Received-SPF: pass (google.com: best guess record for domain of linux-bluetooth-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) client-ip=209.132.180.67;
Date:   Thu, 2 Jan 2020 11:07:27 -0800
From:   Matthias Kaehlcke <mka@chromium.org>
To:     Rocky Liao <rjliao@codeaurora.org>
Cc:     marcel@holtmann.org, johan.hedberg@gmail.com,
        linux-kernel@vger.kernel.org, linux-bluetooth@vger.kernel.org,
        linux-arm-msm@vger.kernel.org
Subject: Re: [PATCH v3 4/4] Bluetooth: hci_qca: Add HCI command timeout
 handling
Message-ID: <20200102190727.GB89495@google.com>
References: <20191225060317.5258-1-rjliao@codeaurora.org>
 <20191227072130.29431-1-rjliao@codeaurora.org>
 <20191227072130.29431-4-rjliao@codeaurora.org>
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Disposition: inline
In-Reply-To: <20191227072130.29431-4-rjliao@codeaurora.org>
User-Agent: Mutt/1.10.1 (2018-07-13)
Sender: linux-bluetooth-owner@vger.kernel.org
Precedence: bulk

Hi Rocky,

On Fri, Dec 27, 2019 at 03:21:30PM +0800, Rocky Liao wrote:
> This patch adds the HCI command timeout handling, it will trigger btsoc
> to report its memory dump via vendor specific events when hit the defined
> max HCI command timeout count. After all the memory dump VSE are sent, the
> btsoc will also send a HCI_HW_ERROR event to host and this will cause a new
> hci down/up process and the btsoc will be re-initialized.
> 
> Signed-off-by: Rocky Liao <rjliao@codeaurora.org>
> ---
> 
> Changes in v2:
> - Fix build error
> Changes in v3:
> - Remove the function declaration 
> - Move the cmd_timeout() callback register to probe()
> - Remove the redundant empty line
> 
>  drivers/bluetooth/hci_qca.c | 45 +++++++++++++++++++++++++++++++++++++
>  1 file changed, 45 insertions(+)
> 
> diff --git a/drivers/bluetooth/hci_qca.c b/drivers/bluetooth/hci_qca.c
> index ca0b38f065e5..026e2e2cdd30 100644
> --- a/drivers/bluetooth/hci_qca.c
> +++ b/drivers/bluetooth/hci_qca.c
> @@ -47,6 +47,8 @@
>  #define IBS_HOST_TX_IDLE_TIMEOUT_MS	2000
>  #define CMD_TRANS_TIMEOUT_MS		100
>  
> +#define QCA_BTSOC_DUMP_CMD	0xFB
> +
>  /* susclk rate */
>  #define SUSCLK_RATE_32KHZ	32768
>  
> @@ -56,6 +58,9 @@
>  /* max retry count when init fails */
>  #define QCA_MAX_INIT_RETRY_COUNT 3
>  
> +/* when hit the max cmd time out count, trigger btsoc dump */
> +#define QCA_MAX_CMD_TIMEOUT_COUNT 3

nit: MAX_CMD_TIMEOUTS?

Similar to QCA_MAX_INIT_RETRY_COUNT on which I commented earlier I don't
think the 'QCA' prefix adds value here. The constant is defined in the driver
itself and isn't related to hardware.

> +
>  enum qca_flags {
>  	QCA_IBS_ENABLED,
>  	QCA_DROP_VENDOR_EVENT,
> @@ -123,6 +128,8 @@ struct qca_data {
>  	u64 rx_votes_off;
>  	u64 votes_on;
>  	u64 votes_off;
> +
> +	u32 cmd_timeout_cnt;

nit: cmd_timeouts?

>  };
>  
>  enum qca_speed_type {
> @@ -1332,6 +1339,11 @@ static int qca_setup(struct hci_uart *hu)
>  	if (!ret) {
>  		set_bit(QCA_IBS_ENABLED, &qca->flags);
>  		qca_debugfs_init(hdev);
> +
> +		/* clear the command time out count every time after
> +		 * initializaiton done
> +		 */
> +		qca->cmd_timeout_cnt = 0;
>  	} else if (ret == -ENOENT) {
>  		/* No patch/nvm-config found, run with original fw/config */
>  		ret = 0;
> @@ -1462,6 +1474,38 @@ static int qca_power_off(struct hci_dev *hdev)
>  	return 0;
>  }
>  
> +static int qca_send_btsoc_dump_cmd(struct hci_uart *hu)
> +{
> +	int err = 0;

The variable is pointless, just return 0 at the end of the function.

> +	struct sk_buff *skb = NULL;
> +	struct qca_data *qca = hu->priv;
> +
> +	BT_DBG("hu %p sending btsoc dump command", hu);
> +
> +	skb = bt_skb_alloc(1, GFP_ATOMIC);
> +	if (!skb) {
> +		BT_ERR("Failed to allocate memory for qca dump command");

"These generic allocation functions all emit a stack dump on failure when used
without __GFP_NOWARN so there is no use in emitting an additional failure
message when NULL is returned."

Documentation/process/coding-style.rst

hence the logging is redundant, drop it.

> +		return -ENOMEM;
> +	}
> +
> +	skb_put_u8(skb, QCA_BTSOC_DUMP_CMD);
> +
> +	skb_queue_tail(&qca->txq, skb);
> +
> +	return err;
> +}
> +
> +static void qca_cmd_timeout(struct hci_dev *hdev)
> +{
> +	struct hci_uart *hu = hci_get_drvdata(hdev);
> +	struct qca_data *qca = hu->priv;
> +
> +	BT_ERR("hu %p hci cmd timeout count=0x%x", hu, ++qca->cmd_timeout_cnt);

Is there any particular reason to print the counter in hex instead of
decimal?

Should this use bt_dev_err() since we have a hdev in this context?

> +
> +	if (qca->cmd_timeout_cnt >= QCA_MAX_CMD_TIMEOUT_COUNT)
> +		qca_send_btsoc_dump_cmd(hu);
> +}
> +
>  static int qca_regulator_enable(struct qca_serdev *qcadev)
>  {
>  	struct qca_power *power = qcadev->bt_power;
> @@ -1605,6 +1649,7 @@ static int qca_serdev_probe(struct serdev_device *serdev)
>  		hdev = qcadev->serdev_hu.hdev;
>  		set_bit(HCI_QUIRK_NON_PERSISTENT_SETUP, &hdev->quirks);
>  		hdev->shutdown = qca_power_off;
> +		hdev->cmd_timeout = qca_cmd_timeout;
>  	}
>  
>  out:	return err;
> -- 
> The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum, a Linux Foundation Collaborative Project