From: Toke Høiland-Jørgensen Date: Thu, 22 Sep 2016 17:04:20 +0000 (+0200) Subject: mac80211: Move reorder-sensitive TX handlers to after TXQ dequeue X-Git-Url: https://git.stricted.de/?a=commitdiff_plain;h=bb42f2d13ffcd0baed7547b37d05add51fcd50e1;p=GitHub%2Fmoto-9609%2Fandroid_kernel_motorola_exynos9610.git mac80211: Move reorder-sensitive TX handlers to after TXQ dequeue The TXQ intermediate queues can cause packet reordering when more than one flow is active to a single station. Since some of the wifi-specific packet handling (notably sequence number and encryption handling) is sensitive to re-ordering, things break if they are applied before the TXQ. This splits up the TX handlers and fast_xmit logic into two parts: An early part and a late part. The former is applied before TXQ enqueue, and the latter after dequeue. The non-TXQ path just applies both parts at once. Because fragments shouldn't be split up or reordered, the fragmentation handler is run after dequeue. Any fragments are then kept in the TXQ and on subsequent dequeues they take precedence over dequeueing from the FQ structure. This approach avoids having to scatter special cases all over the place for when TXQ is enabled, at the cost of making the fast_xmit and TX handler code slightly more complex. Signed-off-by: Toke Høiland-Jørgensen [fix a few code-style nits, make ieee80211_xmit_fast_finish void, remove a useless txq->sta check] Signed-off-by: Johannes Berg --- diff --git a/include/net/mac80211.h b/include/net/mac80211.h index c9f39538ac17..a810dfcb83c2 100644 --- a/include/net/mac80211.h +++ b/include/net/mac80211.h @@ -715,6 +715,7 @@ enum mac80211_tx_info_flags { * frame (PS-Poll or uAPSD). * @IEEE80211_TX_CTRL_RATE_INJECT: This frame is injected with rate information * @IEEE80211_TX_CTRL_AMSDU: This frame is an A-MSDU frame + * @IEEE80211_TX_CTRL_FAST_XMIT: This frame is going through the fast_xmit path * * These flags are used in tx_info->control.flags. */ @@ -723,6 +724,7 @@ enum mac80211_tx_control_flags { IEEE80211_TX_CTRL_PS_RESPONSE = BIT(1), IEEE80211_TX_CTRL_RATE_INJECT = BIT(2), IEEE80211_TX_CTRL_AMSDU = BIT(3), + IEEE80211_TX_CTRL_FAST_XMIT = BIT(4), }; /* diff --git a/net/mac80211/ieee80211_i.h b/net/mac80211/ieee80211_i.h index 2b391f242e58..8f8bddd5c8d8 100644 --- a/net/mac80211/ieee80211_i.h +++ b/net/mac80211/ieee80211_i.h @@ -815,12 +815,14 @@ enum txq_info_flags { * @def_flow: used as a fallback flow when a packet destined to @tin hashes to * a fq_flow which is already owned by a different tin * @def_cvars: codel vars for @def_flow + * @frags: used to keep fragments created after dequeue */ struct txq_info { struct fq_tin tin; struct fq_flow def_flow; struct codel_vars def_cvars; struct codel_stats cstats; + struct sk_buff_head frags; unsigned long flags; /* keep last! */ @@ -1498,6 +1500,13 @@ static inline struct txq_info *to_txq_info(struct ieee80211_txq *txq) return container_of(txq, struct txq_info, txq); } +static inline bool txq_has_queue(struct ieee80211_txq *txq) +{ + struct txq_info *txqi = to_txq_info(txq); + + return !(skb_queue_empty(&txqi->frags) && !txqi->tin.backlog_packets); +} + static inline int ieee80211_bssid_match(const u8 *raddr, const u8 *addr) { return ether_addr_equal(raddr, addr) || diff --git a/net/mac80211/rx.c b/net/mac80211/rx.c index c9489a86e6d6..b2fe725881dc 100644 --- a/net/mac80211/rx.c +++ b/net/mac80211/rx.c @@ -1323,9 +1323,7 @@ static void sta_ps_start(struct sta_info *sta) return; for (tid = 0; tid < ARRAY_SIZE(sta->sta.txq); tid++) { - struct txq_info *txqi = to_txq_info(sta->sta.txq[tid]); - - if (txqi->tin.backlog_packets) + if (txq_has_queue(sta->sta.txq[tid])) set_bit(tid, &sta->txq_buffered_tids); else clear_bit(tid, &sta->txq_buffered_tids); diff --git a/net/mac80211/sta_info.c b/net/mac80211/sta_info.c index 1b1b28ff4fdb..167bff078bdd 100644 --- a/net/mac80211/sta_info.c +++ b/net/mac80211/sta_info.c @@ -1212,12 +1212,10 @@ void ieee80211_sta_ps_deliver_wakeup(struct sta_info *sta) if (sta->sta.txq[0]) { for (i = 0; i < ARRAY_SIZE(sta->sta.txq); i++) { - struct txq_info *txqi = to_txq_info(sta->sta.txq[i]); - - if (!txqi->tin.backlog_packets) + if (!txq_has_queue(sta->sta.txq[i])) continue; - drv_wake_tx_queue(local, txqi); + drv_wake_tx_queue(local, to_txq_info(sta->sta.txq[i])); } } @@ -1649,9 +1647,7 @@ ieee80211_sta_ps_deliver_response(struct sta_info *sta, return; for (tid = 0; tid < ARRAY_SIZE(sta->sta.txq); tid++) { - struct txq_info *txqi = to_txq_info(sta->sta.txq[tid]); - - if (!(tids & BIT(tid)) || txqi->tin.backlog_packets) + if (!(tids & BIT(tid)) || txq_has_queue(sta->sta.txq[tid])) continue; sta_info_recalc_tim(sta); diff --git a/net/mac80211/tx.c b/net/mac80211/tx.c index 378a7a6b6dbe..0ea1b0d02186 100644 --- a/net/mac80211/tx.c +++ b/net/mac80211/tx.c @@ -853,8 +853,7 @@ ieee80211_tx_h_sequence(struct ieee80211_tx_data *tx) tid = *qc & IEEE80211_QOS_CTL_TID_MASK; tx->sta->tx_stats.msdu[tid]++; - if (!tx->sta->sta.txq[0]) - hdr->seq_ctrl = ieee80211_tx_next_seq(tx->sta, tid); + hdr->seq_ctrl = ieee80211_tx_next_seq(tx->sta, tid); return TX_CONTINUE; } @@ -1404,6 +1403,7 @@ void ieee80211_txq_init(struct ieee80211_sub_if_data *sdata, fq_flow_init(&txqi->def_flow); codel_vars_init(&txqi->def_cvars); codel_stats_init(&txqi->cstats); + __skb_queue_head_init(&txqi->frags); txqi->txq.vif = &sdata->vif; @@ -1426,6 +1426,7 @@ void ieee80211_txq_purge(struct ieee80211_local *local, struct fq_tin *tin = &txqi->tin; fq_tin_reset(fq, tin, fq_skb_free_func); + ieee80211_purge_tx_queue(&local->hw, &txqi->frags); } int ieee80211_txq_setup_flows(struct ieee80211_local *local) @@ -1495,6 +1496,47 @@ void ieee80211_txq_teardown_flows(struct ieee80211_local *local) spin_unlock_bh(&fq->lock); } +static bool ieee80211_queue_skb(struct ieee80211_local *local, + struct ieee80211_sub_if_data *sdata, + struct sta_info *sta, + struct sk_buff *skb) +{ + struct ieee80211_tx_info *info = IEEE80211_SKB_CB(skb); + struct fq *fq = &local->fq; + struct ieee80211_vif *vif; + struct txq_info *txqi; + struct ieee80211_sta *pubsta; + + if (!local->ops->wake_tx_queue || + sdata->vif.type == NL80211_IFTYPE_MONITOR) + return false; + + if (sta && sta->uploaded) + pubsta = &sta->sta; + else + pubsta = NULL; + + if (sdata->vif.type == NL80211_IFTYPE_AP_VLAN) + sdata = container_of(sdata->bss, + struct ieee80211_sub_if_data, u.ap); + + vif = &sdata->vif; + txqi = ieee80211_get_txq(local, vif, pubsta, skb); + + if (!txqi) + return false; + + info->control.vif = vif; + + spin_lock_bh(&fq->lock); + ieee80211_txq_enqueue(local, txqi, skb); + spin_unlock_bh(&fq->lock); + + drv_wake_tx_queue(local, txqi); + + return true; +} + static bool ieee80211_tx_frags(struct ieee80211_local *local, struct ieee80211_vif *vif, struct ieee80211_sta *sta, @@ -1502,9 +1544,7 @@ static bool ieee80211_tx_frags(struct ieee80211_local *local, bool txpending) { struct ieee80211_tx_control control = {}; - struct fq *fq = &local->fq; struct sk_buff *skb, *tmp; - struct txq_info *txqi; unsigned long flags; skb_queue_walk_safe(skbs, skb, tmp) { @@ -1519,21 +1559,6 @@ static bool ieee80211_tx_frags(struct ieee80211_local *local, } #endif - txqi = ieee80211_get_txq(local, vif, sta, skb); - if (txqi) { - info->control.vif = vif; - - __skb_unlink(skb, skbs); - - spin_lock_bh(&fq->lock); - ieee80211_txq_enqueue(local, txqi, skb); - spin_unlock_bh(&fq->lock); - - drv_wake_tx_queue(local, txqi); - - continue; - } - spin_lock_irqsave(&local->queue_stop_reason_lock, flags); if (local->queue_stop_reasons[q] || (!txpending && !skb_queue_empty(&local->pending[q]))) { @@ -1654,10 +1679,13 @@ static bool __ieee80211_tx(struct ieee80211_local *local, /* * Invoke TX handlers, return 0 on success and non-zero if the * frame was dropped or queued. + * + * The handlers are split into an early and late part. The latter is everything + * that can be sensitive to reordering, and will be deferred to after packets + * are dequeued from the intermediate queues (when they are enabled). */ -static int invoke_tx_handlers(struct ieee80211_tx_data *tx) +static int invoke_tx_handlers_early(struct ieee80211_tx_data *tx) { - struct ieee80211_tx_info *info = IEEE80211_SKB_CB(tx->skb); ieee80211_tx_result res = TX_DROP; #define CALL_TXH(txh) \ @@ -1675,6 +1703,31 @@ static int invoke_tx_handlers(struct ieee80211_tx_data *tx) if (!ieee80211_hw_check(&tx->local->hw, HAS_RATE_CONTROL)) CALL_TXH(ieee80211_tx_h_rate_ctrl); + txh_done: + if (unlikely(res == TX_DROP)) { + I802_DEBUG_INC(tx->local->tx_handlers_drop); + if (tx->skb) + ieee80211_free_txskb(&tx->local->hw, tx->skb); + else + ieee80211_purge_tx_queue(&tx->local->hw, &tx->skbs); + return -1; + } else if (unlikely(res == TX_QUEUED)) { + I802_DEBUG_INC(tx->local->tx_handlers_queued); + return -1; + } + + return 0; +} + +/* + * Late handlers can be called while the sta lock is held. Handlers that can + * cause packets to be generated will cause deadlock! + */ +static int invoke_tx_handlers_late(struct ieee80211_tx_data *tx) +{ + struct ieee80211_tx_info *info = IEEE80211_SKB_CB(tx->skb); + ieee80211_tx_result res = TX_CONTINUE; + if (unlikely(info->flags & IEEE80211_TX_INTFL_RETRANSMISSION)) { __skb_queue_tail(&tx->skbs, tx->skb); tx->skb = NULL; @@ -1707,6 +1760,15 @@ static int invoke_tx_handlers(struct ieee80211_tx_data *tx) return 0; } +static int invoke_tx_handlers(struct ieee80211_tx_data *tx) +{ + int r = invoke_tx_handlers_early(tx); + + if (r) + return r; + return invoke_tx_handlers_late(tx); +} + bool ieee80211_tx_prepare_skb(struct ieee80211_hw *hw, struct ieee80211_vif *vif, struct sk_buff *skb, int band, struct ieee80211_sta **sta) @@ -1781,7 +1843,13 @@ static bool ieee80211_tx(struct ieee80211_sub_if_data *sdata, info->hw_queue = sdata->vif.hw_queue[skb_get_queue_mapping(skb)]; - if (!invoke_tx_handlers(&tx)) + if (invoke_tx_handlers_early(&tx)) + return false; + + if (ieee80211_queue_skb(local, sdata, tx.sta, tx.skb)) + return true; + + if (!invoke_tx_handlers_late(&tx)) result = __ieee80211_tx(local, &tx.skbs, led_len, tx.sta, txpending); @@ -3125,8 +3193,71 @@ out: return ret; } +/* + * Can be called while the sta lock is held. Anything that can cause packets to + * be generated will cause deadlock! + */ +static void ieee80211_xmit_fast_finish(struct ieee80211_sub_if_data *sdata, + struct sta_info *sta, u8 pn_offs, + struct ieee80211_key *key, + struct sk_buff *skb) +{ + struct ieee80211_tx_info *info = IEEE80211_SKB_CB(skb); + struct ieee80211_hdr *hdr = (void *)skb->data; + u8 tid = IEEE80211_NUM_TIDS; + + if (key) + info->control.hw_key = &key->conf; + + ieee80211_tx_stats(skb->dev, skb->len); + + if (hdr->frame_control & cpu_to_le16(IEEE80211_STYPE_QOS_DATA)) { + tid = skb->priority & IEEE80211_QOS_CTL_TAG1D_MASK; + *ieee80211_get_qos_ctl(hdr) = tid; + hdr->seq_ctrl = ieee80211_tx_next_seq(sta, tid); + } else { + info->flags |= IEEE80211_TX_CTL_ASSIGN_SEQ; + hdr->seq_ctrl = cpu_to_le16(sdata->sequence_number); + sdata->sequence_number += 0x10; + } + + if (skb_shinfo(skb)->gso_size) + sta->tx_stats.msdu[tid] += + DIV_ROUND_UP(skb->len, skb_shinfo(skb)->gso_size); + else + sta->tx_stats.msdu[tid]++; + + info->hw_queue = sdata->vif.hw_queue[skb_get_queue_mapping(skb)]; + + /* statistics normally done by ieee80211_tx_h_stats (but that + * has to consider fragmentation, so is more complex) + */ + sta->tx_stats.bytes[skb_get_queue_mapping(skb)] += skb->len; + sta->tx_stats.packets[skb_get_queue_mapping(skb)]++; + + if (pn_offs) { + u64 pn; + u8 *crypto_hdr = skb->data + pn_offs; + + switch (key->conf.cipher) { + case WLAN_CIPHER_SUITE_CCMP: + case WLAN_CIPHER_SUITE_CCMP_256: + case WLAN_CIPHER_SUITE_GCMP: + case WLAN_CIPHER_SUITE_GCMP_256: + pn = atomic64_inc_return(&key->conf.tx_pn); + crypto_hdr[0] = pn; + crypto_hdr[1] = pn >> 8; + crypto_hdr[4] = pn >> 16; + crypto_hdr[5] = pn >> 24; + crypto_hdr[6] = pn >> 32; + crypto_hdr[7] = pn >> 40; + break; + } + } +} + static bool ieee80211_xmit_fast(struct ieee80211_sub_if_data *sdata, - struct net_device *dev, struct sta_info *sta, + struct sta_info *sta, struct ieee80211_fast_tx *fast_tx, struct sk_buff *skb) { @@ -3177,8 +3308,6 @@ static bool ieee80211_xmit_fast(struct ieee80211_sub_if_data *sdata, return true; } - ieee80211_tx_stats(dev, skb->len + extra_head); - if ((hdr->frame_control & cpu_to_le16(IEEE80211_STYPE_QOS_DATA)) && ieee80211_amsdu_aggregate(sdata, sta, fast_tx, skb)) return true; @@ -3207,24 +3336,7 @@ static bool ieee80211_xmit_fast(struct ieee80211_sub_if_data *sdata, info->flags = IEEE80211_TX_CTL_FIRST_FRAGMENT | IEEE80211_TX_CTL_DONTFRAG | (tid_tx ? IEEE80211_TX_CTL_AMPDU : 0); - - if (hdr->frame_control & cpu_to_le16(IEEE80211_STYPE_QOS_DATA)) { - *ieee80211_get_qos_ctl(hdr) = tid; - if (!sta->sta.txq[0]) - hdr->seq_ctrl = ieee80211_tx_next_seq(sta, tid); - } else { - info->flags |= IEEE80211_TX_CTL_ASSIGN_SEQ; - hdr->seq_ctrl = cpu_to_le16(sdata->sequence_number); - sdata->sequence_number += 0x10; - } - - if (skb_shinfo(skb)->gso_size) - sta->tx_stats.msdu[tid] += - DIV_ROUND_UP(skb->len, skb_shinfo(skb)->gso_size); - else - sta->tx_stats.msdu[tid]++; - - info->hw_queue = sdata->vif.hw_queue[skb_get_queue_mapping(skb)]; + info->control.flags = IEEE80211_TX_CTRL_FAST_XMIT; __skb_queue_head_init(&tx.skbs); @@ -3234,9 +3346,6 @@ static bool ieee80211_xmit_fast(struct ieee80211_sub_if_data *sdata, tx.sta = sta; tx.key = fast_tx->key; - if (fast_tx->key) - info->control.hw_key = &fast_tx->key->conf; - if (!ieee80211_hw_check(&local->hw, HAS_RATE_CONTROL)) { tx.skb = skb; r = ieee80211_tx_h_rate_ctrl(&tx); @@ -3250,31 +3359,11 @@ static bool ieee80211_xmit_fast(struct ieee80211_sub_if_data *sdata, } } - /* statistics normally done by ieee80211_tx_h_stats (but that - * has to consider fragmentation, so is more complex) - */ - sta->tx_stats.bytes[skb_get_queue_mapping(skb)] += skb->len; - sta->tx_stats.packets[skb_get_queue_mapping(skb)]++; + if (ieee80211_queue_skb(local, sdata, sta, skb)) + return true; - if (fast_tx->pn_offs) { - u64 pn; - u8 *crypto_hdr = skb->data + fast_tx->pn_offs; - - switch (fast_tx->key->conf.cipher) { - case WLAN_CIPHER_SUITE_CCMP: - case WLAN_CIPHER_SUITE_CCMP_256: - case WLAN_CIPHER_SUITE_GCMP: - case WLAN_CIPHER_SUITE_GCMP_256: - pn = atomic64_inc_return(&fast_tx->key->conf.tx_pn); - crypto_hdr[0] = pn; - crypto_hdr[1] = pn >> 8; - crypto_hdr[4] = pn >> 16; - crypto_hdr[5] = pn >> 24; - crypto_hdr[6] = pn >> 32; - crypto_hdr[7] = pn >> 40; - break; - } - } + ieee80211_xmit_fast_finish(sdata, sta, fast_tx->pn_offs, + fast_tx->key, skb); if (sdata->vif.type == NL80211_IFTYPE_AP_VLAN) sdata = container_of(sdata->bss, @@ -3294,12 +3383,21 @@ struct sk_buff *ieee80211_tx_dequeue(struct ieee80211_hw *hw, struct sk_buff *skb = NULL; struct fq *fq = &local->fq; struct fq_tin *tin = &txqi->tin; + struct ieee80211_tx_info *info; + struct ieee80211_tx_data tx; + ieee80211_tx_result r; spin_lock_bh(&fq->lock); if (test_bit(IEEE80211_TXQ_STOP, &txqi->flags)) goto out; + /* Make sure fragments stay together. */ + skb = __skb_dequeue(&txqi->frags); + if (skb) + goto out; + +begin: skb = fq_tin_dequeue(fq, tin, fq_tin_dequeue_func); if (!skb) goto out; @@ -3307,16 +3405,46 @@ struct sk_buff *ieee80211_tx_dequeue(struct ieee80211_hw *hw, ieee80211_set_skb_vif(skb, txqi); hdr = (struct ieee80211_hdr *)skb->data; - if (txq->sta && ieee80211_is_data_qos(hdr->frame_control)) { + info = IEEE80211_SKB_CB(skb); + + memset(&tx, 0, sizeof(tx)); + __skb_queue_head_init(&tx.skbs); + tx.local = local; + tx.skb = skb; + tx.sdata = vif_to_sdata(info->control.vif); + + if (txq->sta) + tx.sta = container_of(txq->sta, struct sta_info, sta); + + /* + * The key can be removed while the packet was queued, so need to call + * this here to get the current key. + */ + r = ieee80211_tx_h_select_key(&tx); + if (r != TX_CONTINUE) { + ieee80211_free_txskb(&local->hw, skb); + goto begin; + } + + if (info->control.flags & IEEE80211_TX_CTRL_FAST_XMIT) { struct sta_info *sta = container_of(txq->sta, struct sta_info, sta); - struct ieee80211_tx_info *info = IEEE80211_SKB_CB(skb); + u8 pn_offs = 0; - hdr->seq_ctrl = ieee80211_tx_next_seq(sta, txq->tid); - if (test_bit(IEEE80211_TXQ_AMPDU, &txqi->flags)) - info->flags |= IEEE80211_TX_CTL_AMPDU; - else - info->flags &= ~IEEE80211_TX_CTL_AMPDU; + if (tx.key && + (tx.key->conf.flags & IEEE80211_KEY_FLAG_GENERATE_IV)) + pn_offs = ieee80211_hdrlen(hdr->frame_control); + + ieee80211_xmit_fast_finish(sta->sdata, sta, pn_offs, + tx.key, skb); + } else { + if (invoke_tx_handlers_late(&tx)) + goto begin; + + skb = __skb_dequeue(&tx.skbs); + + if (!skb_queue_empty(&tx.skbs)) + skb_queue_splice_tail(&tx.skbs, &txqi->frags); } out: @@ -3354,7 +3482,7 @@ void __ieee80211_subif_start_xmit(struct sk_buff *skb, fast_tx = rcu_dereference(sta->fast_tx); if (fast_tx && - ieee80211_xmit_fast(sdata, dev, sta, fast_tx, skb)) + ieee80211_xmit_fast(sdata, sta, fast_tx, skb)) goto out; } diff --git a/net/mac80211/util.c b/net/mac80211/util.c index 91754c8dafb2..545c79a42a77 100644 --- a/net/mac80211/util.c +++ b/net/mac80211/util.c @@ -3441,11 +3441,18 @@ void ieee80211_txq_get_depth(struct ieee80211_txq *txq, unsigned long *byte_cnt) { struct txq_info *txqi = to_txq_info(txq); + u32 frag_cnt = 0, frag_bytes = 0; + struct sk_buff *skb; + + skb_queue_walk(&txqi->frags, skb) { + frag_cnt++; + frag_bytes += skb->len; + } if (frame_cnt) - *frame_cnt = txqi->tin.backlog_packets; + *frame_cnt = txqi->tin.backlog_packets + frag_cnt; if (byte_cnt) - *byte_cnt = txqi->tin.backlog_bytes; + *byte_cnt = txqi->tin.backlog_bytes + frag_bytes; } EXPORT_SYMBOL(ieee80211_txq_get_depth);