| Message ID | 1570700763-17319-1-git-send-email-akashast@codeaurora.org (mailing list archive) |
|---|---|
| State | Superseded, archived |
| Headers | show |
| Series | [V2,1/2] tty: serial: qcom_geni_serial: IRQ cleanup | expand |
Quoting Akash Asthana (2019-10-10 02:46:03) > diff --git a/drivers/tty/serial/qcom_geni_serial.c b/drivers/tty/serial/qcom_geni_serial.c > index 14c6306..5180cd8 100644 > --- a/drivers/tty/serial/qcom_geni_serial.c > +++ b/drivers/tty/serial/qcom_geni_serial.c > @@ -1297,11 +1291,21 @@ static int qcom_geni_serial_probe(struct platform_device *pdev) > port->rx_fifo_depth = DEF_FIFO_DEPTH_WORDS; > port->tx_fifo_width = DEF_FIFO_WIDTH_BITS; > > + scnprintf(port->name, sizeof(port->name), "qcom_geni_serial_%s%d", > + (uart_console(uport) ? "console" : "uart"), uport->line); Drop useless parenthesis. Also, it might make more sense to make this a devm_kasprintf() call now. > irq = platform_get_irq(pdev, 0); > if (irq < 0) > return irq; > uport->irq = irq; > > + irq_set_status_flags(uport->irq, IRQ_NOAUTOEN); Is there a reason why we can't always leave the irq enabled and request it later once the uart structure has been fully initialized? > + ret = devm_request_irq(uport->dev, uport->irq, qcom_geni_serial_isr, > + IRQF_TRIGGER_HIGH, port->name, uport); > + if (ret) { > + dev_err(uport->dev, "Failed to get IRQ ret %d\n", ret); > + return ret; > + } > + > uport->private_data = drv; > platform_set_drvdata(pdev, port); > port->handle_rx = console ? handle_rx_console : handle_rx_uart;
On 10/10/2019 7:49 PM, Stephen Boyd wrote: > Quoting Akash Asthana (2019-10-10 02:46:03) >> diff --git a/drivers/tty/serial/qcom_geni_serial.c b/drivers/tty/serial/qcom_geni_serial.c >> index 14c6306..5180cd8 100644 >> --- a/drivers/tty/serial/qcom_geni_serial.c >> +++ b/drivers/tty/serial/qcom_geni_serial.c >> @@ -1297,11 +1291,21 @@ static int qcom_geni_serial_probe(struct platform_device *pdev) >> port->rx_fifo_depth = DEF_FIFO_DEPTH_WORDS; >> port->tx_fifo_width = DEF_FIFO_WIDTH_BITS; >> >> + scnprintf(port->name, sizeof(port->name), "qcom_geni_serial_%s%d", >> + (uart_console(uport) ? "console" : "uart"), uport->line); > Drop useless parenthesis. Also, it might make more sense to make this a > devm_kasprintf() call now. OK >> irq = platform_get_irq(pdev, 0); >> if (irq < 0) >> return irq; >> uport->irq = irq; >> >> + irq_set_status_flags(uport->irq, IRQ_NOAUTOEN); > Is there a reason why we can't always leave the irq enabled and request > it later once the uart structure has been fully initialized? According to current design we are requesting IRQ handler in probe, and we enable and disable it from the startup(port open) and shutdown(port close) function respectively. We need to call for disable_irq in shutdown function because client has closed the port and we don't expect any transfer requests after it. >>request it later once the uart structure has been fully initialized? Is the ask is to move request irq later in probe after the uport is fully initialized? >> + ret = devm_request_irq(uport->dev, uport->irq, qcom_geni_serial_isr, >> + IRQF_TRIGGER_HIGH, port->name, uport); >> + if (ret) { >> + dev_err(uport->dev, "Failed to get IRQ ret %d\n", ret); >> + return ret; >> + } >> + >> uport->private_data = drv; >> platform_set_drvdata(pdev, port); >> port->handle_rx = console ? handle_rx_console : handle_rx_uart;
Quoting Akash Asthana (2019-10-11 00:39:06) > > On 10/10/2019 7:49 PM, Stephen Boyd wrote: > > Quoting Akash Asthana (2019-10-10 02:46:03) > >> diff --git a/drivers/tty/serial/qcom_geni_serial.c b/drivers/tty/serial/qcom_geni_serial.c > >> index 14c6306..5180cd8 100644 > >> --- a/drivers/tty/serial/qcom_geni_serial.c > >> +++ b/drivers/tty/serial/qcom_geni_serial.c > >> @@ -1297,11 +1291,21 @@ static int qcom_geni_serial_probe(struct platform_device *pdev) > >> port->rx_fifo_depth = DEF_FIFO_DEPTH_WORDS; > >> port->tx_fifo_width = DEF_FIFO_WIDTH_BITS; > >> > >> + scnprintf(port->name, sizeof(port->name), "qcom_geni_serial_%s%d", > >> + (uart_console(uport) ? "console" : "uart"), uport->line); > > Drop useless parenthesis. Also, it might make more sense to make this a > > devm_kasprintf() call now. > > OK > > >> irq = platform_get_irq(pdev, 0); > >> if (irq < 0) > >> return irq; > >> uport->irq = irq; > >> > >> + irq_set_status_flags(uport->irq, IRQ_NOAUTOEN); > > Is there a reason why we can't always leave the irq enabled and request > > it later once the uart structure has been fully initialized? > > According to current design we are requesting IRQ handler in probe, and > we enable and disable it from the startup(port open) and shutdown(port > close) function respectively. > > We need to call for disable_irq in shutdown function because client has > closed the port and we don't expect any transfer requests after it. > > >>request it later once the uart structure has been fully initialized? > > Is the ask is to move request irq later in probe after the uport is > fully initialized? Yes I'm wondering if we can get rid of the IRQ_NOAUTOEN and irq_enable/disable bits in here and leave the interrupt enabled all the time.
On 10/16/2019 1:41 AM, Stephen Boyd wrote: > Quoting Akash Asthana (2019-10-11 00:39:06) >> On 10/10/2019 7:49 PM, Stephen Boyd wrote: >>> Quoting Akash Asthana (2019-10-10 02:46:03) >>>> diff --git a/drivers/tty/serial/qcom_geni_serial.c b/drivers/tty/serial/qcom_geni_serial.c >>>> index 14c6306..5180cd8 100644 >>>> --- a/drivers/tty/serial/qcom_geni_serial.c >>>> +++ b/drivers/tty/serial/qcom_geni_serial.c >>>> @@ -1297,11 +1291,21 @@ static int qcom_geni_serial_probe(struct platform_device *pdev) >>>> port->rx_fifo_depth = DEF_FIFO_DEPTH_WORDS; >>>> port->tx_fifo_width = DEF_FIFO_WIDTH_BITS; >>>> >>>> + scnprintf(port->name, sizeof(port->name), "qcom_geni_serial_%s%d", >>>> + (uart_console(uport) ? "console" : "uart"), uport->line); >>> Drop useless parenthesis. Also, it might make more sense to make this a >>> devm_kasprintf() call now. >> OK >> >>>> irq = platform_get_irq(pdev, 0); >>>> if (irq < 0) >>>> return irq; >>>> uport->irq = irq; >>>> >>>> + irq_set_status_flags(uport->irq, IRQ_NOAUTOEN); >>> Is there a reason why we can't always leave the irq enabled and request >>> it later once the uart structure has been fully initialized? >> According to current design we are requesting IRQ handler in probe, and >> we enable and disable it from the startup(port open) and shutdown(port >> close) function respectively. >> >> We need to call for disable_irq in shutdown function because client has >> closed the port and we don't expect any transfer requests after it. >> >>>> request it later once the uart structure has been fully initialized? >> Is the ask is to move request irq later in probe after the uport is >> fully initialized? > Yes I'm wondering if we can get rid of the IRQ_NOAUTOEN and > irq_enable/disable bits in here and leave the interrupt enabled all the > time. Ideally it should work, I will run few experiments to make sure there isn't any spurious interrupts problem after port close is called. If it works, I will post a separate patch for it.
diff --git a/drivers/tty/serial/qcom_geni_serial.c b/drivers/tty/serial/qcom_geni_serial.c index 14c6306..5180cd8 100644 --- a/drivers/tty/serial/qcom_geni_serial.c +++ b/drivers/tty/serial/qcom_geni_serial.c @@ -9,6 +9,7 @@ #include <linux/console.h> #include <linux/io.h> #include <linux/iopoll.h> +#include <linux/irq.h> #include <linux/module.h> #include <linux/of.h> #include <linux/of_device.h> @@ -830,7 +831,7 @@ static void qcom_geni_serial_shutdown(struct uart_port *uport) if (uart_console(uport)) console_stop(uport->cons); - free_irq(uport->irq, uport); + disable_irq(uport->irq); spin_lock_irqsave(&uport->lock, flags); qcom_geni_serial_stop_tx(uport); qcom_geni_serial_stop_rx(uport); @@ -890,21 +891,14 @@ static int qcom_geni_serial_startup(struct uart_port *uport) int ret; struct qcom_geni_serial_port *port = to_dev_port(uport, uport); - scnprintf(port->name, sizeof(port->name), - "qcom_serial_%s%d", - (uart_console(uport) ? "console" : "uart"), uport->line); - if (!port->setup) { ret = qcom_geni_serial_port_setup(uport); if (ret) return ret; } + enable_irq(uport->irq); - ret = request_irq(uport->irq, qcom_geni_serial_isr, IRQF_TRIGGER_HIGH, - port->name, uport); - if (ret) - dev_err(uport->dev, "Failed to get IRQ ret %d\n", ret); - return ret; + return 0; } static unsigned long get_clk_cfg(unsigned long clk_freq) @@ -1297,11 +1291,21 @@ static int qcom_geni_serial_probe(struct platform_device *pdev) port->rx_fifo_depth = DEF_FIFO_DEPTH_WORDS; port->tx_fifo_width = DEF_FIFO_WIDTH_BITS; + scnprintf(port->name, sizeof(port->name), "qcom_geni_serial_%s%d", + (uart_console(uport) ? "console" : "uart"), uport->line); irq = platform_get_irq(pdev, 0); if (irq < 0) return irq; uport->irq = irq; + irq_set_status_flags(uport->irq, IRQ_NOAUTOEN); + ret = devm_request_irq(uport->dev, uport->irq, qcom_geni_serial_isr, + IRQF_TRIGGER_HIGH, port->name, uport); + if (ret) { + dev_err(uport->dev, "Failed to get IRQ ret %d\n", ret); + return ret; + } + uport->private_data = drv; platform_set_drvdata(pdev, port); port->handle_rx = console ? handle_rx_console : handle_rx_uart;
Move ISR registration from startup to probe function to avoid registering it everytime when the port open is called for driver. Signed-off-by: Akash Asthana <akashast@codeaurora.org> --- drivers/tty/serial/qcom_geni_serial.c | 24 ++++++++++++++---------- 1 file changed, 14 insertions(+), 10 deletions(-)