Re: bug in AF_XDP socket cleanup?

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



Hi Július,

The ring clean up routines in i40e and ixgbe are asynchronous, so they
will likely not be ready when you return from the xsk_socket__delete()
call. What happens if you add a sleep(1) before you create it a second
time?

/Magnus

On Wed, Sep 25, 2019 at 5:47 PM Július Milan <Julius.Milan@xxxxxxxxxxxxx> wrote:
>
> Hi folks
>
>
>
> I am stuck with proper cleanup of AF_XDP socket. Not sure if I am doing something wrong or is it a bug.
>
> I create xdp socket, then I remove it and then I am trying to create it again and got EBUSY from bind syscall
>
> of xsk_socket__create.
>
>
>
> My cleanup looks like this:
>
> xsk_socket__delete();
>
> xsk_umem__delete();
>
> bpf_set_link_xdp_fd(ifindex, -1, opt_xdp_flags);
>
>
>
> Attaching simple test case. It’s just reduced and modified xdpsock_user.c form kernel samples.
>
> Tested on kernel v5.3 4d856f72c10ecb060868ed10ff1b1453943fc6c8 and related libbpf.
>
>
>
> Thanks a lot
>
> Julius




[Index of Archives]     [Linux Networking Development]     [Fedora Linux Users]     [Linux SCTP]     [DCCP]     [Gimp]     [Yosemite Campsites]

  Powered by Linux