1BPFTOOL-MAP(8) BPFTOOL-MAP(8)
2
6 bpftool-map - tool for inspection and simple manipulation of eBPF maps
7
9 bpftool [OPTIONS] map COMMAND
10 OPTIONS := { { -j | --json } [{ -p | --pretty }] | { -f | --bpffs }
12 }
13 COMMANDS := { show | list | create | dump | update | lookup | get‐
15 next | delete | pin | help }
16
18 bpftool map { show | list } [MAP]
19 bpftool map create FILE type TYPE key KEY_SIZE value VALUE_SIZE
20 entries MAX_ENTRIES name NAME [flags FLAGS] [inner_map MAP]
21 [dev NAME]
22 bpftool map dump MAP
23 bpftool map update MAP [key DATA] [value VALUE] [UPDATE_FLAGS]
24 bpftool map lookup MAP [key DATA]
25 bpftool map getnext MAP [key DATA]
26 bpftool map delete MAP key DATA
27 bpftool map pin MAP FILE
28 bpftool map event_pipe MAP [cpu N index M]
29 bpftool map peek MAP
30 bpftool map push MAP value VALUE
31 bpftool map pop MAP
32 bpftool map enqueue MAP value VALUE
33 bpftool map dequeue MAP
34 bpftool map freeze MAP
35 bpftool map help
36 MAP := { id MAP_ID | pinned FILE | name MAP_NAME }
38 DATA := { [hex] BYTES }
39 PROG := { id PROG_ID | pinned FILE | tag PROG_TAG | name PROG_NAME }
40 VALUE := { DATA | MAP | PROG }
41 UPDATE_FLAGS := { any | exist | noexist }
42 TYPE := { hash | array | prog_array | perf_event_array | percpu_hash
43 | percpu_array | stack_trace | cgroup_array | lru_hash
44 | lru_percpu_hash | lpm_trie | array_of_maps | hash_of_maps
45 | devmap | devmap_hash | sockmap | cpumap | xskmap | sockhash
46 | cgroup_storage | reuseport_sockarray | percpu_cgroup_storage
47 | queue | stack | sk_storage | struct_ops | ringbuf | inode_storage
48 | task_storage }
49
52 bpftool map { show | list } [MAP]
53 Show information about loaded maps. If MAP is specified show
54 information only about given maps, otherwise list all maps
55 currently loaded on the system. In case of name, MAP may
56 match several maps which will all be shown.
57 Output will start with map ID followed by map type and zero
59 or more named attributes (depending on kernel version).
60 Since Linux 5.8 bpftool is able to discover information about
62 processes that hold open file descriptors (FDs) against BPF
63 maps. On such kernels bpftool will automatically emit this
64 information as well.
65 bpftool map create FILE type TYPE key KEY_SIZE value VALUE_SIZE
67 entries MAX_ENTRIES name NAME [flags FLAGS] [inner_map MAP] [dev
68 NAME]
69 Create a new map with given parameters and pin it to bpffs as
70 FILE.
71 FLAGS should be an integer which is the combination of
73 desired flags, e.g. 1024 for BPF_F_MMAPABLE (see bpf.h UAPI
74 header for existing flags).
75 To create maps of type array-of-maps or hash-of-maps, the
77 inner_map keyword must be used to pass an inner map. The ker‐
78 nel needs it to collect metadata related to the inner maps
79 that the new map will work with.
80 Keyword dev expects a network interface name, and is used to
82 request hardware offload for the map.
83 bpftool map dump MAP
85 Dump all entries in a given MAP. In case of name, MAP may
86 match several maps which will all be dumped.
87 bpftool map update MAP [key DATA] [value VALUE] [UPDATE_FLAGS]
89 Update map entry for a given KEY.
90 UPDATE_FLAGS can be one of: any update existing entry or add
92 if doesn't exit; exist update only if entry already exists;
93 noexist update only if entry doesn't exist.
94 If the hex keyword is provided in front of the bytes
96 sequence, the bytes are parsed as hexadecimal values, even if
97 no "0x" prefix is added. If the keyword is not provided, then
98 the bytes are parsed as decimal values, unless a "0x" prefix
99 (for hexadecimal) or a "0" prefix (for octal) is provided.
100 bpftool map lookup MAP [key DATA]
102 Lookup key in the map.
103 bpftool map getnext MAP [key DATA]
105 Get next key. If key is not specified, get first key.
106 bpftool map delete MAP key DATA
108 Remove entry from the map.
109 bpftool map pin MAP FILE
111 Pin map MAP as FILE.
112 Note: FILE must be located in bpffs mount. It must not con‐
114 tain a dot character ('.'), which is reserved for future
115 extensions of bpffs.
116 bpftool map event_pipe MAP [cpu N index M]
118 Read events from a BPF_MAP_TYPE_PERF_EVENT_ARRAY map.
119 Install perf rings into a perf event array map and dump out‐
121 put of any bpf_perf_event_output() call in the kernel. By
122 default read the number of CPUs on the system and install
123 perf ring for each CPU in the corresponding index in the
124 array.
125 If cpu and index are specified, install perf ring for given
127 cpu at index in the array (single ring).
128 Note that installing a perf ring into an array will silently
130 replace any existing ring. Any other application will stop
131 receiving events if it installed its rings earlier.
132 bpftool map peek MAP
134 Peek next value in the queue or stack.
135 bpftool map push MAP value VALUE
137 Push VALUE onto the stack.
138 bpftool map pop MAP
140 Pop and print value from the stack.
141 bpftool map enqueue MAP value VALUE
143 Enqueue VALUE into the queue.
144 bpftool map dequeue MAP
146 Dequeue and print value from the queue.
147 bpftool map freeze MAP
149 Freeze the map as read-only from user space. Entries from a
150 frozen map can not longer be updated or deleted with the
151 bpf() system call. This operation is not reversible, and the
152 map remains immutable from user space until its destruction.
153 However, read and write permissions for BPF programs to the
154 map remain unchanged.
155 bpftool map help
157 Print short help message.
158
160 -h, --help
161 Print short help message (similar to bpftool help).
162 -V, --version
164 Print version number (similar to bpftool version), and
165 optional features that were included when bpftool was com‐
166 piled. Optional features include linking against libbfd to
167 provide the disassembler for JIT-ted programs (bpftool prog
168 dump jited) and usage of BPF skeletons (some features like
169 bpftool prog profile or showing pids associated to BPF
170 objects may rely on it).
171 -j, --json
173 Generate JSON output. For commands that cannot produce JSON,
174 this option has no effect.
175 -p, --pretty
177 Generate human-readable JSON output. Implies -j.
178 -d, --debug
180 Print all logs available, even debug-level information. This
181 includes logs from libbpf as well as from the verifier, when
182 attempting to load programs.
183 -f, --bpffs
185 Show file names of pinned maps.
186 -n, --nomount
188 Do not automatically attempt to mount any virtual file system
189 (such as tracefs or BPF virtual file system) when necessary.
190
192 # bpftool map show
193 10: hash name some_map flags 0x0
195 key 4B value 8B max_entries 2048 memlock 167936B
196 pids systemd(1)
197 The following three commands are equivalent:
199 # bpftool map update id 10 key hex 20 c4 b7 00 value hex 0f ff ff ab 01 02 03 4c
201 # bpftool map update id 10 key 0x20 0xc4 0xb7 0x00 value 0x0f 0xff 0xff 0xab 0x01 0x02 0x03 0x4c
202 # bpftool map update id 10 key 32 196 183 0 value 15 255 255 171 1 2 3 76
203 # bpftool map lookup id 10 key 0 1 2 3
206 key: 00 01 02 03 value: 00 01 02 03 04 05 06 07
208 # bpftool map dump id 10
210 key: 00 01 02 03 value: 00 01 02 03 04 05 06 07
212 key: 0d 00 07 00 value: 02 00 00 00 01 02 03 04
213 Found 2 elements
214 # bpftool map getnext id 10 key 0 1 2 3
216 key:
218 00 01 02 03
219 next key:
220 0d 00 07 00
221 # mount -t bpf none /sys/fs/bpf/
223 # bpftool map pin id 10 /sys/fs/bpf/map
224 # bpftool map del pinned /sys/fs/bpf/map key 13 00 07 00
225 Note that map update can also be used in order to change the program
228 references hold by a program array map. This can be used, for example,
229 to change the programs used for tail-call jumps at runtime, without
230 having to reload the entry-point program. Below is an example for this
231 use case: we load a program defining a prog array map, and with a main
232 function that contains a tail call to other programs that can be used
233 either to "process" packets or to "debug" processing. Note that the
234 prog array map MUST be pinned into the BPF virtual file system for the
235 map update to work successfully, as kernel flushes prog array maps when
236 they have no more references from user space (and the update would be
237 lost as soon as bpftool exits).
238 # bpftool prog loadall tail_calls.o /sys/fs/bpf/foo type xdp
240 # bpftool prog --bpffs
241 545: xdp name main_func tag 674b4b5597193dc3 gpl
244 loaded_at 2018-12-12T15:02:58+0000 uid 0
245 xlated 240B jited 257B memlock 4096B map_ids 294
246 pinned /sys/fs/bpf/foo/xdp
247 546: xdp name bpf_func_process tag e369a529024751fc gpl
248 loaded_at 2018-12-12T15:02:58+0000 uid 0
249 xlated 200B jited 164B memlock 4096B
250 pinned /sys/fs/bpf/foo/process
251 547: xdp name bpf_func_debug tag 0b597868bc7f0976 gpl
252 loaded_at 2018-12-12T15:02:58+0000 uid 0
253 xlated 200B jited 164B memlock 4096B
254 pinned /sys/fs/bpf/foo/debug
255 # bpftool map
257 294: prog_array name jmp_table flags 0x0
259 key 4B value 4B max_entries 1 memlock 4096B
260 owner_prog_type xdp owner jited
261 # bpftool map pin id 294 /sys/fs/bpf/bar
263 # bpftool map dump pinned /sys/fs/bpf/bar
264 Found 0 elements
267 # bpftool map update pinned /sys/fs/bpf/bar key 0 0 0 0 value pinned /sys/fs/bpf/foo/debug
269 # bpftool map dump pinned /sys/fs/bpf/bar
270 key: 00 00 00 00 value: 22 02 00 00
273 Found 1 element
274
276 bpf(2), bpf-helpers(7), bpftool(8), bpftool-btf(8),
277 bpftool-cgroup(8), bpftool-feature(8), bpftool-gen(8),
278 bpftool-iter(8), bpftool-link(8), bpftool-net(8), bpftool-perf(8),
279 bpftool-prog(8), bpftool-struct_ops(8)
280 BPFTOOL-MAP(8)