Re: [PATCH] bpf: fix a warning message in mark_ptr_not_null_reg()

[Daniel Borkmann]

On 2/16/21 10:10 PM, KP Singh wrote:
> On Tue, Feb 16, 2021 at 8:37 PM Dan Carpenter <dan.carpenter@xxxxxxxxxx> wrote:
> > The WARN_ON() argument is a condition, and it generates a stack trace
> > but it doesn't print the warning.
> >
> > Fixes: 4ddb74165ae5 ("bpf: Extract nullable reg type conversion into a helper function")
> > Signed-off-by: Dan Carpenter <dan.carpenter@xxxxxxxxxx>
> > ---
> >   kernel/bpf/verifier.c | 2 +-
> >   1 file changed, 1 insertion(+), 1 deletion(-)
> >
> > diff --git a/kernel/bpf/verifier.c b/kernel/bpf/verifier.c
> > index 056df6be3e30..bd4d1dfca73c 100644
> > --- a/kernel/bpf/verifier.c
> > +++ b/kernel/bpf/verifier.c
> > @@ -1120,7 +1120,7 @@ static void mark_ptr_not_null_reg(struct bpf_reg_state *reg)
> >                  reg->type = PTR_TO_RDWR_BUF;
> >                  break;
> >          default:
> > -               WARN_ON("unknown nullable register type");
> > +               WARN(1, "unknown nullable register type");
>
> Should we use WARN_ONCE here? Also, I think the fix should be targeted
> for bpf-next as
> the patch that introduced this hasn't made it to bpf yet.
>
> [...]

Usually we have something like `verbose(env, "kernel subsystem misconfigured verifier\n")`,
but in this case we'd need to drag env all the way to here. :/ I agree with WARN_ONCE().