Add the rt linux 4.1.3-rt3 as base
[kvmfornfv.git] / kernel / fs / ext4 / block_validity.c
diff --git a/kernel/fs/ext4/block_validity.c b/kernel/fs/ext4/block_validity.c
new file mode 100644 (file)
index 0000000..3522340
--- /dev/null
@@ -0,0 +1,242 @@
+/*
+ *  linux/fs/ext4/block_validity.c
+ *
+ * Copyright (C) 2009
+ * Theodore Ts'o (tytso@mit.edu)
+ *
+ * Track which blocks in the filesystem are metadata blocks that
+ * should never be used as data blocks by files or directories.
+ */
+
+#include <linux/time.h>
+#include <linux/fs.h>
+#include <linux/namei.h>
+#include <linux/quotaops.h>
+#include <linux/buffer_head.h>
+#include <linux/swap.h>
+#include <linux/pagemap.h>
+#include <linux/blkdev.h>
+#include <linux/slab.h>
+#include "ext4.h"
+
+struct ext4_system_zone {
+       struct rb_node  node;
+       ext4_fsblk_t    start_blk;
+       unsigned int    count;
+};
+
+static struct kmem_cache *ext4_system_zone_cachep;
+
+int __init ext4_init_system_zone(void)
+{
+       ext4_system_zone_cachep = KMEM_CACHE(ext4_system_zone, 0);
+       if (ext4_system_zone_cachep == NULL)
+               return -ENOMEM;
+       return 0;
+}
+
+void ext4_exit_system_zone(void)
+{
+       kmem_cache_destroy(ext4_system_zone_cachep);
+}
+
+static inline int can_merge(struct ext4_system_zone *entry1,
+                    struct ext4_system_zone *entry2)
+{
+       if ((entry1->start_blk + entry1->count) == entry2->start_blk)
+               return 1;
+       return 0;
+}
+
+/*
+ * Mark a range of blocks as belonging to the "system zone" --- that
+ * is, filesystem metadata blocks which should never be used by
+ * inodes.
+ */
+static int add_system_zone(struct ext4_sb_info *sbi,
+                          ext4_fsblk_t start_blk,
+                          unsigned int count)
+{
+       struct ext4_system_zone *new_entry = NULL, *entry;
+       struct rb_node **n = &sbi->system_blks.rb_node, *node;
+       struct rb_node *parent = NULL, *new_node = NULL;
+
+       while (*n) {
+               parent = *n;
+               entry = rb_entry(parent, struct ext4_system_zone, node);
+               if (start_blk < entry->start_blk)
+                       n = &(*n)->rb_left;
+               else if (start_blk >= (entry->start_blk + entry->count))
+                       n = &(*n)->rb_right;
+               else {
+                       if (start_blk + count > (entry->start_blk +
+                                                entry->count))
+                               entry->count = (start_blk + count -
+                                               entry->start_blk);
+                       new_node = *n;
+                       new_entry = rb_entry(new_node, struct ext4_system_zone,
+                                            node);
+                       break;
+               }
+       }
+
+       if (!new_entry) {
+               new_entry = kmem_cache_alloc(ext4_system_zone_cachep,
+                                            GFP_KERNEL);
+               if (!new_entry)
+                       return -ENOMEM;
+               new_entry->start_blk = start_blk;
+               new_entry->count = count;
+               new_node = &new_entry->node;
+
+               rb_link_node(new_node, parent, n);
+               rb_insert_color(new_node, &sbi->system_blks);
+       }
+
+       /* Can we merge to the left? */
+       node = rb_prev(new_node);
+       if (node) {
+               entry = rb_entry(node, struct ext4_system_zone, node);
+               if (can_merge(entry, new_entry)) {
+                       new_entry->start_blk = entry->start_blk;
+                       new_entry->count += entry->count;
+                       rb_erase(node, &sbi->system_blks);
+                       kmem_cache_free(ext4_system_zone_cachep, entry);
+               }
+       }
+
+       /* Can we merge to the right? */
+       node = rb_next(new_node);
+       if (node) {
+               entry = rb_entry(node, struct ext4_system_zone, node);
+               if (can_merge(new_entry, entry)) {
+                       new_entry->count += entry->count;
+                       rb_erase(node, &sbi->system_blks);
+                       kmem_cache_free(ext4_system_zone_cachep, entry);
+               }
+       }
+       return 0;
+}
+
+static void debug_print_tree(struct ext4_sb_info *sbi)
+{
+       struct rb_node *node;
+       struct ext4_system_zone *entry;
+       int first = 1;
+
+       printk(KERN_INFO "System zones: ");
+       node = rb_first(&sbi->system_blks);
+       while (node) {
+               entry = rb_entry(node, struct ext4_system_zone, node);
+               printk("%s%llu-%llu", first ? "" : ", ",
+                      entry->start_blk, entry->start_blk + entry->count - 1);
+               first = 0;
+               node = rb_next(node);
+       }
+       printk("\n");
+}
+
+int ext4_setup_system_zone(struct super_block *sb)
+{
+       ext4_group_t ngroups = ext4_get_groups_count(sb);
+       struct ext4_sb_info *sbi = EXT4_SB(sb);
+       struct ext4_group_desc *gdp;
+       ext4_group_t i;
+       int flex_size = ext4_flex_bg_size(sbi);
+       int ret;
+
+       if (!test_opt(sb, BLOCK_VALIDITY)) {
+               if (EXT4_SB(sb)->system_blks.rb_node)
+                       ext4_release_system_zone(sb);
+               return 0;
+       }
+       if (EXT4_SB(sb)->system_blks.rb_node)
+               return 0;
+
+       for (i=0; i < ngroups; i++) {
+               if (ext4_bg_has_super(sb, i) &&
+                   ((i < 5) || ((i % flex_size) == 0)))
+                       add_system_zone(sbi, ext4_group_first_block_no(sb, i),
+                                       ext4_bg_num_gdb(sb, i) + 1);
+               gdp = ext4_get_group_desc(sb, i, NULL);
+               ret = add_system_zone(sbi, ext4_block_bitmap(sb, gdp), 1);
+               if (ret)
+                       return ret;
+               ret = add_system_zone(sbi, ext4_inode_bitmap(sb, gdp), 1);
+               if (ret)
+                       return ret;
+               ret = add_system_zone(sbi, ext4_inode_table(sb, gdp),
+                               sbi->s_itb_per_group);
+               if (ret)
+                       return ret;
+       }
+
+       if (test_opt(sb, DEBUG))
+               debug_print_tree(EXT4_SB(sb));
+       return 0;
+}
+
+/* Called when the filesystem is unmounted */
+void ext4_release_system_zone(struct super_block *sb)
+{
+       struct ext4_system_zone *entry, *n;
+
+       rbtree_postorder_for_each_entry_safe(entry, n,
+                       &EXT4_SB(sb)->system_blks, node)
+               kmem_cache_free(ext4_system_zone_cachep, entry);
+
+       EXT4_SB(sb)->system_blks = RB_ROOT;
+}
+
+/*
+ * Returns 1 if the passed-in block region (start_blk,
+ * start_blk+count) is valid; 0 if some part of the block region
+ * overlaps with filesystem metadata blocks.
+ */
+int ext4_data_block_valid(struct ext4_sb_info *sbi, ext4_fsblk_t start_blk,
+                         unsigned int count)
+{
+       struct ext4_system_zone *entry;
+       struct rb_node *n = sbi->system_blks.rb_node;
+
+       if ((start_blk <= le32_to_cpu(sbi->s_es->s_first_data_block)) ||
+           (start_blk + count < start_blk) ||
+           (start_blk + count > ext4_blocks_count(sbi->s_es))) {
+               sbi->s_es->s_last_error_block = cpu_to_le64(start_blk);
+               return 0;
+       }
+       while (n) {
+               entry = rb_entry(n, struct ext4_system_zone, node);
+               if (start_blk + count - 1 < entry->start_blk)
+                       n = n->rb_left;
+               else if (start_blk >= (entry->start_blk + entry->count))
+                       n = n->rb_right;
+               else {
+                       sbi->s_es->s_last_error_block = cpu_to_le64(start_blk);
+                       return 0;
+               }
+       }
+       return 1;
+}
+
+int ext4_check_blockref(const char *function, unsigned int line,
+                       struct inode *inode, __le32 *p, unsigned int max)
+{
+       struct ext4_super_block *es = EXT4_SB(inode->i_sb)->s_es;
+       __le32 *bref = p;
+       unsigned int blk;
+
+       while (bref < p+max) {
+               blk = le32_to_cpu(*bref++);
+               if (blk &&
+                   unlikely(!ext4_data_block_valid(EXT4_SB(inode->i_sb),
+                                                   blk, 1))) {
+                       es->s_last_error_block = cpu_to_le64(blk);
+                       ext4_error_inode(inode, function, line, blk,
+                                        "invalid block");
+                       return -EIO;
+               }
+       }
+       return 0;
+}
+