Dan Williams <dan.j.williams@xxxxxxxxx> writes:
> The acpi_map_pxm_to_online_node() helper is used to find the closest
> online node to a given proximity domain. This is used to map devices in
> a proximity domain with no online memory or cpus to the closest online
> node and populate a device's 'numa_node' property. The numa_node
> property allows applications to be migrated "close" to a resource.
>
> In preparation for providing a generic facility to optionally map an
> address range to its closest online node, or the node the range would
> represent were it to be onlined (target_node), up-level the core of
> acpi_map_pxm_to_online_node() to a generic mm/numa helper.
>
> Cc: Michal Hocko <mhocko@xxxxxxxx>
> Cc: "Rafael J. Wysocki" <rjw@xxxxxxxxxxxxx>
> Signed-off-by: Dan Williams <dan.j.williams@xxxxxxxxx>
> ---
> drivers/acpi/numa.c | 41 -----------------------------------------
> include/linux/acpi.h | 23 ++++++++++++++++++++++-
> include/linux/numa.h | 2 ++
> mm/mempolicy.c | 30 ++++++++++++++++++++++++++++++
> 4 files changed, 54 insertions(+), 42 deletions(-)
>
> diff --git a/drivers/acpi/numa.c b/drivers/acpi/numa.c
> index eadbf90e65d1..47b4969d9b93 100644
> --- a/drivers/acpi/numa.c
> +++ b/drivers/acpi/numa.c
> @@ -72,47 +72,6 @@ int acpi_map_pxm_to_node(int pxm)
> }
> EXPORT_SYMBOL(acpi_map_pxm_to_node);
>
> -/**
> - * acpi_map_pxm_to_online_node - Map proximity ID to online node
> - * @pxm: ACPI proximity ID
> - *
> - * This is similar to acpi_map_pxm_to_node(), but always returns an online
> - * node. When the mapped node from a given proximity ID is offline, it
> - * looks up the node distance table and returns the nearest online node.
> - *
> - * ACPI device drivers, which are called after the NUMA initialization has
> - * completed in the kernel, can call this interface to obtain their device
> - * NUMA topology from ACPI tables. Such drivers do not have to deal with
> - * offline nodes. A node may be offline when a device proximity ID is
> - * unique, SRAT memory entry does not exist, or NUMA is disabled, ex.
> - * "numa=off" on x86.
> - */
> -int acpi_map_pxm_to_online_node(int pxm)
> -{
> - int node, min_node;
> -
> - node = acpi_map_pxm_to_node(pxm);
> -
> - if (node == NUMA_NO_NODE)
> - node = 0;
> -
> - min_node = node;
> - if (!node_online(node)) {
> - int min_dist = INT_MAX, dist, n;
> -
> - for_each_online_node(n) {
> - dist = node_distance(node, n);
> - if (dist < min_dist) {
> - min_dist = dist;
> - min_node = n;
> - }
> - }
> - }
> -
> - return min_node;
> -}
> -EXPORT_SYMBOL(acpi_map_pxm_to_online_node);
> -
> static void __init
> acpi_table_print_srat_entry(struct acpi_subtable_header *header)
> {
> diff --git a/include/linux/acpi.h b/include/linux/acpi.h
> index 8b4e516bac00..aeedd09f2f71 100644
> --- a/include/linux/acpi.h
> +++ b/include/linux/acpi.h
> @@ -401,9 +401,30 @@ extern void acpi_osi_setup(char *str);
> extern bool acpi_osi_is_win8(void);
>
> #ifdef CONFIG_ACPI_NUMA
> -int acpi_map_pxm_to_online_node(int pxm);
> int acpi_map_pxm_to_node(int pxm);
> int acpi_get_node(acpi_handle handle);
> +
> +/**
> + * acpi_map_pxm_to_online_node - Map proximity ID to online node
> + * @pxm: ACPI proximity ID
> + *
> + * This is similar to acpi_map_pxm_to_node(), but always returns an online
> + * node. When the mapped node from a given proximity ID is offline, it
> + * looks up the node distance table and returns the nearest online node.
> + *
> + * ACPI device drivers, which are called after the NUMA initialization has
> + * completed in the kernel, can call this interface to obtain their device
> + * NUMA topology from ACPI tables. Such drivers do not have to deal with
> + * offline nodes. A node may be offline when a device proximity ID is
> + * unique, SRAT memory entry does not exist, or NUMA is disabled, ex.
> + * "numa=off" on x86.
> + */
> +static inline int acpi_map_pxm_to_online_node(int pxm)
> +{
> + int node = acpi_map_pxm_to_node(pxm);
> +
> + return numa_map_to_online_node(node);
> +}
> #else
> static inline int acpi_map_pxm_to_online_node(int pxm)
> {
> diff --git a/include/linux/numa.h b/include/linux/numa.h
> index 110b0e5d0fb0..4fd80f42be43 100644
> --- a/include/linux/numa.h
> +++ b/include/linux/numa.h
> @@ -13,4 +13,6 @@
>
> #define NUMA_NO_NODE (-1)
>
> +int numa_map_to_online_node(int node);
> +
> #endif /* _LINUX_NUMA_H */
> diff --git a/mm/mempolicy.c b/mm/mempolicy.c
> index 4ae967bcf954..e2d8dd21ce9d 100644
> --- a/mm/mempolicy.c
> +++ b/mm/mempolicy.c
> @@ -127,6 +127,36 @@ static struct mempolicy default_policy = {
>
> static struct mempolicy preferred_node_policy[MAX_NUMNODES];
>
> +/**
> + * numa_map_to_online_node - Find closest online node
> + * @nid: Node id to start the search
> + *
> + * Lookup the next closest node by distance if @nid is not online.
> + */
> +int numa_map_to_online_node(int node)
> +{
> + int min_node;
> +
> + if (node == NUMA_NO_NODE)
> + node = 0;
The ppc64 variant papr_scm_node return the NUMA_NO_NODE in this case.
Most of the mm helpers can handle with that value . So instead of
forcing node = 0, let the subsystem decide what to do with the
NUMA_NO_NODE value.?
> +
> + min_node = node;
> + if (!node_online(node)) {
> + int min_dist = INT_MAX, dist, n;
> +
> + for_each_online_node(n) {
> + dist = node_distance(node, n);
> + if (dist < min_dist) {
> + min_dist = dist;
> + min_node = n;
> + }
> + }
> + }
> +
> + return min_node;
> +}
> +EXPORT_SYMBOL_GPL(numa_map_to_online_node);
> +
> struct mempolicy *get_task_policy(struct task_struct *p)
> {
> struct mempolicy *pol = p->mempolicy;
Can we also switch papr_scm_node to numa_map_to_online_node()?
-aneesh
