From: Michal Kazior <michal.kazior@tieto.com> To: ath10k@lists.infradead.org Cc: linux-wireless@vger.kernel.org, Michal Kazior <michal.kazior@tieto.com> Subject: [PATCH 1/3] ath10k: fix offchan reliability Date: Mon, 24 Nov 2014 14:58:31 +0100 [thread overview] Message-ID: <1416837513-13386-1-git-send-email-michal.kazior@tieto.com> (raw) New firmware revisions don't need peer creation when doing offchannel tx. Earlier revisions would queue and never release frames without a peer. This prevent new firmware revisions from stopping replenishing wmi-htc tx credits and improves reliability of offchannel tx which would sometimes silently fail. Signed-off-by: Michal Kazior <michal.kazior@tieto.com> --- drivers/net/wireless/ath/ath10k/core.h | 1 + drivers/net/wireless/ath/ath10k/htt.h | 4 +++- drivers/net/wireless/ath/ath10k/htt_tx.c | 7 ++++--- drivers/net/wireless/ath/ath10k/mac.c | 29 +++++++++++++++++++++++------ 4 files changed, 31 insertions(+), 10 deletions(-) diff --git a/drivers/net/wireless/ath/ath10k/core.h b/drivers/net/wireless/ath/ath10k/core.h index 8f86bd3..c510785 100644 --- a/drivers/net/wireless/ath/ath10k/core.h +++ b/drivers/net/wireless/ath/ath10k/core.h @@ -83,6 +83,7 @@ struct ath10k_skb_cb { struct { u8 tid; + u16 freq; bool is_offchan; struct ath10k_htt_txbuf *txbuf; u32 txbuf_paddr; diff --git a/drivers/net/wireless/ath/ath10k/htt.h b/drivers/net/wireless/ath/ath10k/htt.h index 15c58e8..17ca6ac 100644 --- a/drivers/net/wireless/ath/ath10k/htt.h +++ b/drivers/net/wireless/ath/ath10k/htt.h @@ -126,6 +126,7 @@ enum htt_data_tx_ext_tid { * (HL hosts manage queues on the host ) * more_in_batch: only for HL hosts. indicates if more packets are * pending. this allows target to wait and aggregate + * freq: 0 means home channel of given vdev. intended for offchannel */ struct htt_data_tx_desc { u8 flags0; /* %HTT_DATA_TX_DESC_FLAGS0_ */ @@ -133,7 +134,8 @@ struct htt_data_tx_desc { __le16 len; __le16 id; __le32 frags_paddr; - __le32 peerid; + __le16 peerid; + __le16 freq; u8 prefetch[0]; /* start of frame, for FW classification engine */ } __packed; diff --git a/drivers/net/wireless/ath/ath10k/htt_tx.c b/drivers/net/wireless/ath/ath10k/htt_tx.c index 5b7e42f..4bc51d8 100644 --- a/drivers/net/wireless/ath/ath10k/htt_tx.c +++ b/drivers/net/wireless/ath/ath10k/htt_tx.c @@ -554,13 +554,14 @@ int ath10k_htt_tx(struct ath10k_htt *htt, struct sk_buff *msdu) skb_cb->htt.txbuf->cmd_tx.len = __cpu_to_le16(msdu->len); skb_cb->htt.txbuf->cmd_tx.id = __cpu_to_le16(msdu_id); skb_cb->htt.txbuf->cmd_tx.frags_paddr = __cpu_to_le32(frags_paddr); - skb_cb->htt.txbuf->cmd_tx.peerid = __cpu_to_le32(HTT_INVALID_PEERID); + skb_cb->htt.txbuf->cmd_tx.peerid = __cpu_to_le16(HTT_INVALID_PEERID); + skb_cb->htt.txbuf->cmd_tx.freq = __cpu_to_le16(skb_cb->htt.freq); trace_ath10k_htt_tx(ar, msdu_id, msdu->len, vdev_id, tid); ath10k_dbg(ar, ATH10K_DBG_HTT, - "htt tx flags0 %hhu flags1 %hu len %d id %hu frags_paddr %08x, msdu_paddr %08x vdev %hhu tid %hhu\n", + "htt tx flags0 %hhu flags1 %hu len %d id %hu frags_paddr %08x, msdu_paddr %08x vdev %hhu tid %hhu freq %hu\n", flags0, flags1, msdu->len, msdu_id, frags_paddr, - (u32)skb_cb->paddr, vdev_id, tid); + (u32)skb_cb->paddr, vdev_id, tid, skb_cb->htt.freq); ath10k_dbg_dump(ar, ATH10K_DBG_HTT_DUMP, NULL, "htt tx msdu: ", msdu->data, msdu->len); trace_ath10k_tx_hdr(ar, msdu->data, msdu->len); diff --git a/drivers/net/wireless/ath/ath10k/mac.c b/drivers/net/wireless/ath/ath10k/mac.c index 1245ac8..8deae4d 100644 --- a/drivers/net/wireless/ath/ath10k/mac.c +++ b/drivers/net/wireless/ath/ath10k/mac.c @@ -1997,6 +1997,18 @@ static void ath10k_tx_h_add_p2p_noa_ie(struct ath10k *ar, } } +static bool ath10k_mac_need_offchan_tx_work(struct ath10k *ar) +{ + /* FIXME: Not really sure since when the behaviour changed. At some + * point new firmware stopped requiring creation of peer entries for + * offchannel tx (and actually creating them causes issues with wmi-htc + * tx credit replenishment and reliability). Assuming it's at least 3.4 + * because that's when the `freq` was introduced to TX_FRM HTT command. + */ + return !(ar->htt.target_version_major >= 3 && + ar->htt.target_version_minor >= 4); +} + static void ath10k_tx_htt(struct ath10k *ar, struct sk_buff *skb) { struct ieee80211_hdr *hdr = (struct ieee80211_hdr *)skb->data; @@ -2341,16 +2353,21 @@ static void ath10k_tx(struct ieee80211_hw *hw, if (info->flags & IEEE80211_TX_CTL_TX_OFFCHAN) { spin_lock_bh(&ar->data_lock); - ATH10K_SKB_CB(skb)->htt.is_offchan = true; + ATH10K_SKB_CB(skb)->htt.freq = ar->scan.roc_freq; ATH10K_SKB_CB(skb)->vdev_id = ar->scan.vdev_id; spin_unlock_bh(&ar->data_lock); - ath10k_dbg(ar, ATH10K_DBG_MAC, "queued offchannel skb %p\n", - skb); + if (ath10k_mac_need_offchan_tx_work(ar)) { + ATH10K_SKB_CB(skb)->htt.freq = 0; + ATH10K_SKB_CB(skb)->htt.is_offchan = true; - skb_queue_tail(&ar->offchan_tx_queue, skb); - ieee80211_queue_work(hw, &ar->offchan_tx_work); - return; + ath10k_dbg(ar, ATH10K_DBG_MAC, "queued offchannel skb %p\n", + skb); + + skb_queue_tail(&ar->offchan_tx_queue, skb); + ieee80211_queue_work(hw, &ar->offchan_tx_work); + return; + } } ath10k_tx_htt(ar, skb); -- 1.8.5.3
WARNING: multiple messages have this Message-ID (diff)
From: Michal Kazior <michal.kazior@tieto.com> To: ath10k@lists.infradead.org Cc: linux-wireless@vger.kernel.org, Michal Kazior <michal.kazior@tieto.com> Subject: [PATCH 1/3] ath10k: fix offchan reliability Date: Mon, 24 Nov 2014 14:58:31 +0100 [thread overview] Message-ID: <1416837513-13386-1-git-send-email-michal.kazior@tieto.com> (raw) New firmware revisions don't need peer creation when doing offchannel tx. Earlier revisions would queue and never release frames without a peer. This prevent new firmware revisions from stopping replenishing wmi-htc tx credits and improves reliability of offchannel tx which would sometimes silently fail. Signed-off-by: Michal Kazior <michal.kazior@tieto.com> --- drivers/net/wireless/ath/ath10k/core.h | 1 + drivers/net/wireless/ath/ath10k/htt.h | 4 +++- drivers/net/wireless/ath/ath10k/htt_tx.c | 7 ++++--- drivers/net/wireless/ath/ath10k/mac.c | 29 +++++++++++++++++++++++------ 4 files changed, 31 insertions(+), 10 deletions(-) diff --git a/drivers/net/wireless/ath/ath10k/core.h b/drivers/net/wireless/ath/ath10k/core.h index 8f86bd3..c510785 100644 --- a/drivers/net/wireless/ath/ath10k/core.h +++ b/drivers/net/wireless/ath/ath10k/core.h @@ -83,6 +83,7 @@ struct ath10k_skb_cb { struct { u8 tid; + u16 freq; bool is_offchan; struct ath10k_htt_txbuf *txbuf; u32 txbuf_paddr; diff --git a/drivers/net/wireless/ath/ath10k/htt.h b/drivers/net/wireless/ath/ath10k/htt.h index 15c58e8..17ca6ac 100644 --- a/drivers/net/wireless/ath/ath10k/htt.h +++ b/drivers/net/wireless/ath/ath10k/htt.h @@ -126,6 +126,7 @@ enum htt_data_tx_ext_tid { * (HL hosts manage queues on the host ) * more_in_batch: only for HL hosts. indicates if more packets are * pending. this allows target to wait and aggregate + * freq: 0 means home channel of given vdev. intended for offchannel */ struct htt_data_tx_desc { u8 flags0; /* %HTT_DATA_TX_DESC_FLAGS0_ */ @@ -133,7 +134,8 @@ struct htt_data_tx_desc { __le16 len; __le16 id; __le32 frags_paddr; - __le32 peerid; + __le16 peerid; + __le16 freq; u8 prefetch[0]; /* start of frame, for FW classification engine */ } __packed; diff --git a/drivers/net/wireless/ath/ath10k/htt_tx.c b/drivers/net/wireless/ath/ath10k/htt_tx.c index 5b7e42f..4bc51d8 100644 --- a/drivers/net/wireless/ath/ath10k/htt_tx.c +++ b/drivers/net/wireless/ath/ath10k/htt_tx.c @@ -554,13 +554,14 @@ int ath10k_htt_tx(struct ath10k_htt *htt, struct sk_buff *msdu) skb_cb->htt.txbuf->cmd_tx.len = __cpu_to_le16(msdu->len); skb_cb->htt.txbuf->cmd_tx.id = __cpu_to_le16(msdu_id); skb_cb->htt.txbuf->cmd_tx.frags_paddr = __cpu_to_le32(frags_paddr); - skb_cb->htt.txbuf->cmd_tx.peerid = __cpu_to_le32(HTT_INVALID_PEERID); + skb_cb->htt.txbuf->cmd_tx.peerid = __cpu_to_le16(HTT_INVALID_PEERID); + skb_cb->htt.txbuf->cmd_tx.freq = __cpu_to_le16(skb_cb->htt.freq); trace_ath10k_htt_tx(ar, msdu_id, msdu->len, vdev_id, tid); ath10k_dbg(ar, ATH10K_DBG_HTT, - "htt tx flags0 %hhu flags1 %hu len %d id %hu frags_paddr %08x, msdu_paddr %08x vdev %hhu tid %hhu\n", + "htt tx flags0 %hhu flags1 %hu len %d id %hu frags_paddr %08x, msdu_paddr %08x vdev %hhu tid %hhu freq %hu\n", flags0, flags1, msdu->len, msdu_id, frags_paddr, - (u32)skb_cb->paddr, vdev_id, tid); + (u32)skb_cb->paddr, vdev_id, tid, skb_cb->htt.freq); ath10k_dbg_dump(ar, ATH10K_DBG_HTT_DUMP, NULL, "htt tx msdu: ", msdu->data, msdu->len); trace_ath10k_tx_hdr(ar, msdu->data, msdu->len); diff --git a/drivers/net/wireless/ath/ath10k/mac.c b/drivers/net/wireless/ath/ath10k/mac.c index 1245ac8..8deae4d 100644 --- a/drivers/net/wireless/ath/ath10k/mac.c +++ b/drivers/net/wireless/ath/ath10k/mac.c @@ -1997,6 +1997,18 @@ static void ath10k_tx_h_add_p2p_noa_ie(struct ath10k *ar, } } +static bool ath10k_mac_need_offchan_tx_work(struct ath10k *ar) +{ + /* FIXME: Not really sure since when the behaviour changed. At some + * point new firmware stopped requiring creation of peer entries for + * offchannel tx (and actually creating them causes issues with wmi-htc + * tx credit replenishment and reliability). Assuming it's at least 3.4 + * because that's when the `freq` was introduced to TX_FRM HTT command. + */ + return !(ar->htt.target_version_major >= 3 && + ar->htt.target_version_minor >= 4); +} + static void ath10k_tx_htt(struct ath10k *ar, struct sk_buff *skb) { struct ieee80211_hdr *hdr = (struct ieee80211_hdr *)skb->data; @@ -2341,16 +2353,21 @@ static void ath10k_tx(struct ieee80211_hw *hw, if (info->flags & IEEE80211_TX_CTL_TX_OFFCHAN) { spin_lock_bh(&ar->data_lock); - ATH10K_SKB_CB(skb)->htt.is_offchan = true; + ATH10K_SKB_CB(skb)->htt.freq = ar->scan.roc_freq; ATH10K_SKB_CB(skb)->vdev_id = ar->scan.vdev_id; spin_unlock_bh(&ar->data_lock); - ath10k_dbg(ar, ATH10K_DBG_MAC, "queued offchannel skb %p\n", - skb); + if (ath10k_mac_need_offchan_tx_work(ar)) { + ATH10K_SKB_CB(skb)->htt.freq = 0; + ATH10K_SKB_CB(skb)->htt.is_offchan = true; - skb_queue_tail(&ar->offchan_tx_queue, skb); - ieee80211_queue_work(hw, &ar->offchan_tx_work); - return; + ath10k_dbg(ar, ATH10K_DBG_MAC, "queued offchannel skb %p\n", + skb); + + skb_queue_tail(&ar->offchan_tx_queue, skb); + ieee80211_queue_work(hw, &ar->offchan_tx_work); + return; + } } ath10k_tx_htt(ar, skb); -- 1.8.5.3 _______________________________________________ ath10k mailing list ath10k@lists.infradead.org http://lists.infradead.org/mailman/listinfo/ath10k
next reply other threads:[~2014-11-24 14:11 UTC|newest] Thread overview: 8+ messages / expand[flat|nested] mbox.gz Atom feed top 2014-11-24 13:58 Michal Kazior [this message] 2014-11-24 13:58 ` [PATCH 1/3] ath10k: fix offchan reliability Michal Kazior 2014-11-24 13:58 ` [PATCH 2/3] ath10k: make hw roc more reliable Michal Kazior 2014-11-24 13:58 ` Michal Kazior 2014-11-24 13:58 ` [PATCH 3/3] ath10k: fix offchannel cancel failures Michal Kazior 2014-11-24 13:58 ` Michal Kazior 2014-11-26 6:32 ` [PATCH 1/3] ath10k: fix offchan reliability Kalle Valo 2014-11-26 6:32 ` Kalle Valo
Reply instructions: You may reply publicly to this message via plain-text email using any one of the following methods: * Save the following mbox file, import it into your mail client, and reply-to-all from there: mbox Avoid top-posting and favor interleaved quoting: https://en.wikipedia.org/wiki/Posting_style#Interleaved_style * Reply using the --to, --cc, and --in-reply-to switches of git-send-email(1): git send-email \ --in-reply-to=1416837513-13386-1-git-send-email-michal.kazior@tieto.com \ --to=michal.kazior@tieto.com \ --cc=ath10k@lists.infradead.org \ --cc=linux-wireless@vger.kernel.org \ /path/to/YOUR_REPLY https://kernel.org/pub/software/scm/git/docs/git-send-email.html * If your mail client supports setting the In-Reply-To header via mailto: links, try the mailto: linkBe sure your reply has a Subject: header at the top and a blank line before the message body.
This is an external index of several public inboxes, see mirroring instructions on how to clone and mirror all data and code used by this external index.