Re: delayed allocatiou result in Oops

[Mingming Cao <cmm@us.ibm.com>]

On Wed, 2007-06-20 at 12:15 +0400, Alex Tomas wrote:
> Mingming Cao wrote:
> > Hmm, PageMappedToDisk is probably not sufficient enough for pagesize!
> > =blocksize. Is that the reason we need page->private to pass the
> > request?
> 
> PageMappedToDisk isn't enough in that case, definitely. bh is the way
> to track state of each block (this is what i'm implementing now), but
> I think current nobh version (per-page flags are used) is valuable.
> 
> > That's good to know, thanks for the update. So probably above error case
> > handling will be addressed in the new version? 
> 
> well, you actually can move that SetPagePrivate() few lines above
> (Dmitriy already tested this).

Like this?

Index: linux-2.6.22-rc5/fs/ext4/writeback.c
===================================================================
--- linux-2.6.22-rc5.orig/fs/ext4/writeback.c	2007-06-20 16:41:26.000000000 -0700
+++ linux-2.6.22-rc5/fs/ext4/writeback.c	2007-06-20 16:44:10.000000000 -0700
@@ -918,16 +918,16 @@ int ext4_wb_commit_write(struct file *fi
 	wb_debug("commit page %lu (%u-%u) for inode %lu\n",
 			page->index, from, to, inode->i_ino);
 
-	/* mark page private so that we get
-	 * called to invalidate/release page */
-	SetPagePrivate(page);
-
 	if (!PageBooked(page) && !PageMappedToDisk(page)) {
 		/* ->prepare_write() observed that block for this
 		 * page hasn't been allocated yet. there fore it
 		 * asked to reserve block for later allocation */
 		BUG_ON(page->private == 0);
 		page->private = 0;
+		/* mark page private so that we get
+		 * called to invalidate/release page */
+		SetPagePrivate(page);
+
 		err = ext4_wb_reserve_space_page(page, 1);
 		if (err)
 			return err;