From mboxrd@z Thu Jan  1 00:00:00 1970
From: Ian Campbell <ian.campbell@citrix.com>
Subject: Re: [PATCH v7 21/21] xl: vNUMA support
Date: Thu, 12 Mar 2015 10:00:24 +0000
Message-ID: <1426154424.21353.348.camel@citrix.com>
References: <1425905497-30835-1-git-send-email-wei.liu2@citrix.com>
	<1425905497-30835-22-git-send-email-wei.liu2@citrix.com>
	<1426086726.21353.267.camel@citrix.com>
	<20150311201452.GA16468@zion.uk.xensource.com>
Mime-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Content-Transfer-Encoding: 7bit
Return-path: <xen-devel-bounces@lists.xen.org>
In-Reply-To: <20150311201452.GA16468@zion.uk.xensource.com>
List-Unsubscribe: <http://lists.xen.org/cgi-bin/mailman/options/xen-devel>,
	<mailto:xen-devel-request@lists.xen.org?subject=unsubscribe>
List-Post: <mailto:xen-devel@lists.xen.org>
List-Help: <mailto:xen-devel-request@lists.xen.org?subject=help>
List-Subscribe: <http://lists.xen.org/cgi-bin/mailman/listinfo/xen-devel>,
	<mailto:xen-devel-request@lists.xen.org?subject=subscribe>
Sender: xen-devel-bounces@lists.xen.org
Errors-To: xen-devel-bounces@lists.xen.org
To: Wei Liu <wei.liu2@citrix.com>
Cc: Ian Jackson <ian.jackson@eu.citrix.com>, xen-devel@lists.xen.org
List-Id: xen-devel@lists.xenproject.org

On Wed, 2015-03-11 at 20:14 +0000, Wei Liu wrote:
> On Wed, Mar 11, 2015 at 03:12:06PM +0000, Ian Campbell wrote:
> > On Mon, 2015-03-09 at 12:51 +0000, Wei Liu wrote:
> > 
> > > +Each B<VNODE_CONFIG_OPTION> is a quoted key=value pair. Supported
> > > +B<VNODE_CONFIG_OPTION>s are:
> > 
> > Which of these are optional and which are mandatory? All mandatory?
> > 
> 
> Yes, all mandatory for now.

Please can you say so.
> 
> > > +    if (!xlu_cfg_get_long (config, "maxmem", &l, 0))
> > > +        max_memkb1 = l * 1024;
> > 
> > I think you should arrange that if the user has specified maxmem then it
> > has already been parsed and is available in b_info->max_memkb for use in
> > this function. Reparsing it just leaves scope to get out of sync.
> > 
> > In particular this doesn't handle the defaulting of maxmem to memory.
> > 
> 
> This is exactly the reason I do this here.  At this point
> b_info->max_memkb is always set.
> 
> If user has specified maxmem we should check if the configuration he /
> she specifies matches the vnuma value to avoid misconfiguration;
> otherwise we just use vnuma memory as maxmem.
> 
> But see below regarding maxvcpus=.
[...]
> > > +    if (total_cpus != b_info->max_vcpus) {
> > > +        fprintf(stderr, "xl: vnuma vcpus and maxvcpus= mismatch\n");
> > > +        exit(1);
> > > +    }
> > 
> > Can't we do as we do with memory here and set max_vcpus if it isn't
> > already configured?
> > 
> 
> This can be done.  It will require moving parse_vnuma_config to
> beginning of parse_config_data and stubbing out parsing of maxvcpus
> and maxmem if vnuma configuration presents.

I'm afraid I don't follow.

I was thinking the flow would be:

	Call libxl_foo_init to set everything to the default.

        f ( max_mem in cfg )
                cfg->maxmem = v
        if ( memory in cfg )
                cfg->memory = v
        if ( vcpus in cfg )
                cfg->vcpus = v
                
        do numa parsing, using fact that cfg->{maxmem,memory,vcpus} are
        either default value (in which case they are set here) or
        explicitly set to some value, in which case they are sanity
        checked.

        handle maxmem/memory/vcpus/etc still being default values by
        setting the appropriate defaults.

Ian.