netdev - Re: [PATCH iproute2-next] tc/police: Allow 64 bit burst size

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

Message-ID: <CAM0EoM=Q6ewcUbdM_GahUmubxvQeJWAwxPu+3hmC2U1KjPb5_Q@mail.gmail.com>
Date: Fri, 12 Sep 2025 10:31:02 -0400
From: Jamal Hadi Salim <jhs@...atatu.com>
To: Jay Vosburgh <jay.vosburgh@...onical.com>, Victor Nogueira <victor@...atatu.com>
Cc: David Ahern <dsahern@...il.com>, netdev@...r.kernel.org, 
	Stephen Hemminger <stephen@...workplumber.org>
Subject: Re: [PATCH iproute2-next] tc/police: Allow 64 bit burst size

On Fri, Sep 12, 2025 at 12:19 AM Jay Vosburgh
<jay.vosburgh@...onical.com> wrote:
>
> David Ahern <dsahern@...il.com> wrote:
>
> >On 9/9/25 9:32 PM, Jamal Hadi Salim wrote:
> >>
> >> Please run tdc tests. David/Stephen - can we please make this a
> >> requirement for iproute2 tc related changes?
> >
> >I will try to remember to run tdc tests for tc patches. Without an
> >automated setup, there will be misses over time.
> >
> >>
> >> Jay, your patches fail at least one test because you changed the unit outputs.
> >> Either we fix the tdc test or you make your changes backward compatible.
> >> In the future also cc kernel tc maintainers (I only saw this because
> >> someone pointed it to me).
> >> Overall the changes look fine.
> >
> >Sent a patch to add a tc entry to iproute2 maintainers file.
> >
> >You say the change looks fine but at least one test fails meaning
> >changes are requested?
>
>         Yes, I ran the tests and saw one failure, in the following:
>
>         "cmdUnderTest": "$TC actions add action police pkts_rate 1000 pkts_burst
>  200 index 1",
>         "expExitCode": "0",
>         "verifyCmd": "$TC actions ls action police",
>         "matchPattern": "action order [0-9]*:  police 0x1 rate 0bit burst 0b mtu 4096Mb pkts_rate 1000 pkts_burst 200",
>
>         Which is trying to match a returned mtu value of "4096Mb" but
> the new code prints "4Gb"; should be straightforward to change the test
> to accept either returned value.
>
>         Or I can take out the bit that prints sufficiently large values
> in units of Gb, if you've got a preference for leaving that part alone.
> Doing so would ease the lockstep problem between the tests in the kernel
> tree and the change in iproute2.  The numeric formatting isn't the
> important part of the patch set, so I'm ok either way.
>

For backward compat we need to support both. IOW, if someone was using
an older tc then a new kernel should work fine and not fail because of
different output expected. @Victor Nogueira wanna take a crack at
fixing the test? Then when the iproute side is merged as is - both
should work.

cheers,
jamal
>         -J
>
> ---
>         -Jay Vosburgh, jay.vosburgh@...onical.com