Bernhard Mogens Ege wrote:
>We have some trouble with jfs 1.0.24 (kernel 2.4.19, jfs as
>module). Our backup that starts at night at 1 o'clock is usually what
>triggers the bug (as in this case). Here is the log:
>
>Nov 26 01:09:03 boston kernel: jfs_clear_inode, active_ag = 101
>Nov 26 01:09:03 boston kernel: i_ino = 253958, i_mode = 100775
>Nov 26 01:09:03 boston kernel: assert(list_empty(&ji->mp_list))
>Nov 26 01:09:03 boston kernel: kernel BUG at inode.c:51!
>Nov 26 01:09:03 boston kernel: invalid operand: 0000
>Nov 26 01:09:03 boston kernel: CPU: 1
>Nov 26 01:09:03 boston kernel: EIP:
This is fixed in the File System in release 1.1.0. Here is the patch
below that will fix this problem.
We have a bad inode and we have already freed it.
--- /org/fs/jfs/inode.c Fri Nov 15 10:54:32 2002
+++ /fix/fs/jfs/inode.c Fri Nov 15 10:55:29 2002
@@ -39,6 +39,13 @@
{
struct jfs_inode_info *ji = JFS_IP(inode);
+ if (is_bad_inode(inode))
+ /*
+ * We free the fs-dependent structure before making the
+ * inode bad
+ */
+ return;
+
jFYI(1, ("jfs_clear_inode called ip = 0x%p\n", inode));
if (ji->active_ag != -1) {
Thanks,
Steve
_______________________________________________
Jfs-discussion mailing list
[EMAIL PROTECTED]
http://www-124.ibm.com/developerworks/oss/mailman/listinfo/jfs-discussion
