dhcpcd-discuss

Hi thank you for the answer.

After some experiments I discovered that this is not an issue, you are
right.
Probably it's a bug of Conformance test 1.5.0, because with older version
of the certification tool it's working as expected.
I'll contact Apple to explain the problem.

> You should be able to see this as well if you add the word debug to
> /etc/dhcpcd.conf
Interesting, I didn't know this.

About upgrading to 7.x.x, I'm having an issue while building:
*cc1: error: include location "/usr/include" is unsafe for
cross-compilation [-Werror=poison-system-directories]*
*cc1: all warnings being treated as errors*

I saw that the main difference in configure file from 6.x.x is the include
part, so I suppose that this is the problem.

Il giorno gio 10 gen 2019 alle ore 20:30 Roy Marples <roy@xxxxxxxxxxxx> ha
scritto:

> Hi Steffano
>
> Sorry for the late reply, I struggled to get a new Linux VM working.
>
> On 09/01/2019 10:50, Stefano Cappa wrote:
> > Hi everyone, this is my first mail in this mailing list.
> >
> > I'm using dhcpcd 6.11.5 on an embedded device with yocto thud (Linux 4.9
> > based).
> > I'm using the official porting of dhcpcd in open-embedded, but probably
> > this is not relevant.
> >
> > I'm trying to run Apple bonjour conformance test 1.5.0 released at the
> > beginning of December. Everything is perfect with ipv6, but when I try
> > with IPv4 I get en error about arp announcements.
> >
> > Conformance test runs this test (as defined by Apple): "The operator
> > disconnects and reconnects the device's network cable. The test tool
> > verifies that the device repeats its address probing to verify that its
> > address is still unique. The test tool denies the first probe and
> > verifies that the device picks a new address and probes/announces again.
> > It is a warning if the device does not use its original address as its
> > first probing candidate."
> >
> > But it throws this message:" arp src dst address mismatch: Did not
> > receive 2 announcements after 3 unanswered probes" because it is an arp
> > packet with target address equals to the DNS IPv4 address (because I'm
> > running also mdns).
> >
> > As I know this apr behaviour is defined in rfc 5227, but it seems that
> > dhcpcd is not following the specification about 3 probes and 2 arp
> > announcements.
> >
> > I'm not an expert about networking in Linux, so I could saying stupid
> > things.
> >
> > Is this an issue of dhcpcd 6.11.5?
>
> Not that I know of.
> I just spun up a new Arch Linux VM, and tried both dhcpcd-6.11.5 and the
> latest 7.0.8 release.
> I simulated the removal and insertion of the cable by doing:
>         ip link set down dev ens4
>         ip link set up dev ens4
>
> dhcpcd then did 3 ARP probes followed by 2 ARP announcements which I
> verified with wireshark.
> You should be able to see this as well if you add the word debug to
> /etc/dhcpcd.conf
>
> > How can I fix it to pass
> > certification?
>
> I have no idea as from my perspective it works fine.
>
> > Dhcpcd 7 fixes this problem (I cannot try it quickly
> > because the porting is still not available and it isn't so simple).
> Well, if dhcpcd-7 works for you then use dhcpcd-7?
> Porting is actually very easy, why do you think it's not so simple?
>
> Roy
>