On 1/6/20 11:32 PM, Alexei Starovoitov wrote:
On Fri, Jan 03, 2020 at 11:18:28AM -0800, Song Liu wrote:
On Fri, Jan 3, 2020 at 12:27 AM Simon Horman <simon.horman@xxxxxxxxxxxxx> wrote:
On Fri, Jan 03, 2020 at 02:02:33PM +0800, Li RongQing wrote:
bpf_skb_generic_push always returns 0, not need to check
its return, so change its type as void
Signed-off-by: Li RongQing <lirongqing@xxxxxxxxx>
Reviewed-by: Simon Horman <simon.horman@xxxxxxxxxxxxx>
Acked-by: Song Liu <songliubraving@xxxxxx>
---
net/core/filter.c | 30 ++++++++++--------------------
1 file changed, 10 insertions(+), 20 deletions(-)
diff --git a/net/core/filter.c b/net/core/filter.c
index 42fd17c48c5f..1cbac34a4e11 100644
--- a/net/core/filter.c
+++ b/net/core/filter.c
...
@@ -5144,7 +5134,7 @@ BPF_CALL_3(bpf_lwt_seg6_adjust_srh, struct sk_buff *, skb, u32, offset,
if (unlikely(ret < 0))
return ret;
- ret = bpf_skb_net_hdr_push(skb, offset, len);
+ bpf_skb_net_hdr_push(skb, offset, len);
There is a check for (ret < 0) just below this if block.
That is ok becuase in order to get to here (ret < 0) must
be true as per the check a few lines above.
So I think this is ok although the asymmetry with the else arm
of this if statement is not ideal IMHO.
Agreed with this concern. But I cannot think of any free solution. I guess we
will just live with assumption that skb_cow_head() never return >0.
I don't think this patch is worth doing.
I can imagine bpf_skb_generic_push() handling some errors in the future.
compiler can do this optimization job instead.
Yep, fully agree.
