[PATCH] some more fixes for inode.c

[PATCH] some more fixes for inode.c

[Rik van Riel]

Hi Marcelo, David, 

when looking over the 2.4.25-pre6-rmap patch, I found two more
chunks in inode.c that might be needed/useful.  

The first chunk refiles the inode on the inode_unused_pagecache
list if needed, but I'm not 100% sure we need that change, since
maybe only completely unused inodes can end up here. David ?
It should be safe, though.

The second chunk make ssure to also remove the quota stuff for
inodes on the inode_unused_pagecache list, in the same way it
acts the other lists.


diff -Nru a/fs/inode.c b/fs/inode.c
--- a/fs/inode.c	Mon Jan 19 20:44:17 2004
+++ b/fs/inode.c	Mon Jan 19 20:44:17 2004
@@ -1234,10 +1235,8 @@
 				BUG();
 		} else {
 			if (!list_empty(&inode->i_hash)) {
-				if (!(inode->i_state & (I_DIRTY|I_LOCK))) {
-					list_del(&inode->i_list);
-					list_add(&inode->i_list, &inode_unused);
-				}
+				if (!(inode->i_state & (I_DIRTY|I_LOCK))) 
+					__refile_inode(inode);
 				inodes_stat.nr_unused++;
 				spin_unlock(&inode_lock);
 				if (!sb || (sb->s_flags & MS_ACTIVE))
@@ -1414,6 +1413,11 @@
 			remove_inode_dquot_ref(inode, type, &tofree_head);
 	}
 	list_for_each(act_head, &inode_unused) {
+		inode = list_entry(act_head, struct inode, i_list);
+		if (inode->i_sb == sb && IS_QUOTAINIT(inode))
+			remove_inode_dquot_ref(inode, type, &tofree_head);
+	}
+	list_for_each(act_head, &inode_unused_pagecache) {
 		inode = list_entry(act_head, struct inode, i_list);
 		if (inode->i_sb == sb && IS_QUOTAINIT(inode))
 			remove_inode_dquot_ref(inode, type, &tofree_head);

Re: [PATCH] some more fixes for inode.c

[David Woodhouse]

On Tue, 2004-01-20 at 08:06 -0500, Rik van Riel wrote:
> The first chunk refiles the inode on the inode_unused_pagecache
> list if needed, 

Looks sane.

> but I'm not 100% sure we need that change, since
> maybe only completely unused inodes can end up here. David ?

No, I don't see any reason why that should be the case; you can get
iput() called for an inode which happens to have data in the page cache.
This part of the patch is needed.

-- 
dwmw2

Re: [PATCH] some more fixes for inode.c

[Rik van Riel]

On Tue, 20 Jan 2004, David Woodhouse wrote:
> On Tue, 2004-01-20 at 08:06 -0500, Rik van Riel wrote:

> > but I'm not 100% sure we need that change, since
> > maybe only completely unused inodes can end up here. David ?
> 
> No, I don't see any reason why that should be the case; you can get
> iput() called for an inode which happens to have data in the page cache.
> This part of the patch is needed.

OK, thanks for confirming my suspicions.

Marcelo, could you please apply the patch ?

-- 
"Debugging is twice as hard as writing the code in the first place.
Therefore, if you write the code as cleverly as possible, you are,
by definition, not smart enough to debug it." - Brian W. Kernighan

Re: [PATCH] some more fixes for inode.c

[Rusty Russell]

In message <Pine.LNX.4.44.0401200803150.15071-100000@chimarrao.boston.redhat.com> you write:
>  	}
>  	list_for_each(act_head, &inode_unused) {
> +		inode = list_entry(act_head, struct inode, i_list);
> +		if (inode->i_sb == sb && IS_QUOTAINIT(inode))
> +			remove_inode_dquot_ref(inode, type, &tofree_head);
> +	}
> +	list_for_each(act_head, &inode_unused_pagecache) {
>  		inode = list_entry(act_head, struct inode, i_list);

list_for_each_entry() perhaps?

It's in 2.4, as well.

Cheers,
Rusty.
--
  Anyone who quotes me in their sig is an idiot. -- Rusty Russell.