netdev - Re: [PATCH bpf-next 2/2] bpftool: support loading flow dissector

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: Windows password security audit tool. GUI, reports in PDF.

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

Message-ID: <20181107153028.43506cc5@cakuba.netronome.com>
Date:   Wed, 7 Nov 2018 15:30:28 -0800
From:   Jakub Kicinski <jakub.kicinski@...ronome.com>
To:     Stanislav Fomichev <sdf@...ichev.me>
Cc:     Quentin Monnet <quentin.monnet@...ronome.com>,
        Stanislav Fomichev <sdf@...gle.com>, netdev@...r.kernel.org,
        linux-kselftest@...r.kernel.org, ast@...nel.org,
        daniel@...earbox.net, shuah@...nel.org, guro@...com,
        jiong.wang@...ronome.com, bhole_prashant_q7@....ntt.co.jp,
        john.fastabend@...il.com, jbenc@...hat.com,
        treeze.taeung@...il.com, yhs@...com, osk@...com,
        sandipan@...ux.vnet.ibm.com
Subject: Re: [PATCH bpf-next 2/2] bpftool: support loading flow dissector

On Wed, 7 Nov 2018 15:12:07 -0800, Stanislav Fomichev wrote:
> > > I agree, constructing the jmp_table is a bit fragile with all the
> > > dependencies on the order of the progs. I'll drop that and will send a
> > > v2 that pins all the programs from the obj file instead and offloads
> > > jmp_table construction to the user. So the supposed use case would be
> > > something like the following:
> > > 
> > > bpftool prog load bpf_flow.o /sys/fs/bpf/flow type flow_dissector  
> > 
> > Okay.  One more thing - how do we differentiate between mass pin and the
> > existing pin first behaviour?  Should we perhaps add a loadall command
> > or some flag?  
> In v2 I did by program type:
> * flow_dissector -> pin all
> * not flow_dissector -> pin first?
> 
> But we can have loadall or something like:
> load OBJ [pinfirst|pinall] FILE|DIR [type TYPE]
> 
> If we want to add user control, I'd go with loadall command,
> adding more optional flags in between is a mess..

I think user control would be good.  Agreed on loadall being better.