[Linuxppc-users] memmap kernel option on ubuntu 16.04 power8

[Cédric Le Goater]

Not even with a guard record in the PNOR ? 

C.

==On 09/27/2017 05:53 PM, Calvin Sze wrote:
> Checked with BMC people, they said It cannot be done in BMC.
> 
> 
> Calvin Sze
> OpenSystem Enablement/Optimization
> email: calvins at us.ibm.com
> 
> 
> Inactive hide details for "Jenifer Hopper" ---09/27/2017 10:18:36 AM---For FSP based systems you can go into the ASM and deconf"Jenifer Hopper" ---09/27/2017 10:18:36 AM---For FSP based systems you can go into the ASM and deconfigure dimms (System Config -> HW Deconfig ->
> 
> From: "Jenifer Hopper" <jhopper at us.ibm.com>
> To: Brian Horton <brianh at linux.vnet.ibm.com>
> Cc: linuxppc-users at lists.ozlabs.org
> Date: 09/27/2017 10:18 AM
> Subject: Re: [Linuxppc-users] memmap kernel option on ubuntu 16.04 power8
> Sent by: "Linuxppc-users" <linuxppc-users-bounces+calvins=us.ibm.com at lists.ozlabs.org>
> 
> ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
> 
> 
> 
> For FSP based systems you can go into the ASM and deconfigure dimms (System Config -> HW Deconfig -> Mem Deconfig), but I don't know of a way to do that for BMC based systems.
> 
> Thanks,
> Jenifer Hopper
> IBM Linux Technology Center, Linux Optimization_
> __jhopper at us.ibm.com_ <mailto:jhopper at us.ibm.com>
> (512) 286-6701
> 
> -----"Linuxppc-users" <_linuxppc-users-bounces+jhopper=us.ibm.com at lists.ozlabs.org_ <mailto:linuxppc-users-bounces+jhopper=us.ibm.com at lists.ozlabs.org>> wrote: -----
> To: Michael Ellerman <_mpe at ellerman.id.au_ <mailto:mpe at ellerman.id.au>>, _linuxppc-users at lists.ozlabs.org_ <mailto:linuxppc-users at lists.ozlabs.org>
> From: Brian Horton
> Sent by: "Linuxppc-users"
> Date: 09/27/2017 10:12AM
> Subject: Re: [Linuxppc-users] memmap kernel option on ubuntu 16.04 power8
> 
> 
> On 09/27/2017 05:45 AM, Michael Ellerman wrote:
>> Hi Brian,
>>
>> Brian Horton <_brianh at linux.vnet.ibm.com_ <mailto:brianh at linux.vnet.ibm.com>> writes:
>>> Greetings all.
>>>
>>> I'm trying to run some tests with less memory than what's installed. the
>>> 'mem=128G' option works, but it ends up putting all of the memory on the
>>> first numa node:
>>
>> It doesn't put memory anywhere, it just clamps the total memory, and it
>> happens that the first (lowest address) 128G of RAM is all on node 0.
> 
> right, 'put' was a bad word choice. but yes, i see that the affect is
> that the first 128G of RAM is used.
> 
> 
>>
>>> someone said that it worked on RHEL, so maybe it's an Ubuntu-specific bug..
>>
>> OK that would be interesting, but I suspect it's not true :)
> 
> correct, on 2nd look it was mis-information. they confirmed that it
> doesn't work on RHEL either.
> 
>>> thoughts? ideas?
>>
>> AFAIK there's no way to achieve what you're after, without changing
>> kernel code or creating a custom device tree.
>>
>> The mem= option is very primitive, it just clamps the total as I said
>> above, it's not NUMA aware.
>>
>> What hardware/platform are you trying to do this on? Are you familiar
>> with kexec?
> 
> P8, Tuleta and Firestone, baremetal, ppc64le w/ Ubuntu 16.04 HWE kernel.
> 
> No, I haven't looked into anything with kexec - what are my options with
> that? I would prefer to stay away from building my own kernel code -
> i'll probably just live with the testing that i'm doing if that's my
> only option. i was just trying to be a little more 'realistic'.
> 
> Thanks! .bri.
> 
> 
>> cheers
>>
> 
> _______________________________________________
> Linuxppc-users mailing list_
> __Linuxppc-users at lists.ozlabs.org_ <mailto:Linuxppc-users at lists.ozlabs.org>_
> __https://urldefense.proofpoint.com/v2/url?u=https-3A__lists.ozlabs.org_listinfo_linuxppc-2Dusers&d=DwIGaQ&c=jf_iaSHvJObTbx-siA1ZOg&r=59cao60PiAeGQQaaFLkjnWW7CHr6qoRfN8NBRYVc9UQ&m=R9rn3U537CNoKLsBMZtzHp6VJc2mlnNdfmtsygTu4A0&s=o-r0sCpTTOrApxT7FVtnYCyWMpr9QjeO3pt0SW1yJm8&e=_
> 
> _______________________________________________
> Linuxppc-users mailing list
> Linuxppc-users at lists.ozlabs.org
> https://urldefense.proofpoint.com/v2/url?u=https-3A__lists.ozlabs.org_listinfo_linuxppc-2Dusers&d=DwIGaQ&c=jf_iaSHvJObTbx-siA1ZOg&r=XEmD5LeSMcW0pq1ccjREOfGKKSZEo6Uip7M_x_R-NUY&m=WNMxAOec1e_w34zBe3xVYIoJbJamikId375eNQw97IA&s=DGTWVG8bGGNjZ-cfXVvDyGfk8R4CJxIGw84m3Wi56Tk&e= 
> 
> 
> 
> 
> 
> _______________________________________________
> Linuxppc-users mailing list
> Linuxppc-users at lists.ozlabs.org
> https://urldefense.proofpoint.com/v2/url?u=https-3A__lists.ozlabs.org_listinfo_linuxppc-2Dusers&d=DwIGaQ&c=jf_iaSHvJObTbx-siA1ZOg&r=XHJsZhhuWSw9713Fp0ciew&m=TmviNY9vCNXvoo8fJe9fezSUxE-oPeWagE1fD19ZpLU&s=RT48AqWNNoMza32hJQTGnEeLGicTutGj2yrpsbdoYmk&e= 
>