397 .B \fBu64 bpf_sk_cgroup_id(void *\fP\fIsk\fP\fB)\fP
3398 .INDENT 7.0
3399 .TP
3400 .B Description
3401 Return the cgroup v2 id of the socket \fIsk\fP\&.
3402 .sp
3403 \fIsk\fP must be a non\-\fBNULL\fP pointer to a socket, e.g. one
3404 returned from \fBbpf_sk_lookup_xxx\fP(),
3405 \fBbpf_sk_fullsock\fP(), etc. The format of returned id is
3406 same as in \fBbpf_skb_cgroup_id\fP().
3407 .sp
3408 This helper is available only if the kernel was compiled with
3409 the \fBCONFIG_SOCK_CGROUP_DATA\fP configuration option.
3410 .TP
3411 .B Return
3412 The id is returned or 0 in case the id could not be retrieved.
3413 .UNINDENT
3414 .TP
3415 .B \fBu64 bpf_sk_ancestor_cgroup_id(void *\fP\fIsk\fP\fB, int\fP \fIancestor_level\fP\fB)\fP
3416 .INDENT 7.0
3417 .TP
3418 .B Description
3419 Return id of cgroup v2 that is ancestor of cgroup associated
3420 with the \fIsk\fP at the \fIancestor_level\fP\&. The root cgroup is at
3421 \fIancestor_level\fP zero and each step down the hierarchy
3422 increments the level. If \fIancestor_level\fP == level of cgroup
3423 associated with \fIsk\fP, then return value will be same as that
3424 of \fBbpf_sk_cgroup_id\fP().
3425 .sp
3426 The helper is useful to implement policies based on cgroups
3427 that are upper in hierarchy than immediate cgroup associated
3428 with \fIsk\fP\&.
3429 .sp
3430 The format of returned id and helper limitations are same as in
3431 \fBbpf_sk_cgroup_id\fP().
3432 .TP
3433 .B Return
3434 The id is returned or 0 in case the id could not be retrieved.
3435 .UNINDENT
3436 .TP
3437 .B \fBlong bpf_ringbuf_output(void *\fP\fIringbuf\fP\fB, void *\fP\fIdata\fP\fB, u64\fP \fIsize\fP\fB, u64\fP \fIflags\fP\fB)\fP
3438 .INDENT 7.0
3439 .TP
3440 .B Description
3441 Copy \fIsize\fP bytes from \fIdata\fP into a ring buffer \fIringbuf\fP\&.
3442 If \fBBPF_RB_NO_WAKEUP\fP is specified in \fIflags\fP, no notification
3443 of new data availability is sent.
3444 If \fBBPF_RB_FORCE_WAKEUP\fP is specified in \fIflags\fP, notification
3445 of new data availability is sent unconditionally.
3446 If \fB0\fP is specified in \fIflags\fP, an adaptive notification
3447 of new data availability is sent.
3448 .sp
3449 An adaptive notification is a notification sent whenever the user\-space
3450 process has caught up and consumed all available payloads. In case the user\-space
3451 process is still processing a previous payload, then no notification is needed
3452 as it will process the newly added payload automatically.
3453 .TP
3454 .B Return
3455 0 on success, or a negative error in case of failure.
3456 .UNINDENT
3457 .TP
3458 .B \fBvoid *bpf_ringbuf_reserve(void *\fP\fIringbuf\fP\fB, u64\fP \fIsize\fP\fB, u64\fP \fIflags\fP\fB)\fP
3459 .INDENT 7.0
3460 .TP
3461 .B Description
3462 Reserve \fIsize\fP bytes of payload in a ring buffer \fIringbuf\fP\&.
3463 \fIflags\fP must be 0.
3464 .TP
3465 .B Return
3466 Valid pointer with \fIsize\fP bytes of memory available; NULL,
3467 otherwise.
3468 .UNINDENT
3469 .TP
3470 .B \fBvoid bpf_ringbuf_submit(void *\fP\fIdata\fP\fB, u64\fP \fIflags\fP\fB)\fP
3471 .INDENT 7.0
3472 .TP
3473 .B Description
3474 Submit reserved ring buffer sample, pointed to by \fIdata\fP\&.
3475 If \fBBPF_RB_NO_WAKEUP\fP is specified in \fIflags\fP, no notification
3476 of new data availability is sent.
3477 If \fBBPF_RB_FORCE_WAKEUP\fP is specified in \fIflags\fP, notification
3478 of new data availability is sent unconditionally.
3479 If \fB0\fP is specified in \fIflags\fP, an adaptive notification
3480 of new data availability is sent.
3481 .sp
3482 See \(aqbpf_ringbuf_output()\(aq for the definition of adaptive notification.
3483 .TP
3484 .B Return
3485 Nothing. Always succeeds.
3486 .UNINDENT
3487 .TP
3488 .B \fBvoid bpf_ringbuf_discard(void *\fP\fIdata\fP\fB, u64\fP \fIflags\fP\fB)\fP
3489 .INDENT 7.0
3490 .TP
3491 .B Description
3492 Discard reserved ring buffer sample, pointed to by \fIdata\fP\&.
3493 If \fBBPF_RB_NO_WAKEUP\fP is specified in \fIflags\fP, no notification
3494 of new data availability is sent.
3495 If \fBBPF_RB_FORCE_WAKEUP\fP is specified in \fIflags\fP, notification
3496 of new data availability is sent unconditionally.
3497 If \fB0\fP is specified in \fIflags\fP, an adaptive notification
3498 of new data availability is sent.
3499 .sp
3500 See \(aqbpf_ringbuf_output()\(aq for the definition of adaptive notification.
3501 .TP
3502 .B Return
3503 Nothing. Always succeeds.
3504 .UNINDENT
3505 .TP
3506 .B \fBu64 bpf_ringbuf_query(void *\fP\fIringbuf\fP\fB, u64\fP \fIflags\fP\fB)\fP
3507 .INDENT 7.0
3508 .TP
3509 .B Description
3510 Query various characteristics of provided ring buffer. What
3511 exactly is queries is determined by \fIflags\fP:
3512 .INDENT 7.0
3513 .IP \(bu 2
3514 \fBBPF_RB_AVAIL_DATA\fP: Amount of data not yet consumed.
3515 .IP \(bu 2
3516 \fBBPF_RB_RING_SIZE\fP: The size of ring buffer.
3517 .IP \(bu 2
3518 \fBBPF_RB_CONS_POS\fP: Consumer position (can wrap around).
3519 .IP \(bu 2
3520 \fBBPF_RB_PROD_POS\fP: Producer(s) position (can wrap around).
3521 .UNINDENT
3522 .sp
3523 Data returned is just a momentary snapshot of actual values
3524 and could be inaccurate, so this facility should be used to
3525 power heuristics and for reporting, not to make 100% correct
3526 calculation.
3527 .TP
3528 .B Return
3529 Requested value, or 0, if \fIflags\fP are not recognized.
3530 .UNINDENT
3531 .TP
3532 .B \fBlong bpf_csum_level(struct sk_buff *\fP\fIskb\fP\fB, u64\fP \fIlevel\fP\fB)\fP
3533 .INDENT 7.0
3534 .TP
3535 .B Description
3536 Change the skbs checksum level by one layer up or down, or
3537 reset it entirely to none in order to have the stack perform
3538 checksum validation. The level is applicable to the following
3539 protocols: TCP, UDP, GRE, SCTP, FCOE. For example, a decap of
3540 | ETH | IP | UDP | GUE | IP | TCP | into | ETH | IP | TCP |
3541 through \fBbpf_skb_adjust_room\fP() helper with passing in
3542 \fBBPF_F_ADJ_ROOM_NO_CSUM_RESET\fP flag would require one call
3543 to \fBbpf_csum_level\fP() with \fBBPF_CSUM_LEVEL_DEC\fP since
3544 the UDP header is removed. Similarly, an encap of the latter
3545 into the former could be accompanied by a helper call to
3546 \fBbpf_csum_level\fP() with \fBBPF_CSUM_LEVEL_INC\fP if the
3547 skb is still intended to be processed in higher layers of the
3548 stack instead of just egressing at tc.
3549 .sp
3550 There are three supported level settings at this time:
3551 .INDENT 7.0
3552 .IP \(bu 2
3553 \fBBPF_CSUM_LEVEL_INC\fP: Increases skb\->csum_level for skbs
3554 with CHECKSUM_UNNECESSARY.
3555 .IP \(bu 2
3556 \fBBPF_CSUM_LEVEL_DEC\fP: Decreases skb\->csum_level for skbs
3557 with CHECKSUM_UNNECESSARY.
3558 .IP \(bu 2
3559 \fBBPF_CSUM_LEVEL_RESET\fP: Resets skb\->csum_level to 0 and
3560 sets CHECKSUM_NONE to force checksum validation by the stack.
3561 .IP \(bu 2
3562 \fBBPF_CSUM_LEVEL_QUERY\fP: No\-op, returns the current
3563 skb\->csum_level.
3564 .UNINDENT
3565 .TP
3566 .B Return
3567 0 on success, or a negative error in case of failure. In the
3568 case of \fBBPF_CSUM_LEVEL_QUERY\fP, the current skb\->csum_level
3569 is returned or the error code \-EACCES in case the skb is not
3570 subject to CHECKSUM_UNNECESSARY.
3571 .UNINDENT
3572 .TP
3573 .B \fBstruct tcp6_sock *bpf_skc_to_tcp6_sock(void *\fP\fIsk\fP\fB)\fP
3574 .INDENT 7.0
3575 .TP
3576 .B Description
3577 Dynamically cast a \fIsk\fP pointer to a \fItcp6_sock\fP pointer.
3578 .TP
3579 .B Return
3580 \fIsk\fP if casting is valid, or \fBNULL\fP otherwise.
3581 .UNINDENT
3582 .TP
3583 .B \fBstruct tcp_sock *bpf_skc_to_tcp_sock(void *\fP\fIsk\fP\fB)\fP
3584 .INDENT 7.0
3585 .TP
3586 .B Description
3587 Dynamically cast a \fIsk\fP pointer to a \fItcp_sock\fP pointer.
3588 .TP
3589 .B Return
3590 \fIsk\fP if casting is valid, or \fBNULL\fP otherwise.
3591 .UNINDENT
3592 .TP
3593 .B \fBstruct tcp_timewait_sock *bpf_skc_to_tcp_timewait_sock(void *\fP\fIsk\fP\fB)\fP
3594 .INDENT 7.0
3595 .TP
3596 .B Description
3597 Dynamically cast a \fIsk\fP pointer to a \fItcp_timewait_sock\fP pointer.
3598 .TP
3599 .B Return
3600 \fIsk\fP if casting is valid, or \fBNULL\fP otherwise.
3601 .UNINDENT
3602 .TP
3603 .B \fBstruct tcp_request_sock *bpf_skc_to_tcp_request_sock(void *\fP\fIsk\fP\fB)\fP
3604 .INDENT 7.0
3605 .TP
3606 .B Description
3607 Dynamically cast a \fIsk\fP pointer to a \fItcp_request_sock\fP pointer.
3608 .TP
3609 .B Return
3610 \fIsk\fP if casting is valid, or \fBNULL\fP otherwise.
3611 .UNINDENT
3612 .TP
3613 .B \fBstruct udp6_sock *bpf_skc_to_udp6_sock(void *\fP\fIsk\fP\fB)\fP
3614 .INDENT 7.0
3615 .TP
3616 .B Description
3617 Dynamically cast a \fIsk\fP pointer to a \fIudp6_sock\fP pointer.
3618 .TP
3619 .B Return
3620 \fIsk\fP if casting is valid, or \fBNULL\fP otherwise.
3621 .UNINDENT
3622 .TP
3623 .B \fBlong bpf_get_task_stack(struct task_struct *\fP\fItask\fP\fB, void *\fP\fIbuf\fP\fB, u32\fP \fIsize\fP\fB, u64\fP \fIflags\fP\fB)\fP
3624 .INDENT 7.0
3625 .TP
3626 .B Description
3627 Return a user or a kernel stack in bpf program provided buffer.
3628 Note: the user stack will only be populated if the \fItask\fP is
3629 the current task; all other tasks will return \-EOPNOTSUPP.
3630 To achieve this, the helper needs \fItask\fP, which is a valid
3631 pointer to \fBstruct task_struct\fP\&. To store the stacktrace, the
3632 bpf program provides \fIbuf\fP with a nonnegative \fIsize\fP\&.
3633 .sp
3634 The last argument, \fIflags\fP, holds the number of stack frames to
3635 skip (from 0 to 255), masked with
3636 \fBBPF_F_SKIP_FIELD_MASK\fP\&. The next bits can be used to set
3637 the following flags:
3638 .INDENT 7.0
3639 .TP
3640 .B \fBBPF_F_USER_STACK\fP
3641 Collect a user space stack instead of a kernel stack.
3642 The \fItask\fP must be the current task.
3643 .TP
3644 .B \fBBPF_F_USER_BUILD_ID\fP
3645 Collect buildid+offset instead of ips for user stack,
3646 only valid if \fBBPF_F_USER_STACK\fP is also specified.
3647 .UNINDENT
3648 .sp
3649 \fBbpf_get_task_stack\fP() can collect up to
3650 \fBPERF_MAX_STACK_DEPTH\fP both kernel and user frames, subject
3651 to sufficient large buffer size. Note that
3652 this limit can be controlled with the \fBsysctl\fP program, and
3653 that it should be manually increased in order to profile long
3654 user stacks (such as stacks for Java programs). To do so, use:
3655 .INDENT 7.0
3656 .INDENT 3.5
3657 .sp
3658 .EX
3659 # sysctl kernel.perf_event_max_stack=<new value>
3660 .EE
3661 .UNINDENT
3662 .UNINDENT
3663 .TP
3664 .B Return
3665 The non\-negative copied \fIbuf\fP length equal to or less than
3666 \fIsize\fP on success, or a negative error in case of failure.
3667 .UNINDENT
3668 .TP
3669 .B \fBlong bpf_load_hdr_opt(struct bpf_sock_ops *\fP\fIskops\fP\fB, void *\fP\fIsearchby_res\fP\fB, u32\fP \fIlen\fP\fB, u64\fP \fIflags\fP\fB)\fP
3670 .INDENT 7.0
3671 .TP
3672 .B Description
3673 Load header option. Support reading a particular TCP header
3674 option for bpf program (\fBBPF_PROG_TYPE_SOCK_OPS\fP).
3675 .sp
3676 If \fIflags\fP is 0, it will search the option from the
3677 \fIskops\fP\fB\->skb_data\fP\&. The comment in \fBstruct bpf_sock_ops\fP
3678 has details on what skb_data contains under different
3679 \fIskops\fP\fB\->op\fP\&.
3680 .sp
3681 The first byte of the \fIsearchby_res\fP specifies the
3682 kind that it wants to search.
3683 .sp
3684 If the searching kind is an experimental kind
3685 (i.e. 253 or 254 according to RFC6994). It also
3686 needs to specify the \(dqmagic\(dq which is either
3687 2 bytes or 4 bytes. It then also needs to
3688 specify the size of the magic by using
3689 the 2nd byte which is \(dqkind\-length\(dq of a TCP
3690 header option and the \(dqkind\-length\(dq also
3691 includes the first 2 bytes \(dqkind\(dq and \(dqkind\-length\(dq
3692 itself as a normal TCP header option also does.
3693 .sp
3694 For example, to search experimental kind 254 with
3695 2 byte magic 0xeB9F, the searchby_res should be
3696 [ 254, 4, 0xeB, 0x9F, 0, 0, .... 0 ].
3697 .sp
3698 To search for the standard window scale option (3),
3699 the \fIsearchby_res\fP should be [ 3, 0, 0, .... 0 ].
3700 Note, kind\-length must be 0 for regular option.
3701 .sp
3702 Searching for No\-Op (0) and End\-of\-Option\-List (1) are
3703 not supported.
3704 .sp
3705 \fIlen\fP must be at least 2 bytes which is the minimal size
3706 of a header option.
3707 .sp
3708 Supported flags:
3709 .INDENT 7.0
3710 .IP \(bu 2
3711 \fBBPF_LOAD_HDR_OPT_TCP_SYN\fP to search from the
3712 saved_syn packet or the just\-received syn packet.
3713 .UNINDENT
3714 .TP
3715 .B Return
3716 > 0 when found, the header option is copied to \fIsearchby_res\fP\&.
3717 The return value is the total length copied. On failure, a
3718 negative error code is returned:
3719 .sp
3720 \fB\-EINVAL\fP if a parameter is invalid.
3721 .sp
3722 \fB\-ENOMSG\fP if the option is not found.
3723 .sp
3724 \fB\-ENOENT\fP if no syn packet is available when
3725 \fBBPF_LOAD_HDR_OPT_TCP_SYN\fP is used.
3726 .sp
3727 \fB\-ENOSPC\fP if there is not enough space. Only \fIlen\fP number of
3728 bytes are copied.
3729 .sp
3730 \fB\-EFAULT\fP on failure to parse the header options in the
3731 packet.
3732 .sp
3733 \fB\-EPERM\fP if the helper cannot be used under the current
3734 \fIskops\fP\fB\->op\fP\&.
3735 .UNINDENT
3736 .TP
3737 .B \fBlong bpf_store_hdr_opt(struct bpf_sock_ops *\fP\fIskops\fP\fB, const void *\fP\fIfrom\fP\fB, u32\fP \fIlen\fP\fB, u64\fP \fIflags\fP\fB)\fP
3738 .INDENT 7.0
3739 .TP
3740 .B Description
3741 Store header option. The data will be copied
3742 from buffer \fIfrom\fP with length \fIlen\fP to the TCP header.
3743 .sp
3744 The buffer \fIfrom\fP should have the whole option that
3745 includes the kind, kind\-length, and the actual
3746 option data. The \fIlen\fP must be at least kind\-length
3747 long. The kind\-length does not have to be 4 byte
3748 aligned. The kernel will take care of the padding
3749 and setting the 4 bytes aligned value to th\->doff.
3750 .sp
3751 This helper will check for duplicated option
3752 by searching the same option in the outgoing skb.
3753 .sp
3754 This helper can only be called during
3755 \fBBPF_SOCK_OPS_WRITE_HDR_OPT_CB\fP\&.
3756 .TP
3757 .B Return
3758 0 on success, or negative error in case of failure:
3759 .sp
3760 \fB\-EINVAL\fP If param is invalid.
3761 .sp
3762 \fB\-ENOSPC\fP if there is not enough space in the header.
3763 Nothing has been written
3764 .sp
3765 \fB\-EEXIST\fP if the option already exists.
3766 .sp
3767 \fB\-EFAULT\fP on failure to parse the existing header options.
3768 .sp
3769 \fB\-EPERM\fP if the helper cannot be used under the current
3770 \fIskops\fP\fB\->op\fP\&.
3771 .UNINDENT
3772 .TP
3773 .B \fBlong bpf_reserve_hdr_opt(struct bpf_sock_ops *\fP\fIskops\fP\fB, u32\fP \fIlen\fP\fB, u64\fP \fIflags\fP\fB)\fP
3774 .INDENT 7.0
3775 .TP
3776 .B Description
3777 Reserve \fIlen\fP bytes for the bpf header option. The
3778 space will be used by \fBbpf_store_hdr_opt\fP() later in
3779 \fBBPF_SOCK_OPS_WRITE_HDR_OPT_CB\fP\&.
3780 .sp
3781 If \fBbpf_reserve_hdr_opt\fP() is called multiple times,
3782 the total number of bytes will be reserved.
3783 .sp
3784 This helper can only be called during
3785 \fBBPF_SOCK_OPS_HDR_OPT_LEN_CB\fP\&.
3786 .TP
3787 .B Return
3788 0 on success, or negative error in case of failure:
3789 .sp
3790 \fB\-EINVAL\fP if a parameter is invalid.
3791 .sp
3792 \fB\-ENOSPC\fP if there is not enough space in the header.
3793 .sp
3794 \fB\-EPERM\fP if the helper cannot be used under the current
3795 \fIskops\fP\fB\->op\fP\&.
3796 .UNINDENT
3797 .TP
3798 .B \fBvoid *bpf_inode_storage_get(struct bpf_map *\fP\fImap\fP\fB, void *\fP\fIinode\fP\fB, void *\fP\fIvalue\fP\fB, u64\fP \fIflags\fP\fB)\fP
3799 .INDENT 7.0
3800 .TP
3801 .B Description
3802 Get a bpf_local_storage from an \fIinode\fP\&.
3803 .sp
3804 Logically, it could be thought of as getting the value from
3805 a \fImap\fP with \fIinode\fP as the \fBkey\fP\&. From this
3806 perspective, the usage is not much different from
3807 \fBbpf_map_lookup_elem\fP(\fImap\fP, \fB&\fP\fIinode\fP) except this
3808 helper enforces the key must be an inode and the map must also
3809 be a \fBBPF_MAP_TYPE_INODE_STORAGE\fP\&.
3810 .sp
3811 Underneath, the value is stored locally at \fIinode\fP instead of
3812 the \fImap\fP\&. The \fImap\fP is used as the bpf\-local\-storage
3813 \(dqtype\(dq. The bpf\-local\-storage \(dqtype\(dq (i.e. the \fImap\fP) is
3814 searched against all bpf_local_storage residing at \fIinode\fP\&.
3815 .sp
3816 An optional \fIflags\fP (\fBBPF_LOCAL_STORAGE_GET_F_CREATE\fP) can be
3817 used such that a new bpf_local_storage will be
3818 created if one does not exist. \fIvalue\fP can be used
3819 together with \fBBPF_LOCAL_STORAGE_GET_F_CREATE\fP to specify
3820 the initial value of a bpf_local_storage. If \fIvalue\fP is
3821 \fBNULL\fP, the new bpf_local_storage will be zero initialized.
3822 .TP
3823 .B Return
3824 A bpf_local_storage pointer is returned on success.
3825 .sp
3826 \fBNULL\fP if not found or there was an error in adding
3827 a new bpf_local_storage.
3828 .UNINDENT
3829 .TP
3830 .B \fBint bpf_inode_storage_delete(struct bpf_map *\fP\fImap\fP\fB, void *\fP\fIinode\fP\fB)\fP
3831 .INDENT 7.0
3832 .TP
3833 .B Description
3834 Delete a bpf_local_storage from an \fIinode\fP\&.
3835 .TP
3836 .B Return
3837 0 on success.
3838 .sp
3839 \fB\-ENOENT\fP if the bpf_local_storage cannot be found.
3840 .UNINDENT
3841 .TP
3842 .B \fBlong bpf_d_path(const struct path *\fP\fIpath\fP\fB, char *\fP\fIbuf\fP\fB, u32\fP \fIsz\fP\fB)\fP
3843 .INDENT 7.0
3844 .TP
3845 .B Description
3846 Return full path for given \fBstruct path\fP object, which
3847 needs to be the kernel BTF \fIpath\fP object. The path is
3848 returned in the provided buffer \fIbuf\fP of size \fIsz\fP and
3849 is zero terminated.
3850 .TP
3851 .B Return
3852 On success, the strictly positive length of the string,
3853 including the trailing NUL character. On error, a negative
3854 value.
3855 .UNINDENT
3856 .TP
3857 .B \fBlong bpf_copy_from_user(void *\fP\fIdst\fP\fB, u32\fP \fIsize\fP\fB, const void *\fP\fIuser_ptr\fP\fB)\fP
3858 .INDENT 7.0
3859 .TP
3860 .B Description
3861 Read \fIsize\fP bytes from user space address \fIuser_ptr\fP and store
3862 the data in \fIdst\fP\&. This is a wrapper of \fBcopy_from_user\fP().
3863 .TP
3864 .B Return
3865 0 on success, or a negative error in case of failure.
3866 .UNINDENT
3867 .TP
3868 .B \fBlong bpf_snprintf_btf(char *\fP\fIstr\fP\fB, u32\fP \fIstr_size\fP\fB, struct btf_ptr *\fP\fIptr\fP\fB, u32\fP \fIbtf_ptr_size\fP\fB, u64\fP \fIflags\fP\fB)\fP
3869 .INDENT 7.0
3870 .TP
3871 .B Description
3872 Use BTF to store a string representation of \fIptr\fP\->ptr in \fIstr\fP,
3873 using \fIptr\fP\->type_id. This value should specify the type
3874 that \fIptr\fP\->ptr points to. LLVM __builtin_btf_type_id(type, 1)
3875 can be used to look up vmlinux BTF type ids. Traversing the
3876 data structure using BTF, the type information and values are
3877 stored in the first \fIstr_size\fP \- 1 bytes of \fIstr\fP\&. Safe copy of
3878 the pointer data is carried out to avoid kernel crashes during
3879 operation. Smaller types can use string space on the stack;
3880 larger programs can use map data to store the string
3881 representation.
3882 .sp
3883 The string can be subsequently shared with userspace via
3884 bpf_perf_event_output() or ring buffer interfaces.
3885 bpf_trace_printk() is to be avoided as it places too small
3886 a limit on string size to be useful.
3887 .sp
3888 \fIflags\fP is a combination of
3889 .INDENT 7.0
3890 .TP
3891 .B \fBBTF_F_COMPACT\fP
3892 no formatting around type information
3893 .TP
3894 .B \fBBTF_F_NONAME\fP
3895 no struct/union member names/types
3896 .TP
3897 .B \fBBTF_F_PTR_RAW\fP
3898 show raw (unobfuscated) pointer values;
3899 equivalent to printk specifier %px.
3900 .TP
3901 .B \fBBTF_F_ZERO\fP
3902 show zero\-valued struct/union members; they
3903 are not displayed by default
3904 .UNINDENT
3905 .TP
3906 .B Return
3907 The number of bytes that were written (or would have been
3908 written if output had to be truncated due to string size),
3909 or a negative error in cases of failure.
3910 .UNINDENT
3911 .TP
3912 .B \fBlong bpf_seq_printf_btf(struct seq_file *\fP\fIm\fP\fB, struct btf_ptr *\fP\fIptr\fP\fB, u32\fP \fIptr_size\fP\fB, u64\fP \fIflags\fP\fB)\fP
3913 .INDENT 7.0
3914 .TP
3915 .B Description
3916 Use BTF to write to seq_write a string representation of
3917 \fIptr\fP\->ptr, using \fIptr\fP\->type_id as per bpf_snprintf_btf().
3918 \fIflags\fP are identical to those used for bpf_snprintf_btf.
3919 .TP
3920 .B Return
3921 0 on success or a negative error in case of failure.
3922 .UNINDENT
3923 .TP
3924 .B \fBu64 bpf_skb_cgroup_classid(struct sk_buff *\fP\fIskb\fP\fB)\fP
3925 .INDENT 7.0
3926 .TP
3927 .B Description
3928 See \fBbpf_get_cgroup_classid\fP() for the main description.
3929 This helper differs from \fBbpf_get_cgroup_classid\fP() in that
3930 the cgroup v1 net_cls class is retrieved only from the \fIskb\fP\(aqs
3931 associated socket instead of the current process.
3932 .TP
3933 .B Return
3934 The id is returned or 0 in case the id could not be retrieved.
3935 .UNINDENT
3936 .TP
3937 .B \fBlong bpf_redirect_neigh(u32\fP \fIifindex\fP\fB, struct bpf_redir_neigh *\fP\fIparams\fP\fB, int\fP \fIplen\fP\fB, u64\fP \fIflags\fP\fB)\fP
3938 .INDENT 7.0
3939 .TP
3940 .B Description
3941 Redirect the packet to another net device of index \fIifindex\fP
3942 and fill in L2 addresses from neighboring subsystem. This helper
3943 is somewhat similar to \fBbpf_redirect\fP(), except that it
3944 populates L2 addresses as well, meaning, internally, the helper
3945 relies on the neighbor lookup for the L2 address of the nexthop.
3946 .sp
3947 The helper will perform a FIB lookup based on the skb\(aqs
3948 networking header to get the address of the next hop, unless
3949 this is supplied by the caller in the \fIparams\fP argument. The
3950 \fIplen\fP argument indicates the len of \fIparams\fP and should be set
3951 to 0 if \fIparams\fP is NULL.
3952 .sp
3953 The \fIflags\fP argument is reserved and must be 0. The helper is
3954 currently only supported for tc BPF program types, and enabled
3955 for IPv4 and IPv6 protocols.
3956 .TP
3957 .B Return
3958 The helper returns \fBTC_ACT_REDIRECT\fP on success or
3959 \fBTC_ACT_SHOT\fP on error.
3960 .UNINDENT
3961 .TP
3962 .B \fBvoid *bpf_per_cpu_ptr(const void *\fP\fIpercpu_ptr\fP\fB, u32\fP \fIcpu\fP\fB)\fP
3963 .INDENT 7.0
3964 .TP
3965 .B Description
3966 Take a pointer to a percpu ksym, \fIpercpu_ptr\fP, and return a
3967 pointer to the percpu kernel variable on \fIcpu\fP\&. A ksym is an
3968 extern variable decorated with \(aq__ksym\(aq. For ksym, there is a
3969 global var (either static or global) defined of the same name
3970 in the kernel. The ksym is percpu if the global var is percpu.
3971 The returned pointer points to the global percpu var on \fIcpu\fP\&.
3972 .sp
3973 bpf_per_cpu_ptr() has the same semantic as per_cpu_ptr() in the
3974 kernel, except that bpf_per_cpu_ptr() may return NULL. This
3975 happens if \fIcpu\fP is larger than nr_cpu_ids. The caller of
3976 bpf_per_cpu_ptr() must check the returned value.
3977 .TP
3978 .B Return
3979 A pointer pointing to the kernel percpu variable on \fIcpu\fP, or
3980 NULL, if \fIcpu\fP is invalid.
3981 .UNINDENT
3982 .TP
3983 .B \fBvoid *bpf_this_cpu_ptr(const void *\fP\fIpercpu_ptr\fP\fB)\fP
3984 .INDENT 7.0
3985 .TP
3986 .B Description
3987 Take a pointer to a percpu ksym, \fIpercpu_ptr\fP, and return a
3988 pointer to the percpu kernel variable on this cpu. See the
3989 description of \(aqksym\(aq in \fBbpf_per_cpu_ptr\fP().
3990 .sp
3991 bpf_this_cpu_ptr() has the same semantic as this_cpu_ptr() in
3992 the kernel. Different from \fBbpf_per_cpu_ptr\fP(), it would
3993 never return NULL.
3994 .TP
3995 .B Return
3996 A pointer pointing to the kernel percpu variable on this cpu.
3997 .UNINDENT
3998 .TP
3999 .B \fBlong bpf_redirect_peer(u32\fP \fIifindex\fP\fB, u64\fP \fIflags\fP\fB)\fP
4000 .INDENT 7.0
4001 .TP
4002 .B Description
4003 Redirect the packet to another net device of index \fIifindex\fP\&.
4004 This helper is somewhat similar to \fBbpf_redirect\fP(), except
4005 that the redirection happens to the \fIifindex\fP\(aq peer device and
4006 the netns switch takes place from ingress to ingress without
4007 going through the CPU\(aqs backlog queue.
4008 .sp
4009 \fIskb\fP\fB\->mark\fP and \fIskb\fP\fB\->tstamp\fP are not cleared during
4010 the netns switch.
4011 .sp
4012 The \fIflags\fP argument is reserved and must be 0. The helper is
4013 currently only supported for tc BPF program types at the
4014 ingress hook and for veth and netkit target device types. The
4015 peer device must reside in a different network namespace.
4016 .TP
4017 .B Return
4018 The helper returns \fBTC_ACT_REDIRECT\fP on success or
4019 \fBTC_ACT_SHOT\fP on error.
4020 .UNINDENT
4021 .TP
4022 .B \fBvoid *bpf_task_storage_get(struct bpf_map *\fP\fImap\fP\fB, struct task_struct *\fP\fItask\fP\fB, void *\fP\fIvalue\fP\fB, u64\fP \fIflags\fP\fB)\fP
4023 .INDENT 7.0
4024 .TP
4025 .B Description
4026 Get a bpf_local_storage from the \fItask\fP\&.
4027 .sp
4028 Logically, it could be thought of as getting the value from
4029 a \fImap\fP with \fItask\fP as the \fBkey\fP\&. From this
4030 perspective, the usage is not much different from
4031 \fBbpf_map_lookup_elem\fP(\fImap\fP, \fB&\fP\fItask\fP) except this
4032 helper enforces the key must be a task_struct and the map must also
4033 be a \fBBPF_MAP_TYPE_TASK_STORAGE\fP\&.
4034 .sp
4035 Underneath, the value is stored locally at \fItask\fP instead of
4036 the \fImap\fP\&. The \fImap\fP is used as the bpf\-local\-storage
4037 \(dqtype\(dq. The bpf\-local\-storage \(dqtype\(dq (i.e. the \fImap\fP) is
4038 searched against all bpf_local_storage residing at \fItask\fP\&.
4039 .sp
4040 An optional \fIflags\fP (\fBBPF_LOCAL_STORAGE_GET_F_CREATE\fP) can be
4041 used such that a new bpf_local_storage will be
4042 created if one does not exist. \fIvalue\fP can be used
4043 together with \fBBPF_LOCAL_STORAGE_GET_F_CREATE\fP to specify
4044 the initial value of a bpf_local_storage. If \fIvalue\fP is
4045 \fBNULL\fP, the new bpf_local_storage will be zero initialized.
4046 .TP
4047 .B Return
4048 A bpf_local_storage pointer is returned on success.
4049 .sp
4050 \fBNULL\fP if not found or there was an error in adding
4051 a new bpf_local_storage.
4052 .UNINDENT
4053 .TP
4054 .B \fBlong bpf_task_storage_delete(struct bpf_map *\fP\fImap\fP\fB, struct task_struct *\fP\fItask\fP\fB)\fP
4055 .INDENT 7.0
4056 .TP
4057 .B Description
4058 Delete a bpf_local_storage from a \fItask\fP\&.
4059 .TP
4060 .B Return
4061 0 on success.
4062 .sp
4063 \fB\-ENOENT\fP if the bpf_local_storage cannot be found.
4064 .UNINDENT
4065 .TP
4066 .B \fBstruct task_struct *bpf_get_current_task_btf(void)\fP
4067 .INDENT 7.0
4068 .TP
4069 .B Description
4070 Return a BTF pointer to the \(dqcurrent\(dq task.
4071 This pointer can also be used in helpers that accept an
4072 \fIARG_PTR_TO_BTF_ID\fP of type \fItask_struct\fP\&.
4073 .TP
4074 .B Return
4075 Pointer to the current task.
4076 .UNINDENT
4077 .TP
4078 .B \fBlong bpf_bprm_opts_set(struct linux_binprm *\fP\fIbprm\fP\fB, u64\fP \fIflags\fP\fB)\fP
4079 .INDENT 7.0
4080 .TP
4081 .B Description
4082 Set or clear certain options on \fIbprm\fP:
4083 .sp
4084 \fBBPF_F_BPRM_SECUREEXEC\fP Set the secureexec bit
4085 which sets the \fBAT_SECURE\fP auxv for glibc. The bit
4086 is cleared if the flag is not specified.
4087 .TP
4088 .B Return
4089 \fB\-EINVAL\fP if invalid \fIflags\fP are passed, zero otherwise.
4090 .UNINDENT
4091 .TP
4092 .B \fBu64 bpf_ktime_get_coarse_ns(void)\fP
4093 .INDENT 7.0
4094 .TP
4095 .B Description
4096 Return a coarse\-grained version of the time elapsed since
4097 system boot, in nanoseconds. Does not include time the system
4098 was suspended.
4099 .sp
4100 See: \fBclock_gettime\fP(\fBCLOCK_MONOTONIC_COARSE\fP)
4101 .TP
4102 .B Return
4103 Current \fIktime\fP\&.
4104 .UNINDENT
4105 .TP
4106 .B \fBlong bpf_ima_inode_hash(struct inode *\fP\fIinode\fP\fB, void *\fP\fIdst\fP\fB, u32\fP \fIsize\fP\fB)\fP
4107 .INDENT 7.0
4108 .TP
4109 .B Description
4110 Returns the stored IMA hash of the \fIinode\fP (if it\(aqs available).
4111 If the hash is larger than \fIsize\fP, then only \fIsize\fP
4112 bytes will be copied to \fIdst\fP
4113 .TP
4114 .B Return
4115 The \fBhash_algo\fP is returned on success,
4116 \fB\-EOPNOTSUPP\fP if IMA is disabled or \fB\-EINVAL\fP if
4117 invalid arguments are passed.
4118 .UNINDENT
4119 .TP
4120 .B \fBstruct socket *bpf_sock_from_file(struct file *\fP\fIfile\fP\fB)\fP
4121 .INDENT 7.0
4122 .TP
4123 .B Description
4124 If the given file represents a socket, returns the associated
4125 socket.
4126 .TP
4127 .B Return
4128 A pointer to a struct socket on success or NULL if the file is
4129 not a socket.
4130 .UNINDENT
4131 .TP
4132 .B \fBlong bpf_check_mtu(void *\fP\fIctx\fP\fB, u32\fP \fIifindex\fP\fB, u32 *\fP\fImtu_len\fP\fB, s32\fP \fIlen_diff\fP\fB, u64\fP \fIflags\fP\fB)\fP
4133 .INDENT 7.0
4134 .TP
4135 .B Description
4136 Check packet size against exceeding MTU of net device (based
4137 on \fIifindex\fP). This helper will likely be used in combination
4138 with helpers that adjust/change the packet size.
4139 .sp
4140 The argument \fIlen_diff\fP can be used for querying with a planned
4141 size change. This allows to check MTU prior to changing packet
4142 ctx. Providing a \fIlen_diff\fP adjustment that is larger than the
4143 actual packet size (resulting in negative packet size) will in
4144 principle not exceed the MTU, which is why it is not considered
4145 a failure. Other BPF helpers are needed for performing the
4146 planned size change; therefore the responsibility for catching
4147 a negative packet size belongs in those helpers.
4148 .sp
4149 Specifying \fIifindex\fP zero means the MTU check is performed
4150 against the current net device. This is practical if this isn\(aqt
4151 used prior to redirect.
4152 .sp
4153 On input \fImtu_len\fP must be a valid pointer, else verifier will
4154 reject BPF program. If the value \fImtu_len\fP is initialized to
4155 zero then the ctx packet size is use. When value \fImtu_len\fP is
4156 provided as input this specify the L3 length that the MTU check
4157 is done against. Remember XDP and TC length operate at L2, but
4158 this value is L3 as this correlate to MTU and IP\-header tot_len
4159 values which are L3 (similar behavior as bpf_fib_lookup).
4160 .sp
4161 The Linux kernel route table can configure MTUs on a more
4162 specific per route level, which is not provided by this helper.
4163 For route level MTU checks use the \fBbpf_fib_lookup\fP()
4164 helper.
4165 .sp
4166 \fIctx\fP is either \fBstruct xdp_md\fP for XDP programs or
4167 \fBstruct sk_buff\fP for tc cls_act programs.
4168 .sp
4169 The \fIflags\fP argument can be a combination of one or more of the
4170 following values:
4171 .INDENT 7.0
4172 .TP
4173 .B \fBBPF_MTU_CHK_SEGS\fP
4174 This flag will only works for \fIctx\fP \fBstruct sk_buff\fP\&.
4175 If packet context contains extra packet segment buffers
4176 (often knows as GSO skb), then MTU check is harder to
4177 check at this point, because in transmit path it is
4178 possible for the skb packet to get re\-segmented
4179 (depending on net device features). This could still be
4180 a MTU violation, so this flag enables performing MTU
4181 check against segments, with a different violation
4182 return code to tell it apart. Check cannot use len_diff.
4183 .UNINDENT
4184 .sp
4185 On return \fImtu_len\fP pointer contains the MTU value of the net
4186 device. Remember the net device configured MTU is the L3 size,
4187 which is returned here and XDP and TC length operate at L2.
4188 Helper take this into account for you, but remember when using
4189 MTU value in your BPF\-code.
4190 .TP
4191 .B Return
4192 .INDENT 7.0
4193 .IP \(bu 2
4194 0 on success, and populate MTU value in \fImtu_len\fP pointer.
4195 .IP \(bu 2
4196 < 0 if any input argument is invalid (\fImtu_len\fP not updated)
4197 .UNINDENT
4198 .sp
4199 MTU violations return positive values, but also populate MTU
4200 value in \fImtu_len\fP pointer, as this can be needed for
4201 implementing PMTU handing:
4202 .INDENT 7.0
4203 .IP \(bu 2
4204 \fBBPF_MTU_CHK_RET_FRAG_NEEDED\fP
4205 .IP \(bu 2
4206 \fBBPF_MTU_CHK_RET_SEGS_TOOBIG\fP
4207 .UNINDENT
4208 .UNINDENT
4209 .TP
4210 .B \fBlong bpf_for_each_map_elem(struct bpf_map *\fP\fImap\fP\fB, void *\fP\fIcallback_fn\fP\fB, void *\fP\fIcallback_ctx\fP\fB, u64\fP \fIflags\fP\fB)\fP
4211 .INDENT 7.0
4212 .TP
4213 .B Description
4214 For each element in \fBmap\fP, call \fBcallback_fn\fP function with
4215 \fBmap\fP, \fBcallback_ctx\fP and other map\-specific parameters.
4216 The \fBcallback_fn\fP should be a static function and
4217 the \fBcallback_ctx\fP should be a pointer to the stack.
4218 The \fBflags\fP is used to control certain aspects of the helper.
4219 Currently, the \fBflags\fP must be 0.
4220 .sp
4221 The following are a list of supported map types and their
4222 respective expected callback signatures:
4223 .sp
4224 BPF_MAP_TYPE_HASH, BPF_MAP_TYPE_PERCPU_HASH,
4225 BPF_MAP_TYPE_LRU_HASH, BPF_MAP_TYPE_LRU_PERCPU_HASH,
4226 BPF_MAP_TYPE_ARRAY, BPF_MAP_TYPE_PERCPU_ARRAY
4227 .sp
4228 long (*callback_fn)(struct bpf_map *map, const void *key, void *value, void *ctx);
4229 .sp
4230 For per_cpu maps, the map_value is the value on the cpu where the
4231 bpf_prog is running.
4232 .sp
4233 If \fBcallback_fn\fP return 0, the helper will continue to the next
4234 element. If return value is 1, the helper will skip the rest of
4235 elements and return. Other return values are not used now.
4236 .TP
4237 .B Return
4238 The number of traversed map elements for success, \fB\-EINVAL\fP for
4239 invalid \fBflags\fP\&.
4240 .UNINDENT
4241 .TP
4242 .B \fBlong bpf_snprintf(char *\fP\fIstr\fP\fB, u32\fP \fIstr_size\fP\fB, const char *\fP\fIfmt\fP\fB, u64 *\fP\fIdata\fP\fB, u32\fP \fIdata_len\fP\fB)\fP
4243 .INDENT 7.0
4244 .TP
4245 .B Description
4246 Outputs a string into the \fBstr\fP buffer of size \fBstr_size\fP
4247 based on a format string stored in a read\-only map pointed by
4248 \fBfmt\fP\&.
4249 .sp
4250 Each format specifier in \fBfmt\fP corresponds to one u64 element
4251 in the \fBdata\fP array. For strings and pointers where pointees
4252 are accessed, only the pointer values are stored in the \fIdata\fP
4253 array. The \fIdata_len\fP is the size of \fIdata\fP in bytes \- must be
4254 a multiple of 8.
4255 .sp
4256 Formats \fB%s\fP and \fB%p{i,I}{4,6}\fP require to read kernel
4257 memory. Reading kernel memory may fail due to either invalid
4258 address or valid address but requiring a major memory fault. If
4259 reading kernel memory fails, the string for \fB%s\fP will be an
4260 empty string, and the ip address for \fB%p{i,I}{4,6}\fP will be 0.
4261 Not returning error to bpf program is consistent with what
4262 \fBbpf_trace_printk\fP() does for now.
4263 .TP
4264 .B Return
4265 The strictly positive length of the formatted string, including
4266 the trailing zero character. If the return value is greater than
4267 \fBstr_size\fP, \fBstr\fP contains a truncated string, guaranteed to
4268 be zero\-terminated except when \fBstr_size\fP is 0.
4269 .sp
4270 Or \fB\-EBUSY\fP if the per\-CPU memory copy buffer is busy.
4271 .UNINDENT
4272 .TP
4273 .B \fBlong bpf_sys_bpf(u32\fP \fIcmd\fP\fB, void *\fP\fIattr\fP\fB, u32\fP \fIattr_size\fP\fB)\fP
4274 .INDENT 7.0
4275 .TP
4276 .B Description
4277 Execute bpf syscall with given arguments.
4278 .TP
4279 .B Return
4280 A syscall result.
4281 .UNINDENT
4282 .TP
4283 .B \fBlong bpf_btf_find_by_name_kind(char *\fP\fIname\fP\fB, int\fP \fIname_sz\fP\fB, u32\fP \fIkind\fP\fB, int\fP \fIflags\fP\fB)\fP
4284 .INDENT 7.0
4285 .TP
4286 .B Description
4287 Find BTF type with given name and kind in vmlinux BTF or in module\(aqs BTFs.
4288 .TP
4289 .B Return
4290 Returns btf_id and btf_obj_fd in lower and upper 32 bits.
4291 .UNINDENT
4292 .TP
4293 .B \fBlong bpf_sys_close(u32\fP \fIfd\fP\fB)\fP
4294 .INDENT 7.0
4295 .TP
4296 .B Description
4297 Execute close syscall for given FD.
4298 .TP
4299 .B Return
4300 A syscall result.
4301 .UNINDENT
4302 .TP
4303 .B \fBlong bpf_timer_init(struct bpf_timer *\fP\fItimer\fP\fB, struct bpf_map *\fP\fImap\fP\fB, u64\fP \fIflags\fP\fB)\fP
4304 .INDENT 7.0
4305 .TP
4306 .B Description
4307 Initialize the timer.
4308 First 4 bits of \fIflags\fP specify clockid.
4309 Only CLOCK_MONOTONIC, CLOCK_REALTIME, CLOCK_BOOTTIME are allowed.
4310 All other bits of \fIflags\fP are reserved.
4311 The verifier will reject the program if \fItimer\fP is not from
4312 the same \fImap\fP\&.
4313 .TP
4314 .B Return
4315 0 on success.
4316 \fB\-EBUSY\fP if \fItimer\fP is already initialized.
4317 \fB\-EINVAL\fP if invalid \fIflags\fP are passed.
4318 \fB\-EPERM\fP if \fItimer\fP is in a map that doesn\(aqt have any user references.
4319 The user space should either hold a file descriptor to a map with timers
4320 or pin such map in bpffs. When map is unpinned or file descriptor is
4321 closed all timers in the map will be cancelled and freed.
4322 .UNINDENT
4323 .TP
4324 .B \fBlong bpf_timer_set_callback(struct bpf_timer *\fP\fItimer\fP\fB, void *\fP\fIcallback_fn\fP\fB)\fP
4325 .INDENT 7.0
4326 .TP
4327 .B Description
4328 Configure the timer to call \fIcallback_fn\fP static function.
4329 .TP
4330 .B Return
4331 0 on success.
4332 \fB\-EINVAL\fP if \fItimer\fP was not initialized with bpf_timer_init() earlier.
4333 \fB\-EPERM\fP if \fItimer\fP is in a map that doesn\(aqt have any user references.
4334 The user space should either hold a file descriptor to a map with timers
4335 or pin such map in bpffs. When map is unpinned or file descriptor is
4336 closed all timers in the map will be cancelled and freed.
4337 .UNINDENT
4338 .TP
4339 .B \fBlong bpf_timer_start(struct bpf_timer *\fP\fItimer\fP\fB, u64\fP \fInsecs\fP\fB, u64\fP \fIflags\fP\fB)\fP
4340 .INDENT 7.0
4341 .TP
4342 .B Description
4343 Set timer expiration N nanoseconds from the current time. The
4344 configured callback will be invoked in soft irq context on some cpu
4345 and will not repeat unless another bpf_timer_start() is made.
4346 In such case the next invocation can migrate to a different cpu.
4347 Since struct bpf_timer is a field inside map element the map
4348 owns the timer. The bpf_timer_set_callback() will increment refcnt
4349 of BPF program to make sure that callback_fn code stays valid.
4350 When user space reference to a map reaches zero all timers
4351 in a map are cancelled and corresponding program\(aqs refcnts are
4352 decremented. This is done to make sure that Ctrl\-C of a user
4353 process doesn\(aqt leave any timers running. If map is pinned in
4354 bpffs the callback_fn can re\-arm itself indefinitely.
4355 bpf_map_update/delete_elem() helpers and user space sys_bpf commands
4356 cancel and free the timer in the given map element.
4357 The map can contain timers that invoke callback_fn\-s from different
4358 programs. The same callback_fn can serve different timers from
4359 different maps if key/value layout matches across maps.
4360 Every bpf_timer_set_callback() can have different callback_fn.
4361 .sp
4362 \fIflags\fP can be one of:
4363 .INDENT 7.0
4364 .TP
4365 .B \fBBPF_F_TIMER_ABS\fP
4366 Start the timer in absolute expire value instead of the
4367 default relative one.
4368 .TP
4369 .B \fBBPF_F_TIMER_CPU_PIN\fP
4370 Timer will be pinned to the CPU of the caller.
4371 .UNINDENT
4372 .TP
4373 .B Return
4374 0 on success.
4375 \fB\-EINVAL\fP if \fItimer\fP was not initialized with bpf_timer_init() earlier
4376 or invalid \fIflags\fP are passed.
4377 .UNINDENT
4378 .TP
4379 .B \fBlong bpf_timer_cancel(struct bpf_timer *\fP\fItimer\fP\fB)\fP
4380 .INDENT 7.0
4381 .TP
4382 .B Description
4383 Cancel the timer and wait for callback_fn to finish if it was running.
4384 .TP
4385 .B Return
4386 0 if the timer was not active.
4387 1 if the timer was active.
4388 \fB\-EINVAL\fP if \fItimer\fP was not initialized with bpf_timer_init() earlier.
4389 \fB\-EDEADLK\fP if callback_fn tried to call bpf_timer_cancel() on its
4390 own timer which would have led to a deadlock otherwise.
4391 .UNINDENT
4392 .TP
4393 .B \fBu64 bpf_get_func_ip(void *\fP\fIctx\fP\fB)\fP
4394 .INDENT 7.0
4395 .TP
4396 .B Description
4397 Get address of the traced function (for tracing and kprobe programs).
4398 .sp
4399 When called for kprobe program attached as uprobe it returns
4400 probe address for both entry and return uprobe.
4401 .TP
4402 .B Return
4403 Address of the traced function for kprobe.
4404 0 for kprobes placed within the function (not at the entry).
4405 Address of the probe for uprobe and return uprobe.
4406 .UNINDENT
4407 .TP
4408 .B \fBu64 bpf_get_attach_cookie(void *\fP\fIctx\fP\fB)\fP
4409 .INDENT 7.0
4410 .TP
4411 .B Description
4412 Get bpf_cookie value provided (optionally) during the program
4413 attachment. It might be different for each individual
4414 attachment, even if BPF program itself is the same.
4415 Expects BPF program context \fIctx\fP as a first argument.
4416 .INDENT 7.0
4417 .TP
4418 .B Supported for the following program types:
4419 .INDENT 7.0
4420 .IP \(bu 2
4421 kprobe/uprobe;
4422 .IP \(bu 2
4423 tracepoint;
4424 .IP \(bu 2
4425 perf_event.
4426 .UNINDENT
4427 .UNINDENT
4428 .TP
4429 .B Return
4430 Value specified by user at BPF link creation/attachment time
4431 or 0, if it was not specified.
4432 .UNINDENT
4433 .TP
4434 .B \fBlong bpf_task_pt_regs(struct task_struct *\fP\fItask\fP\fB)\fP
4435 .INDENT 7.0
4436 .TP
4437 .B Description
4438 Get the struct pt_regs associated with \fBtask\fP\&.
4439 .TP
4440 .B Return
4441 A pointer to struct pt_regs.
4442 .UNINDENT
4443 .TP
4444 .B \fBlong bpf_get_branch_snapshot(void *\fP\fIentries\fP\fB, u32\fP \fIsize\fP\fB, u64\fP \fIflags\fP\fB)\fP
4445 .INDENT 7.0
4446 .TP
4447 .B Description
4448 Get branch trace from hardware engines like Intel LBR. The
4449 hardware engine is stopped shortly after the helper is
4450 called. Therefore, the user need to filter branch entries
4451 based on the actual use case. To capture branch trace
4452 before the trigger point of the BPF program, the helper
4453 should be called at the beginning of the BPF program.
4454 .sp
4455 The data is stored as struct perf_branch_entry into output
4456 buffer \fIentries\fP\&. \fIsize\fP is the size of \fIentries\fP in bytes.
4457 \fIflags\fP is reserved for now and must be zero.
4458 .TP
4459 .B Return
4460 On success, number of bytes written to \fIbuf\fP\&. On error, a
4461 negative value.
4462 .sp
4463 \fB\-EINVAL\fP if \fIflags\fP is not zero.
4464 .sp
4465 \fB\-ENOENT\fP if architecture does not support branch records.
4466 .UNINDENT
4467 .TP
4468 .B \fBlong bpf_trace_vprintk(const char *\fP\fIfmt\fP\fB, u32\fP \fIfmt_size\fP\fB, const void *\fP\fIdata\fP\fB, u32\fP \fIdata_len\fP\fB)\fP
4469 .INDENT 7.0
4470 .TP
4471 .B Description
4472 Behaves like \fBbpf_trace_printk\fP() helper, but takes an array of u64
4473 to format and can handle more format args as a result.
4474 .sp
4475 Arguments are to be used as in \fBbpf_seq_printf\fP() helper.
4476 .TP
4477 .B Return
4478 The number of bytes written to the buffer, or a negative error
4479 in case of failure.
4480 .UNINDENT
4481 .TP
4482 .B \fBstruct unix_sock *bpf_skc_to_unix_sock(void *\fP\fIsk\fP\fB)\fP
4483 .INDENT 7.0
4484 .TP
4485 .B Description
4486 Dynamically cast a \fIsk\fP pointer to a \fIunix_sock\fP pointer.
4487 .TP
4488 .B Return
4489 \fIsk\fP if casting is valid, or \fBNULL\fP otherwise.
4490 .UNINDENT
4491 .TP
4492 .B \fBlong bpf_kallsyms_lookup_name(const char *\fP\fIname\fP\fB, int\fP \fIname_sz\fP\fB, int\fP \fIflags\fP\fB, u64 *\fP\fIres\fP\fB)\fP
4493 .INDENT 7.0
4494 .TP
4495 .B Description
4496 Get the address of a kernel symbol, returned in \fIres\fP\&. \fIres\fP is
4497 set to 0 if the symbol is not found.
4498 .TP
4499 .B Return
4500 On success, zero. On error, a negative value.
4501 .sp
4502 \fB\-EINVAL\fP if \fIflags\fP is not zero.
4503 .sp
4504 \fB\-EINVAL\fP if string \fIname\fP is not the same size as \fIname_sz\fP\&.
4505 .sp
4506 \fB\-ENOENT\fP if symbol is not found.
4507 .sp
4508 \fB\-EPERM\fP if caller does not have permission to obtain kernel address.
4509 .UNINDENT
4510 .TP
4511 .B \fBlong bpf_find_vma(struct task_struct *\fP\fItask\fP\fB, u64\fP \fIaddr\fP\fB, void *\fP\fIcallback_fn\fP\fB, void *\fP\fIcallback_ctx\fP\fB, u64\fP \fIflags\fP\fB)\fP
4512 .INDENT 7.0
4513 .TP
4514 .B Description
4515 Find vma of \fItask\fP that contains \fIaddr\fP, call \fIcallback_fn\fP
4516 function with \fItask\fP, \fIvma\fP, and \fIcallback_ctx\fP\&.
4517 The \fIcallback_fn\fP should be a static function and
4518 the \fIcallback_ctx\fP should be a pointer to the stack.
4519 The \fIflags\fP is used to control certain aspects of the helper.
4520 Currently, the \fIflags\fP must be 0.
4521 .sp
4522 The expected callback signature is
4523 .sp
4524 long (*callback_fn)(struct task_struct *task, struct vm_area_struct *vma, void *callback_ctx);
4525 .TP
4526 .B Return
4527 0 on success.
4528 \fB\-ENOENT\fP if \fItask\->mm\fP is NULL, or no vma contains \fIaddr\fP\&.
4529 \fB\-EBUSY\fP if failed to try lock mmap_lock.
4530 \fB\-EINVAL\fP for invalid \fBflags\fP\&.
4531 .UNINDENT
4532 .TP
4533 .B \fBlong bpf_loop(u32\fP \fInr_loops\fP\fB, void *\fP\fIcallback_fn\fP\fB, void *\fP\fIcallback_ctx\fP\fB, u64\fP \fIflags\fP\fB)\fP
4534 .INDENT 7.0
4535 .TP
4536 .B Description
4537 For \fBnr_loops\fP, call \fBcallback_fn\fP function
4538 with \fBcallback_ctx\fP as the context parameter.
4539 The \fBcallback_fn\fP should be a static function and
4540 the \fBcallback_ctx\fP should be a pointer to the stack.
4541 The \fBflags\fP is used to control certain aspects of the helper.
4542 Currently, the \fBflags\fP must be 0. Currently, nr_loops is
4543 limited to 1 << 23 (~8 million) loops.
4544 .sp
4545 long (*callback_fn)(u64 index, void *ctx);
4546 .sp
4547 where \fBindex\fP is the current index in the loop. The index
4548 is zero\-indexed.
4549 .sp
4550 If \fBcallback_fn\fP returns 0, the helper will continue to the next
4551 loop. If return value is 1, the helper will skip the rest of
4552 the loops and return. Other return values are not used now,
4553 and will be rejected by the verifier.
4554 .TP
4555 .B Return
4556 The number of loops performed, \fB\-EINVAL\fP for invalid \fBflags\fP,
4557 \fB\-E2BIG\fP if \fBnr_loops\fP exceeds the maximum number of loops.
4558 .UNINDENT
4559 .TP
4560 .B \fBlong bpf_strncmp(const char *\fP\fIs1\fP\fB, u32\fP \fIs1_sz\fP\fB, const char *\fP\fIs2\fP\fB)\fP
4561 .INDENT 7.0
4562 .TP
4563 .B Description
4564 Do strncmp() between \fBs1\fP and \fBs2\fP\&. \fBs1\fP doesn\(aqt need
4565 to be null\-terminated and \fBs1_sz\fP is the maximum storage
4566 size of \fBs1\fP\&. \fBs2\fP must be a read\-only string.
4567 .TP
4568 .B Return
4569 An integer less than, equal to, or greater than zero
4570 if the first \fBs1_sz\fP bytes of \fBs1\fP is found to be
4571 less than, to match, or be greater than \fBs2\fP\&.
4572 .UNINDENT
4573 .TP
4574 .B \fBlong bpf_get_func_arg(void *\fP\fIctx\fP\fB, u32\fP \fIn\fP\fB, u64 *\fP\fIvalue\fP\fB)\fP
4575 .INDENT 7.0
4576 .TP
4577 .B Description
4578 Get \fBn\fP\-th argument register (zero based) of the traced function (for tracing programs)
4579 returned in \fBvalue\fP\&.
4580 .TP
4581 .B Return
4582 0 on success.
4583 \fB\-EINVAL\fP if n >= argument register count of traced function.
4584 .UNINDENT
4585 .TP
4586 .B \fBlong bpf_get_func_ret(void *\fP\fIctx\fP\fB, u64 *\fP\fIvalue\fP\fB)\fP
4587 .INDENT 7.0
4588 .TP
4589 .B Description
4590 Get return value of the traced function (for tracing programs)
4591 in \fBvalue\fP\&.
4592 .TP
4593 .B Return
4594 0 on success.
4595 \fB\-EOPNOTSUPP\fP for tracing programs other than BPF_TRACE_FEXIT or BPF_MODIFY_RETURN.
4596 .UNINDENT
4597 .TP
4598 .B \fBlong bpf_get_func_arg_cnt(void *\fP\fIctx\fP\fB)\fP
4599 .INDENT 7.0
4600 .TP
4601 .B Description
4602 Get number of registers of the traced function (for tracing programs) where
4603 function arguments are stored in these registers.
4604 .TP
4605 .B Return
4606 The number of argument registers of the traced function.
4607 .UNINDENT
4608 .TP
4609 .B \fBint bpf_get_retval(void)\fP
4610 .INDENT 7.0
4611 .TP
4612 .B Description
4613 Get the BPF program\(aqs return value that will be returned to the upper layers.
4614 .sp
4615 This helper is currently supported by cgroup programs and only by the hooks
4616 where BPF program\(aqs return value is returned to the userspace via errno.
4617 .TP
4618 .B Return
4619 The BPF program\(aqs return value.
4620 .UNINDENT
4621 .TP
4622 .B \fBint bpf_set_retval(int\fP \fIretval\fP\fB)\fP
4623 .INDENT 7.0
4624 .TP
4625 .B Description
4626 Set the BPF program\(aqs return value that will be returned to the upper layers.
4627 .sp
4628 This helper is currently supported by cgroup programs and only by the hooks
4629 where BPF program\(aqs return value is returned to the userspace via errno.
4630 .sp
4631 Note that there is the following corner case where the program exports an error
4632 via bpf_set_retval but signals success via \(aqreturn 1\(aq:
4633 .INDENT 7.0
4634 .INDENT 3.5
4635 bpf_set_retval(\-EPERM);
4636 return 1;
4637 .UNINDENT
4638 .UNINDENT
4639 .sp
4640 In this case, the BPF program\(aqs return value will use helper\(aqs \-EPERM. This
4641 still holds true for cgroup/bind{4,6} which supports extra \(aqreturn 3\(aq success case.
4642 .TP
4643 .B Return
4644 0 on success, or a negative error in case of failure.
4645 .UNINDENT
4646 .TP
4647 .B \fBu64 bpf_xdp_get_buff_len(struct xdp_buff *\fP\fIxdp_md\fP\fB)\fP
4648 .INDENT 7.0
4649 .TP
4650 .B Description
4651 Get the total size of a given xdp buff (linear and paged area)
4652 .TP
4653 .B Return
4654 The total size of a given xdp buffer.
4655 .UNINDENT
4656 .TP
4657 .B \fBlong bpf_xdp_load_bytes(struct xdp_buff *\fP\fIxdp_md\fP\fB, u32\fP \fIoffset\fP\fB, void *\fP\fIbuf\fP\fB, u32\fP \fIlen\fP\fB)\fP
4658 .INDENT 7.0
4659 .TP
4660 .B Description
4661 This helper is provided as an easy way to load data from a
4662 xdp buffer. It can be used to load \fIlen\fP bytes from \fIoffset\fP from
4663 the frame associated to \fIxdp_md\fP, into the buffer pointed by
4664 \fIbuf\fP\&.
4665 .TP
4666 .B Return
4667 0 on success, or a negative error in case of failure.
4668 .UNINDENT
4669 .TP
4670 .B \fBlong bpf_xdp_store_bytes(struct xdp_buff *\fP\fIxdp_md\fP\fB, u32\fP \fIoffset\fP\fB, void *\fP\fIbuf\fP\fB, u32\fP \fIlen\fP\fB)\fP
4671 .INDENT 7.0
4672 .TP
4673 .B Description
4674 Store \fIlen\fP bytes from buffer \fIbuf\fP into the frame
4675 associated to \fIxdp_md\fP, at \fIoffset\fP\&.
4676 .TP
4677 .B Return
4678 0 on success, or a negative error in case of failure.
4679 .UNINDENT
4680 .TP
4681 .B \fBlong bpf_copy_from_user_task(void *\fP\fIdst\fP\fB, u32\fP \fIsize\fP\fB, const void *\fP\fIuser_ptr\fP\fB, struct task_struct *\fP\fItsk\fP\fB, u64\fP \fIflags\fP\fB)\fP
4682 .INDENT 7.0
4683 .TP
4684 .B Description
4685 Read \fIsize\fP bytes from user space address \fIuser_ptr\fP in \fItsk\fP\(aqs
4686 address space, and stores the data in \fIdst\fP\&. \fIflags\fP is not
4687 used yet and is provided for future extensibility. This helper
4688 can only be used by sleepable programs.
4689 .TP
4690 .B Return
4691 0 on success, or a negative error in case of failure. On error
4692 \fIdst\fP buffer is zeroed out.
4693 .UNINDENT
4694 .TP
4695 .B \fBlong bpf_skb_set_tstamp(struct sk_buff *\fP\fIskb\fP\fB, u64\fP \fItstamp\fP\fB, u32\fP \fItstamp_type\fP\fB)\fP
4696 .INDENT 7.0
4697 .TP
4698 .B Description
4699 Change the __sk_buff\->tstamp_type to \fItstamp_type\fP
4700 and set \fItstamp\fP to the __sk_buff\->tstamp together.
4701 .sp
4702 If there is no need to change the __sk_buff\->tstamp_type,
4703 the tstamp value can be directly written to __sk_buff\->tstamp
4704 instead.
4705 .sp
4706 BPF_SKB_TSTAMP_DELIVERY_MONO is the only tstamp that
4707 will be kept during bpf_redirect_*(). A non zero
4708 \fItstamp\fP must be used with the BPF_SKB_TSTAMP_DELIVERY_MONO
4709 \fItstamp_type\fP\&.
4710 .sp
4711 A BPF_SKB_TSTAMP_UNSPEC \fItstamp_type\fP can only be used
4712 with a zero \fItstamp\fP\&.
4713 .sp
4714 Only IPv4 and IPv6 skb\->protocol are supported.
4715 .sp
4716 This function is most useful when it needs to set a
4717 mono delivery time to __sk_buff\->tstamp and then
4718 bpf_redirect_*() to the egress of an iface. For example,
4719 changing the (rcv) timestamp in __sk_buff\->tstamp at
4720 ingress to a mono delivery time and then bpf_redirect_*()
4721 to \%<sch_fq@\:phy-dev>\&.
4722 .TP
4723 .B Return
4724 0 on success.
4725 \fB\-EINVAL\fP for invalid input
4726 \fB\-EOPNOTSUPP\fP for unsupported protocol
4727 .UNINDENT
4728 .TP
4729 .B \fBlong bpf_ima_file_hash(struct file *\fP\fIfile\fP\fB, void *\fP\fIdst\fP\fB, u32\fP \fIsize\fP\fB)\fP
4730 .INDENT 7.0
4731 .TP
4732 .B Description
4733 Returns a calculated IMA hash of the \fIfile\fP\&.
4734 If the hash is larger than \fIsize\fP, then only \fIsize\fP
4735 bytes will be copied to \fIdst\fP
4736 .TP
4737 .B Return
4738 The \fBhash_algo\fP is returned on success,
4739 \fB\-EOPNOTSUPP\fP if the hash calculation failed or \fB\-EINVAL\fP if
4740 invalid arguments are passed.
4741 .UNINDENT
4742 .TP
4743 .B \fBvoid *bpf_kptr_xchg(void *\fP\fIdst\fP\fB, void *\fP\fIptr\fP\fB)\fP
4744 .INDENT 7.0
4745 .TP
4746 .B Description
4747 Exchange kptr at pointer \fIdst\fP with \fIptr\fP, and return the old value.
4748 \fIdst\fP can be map value or local kptr. \fIptr\fP can be NULL, otherwise
4749 it must be a referenced pointer which will be released when this helper
4750 is called.
4751 .TP
4752 .B Return
4753 The old value of kptr (which can be NULL). The returned pointer
4754 if not NULL, is a reference which must be released using its
4755 corresponding release function, or moved into a BPF map before
4756 program exit.
4757 .UNINDENT
4758 .TP
4759 .B \fBvoid *bpf_map_lookup_percpu_elem(struct bpf_map *\fP\fImap\fP\fB, const void *\fP\fIkey\fP\fB, u32\fP \fIcpu\fP\fB)\fP
4760 .INDENT 7.0
4761 .TP
4762 .B Description
4763 Perform a lookup in \fIpercpu map\fP for an entry associated to
4764 \fIkey\fP on \fIcpu\fP\&.
4765 .TP
4766 .B Return
4767 Map value associated to \fIkey\fP on \fIcpu\fP, or \fBNULL\fP if no entry
4768 was found or \fIcpu\fP is invalid.
4769 .UNINDENT
4770 .TP
4771 .B \fBstruct mptcp_sock *bpf_skc_to_mptcp_sock(void *\fP\fIsk\fP\fB)\fP
4772 .INDENT 7.0
4773 .TP
4774 .B Description
4775 Dynamically cast a \fIsk\fP pointer to a \fImptcp_sock\fP pointer.
4776 .TP
4777 .B Return
4778 \fIsk\fP if casting is valid, or \fBNULL\fP otherwise.
4779 .UNINDENT
4780 .TP
4781 .B \fBlong bpf_dynptr_from_mem(void *\fP\fIdata\fP\fB, u64\fP \fIsize\fP\fB, u64\fP \fIflags\fP\fB, struct bpf_dynptr *\fP\fIptr\fP\fB)\fP
4782 .INDENT 7.0
4783 .TP
4784 .B Description
4785 Get a dynptr to local memory \fIdata\fP\&.
4786 .sp
4787 \fIdata\fP must be a ptr to a map value.
4788 The maximum \fIsize\fP supported is DYNPTR_MAX_SIZE.
4789 \fIflags\fP is currently unused.
4790 .TP
4791 .B Return
4792 0 on success, \-E2BIG if the size exceeds DYNPTR_MAX_SIZE,
4793 \-EINVAL if flags is not 0.
4794 .UNINDENT
4795 .TP
4796 .B \fBlong bpf_ringbuf_reserve_dynptr(void *\fP\fIringbuf\fP\fB, u32\fP \fIsize\fP\fB, u64\fP \fIflags\fP\fB, struct bpf_dynptr *\fP\fIptr\fP\fB)\fP
4797 .INDENT 7.0
4798 .TP
4799 .B Description
4800 Reserve \fIsize\fP bytes of payload in a ring buffer \fIringbuf\fP
4801 through the dynptr interface. \fIflags\fP must be 0.
4802 .sp
4803 Please note that a corresponding bpf_ringbuf_submit_dynptr or
4804 bpf_ringbuf_discard_dynptr must be called on \fIptr\fP, even if the
4805 reservation fails. This is enforced by the verifier.
4806 .TP
4807 .B Return
4808 0 on success, or a negative error in case of failure.
4809 .UNINDENT
4810 .TP
4811 .B \fBvoid bpf_ringbuf_submit_dynptr(struct bpf_dynptr *\fP\fIptr\fP\fB, u64\fP \fIflags\fP\fB)\fP
4812 .INDENT 7.0
4813 .TP
4814 .B Description
4815 Submit reserved ring buffer sample, pointed to by \fIdata\fP,
4816 through the dynptr interface. This is a no\-op if the dynptr is
4817 invalid/null.
4818 .sp
4819 For more information on \fIflags\fP, please see
4820 \(aqbpf_ringbuf_submit\(aq.
4821 .TP
4822 .B Return
4823 Nothing. Always succeeds.
4824 .UNINDENT
4825 .TP
4826 .B \fBvoid bpf_ringbuf_discard_dynptr(struct bpf_dynptr *\fP\fIptr\fP\fB, u64\fP \fIflags\fP\fB)\fP
4827 .INDENT 7.0
4828 .TP
4829 .B Description
4830 Discard reserved ring buffer sample through the dynptr
4831 interface. This is a no\-op if the dynptr is invalid/null.
4832 .sp
4833 For more information on \fIflags\fP, please see
4834 \(aqbpf_ringbuf_discard\(aq.
4835 .TP
4836 .B Return
4837 Nothing. Always succeeds.
4838 .UNINDENT
4839 .TP
4840 .B \fBlong bpf_dynptr_read(void *\fP\fIdst\fP\fB, u64\fP \fIlen\fP\fB, const struct bpf_dynptr *\fP\fIsrc\fP\fB, u64\fP \fIoffset\fP\fB, u64\fP \fIflags\fP\fB)\fP
4841 .INDENT 7.0
4842 .TP
4843 .B Description
4844 Read \fIlen\fP bytes from \fIsrc\fP into \fIdst\fP, starting from \fIoffset\fP
4845 into \fIsrc\fP\&.
4846 \fIflags\fP is currently unused.
4847 .TP
4848 .B Return
4849 0 on success, \-E2BIG if \fIoffset\fP + \fIlen\fP exceeds the length
4850 of \fIsrc\fP\(aqs data, \-EINVAL if \fIsrc\fP is an invalid dynptr or if
4851 \fIflags\fP is not 0.
4852 .UNINDENT
4853 .TP
4854 .B \fBlong bpf_dynptr_write(const struct bpf_dynptr *\fP\fIdst\fP\fB, u64\fP \fIoffset\fP\fB, void *\fP\fIsrc\fP\fB, u64\fP \fIlen\fP\fB, u64\fP \fIflags\fP\fB)\fP
4855 .INDENT 7.0
4856 .TP
4857 .B Description
4858 Write \fIlen\fP bytes from \fIsrc\fP into \fIdst\fP, starting from \fIoffset\fP
4859 into \fIdst\fP\&.
4860 .sp
4861 \fIflags\fP must be 0 except for skb\-type dynptrs.
4862 .INDENT 7.0
4863 .TP
4864 .B For skb\-type dynptrs:
4865 .INDENT 7.0
4866 .IP \(bu 2
4867 All data slices of the dynptr are automatically
4868 invalidated after \fBbpf_dynptr_write\fP(). This is
4869 because writing may pull the skb and change the
4870 underlying packet buffer.
4871 .IP \(bu 2
4872 For \fIflags\fP, please see the flags accepted by
4873 \fBbpf_skb_store_bytes\fP().
4874 .UNINDENT
4875 .UNINDENT
4876 .TP
4877 .B Return
4878 0 on success, \-E2BIG if \fIoffset\fP + \fIlen\fP exceeds the length
4879 of \fIdst\fP\(aqs data, \-EINVAL if \fIdst\fP is an invalid dynptr or if \fIdst\fP
4880 is a read\-only dynptr or if \fIflags\fP is not correct. For skb\-type dynptrs,
4881 other errors correspond to errors returned by \fBbpf_skb_store_bytes\fP().
4882 .UNINDENT
4883 .TP
4884 .B \fBvoid *bpf_dynptr_data(const struct bpf_dynptr *\fP\fIptr\fP\fB, u64\fP \fIoffset\fP\fB, u64\fP \fIlen\fP\fB)\fP
4885 .INDENT 7.0
4886 .TP
4887 .B Description
4888 Get a pointer to the underlying dynptr data.
4889 .sp
4890 \fIlen\fP must be a statically known value. The returned data slice
4891 is invalidated whenever the dynptr is invalidated.
4892 .sp
4893 skb and xdp type dynptrs may not use bpf_dynptr_data. They should
4894 instead use bpf_dynptr_slice and bpf_dynptr_slice_rdwr.
4895 .TP
4896 .B Return
4897 Pointer to the underlying dynptr data, NULL if the dynptr is
4898 read\-only, if the dynptr is invalid, or if the offset and length
4899 is out of bounds.
4900 .UNINDENT
4901 .TP
4902 .B \fBs64 bpf_tcp_raw_gen_syncookie_ipv4(struct iphdr *\fP\fIiph\fP\fB, struct tcphdr *\fP\fIth\fP\fB, u32\fP \fIth_len\fP\fB)\fP
4903 .INDENT 7.0
4904 .TP
4905 .B Description
4906 Try to issue a SYN cookie for the packet with corresponding
4907 IPv4/TCP headers, \fIiph\fP and \fIth\fP, without depending on a
4908 listening socket.
4909 .sp
4910 \fIiph\fP points to the IPv4 header.
4911 .sp
4912 \fIth\fP points to the start of the TCP header, while \fIth_len\fP
4913 contains the length of the TCP header (at least
4914 \fBsizeof\fP(\fBstruct tcphdr\fP)).
4915 .TP
4916 .B Return
4917 On success, lower 32 bits hold the generated SYN cookie in
4918 followed by 16 bits which hold the MSS value for that cookie,
4919 and the top 16 bits are unused.
4920 .sp
4921 On failure, the returned value is one of the following:
4922 .sp
4923 \fB\-EINVAL\fP if \fIth_len\fP is invalid.
4924 .UNINDENT
4925 .TP
4926 .B \fBs64 bpf_tcp_raw_gen_syncookie_ipv6(struct ipv6hdr *\fP\fIiph\fP\fB, struct tcphdr *\fP\fIth\fP\fB, u32\fP \fIth_len\fP\fB)\fP
4927 .INDENT 7.0
4928 .TP
4929 .B Description
4930 Try to issue a SYN cookie for the packet with corresponding
4931 IPv6/TCP headers, \fIiph\fP and \fIth\fP, without depending on a
4932 listening socket.
4933 .sp
4934 \fIiph\fP points to the IPv6 header.
4935 .sp
4936 \fIth\fP points to the start of the TCP header, while \fIth_len\fP
4937 contains the length of the TCP header (at least
4938 \fBsizeof\fP(\fBstruct tcphdr\fP)).
4939 .TP
4940 .B Return
4941 On success, lower 32 bits hold the generated SYN cookie in
4942 followed by 16 bits which hold the MSS value for that cookie,
4943 and the top 16 bits are unused.
4944 .sp
4945 On failure, the returned value is one of the following:
4946 .sp
4947 \fB\-EINVAL\fP if \fIth_len\fP is invalid.
4948 .sp
4949 \fB\-EPROTONOSUPPORT\fP if CONFIG_IPV6 is not builtin.
4950 .UNINDENT
4951 .TP
4952 .B \fBlong bpf_tcp_raw_check_syncookie_ipv4(struct iphdr *\fP\fIiph\fP\fB, struct tcphdr *\fP\fIth\fP\fB)\fP
4953 .INDENT 7.0
4954 .TP
4955 .B Description
4956 Check whether \fIiph\fP and \fIth\fP contain a valid SYN cookie ACK
4957 without depending on a listening socket.
4958 .sp
4959 \fIiph\fP points to the IPv4 header.
4960 .sp
4961 \fIth\fP points to the TCP header.
4962 .TP
4963 .B Return
4964 0 if \fIiph\fP and \fIth\fP are a valid SYN cookie ACK.
4965 .sp
4966 On failure, the returned value is one of the following:
4967 .sp
4968 \fB\-EACCES\fP if the SYN cookie is not valid.
4969 .UNINDENT
4970 .TP
4971 .B \fBlong bpf_tcp_raw_check_syncookie_ipv6(struct ipv6hdr *\fP\fIiph\fP\fB, struct tcphdr *\fP\fIth\fP\fB)\fP
4972 .INDENT 7.0
4973 .TP
4974 .B Description
4975 Check whether \fIiph\fP and \fIth\fP contain a valid SYN cookie ACK
4976 without depending on a listening socket.
4977 .sp
4978 \fIiph\fP points to the IPv6 header.
4979 .sp
4980 \fIth\fP points to the TCP header.
4981 .TP
4982 .B Return
4983 0 if \fIiph\fP and \fIth\fP are a valid SYN cookie ACK.
4984 .sp
4985 On failure, the returned value is one of the following:
4986 .sp
4987 \fB\-EACCES\fP if the SYN cookie is not valid.
4988 .sp
4989 \fB\-EPROTONOSUPPORT\fP if CONFIG_IPV6 is not builtin.
4990 .UNINDENT
4991 .TP
4992 .B \fBu64 bpf_ktime_get_tai_ns(void)\fP
4993 .INDENT 7.0
4994 .TP
4995 .B Description
4996 A nonsettable system\-wide clock derived from wall\-clock time but
4997 ignoring leap seconds. This clock does not experience
4998 discontinuities and backwards jumps caused by NTP inserting leap
4999 seconds as CLOCK_REALTIME does.
5000 .sp
5001 See: \fBclock_gettime\fP(\fBCLOCK_TAI\fP)
5002 .TP
5003 .B Return
5004 Current \fIktime\fP\&.
5005 .UNINDENT
5006 .TP
5007 .B \fBlong bpf_user_ringbuf_drain(struct bpf_map *\fP\fImap\fP\fB, void *\fP\fIcallback_fn\fP\fB, void *\fP\fIctx\fP\fB, u64\fP \fIflags\fP\fB)\fP
5008 .INDENT 7.0
5009 .TP
5010 .B Description
5011 Drain samples from the specified user ring buffer, and invoke
5012 the provided callback for each such sample:
5013 .sp
5014 long (*callback_fn)(const struct bpf_dynptr *dynptr, void *ctx);
5015 .sp
5016 If \fBcallback_fn\fP returns 0, the helper will continue to try
5017 and drain the next sample, up to a maximum of
5018 BPF_MAX_USER_RINGBUF_SAMPLES samples. If the return value is 1,
5019 the helper will skip the rest of the samples and return. Other
5020 return values are not used now, and will be rejected by the
5021 verifier.
5022 .TP
5023 .B Return
5024 The number of drained samples if no error was encountered while
5025 draining samples, or 0 if no samples were present in the ring
5026 buffer. If a user\-space producer was epoll\-waiting on this map,
5027 and at least one sample was drained, they will receive an event
5028 notification notifying them of available space in the ring
5029 buffer. If the BPF_RB_NO_WAKEUP flag is passed to this
5030 function, no wakeup notification will be sent. If the
5031 BPF_RB_FORCE_WAKEUP flag is passed, a wakeup notification will
5032 be sent even if no sample was drained.
5033 .sp
5034 On failure, the returned value is one of the following:
5035 .sp
5036 \fB\-EBUSY\fP if the ring buffer is contended, and another calling
5037 context was concurrently draining the ring buffer.
5038 .sp
5039 \fB\-EINVAL\fP if user\-space is not properly tracking the ring
5040 buffer due to the producer position not being aligned to 8
5041 bytes, a sample not being aligned to 8 bytes, or the producer
5042 position not matching the advertised length of a sample.
5043 .sp
5044 \fB\-E2BIG\fP if user\-space has tried to publish a sample which is
5045 larger than the size of the ring buffer, or which cannot fit
5046 within a struct bpf_dynptr.
5047 .UNINDENT
5048 .TP
5049 .B \fBvoid *bpf_cgrp_storage_get(struct bpf_map *\fP\fImap\fP\fB, struct cgroup *\fP\fIcgroup\fP\fB, void *\fP\fIvalue\fP\fB, u64\fP \fIflags\fP\fB)\fP
5050 .INDENT 7.0
5051 .TP
5052 .B Description
5053 Get a bpf_local_storage from the \fIcgroup\fP\&.
5054 .sp
5055 Logically, it could be thought of as getting the value from
5056 a \fImap\fP with \fIcgroup\fP as the \fBkey\fP\&. From this
5057 perspective, the usage is not much different from
5058 \fBbpf_map_lookup_elem\fP(\fImap\fP, \fB&\fP\fIcgroup\fP) except this
5059 helper enforces the key must be a cgroup struct and the map must also
5060 be a \fBBPF_MAP_TYPE_CGRP_STORAGE\fP\&.
5061 .sp
5062 In reality, the local\-storage value is embedded directly inside of the
5063 \fIcgroup\fP object itself, rather than being located in the
5064 \fBBPF_MAP_TYPE_CGRP_STORAGE\fP map. When the local\-storage value is
5065 queried for some \fImap\fP on a \fIcgroup\fP object, the kernel will perform an
5066 O(n) iteration over all of the live local\-storage values for that
5067 \fIcgroup\fP object until the local\-storage value for the \fImap\fP is found.
5068 .sp
5069 An optional \fIflags\fP (\fBBPF_LOCAL_STORAGE_GET_F_CREATE\fP) can be
5070 used such that a new bpf_local_storage will be
5071 created if one does not exist. \fIvalue\fP can be used
5072 together with \fBBPF_LOCAL_STORAGE_GET_F_CREATE\fP to specify
5073 the initial value of a bpf_local_storage. If \fIvalue\fP is
5074 \fBNULL\fP, the new bpf_local_storage will be zero initialized.
5075 .TP
5076 .B Return
5077 A bpf_local_storage pointer is returned on success.
5078 .sp
5079 \fBNULL\fP if not found or there was an error in adding
5080 a new bpf_local_storage.
5081 .UNINDENT
5082 .TP
5083 .B \fBlong bpf_cgrp_storage_delete(struct bpf_map *\fP\fImap\fP\fB, struct cgroup *\fP\fIcgroup\fP\fB)\fP
5084 .INDENT 7.0
5085 .TP
5086 .B Description
5087 Delete a bpf_local_storage from a \fIcgroup\fP\&.
5088 .TP
5089 .B Return
5090 0 on success.
5091 .sp
5092 \fB\-ENOENT\fP if the bpf_local_storage cannot be found.
5093 .UNINDENT
5094 .UNINDENT
5095 .SH EXAMPLES
5096 .sp
5097 Example usage for most of the eBPF helpers listed in this manual page are
5098 available within the Linux kernel sources, at the following locations:
5099 .INDENT 0.0
5100 .IP \(bu 2
5101 \fIsamples/bpf/\fP
5102 .IP \(bu 2
5103 \fItools/testing/selftests/bpf/\fP
5104 .UNINDENT
5105 .SH LICENSE
5106 .sp
5107 eBPF programs can have an associated license, passed along with the bytecode
5108 instructions to the kernel when the programs are loaded. The format for that
5109 string is identical to the one in use for kernel modules (Dual licenses, such
5110 as \(dqDual BSD/GPL\(dq, may be used). Some helper functions are only accessible to
5111 programs that are compatible with the GNU General Public License (GNU GPL).
5112 .sp
5113 In order to use such helpers, the eBPF program must be loaded with the correct
5114 license string passed (via \fBattr\fP) to the \fBbpf\fP() system call, and this
5115 generally translates into the C source code of the program containing a line
5116 similar to the following:
5117 .INDENT 0.0
5118 .INDENT 3.5
5119 .sp
5120 .EX
5121 char ____license[] __attribute__((section(\(dqlicense\(dq), used)) = \(dqGPL\(dq;
5122 .EE
5123 .UNINDENT
5124 .UNINDENT
5125 .SH IMPLEMENTATION
5126 .sp
5127 This manual page is an effort to document the existing eBPF helper functions.
5128 But as of this writing, the BPF sub\-system is under heavy development. New eBPF
5129 program or map types are added, along with new helper functions. Some helpers
5130 are occasionally made available for additional program types. So in spite of
5131 the efforts of the community, this page might not be up\-to\-date. If you want to
5132 check by yourself what helper functions exist in your kernel, or what types of
5133 programs they can support, here are some files among the kernel tree that you
5134 may be interested in:
5135 .INDENT 0.0
5136 .IP \(bu 2
5137 \fIinclude/uapi/linux/bpf.h\fP is the main BPF header. It contains the full list
5138 of all helper functions, as well as many other BPF definitions including most
5139 of the flags, structs or constants used by the helpers.
5140 .IP \(bu 2
5141 \fInet/core/filter.c\fP contains the definition of most network\-related helper
5142 functions, and the list of program types from which they can be used.
5143 .IP \(bu 2
5144 \fIkernel/trace/bpf_trace.c\fP is the equivalent for most tracing program\-related
5145 helpers.
5146 .IP \(bu 2
5147 \fIkernel/bpf/verifier.c\fP contains the functions used to check that valid types
5148 of eBPF maps are used with a given helper function.
5149 .IP \(bu 2
5150 \fIkernel/bpf/\fP directory contains other files in which additional helpers are
5151 defined (for cgroups, sockmaps, etc.).
5152 .IP \(bu 2
5153 The bpftool utility can be used to probe the availability of helper functions
5154 on the system (as well as supported program and map types, and a number of
5155 other parameters). To do so, run \fBbpftool feature probe\fP (see
5156 \fBbpftool\-feature\fP(8) for details). Add the \fBunprivileged\fP keyword to
5157 list features available to unprivileged users.
5158 .UNINDENT
5159 .sp
5160 Compatibility between helper functions and program types can generally be found
5161 in the files where helper functions are defined. Look for the \fBstruct
5162 bpf_func_proto\fP objects and for functions returning them: these functions
5163 contain a list of helpers that a given program type can call. Note that the
5164 \fBdefault:\fP label of the \fBswitch ... case\fP used to filter helpers can call
5165 other functions, themselves allowing access to additional helpers. The
5166 requirement for GPL license is also in those \fBstruct bpf_func_proto\fP\&.
5167 .sp
5168 Compatibility between helper functions and map types can be found in the
5169 \fBcheck_map_func_compatibility\fP() function in file \fIkernel/bpf/verifier.c\fP\&.
5170 .sp
5171 Helper functions that invalidate the checks on \fBdata\fP and \fBdata_end\fP
5172 pointers for network processing are listed in function
5173 \fBbpf_helper_changes_pkt_data\fP() in file \fInet/core/filter.c\fP\&.
5174 .SH SEE ALSO
5175 .sp
5176 \fBbpf\fP(2),
5177 \fBbpftool\fP(8),
5178 \fBcgroups\fP(7),
5179 \fBip\fP(8),
5180 \fBperf_event_open\fP(2),
5181 \fBsendmsg\fP(2),
5182 \fBsocket\fP(7),
5183 \fBtc\-bpf\fP(8)
5184 .\" End of generated man page.