ext4: move ext4_fiemap to use iomap framework
authorRitesh Harjani <riteshh@linux.ibm.com>
Fri, 28 Feb 2020 09:26:58 +0000 (14:56 +0530)
committerTheodore Ts'o <tytso@mit.edu>
Sat, 14 Mar 2020 18:43:13 +0000 (14:43 -0400)
This patch moves ext4_fiemap to use iomap framework.
For xattr a new 'ext4_iomap_xattr_ops' is added.

Reported-by: kbuild test robot <lkp@intel.com>
Reviewed-by: Jan Kara <jack@suse.cz>
Reviewed-by: Darrick J. Wong <darrick.wong@oracle.com>
Link: https://lore.kernel.org/r/b9f45c885814fcdd0631747ff0fe08886270828c.1582880246.git.riteshh@linux.ibm.com
Signed-off-by: Ritesh Harjani <riteshh@linux.ibm.com>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
fs/ext4/extents.c
fs/ext4/inline.c

index 2ebc858..5b2cac0 100644 (file)
@@ -28,6 +28,7 @@
 #include <linux/uaccess.h>
 #include <linux/fiemap.h>
 #include <linux/backing-dev.h>
+#include <linux/iomap.h>
 #include "ext4_jbd2.h"
 #include "ext4_extents.h"
 #include "xattr.h"
@@ -90,9 +91,6 @@ static int ext4_split_extent_at(handle_t *handle,
                             int split_flag,
                             int flags);
 
-static int ext4_find_delayed_extent(struct inode *inode,
-                                   struct extent_status *newes);
-
 static int ext4_ext_trunc_restart_fn(struct inode *inode, int *dropped)
 {
        /*
@@ -2130,155 +2128,6 @@ cleanup:
        return err;
 }
 
-static int ext4_fill_fiemap_extents(struct inode *inode,
-                                   ext4_lblk_t block, ext4_lblk_t num,
-                                   struct fiemap_extent_info *fieinfo)
-{
-       struct ext4_ext_path *path = NULL;
-       struct ext4_extent *ex;
-       struct extent_status es;
-       ext4_lblk_t next, next_del, start = 0, end = 0;
-       ext4_lblk_t last = block + num;
-       int exists, depth = 0, err = 0;
-       unsigned int flags = 0;
-       unsigned char blksize_bits = inode->i_sb->s_blocksize_bits;
-
-       while (block < last && block != EXT_MAX_BLOCKS) {
-               num = last - block;
-               /* find extent for this block */
-               down_read(&EXT4_I(inode)->i_data_sem);
-
-               path = ext4_find_extent(inode, block, &path, 0);
-               if (IS_ERR(path)) {
-                       up_read(&EXT4_I(inode)->i_data_sem);
-                       err = PTR_ERR(path);
-                       path = NULL;
-                       break;
-               }
-
-               depth = ext_depth(inode);
-               if (unlikely(path[depth].p_hdr == NULL)) {
-                       up_read(&EXT4_I(inode)->i_data_sem);
-                       EXT4_ERROR_INODE(inode, "path[%d].p_hdr == NULL", depth);
-                       err = -EFSCORRUPTED;
-                       break;
-               }
-               ex = path[depth].p_ext;
-               next = ext4_ext_next_allocated_block(path);
-
-               flags = 0;
-               exists = 0;
-               if (!ex) {
-                       /* there is no extent yet, so try to allocate
-                        * all requested space */
-                       start = block;
-                       end = block + num;
-               } else if (le32_to_cpu(ex->ee_block) > block) {
-                       /* need to allocate space before found extent */
-                       start = block;
-                       end = le32_to_cpu(ex->ee_block);
-                       if (block + num < end)
-                               end = block + num;
-               } else if (block >= le32_to_cpu(ex->ee_block)
-                                       + ext4_ext_get_actual_len(ex)) {
-                       /* need to allocate space after found extent */
-                       start = block;
-                       end = block + num;
-                       if (end >= next)
-                               end = next;
-               } else if (block >= le32_to_cpu(ex->ee_block)) {
-                       /*
-                        * some part of requested space is covered
-                        * by found extent
-                        */
-                       start = block;
-                       end = le32_to_cpu(ex->ee_block)
-                               + ext4_ext_get_actual_len(ex);
-                       if (block + num < end)
-                               end = block + num;
-                       exists = 1;
-               } else {
-                       BUG();
-               }
-               BUG_ON(end <= start);
-
-               if (!exists) {
-                       es.es_lblk = start;
-                       es.es_len = end - start;
-                       es.es_pblk = 0;
-               } else {
-                       es.es_lblk = le32_to_cpu(ex->ee_block);
-                       es.es_len = ext4_ext_get_actual_len(ex);
-                       es.es_pblk = ext4_ext_pblock(ex);
-                       if (ext4_ext_is_unwritten(ex))
-                               flags |= FIEMAP_EXTENT_UNWRITTEN;
-               }
-
-               /*
-                * Find delayed extent and update es accordingly. We call
-                * it even in !exists case to find out whether es is the
-                * last existing extent or not.
-                */
-               next_del = ext4_find_delayed_extent(inode, &es);
-               if (!exists && next_del) {
-                       exists = 1;
-                       flags |= (FIEMAP_EXTENT_DELALLOC |
-                                 FIEMAP_EXTENT_UNKNOWN);
-               }
-               up_read(&EXT4_I(inode)->i_data_sem);
-
-               if (unlikely(es.es_len == 0)) {
-                       EXT4_ERROR_INODE(inode, "es.es_len == 0");
-                       err = -EFSCORRUPTED;
-                       break;
-               }
-
-               /*
-                * This is possible iff next == next_del == EXT_MAX_BLOCKS.
-                * we need to check next == EXT_MAX_BLOCKS because it is
-                * possible that an extent is with unwritten and delayed
-                * status due to when an extent is delayed allocated and
-                * is allocated by fallocate status tree will track both of
-                * them in a extent.
-                *
-                * So we could return a unwritten and delayed extent, and
-                * its block is equal to 'next'.
-                */
-               if (next == next_del && next == EXT_MAX_BLOCKS) {
-                       flags |= FIEMAP_EXTENT_LAST;
-                       if (unlikely(next_del != EXT_MAX_BLOCKS ||
-                                    next != EXT_MAX_BLOCKS)) {
-                               EXT4_ERROR_INODE(inode,
-                                                "next extent == %u, next "
-                                                "delalloc extent = %u",
-                                                next, next_del);
-                               err = -EFSCORRUPTED;
-                               break;
-                       }
-               }
-
-               if (exists) {
-                       err = fiemap_fill_next_extent(fieinfo,
-                               (__u64)es.es_lblk << blksize_bits,
-                               (__u64)es.es_pblk << blksize_bits,
-                               (__u64)es.es_len << blksize_bits,
-                               flags);
-                       if (err < 0)
-                               break;
-                       if (err == 1) {
-                               err = 0;
-                               break;
-                       }
-               }
-
-               block = es.es_lblk + es.es_len;
-       }
-
-       ext4_ext_drop_refs(path);
-       kfree(path);
-       return err;
-}
-
 static int ext4_fill_es_cache_info(struct inode *inode,
                                   ext4_lblk_t block, ext4_lblk_t num,
                                   struct fiemap_extent_info *fieinfo)
@@ -4927,64 +4776,13 @@ int ext4_convert_unwritten_io_end_vec(handle_t *handle, ext4_io_end_t *io_end)
        return ret < 0 ? ret : err;
 }
 
-/*
- * If newes is not existing extent (newes->ec_pblk equals zero) find
- * delayed extent at start of newes and update newes accordingly and
- * return start of the next delayed extent.
- *
- * If newes is existing extent (newes->ec_pblk is not equal zero)
- * return start of next delayed extent or EXT_MAX_BLOCKS if no delayed
- * extent found. Leave newes unmodified.
- */
-static int ext4_find_delayed_extent(struct inode *inode,
-                                   struct extent_status *newes)
-{
-       struct extent_status es;
-       ext4_lblk_t block, next_del;
-
-       if (newes->es_pblk == 0) {
-               ext4_es_find_extent_range(inode, &ext4_es_is_delayed,
-                                         newes->es_lblk,
-                                         newes->es_lblk + newes->es_len - 1,
-                                         &es);
-
-               /*
-                * No extent in extent-tree contains block @newes->es_pblk,
-                * then the block may stay in 1)a hole or 2)delayed-extent.
-                */
-               if (es.es_len == 0)
-                       /* A hole found. */
-                       return 0;
-
-               if (es.es_lblk > newes->es_lblk) {
-                       /* A hole found. */
-                       newes->es_len = min(es.es_lblk - newes->es_lblk,
-                                           newes->es_len);
-                       return 0;
-               }
-
-               newes->es_len = es.es_lblk + es.es_len - newes->es_lblk;
-       }
-
-       block = newes->es_lblk + newes->es_len;
-       ext4_es_find_extent_range(inode, &ext4_es_is_delayed, block,
-                                 EXT_MAX_BLOCKS, &es);
-       if (es.es_len == 0)
-               next_del = EXT_MAX_BLOCKS;
-       else
-               next_del = es.es_lblk;
-
-       return next_del;
-}
-
-static int ext4_xattr_fiemap(struct inode *inode,
-                               struct fiemap_extent_info *fieinfo)
+static int ext4_iomap_xattr_fiemap(struct inode *inode, struct iomap *iomap)
 {
        __u64 physical = 0;
-       __u64 length;
-       __u32 flags = FIEMAP_EXTENT_LAST;
+       __u64 length = 0;
        int blockbits = inode->i_sb->s_blocksize_bits;
        int error = 0;
+       u16 iomap_type;
 
        /* in-inode? */
        if (ext4_test_inode_state(inode, EXT4_STATE_XATTR)) {
@@ -4999,40 +4797,49 @@ static int ext4_xattr_fiemap(struct inode *inode,
                                EXT4_I(inode)->i_extra_isize;
                physical += offset;
                length = EXT4_SB(inode->i_sb)->s_inode_size - offset;
-               flags |= FIEMAP_EXTENT_DATA_INLINE;
                brelse(iloc.bh);
-       } else { /* external block */
+               iomap_type = IOMAP_INLINE;
+       } else if (EXT4_I(inode)->i_file_acl) { /* external block */
                physical = (__u64)EXT4_I(inode)->i_file_acl << blockbits;
                length = inode->i_sb->s_blocksize;
+               iomap_type = IOMAP_MAPPED;
+       } else {
+               /* no in-inode or external block for xattr, so return -ENOENT */
+               error = -ENOENT;
+               goto out;
        }
 
-       if (physical)
-               error = fiemap_fill_next_extent(fieinfo, 0, physical,
-                                               length, flags);
-       return (error < 0 ? error : 0);
+       iomap->addr = physical;
+       iomap->offset = 0;
+       iomap->length = length;
+       iomap->type = iomap_type;
+       iomap->flags = 0;
+out:
+       return error;
 }
 
-static int _ext4_fiemap(struct inode *inode,
-                       struct fiemap_extent_info *fieinfo,
-                       __u64 start, __u64 len,
-                       int (*fill)(struct inode *, ext4_lblk_t,
-                                   ext4_lblk_t,
-                                   struct fiemap_extent_info *))
+static int ext4_iomap_xattr_begin(struct inode *inode, loff_t offset,
+                                 loff_t length, unsigned flags,
+                                 struct iomap *iomap, struct iomap *srcmap)
 {
-       ext4_lblk_t start_blk;
-       u32 ext4_fiemap_flags = FIEMAP_FLAG_SYNC|FIEMAP_FLAG_XATTR;
+       int error;
 
-       int error = 0;
-
-       if (ext4_has_inline_data(inode)) {
-               int has_inline = 1;
+       error = ext4_iomap_xattr_fiemap(inode, iomap);
+       if (error == 0 && (offset >= iomap->length))
+               error = -ENOENT;
+       return error;
+}
 
-               error = ext4_inline_data_fiemap(inode, fieinfo, &has_inline,
-                                               start, len);
+static const struct iomap_ops ext4_iomap_xattr_ops = {
+       .iomap_begin            = ext4_iomap_xattr_begin,
+};
 
-               if (has_inline)
-                       return error;
-       }
+static int _ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
+                       __u64 start, __u64 len, bool from_es_cache)
+{
+       ext4_lblk_t start_blk;
+       u32 ext4_fiemap_flags = FIEMAP_FLAG_SYNC | FIEMAP_FLAG_XATTR;
+       int error = 0;
 
        if (fieinfo->fi_flags & FIEMAP_FLAG_CACHE) {
                error = ext4_ext_precache(inode);
@@ -5041,19 +4848,19 @@ static int _ext4_fiemap(struct inode *inode,
                fieinfo->fi_flags &= ~FIEMAP_FLAG_CACHE;
        }
 
-       /* fallback to generic here if not in extents fmt */
-       if (!(ext4_test_inode_flag(inode, EXT4_INODE_EXTENTS)) &&
-           fill == ext4_fill_fiemap_extents)
-               return generic_block_fiemap(inode, fieinfo, start, len,
-                       ext4_get_block);
-
-       if (fill == ext4_fill_es_cache_info)
+       if (from_es_cache)
                ext4_fiemap_flags &= FIEMAP_FLAG_XATTR;
+
        if (fiemap_check_flags(fieinfo, ext4_fiemap_flags))
                return -EBADR;
 
        if (fieinfo->fi_flags & FIEMAP_FLAG_XATTR) {
-               error = ext4_xattr_fiemap(inode, fieinfo);
+               fieinfo->fi_flags &= ~FIEMAP_FLAG_XATTR;
+               error = iomap_fiemap(inode, fieinfo, start, len,
+                                    &ext4_iomap_xattr_ops);
+       } else if (!from_es_cache) {
+               error = iomap_fiemap(inode, fieinfo, start, len,
+                                    &ext4_iomap_report_ops);
        } else {
                ext4_lblk_t len_blks;
                __u64 last_blk;
@@ -5068,7 +4875,8 @@ static int _ext4_fiemap(struct inode *inode,
                 * Walk the extent tree gathering extent information
                 * and pushing extents back to the user.
                 */
-               error = fill(inode, start_blk, len_blks, fieinfo);
+               error = ext4_fill_es_cache_info(inode, start_blk, len_blks,
+                                               fieinfo);
        }
        return error;
 }
@@ -5076,8 +4884,7 @@ static int _ext4_fiemap(struct inode *inode,
 int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
                __u64 start, __u64 len)
 {
-       return _ext4_fiemap(inode, fieinfo, start, len,
-                           ext4_fill_fiemap_extents);
+       return _ext4_fiemap(inode, fieinfo, start, len, false);
 }
 
 int ext4_get_es_cache(struct inode *inode, struct fiemap_extent_info *fieinfo,
@@ -5093,8 +4900,7 @@ int ext4_get_es_cache(struct inode *inode, struct fiemap_extent_info *fieinfo,
                        return 0;
        }
 
-       return _ext4_fiemap(inode, fieinfo, start, len,
-                           ext4_fill_es_cache_info);
+       return _ext4_fiemap(inode, fieinfo, start, len, true);
 }
 
 
index fad82d0..e7db6ee 100644 (file)
@@ -1857,47 +1857,6 @@ out:
        return error;
 }
 
-int ext4_inline_data_fiemap(struct inode *inode,
-                           struct fiemap_extent_info *fieinfo,
-                           int *has_inline, __u64 start, __u64 len)
-{
-       __u64 physical = 0;
-       __u64 inline_len;
-       __u32 flags = FIEMAP_EXTENT_DATA_INLINE | FIEMAP_EXTENT_NOT_ALIGNED |
-               FIEMAP_EXTENT_LAST;
-       int error = 0;
-       struct ext4_iloc iloc;
-
-       down_read(&EXT4_I(inode)->xattr_sem);
-       if (!ext4_has_inline_data(inode)) {
-               *has_inline = 0;
-               goto out;
-       }
-       inline_len = min_t(size_t, ext4_get_inline_size(inode),
-                          i_size_read(inode));
-       if (start >= inline_len)
-               goto out;
-       if (start + len < inline_len)
-               inline_len = start + len;
-       inline_len -= start;
-
-       error = ext4_get_inode_loc(inode, &iloc);
-       if (error)
-               goto out;
-
-       physical = (__u64)iloc.bh->b_blocknr << inode->i_sb->s_blocksize_bits;
-       physical += (char *)ext4_raw_inode(&iloc) - iloc.bh->b_data;
-       physical += offsetof(struct ext4_inode, i_block);
-
-       brelse(iloc.bh);
-out:
-       up_read(&EXT4_I(inode)->xattr_sem);
-       if (physical)
-               error = fiemap_fill_next_extent(fieinfo, start, physical,
-                                               inline_len, flags);
-       return (error < 0 ? error : 0);
-}
-
 int ext4_inline_data_truncate(struct inode *inode, int *has_inline)
 {
        handle_t *handle;