From: Andrei Emeltchenko <andrei.emeltchenko.news@gmail.com>
To: "Gustavo F. Padovan" <padovan@profusion.mobi>
Cc: linux-bluetooth@vger.kernel.org, marcel@holtmann.org
Subject: Re: [RFC 11/22] Bluetooth: Use RCU to manipulate chan_list
Date: Thu, 26 Jan 2012 15:20:20 +0200 [thread overview]
Message-ID: <20120126132019.GC3300@aemeltch-MOBL1> (raw)
In-Reply-To: <1324157382-1815-12-git-send-email-padovan@profusion.mobi>
Hi Gustavo,
On Sat, Dec 17, 2011 at 07:29:31PM -0200, Gustavo F. Padovan wrote:
> From: "Gustavo F. Padovan" <padovan@profusion.mobi>
>
> Instead of using tasklet_disable() to prevent acess to the channel use, we
> can use RCU and improve the performance of our code.
>
> Signed-off-by: Gustavo F. Padovan <padovan@profusion.mobi>
...
> @@ -986,10 +984,10 @@ int hci_chan_del(struct hci_chan *chan)
>
> void hci_chan_list_flush(struct hci_conn *conn)
> {
> - struct hci_chan *chan, *tmp;
> + struct hci_chan *chan;
>
> BT_DBG("conn %p", conn);
>
> - list_for_each_entry_safe(chan, tmp, &conn->chan_list, list)
> + list_for_each_entry_rcu(chan, &conn->chan_list, list)
> hci_chan_del(chan);
> }
I feel that this code is not exactly correct. hci_chan_del frees chan:
<------8<--------------------------------------------------------
| int hci_chan_del(struct hci_chan *chan)
| {
| struct hci_conn *conn = chan->conn;
| struct hci_dev *hdev = conn->hdev;
|
| BT_DBG("%s conn %p chan %p", hdev->name, conn, chan);
|
| list_del_rcu(&chan->list);
|
| synchronize_rcu();
|
| skb_queue_purge(&chan->data_q);
| kfree(chan);
|
| return 0;
| }
<------8<--------------------------------------------------------
and then list_for_each_entry_rcu references chan->member.next which is not
safe ;)
I've sent patch reverting this change.
Best regards
Andrei Emeltchenko
next prev parent reply other threads:[~2012-01-26 13:20 UTC|newest]
Thread overview: 34+ messages / expand[flat|nested] mbox.gz Atom feed top
2011-12-17 21:29 [RFC 00/22] Bluetooth: change tasklets to workqueue Gustavo F. Padovan
2011-12-17 21:29 ` [RFC 01/22] Bluetooth: Process recv path in a workqueue instead of a tasklet Gustavo F. Padovan
2011-12-17 21:29 ` [RFC 02/22] Bluetooth: Replace spin_lock by mutex in hci_dev Gustavo F. Padovan
2011-12-17 21:29 ` [RFC 03/22] Bluetooth: Use delayed_work for connection timeout Gustavo F. Padovan
2011-12-17 21:29 ` [RFC 04/22] Bluetooth: Use delayed work for advertisiment cache timeout Gustavo F. Padovan
2011-12-17 21:29 ` [RFC 05/22] Bluetooth: hci_conn_auto_accept() doesn't need locking Gustavo F. Padovan
2011-12-17 21:29 ` [RFC 06/22] Bluetooth: Move L2CAP timers to workqueue Gustavo F. Padovan
2011-12-17 21:29 ` [RFC 07/22] Bluetooth: Don't use spin_lock socket lock anymore Gustavo F. Padovan
2011-12-17 21:29 ` [RFC 08/22] Bluetooth: Remove sk_backlog usage from L2CAP Gustavo F. Padovan
2011-12-17 21:29 ` [RFC 09/22] Bluetooth: move hci_task_lock to mutex Gustavo F. Padovan
2011-12-17 21:29 ` [RFC 10/22] Bluetooth: convert chan_lock " Gustavo F. Padovan
2011-12-17 21:29 ` [RFC 11/22] Bluetooth: Use RCU to manipulate chan_list Gustavo F. Padovan
2011-12-17 21:29 ` [RFC 12/22] Bluetooth: convert conn hash to RCU Gustavo F. Padovan
2011-12-17 21:29 ` [RFC 13/22] Bluetooth: Don't disable tasklets to call hdev->notify() Gustavo F. Padovan
2011-12-17 21:29 ` [RFC 14/22] Bluetooth: Move command task to workqueue Gustavo F. Padovan
2011-12-17 21:29 ` [RFC 15/22] Bluetooth: convert tx_task " Gustavo F. Padovan
2011-12-17 21:29 ` [RFC 16/22] Bluetooth: convert info timer to delayed_work Gustavo F. Padovan
2011-12-17 21:29 ` [RFC 17/22] Bluetooth: remove power_on work_struct Gustavo F. Padovan
2011-12-17 21:29 ` [RFC 18/22] Bluetooth: invert locking order in connect path Gustavo F. Padovan
2011-12-17 21:29 ` [RFC 19/22] Bluetooth: Change l2cap chan_list to use RCU Gustavo F. Padovan
2011-12-17 21:29 ` [RFC 20/22] Bluetooth: move power_off to system workqueue Gustavo F. Padovan
2011-12-17 21:29 ` [RFC 21/22] Bluetooth: Use new alloc_workqueue() Gustavo F. Padovan
2011-12-17 21:29 ` [RFC 22/22] Bluetooth: Remove work_add and work_del from hci_sysfs Gustavo F. Padovan
2011-12-17 22:15 ` [RFC 19/22] Bluetooth: Change l2cap chan_list to use RCU Marcel Holtmann
2011-12-19 10:42 ` Andrei Emeltchenko
2011-12-19 13:53 ` Gustavo Padovan
2011-12-17 22:13 ` [RFC 17/22] Bluetooth: remove power_on work_struct Marcel Holtmann
2012-01-26 13:20 ` Andrei Emeltchenko [this message]
2011-12-19 9:58 ` [RFC 10/22] Bluetooth: convert chan_lock to mutex Andrei Emeltchenko
2011-12-19 9:53 ` [RFC 07/22] Bluetooth: Don't use spin_lock socket lock anymore Andrei Emeltchenko
2011-12-19 11:05 ` [RFC 06/22] Bluetooth: Move L2CAP timers to workqueue Andrei Emeltchenko
2011-12-19 12:59 ` Ulisses Furquim
2011-12-17 21:34 ` [RFC 00/22] Bluetooth: change tasklets " Gustavo Padovan
2011-12-17 22:17 ` Marcel Holtmann
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20120126132019.GC3300@aemeltch-MOBL1 \
--to=andrei.emeltchenko.news@gmail.com \
--cc=linux-bluetooth@vger.kernel.org \
--cc=marcel@holtmann.org \
--cc=padovan@profusion.mobi \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox