udf: Replace bkl with the UDF_I(inode)->i_data_sem for protect udf_inode_info struct

Replace bkl with the UDF_I(inode)->i_data_sem rw semaphore in
udf_release_file(), udf_symlink(), udf_symlink_filler(), udf_get_block(),
udf_block_map(), and udf_setattr(). The rule now is that any operation
on regular file's or symlink's extents (or generally allocation information
including goal block) needs to hold i_data_sem.

This work was supported by a hardware donation from the CE Linux Forum.

Signed-off-by: Alessio Igor Bogani <abogani@texware.it>
Signed-off-by: Jan Kara <jack@suse.cz>

6 files changed, 37 insertions, 19 deletions

diff --git a/fs/udf/file.c b/fs/udf/file.c
index 66b9e7e7e4c5..df0c5561cc7e 100644
--- a/fs/udf/file.c
+++ b/fs/udf/file.c
@@ -204,10 +204,10 @@ static int udf_release_file(struct inode *inode, struct file *filp)
 {
         if (filp->f_mode & FMODE_WRITE) {
                 mutex_lock(&inode->i_mutex);
-                lock_kernel();
+                down_write(&UDF_I(inode)->i_data_sem);
                 udf_discard_prealloc(inode);
                 udf_truncate_tail_extent(inode);
-                unlock_kernel();
+                up_write(&UDF_I(inode)->i_data_sem);
                 mutex_unlock(&inode->i_mutex);
         }
         return 0;
diff --git a/fs/udf/inode.c b/fs/udf/inode.c
index fa3c1541151c..b2fe4d7f20eb 100644
--- a/fs/udf/inode.c
+++ b/fs/udf/inode.c
@@ -301,10 +301,9 @@ static int udf_get_block(struct inode *inode, sector_t block,
         err = -EIO;
         new = 0;
         bh = NULL;
-
-        lock_kernel();
-
         iinfo = UDF_I(inode);
+
+        down_write(&iinfo->i_data_sem);
         if (block == iinfo->i_next_alloc_block + 1) {
                 iinfo->i_next_alloc_block++;
                 iinfo->i_next_alloc_goal++;
@@ -323,7 +322,7 @@ static int udf_get_block(struct inode *inode, sector_t block,
         map_bh(bh_result, inode->i_sb, phys);
 
 abort:
-        unlock_kernel();
+        up_write(&iinfo->i_data_sem);
         return err;
 }
 
@@ -1021,16 +1020,16 @@ void udf_truncate(struct inode *inode)
         if (IS_APPEND(inode) || IS_IMMUTABLE(inode))
                 return;
 
-        lock_kernel();
         iinfo = UDF_I(inode);
         if (iinfo->i_alloc_type == ICBTAG_FLAG_AD_IN_ICB) {
+                down_write(&iinfo->i_data_sem);
                 if (inode->i_sb->s_blocksize <
                                 (udf_file_entry_alloc_offset(inode) +
                                  inode->i_size)) {
                         udf_expand_file_adinicb(inode, inode->i_size, &err);
                         if (iinfo->i_alloc_type == ICBTAG_FLAG_AD_IN_ICB) {
                                 inode->i_size = iinfo->i_lenAlloc;
-                                unlock_kernel();
+                                up_write(&iinfo->i_data_sem);
                                 return;
                         } else
                                 udf_truncate_extents(inode);
@@ -1041,10 +1040,13 @@ void udf_truncate(struct inode *inode)
                                 offset - udf_file_entry_alloc_offset(inode));
                         iinfo->i_lenAlloc = inode->i_size;
                 }
+                up_write(&iinfo->i_data_sem);
         } else {
                 block_truncate_page(inode->i_mapping, inode->i_size,
                                     udf_get_block);
+                down_write(&iinfo->i_data_sem);
                 udf_truncate_extents(inode);
+                up_write(&iinfo->i_data_sem);
         }
 
         inode->i_mtime = inode->i_ctime = current_fs_time(inode->i_sb);
@@ -1052,7 +1054,6 @@ void udf_truncate(struct inode *inode)
                 udf_sync_inode(inode);
         else
                 mark_inode_dirty(inode);
-        unlock_kernel();
 }
 
 static void __udf_read_inode(struct inode *inode)
@@ -2043,7 +2044,7 @@ long udf_block_map(struct inode *inode, sector_t block)
         struct extent_position epos = {};
         int ret;
 
-        lock_kernel();
+        down_read(&UDF_I(inode)->i_data_sem);
 
         if (inode_bmap(inode, block, &epos, &eloc, &elen, &offset) ==
                                                 (EXT_RECORDED_ALLOCATED >> 30))
@@ -2051,7 +2052,7 @@ long udf_block_map(struct inode *inode, sector_t block)
         else
                 ret = 0;
 
-        unlock_kernel();
+        up_read(&UDF_I(inode)->i_data_sem);
         brelse(epos.bh);
 
         if (UDF_QUERY_FLAG(inode->i_sb, UDF_FLAG_VARCONV))
diff --git a/fs/udf/namei.c b/fs/udf/namei.c
index 701fcda18415..d5eb000ddddd 100644
--- a/fs/udf/namei.c
+++ b/fs/udf/namei.c
@@ -893,18 +893,18 @@ static int udf_symlink(struct inode *dir, struct dentry *dentry,
         struct udf_inode_info *iinfo;
         struct super_block *sb = dir->i_sb;
 
-        lock_kernel();
         inode = udf_new_inode(dir, S_IFLNK | S_IRWXUGO, &err);
         if (!inode)
                 goto out;
 
+        iinfo = UDF_I(inode);
+        down_write(&iinfo->i_data_sem);
         name = kmalloc(UDF_NAME_LEN, GFP_NOFS);
         if (!name) {
                 err = -ENOMEM;
                 goto out_no_entry;
         }
 
-        iinfo = UDF_I(inode);
         inode->i_data.a_ops = &udf_symlink_aops;
         inode->i_op = &udf_symlink_inode_operations;
 
@@ -1024,6 +1024,7 @@ static int udf_symlink(struct inode *dir, struct dentry *dentry,
         udf_write_fi(dir, &cfi, fi, &fibh, NULL, NULL);
         if (UDF_I(dir)->i_alloc_type == ICBTAG_FLAG_AD_IN_ICB)
                 mark_inode_dirty(dir);
+        up_write(&iinfo->i_data_sem);
         if (fibh.sbh != fibh.ebh)
                 brelse(fibh.ebh);
         brelse(fibh.sbh);
@@ -1032,10 +1033,10 @@ static int udf_symlink(struct inode *dir, struct dentry *dentry,
 
 out:
         kfree(name);
-        unlock_kernel();
         return err;
 
 out_no_entry:
+        up_write(&iinfo->i_data_sem);
         inode_dec_link_count(inode);
         iput(inode);
         goto out;
diff --git a/fs/udf/super.c b/fs/udf/super.c
index d2ec9f31e843..441b892cf85e 100644
--- a/fs/udf/super.c
+++ b/fs/udf/super.c
@@ -135,6 +135,7 @@ static struct inode *udf_alloc_inode(struct super_block *sb)
         ei->i_next_alloc_block = 0;
         ei->i_next_alloc_goal = 0;
         ei->i_strat4096 = 0;
+        init_rwsem(&ei->i_data_sem);
 
         return &ei->vfs_inode;
 }
diff --git a/fs/udf/symlink.c b/fs/udf/symlink.c
index 16064787d2b7..b1d4488b0f14 100644
--- a/fs/udf/symlink.c
+++ b/fs/udf/symlink.c
@@ -27,7 +27,6 @@
 #include <linux/mm.h>
 #include <linux/stat.h>
 #include <linux/pagemap.h>
-#include <linux/smp_lock.h>
 #include <linux/buffer_head.h>
 #include "udf_i.h"
 
@@ -78,13 +77,16 @@ static int udf_symlink_filler(struct file *file, struct page *page)
         int err = -EIO;
         unsigned char *p = kmap(page);
         struct udf_inode_info *iinfo;
+        uint32_t pos;
 
-        lock_kernel();
         iinfo = UDF_I(inode);
+        pos = udf_block_map(inode, 0);
+
+        down_read(&iinfo->i_data_sem);
         if (iinfo->i_alloc_type == ICBTAG_FLAG_AD_IN_ICB) {
                 symlink = iinfo->i_ext.i_data + iinfo->i_lenEAttr;
         } else {
-                bh = sb_bread(inode->i_sb, udf_block_map(inode, 0));
+                bh = sb_bread(inode->i_sb, pos);
 
                 if (!bh)
                         goto out;
@@ -95,14 +97,14 @@ static int udf_symlink_filler(struct file *file, struct page *page)
         udf_pc_to_char(inode->i_sb, symlink, inode->i_size, p);
         brelse(bh);
 
-        unlock_kernel();
+        up_read(&iinfo->i_data_sem);
         SetPageUptodate(page);
         kunmap(page);
         unlock_page(page);
         return 0;
 
 out:
-        unlock_kernel();
+        up_read(&iinfo->i_data_sem);
         SetPageError(page);
         kunmap(page);
         unlock_page(page);
diff --git a/fs/udf/udf_i.h b/fs/udf/udf_i.h
index e58d1de41073..d1bd31ea724e 100644
--- a/fs/udf/udf_i.h
+++ b/fs/udf/udf_i.h
@@ -1,6 +1,18 @@
 #ifndef _UDF_I_H
 #define _UDF_I_H
 
+/*
+ * The i_data_sem and i_mutex serve for protection of allocation information
+ * of a regular files and symlinks. This includes all extents belonging to
+ * the file/symlink, a fact whether data are in-inode or in external data
+ * blocks, preallocation, goal block information... When extents are read,
+ * i_mutex or i_data_sem must be held (for reading is enough in case of
+ * i_data_sem). When extents are changed, i_data_sem must be held for writing
+ * and also i_mutex must be held.
+ *
+ * For directories i_mutex is used for all the necessary protection.
+ */
+
 struct udf_inode_info {
         struct timespec         i_crtime;
         /* Physical address of inode */
@@ -21,6 +33,7 @@ struct udf_inode_info {
                 struct long_ad          *i_lad;
                 __u8            *i_data;
         } i_ext;
+        struct rw_semaphore     i_data_sem;
         struct inode vfs_inode;
 };