RE: [PATCH 25/37] xen/arm: implement bad_srat for Arm NUMA initialization

From: Wei Chen <Wei.Chen@arm.com>
To: Stefano Stabellini <sstabellini@kernel.org>
Cc: "xen-devel@lists.xenproject.org" <xen-devel@lists.xenproject.org>,
	"julien@xen.org" <julien@xen.org>,
	Bertrand Marquis <Bertrand.Marquis@arm.com>,
	"jbeulich@suse.com" <jbeulich@suse.com>,
	"andrew.cooper3@citrix.com" <andrew.cooper3@citrix.com>,
	"roger.pau@citrix.com" <roger.pau@citrix.com>,
	"wl@xen.org" <wl@xen.org>
Subject: RE: [PATCH 25/37] xen/arm: implement bad_srat for Arm NUMA initialization
Date: Fri, 24 Sep 2021 04:45:16 +0000	[thread overview]
Message-ID: <DB9PR08MB68570E6EA33BF7C6E9E23CBD9EA49@DB9PR08MB6857.eurprd08.prod.outlook.com> (raw)
In-Reply-To: <alpine.DEB.2.21.2109231906570.17979@sstabellini-ThinkPad-T480s>

Hi Stefano,

> -----Original Message-----
> From: Stefano Stabellini <sstabellini@kernel.org>
> Sent: 2021年9月24日 10:10
> To: Wei Chen <Wei.Chen@arm.com>
> Cc: xen-devel@lists.xenproject.org; sstabellini@kernel.org; julien@xen.org;
> Bertrand Marquis <Bertrand.Marquis@arm.com>; jbeulich@suse.com;
> andrew.cooper3@citrix.com; roger.pau@citrix.com; wl@xen.org
> Subject: Re: [PATCH 25/37] xen/arm: implement bad_srat for Arm NUMA
> initialization
> 
> On Thu, 23 Sep 2021, Wei Chen wrote:
> > NUMA initialization will parse information from firmware provided
> > static resource affinity table (ACPI SRAT or DTB). bad_srat if a
> > function that will be used when initialization code encounters
> > some unexcepted errors.
> >
> > In this patch, we introduce Arm version bad_srat for NUMA common
> > initialization code to invoke it.
> >
> > Signed-off-by: Wei Chen <wei.chen@arm.com>
> > ---
> >  xen/arch/arm/numa.c | 7 +++++++
> >  1 file changed, 7 insertions(+)
> >
> > diff --git a/xen/arch/arm/numa.c b/xen/arch/arm/numa.c
> > index 3755b01ef4..5209d3de4d 100644
> > --- a/xen/arch/arm/numa.c
> > +++ b/xen/arch/arm/numa.c
> > @@ -18,6 +18,7 @@
> >   *
> >   */
> >  #include <xen/init.h>
> > +#include <xen/nodemask.h>
> >  #include <xen/numa.h>
> >
> >  static uint8_t __read_mostly
> > @@ -25,6 +26,12 @@ node_distance_map[MAX_NUMNODES][MAX_NUMNODES] = {
> >      { 0 }
> >  };
> >
> > +__init void bad_srat(void)
> > +{
> > +    printk(KERN_ERR "NUMA: Firmware SRAT table not used.\n");
> > +    fw_numa = -1;
> > +}
> 
> I realize that the series keeps the "srat" terminology everywhere on DT
> too. I wonder if it is worth replacing srat with something like
> "numa_distance" everywhere as appropriate. I am adding the x86
> maintainers for an opinion.
> 
> If you guys prefer to keep srat (if nothing else, it is concise), I am
> also OK with keeping srat although it is not technically accurate.

I have placed some comments in patch#23 for srat.
I wouldn't like to replace srat by numa_distance. because srat not only
contains distance, and it also includes some affinity information for CPU
or other devices.