[PATCH] add /proc/pid/smaps

Add a "smaps" entry to /proc/pid: show howmuch memory is resident in each
mapping.

People that want to perform a memory consumption analysing can use it
mainly if someone needs to figure out which libraries can be reduced for
embedded systems.  So the new features are the physical size of shared and
clean [or dirty]; private and clean [or dirty].

Take a look the example below:

# cat /proc/4576/smaps

08048000-080dc000 r-xp /bin/bash
Size:               592 KB
Rss:                500 KB
Shared_Clean:       500 KB
Shared_Dirty:         0 KB
Private_Clean:        0 KB
Private_Dirty:        0 KB
080dc000-080e2000 rw-p /bin/bash
Size:                24 KB
Rss:                 24 KB
Shared_Clean:         0 KB
Shared_Dirty:         0 KB
Private_Clean:        0 KB
Private_Dirty:       24 KB
080e2000-08116000 rw-p
Size:               208 KB
Rss:                208 KB
Shared_Clean:         0 KB
Shared_Dirty:         0 KB
Private_Clean:        0 KB
Private_Dirty:      208 KB
b7e2b000-b7e34000 r-xp /lib/tls/libnss_files-2.3.2.so
Size:                36 KB
Rss:                 12 KB
Shared_Clean:        12 KB
Shared_Dirty:         0 KB
Private_Clean:        0 KB
Private_Dirty:        0 KB
...

(Includes a cleanup from "Richard Purdie" <rpurdie@rpsys.net>)

From: Torsten Foertsch <torsten.foertsch@gmx.net>

show_smap calls first show_map and then prints its additional information to
the seq_file.  show_map checks if all it has to print fits into the buffer and
if yes marks the current vma as written.  While that is correct for show_map
it is not for show_smap.  Here the vma should be marked as written only after
the additional information is also written.

The attached patch cures the problem.  It moves the functionality of the
show_map function to a new function show_map_internal that is called with an
additional struct mem_size_stats* argument.  Then show_map calls
show_map_internal with NULL as struct mem_size_stats* whereas show_smap calls
it with a real pointer.  Now the final

	if (m->count < m->size)  /* vma is copied successfully */
		m->version = (vma != get_gate_vma(task))? vma->vm_start: 0;

is done only if the whole entry fits into the buffer.

Signed-off-by: Hugh Dickins <hugh@veritas.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

3 files changed, 245 insertions, 42 deletions

diff --git a/Documentation/filesystems/proc.txt b/Documentation/filesystems/proc.txt
index 6c98f2bd421e79..5024ba7a592c06 100644
--- a/Documentation/filesystems/proc.txt
+++ b/Documentation/filesystems/proc.txt
@@ -133,6 +133,7 @@ Table 1-1: Process specific entries in /proc
  statm   Process memory status information              
  status  Process status in human readable form          
  wchan   If CONFIG_KALLSYMS is set, a pre-decoded wchan
+ smaps	 Extension based on maps, presenting the rss size for each mapped file
 ..............................................................................
 
 For example, to get the status information of a process, all you have to do is
diff --git a/fs/proc/base.c b/fs/proc/base.c
index b796bf90a0b19a..520978e49e9247 100644
--- a/fs/proc/base.c
+++ b/fs/proc/base.c
@@ -11,6 +11,40 @@
  *  go into icache. We cache the reference to task_struct upon lookup too.
  *  Eventually it should become a filesystem in its own. We don't use the
  *  rest of procfs anymore.
+ *
+ *
+ *  Changelog:
+ *  17-Jan-2005
+ *  Allan Bezerra
+ *  Bruna Moreira <bruna.moreira@indt.org.br>
+ *  Edjard Mota <edjard.mota@indt.org.br>
+ *  Ilias Biris <ilias.biris@indt.org.br>
+ *  Mauricio Lin <mauricio.lin@indt.org.br>
+ *
+ *  Embedded Linux Lab - 10LE Instituto Nokia de Tecnologia - INdT
+ *
+ *  A new process specific entry (smaps) included in /proc. It shows the
+ *  size of rss for each memory area. The maps entry lacks information
+ *  about physical memory size (rss) for each mapped file, i.e.,
+ *  rss information for executables and library files.
+ *  This additional information is useful for any tools that need to know
+ *  about physical memory consumption for a process specific library.
+ *
+ *  Changelog:
+ *  21-Feb-2005
+ *  Embedded Linux Lab - 10LE Instituto Nokia de Tecnologia - INdT
+ *  Pud inclusion in the page table walking.
+ *
+ *  ChangeLog:
+ *  10-Mar-2005
+ *  10LE Instituto Nokia de Tecnologia - INdT:
+ *  A better way to walks through the page table as suggested by Hugh Dickins.
+ *
+ *  Simo Piiroinen <simo.piiroinen@nokia.com>:
+ *  Smaps information related to shared, private, clean and dirty pages.
+ *
+ *  Paul Mundt <paul.mundt@nokia.com>:
+ *  Overall revision about smaps.
  */
 
 #include <asm/uaccess.h>
@@ -68,6 +102,7 @@ enum pid_directory_inos {
 	PROC_TGID_NUMA_MAPS,
 	PROC_TGID_MOUNTS,
 	PROC_TGID_WCHAN,
+	PROC_TGID_SMAPS,
 #ifdef CONFIG_SCHEDSTATS
 	PROC_TGID_SCHEDSTAT,
 #endif
@@ -106,6 +141,7 @@ enum pid_directory_inos {
 	PROC_TID_NUMA_MAPS,
 	PROC_TID_MOUNTS,
 	PROC_TID_WCHAN,
+	PROC_TID_SMAPS,
 #ifdef CONFIG_SCHEDSTATS
 	PROC_TID_SCHEDSTAT,
 #endif
@@ -157,6 +193,7 @@ static struct pid_entry tgid_base_stuff[] = {
 	E(PROC_TGID_ROOT,      "root",    S_IFLNK|S_IRWXUGO),
 	E(PROC_TGID_EXE,       "exe",     S_IFLNK|S_IRWXUGO),
 	E(PROC_TGID_MOUNTS,    "mounts",  S_IFREG|S_IRUGO),
+	E(PROC_TGID_SMAPS,     "smaps",   S_IFREG|S_IRUGO),
 #ifdef CONFIG_SECURITY
 	E(PROC_TGID_ATTR,      "attr",    S_IFDIR|S_IRUGO|S_IXUGO),
 #endif
@@ -196,6 +233,7 @@ static struct pid_entry tid_base_stuff[] = {
 	E(PROC_TID_ROOT,       "root",    S_IFLNK|S_IRWXUGO),
 	E(PROC_TID_EXE,        "exe",     S_IFLNK|S_IRWXUGO),
 	E(PROC_TID_MOUNTS,     "mounts",  S_IFREG|S_IRUGO),
+	E(PROC_TID_SMAPS,      "smaps",   S_IFREG|S_IRUGO),
 #ifdef CONFIG_SECURITY
 	E(PROC_TID_ATTR,       "attr",    S_IFDIR|S_IRUGO|S_IXUGO),
 #endif
@@ -544,6 +582,25 @@ static struct file_operations proc_numa_maps_operations = {
 };
 #endif
 
+extern struct seq_operations proc_pid_smaps_op;
+static int smaps_open(struct inode *inode, struct file *file)
+{
+	struct task_struct *task = proc_task(inode);
+	int ret = seq_open(file, &proc_pid_smaps_op);
+	if (!ret) {
+		struct seq_file *m = file->private_data;
+		m->private = task;
+	}
+	return ret;
+}
+
+static struct file_operations proc_smaps_operations = {
+	.open		= smaps_open,
+	.read		= seq_read,
+	.llseek		= seq_lseek,
+	.release	= seq_release,
+};
+
 extern struct seq_operations mounts_op;
 static int mounts_open(struct inode *inode, struct file *file)
 {
@@ -1574,6 +1631,10 @@ static struct dentry *proc_pident_lookup(struct inode *dir,
 		case PROC_TGID_MOUNTS:
 			inode->i_fop = &proc_mounts_operations;
 			break;
+		case PROC_TID_SMAPS:
+		case PROC_TGID_SMAPS:
+			inode->i_fop = &proc_smaps_operations;
+			break;
 #ifdef CONFIG_SECURITY
 		case PROC_TID_ATTR:
 			inode->i_nlink = 2;
diff --git a/fs/proc/task_mmu.c b/fs/proc/task_mmu.c
index 64e84cadfa3c13..c7ef3e48e35bae 100644
--- a/fs/proc/task_mmu.c
+++ b/fs/proc/task_mmu.c
@@ -2,10 +2,13 @@
 #include <linux/hugetlb.h>
 #include <linux/mount.h>
 #include <linux/seq_file.h>
+#include <linux/highmem.h>
 #include <linux/pagemap.h>
 #include <linux/mempolicy.h>
+
 #include <asm/elf.h>
 #include <asm/uaccess.h>
+#include <asm/tlbflush.h>
 #include "internal.h"
 
 char *task_mem(struct mm_struct *mm, char *buffer)
@@ -89,49 +92,58 @@ static void pad_len_spaces(struct seq_file *m, int len)
 	seq_printf(m, "%*c", len, ' ');
 }
 
-static int show_map(struct seq_file *m, void *v)
+struct mem_size_stats
+{
+	unsigned long resident;
+	unsigned long shared_clean;
+	unsigned long shared_dirty;
+	unsigned long private_clean;
+	unsigned long private_dirty;
+};
+
+static int show_map_internal(struct seq_file *m, void *v, struct mem_size_stats *mss)
 {
 	struct task_struct *task = m->private;
-	struct vm_area_struct *map = v;
-	struct mm_struct *mm = map->vm_mm;
-	struct file *file = map->vm_file;
-	int flags = map->vm_flags;
+	struct vm_area_struct *vma = v;
+	struct mm_struct *mm = vma->vm_mm;
+	struct file *file = vma->vm_file;
+	int flags = vma->vm_flags;
 	unsigned long ino = 0;
 	dev_t dev = 0;
 	int len;
 
 	if (file) {
-		struct inode *inode = map->vm_file->f_dentry->d_inode;
+		struct inode *inode = vma->vm_file->f_dentry->d_inode;
 		dev = inode->i_sb->s_dev;
 		ino = inode->i_ino;
 	}
 
 	seq_printf(m, "%08lx-%08lx %c%c%c%c %08lx %02x:%02x %lu %n",
-			map->vm_start,
-			map->vm_end,
+			vma->vm_start,
+			vma->vm_end,
 			flags & VM_READ ? 'r' : '-',
 			flags & VM_WRITE ? 'w' : '-',
 			flags & VM_EXEC ? 'x' : '-',
 			flags & VM_MAYSHARE ? 's' : 'p',
-			map->vm_pgoff << PAGE_SHIFT,
+			vma->vm_pgoff << PAGE_SHIFT,
 			MAJOR(dev), MINOR(dev), ino, &len);
 
 	/*
 	 * Print the dentry name for named mappings, and a
 	 * special [heap] marker for the heap:
 	 */
-	if (map->vm_file) {
+	if (file) {
 		pad_len_spaces(m, len);
-		seq_path(m, file->f_vfsmnt, file->f_dentry, "");
+		seq_path(m, file->f_vfsmnt, file->f_dentry, "\n");
 	} else {
 		if (mm) {
-			if (map->vm_start <= mm->start_brk &&
-						map->vm_end >= mm->brk) {
+			if (vma->vm_start <= mm->start_brk &&
+						vma->vm_end >= mm->brk) {
 				pad_len_spaces(m, len);
 				seq_puts(m, "[heap]");
 			} else {
-				if (map->vm_start <= mm->start_stack &&
-					map->vm_end >= mm->start_stack) {
+				if (vma->vm_start <= mm->start_stack &&
+					vma->vm_end >= mm->start_stack) {
 
 					pad_len_spaces(m, len);
 					seq_puts(m, "[stack]");
@@ -143,24 +155,146 @@ static int show_map(struct seq_file *m, void *v)
 		}
 	}
 	seq_putc(m, '\n');
-	if (m->count < m->size)  /* map is copied successfully */
-		m->version = (map != get_gate_vma(task))? map->vm_start: 0;
+
+	if (mss)
+		seq_printf(m,
+			   "Size:          %8lu kB\n"
+			   "Rss:           %8lu kB\n"
+			   "Shared_Clean:  %8lu kB\n"
+			   "Shared_Dirty:  %8lu kB\n"
+			   "Private_Clean: %8lu kB\n"
+			   "Private_Dirty: %8lu kB\n",
+			   (vma->vm_end - vma->vm_start) >> 10,
+			   mss->resident >> 10,
+			   mss->shared_clean  >> 10,
+			   mss->shared_dirty  >> 10,
+			   mss->private_clean >> 10,
+			   mss->private_dirty >> 10);
+
+	if (m->count < m->size)  /* vma is copied successfully */
+		m->version = (vma != get_gate_vma(task))? vma->vm_start: 0;
 	return 0;
 }
 
+static int show_map(struct seq_file *m, void *v)
+{
+	return show_map_internal(m, v, 0);
+}
+
+static void smaps_pte_range(struct vm_area_struct *vma, pmd_t *pmd,
+				unsigned long addr, unsigned long end,
+				struct mem_size_stats *mss)
+{
+	pte_t *pte, ptent;
+	unsigned long pfn;
+	struct page *page;
+
+	pte = pte_offset_map(pmd, addr);
+	do {
+		ptent = *pte;
+		if (pte_none(ptent) || !pte_present(ptent))
+			continue;
+
+		mss->resident += PAGE_SIZE;
+		pfn = pte_pfn(ptent);
+		if (!pfn_valid(pfn))
+			continue;
+
+		page = pfn_to_page(pfn);
+		if (page_count(page) >= 2) {
+			if (pte_dirty(ptent))
+				mss->shared_dirty += PAGE_SIZE;
+			else
+				mss->shared_clean += PAGE_SIZE;
+		} else {
+			if (pte_dirty(ptent))
+				mss->private_dirty += PAGE_SIZE;
+			else
+				mss->private_clean += PAGE_SIZE;
+		}
+	} while (pte++, addr += PAGE_SIZE, addr != end);
+	pte_unmap(pte - 1);
+	cond_resched_lock(&vma->vm_mm->page_table_lock);
+}
+
+static inline void smaps_pmd_range(struct vm_area_struct *vma, pud_t *pud,
+				unsigned long addr, unsigned long end,
+				struct mem_size_stats *mss)
+{
+	pmd_t *pmd;
+	unsigned long next;
+
+	pmd = pmd_offset(pud, addr);
+	do {
+		next = pmd_addr_end(addr, end);
+		if (pmd_none_or_clear_bad(pmd))
+			continue;
+		smaps_pte_range(vma, pmd, addr, next, mss);
+	} while (pmd++, addr = next, addr != end);
+}
+
+static inline void smaps_pud_range(struct vm_area_struct *vma, pgd_t *pgd,
+				unsigned long addr, unsigned long end,
+				struct mem_size_stats *mss)
+{
+	pud_t *pud;
+	unsigned long next;
+
+	pud = pud_offset(pgd, addr);
+	do {
+		next = pud_addr_end(addr, end);
+		if (pud_none_or_clear_bad(pud))
+			continue;
+		smaps_pmd_range(vma, pud, addr, next, mss);
+	} while (pud++, addr = next, addr != end);
+}
+
+static inline void smaps_pgd_range(struct vm_area_struct *vma,
+				unsigned long addr, unsigned long end,
+				struct mem_size_stats *mss)
+{
+	pgd_t *pgd;
+	unsigned long next;
+
+	pgd = pgd_offset(vma->vm_mm, addr);
+	do {
+		next = pgd_addr_end(addr, end);
+		if (pgd_none_or_clear_bad(pgd))
+			continue;
+		smaps_pud_range(vma, pgd, addr, next, mss);
+	} while (pgd++, addr = next, addr != end);
+}
+
+static int show_smap(struct seq_file *m, void *v)
+{
+	struct vm_area_struct *vma = v;
+	struct mm_struct *mm = vma->vm_mm;
+	struct mem_size_stats mss;
+
+	memset(&mss, 0, sizeof mss);
+
+	if (mm) {
+		spin_lock(&mm->page_table_lock);
+		smaps_pgd_range(vma, vma->vm_start, vma->vm_end, &mss);
+		spin_unlock(&mm->page_table_lock);
+	}
+
+	return show_map_internal(m, v, &mss);
+}
+
 static void *m_start(struct seq_file *m, loff_t *pos)
 {
 	struct task_struct *task = m->private;
 	unsigned long last_addr = m->version;
 	struct mm_struct *mm;
-	struct vm_area_struct *map, *tail_map;
+	struct vm_area_struct *vma, *tail_vma;
 	loff_t l = *pos;
 
 	/*
 	 * We remember last_addr rather than next_addr to hit with
 	 * mmap_cache most of the time. We have zero last_addr at
-	 * the begining and also after lseek. We will have -1 last_addr
-	 * after the end of the maps.
+	 * the beginning and also after lseek. We will have -1 last_addr
+	 * after the end of the vmas.
 	 */
 
 	if (last_addr == -1UL)
@@ -170,47 +304,47 @@ static void *m_start(struct seq_file *m, loff_t *pos)
 	if (!mm)
 		return NULL;
 
-	tail_map = get_gate_vma(task);
+	tail_vma = get_gate_vma(task);
 	down_read(&mm->mmap_sem);
 
 	/* Start with last addr hint */
-	if (last_addr && (map = find_vma(mm, last_addr))) {
-		map = map->vm_next;
+	if (last_addr && (vma = find_vma(mm, last_addr))) {
+		vma = vma->vm_next;
 		goto out;
 	}
 
 	/*
-	 * Check the map index is within the range and do
+	 * Check the vma index is within the range and do
 	 * sequential scan until m_index.
 	 */
-	map = NULL;
+	vma = NULL;
 	if ((unsigned long)l < mm->map_count) {
-		map = mm->mmap;
-		while (l-- && map)
-			map = map->vm_next;
+		vma = mm->mmap;
+		while (l-- && vma)
+			vma = vma->vm_next;
 		goto out;
 	}
 
 	if (l != mm->map_count)
-		tail_map = NULL; /* After gate map */
+		tail_vma = NULL; /* After gate vma */
 
 out:
-	if (map)
-		return map;
+	if (vma)
+		return vma;
 
-	/* End of maps has reached */
-	m->version = (tail_map != NULL)? 0: -1UL;
+	/* End of vmas has been reached */
+	m->version = (tail_vma != NULL)? 0: -1UL;
 	up_read(&mm->mmap_sem);
 	mmput(mm);
-	return tail_map;
+	return tail_vma;
 }
 
 static void m_stop(struct seq_file *m, void *v)
 {
 	struct task_struct *task = m->private;
-	struct vm_area_struct *map = v;
-	if (map && map != get_gate_vma(task)) {
-		struct mm_struct *mm = map->vm_mm;
+	struct vm_area_struct *vma = v;
+	if (vma && vma != get_gate_vma(task)) {
+		struct mm_struct *mm = vma->vm_mm;
 		up_read(&mm->mmap_sem);
 		mmput(mm);
 	}
@@ -219,14 +353,14 @@ static void m_stop(struct seq_file *m, void *v)
 static void *m_next(struct seq_file *m, void *v, loff_t *pos)
 {
 	struct task_struct *task = m->private;
-	struct vm_area_struct *map = v;
-	struct vm_area_struct *tail_map = get_gate_vma(task);
+	struct vm_area_struct *vma = v;
+	struct vm_area_struct *tail_vma = get_gate_vma(task);
 
 	(*pos)++;
-	if (map && (map != tail_map) && map->vm_next)
-		return map->vm_next;
+	if (vma && (vma != tail_vma) && vma->vm_next)
+		return vma->vm_next;
 	m_stop(m, v);
-	return (map != tail_map)? tail_map: NULL;
+	return (vma != tail_vma)? tail_vma: NULL;
 }
 
 struct seq_operations proc_pid_maps_op = {
@@ -236,6 +370,13 @@ struct seq_operations proc_pid_maps_op = {
 	.show	= show_map
 };
 
+struct seq_operations proc_pid_smaps_op = {
+	.start	= m_start,
+	.next	= m_next,
+	.stop	= m_stop,
+	.show	= show_smap
+};
+
 #ifdef CONFIG_NUMA
 
 struct numa_maps {