[bpf-next] bpf: return EOPNOTSUPP when invalid map type in __bpf_tx_xdp_map

a negative value -EOPNOTSUPP should be returned if map->map_type
is invalid although that seems unlikely now, then the caller will
continue to handle buffer, or else the buffer will be leaked

Signed-off-by: Li RongQing <lirongqing@baidu.com>
---
 net/core/filter.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

On Thu, Jan 2, 2020 at 10:26 PM Li RongQing <lirongqing@baidu.com> wrote:
>
> a negative value -EOPNOTSUPP should be returned if map->map_type
> is invalid although that seems unlikely now, then the caller will
> continue to handle buffer, or else the buffer will be leaked
>
> Signed-off-by: Li RongQing <lirongqing@baidu.com>

Acked-by: Song Liu <songliubraving@fb.com>

On 1/3/20 7:25 AM, Li RongQing wrote:
> a negative value -EOPNOTSUPP should be returned if map->map_type
> is invalid although that seems unlikely now, then the caller will
> continue to handle buffer, or else the buffer will be leaked
> 
> Signed-off-by: Li RongQing <lirongqing@baidu.com>
> ---
>   net/core/filter.c | 2 +-
>   1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/net/core/filter.c b/net/core/filter.c
> index 1cbac34a4e11..40fa5905321c 100644
> --- a/net/core/filter.c
> +++ b/net/core/filter.c
> @@ -3512,7 +3512,7 @@ static int __bpf_tx_xdp_map(struct net_device *dev_rx, void *fwd,
>   	case BPF_MAP_TYPE_XSKMAP:
>   		return __xsk_map_redirect(fwd, xdp);
>   	default:
> -		break;
> +		return -EOPNOTSUPP;

So in case of generic XDP we return with -EBADRQC in xdp_do_generic_redirect_map().
I would suggest we adapt the same error code here as well, so it's consistent for
the tracepoint output and not to be confused with -EOPNOTSUPP from other locations
like dev_map_enqueue() when ndo_xdp_xmit is missing or such.

>   	}
>   	return 0;
>   }
>

> -----邮件原件-----
> 发件人: Daniel Borkmann [mailto:daniel@iogearbox.net]
> 发送时间: 2020年1月9日 1:18
> 收件人: Li,Rongqing <lirongqing@baidu.com>
> 抄送: netdev@vger.kernel.org; bpf@vger.kernel.org; brouer@redhat.com
> 主题: Re: [PATCH][bpf-next] bpf: return EOPNOTSUPP when invalid map type in
> __bpf_tx_xdp_map
> 
> On 1/3/20 7:25 AM, Li RongQing wrote:
> > a negative value -EOPNOTSUPP should be returned if map->map_type is
> > invalid although that seems unlikely now, then the caller will
> > continue to handle buffer, or else the buffer will be leaked
> >
> > Signed-off-by: Li RongQing <lirongqing@baidu.com>
> > ---
> >   net/core/filter.c | 2 +-
> >   1 file changed, 1 insertion(+), 1 deletion(-)
> >
> > diff --git a/net/core/filter.c b/net/core/filter.c index
> > 1cbac34a4e11..40fa5905321c 100644
> > --- a/net/core/filter.c
> > +++ b/net/core/filter.c
> > @@ -3512,7 +3512,7 @@ static int __bpf_tx_xdp_map(struct net_device
> *dev_rx, void *fwd,
> >   	case BPF_MAP_TYPE_XSKMAP:
> >   		return __xsk_map_redirect(fwd, xdp);
> >   	default:
> > -		break;
> > +		return -EOPNOTSUPP;
> 
> So in case of generic XDP we return with -EBADRQC in
> xdp_do_generic_redirect_map().
> I would suggest we adapt the same error code here as well, so it's consistent
> for the tracepoint output and not to be confused with -EOPNOTSUPP from other
> locations like dev_map_enqueue() when ndo_xdp_xmit is missing or such.

Ok, I will send v2

Thanks

-Li


> 
> >   	}
> >   	return 0;
> >   }
> >