Attachment 667944 Details for Bug 863424 – queue flush fix

[patch] queue flush fix

0001-iwlwifi-trust-the-SSN-even-for-non-AGG-queues.patch (text/plain), 2.12 KB, created by Emmanuel Grumbach on 2012-12-23 09:48:26 UTC

(hide)

Description:

Filename:

MIME Type:

Creator: Emmanuel Grumbach

Created: 2012-12-23 09:48:26 UTC

Size: 2.12 KB

patch

obsolete

>From f7c0c123e3015349f71b7e371c92e6a59ab04507 Mon Sep 17 00:00:00 2001
>From: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
>Date: Sun, 23 Dec 2012 11:40:53 +0200
>Subject: [PATCH] iwlwifi: trust the SSN even for non AGG queues
>
>The SSN coming from the fw through the Tx response can be
>trusted even for non-AGG queues. The sequence control in
>the Tx response isn't reliable when the frames are flushed.
>
>Relying on the sequence control can cause issues when
>flushing the queues.
>
>Change-Id: Ida59d19d332642a5b75e21bb4b74b59989332985
>Cc: stable@vger.kernel.org
>Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
>---
> drivers/net/wireless/iwlwifi/dvm/tx.c |   23 ++---------------------
> 1 files changed, 2 insertions(+), 21 deletions(-)
>
>diff --git a/drivers/net/wireless/iwlwifi/dvm/tx.c b/drivers/net/wireless/iwlwifi/dvm/tx.c
>index d1f54cd..2b1e569 100644
>--- a/drivers/net/wireless/iwlwifi/dvm/tx.c
>+++ b/drivers/net/wireless/iwlwifi/dvm/tx.c
>@@ -1144,29 +1144,10 @@ int iwlagn_rx_reply_tx(struct iwl_priv *priv, struct iwl_rx_cmd_buffer *rxb,
> 	is_offchannel_skb = false;
> 
> 	if (tx_resp->frame_count == 1) {
>-		u16 next_reclaimed = le16_to_cpu(tx_resp->seq_ctl);
>-		next_reclaimed = SEQ_TO_SN(next_reclaimed + 0x10);
>-
>-		if (is_agg) {
>-			/* If this is an aggregation queue, we can rely on the
>-			 * ssn since the wifi sequence number corresponds to
>-			 * the index in the TFD ring (%256).
>-			 * The seq_ctl is the sequence control of the packet
>-			 * to which this Tx response relates. But if there is a
>-			 * hole in the bitmap of the BA we received, this Tx
>-			 * response may allow to reclaim the hole and all the
>-			 * subsequent packets that were already acked.
>-			 * In that case, seq_ctl != ssn, and the next packet
>-			 * to be reclaimed will be ssn and not seq_ctl.
>-			 */
>-			next_reclaimed = ssn;
>-		}
>-
> 		if (tid != IWL_TID_NON_QOS) {
>-			priv->tid_data[sta_id][tid].next_reclaimed =
>-				next_reclaimed;
>+			priv->tid_data[sta_id][tid].next_reclaimed = ssn;
> 			IWL_DEBUG_TX_REPLY(priv, "Next reclaimed packet:%d\n",
>-						  next_reclaimed);
>+					   ssn);
> 		}
> 
> 		iwl_trans_reclaim(priv->trans, txq_id, ssn, &skbs);
>-- 
>1.7.1
>

Actions: View | Diff

Attachments on bug 863424: 667944 | 670702 | 670947 | 671008 | 671345 | 671662 | 671997 | 672014 | 675302 | 675851 | 746018 | 750255 | 755318 | 764757 | 802340