dhcpcd-discuss

On 01/12/2017 06:58 AM, Roy Marples wrote:
...
> Build dhcpcd with CFLAGS=-g and provide a backtrace would be nice.

I've never done a backtrace before, so I apologize if this isn't what you
asked for. This is from v6.11.5. I got it by typing the following:
./configure CFLAGS="-g"
make
gdb ./dhcpcd
run -B
bt

Program received signal SIGSEGV, Segmentation fault.
0x151384e7 in if_rtmsg (cmd=2 '\002', ifp=0xdfdfdfdf, addrs=4, flags=-538976289, priority=0, dst=0xcf7bd408,
    mask=0xcf7bd3f8, gate=0x0, src=0xcf7bd3d8, mtu=3755991007) at if-bsd.c:381
381             if ((cmd == RTM_ADD || cmd == RTM_DELETE || cmd == RTM_CHANGE) &&
(gdb) bt
#0  0x151384e7 in if_rtmsg (cmd=2 '\002', ifp=0xdfdfdfdf, addrs=4, flags=-538976289, priority=0, dst=0xcf7bd408,
    mask=0xcf7bd3f8, gate=0x0, src=0xcf7bd3d8, mtu=3755991007) at if-bsd.c:381
#1  0x151397a7 in if_route (cmd=2 '\002', rt=0x841a4fc0) at if-bsd.c:805
#2  0x15148c88 in nc_route (ort=0x841a4fc0, nrt=0x841a4000) at ipv4.c:501
#3  0x151497c9 in ipv4_doroute (rt=0x841a4000, nrs=0x7bf8f580) at ipv4.c:785
#4  0x15149a93 in ipv4_buildroutes (ctx=0xcf7bde7c) at ipv4.c:844
#5  0x1514a79d in ipv4_applyaddr (arg=0x823f0c00) at ipv4.c:1258
#6  0x15143172 in dhcp_bind (ifp=0x823f0c00) at dhcp.c:2248
#7  0x1514359b in dhcp_arp_bind (ifp=0x823f0c00) at dhcp.c:2363
#8  0x15145aa1 in dhcp_handledhcp (ifp=0x823f0c00, bootp=0xcf7bd7cc, bootp_len=323, from=0xcf7bd754) at dhcp.c:3082
#9  0x15145ef5 in dhcp_handlepacket (ifp=0x823f0c00, data=0xcf7bd7b0 "", len=351, flags=1) at dhcp.c:3204
#10 0x1514602a in dhcp_readpacket (arg=0x823f0c00) at dhcp.c:3229
#11 0x15128188 in eloop_start (eloop=0x844ac280, signals=0xcf7bdf0c) at eloop.c:935
#12 0x15126263 in main (argc=2, argv=0xcf7be084) at dhcpcd.c:1940

...
> dhcpcd-7 will be a lot less sensitive than dhcpcd-6 to varying BSD differences and hopefully changes.
> As you downloaded via the fossil timeline, can you test the latest trunk code as well? That would be useful in the interim.

The most recent source from the Trunk section does seem to compile and run
fine now on my OpenBSD 5.9 i386 computer.