RE: [PATCH V2] tty: serial: fsl_lpuart: disable transmitter before changing RS485 related registers

[Sherry Sun]

> -----Original Message-----
> From: Jiri Slaby <jirislaby@xxxxxxxxxx>
> Sent: Tuesday, March 11, 2025 11:45 AM
> To: Sherry Sun <sherry.sun@xxxxxxx>; gregkh@xxxxxxxxxxxxxxxxxxx
> Cc: linux-serial@xxxxxxxxxxxxxxx; linux-kernel@xxxxxxxxxxxxxxx;
> imx@xxxxxxxxxxxxxxx; Shenwei Wang <shenwei.wang@xxxxxxx>; Frank Li
> <frank.li@xxxxxxx>
> Subject: Re: [PATCH V2] tty: serial: fsl_lpuart: disable transmitter before
> changing RS485 related registers
> 
> On 11. 03. 25, 3:55, Sherry Sun wrote:
> > According to the LPUART reference manual, TXRTSE and TXRTSPOL of
> MODIR
> > register only can be changed when the transmitter is disabled.
> > So disable the transmitter before changing RS485 related registers and
> > re-enable it after the change is done.
> >
> > Fixes: 67b01837861c ("tty: serial: lpuart: Add RS485 support for
> > 32-bit uart flavour")
> > Signed-off-by: Sherry Sun <sherry.sun@xxxxxxx>
> > Reviewed-by: Frank Li <Frank.Li@xxxxxxx>
> > ---
> > Changes in V2:
> > 1. Add TE bit polling read to ensure TE is really become 0 before proceeding.
> > 2. Add Reviewed-by tag.
> > ---
> >   drivers/tty/serial/fsl_lpuart.c | 17 +++++++++++++++++
> >   1 file changed, 17 insertions(+)
> >
> > diff --git a/drivers/tty/serial/fsl_lpuart.c
> > b/drivers/tty/serial/fsl_lpuart.c index 91d02c55c470..6b2f3a73a367
> > 100644
> > --- a/drivers/tty/serial/fsl_lpuart.c
> > +++ b/drivers/tty/serial/fsl_lpuart.c
> > @@ -1484,6 +1484,19 @@ static int lpuart32_config_rs485(struct
> > uart_port *port, struct ktermios *termio
> >
> >   	unsigned long modem = lpuart32_read(&sport->port, UARTMODIR)
> >   				& ~(UARTMODIR_TXRTSPOL |
> UARTMODIR_TXRTSE);
> 
> This is unrelated, but why is the above ulong?

Hi Jiri, the following lpuart cleanup patch will change all the "unsigned long"
type to the correct u32.

> 
> > +	u32 ctrl;
> > +
> > +	/* TXRTSE and TXRTSPOL only can be changed when transmitter is
> disabled. */
> > +	ctrl = lpuart32_read(&sport->port, UARTCTRL);
> > +	if (ctrl & UARTCTRL_TE) {
> > +		/* wait transmit engin complete */
> 
> wait for the transmit engine to complete

Thanks, will fix.

> 
> > +		lpuart32_wait_bit_set(&sport->port, UARTSTAT,
> UARTSTAT_TC);
> 
> Both this ^^ and:
> 
> > +		lpuart32_write(&sport->port, ctrl & ~UARTCTRL_TE,
> UARTCTRL);
> > +
> > +		while (lpuart32_read(&sport->port, UARTCTRL) &
> UARTCTRL_TE)
> > +			cpu_relax();
> 
> this ^^ are unbound loops in case the hardware gets mad :(.
> 
> Anyway, IIUC, after the TE clear from CTRL by the above write, the TE bit is
> really cleared by the HW from CTRL only after it is really disabled, so has to be
> checked?
> 
> > +	}
> > +

Description of TE bit in LPUART RM: "After this field becomes 0, the field reads 1 until
the transmitter has completed the current character and the TXD pin is tristated".

I also tried to read back the TE bit right after write 0, actually it didn't become 0
immediately, so I added the polling read here.
You can check the discussion between Shenwei and me for more details. Thanks!
https://lore.kernel.org/imx/20250307021950.1000221-1-sherry.sun@xxxxxxx/T/#m14098058b083333213bd74414095aa1ef3299780

Best Regards
Sherry