[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20170612181447.GG19206@htj.duckdns.org>
Date: Mon, 12 Jun 2017 14:14:47 -0400
From: Tejun Heo <tj@...nel.org>
To: Shaohua Li <shli@...nel.org>
Cc: linux-kernel@...r.kernel.org, linux-block@...r.kernel.org,
gregkh@...uxfoundation.org, hch@....de, axboe@...com,
rostedt@...dmis.org, lizefan@...wei.com, Kernel-team@...com,
Shaohua Li <shli@...com>
Subject: Re: [PATCH 02/11] kernfs: use idr instead of ida to manage inode
number
Hello,
On Fri, Jun 02, 2017 at 02:53:55PM -0700, Shaohua Li wrote:
> @@ -630,7 +633,11 @@ static struct kernfs_node *__kernfs_new_node(struct kernfs_root *root,
> if (!kn)
> goto err_out1;
>
> - ret = ida_simple_get(&root->ino_ida, 1, 0, GFP_KERNEL);
> + idr_preload(GFP_KERNEL);
> + spin_lock(&kernfs_idr_lock);
> + ret = idr_alloc(&root->ino_idr, kn, 1, 0, GFP_ATOMIC);
> + spin_unlock(&kernfs_idr_lock);
> + idr_preload_end();
So, this ends up populating the idr w/ a kn which isn't fully
initialized yet, which can lead to bugs which are difficult to hunt
down as we're gonna allow kn's to be looked up through ino / gen. We
probably should allocate with NULL here and populate the actual
pointer with idr_replace() after the kn is fully initialized / online.
Thanks.
--
tejun
Powered by blists - more mailing lists