[1650] in SIPB_Linux_Development
Re: revised patch to (hopefully) fix and/or find afs cache corruption
daemon@ATHENA.MIT.EDU (Theodore Y. Ts'o)
Wed Apr 9 23:03:02 1997
Date: Wed, 9 Apr 1997 23:02:46 -0400
From: "Theodore Y. Ts'o" <tytso@MIT.EDU>
To: Erik Nygren <nygren@MIT.EDU>
Cc: linux-dev@MIT.EDU, linux-afs-bugs@MIT.EDU
In-Reply-To: Erik Nygren's message of Wed, 09 Apr 1997 21:13:33 EDT,
<199704100113.VAA01991@zocalo.mit.edu>
Here's one more patch to fs/inode.c which may help:
----------------------------------------------------------------
--- inode.c.~1~ Thu Sep 4 11:22:24 1980
+++ inode.c Sat Apr 5 22:42:22 1997
@@ -173,6 +173,7 @@
{
struct wait_queue * wait;
+ inode->i_count++;
lock_inode(inode);
truncate_inode_pages(inode, 0);
if (IS_WRITABLE(inode)) {
@@ -182,6 +184,7 @@
remove_inode_hash(inode);
remove_inode_free(inode);
wait = ((volatile struct inode *) inode)->i_wait;
+ inode->i_count--;
if (inode->i_count)
nr_free_inodes++;
memset(inode,0,sizeof(*inode));
----------------------------------------------------------------
This patch was suggested by Stephen Tweedie, to prevent a case where
while the inode is locked and we're waiting for it to become free,
another process manages to grab the inode anyway.
Yes, the change I made to lock_inode() wasn't really necessary.
- Ted
