Re: 2.6.0-test7-mm1

[Kirill Korotaev <dev@sw.ru>]

[-- Attachment #1: Type: text/plain, Size: 390 bytes --]

> Yup.  The "invalidate_inodes-speedup-fixes" and "invalidate_inodes-speedup"
> patches were not so great and need to be reverted.
I found another bug in invalidate_inodes-speedup.patch
introduced by WLI when doing forward porting:

-			list_del(&inode->i_list);
+			list_del(&inode->i_sb_list);

first list_del should be kept!!!

Patch fixing this issue and hugetlbfs is attached.

Kirill

[-- Attachment #2: invalidate_inodes-speedup-fix2.patch --]
[-- Type: text/x-diff, Size: 1143 bytes --]

--- linux-2.6.0-test7-mm1/fs/inode.c.ii	2003-10-17 12:26:59.000000000 +0400
+++ linux-2.6.0-test7-mm1/fs/inode.c	2003-10-17 12:55:54.000000000 +0400
@@ -303,6 +303,7 @@ static int invalidate_list(struct list_h
 		invalidate_inode_buffers(inode);
 		if (!atomic_read(&inode->i_count)) {
 			hlist_del_init(&inode->i_hash);
+			list_del(&inode->i_list);
 			list_del(&inode->i_sb_list);
 			list_add(&inode->i_list, dispose);
 			inode->i_state |= I_FREEING;
--- linux-2.6.0-test7-mm1/fs/hugetlbfs/inode.c.ii	2003-10-17 12:51:51.000000000 +0400
+++ linux-2.6.0-test7-mm1/fs/hugetlbfs/inode.c	2003-10-17 12:52:24.000000000 +0400
@@ -191,6 +191,7 @@ static void hugetlbfs_delete_inode(struc
 
 	hlist_del_init(&inode->i_hash);
 	list_del_init(&inode->i_list);
+	list_del_init(&inode->i_sb_list);
 	inode->i_state |= I_FREEING;
 	inodes_stat.nr_inodes--;
 	spin_unlock(&inode_lock);
@@ -233,6 +234,7 @@ static void hugetlbfs_forget_inode(struc
 	hlist_del_init(&inode->i_hash);
 out_truncate:
 	list_del_init(&inode->i_list);
+	list_del_init(&inode->i_sb_list);
 	inode->i_state |= I_FREEING;
 	inodes_stat.nr_inodes--;
 	spin_unlock(&inode_lock);