Subject: Re: [PATCH] iwlwifi: do not nulify ctx->vif on reset
From: "Guy, Wey-Yi" <wey-yi.w.guy@intel.com>
To: Johannes Berg <johannes@sipsolutions.net>
Cc: Stanislaw Gruszka <sgruszka@redhat.com>,
	Intel Linux Wireless <ilw@linux.intel.com>,
	linux-wireless@vger.kernel.org
In-Reply-To: <1332752906.7081.10.camel@jlt3.sipsolutions.net>
References: <1331651434-4370-1-git-send-email-sgruszka@redhat.com>
	 <1331651730.3329.6.camel@jlt3.sipsolutions.net>
	 <20120314062549.GA2788@redhat.com>
	 <1331708304.3376.0.camel@jlt3.sipsolutions.net>
	 <20120314070748.GB2788@redhat.com> <20120314081424.GC2788@redhat.com>
	 <1332752906.7081.10.camel@jlt3.sipsolutions.net>
Content-Type: text/plain; charset="UTF-8"
Date: Mon, 26 Mar 2012 07:30:32 -0700
Message-ID: <1332772232.27767.14.camel@wwguy-huron> (sfid-20120326_163153_937190_0CA3C239)
Mime-Version: 1.0
Sender: linux-wireless-owner@vger.kernel.org

On Mon, 2012-03-26 at 11:08 +0200, Johannes Berg wrote:
> On Wed, 2012-03-14 at 09:14 +0100, Stanislaw Gruszka wrote:
> > On Wed, Mar 14, 2012 at 08:07:48AM +0100, Stanislaw Gruszka wrote:
> > > I'll update and recheck again.
> > 
> > On updated tree (head 035364916f75151b4b91ea53968c6beba7545317) devices
> > stop working here on any forced reset during TX as well. There are
> > messages like below in dmesg:
> > 
> > wlan3: dropped data frame to not associated station 00:00:00:00:00:00
> > 
> > And one time crash happened too, in:
> > 
> > (gdb) l *(iwl_remove_dynamic_key+0x1f0)
> > 0x15e20 is in iwl_remove_dynamic_key (drivers/net/wireless/iwlwifi/iwl-agn-sta.c:1105).
> > 1100		/*
> > 1101		 * The device expects GTKs for station interfaces to be
> > 1102		 * installed as GTKs for the AP station. If we have no
> > 1103		 * station ID, then use the ap_sta_id in that case.
> > 1104		 */
> > 1105		if (vif->type == NL80211_IFTYPE_STATION && vif_priv->ctx)
> > 1106			return vif_priv->ctx->ap_sta_id;
> > 1107	
> > 1108		return IWL_INVALID_STATION;
> > 1109	}
> > 
> > To reproduce problems, I'm doing "ping -f 192.168.1.1" on one console
> > and run script [1] on other console.
> 
> Ok so I finally got some time to look into this in some detail. It seems
> that the driver assumes a few calls from mac80211 like key settings can
> only happen while the vif pointers are valid, which is correct under
> normal circumstances but due to races with HW reset it's not really true
> there.
> 
> We could fix all of those, but it seems your patch is the better
> approach. Wey, can you pick it up?

yes

Wey
>