Suddenly poweroff in 11-Current r300097

[Kevin Oberman]

On Thu, Jun 2, 2016 at 7:41 AM, Hans Petter Selasky <hps at selasky.org> wrote:

> On 06/02/16 03:07, RayCherng Yu wrote:
>
>> I got a suddenly poweroff in r300097 (and previous revision in April and
>> May) when I built textproc/docproj.
>> My machine is Macbook Pro 13 2011 early. I have checked the Apple website.
>> My bios is the latest version.
>> Actually it also happened in 10.3-STABLE.
>> It happened when the machine load was heavy. Before it shutdown, the fan
>> started to run very loudly. After several seconds (20 or 30 seconds), my
>> laptop shutdown (poweroff directly) suddenly. It seems not happen with the
>> AC power supply connected.
>>
>> I installed both Mac OSX and FreeBSD (dual boot). It never happened in Mac
>> OSX.
>>
>> My dmesg:
>> http://pastebin.com/QjZmbGCB
>>
>> My sysctl hw.acpi:
>>
>> hw.acpi.acline: 0
>> hw.acpi.battery.info_expire: 5
>> hw.acpi.battery.units: 1
>> hw.acpi.battery.state: 1
>> hw.acpi.battery.time: 87
>> hw.acpi.battery.life: 59
>> hw.acpi.cpu.cx_lowest: C8
>> hw.acpi.reset_video: 0
>> hw.acpi.handle_reboot: 1
>> hw.acpi.disable_on_reboot: 0
>> hw.acpi.verbose: 0
>> hw.acpi.s4bios: 0
>> hw.acpi.sleep_delay: 1
>> hw.acpi.suspend_state: S3
>> hw.acpi.standby_state: NONE
>> hw.acpi.lid_switch_state: NONE
>> hw.acpi.sleep_button_state: S3
>> hw.acpi.power_button_state: S5
>> hw.acpi.supported_sleep_state: S3 S4 S5
>>
>>
> Hi,
>
> Do you have a temperature sysctl? Usually FreeBSD will shutdown the system
> if the ACPI temperature exceeds some value. Maybe it would be better to
> reduce the CPU load when the temperature goes up instead of facing a
> shutdown?
>
> --HPS


The relevant information is probably found in dev.cpu. That is where all
temperature information is located as it is per-CPU, not per-system. Of
particular interest is dev.cpu.0.cx_lowest, dev.cpu.0.cx_supported, and
dev.cpu.0.freq_levels. A snapshot of dev.cpu.0 when the fan has cranked up,
but before shutdown would be nice, too.

I see no hw.acpi.thermal information. This is very odd. These values
indicate what the system will do and is doing if it starts getting too hot.

Is coretemp loaded? It is required to see the core temperatures and those
are almost certainly significant. It may account for the lack of thermal
information. Finally, a dmesg might be useful as it will tell us more about
just what thermal control techniques are enabled.

Just to explain a bit on how this should work: when the temperature exceeds
some BIOS defined point, the system should "throttle" by pausing one of
every 8 clock cycles. If that does not fix the problem, the it rests for
two of every 8 and so on until the temperature is reduced. If it continues
to rise and reaches another BIOS set point, it will initiate an emergency
shutdown. If it reaches a CPU defined temperature, the power will shut off
immediately. Note that this is entirely a hardware function with no BIOS or
OS involvement. It should NEVER happen in normal operation as it is
triggered by a significant overtemp that threatens to destroy the CPU. I've
only seen it once when the CPU heat sink came loose on an old P4 system
several years ago.

I should mention that I have zero experience with Apple hardware and it is
possible that they do some things differently than I have seen on other
hardware.
--
Kevin Oberman, Part time kid herder and retired Network Engineer
E-mail: rkoberman at gmail.com
PGP Fingerprint: D03FB98AFA78E3B78C1694B318AB39EF1B055683