Re: [PATCH v5 1/5] Bluetooth: hci_qca: use wait_until_sent() for power pulses

[Matthias Kaehlcke <mka@chromium.org>]

On Fri, Jan 11, 2019 at 08:02:00PM +0530, Balakrishna Godavarthi wrote:
> On 2019-01-11 06:25, Matthias Kaehlcke wrote:
> > On Thu, Jan 10, 2019 at 08:18:37PM +0530, Balakrishna Godavarthi wrote:
> > > Hi Johan,
> > > 
> > > On 2019-01-09 20:08, Johan Hovold wrote:
> > > > On Fri, Dec 21, 2018 at 05:59:47PM -0800, Matthias Kaehlcke wrote:
> > > > > On Thu, Dec 20, 2018 at 08:16:35PM +0530, Balakrishna Godavarthi
> > > > > wrote:
> > > > > > wcn3990 requires a power pulse to turn ON/OFF along with
> > > > > > regulators. Sometimes we are observing the power pulses are sent
> > > > > > out with some time delay, due to queuing these commands. This is
> > > > > > causing synchronization issues with chip, which intern delay the
> > > > > > chip setup or may end up with communication issues.
> > > > > >
> > > > > > Signed-off-by: Balakrishna Godavarthi <bgodavar@codeaurora.org>
> > > > > > ---
> > > > > >  drivers/bluetooth/hci_qca.c | 38 ++++++++++++++-----------------------
> > > > > >  1 file changed, 14 insertions(+), 24 deletions(-)
> > > > > >
> > > > > > diff --git a/drivers/bluetooth/hci_qca.c b/drivers/bluetooth/hci_qca.c
> > > > > > index f036c8f98ea3..5a07c2370289 100644
> > > > > > --- a/drivers/bluetooth/hci_qca.c
> > > > > > +++ b/drivers/bluetooth/hci_qca.c
> > > > > > @@ -1013,11 +1013,9 @@ static inline void host_set_baudrate(struct hci_uart *hu, unsigned int speed)
> > > > > >  		hci_uart_set_baudrate(hu, speed);
> > > > > >  }
> > > > > >
> > > > > > -static int qca_send_power_pulse(struct hci_dev *hdev, u8 cmd)
> > > > > > +static int qca_send_power_pulse(struct hci_uart *hu, u8 cmd)
> > > > > >  {
> > > > > > -	struct hci_uart *hu = hci_get_drvdata(hdev);
> > > > > > -	struct qca_data *qca = hu->priv;
> > > > > > -	struct sk_buff *skb;
> > > > > > +	int ret;
> > > > > >
> > > > > >  	/* These power pulses are single byte command which are sent
> > > > > >  	 * at required baudrate to wcn3990. On wcn3990, we have an external
> > > > > > @@ -1029,19 +1027,16 @@ static int qca_send_power_pulse(struct hci_dev *hdev, u8 cmd)
> > > > > >  	 * save power. Disabling hardware flow control is mandatory while
> > > > > >  	 * sending power pulses to SoC.
> > > > > >  	 */
> > > > > > -	bt_dev_dbg(hdev, "sending power pulse %02x to SoC", cmd);
> > > > > > -
> > > > > > -	skb = bt_skb_alloc(sizeof(cmd), GFP_KERNEL);
> > > > > > -	if (!skb)
> > > > > > -		return -ENOMEM;
> > > > > > -
> > > > > > +	bt_dev_dbg(hu->hdev, "sending power pulse %02x to SoC", cmd);
> > > > > >  	hci_uart_set_flow_control(hu, true);
> > > > > > +	ret = serdev_device_write_buf(hu->serdev, &cmd, sizeof(cmd));
> > > > > > +	if (ret < 0) {
> > > > > > +		bt_dev_err(hu->hdev, "failed to send power pulse %02x to SoC",
> > > > > > +			   cmd);
> > > > > > +		return ret;
> > > > > > +	}
> > > > > >
> > > > > > -	skb_put_u8(skb, cmd);
> > > > > > -	hci_skb_pkt_type(skb) = HCI_COMMAND_PKT;
> > > > > > -
> > > > > > -	skb_queue_tail(&qca->txq, skb);
> > > > > > -	hci_uart_tx_wakeup(hu);
> > > > > > +	serdev_device_wait_until_sent(hu->serdev, 0);
> > > >
> > > > Again, do you really want to wait indefinitely here?
> > > >
> > > [Bala]: these commands are mandatory to turn ON or OFF the chip.
> > >         so blocking to the max time is required.
> > >         these commands are sent during the BT chip ON & OFF.
> > >         in the latest series, i have flushed the uart before sending
> > > this
> > > commands
> > >         so the uart FIFO(as just opened the port before calling this
> > > function) or the circular
> > >         buffer will be empty and also i am disabling the flow
> > > control too.
> > >        https://patchwork.kernel.org/patch/10744435/
> > 
> > The commands may be mandatory for switching the chip on or off, but
> > what is better if there is a problem with sending them (e.g. a buggy
> > UART driver):
> > 
> > 1. wait a reasonable time, report an error
> > 2. wait forever
> > 
> > ?
> > 
> > If the single byte command couldn't be sent after a few milliseconds,
> > it likely never will, waiting forever doesn't fix that. An error
> > report at least provides some information about the problem and the
> > driver is in a not-hanging state.
> > 
> > Cheers
> > 
> > Matthias
> 
> [Bala]: will update this with a bound TIMEOUT value. But wait_until_sent()
> is void return
>         type how could we know that the data is sent out on the lines.

Good point, I didn't check and expected it to return an error. If you
feel really motivated and have maintainer support you could possibly
change the API, however it seems this would be a somewhat larger
change.

I guess the next best thing to do is to proceed as if all data was
sent and if there was a problem it will likely manifest through
another error (especially for the ON pulse), which still seems better
than a hanging driver.

Cheers

Matthias