Re: [PATCH] tty: serial: msm_serial: Fix flow control

[Bjorn Andersson]

On Mon 21 Oct 08:19 PDT 2019, Jeffrey Hugo wrote:

> On Sun, Oct 20, 2019 at 12:28 AM Bjorn Andersson
> <bjorn.andersson@xxxxxxxxxx> wrote:
> >
> > On Sat 19 Oct 14:06 PDT 2019, Jeffrey Hugo wrote:
> >
> > > hci_qca interfaces to the wcn3990 via a uart_dm on the msm8998 mtp and
> > > Lenovo Miix 630 laptop.  As part of initializing the wcn3990, hci_qca
> > > disables flow, configures the uart baudrate, and then reenables flow - at
> > > which point an event is expected to be received over the uart from the
> > > wcn3990.  It is observed that this event comes after the baudrate change
> > > but before hci_qca re-enables flow. This is unexpected, and is a result of
> > > msm_reset() being broken.
> > >
> > > According to the uart_dm hardware documentation, it is recommended that
> > > automatic hardware flow control be enabled by setting RX_RDY_CTL.  Auto
> > > hw flow control will manage RFR based on the configured watermark.  When
> > > there is space to receive data, the hw will assert RFR.  When the watermark
> > > is hit, the hw will de-assert RFR.
> > >
> > > The hardware documentation indicates that RFR can me manually managed via
> > > CR when RX_RDY_CTL is not set.  SET_RFR asserts RFR, and RESET_RFR
> > > de-asserts RFR.
> > >
> > > msm_reset() is broken because after resetting the hardware, it
> > > unconditionally asserts RFR via SET_RFR.  This enables flow regardless of
> > > the current configuration, and would undo a previous flow disable
> > > operation.  It should instead de-assert RFR via RESET_RFR to block flow
> > > until the hardware is reconfigured.  msm_serial should rely on the client
> > > to specify that flow should be enabled, either via mctrl() or the termios
> > > structure, and only assert RFR in response to those triggers.
> > >
> >
> > I traced msm_reset() and msm_set_mctrl() and I get the following:
> >        swapper/0-1     [000] d..1     3.091726: msm_set_mctrl: msm_set_mctrl() reset rfr
> >        swapper/0-1     [000] d..1     3.117046: msm_set_mctrl: msm_set_mctrl() reset rfr
> >        swapper/0-1     [000] d..1     3.125484: msm_set_termios: msm_reset() set rfr
> >        swapper/0-1     [003] d..1     4.491430: msm_set_termios: msm_reset() set rfr
> >        swapper/0-1     [003] d..1     4.491733: msm_set_mctrl: msm_set_mctrl() auto rfr
> >            login-313   [001] d..1    78.010785: msm_set_mctrl: msm_set_mctrl() reset rfr
> >            login-313   [001] d..1    78.011007: msm_set_termios: msm_reset() set rfr
> >            login-313   [001] d..1    78.011021: msm_set_mctrl: msm_set_mctrl() auto rfr
> >            login-313   [001] d..1    78.063330: msm_set_termios: msm_reset() set rfr
> >            login-313   [001] d..1    78.063641: msm_set_termios: msm_reset() set rfr
> >
> > So while your change does make sense for BT, wouldn't this mean that
> > with your patch and a 4-pin UART for the console I would end this dance
> > with receive flow blocked?
> 
> No.  I don't think it occurred to you to consider how RX_RDY_CTL
> factors into this.  RX_RDY_CTL allows the hardware to enable flow when
> the hardware determines it is able to receive data.  RX_RDY_CTL gets
> set in mctrl, and set_termios - essentially when the uart client has
> indicated flow can be enabled.
> 

I see, so per above trace when login does it's rfr toggling RX_RDY_CTL
is already set (and will be maintained), so we'll allow flow at each
part of the process.

> Even though the console is a 2 wire console on msm8998, the driver
> will go through the same flow.  I verified that RX_RDY_CTL is set on
> the console uart after boot, so we won't end up with flow blocked.
> 
> However, this does bring up another issue.  RX_RDY_CTL doesn't get
> unset in msm_reset.  RESET_RX will disable the rx path in the
> hardware, and prevent RX_RDY_CTL from being active, but the rx path is
> enabled in set_baud_rate() before RX_RDY_CTL can be masked out in
> set_termios, so there may be a small window where flow can
> inadvertently enabled.  I'll spin a v2 that also unsets RX_RDY_CTL in
> msm_reset().
> 

Yes, that sounds reasonable. And for the console case the difference
will be that we deassert RFR while the serial is being reset and
reconfigured; which sounds like the right thing to do.

Regards,
Bjorn

> >
> > Regards,
> > Bjorn
> >
> > > Fixes: 04896a77a97b ("msm_serial: serial driver for MSM7K onboard serial peripheral.")
> > > Signed-off-by: Jeffrey Hugo <jeffrey.l.hugo@xxxxxxxxx>
> > > ---
> > >  drivers/tty/serial/msm_serial.c | 2 +-
> > >  1 file changed, 1 insertion(+), 1 deletion(-)
> > >
> > > diff --git a/drivers/tty/serial/msm_serial.c b/drivers/tty/serial/msm_serial.c
> > > index 3657a24913fc..aedabf7646f1 100644
> > > --- a/drivers/tty/serial/msm_serial.c
> > > +++ b/drivers/tty/serial/msm_serial.c
> > > @@ -987,7 +987,7 @@ static void msm_reset(struct uart_port *port)
> > >       msm_write(port, UART_CR_CMD_RESET_ERR, UART_CR);
> > >       msm_write(port, UART_CR_CMD_RESET_BREAK_INT, UART_CR);
> > >       msm_write(port, UART_CR_CMD_RESET_CTS, UART_CR);
> > > -     msm_write(port, UART_CR_CMD_SET_RFR, UART_CR);
> > > +     msm_write(port, UART_CR_CMD_RESET_RFR, UART_CR);
> > >
> > >       /* Disable DM modes */
> > >       if (msm_port->is_uartdm)
> > > --
> > > 2.17.1
> > >