The commit 60f361722ad2 ("serial: fsl_lpuart: Reset prior to
registration") has moved the reset before the port registration,
this change makes the uart_console() in lpuart_global_reset() no sense
since the port is not registered to the system yet. So drop it.
But we sill need to check if the port is used as an early console
before performing the port reset, otherwise the reset will hang the
system.
Fixes: 60f361722ad2 ("serial: fsl_lpuart: Reset prior to registration")
Signed-off-by: Xiaolei Wang <[email protected]>
---
drivers/tty/serial/fsl_lpuart.c | 5 ++++-
1 file changed, 4 insertions(+), 1 deletion(-)
diff --git a/drivers/tty/serial/fsl_lpuart.c b/drivers/tty/serial/fsl_lpuart.c
index 67fa113f77d4..d0efeba5c695 100644
--- a/drivers/tty/serial/fsl_lpuart.c
+++ b/drivers/tty/serial/fsl_lpuart.c
@@ -239,6 +239,8 @@
/* IMX lpuart has four extra unused regs located at the beginning */
#define IMX_REG_OFF 0x10
+static resource_size_t lpuart_earlycon_mapbase;
+
enum lpuart_type {
VF610_LPUART,
LS1021A_LPUART,
@@ -410,7 +412,7 @@ static int lpuart_global_reset(struct lpuart_port *sport)
void __iomem *global_addr;
int ret;
- if (uart_console(port))
+ if ((port->mapbase - 0x10) == lpuart_earlycon_mapbase)
return 0;
ret = clk_prepare_enable(sport->ipg_clk);
@@ -2604,6 +2606,7 @@ static int __init lpuart32_imx_early_console_setup(struct earlycon_device *devic
device->port.iotype = UPIO_MEM32;
device->port.membase += IMX_REG_OFF;
device->con->write = lpuart32_early_write;
+ lpuart_earlycon_mapbase = device->port.mapbase;
return 0;
}
--
2.25.1
> -----Original Message-----
> From: Xiaolei Wang <[email protected]>
> Sent: 2022??10??24?? 17:12
> To: [email protected]; [email protected]; [email protected]
> Cc: [email protected]; Sherry Sun <[email protected]>;
> [email protected]; [email protected]; linux-
> [email protected]
> Subject: [PATCH] serial: fsl_lpuart: Skip the reset for the early console port
>
> The commit 60f361722ad2 ("serial: fsl_lpuart: Reset prior to
> registration") has moved the reset before the port registration, this change
> makes the uart_console() in lpuart_global_reset() no sense since the port is
> not registered to the system yet. So drop it.
> But we sill need to check if the port is used as an early console before
> performing the port reset, otherwise the reset will hang the system.
>
> Fixes: 60f361722ad2 ("serial: fsl_lpuart: Reset prior to registration")
> Signed-off-by: Xiaolei Wang <[email protected]>
> ---
> drivers/tty/serial/fsl_lpuart.c | 5 ++++-
> 1 file changed, 4 insertions(+), 1 deletion(-)
>
> diff --git a/drivers/tty/serial/fsl_lpuart.c b/drivers/tty/serial/fsl_lpuart.c
> index 67fa113f77d4..d0efeba5c695 100644
> --- a/drivers/tty/serial/fsl_lpuart.c
> +++ b/drivers/tty/serial/fsl_lpuart.c
> @@ -239,6 +239,8 @@
> /* IMX lpuart has four extra unused regs located at the beginning */
> #define IMX_REG_OFF 0x10
>
> +static resource_size_t lpuart_earlycon_mapbase;
> +
> enum lpuart_type {
> VF610_LPUART,
> LS1021A_LPUART,
> @@ -410,7 +412,7 @@ static int lpuart_global_reset(struct lpuart_port
> *sport)
> void __iomem *global_addr;
> int ret;
>
> - if (uart_console(port))
> + if ((port->mapbase - 0x10) == lpuart_earlycon_mapbase)
> return 0;
>
> ret = clk_prepare_enable(sport->ipg_clk);
> @@ -2604,6 +2606,7 @@ static int __init
> lpuart32_imx_early_console_setup(struct earlycon_device *devic
> device->port.iotype = UPIO_MEM32;
> device->port.membase += IMX_REG_OFF;
> device->con->write = lpuart32_early_write;
> + lpuart_earlycon_mapbase = device->port.mapbase;
>
Hi, thanks for the fix for lpuart driver, but this looks more like a workaround patch.
Per my understanding, the root cause for this issue is that lpuart_global_reset() break the on-going transfer, such as the earlycon here.
I have sent a patch to fix the lpuart console hang issue, please check https://lore.kernel.org/lkml/[email protected]/
Best Regards
Sherry