All of lore.kernel.org
 help / color / mirror / Atom feed
* [PATCH 1/2] irqdomain: Simple NUMA awareness.
@ 2012-05-21  5:06 ` Paul Mundt
  0 siblings, 0 replies; 15+ messages in thread
From: Paul Mundt @ 2012-05-21  5:06 UTC (permalink / raw)
  To: Grant Likely; +Cc: linux-sh, linux-kernel, Paul Mundt

While common irqdesc allocation is node aware, the irqdomain code is not.

Presently we observe a number of regressions/inconsistencies on
NUMA-capable platforms:

- Platforms using irqdomains with legacy mappings, where the
  irq_descs are allocated node-local and the irqdomain data
  structure is not.

- Drivers implementing irqdomains will lose node locality
  regardless of the underlying struct device's node id.

This plugs in NUMA node id proliferation across the various allocation
callsites by way of_node_to_nid() node lookup. While of_node_to_nid()
does the right thing for OF-capable platforms it doesn't presently handle
the non-DT case. This is trivially dealt with by simply wraping in to
numa_node_id() unconditionally.

Signed-off-by: Paul Mundt <lethal@linux-sh.org>
---
 include/linux/of.h     |   16 ++++++++++------
 kernel/irq/irqdomain.c |   13 ++++++++-----
 2 files changed, 18 insertions(+), 11 deletions(-)

diff --git a/include/linux/of.h b/include/linux/of.h
index fa7fb1d..8650544 100644
--- a/include/linux/of.h
+++ b/include/linux/of.h
@@ -21,7 +21,7 @@
 #include <linux/kref.h>
 #include <linux/mod_devicetable.h>
 #include <linux/spinlock.h>
-
+#include <linux/topology.h>
 #include <asm/byteorder.h>
 #include <asm/errno.h>
 
@@ -158,11 +158,6 @@ static inline unsigned long of_read_ulong(const __be32 *cell, int size)
 
 #define OF_BAD_ADDR	((u64)-1)
 
-#ifndef of_node_to_nid
-static inline int of_node_to_nid(struct device_node *np) { return -1; }
-#define of_node_to_nid of_node_to_nid
-#endif
-
 extern struct device_node *of_find_node_by_name(struct device_node *from,
 	const char *name);
 #define for_each_node_by_name(dn, name) \
@@ -351,6 +346,15 @@ static inline int of_machine_is_compatible(const char *compat)
 #define of_match_node(_matches, _node)	NULL
 #endif /* CONFIG_OF */
 
+#ifndef of_node_to_nid
+static inline int of_node_to_nid(struct device_node *np)
+{
+	return numa_node_id();
+}
+
+#define of_node_to_nid of_node_to_nid
+#endif
+
 /**
  * of_property_read_bool - Findfrom a property
  * @np:		device node from which the property value is to be read.
diff --git a/kernel/irq/irqdomain.c b/kernel/irq/irqdomain.c
index 41c1564..9e99054 100644
--- a/kernel/irq/irqdomain.c
+++ b/kernel/irq/irqdomain.c
@@ -10,6 +10,7 @@
 #include <linux/mutex.h>
 #include <linux/of.h>
 #include <linux/of_address.h>
+#include <linux/topology.h>
 #include <linux/seq_file.h>
 #include <linux/slab.h>
 #include <linux/smp.h>
@@ -45,7 +46,8 @@ static struct irq_domain *irq_domain_alloc(struct device_node *of_node,
 {
 	struct irq_domain *domain;
 
-	domain = kzalloc(sizeof(*domain), GFP_KERNEL);
+	domain = kzalloc_node(sizeof(*domain), GFP_KERNEL,
+			      of_node_to_nid(of_node));
 	if (WARN_ON(!domain))
 		return NULL;
 
@@ -229,7 +231,8 @@ struct irq_domain *irq_domain_add_linear(struct device_node *of_node,
 	struct irq_domain *domain;
 	unsigned int *revmap;
 
-	revmap = kzalloc(sizeof(*revmap) * size, GFP_KERNEL);
+	revmap = kzalloc_node(sizeof(*revmap) * size, GFP_KERNEL,
+			      of_node_to_nid(of_node));
 	if (WARN_ON(!revmap))
 		return NULL;
 
@@ -367,7 +370,7 @@ unsigned int irq_create_direct_mapping(struct irq_domain *domain)
 	BUG_ON(domain = NULL);
 	WARN_ON(domain->revmap_type != IRQ_DOMAIN_MAP_NOMAP);
 
-	virq = irq_alloc_desc_from(1, 0);
+	virq = irq_alloc_desc_from(1, of_node_to_nid(domain->of_node));
 	if (!virq) {
 		pr_debug("create_direct virq allocation failed\n");
 		return 0;
@@ -433,9 +436,9 @@ unsigned int irq_create_mapping(struct irq_domain *domain,
 	hint = hwirq % nr_irqs;
 	if (hint = 0)
 		hint++;
-	virq = irq_alloc_desc_from(hint, 0);
+	virq = irq_alloc_desc_from(hint, of_node_to_nid(domain->of_node));
 	if (virq <= 0)
-		virq = irq_alloc_desc_from(1, 0);
+		virq = irq_alloc_desc_from(1, of_node_to_nid(domain->of_node));
 	if (virq <= 0) {
 		pr_debug("-> virq allocation failed\n");
 		return 0;
-- 
1.7.9.rc0.28.g0e1cf


^ permalink raw reply related	[flat|nested] 15+ messages in thread
* A couple of minor irqdomain fixes
@ 2012-06-15 18:13 Grant Likely
  2012-06-15 18:13 ` [PATCH 1/2] irqdomain: Simple NUMA awareness Grant Likely
  0 siblings, 1 reply; 15+ messages in thread
From: Grant Likely @ 2012-06-15 18:13 UTC (permalink / raw)
  To: linux-kernel

Here are a couple of irqdomain changes that I'm pushing out to
irqdomain/next right away.  The first give irqdomain some NUMA
awareness when allocating irqdescs, and the second is simply dead code
removal.

Grant Likely (1):
      irqdomain: Remove unnecessary test for IRQ_DOMAIN_MAP_LEGACY

Paul Mundt (1):
      irqdomain: Simple NUMA awareness.

 include/linux/of.h     |   15 ++++++++++-----
 kernel/irq/irqdomain.c |   16 +++++++++-------
 2 files changed, 19 insertions(+), 12 deletions(-)

^ permalink raw reply	[flat|nested] 15+ messages in thread
end of thread, other threads:[~2012-06-15 18:13 UTC | newest]

Thread overview: 15+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2012-05-21  5:06 [PATCH 1/2] irqdomain: Simple NUMA awareness Paul Mundt
2012-05-21  5:06 ` Paul Mundt
2012-05-21  5:06 ` [PATCH 2/2] irqdomain: Support for static IRQ mapping and association Paul Mundt
2012-05-21  5:06   ` Paul Mundt
2012-05-26  1:50   ` Grant Likely
2012-05-26  1:50     ` Grant Likely
2012-06-11  3:25     ` Paul Mundt
2012-06-11  3:25       ` Paul Mundt
2012-06-11  5:02       ` Grant Likely
2012-06-11  5:02         ` Grant Likely
2012-06-13  7:31         ` Paul Mundt
2012-06-13  7:31           ` Paul Mundt
2012-05-26  1:10 ` [PATCH 1/2] irqdomain: Simple NUMA awareness Grant Likely
2012-05-26  1:10   ` Grant Likely
  -- strict thread matches above, loose matches on Subject: below --
2012-06-15 18:13 A couple of minor irqdomain fixes Grant Likely
2012-06-15 18:13 ` [PATCH 1/2] irqdomain: Simple NUMA awareness Grant Likely

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.