proc: pagemap: Hold mmap_sem during page walk

In initial design, walk_page_range() was designed just for walking page
table and it didn't require mmap_sem.  Now, find_vma() etc..  are used
in walk_page_range() and we need mmap_sem around it.

This patch adds mmap_sem around walk_page_range().

Because /proc/<pid>/pagemap's callback routine use put_user(), we have
to get rid of it to do sane fix.

Changelog: 2010/Apr/2
 - fixed start_vaddr and end overflow
Changelog: 2010/Apr/1
 - fixed start_vaddr calculation
 - removed unnecessary cast.
 - removed unnecessary change in smaps.
 - use GFP_TEMPORARY instead of GFP_KERNEL

Signed-off-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Matt Mackall <mpm@selenic.com>
Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Brian Swetland <swetland@google.com>
Cc: Dave Hansen <haveblue@us.ibm.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: San Mehat <san@google.com>
[ Fixed kmalloc failure return code as per Matt ]
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

1 files changed, 38 insertions, 49 deletions

diff --git a/fs/proc/task_mmu.c b/fs/proc/task_mmu.c
index 2a1bef9203c6..8e5fae2f7aed 100644
--- a/fs/proc/task_mmu.c
+++ b/fs/proc/task_mmu.c
@@ -404,6 +404,7 @@ static int show_smap(struct seq_file *m, void *v)
 
 	memset(&mss, 0, sizeof mss);
 	mss.vma = vma;
+	/* mmap_sem is held in m_start */
 	if (vma->vm_mm && !is_vm_hugetlb_page(vma))
 		walk_page_range(vma->vm_start, vma->vm_end, &smaps_walk);
 
@@ -550,7 +551,8 @@ const struct file_operations proc_clear_refs_operations = {
 };
 
 struct pagemapread {
-	u64 __user *out, *end;
+	int pos, len;
+	u64 *buffer;
 };
 
 #define PM_ENTRY_BYTES      sizeof(u64)
@@ -573,10 +575,8 @@ struct pagemapread {
 static int add_to_pagemap(unsigned long addr, u64 pfn,
 			  struct pagemapread *pm)
 {
-	if (put_user(pfn, pm->out))
-		return -EFAULT;
-	pm->out++;
-	if (pm->out >= pm->end)
+	pm->buffer[pm->pos++] = pfn;
+	if (pm->pos >= pm->len)
 		return PM_END_OF_BUFFER;
 	return 0;
 }
@@ -674,21 +674,20 @@ static int pagemap_pte_range(pmd_t *pmd, unsigned long addr, unsigned long end,
  * determine which areas of memory are actually mapped and llseek to
  * skip over unmapped regions.
  */
+#define PAGEMAP_WALK_SIZE	(PMD_SIZE)
 static ssize_t pagemap_read(struct file *file, char __user *buf,
 			    size_t count, loff_t *ppos)
 {
 	struct task_struct *task = get_proc_task(file->f_path.dentry->d_inode);
-	struct page **pages, *page;
-	unsigned long uaddr, uend;
 	struct mm_struct *mm;
 	struct pagemapread pm;
-	int pagecount;
 	int ret = -ESRCH;
 	struct mm_walk pagemap_walk = {};
 	unsigned long src;
 	unsigned long svpfn;
 	unsigned long start_vaddr;
 	unsigned long end_vaddr;
+	int copied = 0;
 
 	if (!task)
 		goto out;
@@ -711,35 +710,12 @@ static ssize_t pagemap_read(struct file *file, char __user *buf,
 	if (!mm)
 		goto out_task;
 
-
-	uaddr = (unsigned long)buf & PAGE_MASK;
-	uend = (unsigned long)(buf + count);
-	pagecount = (PAGE_ALIGN(uend) - uaddr) / PAGE_SIZE;
-	ret = 0;
-	if (pagecount == 0)
-		goto out_mm;
-	pages = kcalloc(pagecount, sizeof(struct page *), GFP_KERNEL);
+	pm.len = PM_ENTRY_BYTES * (PAGEMAP_WALK_SIZE >> PAGE_SHIFT);
+	pm.buffer = kmalloc(pm.len, GFP_TEMPORARY);
 	ret = -ENOMEM;
-	if (!pages)
+	if (!pm.buffer)
 		goto out_mm;
 
-	down_read(&current->mm->mmap_sem);
-	ret = get_user_pages(current, current->mm, uaddr, pagecount,
-			     1, 0, pages, NULL);
-	up_read(&current->mm->mmap_sem);
-
-	if (ret < 0)
-		goto out_free;
-
-	if (ret != pagecount) {
-		pagecount = ret;
-		ret = -EFAULT;
-		goto out_pages;
-	}
-
-	pm.out = (u64 __user *)buf;
-	pm.end = (u64 __user *)(buf + count);
-
 	pagemap_walk.pmd_entry = pagemap_pte_range;
 	pagemap_walk.pte_hole = pagemap_pte_hole;
 	pagemap_walk.mm = mm;
@@ -760,23 +736,36 @@ static ssize_t pagemap_read(struct file *file, char __user *buf,
 	 * user buffer is tracked in "pm", and the walk
 	 * will stop when we hit the end of the buffer.
 	 */
-	ret = walk_page_range(start_vaddr, end_vaddr, &pagemap_walk);
-	if (ret == PM_END_OF_BUFFER)
-		ret = 0;
-	/* don't need mmap_sem for these, but this looks cleaner */
-	*ppos += (char __user *)pm.out - buf;
-	if (!ret)
-		ret = (char __user *)pm.out - buf;
-
-out_pages:
-	for (; pagecount; pagecount--) {
-		page = pages[pagecount-1];
-		if (!PageReserved(page))
-			SetPageDirty(page);
-		page_cache_release(page);
+	ret = 0;
+	while (count && (start_vaddr < end_vaddr)) {
+		int len;
+		unsigned long end;
+
+		pm.pos = 0;
+		end = start_vaddr + PAGEMAP_WALK_SIZE;
+		/* overflow ? */
+		if (end < start_vaddr || end > end_vaddr)
+			end = end_vaddr;
+		down_read(&mm->mmap_sem);
+		ret = walk_page_range(start_vaddr, end, &pagemap_walk);
+		up_read(&mm->mmap_sem);
+		start_vaddr = end;
+
+		len = min(count, PM_ENTRY_BYTES * pm.pos);
+		if (copy_to_user(buf, pm.buffer, len) < 0) {
+			ret = -EFAULT;
+			goto out_free;
+		}
+		copied += len;
+		buf += len;
+		count -= len;
 	}
+	*ppos += copied;
+	if (!ret || ret == PM_END_OF_BUFFER)
+		ret = copied;
+
 out_free:
-	kfree(pages);
+	kfree(pm.buffer);
 out_mm:
 	mmput(mm);
 out_task: