DragonflyBSD: CPU overheating on laptop

[Sehnsucht94]

SOLVED

Hi fellow Unix users,
After having received a new laptop (indeed second hand ), I moved the former from OpenBSD to DragonflyBSD, and my new main to FreeBSD (as it comes with Nvidia graphics)

I must admit that my old cheap laptop boasts the BEST performance ever, now that it runs Dragonfly . However here comes my issue: from time time, even when idle, the CPU heats up progressively and rapidly till the point it overtakes 90°C, as witnessed by coretemp turning it off suddenly, and desk literally burning.

SPECS:
-The laptop of interest is an Acer TravelMate 5335, model No: PEW52.
- sysctl -a | egrep -i 'hw.machine|hw.model|hw.ncpu':

hw.machine: x86_64
hw.model: Celeron(R) Dual-Core CPU       T3500  @ 2.10GHz
hw.ncpu: 2

- grep memory /var/run/dmesg.boot:

real memory  = 3146752000 (3000 MB)
avail memory = 2909028352 (2774 MB)
agp0: aperture size is 256M, detected 65532k stolen memory

- doas smartctl -a /dev/da0:

=== START OF INFORMATION SECTION ===
Vendor:               SATA
Product:              Hitachi HTS54503
Revision:             PB3O
User Capacity:        320,072,933,376 bytes [320 GB]
Logical block size:   512 bytes
Serial number:        101028PBN304GTE56DBR
Device type:          disk
Local Time is:        Tue Oct 10 13:50:30 2017 CEST
SMART support is:     Unavailable - device lacks SMART capability.

=== START OF READ SMART DATA SECTION ===
Current Drive Temperature:     0 C
Drive Trip Temperature:        0 C

Error Counter logging not supported

. As you can see unfortunately I cannot monitor HDD temperature.

- acpiconf -i batt0:

Design capacity:	4400 mAh
Last full capacity:	560 mAh
Technology:		secondary (rechargeable)
Design voltage:		11100 mV
Capacity (warn):	220 mAh
Capacity (low):		132 mAh
Low/warn granularity:	264 mAh
Warn/full granularity:	3780 mAh
Model number:		AS10D71
Serial number:		0021
Type:			Lion
OEM info:		Simplo 
State:			high 
Remaining capacity:	100%
Remaining time:		unknown
Present rate:		0 mA (0 mW)
Present voltage:	12301 mV

- pciconf -lv | grep graphics && sysctl -a | grep graphics:

device     = 'Intel Mobile 4 Series Chipset Integrated Graphics Controller'

kernel: [drm] Memory usable by graphics device = 2048M

Despite being well aware that ZFS and HAMMER are targeted to large storage and RAMs, since I had heard on forums of FreeBSD and Arch Linux desktop users formatting their HDD with ZFS, instead of UFS or ext4, I decided to give famous HAMMER a try (with BIOS and MBR P.T.), without deeper reading about it (mea culpa).
cat -v /etc/fstab:

# Device		Mountpoint	FStype	Options		Dump	Pass#
/dev/serno/101028PBN304GTE56DBR.s1a		/boot		ufs	rw	22
/dev/serno/101028PBN304GTE56DBR.s1b		none		swap	sw	00
/dev/serno/101028PBN304GTE56DBR.s1d		/		hammer	rw	11
/dev/serno/101028PBN304GTE56DBR.s1e		/build		hammer	rw	22
/build/usr.obj		/usr/obj		null			rw	00
/build/var.crash	/var/crash		null			rw	00
/build/var.cache	/var/cache		null	rw		0	0
/build/var.spool	/var/spool		null	rw		0	0
/build/var.log		/var/log		null	rw		0	0
/build/var.tmp		/var/tmp		null	rw		0	0
tmpfs			/tmp		tmpfs		rw		0	0
proc			/proc		procfs		rw		0	0

As expected, I noticed HAMMER eats up way too much RAM: on idle, on the same fluxbox-based desktop configuration OpenBSD on UFS (BSD tartition map) consumed less than 400Mb, whereas just a little ago (before writing that post) Dragonfly was using 987Mb. Could be HAMMER the problem?

Anyway, although opening more programs obviously weighs the CPU load down and speeds up overheating, apparently there's no strict linkage between active processes and heating. In fact last couple of time it happened I was:
- downloading a file with aria2 and 1 only terminal opened
- taking notes on word perfect, while listening to radio with mplayer

Other times I have QupZilla, GVim, Insight, Epic5 opened, while building ports, and temperature doesn't go beyond 55 °C.
On the plus side, checking CPU load while it overheats usually reveals a minimal usage percentage, which is quite odd for me.

As, for now I have: sysctl -n hw.sensors.cpu*.temp0
:

60.00 degC (node0 core0 temp), OK
63.00 degC (node0 core1 temp), OK

Which is quite high IMHO (maybe I'm wrong), considering that CPU-FAN has always worked out of the box, and I only running Midori.

In fact, in spite of the temperature, running vmstat 1 | awk '{print $14,$15,$16,$17}' shows that Celeron is idle 99%: :

ctx    us  sy id
1.11K  1   0  99

I tried to grep the temperature of processor's cache thermal sensor as well, loading memtemp, the DragonflyBSD's "device driver for memory thermal sensors", but it seems it supports Intel Cores and Xeons only, as sysctl -n hw.dimminfo returns nothing.

On DragonflyBSD's Handbook, I noticed est deamon is recommended to set and read the CPU frequency through sysctl machdep.est. Sounded great....unfortunately on my system there's no trace of estd in kernel modules nor in the dports tree either.
Perhaps that section of handbook is outdated (there are many and likely just few active volunteer documentation mantainers).

Nonetheless, I decided to download and build Johannes Hofmann's estd from its Homepage, which states:
"This daemon dynamically sets the CPU-frequency on SpeedStep-, PowerNow-, and ACPI P-States enabled CPUs depending on current cpu-utilization. It supports NetBSD 3.0 or later and DragonFly BSD".

I was caught by this statement on its man page:
"Extensive configurability ensures you can
also use it in non-standard situations, eg. when you want to keep the
CPU from running in lower speeds although the system is idle in order
to ensure you have full processing power for interactive applications
that use the CPU in small bursts".

So, I tried to deamonize estd so as to limit maximum CPU frequency to 1 Ghz, in this way: estd -d -M 1000, but the output was:

Cannot guess CPU-scaling technology. (maybe you are missing some kernel-option?)

As estd README does not specifies any required kernel option, I'm basically stuck

If ever were needed, here's my /etc/rc.conf:

hostname="TravelMate"
keymap="it.iso"
nfs_reserved_port_only="YES"
sshd_enable="YES"
nfs_client_enable="YES"
rpc_umntall_enable="NO"
dumpdev="/dev/serno/101028PBN304GTE56DBR.s1b"
wpa_supplicant_program="/usr/local/sbin/wpa_supplicant"
ifconfig_bge0="DHCP"
wlans_urtwn0="wlan0"
ifconfig_wlan0="WPA ssid D-Link  DHCP"
dntpd_enable="YES"
moused_enable="YES"
dbus_enable="YES"

And my /boot/loader.conf:

vfs.root.mountfrom="hammer:serno/101028PBN304GTE56DBR.s1d"
snd_hda_load="YES"
kern.vty=vt
i915_load="YES"
kern.vt.fb.default_mode="1024x768"
legal.realtek.license_ack=1
if_urtwn_load="YES"
hw.psm.synaptics_support="1"
tmpfs_load="YES"
msdos_iconv_load="YES"
if_bridge_load="YES"
if_tap_load="YES"

The bridged network is for qemu.
No particular program is loaded in ~/.xsession (being using XDM)

Well probably I've written too much already, I'm here to provide any further info now lacking.

Thanks in advance for any help

[Oko]

You are unlikely to find any help on this forum for DragonFly BSD. Please send an e-mail to users.AT.dragonflybsd.org. The group is minuscule (less than 10 developers with Matt committing 80% of the stuff) but they are very good at addressing issues. Most likely Matt himself will offer you some advises.

I am too tired to go carefully over your e-mail tonight but on the first glance temperature of little over 60C are within your CPU specification range. HAMMER is not a RAM hog like ZFS (which I run at work). I am running it on my home file server. Some interesting outputs for you

dfly# uname -a
DragonFly dfly.bagdala2.net 4.8-RELEASE DragonFly v4.8.1.5.g69508-RELEASE #9: Mon Aug 28 23:50:01 EDT 2017     root@dfly.bagdala2.net:/usr/obj/usr/src/sys/X86_64_GENERIC  x86_64


dfly# sysctl hw.sensors
hw.sensors.acpi_tz0.temp0: 27.80 degC
hw.sensors.acpi_tz1.temp0: 29.80 degC
hw.sensors.cpu0.temp0: 48.00 degC (node0 core0 temp), OK
hw.sensors.cpu1.temp0: 54.00 degC (node0 core1 temp), OK
hw.sensors.cpu_node0.temp0: 54.00 degC (node0 temp), OK
hw.sensors.dev_idmax: 5

dfly# sysctl -a | egrep -i 'hw.machine|hw.model|hw.ncpu'
hw.machine: x86_64
hw.model: Intel(R) Celeron(R) CPU 1037U @ 1.80GHz
hw.ncpu: 2
hw.machine_arch: x86_64


dfly# mount
ROOT on / (hammer, noatime, local)
devfs on /dev (devfs, nosymfollow, local)
/dev/serno/B620550018.s1a on /boot (ufs, local)
/pfs/@@-1:00001 on /var (null, local)
/pfs/@@-1:00002 on /tmp (null, local)
/pfs/@@-1:00003 on /home (null, local)
/pfs/@@-1:00004 on /usr/obj (null, local)
/pfs/@@-1:00005 on /var/crash (null, local)
/pfs/@@-1:00006 on /var/tmp (null, local)
procfs on /proc (procfs, local)
DATA on /data (hammer, noatime, local)
BACKUP on /backup (hammer, read-only, noatime, local)
/data/pfs/@@-1:00001 on /data/backups (null, local)
/data/pfs/@@-1:00002 on /data/nfs (null, NFS exported, local)


dfly# df -h
Filesystem                  Size   Used  Avail Capacity  Mounted on
ROOT                       21.7G  9475M  12.5G    43%    /
devfs                      1024B  1024B     0B   100%    /dev
/dev/serno/B620550018.s1a  1008M   448M   480M    48%    /boot
/pfs/@@-1:00001            21.7G  9475M  12.5G    43%    /var
/pfs/@@-1:00002            21.7G  9475M  12.5G    43%    /tmp
/pfs/@@-1:00003            21.7G  9475M  12.5G    43%    /home
/pfs/@@-1:00004            21.7G  9475M  12.5G    43%    /usr/obj
/pfs/@@-1:00005            21.7G  9475M  12.5G    43%    /var/crash
/pfs/@@-1:00006            21.7G  9475M  12.5G    43%    /var/tmp
procfs                     4096B  4096B     0B   100%    /proc
DATA                       2792G   156G  2637G     6%    /data
BACKUP                     2792G   507G  2285G    18%    /backup
/data/pfs/@@-1:00001       2792G   156G  2637G     6%    /data/backups
/data/pfs/@@-1:00002       2792G   156G  2637G     6%    /data/nfs

dfly# vmstat 
--procs-- ---memory-- -------paging------ --disks-- -----faults------ ---cpu---
  r  b  w   fre   flt   re   pi   po   fr  da0  da1   int   sys   ctx us sy id
  0  1  0 308M  117M 1.5K 3.0K 4.2K 219M    0    0  639M  356M  577M  0  0 100


load averages:  0.01,  0.01,  0.00;               up 11+22:17:05       22:10:00
46 processes: 1 running, 46 active
CPU states:  0.0% user,  0.0% nice,  0.0% system,  0.0% interrupt,  100% idle
Memory: 3322M Active, 1900M Inact, 2236M Wired, 233M Cache, 772M Buf, 74M Free
Swap: 6610M Total, 20M Used, 6590M Free

[Sehnsucht94]

EDIT

OKO I've just found out where the problem stood.

It was related to my urtwn driver.

Since:
- urtwn on DrahonflyBSD complained about altq_maxlen not being set when attaching to my WiFi dongle

- wireless connection was unstable and went up and down

I had asked about that on #dragonflybsd IRC channel, and mods suggested me to patch the driver. I struggled to do so even following their indications, but finally had made it through.
However this was making my dhclient going crazy when downloading large files. I went back to native driver,connevcultion is loose,but no more overheating now

.................................................. ....... ..............

Thanks very much for your reply Oko,
I can see you've got a similar configuration on the same processor. Judging from your CPU temp, my temperature on idle, despite being a little bit higher than yours, shall be considered more or less normal .

The only thing that still doesn't add up are these sudden, intense overheatings and poweroffs.

I'll try to disable some services, look in dmesg and logs in order to find out a clue, then if I can't work it out myself, I'll probably subscribe to the mailing list.

By the way, I'm starting to love this OS so much.
As I mentioned, my laptop has never been so fast. Dragonfly provides many interesting tools and hammer seems very stable. It's a pity that it doesn't get the attention it deserves.

Still seems these 10 developers have been doing a superb job so far.