| Message ID | 20220823132236.65122-1-donald.hunter@gmail.com (mailing list archive) |
|---|---|
| State | Superseded |
| Delegated to: | BPF |
| Headers | show |
| Series | [bpf-next] Add table of BPF program types to docs | expand |
On Tue, Aug 23, 2022 at 9:56 AM Donald Hunter <donald.hunter@gmail.com> wrote: > > Extend the BPF program types documentation with a table of > program types, attach points and ELF section names. > > The program_types.csv file is generated from tools/lib/bpf/libbpf.c > and a script is included for regenerating the .csv file. > > I have not integrated the script into the doc build but if that > is desirable then please suggest the preferred way to do so. > > Signed-off-by: Donald Hunter <donald.hunter@gmail.com> > --- It does seem cleaner to generate this .csv during docs build, instead of having to manually regenerate it all the time? Should we also put it under Documentation/bpf/libbpf/ as it's libbpf-specific? Having it under libbpf subdir would also make it simpler to expose it in libbpf docs at libbpf.readthedocs.io/ We can probably also establish some special comment format next to SEC_DEF() to specify the format of those "extras", I think it would be useful for users. WDYT? CC'ing Grant as well, who worked on building libbpf docs. > Documentation/bpf/program_types.csv | 82 +++++++++++++++++++++++++++++ > Documentation/bpf/programs.rst | 15 ++++++ > scripts/gen-bpf-progtypes.sh | 21 ++++++++ > 3 files changed, 118 insertions(+) > create mode 100644 Documentation/bpf/program_types.csv > create mode 100755 scripts/gen-bpf-progtypes.sh > > diff --git a/Documentation/bpf/program_types.csv b/Documentation/bpf/program_types.csv > new file mode 100644 > index 000000000000..adec046b0bde > --- /dev/null > +++ b/Documentation/bpf/program_types.csv > @@ -0,0 +1,82 @@ > +Program Type,Attach Type,ELF Section Name,Sleepable > +``BPF_PROG_TYPE_SOCKET_FILTER``,,``socket``, > +``BPF_PROG_TYPE_SK_REUSEPORT``,``BPF_SK_REUSEPORT_SELECT_OR_MIGRATE``,``sk_reuseport/migrate``, > +``BPF_PROG_TYPE_SK_REUSEPORT``,``BPF_SK_REUSEPORT_SELECT``,``sk_reuseport``, > +``BPF_PROG_TYPE_KPROBE``,,``kprobe+``, > +``BPF_PROG_TYPE_KPROBE``,,``uprobe+``, > +``BPF_PROG_TYPE_KPROBE``,,``uprobe.s+``,Yes > +``BPF_PROG_TYPE_KPROBE``,,``kretprobe+``, > +``BPF_PROG_TYPE_KPROBE``,,``uretprobe+``, > +``BPF_PROG_TYPE_KPROBE``,,``uretprobe.s+``,Yes > +``BPF_PROG_TYPE_KPROBE``,``BPF_TRACE_KPROBE_MULTI``,``kprobe.multi+``, > +``BPF_PROG_TYPE_KPROBE``,``BPF_TRACE_KPROBE_MULTI``,``kretprobe.multi+``, > +``BPF_PROG_TYPE_KPROBE``,,``ksyscall+``, > +``BPF_PROG_TYPE_KPROBE``,,``kretsyscall+``, > +``BPF_PROG_TYPE_KPROBE``,,``usdt+``, > +``BPF_PROG_TYPE_SCHED_CLS``,,``tc``, > +``BPF_PROG_TYPE_SCHED_CLS``,,``classifier``, > +``BPF_PROG_TYPE_SCHED_ACT``,,``action``, > +``BPF_PROG_TYPE_TRACEPOINT``,,``tracepoint+``, > +``BPF_PROG_TYPE_TRACEPOINT``,,``tp+``, > +``BPF_PROG_TYPE_RAW_TRACEPOINT``,,``raw_tracepoint+``, > +``BPF_PROG_TYPE_RAW_TRACEPOINT``,,``raw_tp+``, > +``BPF_PROG_TYPE_RAW_TRACEPOINT_WRITABLE``,,``raw_tracepoint.w+``, > +``BPF_PROG_TYPE_RAW_TRACEPOINT_WRITABLE``,,``raw_tp.w+``, > +``BPF_PROG_TYPE_TRACING``,``BPF_TRACE_RAW_TP``,``tp_btf+``, > +``BPF_PROG_TYPE_TRACING``,``BPF_TRACE_FENTRY``,``fentry+``, > +``BPF_PROG_TYPE_TRACING``,``BPF_MODIFY_RETURN``,``fmod_ret+``, > +``BPF_PROG_TYPE_TRACING``,``BPF_TRACE_FEXIT``,``fexit+``, > +``BPF_PROG_TYPE_TRACING``,``BPF_TRACE_FENTRY``,``fentry.s+``,Yes > +``BPF_PROG_TYPE_TRACING``,``BPF_MODIFY_RETURN``,``fmod_ret.s+``,Yes > +``BPF_PROG_TYPE_TRACING``,``BPF_TRACE_FEXIT``,``fexit.s+``,Yes > +``BPF_PROG_TYPE_EXT``,,``freplace+``, > +``BPF_PROG_TYPE_LSM``,``BPF_LSM_MAC``,``lsm+``, > +``BPF_PROG_TYPE_LSM``,``BPF_LSM_MAC``,``lsm.s+``,Yes > +``BPF_PROG_TYPE_LSM``,``BPF_LSM_CGROUP``,``lsm_cgroup+``, > +``BPF_PROG_TYPE_TRACING``,``BPF_TRACE_ITER``,``iter+``, > +``BPF_PROG_TYPE_TRACING``,``BPF_TRACE_ITER``,``iter.s+``,Yes > +``BPF_PROG_TYPE_SYSCALL``,,``syscall``,Yes > +``BPF_PROG_TYPE_XDP``,``BPF_XDP_DEVMAP``,``xdp.frags/devmap``, > +``BPF_PROG_TYPE_XDP``,``BPF_XDP_DEVMAP``,``xdp/devmap``, > +``BPF_PROG_TYPE_XDP``,``BPF_XDP_CPUMAP``,``xdp.frags/cpumap``, > +``BPF_PROG_TYPE_XDP``,``BPF_XDP_CPUMAP``,``xdp/cpumap``, > +``BPF_PROG_TYPE_XDP``,``BPF_XDP``,``xdp.frags``, > +``BPF_PROG_TYPE_XDP``,``BPF_XDP``,``xdp``, > +``BPF_PROG_TYPE_PERF_EVENT``,,``perf_event``, > +``BPF_PROG_TYPE_LWT_IN``,,``lwt_in``, > +``BPF_PROG_TYPE_LWT_OUT``,,``lwt_out``, > +``BPF_PROG_TYPE_LWT_XMIT``,,``lwt_xmit``, > +``BPF_PROG_TYPE_LWT_SEG6LOCAL``,,``lwt_seg6local``, > +``BPF_PROG_TYPE_SOCK_OPS``,``BPF_CGROUP_SOCK_OPS``,``sockops``, > +``BPF_PROG_TYPE_SK_SKB``,``BPF_SK_SKB_STREAM_PARSER``,``sk_skb/stream_parser``, > +``BPF_PROG_TYPE_SK_SKB``,``BPF_SK_SKB_STREAM_VERDICT``,``sk_skb/stream_verdict``, > +``BPF_PROG_TYPE_SK_SKB``,,``sk_skb``, > +``BPF_PROG_TYPE_SK_MSG``,``BPF_SK_MSG_VERDICT``,``sk_msg``, > +``BPF_PROG_TYPE_LIRC_MODE2``,``BPF_LIRC_MODE2``,``lirc_mode2``, > +``BPF_PROG_TYPE_FLOW_DISSECTOR``,``BPF_FLOW_DISSECTOR``,``flow_dissector``, > +``BPF_PROG_TYPE_CGROUP_SKB``,``BPF_CGROUP_INET_INGRESS``,``cgroup_skb/ingress``, > +``BPF_PROG_TYPE_CGROUP_SKB``,``BPF_CGROUP_INET_EGRESS``,``cgroup_skb/egress``, > +``BPF_PROG_TYPE_CGROUP_SKB``,,``cgroup/skb``, > +``BPF_PROG_TYPE_CGROUP_SOCK``,``BPF_CGROUP_INET_SOCK_CREATE``,``cgroup/sock_create``, > +``BPF_PROG_TYPE_CGROUP_SOCK``,``BPF_CGROUP_INET_SOCK_RELEASE``,``cgroup/sock_release``, > +``BPF_PROG_TYPE_CGROUP_SOCK``,``BPF_CGROUP_INET_SOCK_CREATE``,``cgroup/sock``, > +``BPF_PROG_TYPE_CGROUP_SOCK``,``BPF_CGROUP_INET4_POST_BIND``,``cgroup/post_bind4``, > +``BPF_PROG_TYPE_CGROUP_SOCK``,``BPF_CGROUP_INET6_POST_BIND``,``cgroup/post_bind6``, > +``BPF_PROG_TYPE_CGROUP_SOCK_ADDR``,``BPF_CGROUP_INET4_BIND``,``cgroup/bind4``, > +``BPF_PROG_TYPE_CGROUP_SOCK_ADDR``,``BPF_CGROUP_INET6_BIND``,``cgroup/bind6``, > +``BPF_PROG_TYPE_CGROUP_SOCK_ADDR``,``BPF_CGROUP_INET4_CONNECT``,``cgroup/connect4``, > +``BPF_PROG_TYPE_CGROUP_SOCK_ADDR``,``BPF_CGROUP_INET6_CONNECT``,``cgroup/connect6``, > +``BPF_PROG_TYPE_CGROUP_SOCK_ADDR``,``BPF_CGROUP_UDP4_SENDMSG``,``cgroup/sendmsg4``, > +``BPF_PROG_TYPE_CGROUP_SOCK_ADDR``,``BPF_CGROUP_UDP6_SENDMSG``,``cgroup/sendmsg6``, > +``BPF_PROG_TYPE_CGROUP_SOCK_ADDR``,``BPF_CGROUP_UDP4_RECVMSG``,``cgroup/recvmsg4``, > +``BPF_PROG_TYPE_CGROUP_SOCK_ADDR``,``BPF_CGROUP_UDP6_RECVMSG``,``cgroup/recvmsg6``, > +``BPF_PROG_TYPE_CGROUP_SOCK_ADDR``,``BPF_CGROUP_INET4_GETPEERNAME``,``cgroup/getpeername4``, > +``BPF_PROG_TYPE_CGROUP_SOCK_ADDR``,``BPF_CGROUP_INET6_GETPEERNAME``,``cgroup/getpeername6``, > +``BPF_PROG_TYPE_CGROUP_SOCK_ADDR``,``BPF_CGROUP_INET4_GETSOCKNAME``,``cgroup/getsockname4``, > +``BPF_PROG_TYPE_CGROUP_SOCK_ADDR``,``BPF_CGROUP_INET6_GETSOCKNAME``,``cgroup/getsockname6``, > +``BPF_PROG_TYPE_CGROUP_SYSCTL``,``BPF_CGROUP_SYSCTL``,``cgroup/sysctl``, > +``BPF_PROG_TYPE_CGROUP_SOCKOPT``,``BPF_CGROUP_GETSOCKOPT``,``cgroup/getsockopt``, > +``BPF_PROG_TYPE_CGROUP_SOCKOPT``,``BPF_CGROUP_SETSOCKOPT``,``cgroup/setsockopt``, > +``BPF_PROG_TYPE_CGROUP_DEVICE``,``BPF_CGROUP_DEVICE``,``cgroup/dev``, > +``BPF_PROG_TYPE_STRUCT_OPS``,,``struct_ops+``, > +``BPF_PROG_TYPE_SK_LOOKUP``,``BPF_SK_LOOKUP``,``sk_lookup``, > diff --git a/Documentation/bpf/programs.rst b/Documentation/bpf/programs.rst > index 620eb667ac7a..71448fe0b955 100644 > --- a/Documentation/bpf/programs.rst > +++ b/Documentation/bpf/programs.rst > @@ -7,3 +7,18 @@ Program Types > :glob: > > prog_* > + > +Program Types and libbpf > +======================== > + > +The table below lists the program types, their attach types where relevant and the ELF section > +names supported by libbpf for them. The ELF section names follow these rules: > + > +- ``type`` is an exact match, e.g. ``SEC("socket")`` > +- ``type+`` means it can be either exact ``SEC("type")`` or well-formed ``SEC("type/extras")`` > + with a ‘``/``’ separator, e.g. ``SEC("tracepoint/syscalls/sys_enter_open")`` > + > +.. csv-table:: Program Types and Their ELF Section Names > + :file: program_types.csv > + :widths: 40 30 20 10 > + :header-rows: 1 > diff --git a/scripts/gen-bpf-progtypes.sh b/scripts/gen-bpf-progtypes.sh > new file mode 100755 > index 000000000000..fb9650bd5c1b > --- /dev/null > +++ b/scripts/gen-bpf-progtypes.sh > @@ -0,0 +1,21 @@ > +#!/bin/bash > +# SPDX-License-Identifier: GPL-2.0-only > +# > +# Copyright (C) Red Hat. > +# > +# Generate a .csv table of BPF program types > + > +if ! [ -d "tools/lib/bpf" -a -d "Documentation" ]; then > + echo "Run from top level of kernel tree" > + exit 1 > +fi > + > +awk -F'[",[:space:]]+' \ > + 'BEGIN { print "Program Type,Attach Type,ELF Section Name,Sleepable" } > + /SEC_DEF\(\"/ && !/SEC_DEPRECATED/ { > + type = "``BPF_PROG_TYPE_" $4 "``" > + attach = index($5, "0") ? "" : "``" $5 "``"; > + section = "``" $3 "``" > + sleepable = index($0, "SEC_SLEEPABLE") ? "Yes" : ""; > + print type "," attach "," section "," sleepable }' \ > +tools/lib/bpf/libbpf.c > Documentation/bpf/program_types.csv > -- > 2.35.1 >
Andrii Nakryiko <andrii.nakryiko@gmail.com> writes: > On Tue, Aug 23, 2022 at 9:56 AM Donald Hunter <donald.hunter@gmail.com> wrote: >> >> Extend the BPF program types documentation with a table of >> program types, attach points and ELF section names. >> >> The program_types.csv file is generated from tools/lib/bpf/libbpf.c >> and a script is included for regenerating the .csv file. >> >> I have not integrated the script into the doc build but if that >> is desirable then please suggest the preferred way to do so. >> >> Signed-off-by: Donald Hunter <donald.hunter@gmail.com> >> --- > > It does seem cleaner to generate this .csv during docs build, instead > of having to manually regenerate it all the time? Should we also put > it under Documentation/bpf/libbpf/ as it's libbpf-specific? Having it > under libbpf subdir would also make it simpler to expose it in libbpf > docs at libbpf.readthedocs.io/ Agreed about generating the .csv as part of the doc build. I will look at adding it to the docs Makefile. I'm happy to put it in Documentation/bpf/libbpf and link to it from Documentation/bpf/programs.rst. > We can probably also establish some special comment format next to > SEC_DEF() to specify the format of those "extras", I think it would be > useful for users. WDYT? Yes this would be a useful addition. Are the extras always for auto-attach? If so, then I can add that to the rules. I'd prefer to modify the existing ELF section name column to replace '+' with extras since the table is already wide. > CC'ing Grant as well, who worked on building libbpf docs.
On Wed, Aug 24, 2022 at 3:25 AM Donald Hunter <donald.hunter@gmail.com> wrote: > > Andrii Nakryiko <andrii.nakryiko@gmail.com> writes: > > > On Tue, Aug 23, 2022 at 9:56 AM Donald Hunter <donald.hunter@gmail.com> wrote: > >> > >> Extend the BPF program types documentation with a table of > >> program types, attach points and ELF section names. > >> > >> The program_types.csv file is generated from tools/lib/bpf/libbpf.c > >> and a script is included for regenerating the .csv file. > >> > >> I have not integrated the script into the doc build but if that > >> is desirable then please suggest the preferred way to do so. > >> > >> Signed-off-by: Donald Hunter <donald.hunter@gmail.com> > >> --- > > > > It does seem cleaner to generate this .csv during docs build, instead > > of having to manually regenerate it all the time? Should we also put > > it under Documentation/bpf/libbpf/ as it's libbpf-specific? Having it > > under libbpf subdir would also make it simpler to expose it in libbpf > > docs at libbpf.readthedocs.io/ > > Agreed about generating the .csv as part of the doc build. I will look > at adding it to the docs Makefile. > > I'm happy to put it in Documentation/bpf/libbpf and link to it from > Documentation/bpf/programs.rst. > > > We can probably also establish some special comment format next to > > SEC_DEF() to specify the format of those "extras", I think it would be > > useful for users. WDYT? > > Yes this would be a useful addition. Are the extras always for > auto-attach? If so, then I can add that to the rules. I think so. I can't recall any program type that can be auto-attached with just its type. > > I'd prefer to modify the existing ELF section name column to replace '+' > with extras since the table is already wide. > > > CC'ing Grant as well, who worked on building libbpf docs.
diff --git a/Documentation/bpf/program_types.csv b/Documentation/bpf/program_types.csv new file mode 100644 index 000000000000..adec046b0bde --- /dev/null +++ b/Documentation/bpf/program_types.csv @@ -0,0 +1,82 @@ +Program Type,Attach Type,ELF Section Name,Sleepable +``BPF_PROG_TYPE_SOCKET_FILTER``,,``socket``, +``BPF_PROG_TYPE_SK_REUSEPORT``,``BPF_SK_REUSEPORT_SELECT_OR_MIGRATE``,``sk_reuseport/migrate``, +``BPF_PROG_TYPE_SK_REUSEPORT``,``BPF_SK_REUSEPORT_SELECT``,``sk_reuseport``, +``BPF_PROG_TYPE_KPROBE``,,``kprobe+``, +``BPF_PROG_TYPE_KPROBE``,,``uprobe+``, +``BPF_PROG_TYPE_KPROBE``,,``uprobe.s+``,Yes +``BPF_PROG_TYPE_KPROBE``,,``kretprobe+``, +``BPF_PROG_TYPE_KPROBE``,,``uretprobe+``, +``BPF_PROG_TYPE_KPROBE``,,``uretprobe.s+``,Yes +``BPF_PROG_TYPE_KPROBE``,``BPF_TRACE_KPROBE_MULTI``,``kprobe.multi+``, +``BPF_PROG_TYPE_KPROBE``,``BPF_TRACE_KPROBE_MULTI``,``kretprobe.multi+``, +``BPF_PROG_TYPE_KPROBE``,,``ksyscall+``, +``BPF_PROG_TYPE_KPROBE``,,``kretsyscall+``, +``BPF_PROG_TYPE_KPROBE``,,``usdt+``, +``BPF_PROG_TYPE_SCHED_CLS``,,``tc``, +``BPF_PROG_TYPE_SCHED_CLS``,,``classifier``, +``BPF_PROG_TYPE_SCHED_ACT``,,``action``, +``BPF_PROG_TYPE_TRACEPOINT``,,``tracepoint+``, +``BPF_PROG_TYPE_TRACEPOINT``,,``tp+``, +``BPF_PROG_TYPE_RAW_TRACEPOINT``,,``raw_tracepoint+``, +``BPF_PROG_TYPE_RAW_TRACEPOINT``,,``raw_tp+``, +``BPF_PROG_TYPE_RAW_TRACEPOINT_WRITABLE``,,``raw_tracepoint.w+``, +``BPF_PROG_TYPE_RAW_TRACEPOINT_WRITABLE``,,``raw_tp.w+``, +``BPF_PROG_TYPE_TRACING``,``BPF_TRACE_RAW_TP``,``tp_btf+``, +``BPF_PROG_TYPE_TRACING``,``BPF_TRACE_FENTRY``,``fentry+``, +``BPF_PROG_TYPE_TRACING``,``BPF_MODIFY_RETURN``,``fmod_ret+``, +``BPF_PROG_TYPE_TRACING``,``BPF_TRACE_FEXIT``,``fexit+``, +``BPF_PROG_TYPE_TRACING``,``BPF_TRACE_FENTRY``,``fentry.s+``,Yes +``BPF_PROG_TYPE_TRACING``,``BPF_MODIFY_RETURN``,``fmod_ret.s+``,Yes +``BPF_PROG_TYPE_TRACING``,``BPF_TRACE_FEXIT``,``fexit.s+``,Yes +``BPF_PROG_TYPE_EXT``,,``freplace+``, +``BPF_PROG_TYPE_LSM``,``BPF_LSM_MAC``,``lsm+``, +``BPF_PROG_TYPE_LSM``,``BPF_LSM_MAC``,``lsm.s+``,Yes +``BPF_PROG_TYPE_LSM``,``BPF_LSM_CGROUP``,``lsm_cgroup+``, +``BPF_PROG_TYPE_TRACING``,``BPF_TRACE_ITER``,``iter+``, +``BPF_PROG_TYPE_TRACING``,``BPF_TRACE_ITER``,``iter.s+``,Yes +``BPF_PROG_TYPE_SYSCALL``,,``syscall``,Yes +``BPF_PROG_TYPE_XDP``,``BPF_XDP_DEVMAP``,``xdp.frags/devmap``, +``BPF_PROG_TYPE_XDP``,``BPF_XDP_DEVMAP``,``xdp/devmap``, +``BPF_PROG_TYPE_XDP``,``BPF_XDP_CPUMAP``,``xdp.frags/cpumap``, +``BPF_PROG_TYPE_XDP``,``BPF_XDP_CPUMAP``,``xdp/cpumap``, +``BPF_PROG_TYPE_XDP``,``BPF_XDP``,``xdp.frags``, +``BPF_PROG_TYPE_XDP``,``BPF_XDP``,``xdp``, +``BPF_PROG_TYPE_PERF_EVENT``,,``perf_event``, +``BPF_PROG_TYPE_LWT_IN``,,``lwt_in``, +``BPF_PROG_TYPE_LWT_OUT``,,``lwt_out``, +``BPF_PROG_TYPE_LWT_XMIT``,,``lwt_xmit``, +``BPF_PROG_TYPE_LWT_SEG6LOCAL``,,``lwt_seg6local``, +``BPF_PROG_TYPE_SOCK_OPS``,``BPF_CGROUP_SOCK_OPS``,``sockops``, +``BPF_PROG_TYPE_SK_SKB``,``BPF_SK_SKB_STREAM_PARSER``,``sk_skb/stream_parser``, +``BPF_PROG_TYPE_SK_SKB``,``BPF_SK_SKB_STREAM_VERDICT``,``sk_skb/stream_verdict``, +``BPF_PROG_TYPE_SK_SKB``,,``sk_skb``, +``BPF_PROG_TYPE_SK_MSG``,``BPF_SK_MSG_VERDICT``,``sk_msg``, +``BPF_PROG_TYPE_LIRC_MODE2``,``BPF_LIRC_MODE2``,``lirc_mode2``, +``BPF_PROG_TYPE_FLOW_DISSECTOR``,``BPF_FLOW_DISSECTOR``,``flow_dissector``, +``BPF_PROG_TYPE_CGROUP_SKB``,``BPF_CGROUP_INET_INGRESS``,``cgroup_skb/ingress``, +``BPF_PROG_TYPE_CGROUP_SKB``,``BPF_CGROUP_INET_EGRESS``,``cgroup_skb/egress``, +``BPF_PROG_TYPE_CGROUP_SKB``,,``cgroup/skb``, +``BPF_PROG_TYPE_CGROUP_SOCK``,``BPF_CGROUP_INET_SOCK_CREATE``,``cgroup/sock_create``, +``BPF_PROG_TYPE_CGROUP_SOCK``,``BPF_CGROUP_INET_SOCK_RELEASE``,``cgroup/sock_release``, +``BPF_PROG_TYPE_CGROUP_SOCK``,``BPF_CGROUP_INET_SOCK_CREATE``,``cgroup/sock``, +``BPF_PROG_TYPE_CGROUP_SOCK``,``BPF_CGROUP_INET4_POST_BIND``,``cgroup/post_bind4``, +``BPF_PROG_TYPE_CGROUP_SOCK``,``BPF_CGROUP_INET6_POST_BIND``,``cgroup/post_bind6``, +``BPF_PROG_TYPE_CGROUP_SOCK_ADDR``,``BPF_CGROUP_INET4_BIND``,``cgroup/bind4``, +``BPF_PROG_TYPE_CGROUP_SOCK_ADDR``,``BPF_CGROUP_INET6_BIND``,``cgroup/bind6``, +``BPF_PROG_TYPE_CGROUP_SOCK_ADDR``,``BPF_CGROUP_INET4_CONNECT``,``cgroup/connect4``, +``BPF_PROG_TYPE_CGROUP_SOCK_ADDR``,``BPF_CGROUP_INET6_CONNECT``,``cgroup/connect6``, +``BPF_PROG_TYPE_CGROUP_SOCK_ADDR``,``BPF_CGROUP_UDP4_SENDMSG``,``cgroup/sendmsg4``, +``BPF_PROG_TYPE_CGROUP_SOCK_ADDR``,``BPF_CGROUP_UDP6_SENDMSG``,``cgroup/sendmsg6``, +``BPF_PROG_TYPE_CGROUP_SOCK_ADDR``,``BPF_CGROUP_UDP4_RECVMSG``,``cgroup/recvmsg4``, +``BPF_PROG_TYPE_CGROUP_SOCK_ADDR``,``BPF_CGROUP_UDP6_RECVMSG``,``cgroup/recvmsg6``, +``BPF_PROG_TYPE_CGROUP_SOCK_ADDR``,``BPF_CGROUP_INET4_GETPEERNAME``,``cgroup/getpeername4``, +``BPF_PROG_TYPE_CGROUP_SOCK_ADDR``,``BPF_CGROUP_INET6_GETPEERNAME``,``cgroup/getpeername6``, +``BPF_PROG_TYPE_CGROUP_SOCK_ADDR``,``BPF_CGROUP_INET4_GETSOCKNAME``,``cgroup/getsockname4``, +``BPF_PROG_TYPE_CGROUP_SOCK_ADDR``,``BPF_CGROUP_INET6_GETSOCKNAME``,``cgroup/getsockname6``, +``BPF_PROG_TYPE_CGROUP_SYSCTL``,``BPF_CGROUP_SYSCTL``,``cgroup/sysctl``, +``BPF_PROG_TYPE_CGROUP_SOCKOPT``,``BPF_CGROUP_GETSOCKOPT``,``cgroup/getsockopt``, +``BPF_PROG_TYPE_CGROUP_SOCKOPT``,``BPF_CGROUP_SETSOCKOPT``,``cgroup/setsockopt``, +``BPF_PROG_TYPE_CGROUP_DEVICE``,``BPF_CGROUP_DEVICE``,``cgroup/dev``, +``BPF_PROG_TYPE_STRUCT_OPS``,,``struct_ops+``, +``BPF_PROG_TYPE_SK_LOOKUP``,``BPF_SK_LOOKUP``,``sk_lookup``, diff --git a/Documentation/bpf/programs.rst b/Documentation/bpf/programs.rst index 620eb667ac7a..71448fe0b955 100644 --- a/Documentation/bpf/programs.rst +++ b/Documentation/bpf/programs.rst @@ -7,3 +7,18 @@ Program Types :glob: prog_* + +Program Types and libbpf +======================== + +The table below lists the program types, their attach types where relevant and the ELF section +names supported by libbpf for them. The ELF section names follow these rules: + +- ``type`` is an exact match, e.g. ``SEC("socket")`` +- ``type+`` means it can be either exact ``SEC("type")`` or well-formed ``SEC("type/extras")`` + with a ‘``/``’ separator, e.g. ``SEC("tracepoint/syscalls/sys_enter_open")`` + +.. csv-table:: Program Types and Their ELF Section Names + :file: program_types.csv + :widths: 40 30 20 10 + :header-rows: 1 diff --git a/scripts/gen-bpf-progtypes.sh b/scripts/gen-bpf-progtypes.sh new file mode 100755 index 000000000000..fb9650bd5c1b --- /dev/null +++ b/scripts/gen-bpf-progtypes.sh @@ -0,0 +1,21 @@ +#!/bin/bash +# SPDX-License-Identifier: GPL-2.0-only +# +# Copyright (C) Red Hat. +# +# Generate a .csv table of BPF program types + +if ! [ -d "tools/lib/bpf" -a -d "Documentation" ]; then + echo "Run from top level of kernel tree" + exit 1 +fi + +awk -F'[",[:space:]]+' \ + 'BEGIN { print "Program Type,Attach Type,ELF Section Name,Sleepable" } + /SEC_DEF\(\"/ && !/SEC_DEPRECATED/ { + type = "``BPF_PROG_TYPE_" $4 "``" + attach = index($5, "0") ? "" : "``" $5 "``"; + section = "``" $3 "``" + sleepable = index($0, "SEC_SLEEPABLE") ? "Yes" : ""; + print type "," attach "," section "," sleepable }' \ +tools/lib/bpf/libbpf.c > Documentation/bpf/program_types.csv
Extend the BPF program types documentation with a table of program types, attach points and ELF section names. The program_types.csv file is generated from tools/lib/bpf/libbpf.c and a script is included for regenerating the .csv file. I have not integrated the script into the doc build but if that is desirable then please suggest the preferred way to do so. Signed-off-by: Donald Hunter <donald.hunter@gmail.com> --- Documentation/bpf/program_types.csv | 82 +++++++++++++++++++++++++++++ Documentation/bpf/programs.rst | 15 ++++++ scripts/gen-bpf-progtypes.sh | 21 ++++++++ 3 files changed, 118 insertions(+) create mode 100644 Documentation/bpf/program_types.csv create mode 100755 scripts/gen-bpf-progtypes.sh