From mboxrd@z Thu Jan  1 00:00:00 1970
Return-Path: <linux-wpan-owner@vger.kernel.org>
Received: from mail-wi0-f171.google.com ([209.85.212.171]:36843 "EHLO
	mail-wi0-f171.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org
	with ESMTP id S1750778AbbEIQnD (ORCPT
	<rfc822;linux-wpan@vger.kernel.org>); Sat, 9 May 2015 12:43:03 -0400
Received: by wizk4 with SMTP id k4so61960517wiz.1
        for <linux-wpan@vger.kernel.org>; Sat, 09 May 2015 09:43:01 -0700 (PDT)
Date: Sat, 9 May 2015 18:42:54 +0200
From: Alexander Aring <alex.aring@gmail.com>
Subject: Re: Ieee802154 socket problem with SIOCGIFADDR
Message-ID: <20150509164227.GA694@omega>
References: <554E1C92.7040306@sssup.it>
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Disposition: inline
In-Reply-To: <554E1C92.7040306@sssup.it>
Sender: linux-wpan-owner@vger.kernel.org
List-ID: <linux-wpan.vger.kernel.org>
To: Matteo Petracca <matteo.petracca@sssup.it>
Cc: linux-wpan@vger.kernel.org

Hi,

On Sat, May 09, 2015 at 04:41:22PM +0200, Matteo Petracca wrote:
> Dear all,
> I am working on a BeagleboneBlack with transceiver at86rf233.
> 
> When using ieee802154 sockets in my code I have:
> 
> ret = ioctl(socketfd, SIOCGIFADDR, &ifr);
> 
> and everything works fine with kernel 3.8.13.
> 
> In kernel 4.0.1 the request fails with the following kernel messages:
> 
> [  846.038365] RTNL: assertion failed at net/mac802154/iface.c (65)
> [  846.044629] CPU: 0 PID: 2350 Comm: test Not tainted 4.0.1-bone1 #1
> [  846.044653] Hardware name: Generic AM33XX (Flattened Device Tree)
> [  846.044747] [<c0011f35>] (unwind_backtrace) from [<c00101b5>]
> (show_stack+0x11/0x14)
> [  846.044876] [<c00101b5>] (show_stack) from [<bf8502f3>]
> (mac802154_wpan_ioctl+0xb6/0x100 [mac802154])
> [  846.044972] [<bf8502f3>] (mac802154_wpan_ioctl [mac802154]) from
> [<bf8b738f>] (ieee802154_sock_ioctl+0xf6/0x14c [ieee802154_socket])
> [  846.045044] [<bf8b738f>] (ieee802154_sock_ioctl [ieee802154_socket]) from
> [<c05195fb>] (sock_ioctl+0xd3/0x1bc)
> [  846.045094] [<c05195fb>] (sock_ioctl) from [<c00fca4d>]
> (do_vfs_ioctl+0x2cd/0x400)
> [  846.045133] [<c00fca4d>] (do_vfs_ioctl) from [<c00fcbcd>]
> (SyS_ioctl+0x4d/0x58)
> [  846.045172] [<c00fcbcd>] (SyS_ioctl) from [<c000d641>]
> (ret_fast_syscall+0x1/0x44)
> 
> Do you have any idea how to solve the problem?
> 

I did this assertion there because I supposed that the rtnl lock is hold
while calling this callback. Now I see it isn't...

We need to hold the rtnl lock there because we accessing the mib values.

Try this:

diff --git a/net/mac802154/iface.c b/net/mac802154/iface.c
index 91b75ab..2a58788 100644
--- a/net/mac802154/iface.c
+++ b/net/mac802154/iface.c
@@ -62,8 +62,7 @@ mac802154_wpan_ioctl(struct net_device *dev, struct ifreq *ifr, int cmd)
                (struct sockaddr_ieee802154 *)&ifr->ifr_addr;
        int err = -ENOIOCTLCMD;
 
-       ASSERT_RTNL();
-
+       rtnl_lock();
        spin_lock_bh(&sdata->mib_lock);
 
        switch (cmd) {
@@ -90,6 +89,7 @@ mac802154_wpan_ioctl(struct net_device *dev, struct ifreq *ifr, int cmd)
        case SIOCSIFADDR:
                if (netif_running(dev)) {
                        spin_unlock_bh(&sdata->mib_lock);
+                       rtnl_unlock();
                        return -EBUSY;
                }
 
@@ -112,6 +112,7 @@ mac802154_wpan_ioctl(struct net_device *dev, struct ifreq *ifr, int cmd)
        }
 
        spin_unlock_bh(&sdata->mib_lock);
+       rtnl_unlock();
        return err;
 }


I will put it in my queue for next patch series to remove the mib/pib
spinlocks. We have only the rtnl lock now.

- Alex