Wed, Oct 25, 2017 at 04:07:26PM CEST, ogerl...@mellanox.com wrote:
>On 10/25/2017 4:58 PM, Jiri Pirko wrote:
>> Wed, Oct 25, 2017 at 03:44:40PM CEST, ogerl...@mellanox.com wrote:
>> > If we failed to offload a flow to HW, we should not be attempting to delete
>> > it from the HW. Also, on this case, we should be err-ing only if the flow
>> > is
>> > not is SW, fix both issues.
>> >
>> > Fixes: 717503b9cf57 ('net: sched: convert cls_flower->egress_dev users to
>> > tc_setup_cb_egdev infra')
>> > Signed-off-by: Or Gerlitz <ogerl...@mellanox.com>
>> > ---
>> > net/sched/cls_flower.c | 11 ++++-------
>> > 1 file changed, 4 insertions(+), 7 deletions(-)
>> >
>> > diff --git a/net/sched/cls_flower.c b/net/sched/cls_flower.c
>> > index 16f58ab..b98e0cb 100644
>> > --- a/net/sched/cls_flower.c
>> > +++ b/net/sched/cls_flower.c
>> > @@ -230,15 +230,12 @@ static int fl_hw_replace_filter(struct tcf_proto *tp,
>> >
>> > err = tc_setup_cb_call(block, &f->exts, TC_SETUP_CLSFLOWER,
>> > &cls_flower, skip_sw);
>> > - if (err < 0) {
>> > - fl_hw_destroy_filter(tp, f);
>> > - return err;
>>
>> As I wrote in the other thread: Yes, that is intentional. The thing is,
>> there might be multiple block callbacks registered and to be called. If
>> there is a fail with one, we need to cleanup all. So in your case you have 1
>> cb registered, that means that in case of an error during insertion, you
>> will get cb called to remove. Driver has to take care of that. I was
>> checking that and was under impression that mlx5 deals with that.
>
>I see, what about the other line I deleted of blankly returning err no matter
>regardless if we' re on skip_sw or not, do you agree this fix is needed, also
>see below
No. That is not needed. The current behaviour with the skip_sw is the
same as the original. I don't understand why you want to change it. I
also don't undestand why you do 2 things in one patch.
>> > - } else if (err > 0) {
>> > +
>> > + if (err > 0)
>> > f->flags |= TCA_CLS_FLAGS_IN_HW;
>> > - }
>> >
>> > - if (skip_sw && !(f->flags & TCA_CLS_FLAGS_IN_HW))
>> > - return -EINVAL;
>> > + if (skip_sw)
>> > + return err;
>> >
>> > return 0;
>> > }
>
>here's too, I don't see why we should return -EINVAL etc, and what's wrong
>with the code as it wad before your patch, I just returned it to how it was
>before which I think is correct
>
>Or.
