From: Marek Lindner <lindner_marek@yahoo.de>
To: The list for a Better Approach To Mobile Ad-hoc Networking
<b.a.t.m.a.n@lists.open-mesh.org>
Subject: Re: [B.A.T.M.A.N.] [PATCH 5/7] batman-adv: Make bat_priv->curr_gw an rcu protected pointer
Date: Tue, 8 Feb 2011 14:18:37 +0100 [thread overview]
Message-ID: <201102081418.38151.lindner_marek@yahoo.de> (raw)
In-Reply-To: <1296832896-30081-6-git-send-email-linus.luessing@ascom.ch>
On Friday 04 February 2011 16:21:34 Linus Lüssing wrote:
> + if (curr_gw) {
> + rcu_read_unlock();
> return;
> + }
>
> - rcu_read_lock();
> if (hlist_empty(&bat_priv->gw_list)) {
> - rcu_read_unlock();
>
> - if (bat_priv->curr_gw) {
> + if (curr_gw) {
> + rcu_read_unlock();
> bat_dbg(DBG_BATMAN, bat_priv,
> "Removing selected gateway - "
> "no gateway in range\n");
> gw_deselect(bat_priv);
> }
> + else
> + rcu_read_unlock();
This is a bit odd here - we return if curr_gw is not NULL and later we print a
message if curr_gw is not NULL ?
The issue existed before your patch came along.
> void gw_check_election(struct bat_priv *bat_priv, struct orig_node
> *orig_node) {
> - struct gw_node *curr_gateway_tmp = bat_priv->curr_gw;
> + struct gw_node *curr_gateway_tmp;
> uint8_t gw_tq_avg, orig_tq_avg;
>
> + rcu_read_lock();
> + curr_gateway_tmp = rcu_dereference(bat_priv->curr_gw);
> if (!curr_gateway_tmp)
> - return;
> + goto rcu_unlock;
>
> if (!curr_gateway_tmp->orig_node)
> goto deselect;
It seems if we jump to "deselect" here the rcu_lock() will never be unlocked.
> @@ -316,7 +342,7 @@ void gw_node_purge(struct bat_priv *bat_priv)
> - if (bat_priv->curr_gw == gw_node)
> + if (rcu_dereference(bat_priv->curr_gw) == gw_node)
> gw_deselect(bat_priv);
At this point we neither hold a rcu_lock() nor the newly created spinlock ...
> spinlock_t gw_list_lock; /* protects gw_list */
> + spinlock_t curr_gw_lock; /* protects curr_gw updates */
Would speak anything against re-using the gw_list_lock ?
Regards,
Marek
next prev parent reply other threads:[~2011-02-08 13:18 UTC|newest]
Thread overview: 44+ messages / expand[flat|nested] mbox.gz Atom feed top
2011-01-30 1:52 [B.A.T.M.A.N.] Reordered rcu refcounting Sven Eckelmann
2011-01-30 1:52 ` [B.A.T.M.A.N.] [PATCH 1/4] batman-adv: Correct rcu refcounting for gw_node Sven Eckelmann
2011-02-02 17:37 ` [B.A.T.M.A.N.] [PATCH] " Linus Lüssing
2011-02-02 19:49 ` Marek Lindner
2011-02-02 20:43 ` Linus Lüssing
2011-02-02 21:42 ` Sven Eckelmann
2011-02-03 0:19 ` Marek Lindner
2011-02-03 9:55 ` Linus Lüssing
2011-02-03 10:01 ` Sven Eckelmann
2011-02-03 14:43 ` [B.A.T.M.A.N.] [PATCH 1/3] " Linus Lüssing
2011-02-03 17:56 ` Sven Eckelmann
2011-02-03 14:43 ` [B.A.T.M.A.N.] [PATCH 2/3] batman-adv: add __rcu annotations " Linus Lüssing
2011-02-03 14:43 ` [B.A.T.M.A.N.] [PATCH 3/3] batman-adv: Increase orig_node refcount before releasing rcu read lock Linus Lüssing
2011-01-30 1:52 ` [B.A.T.M.A.N.] [PATCH 2/4] batman-adv: Correct rcu refcounting for softif_neigh Sven Eckelmann
2011-01-30 1:52 ` [B.A.T.M.A.N.] [PATCH 3/4] batman-adv: Correct rcu refcounting for batman_if Sven Eckelmann
2011-01-30 1:52 ` [B.A.T.M.A.N.] [PATCH 4/4] batman-adv: Correct rcu refcounting for neigh_node Sven Eckelmann
2011-01-30 15:20 ` [B.A.T.M.A.N.] [PATCHv2 " Sven Eckelmann
2011-02-03 17:02 ` [B.A.T.M.A.N.] Reordered rcu refcounting Marek Lindner
2011-02-03 17:03 ` [B.A.T.M.A.N.] [PATCH 1/4] batman-adv: Correct rcu refcounting for gw_node Marek Lindner
2011-02-03 17:03 ` [B.A.T.M.A.N.] [PATCH 2/4] batman-adv: Correct rcu refcounting for softif_neigh Marek Lindner
2011-02-03 17:03 ` [B.A.T.M.A.N.] [PATCH 3/4] batman-adv: Correct rcu refcounting for batman_if Marek Lindner
2011-02-03 17:03 ` [B.A.T.M.A.N.] [PATCH 4/4] batman-adv: Correct rcu refcounting for neigh_node Marek Lindner
2011-02-04 12:59 ` [B.A.T.M.A.N.] [PATCHv2 " Marek Lindner
2011-02-04 15:21 ` [B.A.T.M.A.N.] Reordered rcu refcounting, v3 Linus Lüssing
2011-02-04 15:21 ` [B.A.T.M.A.N.] [PATCH 1/7] batman-adv: Correct rcu refcounting for gw_node Linus Lüssing
2011-02-10 11:01 ` Linus Lüssing
2011-02-04 15:21 ` [B.A.T.M.A.N.] [PATCH 2/7] batman-adv: Correct rcu refcounting for softif_neigh Linus Lüssing
2011-02-10 12:45 ` Linus Lüssing
2011-02-10 13:57 ` Marek Lindner
2011-02-12 21:23 ` Linus Lüssing
2011-02-04 15:21 ` [B.A.T.M.A.N.] [PATCH 3/7] batman-adv: Correct rcu refcounting for batman_if Linus Lüssing
2011-02-04 15:21 ` [B.A.T.M.A.N.] [PATCH 4/7] batman-adv: Correct rcu refcounting for neigh_node Linus Lüssing
2011-02-04 15:21 ` [B.A.T.M.A.N.] [PATCH 5/7] batman-adv: Make bat_priv->curr_gw an rcu protected pointer Linus Lüssing
2011-02-08 13:18 ` Marek Lindner [this message]
2011-02-10 10:42 ` Linus Lüssing
2011-02-10 14:25 ` Marek Lindner
2011-02-12 21:21 ` Linus Lüssing
2011-02-12 21:21 ` [B.A.T.M.A.N.] [PATCHv2 " Linus Lüssing
2011-02-13 21:09 ` Marek Lindner
2011-02-04 15:21 ` [B.A.T.M.A.N.] [PATCH 6/7] batman-adv: add __rcu annotations for gw_node Linus Lüssing
2011-02-13 21:10 ` Marek Lindner
2011-02-04 15:21 ` [B.A.T.M.A.N.] [PATCH 7/7] batman-adv: Increase orig_node refcount before releasing rcu read lock Linus Lüssing
2011-02-04 16:33 ` [B.A.T.M.A.N.] [PATCHv2 " Linus Lüssing
2011-02-13 21:11 ` [B.A.T.M.A.N.] [PATCH " Marek Lindner
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=201102081418.38151.lindner_marek@yahoo.de \
--to=lindner_marek@yahoo.de \
--cc=b.a.t.m.a.n@lists.open-mesh.org \
/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;
as well as URLs for NNTP newsgroup(s).