Re: 1352 NUL bytes at the end of a page? (was Re: Assertion `s && s->tree' failed: The saga continues.)

[Chris Mason <mason@suse.com>]

On Mon, 2004-05-17 at 11:23, Chris Mason wrote:

> You've described it correctly for reiserfs though, we unlock the page
> too soon.  I'll fix the page locking for reiserfs_file_write.  Steven,
> we need to figure out why you're seeing this on ext3.  

Steven, could you give this a try as well?  It is against 2.6.6-mm3, but
should work against vanilla too:

reiserfs_file_write unlocks the pages it operated on before updating
i_size.  This can lead to races with writepage, who checks i_size when
deciding how much of the file to zero out.

This patch also replaces SetPageReferenced with mark_page_accessed() in
reiserfs_file_write

Index: linux.mm/fs/reiserfs/file.c
===================================================================
--- linux.mm.orig/fs/reiserfs/file.c	2004-05-17 13:42:02.000000000 -0400
+++ linux.mm/fs/reiserfs/file.c	2004-05-17 16:22:35.135105528 -0400
@@ -10,6 +10,7 @@
 #include <linux/smp_lock.h>
 #include <asm/uaccess.h>
 #include <linux/pagemap.h>
+#include <linux/swap.h>
 #include <linux/writeback.h>
 #include <linux/blkdev.h>
 #include <linux/buffer_head.h>
@@ -678,10 +679,6 @@
 			     // we only remember error status to report it on
 			     // exit.
 	write_bytes-=count;
-	SetPageReferenced(page);
-	unlock_page(page); // We unlock the page as it was locked by earlier call
-			  // to grab_cache_page
-	page_cache_release(page);
     }
     /* now that we've gotten all the ordered buffers marked dirty,
      * we can safely update i_size and close any running transaction
@@ -718,6 +715,17 @@
 	reiserfs_write_unlock(inode->i_sb);
     }
     th->t_trans_id = 0;
+
+    /* 
+     * we have to unlock the pages after updating i_size, otherwise
+     * we race with writepage
+     */
+    for ( i = 0; i < num_pages ; i++) {
+	struct page *page=prepared_pages[i];
+	unlock_page(page); 
+	mark_page_accessed(page);
+	page_cache_release(page);
+    }
     return retval;
 }