System hangs on cold boot

[axelgenus]

Hi guys, 

I installed Funtoo on my home machine a couple of months ago. It runs smoothly but I have a really annoying problem: the computer hangs completely (no keyboard, no remote access through SSH...) when it cold boots. It may take up to three consecutive resets to make it work. The kernel seems to boot fine. When the booting procedure reaches OpenRC the system sits at "waiting for uevents to be completed" for a very long time and eventually hangs there forever. Sometimes it manages to go on after 20-30 seconds but as soon as XDM starts the computer hangs for good. When the booting procedure completes (it reaches the login screen) I can reboot the computer dozen of times and it keeps working flawlessy: no hangs, no lags, nothing... 

 

The most annoying part is that I can't manage to get a log of what's going on. I use syslog-ng but nothing seems to be breaking the system. I compiled the kernel myself and built-in all the firmware needed (AMD ucode, NIC patch, video card firmware). I don't use INITRD. 

 

Here my specs: 

 

ASUS SABERTOOTH 990FX/R2.0 

AMD FX-6350 Black Edition 

Kingston 4x4GB DDR3 1333MHz 

Kingston SSDNow! V300 128GB 

Sapphire AMD Radeon R9 280X 

 

Can you please help me? I'm out of ideas...   :(

 

cat /proc/cpuinfo -> http://pastebin.com/b89PV9Uf

dmesg -> http://pastebin.com/NtRHwm5s

zcat /proc/config.gz -> http://pastebin.com/SSrqNfeV

lsmod -> http://pastebin.com/R0NiQr0F

lspci -> http://pastebin.com/JDz3EUUQ

lsusb -> http://pastebin.com/cEayLFjf

 

PS: I am running Funtoo current/pure64/piledriver.

[Oleg Vinichenko]

some pastes are removed.

[axelgenus]

some pastes are removed.

 

Ok, now they should be fixed. ;)

[Oleg Vinichenko]

see no suspicious things in dmesg. Please, add /var/log/mesages. Enable logging of openrc services start-up. It is in /etc/rc.conf, uncomment logging directive. Reboot, and provide logs. Was it a permanent trouble or started at some point?

[axelgenus]

Was it a permanent trouble or started at some point?

 

Well, actually it started on Windows: I had a dual boot system (Windows 8.1/Archlinux) and sometimes it hung at Windows login screen but since Linux worked fine I did not take care of it (I used Linux 90% of the time). Eventually I got bored by the problem because it started to happen on every cold boot and also on Linux. I then replaced the configuration with the current one (only Funtoo) but it did not solved the problem. I loved Gentoo and waited far too long to try Funtoo. ;)

 

I will provide the other info you asked for ASAP. In the meanwhile I changed my kernel configuration in order to enable the SysRq magic key and I compiled/configured netconsole. Now if the system hangs again I can retrive the dmesg from my seedbox (hopefully it will contain some useful information about the problem). ;)

[Oleg Vinichenko]

ok, it's maybe needed to run S.M.A.R.T tests for your drives

[axelgenus]

Here it goes:

 

rc.log -> http://pastebin.com/vtDd706Q

smartctl -i /dev/sda -> http://pastebin.com/tW25fTH3

smartctl -H /dev/sda -> http://pastebin.com/CtVb2STU

[Oleg Vinichenko]

smartctl is not helpful. Please, run test check and collect a test data.

[axelgenus]

Maybe I get what the fuss is about. In the dmesg I found a line saying:

fb: switching to radeondrmfb from EFI VGA

It seemed redundant to me to have two different framebuffer drivers just to switch from one another in about second... so I disabled EFIfb and no hangs occurred since two days ago.

 

The thing I noticed after this change in the kernel configuration is that the display goes offline (sleep) as soon as GDM starts but I can live with it.
 

I'm going to follow your advise though but what is the command to gather those SMART logs? I know about smartctl... is there another utility?

 

PS: Now that I think about it, Windows 7 and above use EFI framebuffer (if installed in UEFI mode) to show the Windows logo while loading the OS and then it activates the AMD driver... I guess it is a VGA BIOS bug but I'm not sure.

[Oleg Vinichenko]

smartctl -S will run short test

[axelgenus]

My system disk is a SSD. I started the test but when it finished it I was not able to retrive the log... I tried with "-l error", "-l ssd" and "-l selftest" but these are the results:

 

# smartctl -l selftest /dev/sda
smartctl 6.4 2015-06-04 r4109 [x86_64-linux-4.1.4-gentoo-debug] (local build)
Copyright (C) 2002-15, Bruce Allen, Christian Franke, www.smartmontools.org


=== START OF READ SMART DATA SECTION ===
SMART Self-test Log not supported

# smartctl -l ssd /dev/sda
smartctl 6.4 2015-06-04 r4109 [x86_64-linux-4.1.4-gentoo-debug] (local build)
Copyright (C) 2002-15, Bruce Allen, Christian Franke, www.smartmontools.org


Device Statistics Log has only 0x01 pages

# smartctl -l error /dev/sda
smartctl 6.4 2015-06-04 r4109 [x86_64-linux-4.1.4-gentoo-debug] (local build)
Copyright (C) 2002-15, Bruce Allen, Christian Franke, www.smartmontools.org


=== START OF READ SMART DATA SECTION ===
SMART Error Log not supported

 

This is the complete output with "-a" argument:

 

smartctl -a /dev/sda -> http://pastebin.com/Wr3EWwA6

[axelgenus]

After some investigation (and a full system installation) I finally noticed something strange in the logs:

Aug 16 20:53:54 [sensord] sensord started
Aug 16 20:53:54 [sensord] Sensor alarm: Chip it8721-isa-0290: in0: +2.86 V (min = +2.74 V, max = +3.02 V) [ALARM]
Aug 16 20:53:54 [sensord] Sensor alarm: Chip it8721-isa-0290: in2: +0.83 V (min = +0.80 V, max = +1.60 V) [ALARM]
Aug 16 20:53:54 [sensord] Sensor alarm: Chip it8721-isa-0290: +3.3V: +3.34 V (min = +3.19 V, max = +3.60 V) [ALARM]
Aug 16 20:53:54 [sensord] Sensor alarm: Chip it8721-isa-0290: in4: +0.17 V (min = +3.01 V, max = +1.06 V) [ALARM]
Aug 16 20:53:54 [sensord] Sensor alarm: Chip it8721-isa-0290: in6: +1.78 V (min = +2.29 V, max = +2.98 V) [ALARM]
Aug 16 20:53:54 [sensord] Sensor alarm: Chip it8721-isa-0290: 3VSB: +0.02 V (min = +2.21 V, max = +2.69 V) [ALARM]
Aug 16 20:53:54 [sensord] Sensor alarm: Chip it8721-isa-0290: fan3: 489 RPM (min = 600 RPM) [ALARM]
Aug 16 20:53:54 [sensord] Sensor alarm: Chip it8721-isa-0290: temp1: 29.0 C (min = 0.0 C, max = 70.0 C) [ALARM]
Aug 16 20:53:54 [sensord] Sensor alarm: Chip it8721-isa-0290: temp2: 27.0 C (min = 0.0 C, max = 60.0 C) [ALARM]
Aug 16 20:53:54 [sensord] Chip: radeon-pci-0100
Aug 16 20:53:54 [sensord] Adapter: PCI adapter
Aug 16 20:53:54 [sensord] temp1: 35.0 C
Aug 16 20:53:54 [sensord] Chip: fam15h_power-pci-00c4
Aug 16 20:53:54 [sensord] Adapter: PCI adapter
Aug 16 20:53:54 [sensord] Chip: it8721-isa-0290
Aug 16 20:53:54 [sensord] Adapter: ISA adapter
Aug 16 20:53:54 [sensord] in0: +2.86 V (min = +2.74 V, max = +3.02 V) [ALARM]
Aug 16 20:53:54 [sensord] in1: +2.76 V (min = +2.78 V, max = +3.06 V)
Aug 16 20:53:54 [sensord] in2: +0.83 V (min = +0.80 V, max = +1.60 V) [ALARM]
Aug 16 20:53:54 [sensord] +3.3V: +3.34 V (min = +3.19 V, max = +3.60 V) [ALARM]
Aug 16 20:53:54 [sensord] in4: +0.17 V (min = +3.01 V, max = +1.06 V) [ALARM]
Aug 16 20:53:54 [sensord] in5: +2.51 V (min = +2.20 V, max = +2.80 V)
Aug 16 20:53:54 [sensord] in6: +1.78 V (min = +2.29 V, max = +2.98 V) [ALARM]
Aug 16 20:53:54 [sensord] 3VSB: +0.02 V (min = +2.21 V, max = +2.69 V) [ALARM]
Aug 16 20:53:54 [sensord] Vbat: +3.31 V
Aug 16 20:53:54 [sensord] fan1: 1366 RPM (min = 600 RPM)
Aug 16 20:53:54 [sensord] fan2: 1320 RPM (min = 600 RPM)
Aug 16 20:53:54 [sensord] fan3: 489 RPM (min = 600 RPM) [ALARM]
Aug 16 20:53:54 [sensord] temp1: 29.0 C (min = 0.0 C, max = 70.0 C) [ALARM]
Aug 16 20:53:54 [sensord] temp2: 27.0 C (min = 0.0 C, max = 60.0 C) [ALARM]
Aug 16 20:53:54 [sensord] temp3: -128.0 C (min = 100.0 C, max = -50.0 C)
Aug 16 20:53:54 [sensord] Chip: k10temp-pci-00c3
Aug 16 20:53:54 [sensord] Adapter: PCI adapter
Aug 16 20:53:54 [sensord] temp1: 12.5 C
Aug 16 20:54:54 [sensord] Sensor alarm: Chip it8721-isa-0290: in1: +2.76 V (min = +2.78 V, max = +3.06 V) [ALARM]
Aug 16 20:54:54 [sensord] Sensor alarm: Chip it8721-isa-0290: in4: +0.17 V (min = +3.01 V, max = +1.06 V) [ALARM]
Aug 16 20:54:54 [sensord] Sensor alarm: Chip it8721-isa-0290: in6: +1.78 V (min = +2.29 V, max = +2.98 V) [ALARM]
Aug 16 20:54:54 [sensord] Sensor alarm: Chip it8721-isa-0290: 3VSB: +0.02 V (min = +2.21 V, max = +2.69 V) [ALARM]
Aug 16 20:54:54 [sensord] Sensor alarm: Chip it8721-isa-0290: fan3: 505 RPM (min = 600 RPM) [ALARM]
Aug 16 20:55:54 [sensord] Sensor alarm: Chip it8721-isa-0290: in1: +2.76 V (min = +2.78 V, max = +3.06 V) [ALARM]
Aug 16 20:55:54 [sensord] Sensor alarm: Chip it8721-isa-0290: in4: +0.17 V (min = +3.01 V, max = +1.06 V) [ALARM]
Aug 16 20:55:54 [sensord] Sensor alarm: Chip it8721-isa-0290: in6: +1.78 V (min = +2.29 V, max = +2.98 V) [ALARM]
Aug 16 20:55:54 [sensord] Sensor alarm: Chip it8721-isa-0290: 3VSB: +0.02 V (min = +2.21 V, max = +2.69 V) [ALARM]
Aug 16 20:55:54 [sensord] Sensor alarm: Chip it8721-isa-0290: fan3: 524 RPM (min = 600 RPM) [ALARM]

I guess my power supply is going to die soon...  :(

[axelgenus]

Ok guys... it seems I found out what's going on with my system. Let's begin with "no, it is not an hardware fault"... well, mostly it isn't. I tried replacing my PSU but it did not fix the hang-ups. I also tried to switch distro (from Funtoo to Gentoo and then some LiveUSB distros) but nothing seemed to help (I thought about the opensource radeon driver locking the system and with Funtoo I was not able to install fglrx). Then, this morning a lightning stroke me and I was like: "let's try to clear the CMOS"... well it did not help but I noticed one thing: the mobo did not recognized the RAM timings and voltage correctly! I had to activate some ASUS thing called "D.O.C.P." and voil?... I get the right profile in the preferences and the system is rock solid since then... it's even f*cking faster! My Kingston kit is even in the damned QVL... WAY TO GO ASUS! :angry:

I was like this the whole morning:

 

EDIT: just to be clear. My RAM supports both JEDEC (standard) and XMP but the former is DDR3-1600 11-11-11 @1.5V and the latter is DDR3-1600 9-9-9 @1.65V. The BIOS set this profile: DDR3-1333 JEDEC 9-9-9-28 @ 1.5V. WTF?!?