perf ftrace: Add -b/--use-bpf option for latency subcommand
authorNamhyung Kim <namhyung@kernel.org>
Wed, 15 Dec 2021 18:51:53 +0000 (10:51 -0800)
committerArnaldo Carvalho de Melo <acme@redhat.com>
Thu, 16 Dec 2021 15:18:12 +0000 (12:18 -0300)
commit177f4eac7fb7fe5c70fef30dd6c4ef8f81cf7776
tree294005e620450ccb2ea2742ae2e0439adc2088f3
parent53be50282269b46c678ae5a9f54acf7416a10dbb
perf ftrace: Add -b/--use-bpf option for latency subcommand

The -b/--use-bpf option is to use BPF to get latency info of kernel
functions.  It'd have better performance impact and I observed that
latency of same function is smaller than before when using BPF.

Committer testing:

  # strace -e bpf perf ftrace latency -b -T __handle_mm_fault -a sleep 1
  bpf(BPF_PROG_LOAD, {prog_type=BPF_PROG_TYPE_SOCKET_FILTER, insn_cnt=2, insns=0x7fff51914e00, license="GPL", log_level=0, log_size=0, log_buf=NULL, kern_version=KERNEL_VERSION(0, 0, 0), prog_flags=0, prog_name="", prog_ifindex=0, expected_attach_type=BPF_CGROUP_INET_INGRESS, prog_btf_fd=0, func_info_rec_size=0, func_info=NULL, func_info_cnt=0, line_info_rec_size=0, line_info=NULL, line_info_cnt=0, attach_btf_id=0, attach_prog_fd=0}, 128) = 3
  bpf(BPF_BTF_LOAD, {btf="\237\353\1\0\30\0\0\0\0\0\0\0\20\0\0\0\20\0\0\0\5\0\0\0\1\0\0\0\0\0\0\1"..., btf_log_buf=NULL, btf_size=45, btf_log_size=0, btf_log_level=0}, 128) = 3
  bpf(BPF_BTF_LOAD, {btf="\237\353\1\0\30\0\0\0\0\0\0\0000\0\0\0000\0\0\0\t\0\0\0\1\0\0\0\0\0\0\1"..., btf_log_buf=NULL, btf_size=81, btf_log_size=0, btf_log_level=0}, 128) = 3
  bpf(BPF_BTF_LOAD, {btf="\237\353\1\0\30\0\0\0\0\0\0\08\0\0\08\0\0\0\t\0\0\0\0\0\0\0\0\0\0\1"..., btf_log_buf=NULL, btf_size=89, btf_log_size=0, btf_log_level=0}, 128) = 3
  bpf(BPF_BTF_LOAD, {btf="\237\353\1\0\30\0\0\0\0\0\0\0\f\0\0\0\f\0\0\0\7\0\0\0\1\0\0\0\0\0\0\20"..., btf_log_buf=NULL, btf_size=43, btf_log_size=0, btf_log_level=0}, 128) = 3
  bpf(BPF_BTF_LOAD, {btf="\237\353\1\0\30\0\0\0\0\0\0\0000\0\0\0000\0\0\0\t\0\0\0\1\0\0\0\0\0\0\1"..., btf_log_buf=NULL, btf_size=81, btf_log_size=0, btf_log_level=0}, 128) = 3
  bpf(BPF_BTF_LOAD, {btf="\237\353\1\0\30\0\0\0\0\0\0\0000\0\0\0000\0\0\0\5\0\0\0\0\0\0\0\0\0\0\1"..., btf_log_buf=NULL, btf_size=77, btf_log_size=0, btf_log_level=0}, 128) = -1 EINVAL (Invalid argument)
  bpf(BPF_BTF_LOAD, {btf="\237\353\1\0\30\0\0\0\0\0\0\0\350\2\0\0\350\2\0\0\353\2\0\0\0\0\0\0\0\0\0\2"..., btf_log_buf=NULL, btf_size=1515, btf_log_size=0, btf_log_level=0}, 128) = 3
  bpf(BPF_MAP_CREATE, {map_type=BPF_MAP_TYPE_ARRAY, key_size=4, value_size=32, max_entries=1, map_flags=0, inner_map_fd=0, map_name="", map_ifindex=0, btf_fd=0, btf_key_type_id=0, btf_value_type_id=0, btf_vmlinux_value_type_id=0}, 128) = 4
  bpf(BPF_PROG_LOAD, {prog_type=BPF_PROG_TYPE_SOCKET_FILTER, insn_cnt=5, insns=0x7fff51914c30, license="GPL", log_level=0, log_size=0, log_buf=NULL, kern_version=KERNEL_VERSION(0, 0, 0), prog_flags=0, prog_name="", prog_ifindex=0, expected_attach_type=BPF_CGROUP_INET_INGRESS, prog_btf_fd=0, func_info_rec_size=0, func_info=NULL, func_info_cnt=0, line_info_rec_size=0, line_info=NULL, line_info_cnt=0, attach_btf_id=0, attach_prog_fd=0}, 128) = 5
  bpf(BPF_MAP_CREATE, {map_type=BPF_MAP_TYPE_ARRAY, key_size=4, value_size=4, max_entries=1, map_flags=BPF_F_MMAPABLE, inner_map_fd=0, map_name="", map_ifindex=0, btf_fd=0, btf_key_type_id=0, btf_value_type_id=0, btf_vmlinux_value_type_id=0}, 128) = 4
  bpf(BPF_PROG_LOAD, {prog_type=BPF_PROG_TYPE_SOCKET_FILTER, insn_cnt=2, insns=0x7fff51914a80, license="GPL", log_level=0, log_size=0, log_buf=NULL, kern_version=KERNEL_VERSION(0, 0, 0), prog_flags=0, prog_name="test", prog_ifindex=0, expected_attach_type=BPF_CGROUP_INET_INGRESS, prog_btf_fd=0, func_info_rec_size=0, func_info=NULL, func_info_cnt=0, line_info_rec_size=0, line_info=NULL, line_info_cnt=0, attach_btf_id=0, attach_prog_fd=0}, 128) = 4
  bpf(BPF_MAP_CREATE, {map_type=BPF_MAP_TYPE_HASH, key_size=8, value_size=8, max_entries=10000, map_flags=0, inner_map_fd=0, map_name="functime", map_ifindex=0, btf_fd=3, btf_key_type_id=0, btf_value_type_id=0, btf_vmlinux_value_type_id=0}, 128) = 4
  bpf(BPF_MAP_CREATE, {map_type=BPF_MAP_TYPE_HASH, key_size=4, value_size=1, max_entries=1, map_flags=0, inner_map_fd=0, map_name="cpu_filter", map_ifindex=0, btf_fd=3, btf_key_type_id=0, btf_value_type_id=0, btf_vmlinux_value_type_id=0}, 128) = 5
  bpf(BPF_MAP_CREATE, {map_type=BPF_MAP_TYPE_HASH, key_size=4, value_size=1, max_entries=1, map_flags=0, inner_map_fd=0, map_name="task_filter", map_ifindex=0, btf_fd=3, btf_key_type_id=0, btf_value_type_id=0, btf_vmlinux_value_type_id=0}, 128) = 7
  bpf(BPF_MAP_CREATE, {map_type=BPF_MAP_TYPE_PERCPU_ARRAY, key_size=4, value_size=8, max_entries=22, map_flags=0, inner_map_fd=0, map_name="latency", map_ifindex=0, btf_fd=3, btf_key_type_id=0, btf_value_type_id=0, btf_vmlinux_value_type_id=0}, 128) = 8
  bpf(BPF_MAP_CREATE, {map_type=BPF_MAP_TYPE_ARRAY, key_size=4, value_size=4, max_entries=1, map_flags=BPF_F_MMAPABLE, inner_map_fd=0, map_name="func_lat.bss", map_ifindex=0, btf_fd=3, btf_key_type_id=0, btf_value_type_id=30, btf_vmlinux_value_type_id=0}, 128) = 9
  bpf(BPF_MAP_UPDATE_ELEM, {map_fd=9, key=0x7fff51914c40, value=0x7f6e99be2000, flags=BPF_ANY}, 128) = 0
  bpf(BPF_PROG_LOAD, {prog_type=BPF_PROG_TYPE_KPROBE, insn_cnt=18, insns=0x11e4160, license="", log_level=0, log_size=0, log_buf=NULL, kern_version=KERNEL_VERSION(5, 14, 16), prog_flags=0, prog_name="func_begin", prog_ifindex=0, expected_attach_type=BPF_CGROUP_INET_INGRESS, prog_btf_fd=3, func_info_rec_size=8, func_info=0x11dfc50, func_info_cnt=1, line_info_rec_size=16, line_info=0x11e04c0, line_info_cnt=9, attach_btf_id=0, attach_prog_fd=0}, 128) = 10
  bpf(BPF_PROG_LOAD, {prog_type=BPF_PROG_TYPE_KPROBE, insn_cnt=99, insns=0x11ded70, license="", log_level=0, log_size=0, log_buf=NULL, kern_version=KERNEL_VERSION(5, 14, 16), prog_flags=0, prog_name="func_end", prog_ifindex=0, expected_attach_type=BPF_CGROUP_INET_INGRESS, prog_btf_fd=3, func_info_rec_size=8, func_info=0x11dfc70, func_info_cnt=1, line_info_rec_size=16, line_info=0x11f6e10, line_info_cnt=20, attach_btf_id=0, attach_prog_fd=0}, 128) = 11
  bpf(BPF_PROG_LOAD, {prog_type=BPF_PROG_TYPE_TRACEPOINT, insn_cnt=2, insns=0x7fff51914a80, license="GPL", log_level=0, log_size=0, log_buf=NULL, kern_version=KERNEL_VERSION(0, 0, 0), prog_flags=0, prog_name="", prog_ifindex=0, expected_attach_type=BPF_CGROUP_INET_INGRESS, prog_btf_fd=0, func_info_rec_size=0, func_info=NULL, func_info_cnt=0, line_info_rec_size=0, line_info=NULL, line_info_cnt=0, attach_btf_id=0, attach_prog_fd=0}, 128) = 13
  bpf(BPF_LINK_CREATE, {link_create={prog_fd=13, target_fd=-1, attach_type=0x29 /* BPF_??? */, flags=0}}, 128) = -1 EINVAL (Invalid argument)
  --- SIGCHLD {si_signo=SIGCHLD, si_code=CLD_EXITED, si_pid=1699992, si_uid=0, si_status=0, si_utime=0, si_stime=0} ---
  bpf(BPF_MAP_LOOKUP_ELEM, {map_fd=8, key=0x7fff51914f84, value=0x11f6fa0, flags=BPF_ANY}, 128) = 0
  bpf(BPF_MAP_LOOKUP_ELEM, {map_fd=8, key=0x7fff51914f84, value=0x11f6fa0, flags=BPF_ANY}, 128) = 0
  bpf(BPF_MAP_LOOKUP_ELEM, {map_fd=8, key=0x7fff51914f84, value=0x11f6fa0, flags=BPF_ANY}, 128) = 0
  bpf(BPF_MAP_LOOKUP_ELEM, {map_fd=8, key=0x7fff51914f84, value=0x11f6fa0, flags=BPF_ANY}, 128) = 0
  bpf(BPF_MAP_LOOKUP_ELEM, {map_fd=8, key=0x7fff51914f84, value=0x11f6fa0, flags=BPF_ANY}, 128) = 0
  bpf(BPF_MAP_LOOKUP_ELEM, {map_fd=8, key=0x7fff51914f84, value=0x11f6fa0, flags=BPF_ANY}, 128) = 0
  bpf(BPF_MAP_LOOKUP_ELEM, {map_fd=8, key=0x7fff51914f84, value=0x11f6fa0, flags=BPF_ANY}, 128) = 0
  bpf(BPF_MAP_LOOKUP_ELEM, {map_fd=8, key=0x7fff51914f84, value=0x11f6fa0, flags=BPF_ANY}, 128) = 0
  bpf(BPF_MAP_LOOKUP_ELEM, {map_fd=8, key=0x7fff51914f84, value=0x11f6fa0, flags=BPF_ANY}, 128) = 0
  bpf(BPF_MAP_LOOKUP_ELEM, {map_fd=8, key=0x7fff51914f84, value=0x11f6fa0, flags=BPF_ANY}, 128) = 0
  bpf(BPF_MAP_LOOKUP_ELEM, {map_fd=8, key=0x7fff51914f84, value=0x11f6fa0, flags=BPF_ANY}, 128) = 0
  bpf(BPF_MAP_LOOKUP_ELEM, {map_fd=8, key=0x7fff51914f84, value=0x11f6fa0, flags=BPF_ANY}, 128) = 0
  bpf(BPF_MAP_LOOKUP_ELEM, {map_fd=8, key=0x7fff51914f84, value=0x11f6fa0, flags=BPF_ANY}, 128) = 0
  bpf(BPF_MAP_LOOKUP_ELEM, {map_fd=8, key=0x7fff51914f84, value=0x11f6fa0, flags=BPF_ANY}, 128) = 0
  bpf(BPF_MAP_LOOKUP_ELEM, {map_fd=8, key=0x7fff51914f84, value=0x11f6fa0, flags=BPF_ANY}, 128) = 0
  bpf(BPF_MAP_LOOKUP_ELEM, {map_fd=8, key=0x7fff51914f84, value=0x11f6fa0, flags=BPF_ANY}, 128) = 0
  bpf(BPF_MAP_LOOKUP_ELEM, {map_fd=8, key=0x7fff51914f84, value=0x11f6fa0, flags=BPF_ANY}, 128) = 0
  bpf(BPF_MAP_LOOKUP_ELEM, {map_fd=8, key=0x7fff51914f84, value=0x11f6fa0, flags=BPF_ANY}, 128) = 0
  bpf(BPF_MAP_LOOKUP_ELEM, {map_fd=8, key=0x7fff51914f84, value=0x11f6fa0, flags=BPF_ANY}, 128) = 0
  bpf(BPF_MAP_LOOKUP_ELEM, {map_fd=8, key=0x7fff51914f84, value=0x11f6fa0, flags=BPF_ANY}, 128) = 0
  bpf(BPF_MAP_LOOKUP_ELEM, {map_fd=8, key=0x7fff51914f84, value=0x11f6fa0, flags=BPF_ANY}, 128) = 0
  bpf(BPF_MAP_LOOKUP_ELEM, {map_fd=8, key=0x7fff51914f84, value=0x11f6fa0, flags=BPF_ANY}, 128) = 0
  #   DURATION     |      COUNT | GRAPH                                          |
       0 - 1    us |         52 | ###################                            |
       1 - 2    us |         36 | #############                                  |
       2 - 4    us |         24 | #########                                      |
       4 - 8    us |          7 | ##                                             |
       8 - 16   us |          1 |                                                |
      16 - 32   us |          0 |                                                |
      32 - 64   us |          0 |                                                |
      64 - 128  us |          0 |                                                |
     128 - 256  us |          0 |                                                |
     256 - 512  us |          0 |                                                |
     512 - 1024 us |          0 |                                                |
       1 - 2    ms |          0 |                                                |
       2 - 4    ms |          0 |                                                |
       4 - 8    ms |          0 |                                                |
       8 - 16   ms |          0 |                                                |
      16 - 32   ms |          0 |                                                |
      32 - 64   ms |          0 |                                                |
      64 - 128  ms |          0 |                                                |
     128 - 256  ms |          0 |                                                |
     256 - 512  ms |          0 |                                                |
     512 - 1024 ms |          0 |                                                |
       1 - ...   s |          0 |                                                |
  +++ exited with 0 +++
  #

Signed-off-by: Namhyung Kim <namhyung@kernel.org>
Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Athira Jajeev <atrajeev@linux.vnet.ibm.com>
Cc: Changbin Du <changbin.du@gmail.com>
Cc: Ian Rogers <irogers@google.com>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Jiri Olsa <jolsa@redhat.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Song Liu <songliubraving@fb.com>
Cc: Stephane Eranian <eranian@google.com>
Link: https://lore.kernel.org/r/20211215185154.360314-5-namhyung@kernel.org
[ Add missing util/cpumap.h include and removed unused 'fd' variable ]
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
tools/perf/Makefile.perf
tools/perf/builtin-ftrace.c
tools/perf/util/Build
tools/perf/util/bpf_ftrace.c [new file with mode: 0644]
tools/perf/util/bpf_skel/func_latency.bpf.c [new file with mode: 0644]
tools/perf/util/ftrace.h [new file with mode: 0644]