dhcpcd-discuss

Thanks Roy,

On 3/8/2017 10:37 PM, Roy Marples wrote:
> Hi Shahid
>
> On 07/03/2017 20:57, Shahid Mahmood wrote:
> > Hi Roy,
> > There is a crash found during restart of dhcpcd. Brief bt:
> > ...
> >
> >     Core was generated by `dhcpcd -4 -f /var/run/dhcp/dhcpcd.conf -c
> >     /bin/dhcpcd-run-hooks --l'.
> >     Program terminated with signal SIGSEGV, Segmentation fault.
> >     ...
> >
> >     #0  0x3d737264 in ?? ()
> >     #1  0x0001e0d4 in arp_packet (len=<optimized out>, 
> > data=0xbee65938 "",
> >         ifp=<optimized out>) at arp.c:180
> >     #2  arp_read (arg=0xfdc108) at arp.c:206
>
> So arp.c:180 calls out to here:
> https://dev.marples.name/source/dhcpcd/browse/master/arp.c;69a91daff8a4defc6131522437af5645ddf8de18$180 
>
>
> astate->conflicted_cb(astate, &arm);
>
> Now, because the BT shows arp_read then there is debug info in the 
> dhcpcd binary, or it's not fully stripped.
> That means the function pointed to by astate->conflicted_cb should be 
> logged as well, but isn't.
The actual binary that crashed was stripped. gdb was later run using an 
unstripped binary that was rebuilt later (unfortunately, original 
unstripped binary was not available).
However, being in conflict state is a hint that I can use in the attempt 
to reproduce issue.
> So I think that's a corrupt stack somehow. Is this isolated to one or 
> many units? If one, maybe a memory test would be in order.
Unfortunately this was a one-time incident. I am trying to reproduce it 
but no luck so far. I'll setup valgrind.
I notice the dhcpcd command line shown in the core file is truncated 
(missing eth adaptor name as it was ought to be).
I recall seeing a crash in an older version when dhcpcd was launched 
without any parameters. But now, when I try that (ie. run dhcpcd without 
specifying the ethernet adaptor at command line), I get this:
        # dhcpcd
        dev: loaded udev
        no valid interfaces found
        no interfaces have a carrier
        forked to background, child pid 18565

Whereas, this works:

   # dhcpcd eth0
   all: IPv6 kernel autoconf disabled
   eth0: IPv6 kernel autoconf disabled
   DUID xxx
   eth0: IAID xxx
   eth0: adding address xxx
   if_addaddress6: Operation not supported
   eth0: rebinding lease of 10.138.179.11
   eth0: leased 10.138.179.11 for 30 seconds
   eth0: changing route to 10.138.179.0/27
   eth0: changing default route via 10.138.179.1
   forked to background, child pid 18614

Everything works in the later case. May not be related to the actual 
crash, but still unexpected. (The adaptor is POE, if that matters)
> Roy
-shahid