]> Pileus Git - ~andy/linux/blobdiff - fs/proc/task_mmu.c
Merge branch 'core' of git://git.kernel.org/pub/scm/linux/kernel/git/rric/oprofile...
[~andy/linux] / fs / proc / task_mmu.c
index 318d8654989bf70479321fb8599b0dcf9ac29365..25b6a887adb916a93d20aa092481f67df451ff8c 100644 (file)
@@ -211,7 +211,7 @@ static void show_map_vma(struct seq_file *m, struct vm_area_struct *vma)
 {
        struct mm_struct *mm = vma->vm_mm;
        struct file *file = vma->vm_file;
-       int flags = vma->vm_flags;
+       vm_flags_t flags = vma->vm_flags;
        unsigned long ino = 0;
        unsigned long long pgoff = 0;
        unsigned long start, end;
@@ -536,15 +536,17 @@ static ssize_t clear_refs_write(struct file *file, const char __user *buf,
        char buffer[PROC_NUMBUF];
        struct mm_struct *mm;
        struct vm_area_struct *vma;
-       long type;
+       int type;
+       int rv;
 
        memset(buffer, 0, sizeof(buffer));
        if (count > sizeof(buffer) - 1)
                count = sizeof(buffer) - 1;
        if (copy_from_user(buffer, buf, count))
                return -EFAULT;
-       if (strict_strtol(strstrip(buffer), 10, &type))
-               return -EINVAL;
+       rv = kstrtoint(strstrip(buffer), 10, &type);
+       if (rv < 0)
+               return rv;
        if (type < CLEAR_REFS_ALL || type > CLEAR_REFS_MAPPED)
                return -EINVAL;
        task = get_proc_task(file->f_path.dentry->d_inode);
@@ -769,18 +771,12 @@ static ssize_t pagemap_read(struct file *file, char __user *buf,
        if (!task)
                goto out;
 
-       mm = mm_for_maps(task);
-       ret = PTR_ERR(mm);
-       if (!mm || IS_ERR(mm))
-               goto out_task;
-
        ret = -EINVAL;
        /* file position must be aligned */
        if ((*ppos % PM_ENTRY_BYTES) || (count % PM_ENTRY_BYTES))
                goto out_task;
 
        ret = 0;
-
        if (!count)
                goto out_task;
 
@@ -788,7 +784,12 @@ static ssize_t pagemap_read(struct file *file, char __user *buf,
        pm.buffer = kmalloc(pm.len, GFP_TEMPORARY);
        ret = -ENOMEM;
        if (!pm.buffer)
-               goto out_mm;
+               goto out_task;
+
+       mm = mm_for_maps(task);
+       ret = PTR_ERR(mm);
+       if (!mm || IS_ERR(mm))
+               goto out_free;
 
        pagemap_walk.pmd_entry = pagemap_pte_range;
        pagemap_walk.pte_hole = pagemap_pte_hole;
@@ -831,7 +832,7 @@ static ssize_t pagemap_read(struct file *file, char __user *buf,
                len = min(count, PM_ENTRY_BYTES * pm.pos);
                if (copy_to_user(buf, pm.buffer, len)) {
                        ret = -EFAULT;
-                       goto out_free;
+                       goto out_mm;
                }
                copied += len;
                buf += len;
@@ -841,10 +842,10 @@ static ssize_t pagemap_read(struct file *file, char __user *buf,
        if (!ret || ret == PM_END_OF_BUFFER)
                ret = copied;
 
-out_free:
-       kfree(pm.buffer);
 out_mm:
        mmput(mm);
+out_free:
+       kfree(pm.buffer);
 out_task:
        put_task_struct(task);
 out:
@@ -858,7 +859,192 @@ const struct file_operations proc_pagemap_operations = {
 #endif /* CONFIG_PROC_PAGE_MONITOR */
 
 #ifdef CONFIG_NUMA
-extern int show_numa_map(struct seq_file *m, void *v);
+
+struct numa_maps {
+       struct vm_area_struct *vma;
+       unsigned long pages;
+       unsigned long anon;
+       unsigned long active;
+       unsigned long writeback;
+       unsigned long mapcount_max;
+       unsigned long dirty;
+       unsigned long swapcache;
+       unsigned long node[MAX_NUMNODES];
+};
+
+struct numa_maps_private {
+       struct proc_maps_private proc_maps;
+       struct numa_maps md;
+};
+
+static void gather_stats(struct page *page, struct numa_maps *md, int pte_dirty)
+{
+       int count = page_mapcount(page);
+
+       md->pages++;
+       if (pte_dirty || PageDirty(page))
+               md->dirty++;
+
+       if (PageSwapCache(page))
+               md->swapcache++;
+
+       if (PageActive(page) || PageUnevictable(page))
+               md->active++;
+
+       if (PageWriteback(page))
+               md->writeback++;
+
+       if (PageAnon(page))
+               md->anon++;
+
+       if (count > md->mapcount_max)
+               md->mapcount_max = count;
+
+       md->node[page_to_nid(page)]++;
+}
+
+static int gather_pte_stats(pmd_t *pmd, unsigned long addr,
+               unsigned long end, struct mm_walk *walk)
+{
+       struct numa_maps *md;
+       spinlock_t *ptl;
+       pte_t *orig_pte;
+       pte_t *pte;
+
+       md = walk->private;
+       orig_pte = pte = pte_offset_map_lock(walk->mm, pmd, addr, &ptl);
+       do {
+               struct page *page;
+               int nid;
+
+               if (!pte_present(*pte))
+                       continue;
+
+               page = vm_normal_page(md->vma, addr, *pte);
+               if (!page)
+                       continue;
+
+               if (PageReserved(page))
+                       continue;
+
+               nid = page_to_nid(page);
+               if (!node_isset(nid, node_states[N_HIGH_MEMORY]))
+                       continue;
+
+               gather_stats(page, md, pte_dirty(*pte));
+
+       } while (pte++, addr += PAGE_SIZE, addr != end);
+       pte_unmap_unlock(orig_pte, ptl);
+       return 0;
+}
+#ifdef CONFIG_HUGETLB_PAGE
+static int gather_hugetbl_stats(pte_t *pte, unsigned long hmask,
+               unsigned long addr, unsigned long end, struct mm_walk *walk)
+{
+       struct numa_maps *md;
+       struct page *page;
+
+       if (pte_none(*pte))
+               return 0;
+
+       page = pte_page(*pte);
+       if (!page)
+               return 0;
+
+       md = walk->private;
+       gather_stats(page, md, pte_dirty(*pte));
+       return 0;
+}
+
+#else
+static int gather_hugetbl_stats(pte_t *pte, unsigned long hmask,
+               unsigned long addr, unsigned long end, struct mm_walk *walk)
+{
+       return 0;
+}
+#endif
+
+/*
+ * Display pages allocated per node and memory policy via /proc.
+ */
+static int show_numa_map(struct seq_file *m, void *v)
+{
+       struct numa_maps_private *numa_priv = m->private;
+       struct proc_maps_private *proc_priv = &numa_priv->proc_maps;
+       struct vm_area_struct *vma = v;
+       struct numa_maps *md = &numa_priv->md;
+       struct file *file = vma->vm_file;
+       struct mm_struct *mm = vma->vm_mm;
+       struct mm_walk walk = {};
+       struct mempolicy *pol;
+       int n;
+       char buffer[50];
+
+       if (!mm)
+               return 0;
+
+       /* Ensure we start with an empty set of numa_maps statistics. */
+       memset(md, 0, sizeof(*md));
+
+       md->vma = vma;
+
+       walk.hugetlb_entry = gather_hugetbl_stats;
+       walk.pmd_entry = gather_pte_stats;
+       walk.private = md;
+       walk.mm = mm;
+
+       pol = get_vma_policy(proc_priv->task, vma, vma->vm_start);
+       mpol_to_str(buffer, sizeof(buffer), pol, 0);
+       mpol_cond_put(pol);
+
+       seq_printf(m, "%08lx %s", vma->vm_start, buffer);
+
+       if (file) {
+               seq_printf(m, " file=");
+               seq_path(m, &file->f_path, "\n\t= ");
+       } else if (vma->vm_start <= mm->brk && vma->vm_end >= mm->start_brk) {
+               seq_printf(m, " heap");
+       } else if (vma->vm_start <= mm->start_stack &&
+                       vma->vm_end >= mm->start_stack) {
+               seq_printf(m, " stack");
+       }
+
+       walk_page_range(vma->vm_start, vma->vm_end, &walk);
+
+       if (!md->pages)
+               goto out;
+
+       if (md->anon)
+               seq_printf(m, " anon=%lu", md->anon);
+
+       if (md->dirty)
+               seq_printf(m, " dirty=%lu", md->dirty);
+
+       if (md->pages != md->anon && md->pages != md->dirty)
+               seq_printf(m, " mapped=%lu", md->pages);
+
+       if (md->mapcount_max > 1)
+               seq_printf(m, " mapmax=%lu", md->mapcount_max);
+
+       if (md->swapcache)
+               seq_printf(m, " swapcache=%lu", md->swapcache);
+
+       if (md->active < md->pages && !is_vm_hugetlb_page(vma))
+               seq_printf(m, " active=%lu", md->active);
+
+       if (md->writeback)
+               seq_printf(m, " writeback=%lu", md->writeback);
+
+       for_each_node_state(n, N_HIGH_MEMORY)
+               if (md->node[n])
+                       seq_printf(m, " N%d=%lu", n, md->node[n]);
+out:
+       seq_putc(m, '\n');
+
+       if (m->count < m->size)
+               m->version = (vma != proc_priv->tail_vma) ? vma->vm_start : 0;
+       return 0;
+}
 
 static const struct seq_operations proc_pid_numa_maps_op = {
         .start  = m_start,
@@ -869,7 +1055,20 @@ static const struct seq_operations proc_pid_numa_maps_op = {
 
 static int numa_maps_open(struct inode *inode, struct file *file)
 {
-       return do_maps_open(inode, file, &proc_pid_numa_maps_op);
+       struct numa_maps_private *priv;
+       int ret = -ENOMEM;
+       priv = kzalloc(sizeof(*priv), GFP_KERNEL);
+       if (priv) {
+               priv->proc_maps.pid = proc_pid(inode);
+               ret = seq_open(file, &proc_pid_numa_maps_op);
+               if (!ret) {
+                       struct seq_file *m = file->private_data;
+                       m->private = priv;
+               } else {
+                       kfree(priv);
+               }
+       }
+       return ret;
 }
 
 const struct file_operations proc_numa_maps_operations = {
@@ -878,4 +1077,4 @@ const struct file_operations proc_numa_maps_operations = {
        .llseek         = seq_lseek,
        .release        = seq_release_private,
 };
-#endif
+#endif /* CONFIG_NUMA */