On Thu, Apr 28, 2022 at 11:22 AM Andrii Nakryiko
<andrii.nakryiko@xxxxxxxxx> wrote:
>
> On Thu, Apr 28, 2022 at 9:54 AM Delyan Kratunov <delyank@xxxxxx> wrote:
> >
> > uprobe and kprobe programs have the same program type, KPROBE, which is
> > currently not allowed to load sleepable programs.
> >
> > To avoid adding a new UPROBE type, we instead allow sleepable KPROBE
> > programs to load and defer the is-it-actually-a-uprobe-program check
> > to attachment time, where we're already validating the corresponding
> > perf_event.
> >
> > A corollary of this patch is that you can now load a sleepable kprobe
> > program but cannot attach it.
> >
> > Signed-off-by: Delyan Kratunov <delyank@xxxxxx>
> > ---
> > kernel/bpf/syscall.c | 8 ++++++++
> > kernel/bpf/verifier.c | 4 ++--
> > 2 files changed, 10 insertions(+), 2 deletions(-)
> >
> > diff --git a/kernel/bpf/syscall.c b/kernel/bpf/syscall.c
> > index e9e3e49c0eb7..3ce923f489d7 100644
> > --- a/kernel/bpf/syscall.c
> > +++ b/kernel/bpf/syscall.c
> > @@ -3009,6 +3009,14 @@ static int bpf_perf_link_attach(const union bpf_attr *attr, struct bpf_prog *pro
> > }
> >
> > event = perf_file->private_data;
> > + if (prog->aux->sleepable) {
> > + if (!event->tp_event || (event->tp_event->flags & TRACE_EVENT_FL_UPROBE) == 0) {
>
> so far TRACE_EVENT_FL_UPROBE was contained to within kernel/trace so
> far, maybe it's better to instead expose a helper function to check if
> perf_event represents uprobe?
or we can move prog->aux->sleepable check down into
perf_event_set_bpf_prog().
Which is probably cleaner.
We check other prog flags there.
