dhcpcd-discuss

Hi Ross

On 02/11/2019 20:08, Ross North wrote:
> I am experiencing a seg fault with rb_tree_insert_node ever since 
> upgrading to dhcpcd version 8 and up. Version 7 works without issue.
> This only occurs when the /-b flag/ is used and the /routers=/ option is 
> set in the config to assign a static address.
> Config file (default file plus the following appended):
> static ip_address=192.168.0.150
> static routers=192.168.0.1
> static domain_name_servers=1.1.1.1 1.0.0.1
> Config file (no crash):
> static ip_address=192.168.0.150
> static domain_name_servers=1.1.1.1 1.0.0.1
> (Starting without -b flag)
> [user@srv1 ~]$ sudo pkill dhcpcd
> [user@srv1 ~]$ sudo lldb dhcpcd --
> (lldb) target create "dhcpcd"
> Current executable set to 'dhcpcd' (x86_64).
> (lldb) run
> Process 1046 launched: '/usr/bin/dhcpcd' (x86_64)
> dev: loaded udev
> Process 1046 stopped and restarted: thread 1 received signal: SIGCHLD
> Process 1046 stopped and restarted: thread 1 received signal: SIGCHLD
> DUID 00:04:00:00:00:00:00:00:00:00:00:00:30:9c:23:21:92:bb
> enp2s0: IAID 23:21:92:bb
> enp2s0: using static address 192.168.0.140/24
> enp2s0: adding route to 192.168.0.0/24
> enp2s0: adding default route via 192.168.0.1
> Process 1046 stopped and restarted: thread 1 received signal: SIGCHLD
> forked to background, child pid 1062
> Process 1046 exited with status = 0 (0x00000000)
> (lldb) exit
> (Starting with a -b flag - invalid address)
> [user@srv1 ~]$ sudo pkill dhcpcd
> [user@srv1 ~]$ sudo lldb dhcpcd -- -b
> (lldb) target create "dhcpcd"
> Current executable set to 'dhcpcd' (x86_64).
> (lldb) settings set -- target.run-args  "-b"
> (lldb) run
> Process 1081 launched: '/usr/bin/dhcpcd' (x86_64)
> dev: loaded udev
> forked to background, child pid 1084
> Process 1081 stopped
> * thread #1, name = 'dhcpcd', stop reason = signal SIGSEGV: invalid 
> address (fault address: 0x10)
>      frame #0: 0x000055555558f6dd dhcpcd`rb_tree_insert_node + 29
> dhcpcd`rb_tree_insert_node:
> ->  0x55555558f6dd <+29>: movq   0x10(%r15), %rax
>      0x55555558f6e1 <+33>: movq   (%r15), %rbx
>      0x55555558f6e4 <+36>: leaq   (%rsi,%rax), %rbp
>      0x55555558f6e8 <+40>: testq  %r14, %r14
> (lldb) bt
> * thread #1, name = 'dhcpcd', stop reason = signal SIGSEGV: invalid 
> address (fault address: 0x10)
>    * frame #0: 0x000055555558f6dd dhcpcd`rb_tree_insert_node + 29
>      frame #1: 0x000055555556c54a dhcpcd`rt_headclear0.part.0 + 106
>      frame #2: 0x000055555556acf6 dhcpcd`free_options + 246
>      frame #3: 0x000055555555e93e dhcpcd`main + 2286
>      frame #4: 0x00007ffff7e10153 libc.so.6`__libc_start_main + 243
>      frame #5: 0x000055555555f47e dhcpcd`_start + 46
> (lldb)
> In both scenarios (non-crash and crash) the master dhcpcd process seems 
> to have been started before crashing/exiting.
> I am using the latest Arch Linux build 
> (https://www.archlinux.org/packages/core/x86_64/dhcpcd/) and seems to be 
> an issue for a few users.
> I saw a similar thread but that patch did not fix the crash for me.
> Any ideas how to fix this? Thanks.

Fixed here:
https://roy.marples.name/cgit/dhcpcd.git/commit/?id=9ffb2a62630662bf060769896230f38ad7a98390

Roy