| Message ID | 20220222170600.611515-1-jolsa@kernel.org (mailing list archive) |
|---|---|
| Headers | show |
| Series | bpf: Add kprobe multi link | expand |
On Tue, Feb 22, 2022 at 9:06 AM Jiri Olsa <jolsa@kernel.org> wrote: > > hi, > this patchset adds new link type BPF_TRACE_KPROBE_MULTI that attaches > kprobe program through fprobe API [1] instroduced by Masami. > > The fprobe API allows to attach probe on multiple functions at once very > fast, because it works on top of ftrace. On the other hand this limits > the probe point to the function entry or return. > > > With bpftrace support I see following attach speed: > > # perf stat --null -r 5 ./src/bpftrace -e 'kprobe:x* { } i:ms:1 { exit(); } ' > Attaching 2 probes... > Attaching 3342 functions > ... > > 1.4960 +- 0.0285 seconds time elapsed ( +- 1.91% ) > > > v2 changes: > - based on latest fprobe changes [1] > - renaming the uapi interface to kprobe multi > - adding support for sort_r to pass user pointer for swap functions > and using that in cookie support to keep just single functions array > - moving new link to kernel/trace/bpf_trace.c file > - using single fprobe callback function for entry and exit > - using kvzalloc, libbpf_ensure_mem functions > - adding new k[ret]probe.multi sections instead of using current kprobe > - used glob_match from test_progs.c, added '?' matching > - move bpf_get_func_ip verifier inline change to seprate change > - couple of other minor fixes > > I think it's shaping up pretty well. Great work, Jiri! Can't wait to adopt this in retsnoop. See below about dependency on Masami's patches. > Also available at: > https://git.kernel.org/pub/scm/linux/kernel/git/jolsa/perf.git > bpf/kprobe_multi > > thanks, > jirka > > > [1] https://lore.kernel.org/bpf/164458044634.586276.3261555265565111183.stgit@devnote2/ Masami, Jiri, Steven, what would be the logistics here? What's the plan for getting this upstream? Any idea about timelines? I really hope it won't take as long as it took for kretprobe stack trace capturing fixes last year to land. Can we take Masami's changes through bpf-next tree? If yes, Steven, can you please review and give your acks? Thanks for understanding! > --- > Jiri Olsa (10): > lib/sort: Add priv pointer to swap function > bpf: Add multi kprobe link > bpf: Add bpf_get_func_ip kprobe helper for multi kprobe link > bpf: Add support to inline bpf_get_func_ip helper on x86 > bpf: Add cookie support to programs attached with kprobe multi link > libbpf: Add libbpf_kallsyms_parse function > libbpf: Add bpf_link_create support for multi kprobes > libbpf: Add bpf_program__attach_kprobe_opts support for multi kprobes > selftest/bpf: Add kprobe_multi attach test > selftest/bpf: Add kprobe_multi test for bpf_cookie values > > include/linux/bpf_types.h | 1 + > include/linux/sort.h | 4 +- > include/linux/trace_events.h | 6 ++ > include/linux/types.h | 1 + > include/uapi/linux/bpf.h | 14 ++++ > kernel/bpf/syscall.c | 26 ++++++-- > kernel/bpf/verifier.c | 21 +++++- > kernel/trace/bpf_trace.c | 331 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++- > lib/sort.c | 42 +++++++++--- > tools/include/uapi/linux/bpf.h | 14 ++++ > tools/lib/bpf/bpf.c | 7 ++ > tools/lib/bpf/bpf.h | 9 ++- > tools/lib/bpf/libbpf.c | 192 +++++++++++++++++++++++++++++++++++++++++++++-------- > tools/lib/bpf/libbpf_internal.h | 5 ++ > tools/testing/selftests/bpf/prog_tests/bpf_cookie.c | 72 ++++++++++++++++++++ > tools/testing/selftests/bpf/prog_tests/kprobe_multi_test.c | 115 ++++++++++++++++++++++++++++++++ > tools/testing/selftests/bpf/progs/kprobe_multi.c | 58 ++++++++++++++++ > tools/testing/selftests/bpf/progs/kprobe_multi_bpf_cookie.c | 62 +++++++++++++++++ > 18 files changed, 930 insertions(+), 50 deletions(-) > create mode 100644 tools/testing/selftests/bpf/prog_tests/kprobe_multi_test.c > create mode 100644 tools/testing/selftests/bpf/progs/kprobe_multi.c > create mode 100644 tools/testing/selftests/bpf/progs/kprobe_multi_bpf_cookie.c
On Fri, 4 Mar 2022 15:10:55 -0800 Andrii Nakryiko <andrii.nakryiko@gmail.com> wrote: > Masami, Jiri, Steven, what would be the logistics here? What's the > plan for getting this upstream? Any idea about timelines? I really > hope it won't take as long as it took for kretprobe stack trace > capturing fixes last year to land. Can we take Masami's changes > through bpf-next tree? If yes, Steven, can you please review and give > your acks? Thanks for understanding! Yeah, I'll start looking at it this week. I just started a new job and that's been taking up a lot of my time and limiting what I can look at upstream. -- Steve
On Sat, 5 Mar 2022 20:09:39 -0500 Steven Rostedt <rostedt@goodmis.org> wrote: > On Fri, 4 Mar 2022 15:10:55 -0800 > Andrii Nakryiko <andrii.nakryiko@gmail.com> wrote: > > > Masami, Jiri, Steven, what would be the logistics here? What's the > > plan for getting this upstream? Any idea about timelines? I really > > hope it won't take as long as it took for kretprobe stack trace > > capturing fixes last year to land. Can we take Masami's changes > > through bpf-next tree? If yes, Steven, can you please review and give > > your acks? Thanks for understanding! > > Yeah, I'll start looking at it this week. I just started a new job and > that's been taking up a lot of my time and limiting what I can look at > upstream. Let me update my series, I found some issues in the selftest. I'll send v9 soon. Thank you!
On Sat, Mar 5, 2022 at 5:33 PM Masami Hiramatsu <masami.hiramatsu@gmail.com> wrote: > > On Sat, 5 Mar 2022 20:09:39 -0500 > Steven Rostedt <rostedt@goodmis.org> wrote: > > > On Fri, 4 Mar 2022 15:10:55 -0800 > > Andrii Nakryiko <andrii.nakryiko@gmail.com> wrote: > > > > > Masami, Jiri, Steven, what would be the logistics here? What's the > > > plan for getting this upstream? Any idea about timelines? I really > > > hope it won't take as long as it took for kretprobe stack trace > > > capturing fixes last year to land. Can we take Masami's changes > > > through bpf-next tree? If yes, Steven, can you please review and give > > > your acks? Thanks for understanding! > > > > Yeah, I'll start looking at it this week. I just started a new job and Thanks, Steven. Greatly appreciated! > > that's been taking up a lot of my time and limiting what I can look at > > upstream. > > Let me update my series, I found some issues in the selftest. > I'll send v9 soon. I haven't checked, but if you haven't based your patches off of bpf-next tree, please do so in the next revision, so that we can land your patch set and Jiri's patch set on top without any issues. Thanks! > > Thank you! > > -- > Masami Hiramatsu
hi, this patchset adds new link type BPF_TRACE_KPROBE_MULTI that attaches kprobe program through fprobe API [1] instroduced by Masami. The fprobe API allows to attach probe on multiple functions at once very fast, because it works on top of ftrace. On the other hand this limits the probe point to the function entry or return. With bpftrace support I see following attach speed: # perf stat --null -r 5 ./src/bpftrace -e 'kprobe:x* { } i:ms:1 { exit(); } ' Attaching 2 probes... Attaching 3342 functions ... 1.4960 +- 0.0285 seconds time elapsed ( +- 1.91% ) v2 changes: - based on latest fprobe changes [1] - renaming the uapi interface to kprobe multi - adding support for sort_r to pass user pointer for swap functions and using that in cookie support to keep just single functions array - moving new link to kernel/trace/bpf_trace.c file - using single fprobe callback function for entry and exit - using kvzalloc, libbpf_ensure_mem functions - adding new k[ret]probe.multi sections instead of using current kprobe - used glob_match from test_progs.c, added '?' matching - move bpf_get_func_ip verifier inline change to seprate change - couple of other minor fixes Also available at: https://git.kernel.org/pub/scm/linux/kernel/git/jolsa/perf.git bpf/kprobe_multi thanks, jirka [1] https://lore.kernel.org/bpf/164458044634.586276.3261555265565111183.stgit@devnote2/ --- Jiri Olsa (10): lib/sort: Add priv pointer to swap function bpf: Add multi kprobe link bpf: Add bpf_get_func_ip kprobe helper for multi kprobe link bpf: Add support to inline bpf_get_func_ip helper on x86 bpf: Add cookie support to programs attached with kprobe multi link libbpf: Add libbpf_kallsyms_parse function libbpf: Add bpf_link_create support for multi kprobes libbpf: Add bpf_program__attach_kprobe_opts support for multi kprobes selftest/bpf: Add kprobe_multi attach test selftest/bpf: Add kprobe_multi test for bpf_cookie values include/linux/bpf_types.h | 1 + include/linux/sort.h | 4 +- include/linux/trace_events.h | 6 ++ include/linux/types.h | 1 + include/uapi/linux/bpf.h | 14 ++++ kernel/bpf/syscall.c | 26 ++++++-- kernel/bpf/verifier.c | 21 +++++- kernel/trace/bpf_trace.c | 331 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++- lib/sort.c | 42 +++++++++--- tools/include/uapi/linux/bpf.h | 14 ++++ tools/lib/bpf/bpf.c | 7 ++ tools/lib/bpf/bpf.h | 9 ++- tools/lib/bpf/libbpf.c | 192 +++++++++++++++++++++++++++++++++++++++++++++-------- tools/lib/bpf/libbpf_internal.h | 5 ++ tools/testing/selftests/bpf/prog_tests/bpf_cookie.c | 72 ++++++++++++++++++++ tools/testing/selftests/bpf/prog_tests/kprobe_multi_test.c | 115 ++++++++++++++++++++++++++++++++ tools/testing/selftests/bpf/progs/kprobe_multi.c | 58 ++++++++++++++++ tools/testing/selftests/bpf/progs/kprobe_multi_bpf_cookie.c | 62 +++++++++++++++++ 18 files changed, 930 insertions(+), 50 deletions(-) create mode 100644 tools/testing/selftests/bpf/prog_tests/kprobe_multi_test.c create mode 100644 tools/testing/selftests/bpf/progs/kprobe_multi.c create mode 100644 tools/testing/selftests/bpf/progs/kprobe_multi_bpf_cookie.c