netdev - Re: [PATCH 1/2] tools: bpftool: add net load command to load XDP on interface

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite for Android: free password hash cracker in your pocket

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <CAEKGpziq3fGWHxibHSkM48sR=6A-_oBJkiy3kKRQq4m-6s-2Mg@mail.gmail.com>
Date:   Thu, 1 Aug 2019 03:23:40 +0900
From:   "Daniel T. Lee" <danieltimlee@...il.com>
To:     Jesper Dangaard Brouer <brouer@...hat.com>
Cc:     Daniel Borkmann <daniel@...earbox.net>,
        Alexei Starovoitov <ast@...nel.org>,
        netdev <netdev@...r.kernel.org>,
        Quentin Monnet <quentin.monnet@...ronome.com>
Subject: Re: [PATCH 1/2] tools: bpftool: add net load command to load XDP on interface

On Wed, Jul 31, 2019 at 7:08 PM Jesper Dangaard Brouer
<brouer@...hat.com> wrote:
>
> On Wed, 31 Jul 2019 03:48:20 +0900
> "Daniel T. Lee" <danieltimlee@...il.com> wrote:
>
> > By this commit, using `bpftool net load`, user can load XDP prog on
> > interface. New type of enum 'net_load_type' has been made, as stated at
> > cover-letter, the meaning of 'load' is, prog will be loaded on interface.
>
> Why the keyword "load" ?
> Why not "attach" (and "detach")?
>
> For BPF there is a clear distinction between the "load" and "attach"
> steps.  I know this is under subcommand "net", but to follow the
> conversion of other subcommands e.g. "prog" there are both "load" and
> "attach" commands.
>
>
> > BPF prog will be loaded through libbpf 'bpf_set_link_xdp_fd'.
>
> Again this is a "set" operation, not a "load" operation.

>From earlier at cover-letter, I thought using the same word 'load' might give
confusion since XDP program is not considered as 'bpf_attach_type' and can't
be attached with 'BPF_PROG_ATTACH'.

But, according to the feedback from you and Andrii Nakryiko, replacing
the word 'load' as 'attach' would be more clear and more consistent.

> > Signed-off-by: Daniel T. Lee <danieltimlee@...il.com>
>
> [...]
> >  static int do_show(int argc, char **argv)
> >  {
> >       struct bpf_attach_info attach_info = {};
> > @@ -305,13 +405,17 @@ static int do_help(int argc, char **argv)
> >
> >       fprintf(stderr,
> >               "Usage: %s %s { show | list } [dev <devname>]\n"
> > +             "       %s %s load PROG LOAD_TYPE <devname>\n"
>
> The "PROG" here does it correspond to the 'bpftool prog' syntax?:
>
>  PROG := { id PROG_ID | pinned FILE | tag PROG_TAG }
>

Yes. By using the same 'prog_parse_fd' from 'bpftool prog',
user can 'attach' XDP prog with id, pinned file or tag.

> >               "       %s %s help\n"
> > +             "\n"
> > +             "       " HELP_SPEC_PROGRAM "\n"
> > +             "       LOAD_TYPE := { xdp | xdpgeneric | xdpdrv | xdpoffload }\n"
> >               "Note: Only xdp and tc attachments are supported now.\n"
> >               "      For progs attached to cgroups, use \"bpftool cgroup\"\n"
> >               "      to dump program attachments. For program types\n"
> >               "      sk_{filter,skb,msg,reuseport} and lwt/seg6, please\n"
> >               "      consult iproute2.\n",
>
>
> --
> Best regards,
>   Jesper Dangaard Brouer
>   MSc.CS, Principal Kernel Engineer at Red Hat
>   LinkedIn: http://www.linkedin.com/in/brouer

And about the enum 'NET_LOAD_TYPE_XDP_DRIVE',
'DRIVER' looks more clear to understand.

Will change to it right away.

Thanks for the review.