On Fri, Jan 02, 2026 at 02:52:25PM +0000, Will Deacon wrote:
> On Wed, Nov 05, 2025 at 01:59:23PM +0100, Jiri Olsa wrote:
> > hi,
> > Mahe reported issue with bpf_override_return helper not working
> > when executed from kprobe.multi bpf program on arm.
> >
> > The problem seems to be that on arm we use alternate storage for
> > pt_regs object that is passed to bpf_prog_run and if any register
> > is changed (which is the case of bpf_override_return) it's not
> > propagated back to actual pt_regs object.
> >
> > The change below seems to fix the issue, but I have no idea if
> > that's proper fix for arm, thoughts?
> >
> > I'm attaching selftest to actually test bpf_override_return helper
> > functionality, because currently we only test that we are able to
> > attach a program with it, but not the override itself.
> >
> > thanks,
> > jirka
> >
> >
> > ---
> > arch/arm64/include/asm/ftrace.h | 11 +++++++++++
> > include/linux/ftrace.h | 3 +++
> > kernel/trace/bpf_trace.c | 1 +
> > 3 files changed, 15 insertions(+)
> >
> > diff --git a/arch/arm64/include/asm/ftrace.h
> > b/arch/arm64/include/asm/ftrace.h
> > index ba7cf7fec5e9..ad6cf587885c 100644
> > --- a/arch/arm64/include/asm/ftrace.h
> > +++ b/arch/arm64/include/asm/ftrace.h
> > @@ -157,6 +157,17 @@ ftrace_partial_regs(const struct ftrace_regs *fregs,
> > struct pt_regs *regs)
> > return regs;
> > }
> >
> > +static __always_inline void
> > +ftrace_partial_regs_fix(const struct ftrace_regs *fregs, struct pt_regs
> > *regs)
> > +{
> > + struct __arch_ftrace_regs *afregs = arch_ftrace_regs(fregs);
> > +
> > + if (afregs->pc != regs->pc) {
> > + afregs->pc = regs->pc;
> > + afregs->regs[0] = regs->regs[0];
> > + }
> > +}
>
> This looks a bit grotty to me and presumably other architectures would
> need similar treatement. Wouldn't it be cleaner to reuse the existing
> API instead? For example, by calling ftrace_regs_set_instruction_pointer()
> and ftrace_regs_set_return_value() to update the relevant registers from
> the core code?
I knew I forgot some change.. thanks for replying
ftrace_partial_regs is overloaded in arm64 and because of that we need
to propagate the change to pt_regs, so I think the ftrace_partial_regs_fix
code is arm64 specific, so can't see that in core code
also wrt ftrace_partial_regs_fix name, I was thinking it might be better
to have begin/end functions, like:
ftrace_partial_regs_begin
ftrace_partial_regs_end
thanks,
jirka
--- a/kernel/trace/bpf_trace.c
+++ b/kernel/trace/bpf_trace.c
@@ -2560,10 +2560,11 @@ kprobe_multi_link_prog_run(struct bpf_kprobe_multi_link
*link,
}
rcu_read_lock();
- regs = ftrace_partial_regs(fregs, bpf_kprobe_multi_pt_regs_ptr());
+ regs = ftrace_partial_regs_begin(fregs, bpf_kprobe_multi_pt_regs_ptr());
old_run_ctx = bpf_set_run_ctx(&run_ctx.session_ctx.run_ctx);
err = bpf_prog_run(link->link.prog, regs);
bpf_reset_run_ctx(old_run_ctx);
+ ftrace_partial_regs_end(fregs, bpf_kprobe_multi_pt_regs_ptr());
rcu_read_unlock();
out:
