From mboxrd@z Thu Jan 1 00:00:00 1970 From: Andrew Cooper Subject: Re: [PATCH v4 1/9] numa: __node_distance() should return u8 Date: Tue, 10 Mar 2015 11:53:04 +0000 Message-ID: <54FEDB20.6030800@citrix.com> References: <1425954475-4913-1-git-send-email-boris.ostrovsky@oracle.com> <1425954475-4913-2-git-send-email-boris.ostrovsky@oracle.com> Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit Return-path: In-Reply-To: <1425954475-4913-2-git-send-email-boris.ostrovsky@oracle.com> List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Sender: xen-devel-bounces@lists.xen.org Errors-To: xen-devel-bounces@lists.xen.org To: Boris Ostrovsky , jbeulich@suse.com, keir@xen.org, ian.jackson@eu.citrix.com, stefano.stabellini@eu.citrix.com, ian.campbell@citrix.com, wei.liu2@citrix.com Cc: elena.ufimtseva@oracle.com, dario.faggioli@citrix.com, xen-devel@lists.xen.org List-Id: xen-devel@lists.xenproject.org On 10/03/15 02:27, Boris Ostrovsky wrote: > SLIT values are byte-sized and some of them (0-9 and 255) have > special meaning. Adjust __node_distance() to reflect this and > modify scrub_heap_pages() to deal with __node_distance() returning > an invalid SLIT entry. > > Signed-off-by: Boris Ostrovsky You also need to teach XEN_SYSCTL_numainfo about the new NUMA_NO_DISTANCE. > --- > xen/arch/x86/srat.c | 15 +++++++++++---- > xen/common/page_alloc.c | 4 ++-- > xen/include/asm-x86/numa.h | 2 +- > xen/include/xen/numa.h | 3 ++- > 4 files changed, 16 insertions(+), 8 deletions(-) > > diff --git a/xen/arch/x86/srat.c b/xen/arch/x86/srat.c > index dfabba3..aa2eda3 100644 > --- a/xen/arch/x86/srat.c > +++ b/xen/arch/x86/srat.c > @@ -496,14 +496,21 @@ static unsigned node_to_pxm(nodeid_t n) > return 0; > } > > -int __node_distance(nodeid_t a, nodeid_t b) > +u8 __node_distance(nodeid_t a, nodeid_t b) > { > - int index; > + u8 slit_val; > > if (!acpi_slit) > return a == b ? 10 : 20; > - index = acpi_slit->locality_count * node_to_pxm(a); > - return acpi_slit->entry[index + node_to_pxm(b)]; > + > + slit_val = acpi_slit->entry[acpi_slit->locality_count * node_to_pxm(a) + > + node_to_pxm(b)]; This would be easier to read if you kept the old index temporary (although making it a u64). ~Andrew > + > + /* ACPI defines 0xff as an unreachable node and 0-9 are undefined */ > + if ((slit_val == 0xff) || (slit_val <= 9)) > + return NUMA_NO_DISTANCE; > + else > + return slit_val; > } > > EXPORT_SYMBOL(__node_distance); > diff --git a/xen/common/page_alloc.c b/xen/common/page_alloc.c > index d96d25b..0f0ca56 100644 > --- a/xen/common/page_alloc.c > +++ b/xen/common/page_alloc.c > @@ -1435,13 +1435,13 @@ void __init scrub_heap_pages(void) > /* Figure out which NODE CPUs are close. */ > for_each_online_node ( j ) > { > - int distance; > + u8 distance; > > if ( cpumask_empty(&node_to_cpumask(j)) ) > continue; > > distance = __node_distance(i, j); > - if ( distance < last_distance ) > + if ( (distance < last_distance) && (distance != NUMA_NO_DISTANCE) ) > { > last_distance = distance; > best_node = j; > diff --git a/xen/include/asm-x86/numa.h b/xen/include/asm-x86/numa.h > index cc5b5d1..7a489d3 100644 > --- a/xen/include/asm-x86/numa.h > +++ b/xen/include/asm-x86/numa.h > @@ -85,6 +85,6 @@ extern int valid_numa_range(u64 start, u64 end, nodeid_t node); > #endif > > void srat_parse_regions(u64 addr); > -extern int __node_distance(nodeid_t a, nodeid_t b); > +extern u8 __node_distance(nodeid_t a, nodeid_t b); > > #endif > diff --git a/xen/include/xen/numa.h b/xen/include/xen/numa.h > index ac4b391..7aef1a8 100644 > --- a/xen/include/xen/numa.h > +++ b/xen/include/xen/numa.h > @@ -7,7 +7,8 @@ > #define NODES_SHIFT 0 > #endif > > -#define NUMA_NO_NODE 0xFF > +#define NUMA_NO_NODE 0xFF > +#define NUMA_NO_DISTANCE 0xFF > > #define MAX_NUMNODES (1 << NODES_SHIFT) >