dhcpcd-discuss

Am 16.10.19 um 08:08 schrieb Giorgio Dal Molin:
> > On October 15, 2019 at 4:29 PM Roy Marples <roy@xxxxxxxxxxxx> wrote:
> >
> >
> > On 15/10/2019 10:29, Christian Schneider wrote:
> > > Hi, yesterday I installed dhcpcd-8.1.0, and all dhcp request it does
> > > look like this:
> > >
> > > $ dhcpcd enp9s0
> > > DUID 00:04:03:1b:02:1c:04:0d:05:0e:f2:06:bf:07:00:08:00:09
> > > enp9s0: IAID 6b:98:89:55
> > > enp9s0: adding address fe80::9e58:3780:93c4:149b
> > > enp9s0: soliciting an IPv6 router
> > > enp9s0: soliciting a DHCP lease
> > > enp9s0: checksum failure from 192.168.10.2
> > > enp9s0: checksum failure from 192.168.10.2
> > > enp9s0: probing for an IPv4LL address
> > > enp9s0: using IPv4LL address 169.254.131.206
> > > enp9s0: adding route to 169.254.0.0/16
> > > enp9s0: adding default route
> > > forked to background, child pid 2327869
> > >
> > > what's going on? is this a bug? can I get more verbose output from dhcpcd?
> >
> > So far it could be a dhcpcd bug, it could be a compiler bug.
> > I've attached a patch to another email thread here today which might help.
> >
> > If not, try adding -fno-strict-aliasing to CFLAGS.
> >
> > Roy
>
> Hi Roy,
>
> yesterday, caught in my compiler tests rush, I've missed this mail from you.
>
> I tested it now and it works: compiling with gcc 9.2.0 and with
> CFLAGS="-O2 -march=native -fno-strict-aliasing" generates a dhcpcd executable
> without udp checksum failures.
>
> giorgio

Hi all, I tried with -fno-strict-aliasing, too, and then dhcpcd is fine.
Thx for the help.

If you want the patch to be tested, I can give it a try, if you want.

BR, Christian