All of lore.kernel.org
 help / color / mirror / Atom feed
From: Thiago Jung Bauermann <bauerman@linux.vnet.ibm.com>
To: linuxppc-dev@lists.ozlabs.org
Cc: Michael Ellerman <mpe@ellerman.id.au>,
	Sukadev Bhattiprolu <sukadev@linux.vnet.ibm.com>,
	Paul Clarke <pc@us.ibm.com>,
	linux-kernel@vger.kernel.org
Subject: Re: [PATCH 2/2] powerpc/pseries: Dynamically increase RMA size
Date: Wed, 01 Feb 2017 15:49:23 -0200	[thread overview]
Message-ID: <2499748.LKgOIUjhgo@morokweng> (raw)
In-Reply-To: <87wpdatqrt.fsf@concordia.ellerman.id.au>

Hello,

Am Mittwoch, 1. Februar 2017, 16:37:58 BRST schrieb Michael Ellerman:
> Sukadev Bhattiprolu <sukadev@linux.vnet.ibm.com> writes:
> > Paul Clarke [pc@us.ibm.com] wrote:
> > ---
> >=20
> > From f9e9e8460206bc3fa7eaa741b9a2bde22870b9e0 Mon Sep 17 00:00:00 2001
>=20
> I know it's been a while but I think it would still be good to get this
> in a shape that we can merge it.

Sorry if this has been tried and didn't work or if I'm missing something=20
obvious:

Instead of this method of trying a small RMA size and rebooting to try a=20
bigger size, could the "min RMA percentage of total RAM" field of the=20
ibm_architecture_vec be used?

LoPAPR says that "The Initial size of the RMA is set to the greater of the=
=20
values indicated by bytes 24-27 [min RMA] or 32 [min RMA percentage of tota=
l=20
RAM] of option vector number 2 =E2=80=9COpen Firmware=E2=80=9D or minimum R=
MA size supported=20
by the platform and capped by the maximum memory defined for the partition =
and=20
the maximum size of the RMA supported by the platform. The respective selec=
ted=20
values are reported in the length of the first memory property."

My understanding is that these patches are intended for big guests with man=
y=20
processors, but the RMA size isn't changed to 512MB outright because of=20
worries that it could affect smaller guests. Since guests with many process=
ors=20
tend to have more RAM as well, specifying a min RMA size of 256MB and a min=
=20
RMA percentage of, say, 10% or 20% could make the host automatically alloca=
te=20
an adequate RMA size in the first boot.

=2D-=20
Thiago Jung Bauermann
IBM Linux Technology Center

WARNING: multiple messages have this Message-ID (diff)
From: Thiago Jung Bauermann <bauerman@linux.vnet.ibm.com>
To: linuxppc-dev@lists.ozlabs.org
Cc: Michael Ellerman <mpe@ellerman.id.au>,
	Sukadev Bhattiprolu <sukadev@linux.vnet.ibm.com>,
	Paul Clarke <pc@us.ibm.com>,
	linux-kernel@vger.kernel.org
Subject: Re: [PATCH 2/2] powerpc/pseries: Dynamically increase RMA size
Date: Wed, 01 Feb 2017 15:49:23 -0200	[thread overview]
Message-ID: <2499748.LKgOIUjhgo@morokweng> (raw)
In-Reply-To: <87wpdatqrt.fsf@concordia.ellerman.id.au>

Hello,

Am Mittwoch, 1. Februar 2017, 16:37:58 BRST schrieb Michael Ellerman:
> Sukadev Bhattiprolu <sukadev@linux.vnet.ibm.com> writes:
> > Paul Clarke [pc@us.ibm.com] wrote:
> > ---
> > 
> > From f9e9e8460206bc3fa7eaa741b9a2bde22870b9e0 Mon Sep 17 00:00:00 2001
> 
> I know it's been a while but I think it would still be good to get this
> in a shape that we can merge it.

Sorry if this has been tried and didn't work or if I'm missing something 
obvious:

Instead of this method of trying a small RMA size and rebooting to try a 
bigger size, could the "min RMA percentage of total RAM" field of the 
ibm_architecture_vec be used?

LoPAPR says that "The Initial size of the RMA is set to the greater of the 
values indicated by bytes 24-27 [min RMA] or 32 [min RMA percentage of total 
RAM] of option vector number 2 “Open Firmware” or minimum RMA size supported 
by the platform and capped by the maximum memory defined for the partition and 
the maximum size of the RMA supported by the platform. The respective selected 
values are reported in the length of the first memory property."

My understanding is that these patches are intended for big guests with many 
processors, but the RMA size isn't changed to 512MB outright because of 
worries that it could affect smaller guests. Since guests with many processors 
tend to have more RAM as well, specifying a min RMA size of 256MB and a min 
RMA percentage of, say, 10% or 20% could make the host automatically allocate 
an adequate RMA size in the first boot.

-- 
Thiago Jung Bauermann
IBM Linux Technology Center

  reply	other threads:[~2017-02-01 17:49 UTC|newest]

Thread overview: 10+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2016-08-05  6:10 [PATCH 1/2] powerpc/pseries: Use a helper to fixup nr_cores Sukadev Bhattiprolu
2016-08-05  6:14 ` [PATCH 2/2] powerpc/pseries: Dynamically increase RMA size Sukadev Bhattiprolu
2016-08-05 13:28   ` kbuild test robot
2016-08-05 18:30   ` Sukadev Bhattiprolu
2016-08-05 19:04     ` Paul Clarke
2016-08-09 17:11       ` Sukadev Bhattiprolu
2017-02-01  5:37         ` Michael Ellerman
2017-02-01 17:49           ` Thiago Jung Bauermann [this message]
2017-02-01 17:49             ` Thiago Jung Bauermann
2017-02-01 18:11             ` Sukadev Bhattiprolu

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=2499748.LKgOIUjhgo@morokweng \
    --to=bauerman@linux.vnet.ibm.com \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linuxppc-dev@lists.ozlabs.org \
    --cc=mpe@ellerman.id.au \
    --cc=pc@us.ibm.com \
    --cc=sukadev@linux.vnet.ibm.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.