Re: [bpf-next v7 1/3] bpftool: Add auto_attach for bpf prog load|loadall
From: Quentin Monnet
Date: Mon Oct 10 2022 - 04:40:17 EST
Sat Oct 08 2022 06:16:42 GMT+0100 ~ wangyufen <wangyufen@xxxxxxxxxx>
>
> 在 2022/10/1 0:26, Quentin Monnet 写道:
>> Tue Sep 27 2022 12:21:14 GMT+0100 ~ Wang Yufen <wangyufen@xxxxxxxxxx>
>>> Add auto_attach optional to support one-step load-attach-pin_link.
>> Nit: Now "autoattach" instead of "auto_attach". Same in commit title.
> will change in v8, thanks.
>>
>>> For example,
>>> $ bpftool prog loadall test.o /sys/fs/bpf/test autoattach
>>>
>>> $ bpftool link
>>> 26: tracing name test1 tag f0da7d0058c00236 gpl
>>> loaded_at 2022-09-09T21:39:49+0800 uid 0
>>> xlated 88B jited 55B memlock 4096B map_ids 3
>>> btf_id 55
>>> 28: kprobe name test3 tag 002ef1bef0723833 gpl
>>> loaded_at 2022-09-09T21:39:49+0800 uid 0
>>> xlated 88B jited 56B memlock 4096B map_ids 3
>>> btf_id 55
>>> 57: tracepoint name oncpu tag 7aa55dfbdcb78941 gpl
>>> loaded_at 2022-09-09T21:41:32+0800 uid 0
>>> xlated 456B jited 265B memlock 4096B map_ids 17,13,14,15
>>> btf_id 82
>>>
>>> $ bpftool link
>>> 1: tracing prog 26
>>> prog_type tracing attach_type trace_fentry
>>> 3: perf_event prog 28
>>> 10: perf_event prog 57
>>>
>>> The autoattach optional can support tracepoints, k(ret)probes,
>>> u(ret)probes.
>>>
>>> Signed-off-by: Wei Yongjun <weiyongjun1@xxxxxxxxxx>
>>> Signed-off-by: Wang Yufen <wangyufen@xxxxxxxxxx>
>>> ---
>>> v6 -> v7: add info msg print and update doc for the skip program
>>> v5 -> v6: skip the programs not supporting auto-attach,
>>> and change optional name from "auto_attach" to "autoattach"
>>> v4 -> v5: some formatting nits of doc
>>> v3 -> v4: rename functions, update doc, bash and do_help()
>>> v2 -> v3: switch to extend prog load command instead of extend perf
>>> v2:
>>> https://patchwork.kernel.org/project/netdevbpf/patch/20220824033837.458197-1-weiyongjun1@xxxxxxxxxx/
>>> v1:
>>> https://patchwork.kernel.org/project/netdevbpf/patch/20220816151725.153343-1-weiyongjun1@xxxxxxxxxx/
>>> tools/bpf/bpftool/prog.c | 81
>>> ++++++++++++++++++++++++++++++++++++++++++++++--
>>> 1 file changed, 79 insertions(+), 2 deletions(-)
>>>
>>> diff --git a/tools/bpf/bpftool/prog.c b/tools/bpf/bpftool/prog.c
>>> index c81362a..84eced8 100644
>>> --- a/tools/bpf/bpftool/prog.c
>>> +++ b/tools/bpf/bpftool/prog.c
>>> @@ -1453,6 +1453,72 @@ static int do_run(int argc, char **argv)
>>> return ret;
>>> }
>>> +static int
>>> +auto_attach_program(struct bpf_program *prog, const char *path)
>>> +{
>>> + struct bpf_link *link;
>>> + int err;
>>> +
>>> + link = bpf_program__attach(prog);
>>> + if (!link)
>>> + return -1;
>>> +
>>> + err = bpf_link__pin(link, path);
>>> + if (err) {
>>> + bpf_link__destroy(link);
>>> + return err;
>>> + }
>>> + return 0;
>>> +}
>>> +
>>> +static int pathname_concat(const char *path, const char *name, char
>>> *buf)
>>> +{
>>> + int len;
>>> +
>>> + len = snprintf(buf, PATH_MAX, "%s/%s", path, name);
>>> + if (len < 0)
>>> + return -EINVAL;
>>> + if (len >= PATH_MAX)
>>> + return -ENAMETOOLONG;
>>> +
>>> + return 0;
>>> +}
>>> +
>>> +static int
>>> +auto_attach_programs(struct bpf_object *obj, const char *path)
>>> +{
>>> + struct bpf_program *prog;
>>> + char buf[PATH_MAX];
>>> + int err;
>>> +
>>> + bpf_object__for_each_program(prog, obj) {
>>> + err = pathname_concat(path, bpf_program__name(prog), buf);
>>> + if (err)
>>> + goto err_unpin_programs;
>>> +
>>> + err = auto_attach_program(prog, buf);
>>> + if (!err)
>>> + continue;
>>> + if (errno == EOPNOTSUPP)
>>> + p_info("Program %s does not support autoattach",
>>> + bpf_program__name(prog));
>>> + else
>>> + goto err_unpin_programs
>> With this code, if auto-attach fails, then we skip this program and move
>> on to the next. That's an improvement, but in that case the program
>> won't remain loaded in the kernel after bpftool exits. My suggestion in
>> my previous message (sorry if it was not clear) was to fall back to
>> regular pinning in that case (bpf_obj_pin()), along with the p_info()
>> message, so we can have the program pinned but not attached and let the
>> user know. If regular pinning fails as well, then we should unpin all
>> and error out, for consistency with bpf_object__pin_programs().
>>
>> And in that case, the (errno == EOPNOTSUPP) with fallback to regular
>> pinning could maybe be moved into auto_attach_program(), so that
>> auto-attaching single programs can use the fallback too?
>>
>> Thanks,
>> Quentin
>
> If I understand correctly, can we just check link? as following:
Yes, this is exactly what I meant
>
> --- a/tools/bpf/bpftool/prog.c
> +++ b/tools/bpf/bpftool/prog.c
> @@ -1460,9 +1460,10 @@ static int do_run(int argc, char **argv)
> int err;
>
> link = bpf_program__attach(prog);
> - if (!link)
> - return -1;
> -
> + if (!link) {
> + p_info("Program %s attach failed",
> bpf_program__name(prog));
> + return bpf_obj_pin(bpf_program__fd(prog), path);
> + }
> err = bpf_link__pin(link, path);
> if (err) {
> bpf_link__destroy(link);
> @@ -1499,9 +1500,6 @@ static int pathname_concat(const char *path, const
> char *name, char *buf)
> err = auto_attach_program(prog, buf);
> if (!err)
> continue;
> - if (errno == EOPNOTSUPP)
> - p_info("Program %s does not support autoattach",
p_info("Program %s does not support autoattach, falling back to pinning"
> - bpf_program__name(prog));
> else
> goto err_unpin_programs;
> }
>
>
> and the doc is modified as follows:
>
> If the program does not support autoattach, will do regular pin along
> with an
> info message such as "Program %s attach failed". If the *OBJ* contains
> multiple
> programs and **loadall** is used, if the program A in these programs
> does not
> support autoattach, the program A will do regular pin along with an info
> message,
> and continue to autoattach the next program.
Not sure the "program A" designation helps too much, I'd simply write this:
"If a program does not support autoattach, bpftool falls back to regular
pinning for that program instead."
Which should be enough for both the "load" and "loadall" behaviours? I
wouldn't mention the help message in the docs (the p_info() won't show
up in the JSON output for example).
Looks good otherwise, thanks!