[Solved] conky 1.10.2-1 segfaults (fixed in 1.10.3)

root · 2016-05-04 01:13:57

After upgrading conky from 1.10.1-2 to 1.10.2-1 it didn't start when running it.
In order to see if there was any additional information on the reason for this behaviour I ran it in a terminal and saw that it ran normally and was forked to the background.
The strange thing was that after the forking line there was an additional blank one.

As an additional test I moved my .conkyrc file to a backup one and tried to run it with its built-in config. This time after

conky: drawing to created window (0x1a00001)

I am presented with a segfault message. Checking the journal it reads as follows,

kernel: conky[9997]: segfault at 0 ip b6fa6530 sp b2327054 error 4 in libc-2.23.so[b6f30000+1ae000]

As the segfault notification appears instead of the one related to double buffer

conky: drawing to double buffer

I wondered if therefore the segfault may be related to this setting in the config file. Thus I changed it to 'false' and conky reloaded itself. I changed it back to 'true' and, unexpectedly, again conky showed up.
It wasn't until I used pkill that conky did not show up (with a punctual exception and equally fast disappearance for unknown reasons) when run with either setting value.
Also, now when run in a terminal the empty line instead of at the end of conky's loading process (after forking to the background) appears under the terminal prompt.
The segfaults message is the same in the journal though (with variation in the final code part).

Once I downgrade the package to its previous version, conky performs as usual.

Anyone else is experiencing this?

Last edited by root (2016-07-09 02:59:31)

Leonardo19 · 2016-05-04 08:44:30

The same result by launching conky or conky-git from AUR as ordinary user.
Xfce4, i686, Intel integrated graphics.

Last edited by Leonardo19 (2016-05-04 09:45:54)

perseus · 2016-05-04 21:28:00

Confirmed on my 64 bit laptop. I also have this desktop still running 32 bit (using a very similar config), which I have just updated and which does not segfault.

root · 2016-05-05 01:02:56

perseus wrote:

Confirmed on my 64 bit laptop. I also have this desktop still running 32 bit (using a very similar config), which I have just updated and which does not segfault.

Did you restart conky after updating it on the 32bit system?

If you did, running it in a terminal performs as usual or does it present the symptons I described?

perseus · 2016-05-05 07:34:52

Yes, I killed the running instance and restarted from a terminal. I see the usual:

conky: desktop window (25f) is root window
conky: window type - normal
conky: drawing to created window (0x400001)
conky: drawing to double buffer
conky: forked to background, pid is 21222

I just did it again to confirm. I have no core dumps or other indications of error in my logs.

I have not rebooted the machine itself yet. It needs to run 24/7 and this is not the moment to do it.

By the way, I am subscribed to this topic, but did not receive an e-mail re your post. I tried e-mailing myself from here and that is not working either. Any furtther resposnses may be slow due to this.

jasonwryan · 2016-05-05 07:38:22

perseus wrote:

By the way, I am subscribed to this topic, but did not receive an e-mail re your post. I tried e-mailing myself from here and that is not working either. Any furtther resposnses may be slow due to this.

You won't receive any mails from us: Yahoo doesn't deliver our mail

perseus · 2016-05-05 07:49:57

Many thanks jasonwryan It is a little while since I posted. Never had that issue before. Dummies. I will change e-mail addy later.

seeder · 2016-05-07 14:45:02

I can confirm this on i686 machine

[ 288.640499] conky[614]: segfault at 704b ip b6f4ebec sp b4bbe100 error 4 in libc-2.23.so[b6eea000+1ae000]

Mr Green · 2016-05-07 15:10:17

Same here on a x86_64 system... core dumps version 1.10-2

grepfor · 2016-05-07 16:38:35

Same here, on i686, Xorg 1.18.3, kernel 4.5.1-1-ARCH #1 SMP PREEMPT:

May 07 10:14:17 nu kernel: conky[604]: segfault at 545a ip b6ea2bec sp b3a3e100 error 4 in libc-2.23.so[b6e3e000+1ae000]
May 07 10:14:17 nu systemd[1]: Started Process Core Dump (PID 607/UID 0).
May 07 10:14:26 nu systemd-coredump[608]: Process 598 (conky) of user 501 dumped core.

       Stack trace of thread 604: 
       #0  0x00000000b6ea2bec _IO_feof (libc.so.6)
       #1  0x00000000080b8629 _Z11update_statv (conky)
       #2  0x00000000080b8998 _Z16update_cpu_usagev (conky)
       #3  0x00000000080b0795 _ZN5conky4priv13callback_base13start_routineEv (conky)
       #4  0x00000000b70b153e execute_native_thread_routine_compat (libstdc++.so.6)
       #5  0x00000000b76a12f1 start_thread (libpthread.so.0)
       #6  0x00000000b6f24c7e __clone (libc.so.6)

       Stack trace of thread 598: 
       #0  0x00000000b76dcd8d __kernel_vsyscall (linux-gate.so.1)
       #1  0x00000000b76a9186 do_futex_wait.constprop.2 (libpthread.so.0)
       #2  0x00000000b76a9277 __new_sem_wait_slow.constprop.1 (libpthread.so.0)
       #3  0x00000000080b0ffb _ZN9semaphore4waitEv (conky)
       #4  0x00000000080731ce _Z12update_stuffv (conky)
       #5  0x000000000807affb generate_text (conky)
       #6  0x000000000807c032 main_loop (conky)
       #7  0x000000000806c004 main (conky)
       #8  0x00000000b6e56536 __libc_start_main (libc.so.6)
       #9  0x000000000806ff39 _start (conky)

root: Are you going to file a bug report on this? If not, just let me know, I will do so. Thx.

Mr Green · 2016-05-08 06:08:54

Shot in the dark here, could it be that with conky changing to Luasyntax old config files would not work correctly?

grepfor · 2016-05-08 07:09:31

I see the same segfault behavior even with no .conkyrc present, so guessing it is not a config file issue.

Btw, what window mgr you guys using when you see this segfault? Here, fvwm 2.6.5.

Last edited by grepfor (2016-05-08 07:31:54)

berbae · 2016-05-08 08:29:36

conky 1.10.2-1 doesn't segfault here under openbox wm.
It just outputs these lines when starting up:

conky: desktop window (d4) is root window
conky: window type - desktop
conky: drawing to created window (0x600001)
conky: drawing to double buffer
conky: forked to background, pid is 5226

It is started with just
conky
in the ~/.config/openbox/autostart

perseus · 2016-05-08 08:45:22

grepfor wrote:

... what window mgr you guys using when you see this segfault? Here, fvwm 2.6.5.

Current FluxBox on both of my machines.

V1del · 2016-05-08 11:25:53

Posted stacktraces contain mention of futex locks, are any of you on a pre-Skylake intel i-series cpu, and if so, have you done the Microcode update?

grepfor · 2016-05-08 14:35:39

V1del wrote:

Posted stacktraces contain mention of futex locks, are any of you on a pre-Skylake intel i-series cpu, and if so, have you done the Microcode update?

Good suggestion, thanks. Alas, afaict from the instructions link you provided above, microcode on my setup is the latest:
CPU signature 0x000006d8 and microcode version 0x20. Nothing listed under "selected microcodes" using iucode_tool.

Last edited by grepfor (2016-05-08 14:37:24)

henk · 2016-05-09 15:59:34

Updated today, was curious how conky would behave, but it started up normally.
FWIW, I start conky delayed:

sh -c 'sleep 12;sh -c "conky"'

part of my config:

conky.config = {
        -- Use double buffering (eliminates flickering)
	double_buffer = true,

	-- Run conky in the background
	background = true,

        -- Use Xft (anti-aliased font and stuff)
        use_xft = true,
        font = 'Sans:size= 8',
        xftalpha = 1,
        uppercase = false,
        override_utf8_locale = false,

	-- Prevent text from moving around while using a mono font
	use_spacer = 'left',

        -- Update interval in seconds
        update_interval = 1.0,

	-- Set to zero to run forever
	total_run_times = 0,

	-- Subtract file system buffers from used memory
	no_buffers = true,

	-- Number of samples to take for CPU and network readings
	cpu_avg_samples = 2,

	-- Makes conky window transparent
	own_window = true,
	own_window_class = 'Conky',
        own_window_type = 'desktop',
	own_window_transparent = true,
	own_window_hints = 'undecorated,below,skip_taskbar,sticky,skip_pager',

	-- Window size and position
        minimum_width = 300,
        minimum_height = 970,
        maximum_width = 300,
        alignment = 'top_right',
        gap_x = 20,
        gap_y = 80,

	-- Default color and border settings
        draw_shades = true,
        draw_outline = false,
        draw_borders = false,
        draw_graph_borders = false,
        default_color = 'white',
        default_shade_color = 'black',
        default_outline_color = 'white'
}

Using XFCE BTW

Last edited by henk (2016-05-09 16:03:34)

grepfor · 2016-05-09 20:51:48

Config file bisection seems to indicate that the $cpu variable is implicated in the segfault, at least on my setup.

Can one of you guys who is also experiencing this issue try to verify this?

This minimal config file behaves as expected:

   alignment top_right
   TEXT
   FOO
   Without cpu variable: 
   BAR

This one segfaults:

   alignment top_right
   TEXT
   FOO
   With cpu variable: $cpu
   BAR

iishikawa · 2016-05-10 15:15:26

@grepfor
I'm facing the same issue.
The code you posted made the same situation happen.
I think $cpu (or $cpubar) variable is related to this segfault.

grepfor · 2016-05-10 15:27:20

Ok thanks @iishikawa.

I'll file an upstream bug report sometime in the next few days, busy now. If someone else has the time, go for it.

Mr Green · 2016-05-12 06:24:58

Can confirm that cpubar causes issue, thanks grepfor

Tsu Jan · 2016-05-12 18:22:13

Yesterday Conky was upgraded to 1.10.2-1 in Debian too. No segfault here but in so many years of using Conky, I hadn't seen such a thing: two of the cpu bars created very high graphs that covered the whole height of Conky's window. So, I confirm that cpubar has a bug. I downgraded to 1.10.1 for now.

DanielH · 2016-05-13 02:59:42

I can confirm conky 1.10.2-1 segfault on old single core Intel Celeron using cpu* variables, view https://bbs.archlinux.org/viewtopic.php?id=212506
There is not segfaults using ${cpu cpuN} and ${cpubar cpuN} on Core2 Duo T8100 and Atom N450 machines.
No clues...

grepfor · 2016-05-14 14:52:58

Looks like this just got reported to the devs in the last day or so. There's an associated pull request
so the issue will probably be fixed next release.

https://github.com/brndnmtthws/conky/pull/257

Thx to all for input.

Arch Linux

#1 2016-05-04 01:13:57

[Solved] conky 1.10.2-1 segfaults (fixed in 1.10.3)

#2 2016-05-04 08:44:30

Re: [Solved] conky 1.10.2-1 segfaults (fixed in 1.10.3)

#3 2016-05-04 21:28:00

Re: [Solved] conky 1.10.2-1 segfaults (fixed in 1.10.3)

#4 2016-05-05 01:02:56

Re: [Solved] conky 1.10.2-1 segfaults (fixed in 1.10.3)

#5 2016-05-05 07:34:52

Re: [Solved] conky 1.10.2-1 segfaults (fixed in 1.10.3)

#6 2016-05-05 07:38:22

Re: [Solved] conky 1.10.2-1 segfaults (fixed in 1.10.3)

#7 2016-05-05 07:49:57

Re: [Solved] conky 1.10.2-1 segfaults (fixed in 1.10.3)

#8 2016-05-07 14:45:02

Re: [Solved] conky 1.10.2-1 segfaults (fixed in 1.10.3)

#9 2016-05-07 15:10:17

Re: [Solved] conky 1.10.2-1 segfaults (fixed in 1.10.3)

#10 2016-05-07 16:38:35

Re: [Solved] conky 1.10.2-1 segfaults (fixed in 1.10.3)

#11 2016-05-08 06:08:54

Re: [Solved] conky 1.10.2-1 segfaults (fixed in 1.10.3)

#12 2016-05-08 07:09:31

Re: [Solved] conky 1.10.2-1 segfaults (fixed in 1.10.3)

#13 2016-05-08 08:29:36

Re: [Solved] conky 1.10.2-1 segfaults (fixed in 1.10.3)

#14 2016-05-08 08:45:22

Re: [Solved] conky 1.10.2-1 segfaults (fixed in 1.10.3)

#15 2016-05-08 11:25:53

Re: [Solved] conky 1.10.2-1 segfaults (fixed in 1.10.3)

#16 2016-05-08 14:35:39

Re: [Solved] conky 1.10.2-1 segfaults (fixed in 1.10.3)

#17 2016-05-09 15:59:34

Re: [Solved] conky 1.10.2-1 segfaults (fixed in 1.10.3)

#18 2016-05-09 20:51:48

Re: [Solved] conky 1.10.2-1 segfaults (fixed in 1.10.3)

#19 2016-05-10 15:15:26

Re: [Solved] conky 1.10.2-1 segfaults (fixed in 1.10.3)

#20 2016-05-10 15:27:20

Re: [Solved] conky 1.10.2-1 segfaults (fixed in 1.10.3)

#21 2016-05-12 06:24:58

Re: [Solved] conky 1.10.2-1 segfaults (fixed in 1.10.3)

#22 2016-05-12 18:22:13

Re: [Solved] conky 1.10.2-1 segfaults (fixed in 1.10.3)

#23 2016-05-13 02:59:42

Re: [Solved] conky 1.10.2-1 segfaults (fixed in 1.10.3)

#24 2016-05-14 14:52:58

Re: [Solved] conky 1.10.2-1 segfaults (fixed in 1.10.3)

Board footer