afs: Fix directory read/modify race

Because parsing of the directory wasn't being done under any sort of lock,
the pages holding the directory content can get invalidated whilst the
parsing is ongoing.

Further, the directory page check function gets called outside of the page
lock, so if the page gets cleared or updated, this may return reports of
bad magic numbers in the directory page.

Also, the directory may change size whilst checking and parsing are
ongoing, so more care needs to be taken here.

Fix this by:

 (1) Perform the page check from the page filling function before we set
     PageUptodate and drop the page lock.

 (2) Check for the file having shrunk and the page having been abandoned
     before checking the page contents.

 (3) Lock the page whilst parsing it for the directory iterator.

Whilst we're at it, add a tracepoint to report check failure.

Signed-off-by: David Howells <dhowells@redhat.com>

diff --git a/fs/afs/dir.c b/fs/afs/dir.c
index 53f3917440e7c4d642a54eadd6534e6848ed95a8..ecda0e6a9f7e03e9803cca29adf276974f9a7477 100644 (file)
--- a/fs/afs/dir.c
+++ b/fs/afs/dir.c
@@ -130,10 +130,11 @@ struct afs_lookup_cookie {
 /*
  * check that a directory page is valid
  */
-static inline bool afs_dir_check_page(struct inode *dir, struct page *page)
+bool afs_dir_check_page(struct inode *dir, struct page *page)
 {
        struct afs_dir_page *dbuf;
-       loff_t latter;
+       struct afs_vnode *vnode = AFS_FS_I(dir);
+       loff_t latter, i_size, off;
        int tmp, qty;
 
 #if 0
@@ -150,8 +151,15 @@ static inline bool afs_dir_check_page(struct inode *dir, struct page *page)
        }
 #endif
 
-       /* determine how many magic numbers there should be in this page */
-       latter = dir->i_size - page_offset(page);
+       /* Determine how many magic numbers there should be in this page, but
+        * we must take care because the directory may change size under us.
+        */
+       off = page_offset(page);
+       i_size = i_size_read(dir);
+       if (i_size <= off)
+               goto checked;
+
+       latter = i_size - off;
        if (latter >= PAGE_SIZE)
                qty = PAGE_SIZE;
        else
@@ -162,13 +170,15 @@ static inline bool afs_dir_check_page(struct inode *dir, struct page *page)
        dbuf = page_address(page);
        for (tmp = 0; tmp < qty; tmp++) {
                if (dbuf->blocks[tmp].pagehdr.magic != AFS_DIR_MAGIC) {
-                       printk("kAFS: %s(%lu): bad magic %d/%d is %04hx\n",
+                       printk("kAFS: %s(%lx): bad magic %d/%d is %04hx\n",
                               __func__, dir->i_ino, tmp, qty,
                               ntohs(dbuf->blocks[tmp].pagehdr.magic));
+                       trace_afs_dir_check_failed(vnode, off, i_size);
                        goto error;
                }
        }
 
+checked:
        SetPageChecked(page);
        return true;
 
@@ -183,6 +193,7 @@ error:
 static inline void afs_dir_put_page(struct page *page)
 {
        kunmap(page);
+       unlock_page(page);
        put_page(page);
 }
 
@@ -197,9 +208,10 @@ static struct page *afs_dir_get_page(struct inode *dir, unsigned long index,
 
        page = read_cache_page(dir->i_mapping, index, afs_page_filler, key);
        if (!IS_ERR(page)) {
+               lock_page(page);
                kmap(page);
                if (unlikely(!PageChecked(page))) {
-                       if (PageError(page) || !afs_dir_check_page(dir, page))
+                       if (PageError(page))
                                goto fail;
                }
        }
@@ -384,8 +396,7 @@ out:
  */
 static int afs_readdir(struct file *file, struct dir_context *ctx)
 {
-       return afs_dir_iterate(file_inode(file),
-                             ctx, file->private_data);
+       return afs_dir_iterate(file_inode(file), ctx, file->private_data);
 }
 
 /*

diff --git a/fs/afs/file.c b/fs/afs/file.c
index 1f26ac9f816d60609db0011471f7be825ce53e85..5786f68f87f143e4d3a8c786f27382f8113328f9 100644 (file)
--- a/fs/afs/file.c
+++ b/fs/afs/file.c
@@ -232,6 +232,11 @@ int afs_page_filler(void *data, struct page *page)
                 * page */
                ret = afs_fetch_data(vnode, key, req);
                afs_put_read(req);
+
+               if (ret >= 0 && S_ISDIR(inode->i_mode) &&
+                   !afs_dir_check_page(inode, page))
+                       ret = -EIO;
+
                if (ret < 0) {
                        if (ret == -ENOENT) {
                                _debug("got NOENT from server"

diff --git a/fs/afs/internal.h b/fs/afs/internal.h
index aad12546e0ea61725a0dddbd23b2bc4669486515..6aa6e9957c447437d4f4a748ff1c7e18c84343b2 100644 (file)
--- a/fs/afs/internal.h
+++ b/fs/afs/internal.h
@@ -622,6 +622,7 @@ extern bool afs_cm_incoming_call(struct afs_call *);
 /*
  * dir.c
  */
+extern bool afs_dir_check_page(struct inode *, struct page *);
 extern const struct inode_operations afs_dir_inode_operations;
 extern const struct dentry_operations afs_fs_dentry_operations;
 extern const struct file_operations afs_dir_file_operations;

diff --git a/include/trace/events/afs.h b/include/trace/events/afs.h
index c780f4c39a53f7f59d5be937cff68a008f9e0823..9cfb7657b72c7506618d1c68795b635522c96b1d 100644 (file)
--- a/include/trace/events/afs.h
+++ b/include/trace/events/afs.h
@@ -381,6 +381,27 @@ TRACE_EVENT(afs_sent_pages,
                      __entry->cursor, __entry->ret)
            );
 
+TRACE_EVENT(afs_dir_check_failed,
+           TP_PROTO(struct afs_vnode *vnode, loff_t off, loff_t i_size),
+
+           TP_ARGS(vnode, off, i_size),
+
+           TP_STRUCT__entry(
+                   __field(struct afs_vnode *,         vnode           )
+                   __field(loff_t,                     off             )
+                   __field(loff_t,                     i_size          )
+                            ),
+
+           TP_fast_assign(
+                   __entry->vnode = vnode;
+                   __entry->off = off;
+                   __entry->i_size = i_size;
+                          ),
+
+           TP_printk("vn=%p %llx/%llx",
+                     __entry->vnode, __entry->off, __entry->i_size)
+           );
+
 #endif /* _TRACE_AFS_H */
 
 /* This part must be outside protection */