[PATCH] Fold numa_maps into mempolicies.c

First discussed at http://marc.theaimsgroup.com/?t=113149255100001&r=1&w=2 - Use the check_range() in mempolicy.c to gather statistics. - Improve the numa_maps code in general and fix some comments. Signed-off-by: Christoph Lameter <clameter@sgi.com> Cc: Andi Kleen <ak@muc.de> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] Fold numa_maps into mempolicies.c
First discussed at http://marc.theaimsgroup.com/?t=113149255100001&r=1&w=2 - Use the check_range() in mempolicy.c to gather statistics. - Improve the numa_maps code in general and fix some comments. Signed-off-by: Christoph Lameter <clameter@sgi.com> Cc: Andi Kleen <ak@muc.de> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Christoph Lameter · Linus Torvalds
1 parent 38e35860db
Showing 2 changed files with 142 additions and 123 deletions Side-by-side Diff
fs/proc/task_mmu.c
mm/mempolicy.c
@@ -390,130 +390,13 @@
 };
  
 #ifdef CONFIG_NUMA
+extern int show_numa_map(struct seq_file *m, void *v);
  
-struct numa_maps {
-	unsigned long pages;
-	unsigned long anon;
-	unsigned long mapped;
-	unsigned long mapcount_max;
-	unsigned long node[MAX_NUMNODES];
-};
-
-/*
- * Calculate numa node maps for a vma
- */
-static struct numa_maps *get_numa_maps(struct vm_area_struct *vma)
-{
-	int i;
-	struct page *page;
-	unsigned long vaddr;
-	struct numa_maps *md = kmalloc(sizeof(struct numa_maps), GFP_KERNEL);
-
-	if (!md)
-		return NULL;
-	md->pages = 0;
-	md->anon = 0;
-	md->mapped = 0;
-	md->mapcount_max = 0;
-	for_each_node(i)
-		md->node[i] =0;
-
- 	for (vaddr = vma->vm_start; vaddr < vma->vm_end; vaddr += PAGE_SIZE) {
-		page = follow_page(vma, vaddr, 0);
-		if (page) {
-			int count = page_mapcount(page);
-
-			if (count)
-				md->mapped++;
-			if (count > md->mapcount_max)
-				md->mapcount_max = count;
-			md->pages++;
-			if (PageAnon(page))
-				md->anon++;
-			md->node[page_to_nid(page)]++;
-		}
-		cond_resched();
-	}
-	return md;
-}
-
-static int show_numa_map(struct seq_file *m, void *v)
-{
-	struct task_struct *task = m->private;
-	struct vm_area_struct *vma = v;
-	struct mempolicy *pol;
-	struct numa_maps *md;
-	struct zone **z;
-	int n;
-	int first;
-
-	if (!vma->vm_mm)
-		return 0;
-
-	md = get_numa_maps(vma);
-	if (!md)
-		return 0;
-
-	seq_printf(m, "%08lx", vma->vm_start);
-	pol = get_vma_policy(task, vma, vma->vm_start);
-	/* Print policy */
-	switch (pol->policy) {
-	case MPOL_PREFERRED:
-		seq_printf(m, " prefer=%d", pol->v.preferred_node);
-		break;
-	case MPOL_BIND:
-		seq_printf(m, " bind={");
-		first = 1;
-		for (z = pol->v.zonelist->zones; *z; z++) {
-
-			if (!first)
-				seq_putc(m, ',');
-			else
-				first = 0;
-			seq_printf(m, "%d/%s", (*z)->zone_pgdat->node_id,
-					(*z)->name);
-		}
-		seq_putc(m, '}');
-		break;
-	case MPOL_INTERLEAVE:
-		seq_printf(m, " interleave={");
-		first = 1;
-		for_each_node(n) {
-			if (node_isset(n, pol->v.nodes)) {
-				if (!first)
-					seq_putc(m,',');
-				else
-					first = 0;
-				seq_printf(m, "%d",n);
-			}
-		}
-		seq_putc(m, '}');
-		break;
-	default:
-		seq_printf(m," default");
-		break;
-	}
-	seq_printf(m, " MaxRef=%lu Pages=%lu Mapped=%lu",
-			md->mapcount_max, md->pages, md->mapped);
-	if (md->anon)
-		seq_printf(m," Anon=%lu",md->anon);
-
-	for_each_online_node(n) {
-		if (md->node[n])
-			seq_printf(m, " N%d=%lu", n, md->node[n]);
-	}
-	seq_putc(m, '\n');
-	kfree(md);
-	if (m->count < m->size)  /* vma is copied successfully */
-		m->version = (vma != get_gate_vma(task)) ? vma->vm_start : 0;
-	return 0;
-}
-
 struct seq_operations proc_pid_numa_maps_op = {
-	.start	= m_start,
-	.next	= m_next,
-	.stop	= m_stop,
-	.show	= show_numa_map
+        .start  = m_start,
+        .next   = m_next,
+        .stop   = m_stop,
+        .show   = show_numa_map
 };
 #endif
@@ -84,6 +84,8 @@
 #include <linux/compat.h>
 #include <linux/mempolicy.h>
 #include <linux/swap.h>
+#include <linux/seq_file.h>
+#include <linux/proc_fs.h>
  
 #include <asm/tlbflush.h>
 #include <asm/uaccess.h>
@@ -91,6 +93,7 @@
 /* Internal flags */
 #define MPOL_MF_DISCONTIG_OK (MPOL_MF_INTERNAL << 0)	/* Skip checks for continuous vmas */
 #define MPOL_MF_INVERT (MPOL_MF_INTERNAL << 1)		/* Invert check for nodemask */
+#define MPOL_MF_STATS (MPOL_MF_INTERNAL << 2)		/* Gather statistics */
  
 static kmem_cache_t *policy_cache;
 static kmem_cache_t *sn_cache;
@@ -228,6 +231,8 @@
 	}
 }
  
+static void gather_stats(struct page *, void *);
+
 /* Scan through pages checking if pages follow certain conditions. */
 static int check_pte_range(struct vm_area_struct *vma, pmd_t *pmd,
 		unsigned long addr, unsigned long end,
@@ -252,7 +257,9 @@
 		if (node_isset(nid, *nodes) == !!(flags & MPOL_MF_INVERT))
 			continue;
  
-		if (flags & (MPOL_MF_MOVE | MPOL_MF_MOVE_ALL))
+		if (flags & MPOL_MF_STATS)
+			gather_stats(page, private);
+		else if (flags & (MPOL_MF_MOVE | MPOL_MF_MOVE_ALL))
 			migrate_page_add(vma, page, private, flags);
 		else
 			break;
@@ -1459,5 +1466,133 @@
 void numa_policy_rebind(const nodemask_t *old, const nodemask_t *new)
 {
 	rebind_policy(current->mempolicy, old, new);
+}
+
+/*
+ * Display pages allocated per node and memory policy via /proc.
+ */
+
+static const char *policy_types[] = { "default", "prefer", "bind",
+				      "interleave" };
+
+/*
+ * Convert a mempolicy into a string.
+ * Returns the number of characters in buffer (if positive)
+ * or an error (negative)
+ */
+static inline int mpol_to_str(char *buffer, int maxlen, struct mempolicy *pol)
+{
+	char *p = buffer;
+	int l;
+	nodemask_t nodes;
+	int mode = pol ? pol->policy : MPOL_DEFAULT;
+
+	switch (mode) {
+	case MPOL_DEFAULT:
+		nodes_clear(nodes);
+		break;
+
+	case MPOL_PREFERRED:
+		nodes_clear(nodes);
+		node_set(pol->v.preferred_node, nodes);
+		break;
+
+	case MPOL_BIND:
+		get_zonemask(pol, &nodes);
+		break;
+
+	case MPOL_INTERLEAVE:
+		nodes = pol->v.nodes;
+		break;
+
+	default:
+		BUG();
+		return -EFAULT;
+	}
+
+	l = strlen(policy_types[mode]);
+ 	if (buffer + maxlen < p + l + 1)
+ 		return -ENOSPC;
+
+	strcpy(p, policy_types[mode]);
+	p += l;
+
+	if (!nodes_empty(nodes)) {
+		if (buffer + maxlen < p + 2)
+			return -ENOSPC;
+		*p++ = '=';
+	 	p += nodelist_scnprintf(p, buffer + maxlen - p, nodes);
+	}
+	return p - buffer;
+}
+
+struct numa_maps {
+	unsigned long pages;
+	unsigned long anon;
+	unsigned long mapped;
+	unsigned long mapcount_max;
+	unsigned long node[MAX_NUMNODES];
+};
+
+static void gather_stats(struct page *page, void *private)
+{
+	struct numa_maps *md = private;
+	int count = page_mapcount(page);
+
+	if (count)
+		md->mapped++;
+
+	if (count > md->mapcount_max)
+		md->mapcount_max = count;
+
+	md->pages++;
+
+	if (PageAnon(page))
+		md->anon++;
+
+	md->node[page_to_nid(page)]++;
+	cond_resched();
+}
+
+int show_numa_map(struct seq_file *m, void *v)
+{
+	struct task_struct *task = m->private;
+	struct vm_area_struct *vma = v;
+	struct numa_maps *md;
+	int n;
+	char buffer[50];
+
+	if (!vma->vm_mm)
+		return 0;
+
+	md = kzalloc(sizeof(struct numa_maps), GFP_KERNEL);
+	if (!md)
+		return 0;
+
+	check_pgd_range(vma, vma->vm_start, vma->vm_end,
+		    &node_online_map, MPOL_MF_STATS, md);
+
+	if (md->pages) {
+		mpol_to_str(buffer, sizeof(buffer),
+			    get_vma_policy(task, vma, vma->vm_start));
+
+		seq_printf(m, "%08lx %s pages=%lu mapped=%lu maxref=%lu",
+			   vma->vm_start, buffer, md->pages,
+			   md->mapped, md->mapcount_max);
+
+		if (md->anon)
+			seq_printf(m," anon=%lu",md->anon);
+
+		for_each_online_node(n)
+			if (md->node[n])
+				seq_printf(m, " N%d=%lu", n, md->node[n]);
+
+		seq_putc(m, '\n');
+	}
+	kfree(md);
+
+	if (m->count < m->size)
+		m->version = (vma != get_gate_vma(task)) ? vma->vm_start : 0;
+	return 0;
 }
...	...	@@ -390,130 +390,13 @@
390	390	};
391	391
392	392	#ifdef CONFIG_NUMA
	393	+extern int show_numa_map(struct seq_file m, void v);
393	394
394		-struct numa_maps {
395		- unsigned long pages;
396		- unsigned long anon;
397		- unsigned long mapped;
398		- unsigned long mapcount_max;
399		- unsigned long node[MAX_NUMNODES];
400		-};
401		-
402		-/*
403		- * Calculate numa node maps for a vma
404		- */
405		-static struct numa_maps get_numa_maps(struct vm_area_struct vma)
406		-{
407		- int i;
408		- struct page *page;
409		- unsigned long vaddr;
410		- struct numa_maps *md = kmalloc(sizeof(struct numa_maps), GFP_KERNEL);
411		-
412		- if (!md)
413		- return NULL;
414		- md->pages = 0;
415		- md->anon = 0;
416		- md->mapped = 0;
417		- md->mapcount_max = 0;
418		- for_each_node(i)
419		- md->node[i] =0;
420		-
421		- for (vaddr = vma->vm_start; vaddr < vma->vm_end; vaddr += PAGE_SIZE) {
422		- page = follow_page(vma, vaddr, 0);
423		- if (page) {
424		- int count = page_mapcount(page);
425		-
426		- if (count)
427		- md->mapped++;
428		- if (count > md->mapcount_max)
429		- md->mapcount_max = count;
430		- md->pages++;
431		- if (PageAnon(page))
432		- md->anon++;
433		- md->node[page_to_nid(page)]++;
434		- }
435		- cond_resched();
436		- }
437		- return md;
438		-}
439		-
440		-static int show_numa_map(struct seq_file m, void v)
441		-{
442		- struct task_struct *task = m->private;
443		- struct vm_area_struct *vma = v;
444		- struct mempolicy *pol;
445		- struct numa_maps *md;
446		- struct zone **z;
447		- int n;
448		- int first;
449		-
450		- if (!vma->vm_mm)
451		- return 0;
452		-
453		- md = get_numa_maps(vma);
454		- if (!md)
455		- return 0;
456		-
457		- seq_printf(m, "%08lx", vma->vm_start);
458		- pol = get_vma_policy(task, vma, vma->vm_start);
459		- /* Print policy */
460		- switch (pol->policy) {
461		- case MPOL_PREFERRED:
462		- seq_printf(m, " prefer=%d", pol->v.preferred_node);
463		- break;
464		- case MPOL_BIND:
465		- seq_printf(m, " bind={");
466		- first = 1;
467		- for (z = pol->v.zonelist->zones; *z; z++) {
468		-
469		- if (!first)
470		- seq_putc(m, ',');
471		- else
472		- first = 0;
473		- seq_printf(m, "%d/%s", (*z)->zone_pgdat->node_id,
474		- (*z)->name);
475		- }
476		- seq_putc(m, '}');
477		- break;
478		- case MPOL_INTERLEAVE:
479		- seq_printf(m, " interleave={");
480		- first = 1;
481		- for_each_node(n) {
482		- if (node_isset(n, pol->v.nodes)) {
483		- if (!first)
484		- seq_putc(m,',');
485		- else
486		- first = 0;
487		- seq_printf(m, "%d",n);
488		- }
489		- }
490		- seq_putc(m, '}');
491		- break;
492		- default:
493		- seq_printf(m," default");
494		- break;
495		- }
496		- seq_printf(m, " MaxRef=%lu Pages=%lu Mapped=%lu",
497		- md->mapcount_max, md->pages, md->mapped);
498		- if (md->anon)
499		- seq_printf(m," Anon=%lu",md->anon);
500		-
501		- for_each_online_node(n) {
502		- if (md->node[n])
503		- seq_printf(m, " N%d=%lu", n, md->node[n]);
504		- }
505		- seq_putc(m, '\n');
506		- kfree(md);
507		- if (m->count < m->size) /* vma is copied successfully */
508		- m->version = (vma != get_gate_vma(task)) ? vma->vm_start : 0;
509		- return 0;
510		-}
511		-
512	395	struct seq_operations proc_pid_numa_maps_op = {
513		- .start = m_start,
514		- .next = m_next,
515		- .stop = m_stop,
516		- .show = show_numa_map
	396	+ .start = m_start,
	397	+ .next = m_next,
	398	+ .stop = m_stop,
	399	+ .show = show_numa_map
517	400	};
518	401	#endif
...	...	@@ -84,6 +84,8 @@
84	84	#include <linux/compat.h>
85	85	#include <linux/mempolicy.h>
86	86	#include <linux/swap.h>
	87	+#include <linux/seq_file.h>
	88	+#include <linux/proc_fs.h>
87	89
88	90	#include <asm/tlbflush.h>
89	91	#include <asm/uaccess.h>
...	...	@@ -91,6 +93,7 @@
91	93	/* Internal flags */
92	94	#define MPOL_MF_DISCONTIG_OK (MPOL_MF_INTERNAL << 0) /* Skip checks for continuous vmas */
93	95	#define MPOL_MF_INVERT (MPOL_MF_INTERNAL << 1) /* Invert check for nodemask */
	96	+#define MPOL_MF_STATS (MPOL_MF_INTERNAL << 2) /* Gather statistics */
94	97
95	98	static kmem_cache_t *policy_cache;
96	99	static kmem_cache_t *sn_cache;
...	...	@@ -228,6 +231,8 @@
228	231	}
229	232	}
230	233
	234	+static void gather_stats(struct page , void );
	235	+
231	236	/* Scan through pages checking if pages follow certain conditions. */
232	237	static int check_pte_range(struct vm_area_struct vma, pmd_t pmd,
233	238	unsigned long addr, unsigned long end,
...	...	@@ -252,7 +257,9 @@
252	257	if (node_isset(nid, *nodes) == !!(flags & MPOL_MF_INVERT))
253	258	continue;
254	259
255		- if (flags & (MPOL_MF_MOVE \| MPOL_MF_MOVE_ALL))
	260	+ if (flags & MPOL_MF_STATS)
	261	+ gather_stats(page, private);
	262	+ else if (flags & (MPOL_MF_MOVE \| MPOL_MF_MOVE_ALL))
256	263	migrate_page_add(vma, page, private, flags);
257	264	else
258	265	break;
...	...	@@ -1459,5 +1466,133 @@
1459	1466	void numa_policy_rebind(const nodemask_t old, const nodemask_t new)
1460	1467	{
1461	1468	rebind_policy(current->mempolicy, old, new);
	1469	+}
	1470	+
	1471	+/*
	1472	+ * Display pages allocated per node and memory policy via /proc.
	1473	+ */
	1474	+
	1475	+static const char *policy_types[] = { "default", "prefer", "bind",
	1476	+ "interleave" };
	1477	+
	1478	+/*
	1479	+ * Convert a mempolicy into a string.
	1480	+ * Returns the number of characters in buffer (if positive)
	1481	+ * or an error (negative)
	1482	+ */
	1483	+static inline int mpol_to_str(char buffer, int maxlen, struct mempolicy pol)
	1484	+{
	1485	+ char *p = buffer;
	1486	+ int l;
	1487	+ nodemask_t nodes;
	1488	+ int mode = pol ? pol->policy : MPOL_DEFAULT;
	1489	+
	1490	+ switch (mode) {
	1491	+ case MPOL_DEFAULT:
	1492	+ nodes_clear(nodes);
	1493	+ break;
	1494	+
	1495	+ case MPOL_PREFERRED:
	1496	+ nodes_clear(nodes);
	1497	+ node_set(pol->v.preferred_node, nodes);
	1498	+ break;
	1499	+
	1500	+ case MPOL_BIND:
	1501	+ get_zonemask(pol, &nodes);
	1502	+ break;
	1503	+
	1504	+ case MPOL_INTERLEAVE:
	1505	+ nodes = pol->v.nodes;
	1506	+ break;
	1507	+
	1508	+ default:
	1509	+ BUG();
	1510	+ return -EFAULT;
	1511	+ }
	1512	+
	1513	+ l = strlen(policy_types[mode]);
	1514	+ if (buffer + maxlen < p + l + 1)
	1515	+ return -ENOSPC;
	1516	+
	1517	+ strcpy(p, policy_types[mode]);
	1518	+ p += l;
	1519	+
	1520	+ if (!nodes_empty(nodes)) {
	1521	+ if (buffer + maxlen < p + 2)
	1522	+ return -ENOSPC;
	1523	+ *p++ = '=';
	1524	+ p += nodelist_scnprintf(p, buffer + maxlen - p, nodes);
	1525	+ }
	1526	+ return p - buffer;
	1527	+}
	1528	+
	1529	+struct numa_maps {
	1530	+ unsigned long pages;
	1531	+ unsigned long anon;
	1532	+ unsigned long mapped;
	1533	+ unsigned long mapcount_max;
	1534	+ unsigned long node[MAX_NUMNODES];
	1535	+};
	1536	+
	1537	+static void gather_stats(struct page page, void private)
	1538	+{
	1539	+ struct numa_maps *md = private;
	1540	+ int count = page_mapcount(page);
	1541	+
	1542	+ if (count)
	1543	+ md->mapped++;
	1544	+
	1545	+ if (count > md->mapcount_max)
	1546	+ md->mapcount_max = count;
	1547	+
	1548	+ md->pages++;
	1549	+
	1550	+ if (PageAnon(page))
	1551	+ md->anon++;
	1552	+
	1553	+ md->node[page_to_nid(page)]++;
	1554	+ cond_resched();
	1555	+}
	1556	+
	1557	+int show_numa_map(struct seq_file m, void v)
	1558	+{
	1559	+ struct task_struct *task = m->private;
	1560	+ struct vm_area_struct *vma = v;
	1561	+ struct numa_maps *md;
	1562	+ int n;
	1563	+ char buffer[50];
	1564	+
	1565	+ if (!vma->vm_mm)
	1566	+ return 0;
	1567	+
	1568	+ md = kzalloc(sizeof(struct numa_maps), GFP_KERNEL);
	1569	+ if (!md)
	1570	+ return 0;
	1571	+
	1572	+ check_pgd_range(vma, vma->vm_start, vma->vm_end,
	1573	+ &node_online_map, MPOL_MF_STATS, md);
	1574	+
	1575	+ if (md->pages) {
	1576	+ mpol_to_str(buffer, sizeof(buffer),
	1577	+ get_vma_policy(task, vma, vma->vm_start));
	1578	+
	1579	+ seq_printf(m, "%08lx %s pages=%lu mapped=%lu maxref=%lu",
	1580	+ vma->vm_start, buffer, md->pages,
	1581	+ md->mapped, md->mapcount_max);
	1582	+
	1583	+ if (md->anon)
	1584	+ seq_printf(m," anon=%lu",md->anon);
	1585	+
	1586	+ for_each_online_node(n)
	1587	+ if (md->node[n])
	1588	+ seq_printf(m, " N%d=%lu", n, md->node[n]);
	1589	+
	1590	+ seq_putc(m, '\n');
	1591	+ }
	1592	+ kfree(md);
	1593	+
	1594	+ if (m->count < m->size)
	1595	+ m->version = (vma != get_gate_vma(task)) ? vma->vm_start : 0;
	1596	+ return 0;
1462	1597	}