| Message ID | 20191115075942.120943-1-ming.chen@watchguard.com (mailing list archive) |
|---|---|
| State | Superseded |
| Delegated to: | Johannes Berg |
| Headers | show |
| Series | [v3] mac80211: Drop the packets whose source or destination mac address is empty | expand |
Ming Chen <ming032217@gmail.com> writes: > We occasionally found ath9k could receive some packets from Linux IP stack > with empty source and destination mac address, How does that happen? > which will result in the driver cannot find the station node in TX > complete. And thus, the driver will complete this buffer but without > updating the block ack window. If it can't find the station, how is the packet transmitted (and affecting the BA window) in the first place? -Toke
Hello Toke, Thanks for the review. Please see my inline comments. Ming > -----Original Message----- > From: Toke Høiland-Jørgensen <toke@redhat.com> > Sent: Friday, November 15, 2019 3:08 AM > To: Ming Chen <ming032217@gmail.com>; Johannes Berg > <johannes@sipsolutions.net> > Cc: linux-wireless@vger.kernel.org; Ming Chen > <Ming.Chen@watchguard.com> > Subject: Re: [PATCH v3] mac80211: Drop the packets whose source or > destination mac address is empty > > Ming Chen <ming032217@gmail.com> writes: > > > We occasionally found ath9k could receive some packets from Linux IP > > stack with empty source and destination mac address, > > How does that happen? [Ming Chen] To be honest, I not quite sure how it happen. When the radio works as a wireless client, I cloud occasionally reproduce this issue by disabling DHCP and configure a static IP for wireless client, and also try to ping to its gateway without break. My kernel version is 4.14. > > > which will result in the driver cannot find the station node in TX > > complete. And thus, the driver will complete this buffer but without > > updating the block ack window. > > If it can't find the station, how is the packet transmitted (and affecting the BA > window) in the first place? [Ming Chen] Please see the new changelog of version 4. > > -Toke
diff --git a/net/mac80211/tx.c b/net/mac80211/tx.c index db38be1b75fa..b18745a3f6b0 100644 --- a/net/mac80211/tx.c +++ b/net/mac80211/tx.c @@ -2489,6 +2489,13 @@ static struct sk_buff *ieee80211_build_hdr(struct ieee80211_sub_if_data *sdata, if (IS_ERR(sta)) sta = NULL; + /* drop this skb when source mac or destination mac is empty */ + if (is_zero_ether_addr(skb->data) || + is_zero_ether_addr(skb->data + ETH_ALEN)) { + ret = -ENOTCONN; + goto free; + } + #ifdef CONFIG_MAC80211_DEBUGFS if (local->force_tx_status) info_flags |= IEEE80211_TX_CTL_REQ_TX_STATUS; @@ -3435,6 +3442,11 @@ static bool ieee80211_xmit_fast(struct ieee80211_sub_if_data *sdata, if (skb->sk && skb_shinfo(skb)->tx_flags & SKBTX_WIFI_STATUS) return false; + /* drop this skb when source mac or destination mac is empty */ + if (is_zero_ether_addr(skb->data) || + is_zero_ether_addr(skb->data + ETH_ALEN)) + return false; + if (hdr->frame_control & cpu_to_le16(IEEE80211_STYPE_QOS_DATA)) { tid = skb->priority & IEEE80211_QOS_CTL_TAG1D_MASK; tid_tx = rcu_dereference(sta->ampdu_mlme.tid_tx[tid]);
We occasionally found ath9k could receive some packets from Linux IP stack with empty source and destination mac address,which will result in the driver cannot find the station node in TX complete. And thus, the driver will complete this buffer but without updating the block ack window. To fix this issue, we should drop this kind of error packet before it goes into the driver. Signed-off-by: Ming Chen <ming.chen@watchguard.com> --- v3: -Fix s-o-b location v2: -According to review feedback, use the is_zero_ether_addr to check if the mac address is empty. --- net/mac80211/tx.c | 12 ++++++++++++ 1 file changed, 12 insertions(+)