linux-wireless.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
* [PATCH] ath10k: Fill the channel survey results for WCN3990 correctly
@ 2018-10-03 10:57 Rakesh Pillai
  2018-10-12 14:34 ` Kalle Valo
       [not found] ` <20181012143434.3F49060A98@smtp.codeaurora.org>
  0 siblings, 2 replies; 4+ messages in thread
From: Rakesh Pillai @ 2018-10-03 10:57 UTC (permalink / raw)
  To: ath10k; +Cc: linux-wireless, Rakesh Pillai, Surabhi Vishnoi

The host driver currently expects the channel info event to be
received in pairs for all the channels, i.e. the first
chan_info event for a particular channel will not have the
COMPLETE flag set and the second chan_info event for the
same channel will have the COMPLETE flag set.

The HL2.0 firmware sends only one channel info event per channel
which is scanned without the COMPLETE flag set. After sending the
chan_info_event for all the channels, the HL2.0 firmware sends a
chan_info_event with COMPLETE flag set to indicate the completion
of the channel info event.

The firmware does not indicate this behavior with any service bitmap
and hence a new firmware feature flag is used to handle the modified
parsing of the channel info events, in the host driver, for the
firmware which sends single channel info event per scanned channel.

Tested HW: WCN3990
Tested FW: WLAN.HL.2.0-01188-QCAHLSWMTPLZ-1

Co-developed-by: Surabhi Vishnoi <svishnoi@codeaurora.org>
Signed-off-by: Surabhi Vishnoi <svishnoi@codeaurora.org>
Signed-off-by: Rakesh Pillai <pillair@codeaurora.org>
---
 drivers/net/wireless/ath/ath10k/core.c    |   1 +
 drivers/net/wireless/ath/ath10k/core.h    |   3 +
 drivers/net/wireless/ath/ath10k/wmi-tlv.c |   3 +
 drivers/net/wireless/ath/ath10k/wmi-tlv.h |  10 +++
 drivers/net/wireless/ath/ath10k/wmi.c     | 130 +++++++++++++++++++++---------
 drivers/net/wireless/ath/ath10k/wmi.h     |  12 +++
 6 files changed, 122 insertions(+), 37 deletions(-)

diff --git a/drivers/net/wireless/ath/ath10k/core.c b/drivers/net/wireless/ath/ath10k/core.c
index 203f309..218b44f 100644
--- a/drivers/net/wireless/ath/ath10k/core.c
+++ b/drivers/net/wireless/ath/ath10k/core.c
@@ -594,6 +594,7 @@ static const char *const ath10k_core_fw_feature_str[] = {
 	[ATH10K_FW_FEATURE_NO_PS] = "no-ps",
 	[ATH10K_FW_FEATURE_MGMT_TX_BY_REF] = "mgmt-tx-by-reference",
 	[ATH10K_FW_FEATURE_NON_BMI] = "non-bmi",
+	[ATH10K_FW_FEATURE_SINGLE_CHAN_INFO_PER_CHANNEL] = "single-chan-info-per-channel",
 };
 
 static unsigned int ath10k_core_get_fw_feature_str(char *buf,
diff --git a/drivers/net/wireless/ath/ath10k/core.h b/drivers/net/wireless/ath/ath10k/core.h
index c76af34..3c34560 100644
--- a/drivers/net/wireless/ath/ath10k/core.h
+++ b/drivers/net/wireless/ath/ath10k/core.h
@@ -760,6 +760,9 @@ enum ath10k_fw_features {
 	/* Firmware load is done externally, not by bmi */
 	ATH10K_FW_FEATURE_NON_BMI = 19,
 
+	/* Firmware sends only one chan_info event per channel */
+	ATH10K_FW_FEATURE_SINGLE_CHAN_INFO_PER_CHANNEL = 20,
+
 	/* keep last */
 	ATH10K_FW_FEATURE_COUNT,
 };
diff --git a/drivers/net/wireless/ath/ath10k/wmi-tlv.c b/drivers/net/wireless/ath/ath10k/wmi-tlv.c
index df8f673..442184d 100644
--- a/drivers/net/wireless/ath/ath10k/wmi-tlv.c
+++ b/drivers/net/wireless/ath/ath10k/wmi-tlv.c
@@ -824,6 +824,9 @@ static int ath10k_wmi_tlv_op_pull_ch_info_ev(struct ath10k *ar,
 	arg->noise_floor = ev->noise_floor;
 	arg->rx_clear_count = ev->rx_clear_count;
 	arg->cycle_count = ev->cycle_count;
+	if (test_bit(ATH10K_FW_FEATURE_SINGLE_CHAN_INFO_PER_CHANNEL,
+		     ar->running_fw->fw_file.fw_features))
+		arg->mac_clk_mhz = ev->mac_clk_mhz;
 
 	kfree(tb);
 	return 0;
diff --git a/drivers/net/wireless/ath/ath10k/wmi-tlv.h b/drivers/net/wireless/ath/ath10k/wmi-tlv.h
index 8014fc8..b529955 100644
--- a/drivers/net/wireless/ath/ath10k/wmi-tlv.h
+++ b/drivers/net/wireless/ath/ath10k/wmi-tlv.h
@@ -1582,6 +1582,16 @@ struct ath10k_mgmt_tx_pkt_addr {
 
 #define WMI_TLV_FLAG_MGMT_BUNDLE_TX_COMPL	BIT(9)
 
+struct chan_info_params {
+	u32 err_code;
+	u32 freq;
+	u32 cmd_flags;
+	u32 noise_floor;
+	u32 rx_clear_count;
+	u32 cycle_count;
+	u32 mac_clk_mhz;
+};
+
 struct wmi_tlv_mgmt_tx_compl_ev {
 	__le32 desc_id;
 	__le32 status;
diff --git a/drivers/net/wireless/ath/ath10k/wmi.c b/drivers/net/wireless/ath/ath10k/wmi.c
index 8f62a85..37607fa 100644
--- a/drivers/net/wireless/ath/ath10k/wmi.c
+++ b/drivers/net/wireless/ath/ath10k/wmi.c
@@ -2580,60 +2580,71 @@ static int ath10k_wmi_10_4_op_pull_ch_info_ev(struct ath10k *ar,
 	return 0;
 }
 
-void ath10k_wmi_event_chan_info(struct ath10k *ar, struct sk_buff *skb)
+/*
+ * Handle the channel info event for firmware which only sends one
+ * chan_info event per scanned channel.
+ */
+static void ath10k_wmi_event_chan_info_unpaired(struct ath10k *ar,
+						struct chan_info_params *params)
 {
-	struct wmi_ch_info_ev_arg arg = {};
 	struct survey_info *survey;
-	u32 err_code, freq, cmd_flags, noise_floor, rx_clear_count, cycle_count;
-	int idx, ret;
+	u64 clock_freq;
+	int idx;
 
-	ret = ath10k_wmi_pull_ch_info(ar, skb, &arg);
-	if (ret) {
-		ath10k_warn(ar, "failed to parse chan info event: %d\n", ret);
+	if (params->cmd_flags & WMI_CHAN_INFO_FLAG_COMPLETE) {
+		ath10k_dbg(ar, ATH10K_DBG_WMI, "chan info report completed\n");
 		return;
 	}
 
-	err_code = __le32_to_cpu(arg.err_code);
-	freq = __le32_to_cpu(arg.freq);
-	cmd_flags = __le32_to_cpu(arg.cmd_flags);
-	noise_floor = __le32_to_cpu(arg.noise_floor);
-	rx_clear_count = __le32_to_cpu(arg.rx_clear_count);
-	cycle_count = __le32_to_cpu(arg.cycle_count);
+	idx = freq_to_idx(ar, params->freq);
+	if (idx >= ARRAY_SIZE(ar->survey)) {
+		ath10k_warn(ar, "chan info: invalid frequency %d (idx %d out of bounds)\n",
+			    params->freq, idx);
+		return;
+	}
 
-	ath10k_dbg(ar, ATH10K_DBG_WMI,
-		   "chan info err_code %d freq %d cmd_flags %d noise_floor %d rx_clear_count %d cycle_count %d\n",
-		   err_code, freq, cmd_flags, noise_floor, rx_clear_count,
-		   cycle_count);
+	survey = &ar->survey[idx];
 
-	spin_lock_bh(&ar->data_lock);
+	if (!params->mac_clk_mhz || !survey)
+		return;
 
-	switch (ar->scan.state) {
-	case ATH10K_SCAN_IDLE:
-	case ATH10K_SCAN_STARTING:
-		ath10k_warn(ar, "received chan info event without a scan request, ignoring\n");
-		goto exit;
-	case ATH10K_SCAN_RUNNING:
-	case ATH10K_SCAN_ABORTING:
-		break;
-	}
+	clock_freq = params->mac_clk_mhz * 1000;
+	memset(survey, 0, sizeof(*survey));
 
-	idx = freq_to_idx(ar, freq);
+	survey->noise = params->noise_floor;
+	survey->time = params->cycle_count / clock_freq;
+	survey->time_busy = params->rx_clear_count / clock_freq;
+	survey->filled |= SURVEY_INFO_NOISE_DBM | SURVEY_INFO_TIME |
+			  SURVEY_INFO_TIME_BUSY;
+}
+
+/*
+ * Handle the channel info event for firmware which sends chan_info
+ * event in pairs(start and stop events) for every scanned channel.
+ */
+static void ath10k_wmi_event_chan_info_paired(struct ath10k *ar,
+					      struct chan_info_params *params)
+{
+	struct survey_info *survey;
+	int idx;
+
+	idx = freq_to_idx(ar, params->freq);
 	if (idx >= ARRAY_SIZE(ar->survey)) {
 		ath10k_warn(ar, "chan info: invalid frequency %d (idx %d out of bounds)\n",
-			    freq, idx);
-		goto exit;
+			    params->freq, idx);
+		return;
 	}
 
-	if (cmd_flags & WMI_CHAN_INFO_FLAG_COMPLETE) {
+	if (params->cmd_flags & WMI_CHAN_INFO_FLAG_COMPLETE) {
 		if (ar->ch_info_can_report_survey) {
 			survey = &ar->survey[idx];
-			survey->noise = noise_floor;
+			survey->noise = params->noise_floor;
 			survey->filled = SURVEY_INFO_NOISE_DBM;
 
 			ath10k_hw_fill_survey_time(ar,
 						   survey,
-						   cycle_count,
-						   rx_clear_count,
+						   params->cycle_count,
+						   params->rx_clear_count,
 						   ar->survey_last_cycle_count,
 						   ar->survey_last_rx_clear_count);
 		}
@@ -2643,11 +2654,56 @@ void ath10k_wmi_event_chan_info(struct ath10k *ar, struct sk_buff *skb)
 		ar->ch_info_can_report_survey = true;
 	}
 
-	if (!(cmd_flags & WMI_CHAN_INFO_FLAG_PRE_COMPLETE)) {
-		ar->survey_last_rx_clear_count = rx_clear_count;
-		ar->survey_last_cycle_count = cycle_count;
+	if (!(params->cmd_flags & WMI_CHAN_INFO_FLAG_PRE_COMPLETE)) {
+		ar->survey_last_rx_clear_count = params->rx_clear_count;
+		ar->survey_last_cycle_count = params->cycle_count;
+	}
+}
+
+void ath10k_wmi_event_chan_info(struct ath10k *ar, struct sk_buff *skb)
+{
+	struct chan_info_params ch_info_param;
+	struct wmi_ch_info_ev_arg arg = {};
+	int ret;
+
+	ret = ath10k_wmi_pull_ch_info(ar, skb, &arg);
+	if (ret) {
+		ath10k_warn(ar, "failed to parse chan info event: %d\n", ret);
+		return;
+	}
+
+	ch_info_param.err_code = __le32_to_cpu(arg.err_code);
+	ch_info_param.freq = __le32_to_cpu(arg.freq);
+	ch_info_param.cmd_flags = __le32_to_cpu(arg.cmd_flags);
+	ch_info_param.noise_floor = __le32_to_cpu(arg.noise_floor);
+	ch_info_param.rx_clear_count = __le32_to_cpu(arg.rx_clear_count);
+	ch_info_param.cycle_count = __le32_to_cpu(arg.cycle_count);
+	ch_info_param.mac_clk_mhz = __le32_to_cpu(arg.mac_clk_mhz);
+
+	ath10k_dbg(ar, ATH10K_DBG_WMI,
+		   "chan info err_code %d freq %d cmd_flags %d noise_floor %d rx_clear_count %d cycle_count %d\n",
+		   ch_info_param.err_code, ch_info_param.freq, ch_info_param.cmd_flags,
+		   ch_info_param.noise_floor, ch_info_param.rx_clear_count,
+		   ch_info_param.cycle_count);
+
+	spin_lock_bh(&ar->data_lock);
+
+	switch (ar->scan.state) {
+	case ATH10K_SCAN_IDLE:
+	case ATH10K_SCAN_STARTING:
+		ath10k_warn(ar, "received chan info event without a scan request, ignoring\n");
+		goto exit;
+	case ATH10K_SCAN_RUNNING:
+	case ATH10K_SCAN_ABORTING:
+		break;
 	}
 
+	if (test_bit(ATH10K_FW_FEATURE_SINGLE_CHAN_INFO_PER_CHANNEL,
+		     ar->running_fw->fw_file.fw_features))
+		ath10k_wmi_event_chan_info_unpaired(ar, &ch_info_param);
+	else
+		ath10k_wmi_event_chan_info_paired(ar, &ch_info_param);
+
 exit:
 	spin_unlock_bh(&ar->data_lock);
 }
diff --git a/drivers/net/wireless/ath/ath10k/wmi.h b/drivers/net/wireless/ath/ath10k/wmi.h
index 1287687..e8e471e 100644
--- a/drivers/net/wireless/ath/ath10k/wmi.h
+++ b/drivers/net/wireless/ath/ath10k/wmi.h
@@ -6441,6 +6441,14 @@ struct wmi_chan_info_event {
 	__le32 noise_floor;
 	__le32 rx_clear_count;
 	__le32 cycle_count;
+	__le32 chan_tx_pwr_range;
+	__le32 chan_tx_pwr_tp;
+	__le32 rx_frame_count;
+	__le32 my_bss_rx_cycle_count;
+	__le32 rx_11b_mode_data_duration;
+	__le32 tx_frame_cnt;
+	__le32 mac_clk_mhz;
+
 } __packed;
 
 struct wmi_10_4_chan_info_event {
@@ -6675,6 +6683,10 @@ struct wmi_ch_info_ev_arg {
 	__le32 chan_tx_pwr_range;
 	__le32 chan_tx_pwr_tp;
 	__le32 rx_frame_count;
+	__le32 my_bss_rx_cycle_count;
+	__le32 rx_11b_mode_data_duration;
+	__le32 tx_frame_cnt;
+	__le32 mac_clk_mhz;
 };
 
 /* From 10.4 firmware, not sure all have the same values. */
-- 
2.7.4


^ permalink raw reply related	[flat|nested] 4+ messages in thread

* Re: [PATCH] ath10k: Fill the channel survey results for WCN3990 correctly
  2018-10-03 10:57 [PATCH] ath10k: Fill the channel survey results for WCN3990 correctly Rakesh Pillai
@ 2018-10-12 14:34 ` Kalle Valo
       [not found] ` <20181012143434.3F49060A98@smtp.codeaurora.org>
  1 sibling, 0 replies; 4+ messages in thread
From: Kalle Valo @ 2018-10-12 14:34 UTC (permalink / raw)
  To: Rakesh Pillai; +Cc: ath10k, linux-wireless, Rakesh Pillai, Surabhi Vishnoi

Rakesh Pillai <pillair@codeaurora.org> wrote:

> The host driver currently expects the channel info event to be
> received in pairs for all the channels, i.e. the first
> chan_info event for a particular channel will not have the
> COMPLETE flag set and the second chan_info event for the
> same channel will have the COMPLETE flag set.
> 
> The HL2.0 firmware sends only one channel info event per channel
> which is scanned without the COMPLETE flag set. After sending the
> chan_info_event for all the channels, the HL2.0 firmware sends a
> chan_info_event with COMPLETE flag set to indicate the completion
> of the channel info event.
> 
> The firmware does not indicate this behavior with any service bitmap
> and hence a new firmware feature flag is used to handle the modified
> parsing of the channel info events, in the host driver, for the
> firmware which sends single channel info event per scanned channel.
> 
> Tested HW: WCN3990
> Tested FW: WLAN.HL.2.0-01188-QCAHLSWMTPLZ-1
> 
> Co-developed-by: Surabhi Vishnoi <svishnoi@codeaurora.org>
> Signed-off-by: Surabhi Vishnoi <svishnoi@codeaurora.org>
> Signed-off-by: Rakesh Pillai <pillair@codeaurora.org>
> Signed-off-by: Kalle Valo <kvalo@codeaurora.org>

Kbuild bot reported some errors:

>> ERROR: "__udivdi3" [drivers/net/wireless/ath/ath10k/ath10k_core.ko] undefined!
>> ERROR: "__divdi3" [drivers/net/wireless/ath/ath10k/ath10k_core.ko] undefined!

Need to investigate those if they are valid or problem with the bot.

-- 
https://patchwork.kernel.org/patch/10624567/

https://wireless.wiki.kernel.org/en/developers/documentation/submittingpatches


^ permalink raw reply	[flat|nested] 4+ messages in thread

* Re: [PATCH] ath10k: Fill the channel survey results for WCN3990 correctly
       [not found] ` <20181012143434.3F49060A98@smtp.codeaurora.org>
@ 2018-10-12 15:53   ` Kalle Valo
  2018-10-17 11:21     ` Rakesh Pillai
  0 siblings, 1 reply; 4+ messages in thread
From: Kalle Valo @ 2018-10-12 15:53 UTC (permalink / raw)
  To: Rakesh Pillai; +Cc: linux-wireless, ath10k, Surabhi Vishnoi

Kalle Valo <kvalo@codeaurora.org> writes:

> Rakesh Pillai <pillair@codeaurora.org> wrote:
>
>> The host driver currently expects the channel info event to be
>> received in pairs for all the channels, i.e. the first
>> chan_info event for a particular channel will not have the
>> COMPLETE flag set and the second chan_info event for the
>> same channel will have the COMPLETE flag set.
>> 
>> The HL2.0 firmware sends only one channel info event per channel
>> which is scanned without the COMPLETE flag set. After sending the
>> chan_info_event for all the channels, the HL2.0 firmware sends a
>> chan_info_event with COMPLETE flag set to indicate the completion
>> of the channel info event.
>> 
>> The firmware does not indicate this behavior with any service bitmap
>> and hence a new firmware feature flag is used to handle the modified
>> parsing of the channel info events, in the host driver, for the
>> firmware which sends single channel info event per scanned channel.
>> 
>> Tested HW: WCN3990
>> Tested FW: WLAN.HL.2.0-01188-QCAHLSWMTPLZ-1
>> 
>> Co-developed-by: Surabhi Vishnoi <svishnoi@codeaurora.org>
>> Signed-off-by: Surabhi Vishnoi <svishnoi@codeaurora.org>
>> Signed-off-by: Rakesh Pillai <pillair@codeaurora.org>
>> Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
>
> Kbuild bot reported some errors:
>
>>> ERROR: "__udivdi3" [drivers/net/wireless/ath/ath10k/ath10k_core.ko] undefined!
>>> ERROR: "__divdi3" [drivers/net/wireless/ath/ath10k/ath10k_core.ko] undefined!
>
> Need to investigate those if they are valid or problem with the bot.

Arnd explained the problem in the other thread:

  "I think the problem is that 'clock_freq' is a 'u64', which means we
   can't use a normal 32-bit division. It looks like it cannot actually
   go beyond the range of a u32, so changing the type should be
   sufficient."

Please investigate that and submit v2.

-- 
Kalle Valo

^ permalink raw reply	[flat|nested] 4+ messages in thread

* Re: [PATCH] ath10k: Fill the channel survey results for WCN3990 correctly
  2018-10-12 15:53   ` Kalle Valo
@ 2018-10-17 11:21     ` Rakesh Pillai
  0 siblings, 0 replies; 4+ messages in thread
From: Rakesh Pillai @ 2018-10-17 11:21 UTC (permalink / raw)
  To: Kalle Valo; +Cc: linux-wireless, ath10k, Surabhi Vishnoi

Hi Kalle,

I have removed the clk_freq variable and now there will be u32 division 
only.
This change has been sent in v2.

Thanks,
Rakesh Pillai.

On 2018-10-12 21:23, Kalle Valo wrote:
> Kalle Valo <kvalo@codeaurora.org> writes:
> 
>> Rakesh Pillai <pillair@codeaurora.org> wrote:
>> 
>>> The host driver currently expects the channel info event to be
>>> received in pairs for all the channels, i.e. the first
>>> chan_info event for a particular channel will not have the
>>> COMPLETE flag set and the second chan_info event for the
>>> same channel will have the COMPLETE flag set.
>>> 
>>> The HL2.0 firmware sends only one channel info event per channel
>>> which is scanned without the COMPLETE flag set. After sending the
>>> chan_info_event for all the channels, the HL2.0 firmware sends a
>>> chan_info_event with COMPLETE flag set to indicate the completion
>>> of the channel info event.
>>> 
>>> The firmware does not indicate this behavior with any service bitmap
>>> and hence a new firmware feature flag is used to handle the modified
>>> parsing of the channel info events, in the host driver, for the
>>> firmware which sends single channel info event per scanned channel.
>>> 
>>> Tested HW: WCN3990
>>> Tested FW: WLAN.HL.2.0-01188-QCAHLSWMTPLZ-1
>>> 
>>> Co-developed-by: Surabhi Vishnoi <svishnoi@codeaurora.org>
>>> Signed-off-by: Surabhi Vishnoi <svishnoi@codeaurora.org>
>>> Signed-off-by: Rakesh Pillai <pillair@codeaurora.org>
>>> Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
>> 
>> Kbuild bot reported some errors:
>> 
>>>> ERROR: "__udivdi3" [drivers/net/wireless/ath/ath10k/ath10k_core.ko] 
>>>> undefined!
>>>> ERROR: "__divdi3" [drivers/net/wireless/ath/ath10k/ath10k_core.ko] 
>>>> undefined!
>> 
>> Need to investigate those if they are valid or problem with the bot.
> 
> Arnd explained the problem in the other thread:
> 
>   "I think the problem is that 'clock_freq' is a 'u64', which means we
>    can't use a normal 32-bit division. It looks like it cannot actually
>    go beyond the range of a u32, so changing the type should be
>    sufficient."
> 
> Please investigate that and submit v2.

^ permalink raw reply	[flat|nested] 4+ messages in thread

end of thread, other threads:[~2018-10-17 11:21 UTC | newest]

Thread overview: 4+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2018-10-03 10:57 [PATCH] ath10k: Fill the channel survey results for WCN3990 correctly Rakesh Pillai
2018-10-12 14:34 ` Kalle Valo
     [not found] ` <20181012143434.3F49060A98@smtp.codeaurora.org>
2018-10-12 15:53   ` Kalle Valo
2018-10-17 11:21     ` Rakesh Pillai

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).