[PATCH bpf-next 2/3] bpf, arm64: Add JIT support for stack arguments

[Alexei Starovoitov]

On Tue, Apr 21, 2026 at 4:53 AM Puranjay Mohan <puranjay12 at gmail.com> wrote:
>
> On Tue, Apr 21, 2026 at 3:58 AM Alexei Starovoitov
> <alexei.starovoitov at gmail.com> wrote:
> >
> > On Mon, Apr 20, 2026 at 8:36 AM Puranjay Mohan <puranjay at kernel.org> wrote:
> > >
> >
> > nice and clean. I like how it maps to arm64 calling convention.
> >
> > > +       if (prog->aux->stack_arg_depth > prog->aux->incoming_stack_arg_depth) {
> > > +               u16 outgoing = prog->aux->stack_arg_depth - prog->aux->incoming_stack_arg_depth;
> > > +               int nr_on_stack = outgoing / sizeof(u64) - NR_STACK_ARG_REGS;
> > > +
> > > +               if (nr_on_stack > 0)
> > > +                       ctx.stack_arg_size = round_up(nr_on_stack * sizeof(u64), 16);
> > > +       }
> >
> > I'm struggling to understand this part.
> > Why do this when this func calls more than what callee passed in?
> > Looks fishy. I'd like to see selftests with more than 6,7,8 args.
> > Because only then this logic will kick in?
>
> Your confusion stems from the naming of "incoming_stack_arg_depth" and
> "stack_arg_depth" (this should be called total_stack_arg_depth in my
> opinion)
>
> So, if you see fixups.c
>
>                 func[i]->aux->incoming_stack_arg_depth =
> env->subprog_info[i].incoming_stack_arg_depth;
>                 func[i]->aux->stack_arg_depth =
> env->subprog_info[i].incoming_stack_arg_depth +
>
> env->subprog_info[i].outgoing_stack_arg_depth;
>
> prog->aux->stack_arg_depth doesn't store outgoing stack depth, rather
> it has the sum of both incoming and outgoing, that means if a func
> doesn't call any function with more than 5 arguments but receives more
> than five arguments, incoming_stack_arg_depth will be equal to
> stack_arg_depth.

Ohh. That's indeed all too confusing.
See my response to Yonghong.
I think stack_arg_depth should mean outgoing
and incoming_stack_arg_depth should mean incoming only and
it shouldn't be even used by JIT.
That memory was allocated by caller, so to JIT this callee
the conversion of r11+const is straightforward and no checks necessary.

> if (prog->aux->stack_arg_depth > prog->aux->incoming_stack_arg_depth)
>
> This check is for - "Does this function call any function with more
> than 5 arguments", if yes, is it more than 8? if yes allocate stack
> space, otherwise stack space is not needed because argument 6,7,8 can
> live in arm64 registers.

I think it should really be one check based on stack_arg_depth.