Re: [PATCHv2 bpf-next 09/11] bpf: Remove trace_in argument from __bpf_get_stack

[Jiri Olsa <olsajiri@gmail.com>]

On Thu, Jul 30, 2026 at 04:06:14PM -0700, Andrii Nakryiko wrote:

SNIP

> >  BPF_CALL_4(bpf_get_stack_pe, struct bpf_perf_event_data_kern *, ctx,
> >            void *, buf, u32, size, u64, flags)
> >  {
> > @@ -947,7 +970,7 @@ BPF_CALL_4(bpf_get_stack_pe, struct bpf_perf_event_data_kern *, ctx,
> >         int err = -EINVAL;
> >
> >         if (!(event->attr.sample_type & PERF_SAMPLE_CALLCHAIN))
> > -               return __bpf_get_stack(regs, NULL, NULL, buf, size, flags, false /* !may_fault */);
> > +               return __bpf_get_stack(regs, NULL, buf, size, flags, false /* !may_fault */);
> >
> >         if (unlikely(flags & ~(BPF_F_SKIP_FIELD_MASK | BPF_F_USER_STACK |
> >                                BPF_F_USER_BUILD_ID)))
> > @@ -966,7 +989,7 @@ BPF_CALL_4(bpf_get_stack_pe, struct bpf_perf_event_data_kern *, ctx,
> >
> >         if (kernel) {
> >                 trace->nr = nr_kernel;
> 
> this whole count_kernel_ip() logic above, why do we have it? I don't
> think we can have combined user and kernel stack trace, so how can we
> end up with PERF_CONTEXT_USER "ip" at all? I might be missing

the callchain comes from perf event's callchain and that can
have both kernel and user part

> something subtle here, but can you please check again if this whole
> trace->nr calculation/override is even necessary.

as I wrote in the other reply I don't think we need to change
trace->nr directly, just pass the needed nr/count as argument

jirka

> 
> > -               err = __bpf_get_stack(regs, NULL, trace, buf, size, flags, false /* !may_fault */);
> > +               err = __bpf_get_stack_pe(trace, buf, size, flags);
> >
> >         } else { /* user */
> >                 u64 skip = flags & BPF_F_SKIP_FIELD_MASK;
> > @@ -974,9 +997,8 @@ BPF_CALL_4(bpf_get_stack_pe, struct bpf_perf_event_data_kern *, ctx,
> >                 skip += nr_kernel;
> >                 if (skip > BPF_F_SKIP_FIELD_MASK)
> >                         goto clear;
> > -
> >                 flags = (flags & ~BPF_F_SKIP_FIELD_MASK) | skip;
> > -               err = __bpf_get_stack(regs, NULL, trace, buf, size, flags, false /* !may_fault */);
> > +               err = __bpf_get_stack_pe(trace, buf, size, flags);
> >         }
> >
> >         /* restore nr */
> > --
> > 2.54.0
> >