jffs_remove calls d_delete(dentry). vfs_unlink then tries to access dentry->d_inode->i_sem
and calls d_delete(dentry).
Suggested patch:
--- inode-v23.c 20 Nov 2002 11:57:45 -0000 1.4
+++ inode-v23.c 20 Nov 2002 19:45:07 -0000
@@ -1063,8 +1063,6 @@
inode->i_ctime = dir->i_ctime;
mark_inode_dirty(inode);
- d_delete(dentry); /* This also frees the inode */
-
result = 0;
jffs_remove_end:
return result;
Can the JFFS guys verify that this doesn't cause a memory leak?
/Mikael