Re: [PATCH bpf-next] bpf: Add show_fdinfo for perf_event

[Tao Chen <chen.dylane@linux.dev>]

在 2025/6/6 02:41, Andrii Nakryiko 写道:
> On Wed, Jun 4, 2025 at 9:37 AM Tao Chen <chen.dylane@linux.dev> wrote:
>>
>> After commit 1b715e1b0ec5 ("bpf: Support ->fill_link_info for perf_event") add
>> perf_event info, we can also show the info with the method of cat /proc/[fd]/fdinfo.
>>
>> kprobe fdinfo:
>> link_type:      perf
>> link_id:        2
>> prog_tag:       bcf7977d3b93787c
>> prog_id:        18
>> name:   bpf_fentry_test1
>> offset: 0
>> missed: 0
>> addr:   ffffffffaea8d134
>> event_type:     3
>> cookie: 3735928559
>>
>> uprobe fdinfo:
>> link_type:      perf
>> link_id:        6
>> prog_tag:       bcf7977d3b93787c
>> prog_id:        7
>> name:   /proc/self/exe
>> offset: 6507541
>> event_type:     1
>> cookie: 3735928559
>>
>> tracepoint fdinfo:
>> link_type:      perf
>> link_id:        4
>> prog_tag:       bcf7977d3b93787c
>> prog_id:        8
>> tp_name:        sched_switch
>> event_type:     5
>> cookie: 3735928559
>>
>> perf_event fdinfo:
>> link_type:      perf
>> link_id:        5
>> prog_tag:       bcf7977d3b93787c
>> prog_id:        9
>> type:   1
>> config: 2
>> event_type:     6
>> cookie: 3735928559
>>
>> Signed-off-by: Tao Chen <chen.dylane@linux.dev>
>> ---
>>   kernel/bpf/syscall.c | 126 +++++++++++++++++++++++++++++++++++++++++++
>>   1 file changed, 126 insertions(+)
>>
>> diff --git a/kernel/bpf/syscall.c b/kernel/bpf/syscall.c
>> index 9794446bc8..9af54852eb 100644
>> --- a/kernel/bpf/syscall.c
>> +++ b/kernel/bpf/syscall.c
>> @@ -3793,6 +3793,35 @@ static int bpf_perf_link_fill_kprobe(const struct perf_event *event,
>>          info->perf_event.kprobe.cookie = event->bpf_cookie;
>>          return 0;
>>   }
>> +
>> +static void bpf_perf_link_fdinfo_kprobe(const struct perf_event *event,
>> +                                       struct seq_file *seq)
>> +{
>> +       const char *name;
>> +       int err;
>> +       u32 prog_id, type;
>> +       u64 offset, addr;
>> +       unsigned long missed;
>> +
>> +       err = bpf_get_perf_event_info(event, &prog_id, &type, &name,
>> +                                     &offset, &addr, &missed);
>> +       if (err)
>> +               return;
>> +
>> +       if (type == BPF_FD_TYPE_KRETPROBE)
>> +               type = BPF_PERF_EVENT_KRETPROBE;
>> +       else
>> +               type = BPF_PERF_EVENT_KPROBE;
> 
> maybe use "kretprobe" and "kprobe" strings?
> 

It looks more readable, i will change it in v2, thanks.

>> +
>> +       seq_printf(seq,
>> +                  "name:\t%s\n"
>> +                  "offset:\t%llu\n"
> 
> llx, hex makes most sense (we had similar discussion within the
> context of bpftool reporting)
> 
> pw-bot: cr
> 
>> +                  "missed:\t%lu\n"
>> +                  "addr:\t%llx\n"
> 
> ditto, address -> hex
> 

will change it in v2, thanks.

>> +                  "event_type:\t%u\n"
>> +                  "cookie:\t%llu\n",
>> +                  name, offset, missed, addr, type, event->bpf_cookie);
>> +}
>>   #endif
>>
>>   #ifdef CONFIG_UPROBE_EVENTS
>> @@ -3820,6 +3849,34 @@ static int bpf_perf_link_fill_uprobe(const struct perf_event *event,
>>          info->perf_event.uprobe.cookie = event->bpf_cookie;
>>          return 0;
>>   }
>> +
>> +static void bpf_perf_link_fdinfo_uprobe(const struct perf_event *event,
>> +                                       struct seq_file *seq)
>> +{
>> +       const char *name;
>> +       int err;
>> +       u32 prog_id, type;
>> +       u64 offset, addr;
>> +       unsigned long missed;
>> +
>> +       err = bpf_get_perf_event_info(event, &prog_id, &type, &name,
>> +                                     &offset, &addr, &missed);
>> +       if (err)
>> +               return;
>> +
>> +       if (type == BPF_FD_TYPE_URETPROBE)
>> +               type = BPF_PERF_EVENT_URETPROBE;
>> +       else
>> +               type = BPF_PERF_EVENT_UPROBE;
> 
> strings, just as above
> 

get it.

>> +
>> +       seq_printf(seq,
>> +                  "name:\t%s\n"
>> +                  "offset:\t%llu\n"
> 
> hex

get it.

> 
>> +                  "event_type:\t%u\n"
>> +                  "cookie:\t%llu\n",
>> +                  name, offset, type, event->bpf_cookie);
>> +
>> +}
>>   #endif
>>
>>   static int bpf_perf_link_fill_probe(const struct perf_event *event,
>> @@ -3888,10 +3945,79 @@ static int bpf_perf_link_fill_link_info(const struct bpf_link *link,
>>          }
>>   }
>>
>> +static void bpf_perf_event_link_show_fdinfo(const struct perf_event *event,
>> +                                           struct seq_file *seq)
>> +{
>> +       seq_printf(seq,
>> +                  "type:\t%u\n"
>> +                  "config:\t%llu\n"
>> +                  "event_type:\t%u\n"
> 
> string?
> 

sure.

>> +                  "cookie:\t%llu\n",
>> +                  event->attr.type, event->attr.config,
>> +                  BPF_PERF_EVENT_EVENT, event->bpf_cookie);
>> +}
>> +
>> +static void bpf_tracepoint_link_show_fdinfo(const struct perf_event *event,
>> +                                           struct seq_file *seq)
>> +{
>> +       int err;
>> +       const char *name;
>> +       u32 prog_id;
>> +
>> +       err = bpf_get_perf_event_info(event, &prog_id, NULL, &name, NULL,
>> +                                     NULL, NULL);
>> +       if (err)
>> +               return;
>> +
>> +       seq_printf(seq,
>> +                  "tp_name:\t%s\n"
>> +                  "event_type:\t%u\n"
> 
> string
> 

get it.

>> +                  "cookie:\t%llu\n",
>> +                  name, BPF_PERF_EVENT_TRACEPOINT, event->bpf_cookie);
>> +}
>> +
>> +static void bpf_probe_link_show_fdinfo(const struct perf_event *event,
>> +                                      struct seq_file *seq)
>> +{
>> +#ifdef CONFIG_KPROBE_EVENTS
>> +       if (event->tp_event->flags & TRACE_EVENT_FL_KPROBE)
>> +               return bpf_perf_link_fdinfo_kprobe(event, seq);
>> +#endif
>> +
>> +#ifdef CONFIG_UPROBE_EVENTS
>> +       if (event->tp_event->flags & TRACE_EVENT_FL_UPROBE)
>> +               return bpf_perf_link_fdinfo_uprobe(event, seq);
>> +#endif
>> +}
>> +
>> +static void bpf_perf_link_show_fdinfo(const struct bpf_link *link,
>> +                                     struct seq_file *seq)
>> +{
>> +       struct bpf_perf_link *perf_link;
>> +       const struct perf_event *event;
>> +
>> +       perf_link = container_of(link, struct bpf_perf_link, link);
>> +       event = perf_get_event(perf_link->perf_file);
>> +       if (IS_ERR(event))
>> +               return;
>> +
>> +       switch (event->prog->type) {
>> +       case BPF_PROG_TYPE_PERF_EVENT:
>> +               return bpf_perf_event_link_show_fdinfo(event, seq);
>> +       case BPF_PROG_TYPE_TRACEPOINT:
>> +               return bpf_tracepoint_link_show_fdinfo(event, seq);
>> +       case BPF_PROG_TYPE_KPROBE:
>> +               return bpf_probe_link_show_fdinfo(event, seq);
>> +       default:
>> +               return;
>> +       }
>> +}
>> +
>>   static const struct bpf_link_ops bpf_perf_link_lops = {
>>          .release = bpf_perf_link_release,
>>          .dealloc = bpf_perf_link_dealloc,
>>          .fill_link_info = bpf_perf_link_fill_link_info,
>> +       .show_fdinfo = bpf_perf_link_show_fdinfo,
>>   };
>>
>>   static int bpf_perf_link_attach(const union bpf_attr *attr, struct bpf_prog *prog)
>> --
>> 2.43.0
>>


-- 
Best Regards
Tao Chen