Re: [Powertop] Fwd: Suggested tunables addition: sysctl kernel.core_pattern

Re: [Powertop] Fwd: Suggested tunables addition: sysctl kernel.core_pattern

[Arjan van de Ven]

[-- Attachment #1: Type: text/plain, Size: 216 bytes --]

On 1/4/2016 4:49 PM, George Barrett wrote:
> I primarily use my laptop for hacking on my own little projects, so this occurs fairly frequently

hmm but wouldn't you WANT the core dumps?

(I know I do ;-) )

Re: [Powertop] Fwd: Suggested tunables addition: sysctl kernel.core_pattern

[George Barrett]

[-- Attachment #1: Type: text/plain, Size: 1129 bytes --]

I primarily use my laptop for hacking on my own little projects, so 
this occurs fairly frequently

On Tue, Jan 5, 2016 at 11:47 AM, Magnus Fromreide 
<magfr(a)lysator.liu.se> wrote:
> On Sat, Jan 02, 2016 at 09:00:57PM +1100, George Barrett wrote:
>>  Hi, first off thanks for the great tool!
>> 
>>  I currently rely on a systemd unit file present in the Fedora 
>> package for
>>  powertop to do most of my laptop's power saving config on boot, and 
>> it works
>>  great. However, I've noticed a missing tunable which might be 
>> helpful for
>>  some folks: setting the kernel.core_pattern sysctl parameter to 
>> |/dev/null
>>  or similar. On my x220, this parameter set to it's default causes
>>  systemd-coredump to peg a single core, sometimes for quite an 
>> extended
>>  period of time. On many devices this would cause substantially 
>> greater heat
>>  output, and most importantly substantially higher power consumption.
> 
> I have to ask, why do you get enough core dumps for this to be a 
> problem?
> Wouldn't a better solution be to fix the application that dumps core?
> 
> /MF

Re: [Powertop] Fwd: Suggested tunables addition: sysctl kernel.core_pattern

[Magnus Fromreide]

[-- Attachment #1: Type: text/plain, Size: 874 bytes --]

On Sat, Jan 02, 2016 at 09:00:57PM +1100, George Barrett wrote:
> Hi, first off thanks for the great tool!
> 
> I currently rely on a systemd unit file present in the Fedora package for
> powertop to do most of my laptop's power saving config on boot, and it works
> great. However, I've noticed a missing tunable which might be helpful for
> some folks: setting the kernel.core_pattern sysctl parameter to |/dev/null
> or similar. On my x220, this parameter set to it's default causes
> systemd-coredump to peg a single core, sometimes for quite an extended
> period of time. On many devices this would cause substantially greater heat
> output, and most importantly substantially higher power consumption.

I have to ask, why do you get enough core dumps for this to be a problem?
Wouldn't a better solution be to fix the application that dumps core?

/MF

[Powertop] Fwd: Suggested tunables addition: sysctl kernel.core_pattern

[George Barrett]

[-- Attachment #1: Type: text/plain, Size: 994 bytes --]

Hi, first off thanks for the great tool!

I currently rely on a systemd unit file present in the Fedora package 
for powertop to do most of my laptop's power saving config on boot, and 
it works great. However, I've noticed a missing tunable which might be 
helpful for some folks: setting the kernel.core_pattern sysctl 
parameter to |/dev/null or similar. On my x220, this parameter set to 
it's default causes systemd-coredump to peg a single core, sometimes 
for quite an extended period of time. On many devices this would cause 
substantially greater heat output, and most importantly substantially 
higher power consumption.

I had a quick browse of the source code to see if this addition was 
trivial enough for me to submit a patch over, but from what I could see 
the existing tunables seem to work exclusively with sysfs. 
Unfortunately I don't write much C++, so I apologise for sending this 
particular email rather than submitting a patch for y'all.

Thanks