[PATCH] cpuset: top_cpuset tracks hotplug changes to node_online_map

Change the list of memory nodes allowed to tasks in the top (root) nodeset to dynamically track what cpus are online, using a call to a cpuset hook from the memory hotplug code. Make this top cpus file read-only. On systems that have cpusets configured in their kernel, but that aren't actively using cpusets (for some distros, this covers the majority of systems) all tasks end up in the top cpuset. If that system does support memory hotplug, then these tasks cannot make use of memory nodes that are added after system boot, because the memory nodes are not allowed in the top cpuset. This is a surprising regression over earlier kernels that didn't have cpusets enabled. One key motivation for this change is to remain consistent with the behaviour for the top_cpuset's 'cpus', which is also read-only, and which automatically tracks the cpu_online_map. This change also has the minor benefit that it fixes a long standing, little noticed, minor bug in cpusets. The cpuset performance tweak to short circuit the cpuset_zone_allowed() check on systems with just a single cpuset (see 'number_of_cpusets', in linux/cpuset.h) meant that simply changing the 'mems' of the top_cpuset had no affect, even though the change (the write system call) appeared to succeed. With the following change, that write to the 'mems' file fails -EACCES, and the 'mems' file stubbornly refuses to be changed via user space writes. Thus no one should be mislead into thinking they've changed the top_cpusets's 'mems' when in affect they haven't. In order to keep the behaviour of cpusets consistent between systems actively making use of them and systems not using them, this patch changes the behaviour of the 'mems' file in the top (root) cpuset, making it read only, and making it automatically track the value of node_online_map. Thus tasks in the top cpuset will have automatic use of hot plugged memory nodes allowed by their cpuset. [akpm@osdl.org: build fix] [bunk@stusta.de: build fix] Signed-off-by: Paul Jackson <pj@sgi.com> Signed-off-by: Adrian Bunk <bunk@stusta.de> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] cpuset: top_cpuset tracks hotplug changes to node_online_map
Change the list of memory nodes allowed to tasks in the top (root) nodeset to dynamically track what cpus are online, using a call to a cpuset hook from the memory hotplug code. Make this top cpus file read-only. On systems that have cpusets configured in their kernel, but that aren't actively using cpusets (for some distros, this covers the majority of systems) all tasks end up in the top cpuset. If that system does support memory hotplug, then these tasks cannot make use of memory nodes that are added after system boot, because the memory nodes are not allowed in the top cpuset. This is a surprising regression over earlier kernels that didn't have cpusets enabled. One key motivation for this change is to remain consistent with the behaviour for the top_cpuset's 'cpus', which is also read-only, and which automatically tracks the cpu_online_map. This change also has the minor benefit that it fixes a long standing, little noticed, minor bug in cpusets. The cpuset performance tweak to short circuit the cpuset_zone_allowed() check on systems with just a single cpuset (see 'number_of_cpusets', in linux/cpuset.h) meant that simply changing the 'mems' of the top_cpuset had no affect, even though the change (the write system call) appeared to succeed. With the following change, that write to the 'mems' file fails -EACCES, and the 'mems' file stubbornly refuses to be changed via user space writes. Thus no one should be mislead into thinking they've changed the top_cpusets's 'mems' when in affect they haven't. In order to keep the behaviour of cpusets consistent between systems actively making use of them and systems not using them, this patch changes the behaviour of the 'mems' file in the top (root) cpuset, making it read only, and making it automatically track the value of node_online_map. Thus tasks in the top cpuset will have automatic use of hot plugged memory nodes allowed by their cpuset. [akpm@osdl.org: build fix] [bunk@stusta.de: build fix] Signed-off-by: Paul Jackson <pj@sgi.com> Signed-off-by: Adrian Bunk <bunk@stusta.de> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Paul Jackson · Linus Torvalds
1 parent af3ffa6758
Showing 4 changed files with 37 additions and 8 deletions Side-by-side Diff
Documentation/cpusets.txt
include/linux/cpuset.h
kernel/cpuset.c
mm/memory_hotplug.c
@@ -217,11 +217,11 @@
 to represent the cpuset hierarchy provides for a familiar permission
 and name space for cpusets, with a minimum of additional kernel code.
  
-The cpus file in the root (top_cpuset) cpuset is read-only.
-It automatically tracks the value of cpu_online_map, using a CPU
-hotplug notifier.  If and when memory nodes can be hotplugged,
-we expect to make the mems file in the root cpuset read-only
-as well, and have it track the value of node_online_map.
+The cpus and mems files in the root (top_cpuset) cpuset are
+read-only.  The cpus file automatically tracks the value of
+cpu_online_map using a CPU hotplug notifier, and the mems file
+automatically tracks the value of node_online_map using the
+cpuset_track_online_nodes() hook.
  
  
 1.4 What are exclusive cpusets ?
@@ -63,6 +63,8 @@
 	return current->flags & PF_SPREAD_SLAB;
 }
  
+extern void cpuset_track_online_nodes(void);
+
 #else /* !CONFIG_CPUSETS */
  
 static inline int cpuset_init_early(void) { return 0; }
@@ -125,6 +127,8 @@
 {
 	return 0;
 }
+
+static inline void cpuset_track_online_nodes(void) {}
  
 #endif /* !CONFIG_CPUSETS */
  
@@ -912,6 +912,10 @@
 	int fudge;
 	int retval;
  
+	/* top_cpuset.mems_allowed tracks node_online_map; it's read-only */
+	if (cs == &top_cpuset)
+		return -EACCES;
+
 	trialcs = *cs;
 	retval = nodelist_parse(buf, trialcs.mems_allowed);
 	if (retval < 0)
@@ -2042,9 +2046,8 @@
  * (of no affect) on systems that are actively using CPU hotplug
  * but making no active use of cpusets.
  *
- * This handles CPU hotplug (cpuhp) events.  If someday Memory
- * Nodes can be hotplugged (dynamically changing node_online_map)
- * then we should handle that too, perhaps in a similar way.
+ * This routine ensures that top_cpuset.cpus_allowed tracks
+ * cpu_online_map on each CPU hotplug (cpuhp) event.
  */
  
 #ifdef CONFIG_HOTPLUG_CPU
@@ -2060,6 +2063,25 @@
 	mutex_unlock(&manage_mutex);
  
 	return 0;
+}
+#endif
+
+/*
+ * Keep top_cpuset.mems_allowed tracking node_online_map.
+ * Call this routine anytime after you change node_online_map.
+ * See also the previous routine cpuset_handle_cpuhp().
+ */
+
+#ifdef CONFIG_MEMORY_HOTPLUG
+void cpuset_track_online_nodes()
+{
+	mutex_lock(&manage_mutex);
+	mutex_lock(&callback_mutex);
+
+	top_cpuset.mems_allowed = node_online_map;
+
+	mutex_unlock(&callback_mutex);
+	mutex_unlock(&manage_mutex);
 }
 #endif
  
@@ -21,6 +21,7 @@
 #include <linux/highmem.h>
 #include <linux/vmalloc.h>
 #include <linux/ioport.h>
+#include <linux/cpuset.h>
  
 #include <asm/tlbflush.h>
  
@@ -282,6 +283,8 @@
  
 	/* we online node here. we can't roll back from here. */
 	node_set_online(nid);
+
+	cpuset_track_online_nodes();
  
 	if (new_pgdat) {
 		ret = register_one_node(nid);
...	...	@@ -217,11 +217,11 @@
217	217	to represent the cpuset hierarchy provides for a familiar permission
218	218	and name space for cpusets, with a minimum of additional kernel code.
219	219
220		-The cpus file in the root (top_cpuset) cpuset is read-only.
221		-It automatically tracks the value of cpu_online_map, using a CPU
222		-hotplug notifier. If and when memory nodes can be hotplugged,
223		-we expect to make the mems file in the root cpuset read-only
224		-as well, and have it track the value of node_online_map.
	220	+The cpus and mems files in the root (top_cpuset) cpuset are
	221	+read-only. The cpus file automatically tracks the value of
	222	+cpu_online_map using a CPU hotplug notifier, and the mems file
	223	+automatically tracks the value of node_online_map using the
	224	+cpuset_track_online_nodes() hook.
225	225
226	226
227	227	1.4 What are exclusive cpusets ?
...	...	@@ -63,6 +63,8 @@
63	63	return current->flags & PF_SPREAD_SLAB;
64	64	}
65	65
	66	+extern void cpuset_track_online_nodes(void);
	67	+
66	68	#else /* !CONFIG_CPUSETS */
67	69
68	70	static inline int cpuset_init_early(void) { return 0; }
...	...	@@ -125,6 +127,8 @@
125	127	{
126	128	return 0;
127	129	}
	130	+
	131	+static inline void cpuset_track_online_nodes(void) {}
128	132
129	133	#endif /* !CONFIG_CPUSETS */
130	134
...	...	@@ -912,6 +912,10 @@
912	912	int fudge;
913	913	int retval;
914	914
	915	+ /* top_cpuset.mems_allowed tracks node_online_map; it's read-only */
	916	+ if (cs == &top_cpuset)
	917	+ return -EACCES;
	918	+
915	919	trialcs = *cs;
916	920	retval = nodelist_parse(buf, trialcs.mems_allowed);
917	921	if (retval < 0)
...	...	@@ -2042,9 +2046,8 @@
2042	2046	* (of no affect) on systems that are actively using CPU hotplug
2043	2047	* but making no active use of cpusets.
2044	2048	*
2045		- * This handles CPU hotplug (cpuhp) events. If someday Memory
2046		- * Nodes can be hotplugged (dynamically changing node_online_map)
2047		- * then we should handle that too, perhaps in a similar way.
	2049	+ * This routine ensures that top_cpuset.cpus_allowed tracks
	2050	+ * cpu_online_map on each CPU hotplug (cpuhp) event.
2048	2051	*/
2049	2052
2050	2053	#ifdef CONFIG_HOTPLUG_CPU
...	...	@@ -2060,6 +2063,25 @@
2060	2063	mutex_unlock(&manage_mutex);
2061	2064
2062	2065	return 0;
	2066	+}
	2067	+#endif
	2068	+
	2069	+/*
	2070	+ * Keep top_cpuset.mems_allowed tracking node_online_map.
	2071	+ * Call this routine anytime after you change node_online_map.
	2072	+ * See also the previous routine cpuset_handle_cpuhp().
	2073	+ */
	2074	+
	2075	+#ifdef CONFIG_MEMORY_HOTPLUG
	2076	+void cpuset_track_online_nodes()
	2077	+{
	2078	+ mutex_lock(&manage_mutex);
	2079	+ mutex_lock(&callback_mutex);
	2080	+
	2081	+ top_cpuset.mems_allowed = node_online_map;
	2082	+
	2083	+ mutex_unlock(&callback_mutex);
	2084	+ mutex_unlock(&manage_mutex);
2063	2085	}
2064	2086	#endif
2065	2087
...	...	@@ -21,6 +21,7 @@
21	21	#include <linux/highmem.h>
22	22	#include <linux/vmalloc.h>
23	23	#include <linux/ioport.h>
	24	+#include <linux/cpuset.h>
24	25
25	26	#include <asm/tlbflush.h>
26	27
...	...	@@ -282,6 +283,8 @@
282	283
283	284	/* we online node here. we can't roll back from here. */
284	285	node_set_online(nid);
	286	+
	287	+ cpuset_track_online_nodes();
285	288
286	289	if (new_pgdat) {
287	290	ret = register_one_node(nid);