1================ 2bpftool-prog 3================ 4------------------------------------------------------------------------------- 5tool for inspection and simple manipulation of eBPF progs 6------------------------------------------------------------------------------- 7 8:Manual section: 8 9 10SYNOPSIS 11======== 12 13 **bpftool** [*OPTIONS*] **prog** *COMMAND* 14 15 *OPTIONS* := { { **-j** | **--json** } [{ **-p** | **--pretty** }] | { **-d** | **--debug** } | 16 { **-f** | **--bpffs** } | { **-m** | **--mapcompat** } | { **-n** | **--nomount** } | 17 { **-L** | **--use-loader** } } 18 19 *COMMANDS* := 20 { **show** | **list** | **dump xlated** | **dump jited** | **pin** | **load** 21 | **loadall** | **help** } 22 23PROG COMMANDS 24============= 25 26| **bpftool** **prog** { **show** | **list** } [*PROG*] 27| **bpftool** **prog dump xlated** *PROG* [{**file** *FILE* | **opcodes** | **visual** | **linum**}] 28| **bpftool** **prog dump jited** *PROG* [{**file** *FILE* | **opcodes** | **linum**}] 29| **bpftool** **prog pin** *PROG* *FILE* 30| **bpftool** **prog** { **load** | **loadall** } *OBJ* *PATH* [**type** *TYPE*] [**map** {**idx** *IDX* | **name** *NAME*} *MAP*] [**dev** *NAME*] [**pinmaps** *MAP_DIR*] 31| **bpftool** **prog attach** *PROG* *ATTACH_TYPE* [*MAP*] 32| **bpftool** **prog detach** *PROG* *ATTACH_TYPE* [*MAP*] 33| **bpftool** **prog tracelog** 34| **bpftool** **prog run** *PROG* **data_in** *FILE* [**data_out** *FILE* [**data_size_out** *L*]] [**ctx_in** *FILE* [**ctx_out** *FILE* [**ctx_size_out** *M*]]] [**repeat** *N*] 35| **bpftool** **prog profile** *PROG* [**duration** *DURATION*] *METRICs* 36| **bpftool** **prog help** 37| 38| *MAP* := { **id** *MAP_ID* | **pinned** *FILE* } 39| *PROG* := { **id** *PROG_ID* | **pinned** *FILE* | **tag** *PROG_TAG* | **name** *PROG_NAME* } 40| *TYPE* := { 41| **socket** | **kprobe** | **kretprobe** | **classifier** | **action** | 42| **tracepoint** | **raw_tracepoint** | **xdp** | **perf_event** | **cgroup/skb** | 43| **cgroup/sock** | **cgroup/dev** | **lwt_in** | **lwt_out** | **lwt_xmit** | 44| **lwt_seg6local** | **sockops** | **sk_skb** | **sk_msg** | **lirc_mode2** | 45| **cgroup/bind4** | **cgroup/bind6** | **cgroup/post_bind4** | **cgroup/post_bind6** | 46| **cgroup/connect4** | **cgroup/connect6** | **cgroup/getpeername4** | **cgroup/getpeername6** | 47| **cgroup/getsockname4** | **cgroup/getsockname6** | **cgroup/sendmsg4** | **cgroup/sendmsg6** | 48| **cgroup/recvmsg4** | **cgroup/recvmsg6** | **cgroup/sysctl** | 49| **cgroup/getsockopt** | **cgroup/setsockopt** | **cgroup/sock_release** | 50| **struct_ops** | **fentry** | **fexit** | **freplace** | **sk_lookup** 51| } 52| *ATTACH_TYPE* := { 53| **msg_verdict** | **skb_verdict** | **stream_verdict** | **stream_parser** | **flow_dissector** 54| } 55| *METRICs* := { 56| **cycles** | **instructions** | **l1d_loads** | **llc_misses** | 57| **itlb_misses** | **dtlb_misses** 58| } 59 60 61DESCRIPTION 62=========== 63 **bpftool prog { show | list }** [*PROG*] 64 Show information about loaded programs. If *PROG* is 65 specified show information only about given programs, 66 otherwise list all programs currently loaded on the system. 67 In case of **tag** or **name**, *PROG* may match several 68 programs which will all be shown. 69 70 Output will start with program ID followed by program type and 71 zero or more named attributes (depending on kernel version). 72 73 Since Linux 5.1 the kernel can collect statistics on BPF 74 programs (such as the total time spent running the program, 75 and the number of times it was run). If available, bpftool 76 shows such statistics. However, the kernel does not collect 77 them by defaults, as it slightly impacts performance on each 78 program run. Activation or deactivation of the feature is 79 performed via the **kernel.bpf_stats_enabled** sysctl knob. 80 81 Since Linux 5.8 bpftool is able to discover information about 82 processes that hold open file descriptors (FDs) against BPF 83 programs. On such kernels bpftool will automatically emit this 84 information as well. 85 86 **bpftool prog dump xlated** *PROG* [{ **file** *FILE* | **opcodes** | **visual** | **linum** }] 87 Dump eBPF instructions of the programs from the kernel. By 88 default, eBPF will be disassembled and printed to standard 89 output in human-readable format. In this case, **opcodes** 90 controls if raw opcodes should be printed as well. 91 92 In case of **tag** or **name**, *PROG* may match several 93 programs which will all be dumped. However, if **file** or 94 **visual** is specified, *PROG* must match a single program. 95 96 If **file** is specified, the binary image will instead be 97 written to *FILE*. 98 99 If **visual** is specified, control flow graph (CFG) will be 100 built instead, and eBPF instructions will be presented with 101 CFG in DOT format, on standard output. 102 103 If the programs have line_info available, the source line will 104 be displayed by default. If **linum** is specified, 105 the filename, line number and line column will also be 106 displayed on top of the source line. 107 108 **bpftool prog dump jited** *PROG* [{ **file** *FILE* | **opcodes** | **linum** }] 109 Dump jited image (host machine code) of the program. 110 111 If *FILE* is specified image will be written to a file, 112 otherwise it will be disassembled and printed to stdout. 113 *PROG* must match a single program when **file** is specified. 114 115 **opcodes** controls if raw opcodes will be printed. 116 117 If the prog has line_info available, the source line will 118 be displayed by default. If **linum** is specified, 119 the filename, line number and line column will also be 120 displayed on top of the source line. 121 122 **bpftool prog pin** *PROG* *FILE* 123 Pin program *PROG* as *FILE*. 124 125 Note: *FILE* must be located in *bpffs* mount. It must not 126 contain a dot character ('.'), which is reserved for future 127 extensions of *bpffs*. 128 129 **bpftool prog { load | loadall }** *OBJ* *PATH* [**type** *TYPE*] [**map** {**idx** *IDX* | **name** *NAME*} *MAP*] [**dev** *NAME*] [**pinmaps** *MAP_DIR*] 130 Load bpf program(s) from binary *OBJ* and pin as *PATH*. 131 **bpftool prog load** pins only the first program from the 132 *OBJ* as *PATH*. **bpftool prog loadall** pins all programs 133 from the *OBJ* under *PATH* directory. 134 **type** is optional, if not specified program type will be 135 inferred from section names. 136 By default bpftool will create new maps as declared in the ELF 137 object being loaded. **map** parameter allows for the reuse 138 of existing maps. It can be specified multiple times, each 139 time for a different map. *IDX* refers to index of the map 140 to be replaced in the ELF file counting from 0, while *NAME* 141 allows to replace a map by name. *MAP* specifies the map to 142 use, referring to it by **id** or through a **pinned** file. 143 If **dev** *NAME* is specified program will be loaded onto 144 given networking device (offload). 145 Optional **pinmaps** argument can be provided to pin all 146 maps under *MAP_DIR* directory. 147 148 Note: *PATH* must be located in *bpffs* mount. It must not 149 contain a dot character ('.'), which is reserved for future 150 extensions of *bpffs*. 151 152 **bpftool prog attach** *PROG* *ATTACH_TYPE* [*MAP*] 153 Attach bpf program *PROG* (with type specified by 154 *ATTACH_TYPE*). Most *ATTACH_TYPEs* require a *MAP* 155 parameter, with the exception of *flow_dissector* which is 156 attached to current networking name space. 157 158 **bpftool prog detach** *PROG* *ATTACH_TYPE* [*MAP*] 159 Detach bpf program *PROG* (with type specified by 160 *ATTACH_TYPE*). Most *ATTACH_TYPEs* require a *MAP* 161 parameter, with the exception of *flow_dissector* which is 162 detached from the current networking name space. 163 164 **bpftool prog tracelog** 165 Dump the trace pipe of the system to the console (stdout). 166 Hit <Ctrl+C> to stop printing. BPF programs can write to this 167 trace pipe at runtime with the **bpf_trace_printk**\ () helper. 168 This should be used only for debugging purposes. For 169 streaming data from BPF programs to user space, one can use 170 perf events (see also **bpftool-map**\ (8)). 171 172 **bpftool prog run** *PROG* **data_in** *FILE* [**data_out** *FILE* [**data_size_out** *L*]] [**ctx_in** *FILE* [**ctx_out** *FILE* [**ctx_size_out** *M*]]] [**repeat** *N*] 173 Run BPF program *PROG* in the kernel testing infrastructure 174 for BPF, meaning that the program works on the data and 175 context provided by the user, and not on actual packets or 176 monitored functions etc. Return value and duration for the 177 test run are printed out to the console. 178 179 Input data is read from the *FILE* passed with **data_in**. 180 If this *FILE* is "**-**", input data is read from standard 181 input. Input context, if any, is read from *FILE* passed with 182 **ctx_in**. Again, "**-**" can be used to read from standard 183 input, but only if standard input is not already in use for 184 input data. If a *FILE* is passed with **data_out**, output 185 data is written to that file. Similarly, output context is 186 written to the *FILE* passed with **ctx_out**. For both 187 output flows, "**-**" can be used to print to the standard 188 output (as plain text, or JSON if relevant option was 189 passed). If output keywords are omitted, output data and 190 context are discarded. Keywords **data_size_out** and 191 **ctx_size_out** are used to pass the size (in bytes) for the 192 output buffers to the kernel, although the default of 32 kB 193 should be more than enough for most cases. 194 195 Keyword **repeat** is used to indicate the number of 196 consecutive runs to perform. Note that output data and 197 context printed to files correspond to the last of those 198 runs. The duration printed out at the end of the runs is an 199 average over all runs performed by the command. 200 201 Not all program types support test run. Among those which do, 202 not all of them can take the **ctx_in**/**ctx_out** 203 arguments. bpftool does not perform checks on program types. 204 205 **bpftool prog profile** *PROG* [**duration** *DURATION*] *METRICs* 206 Profile *METRICs* for bpf program *PROG* for *DURATION* 207 seconds or until user hits <Ctrl+C>. *DURATION* is optional. 208 If *DURATION* is not specified, the profiling will run up to 209 **UINT_MAX** seconds. 210 211 **bpftool prog help** 212 Print short help message. 213 214OPTIONS 215======= 216 .. include:: common_options.rst 217 218 -f, --bpffs 219 When showing BPF programs, show file names of pinned 220 programs. 221 222 -m, --mapcompat 223 Allow loading maps with unknown map definitions. 224 225 -n, --nomount 226 Do not automatically attempt to mount any virtual file system 227 (such as tracefs or BPF virtual file system) when necessary. 228 229 -L, --use-loader 230 Load program as a "loader" program. This is useful to debug 231 the generation of such programs. When this option is in 232 use, bpftool attempts to load the programs from the object 233 file into the kernel, but does not pin them (therefore, the 234 *PATH* must not be provided). 235 236 When combined with the **-d**\ \|\ **--debug** option, 237 additional debug messages are generated, and the execution 238 of the loader program will use the **bpf_trace_printk**\ () 239 helper to log each step of loading BTF, creating the maps, 240 and loading the programs (see **bpftool prog tracelog** as 241 a way to dump those messages). 242 243EXAMPLES 244======== 245**# bpftool prog show** 246 247:: 248 249 10: xdp name some_prog tag 005a3d2123620c8b gpl run_time_ns 81632 run_cnt 10 250 loaded_at 2017-09-29T20:11:00+0000 uid 0 251 xlated 528B jited 370B memlock 4096B map_ids 10 252 pids systemd(1) 253 254**# bpftool --json --pretty prog show** 255 256:: 257 258 [{ 259 "id": 10, 260 "type": "xdp", 261 "tag": "005a3d2123620c8b", 262 "gpl_compatible": true, 263 "run_time_ns": 81632, 264 "run_cnt": 10, 265 "loaded_at": 1506715860, 266 "uid": 0, 267 "bytes_xlated": 528, 268 "jited": true, 269 "bytes_jited": 370, 270 "bytes_memlock": 4096, 271 "map_ids": [10 272 ], 273 "pids": [{ 274 "pid": 1, 275 "comm": "systemd" 276 } 277 ] 278 } 279 ] 280 281| 282| **# bpftool prog dump xlated id 10 file /tmp/t** 283| **$ ls -l /tmp/t** 284 285:: 286 287 -rw------- 1 root root 560 Jul 22 01:42 /tmp/t 288 289**# bpftool prog dump jited tag 005a3d2123620c8b** 290 291:: 292 293 0: push %rbp 294 1: mov %rsp,%rbp 295 2: sub $0x228,%rsp 296 3: sub $0x28,%rbp 297 4: mov %rbx,0x0(%rbp) 298 299| 300| **# mount -t bpf none /sys/fs/bpf/** 301| **# bpftool prog pin id 10 /sys/fs/bpf/prog** 302| **# bpftool prog load ./my_prog.o /sys/fs/bpf/prog2** 303| **# ls -l /sys/fs/bpf/** 304 305:: 306 307 -rw------- 1 root root 0 Jul 22 01:43 prog 308 -rw------- 1 root root 0 Jul 22 01:44 prog2 309 310**# bpftool prog dump jited pinned /sys/fs/bpf/prog opcodes** 311 312:: 313 314 0: push %rbp 315 55 316 1: mov %rsp,%rbp 317 48 89 e5 318 4: sub $0x228,%rsp 319 48 81 ec 28 02 00 00 320 b: sub $0x28,%rbp 321 48 83 ed 28 322 f: mov %rbx,0x0(%rbp) 323 48 89 5d 00 324 325| 326| **# bpftool prog load xdp1_kern.o /sys/fs/bpf/xdp1 type xdp map name rxcnt id 7** 327| **# bpftool prog show pinned /sys/fs/bpf/xdp1** 328 329:: 330 331 9: xdp name xdp_prog1 tag 539ec6ce11b52f98 gpl 332 loaded_at 2018-06-25T16:17:31-0700 uid 0 333 xlated 488B jited 336B memlock 4096B map_ids 7 334 335**# rm /sys/fs/bpf/xdp1** 336 337| 338| **# bpftool prog profile id 337 duration 10 cycles instructions llc_misses** 339 340:: 341 342 51397 run_cnt 343 40176203 cycles (83.05%) 344 42518139 instructions # 1.06 insns per cycle (83.39%) 345 123 llc_misses # 2.89 LLC misses per million insns (83.15%) 346 347| 348| Output below is for the trace logs. 349| Run in separate terminals: 350| **# bpftool prog tracelog** 351| **# bpftool prog load -L -d file.o** 352 353:: 354 355 bpftool-620059 [004] d... 2634685.517903: bpf_trace_printk: btf_load size 665 r=5 356 bpftool-620059 [004] d... 2634685.517912: bpf_trace_printk: map_create sample_map idx 0 type 2 value_size 4 value_btf_id 0 r=6 357 bpftool-620059 [004] d... 2634685.517997: bpf_trace_printk: prog_load sample insn_cnt 13 r=7 358 bpftool-620059 [004] d... 2634685.517999: bpf_trace_printk: close(5) = 0 359