Re: [PATCH bpf-next] bpf: Allow bpf_d_path in perf_event_mmap

[Florent Revest]

On Fri, Oct 29, 2021 at 5:17 PM Hengqi Chen <hengqi.chen@xxxxxxxxx> wrote:
>
>
>
> On 10/29/21 6:46 AM, Martin KaFai Lau wrote:
> > On Thu, Oct 28, 2021 at 06:43:57PM +0200, Florent Revest wrote:
> >> Allow the helper to be called from the perf_event_mmap hook. This is
> >> convenient to lookup vma->vm_file and implement a similar logic as
> >> perf_event_mmap_event in BPF.
> > From struct vm_area_struct:
> >       struct file * vm_file;          /* File we map to (can be NULL). */
> >
> > Under perf_event_mmap, vm_file won't be NULL or bpf_d_path can handle it?
> >
>
> Hmm, is perf_event_mmap a proper tracing target ?
> It does not appear in /sys/kernel/debug/tracing/available_filter_functions.
>
> I tried using kprobe/fentry to attach to it, both failed.

Good observation! :) In the bpf-next tree, perf_event_mmap is still
not exposed to tracing because the kernel/events/Makefile specifically
removes CC_FLAGS_FTRACE for core.c.
However, Song sent a patch to expose the functions of
kernel/events/core.c to tracing:
https://git.kernel.org/pub/scm/linux/kernel/git/peterz/queue.git/commit/kernel/events/Makefile?id=79df45731da68772d2285265864a52c900b8c65f

That patch is going through Peter Zijlstra's tree so it should
percolate down to Linus tree probably at ~the same time as this patch
(if it gets in :)). I tested this on bpf-next with Song's patch
cherry-picked.