| Message ID | 20200313134635.1.Icf54c533065306b02b880c46dfd401d8db34e213@changeid (mailing list archive) |
|---|---|
| State | Accepted |
| Commit | e83766334f96b3396a71c7baa3b0b53dfd5190cd |
| Headers | show |
| Series | [1/2] tty: serial: qcom_geni_serial: No need to stop tx/rx on UART shutdown | expand |
Quoting Douglas Anderson (2020-03-13 13:46:51) > On a board using qcom_geni_serial I found that I could no longer > interact with kdb if I got a crash after the "agetty" running on the > same serial port was killed. This meant that various classes of > crashes that happened at reboot time were undebuggable. > > Reading through the code, I couldn't figure out why qcom_geni_serial > felt the need to run so much code at port shutdown time. All we need > to do is disable the interrupt. > > After I make this change then a hardcoded kgdb_breakpoint in some late > shutdown code now allows me to interact with the debugger. I also > could freely close / re-open the port without problems. > > Fixes: c4f528795d1a ("tty: serial: msm_geni_serial: Add serial driver support for GENI based QUP") > Signed-off-by: Douglas Anderson <dianders@chromium.org> > --- Matches what is described in Documentation/driver-api/serial/driver.rst Reviewed-by: Stephen Boyd <swboyd@chromium.org>
On 3/14/2020 2:16 AM, Douglas Anderson wrote: > On a board using qcom_geni_serial I found that I could no longer > interact with kdb if I got a crash after the "agetty" running on the > same serial port was killed. This meant that various classes of > crashes that happened at reboot time were undebuggable. > > Reading through the code, I couldn't figure out why qcom_geni_serial > felt the need to run so much code at port shutdown time. All we need > to do is disable the interrupt. > > After I make this change then a hardcoded kgdb_breakpoint in some late > shutdown code now allows me to interact with the debugger. I also > could freely close / re-open the port without problems. > > Fixes: c4f528795d1a ("tty: serial: msm_geni_serial: Add serial driver support for GENI based QUP") > Signed-off-by: Douglas Anderson <dianders@chromium.org> > --- Looks good to me. Reviewed-by: Akash Asthana <akashast@codeaurora.org>
diff --git a/drivers/tty/serial/qcom_geni_serial.c b/drivers/tty/serial/qcom_geni_serial.c index 272bae0eebc7..09d8612517aa 100644 --- a/drivers/tty/serial/qcom_geni_serial.c +++ b/drivers/tty/serial/qcom_geni_serial.c @@ -827,17 +827,11 @@ static void get_tx_fifo_size(struct qcom_geni_serial_port *port) static void qcom_geni_serial_shutdown(struct uart_port *uport) { - unsigned long flags; - /* Stop the console before stopping the current tx */ if (uart_console(uport)) console_stop(uport->cons); disable_irq(uport->irq); - spin_lock_irqsave(&uport->lock, flags); - qcom_geni_serial_stop_tx(uport); - qcom_geni_serial_stop_rx(uport); - spin_unlock_irqrestore(&uport->lock, flags); } static int qcom_geni_serial_port_setup(struct uart_port *uport)
On a board using qcom_geni_serial I found that I could no longer interact with kdb if I got a crash after the "agetty" running on the same serial port was killed. This meant that various classes of crashes that happened at reboot time were undebuggable. Reading through the code, I couldn't figure out why qcom_geni_serial felt the need to run so much code at port shutdown time. All we need to do is disable the interrupt. After I make this change then a hardcoded kgdb_breakpoint in some late shutdown code now allows me to interact with the debugger. I also could freely close / re-open the port without problems. Fixes: c4f528795d1a ("tty: serial: msm_geni_serial: Add serial driver support for GENI based QUP") Signed-off-by: Douglas Anderson <dianders@chromium.org> --- drivers/tty/serial/qcom_geni_serial.c | 6 ------ 1 file changed, 6 deletions(-)