On 27 June 2014 16:17, Kalle Valo [off-list ref] wrote:

Michal Kazior [off-list ref] writes:

quoted

Firmware doesn't perform Rx reordering so it is
left to the host driver to do that.

Use mac80211 to perform reordering instead of
re-inventing the wheel.

This fixes TCP throughput issues in some
environments (notably Windows stations connecting
to ath10k AP).

Reported-By: Denton Gentry <redacted>
Signed-off-by: Michal Kazior <redacted>

[...]

quoted

--- a/drivers/net/wireless/ath/ath10k/htt_rx.c
+++ b/drivers/net/wireless/ath/ath10k/htt_rx.c

@@ -21,6 +21,7 @@
 #include "txrx.h"
 #include "debug.h"
 #include "trace.h"
+#include "mac.h"

Why this? At least from a quick search I didn't see any reason to add this.

ath10k_get_arvif()

quoted

 #include <linux/log2.h>

@@ -1533,10 +1534,95 @@ void ath10k_htt_t2h_msg_handler(struct ath10k *ar, struct sk_buff *skb)
      case HTT_T2H_MSG_TYPE_STATS_CONF:
              trace_ath10k_htt_stats(skb->data, skb->len);
              break;
+     case HTT_T2H_MSG_TYPE_RX_ADDBA: {
+             struct ath10k *ar = htt->ar;
+             struct htt_rx_addba *ev = &resp->rx_addba;
+             struct ath10k_peer *peer;
+             struct ath10k_vif *arvif;
+             u16 info0, tid, peer_id;
+
+             info0 = __le32_to_cpu(ev->info0);
+             tid = MS(info0, HTT_RX_BA_INFO0_TID);
+             peer_id = MS(info0, HTT_RX_BA_INFO0_PEER_ID);

This event handling is pretty long. I think it would be better to have a
separate function for this event.

quoted

+     case HTT_T2H_MSG_TYPE_RX_DELBA: {
+             struct ath10k *ar = htt->ar;
+             struct htt_rx_addba *ev = &resp->rx_addba;
+             struct ath10k_peer *peer;
+             struct ath10k_vif *arvif;
+             u16 info0, tid, peer_id;

And the same for this event.

Good idea. I'll do that.


Michał