2015-12-01 01:52:02

by Rafael J. Wysocki

[permalink] [raw]
Subject: Re: [PATCH 3/3] serial: amba-pl011: add ACPI support to AMBA probe

On Wednesday, September 30, 2015 11:38:50 AM Graeme Gregory wrote:
> In ACPI this device is only defined in SBSA mode so
> if we are probing from ACPI use this mode.
>
> Cc: Rafael J. Wysocki <[email protected]>
> Cc: Len Brown <[email protected]>
> Cc: Russell King <[email protected]>
> Cc: Greg Kroah-Hartman <[email protected]>
> Signed-off-by: Graeme Gregory <[email protected]>
> ---
> drivers/tty/serial/amba-pl011.c | 32 +++++++++++++++++++++-----------
> 1 file changed, 21 insertions(+), 11 deletions(-)
>
> diff --git a/drivers/tty/serial/amba-pl011.c b/drivers/tty/serial/amba-pl011.c
> index fd27e98..55209aa 100644
> --- a/drivers/tty/serial/amba-pl011.c
> +++ b/drivers/tty/serial/amba-pl011.c
> @@ -2368,18 +2368,28 @@ static int pl011_probe(struct amba_device *dev, const struct amba_id *id)
> if (!uap)
> return -ENOMEM;
>
> - uap->clk = devm_clk_get(&dev->dev, NULL);
> - if (IS_ERR(uap->clk))
> - return PTR_ERR(uap->clk);
> -
> - uap->vendor = vendor;
> - uap->lcrh_rx = vendor->lcrh_rx;
> - uap->lcrh_tx = vendor->lcrh_tx;
> - uap->fifosize = vendor->get_fifosize(dev);
> + /* ACPI only defines SBSA variant */
> + if (!ACPI_COMPANION(&dev->dev)) {

It would read more straightforward if you did

if (ACPI_COMPANION(&dev->dev)) {
<handle the ACPI case>
} else {
<handle the non-ACPI case>
}

> + uap->clk = devm_clk_get(&dev->dev, NULL);
> + if (IS_ERR(uap->clk))
> + return PTR_ERR(uap->clk);
> +
> + uap->vendor = vendor;
> + uap->lcrh_rx = vendor->lcrh_rx;
> + uap->lcrh_tx = vendor->lcrh_tx;
> + uap->fifosize = vendor->get_fifosize(dev);
> + uap->port.ops = &amba_pl011_pops;
> + snprintf(uap->type, sizeof(uap->type), "PL011 rev%u",
> + amba_rev(dev));
> + } else {
> + uap->vendor = &vendor_sbsa;
> + uap->fifosize = 32;
> + uap->port.ops = &sbsa_uart_pops;
> + uap->fixed_baud = 115200;
> +
> + snprintf(uap->type, sizeof(uap->type), "SBSA");

This looks sort of heavy-handed.

Is this the only possible configuration of the device in the ACPI case?

> + }
> uap->port.irq = dev->irq[0];
> - uap->port.ops = &amba_pl011_pops;
> -
> - snprintf(uap->type, sizeof(uap->type), "PL011 rev%u", amba_rev(dev));
>
> ret = pl011_setup_port(&dev->dev, uap, &dev->res, portnr);
> if (ret)

Thanks,
Rafael


2015-12-01 12:21:06

by Graeme Gregory

[permalink] [raw]
Subject: Re: [PATCH 3/3] serial: amba-pl011: add ACPI support to AMBA probe

On Tue, Dec 01, 2015 at 03:21:47AM +0100, Rafael J. Wysocki wrote:
> On Wednesday, September 30, 2015 11:38:50 AM Graeme Gregory wrote:
> > In ACPI this device is only defined in SBSA mode so
> > if we are probing from ACPI use this mode.
> >
> > Cc: Rafael J. Wysocki <[email protected]>
> > Cc: Len Brown <[email protected]>
> > Cc: Russell King <[email protected]>
> > Cc: Greg Kroah-Hartman <[email protected]>
> > Signed-off-by: Graeme Gregory <[email protected]>
> > ---
> > drivers/tty/serial/amba-pl011.c | 32 +++++++++++++++++++++-----------
> > 1 file changed, 21 insertions(+), 11 deletions(-)
> >
> > diff --git a/drivers/tty/serial/amba-pl011.c b/drivers/tty/serial/amba-pl011.c
> > index fd27e98..55209aa 100644
> > --- a/drivers/tty/serial/amba-pl011.c
> > +++ b/drivers/tty/serial/amba-pl011.c
> > @@ -2368,18 +2368,28 @@ static int pl011_probe(struct amba_device *dev, const struct amba_id *id)
> > if (!uap)
> > return -ENOMEM;
> >
> > - uap->clk = devm_clk_get(&dev->dev, NULL);
> > - if (IS_ERR(uap->clk))
> > - return PTR_ERR(uap->clk);
> > -
> > - uap->vendor = vendor;
> > - uap->lcrh_rx = vendor->lcrh_rx;
> > - uap->lcrh_tx = vendor->lcrh_tx;
> > - uap->fifosize = vendor->get_fifosize(dev);
> > + /* ACPI only defines SBSA variant */
> > + if (!ACPI_COMPANION(&dev->dev)) {
>
> It would read more straightforward if you did
>
> if (ACPI_COMPANION(&dev->dev)) {
> <handle the ACPI case>
> } else {
> <handle the non-ACPI case>
> }
>

This was something I debated about whether to put the ACPI case or the
common case first. I can certainly reverse it.

> > + uap->clk = devm_clk_get(&dev->dev, NULL);
> > + if (IS_ERR(uap->clk))
> > + return PTR_ERR(uap->clk);
> > +
> > + uap->vendor = vendor;
> > + uap->lcrh_rx = vendor->lcrh_rx;
> > + uap->lcrh_tx = vendor->lcrh_tx;
> > + uap->fifosize = vendor->get_fifosize(dev);
> > + uap->port.ops = &amba_pl011_pops;
> > + snprintf(uap->type, sizeof(uap->type), "PL011 rev%u",
> > + amba_rev(dev));
> > + } else {
> > + uap->vendor = &vendor_sbsa;
> > + uap->fifosize = 32;
> > + uap->port.ops = &sbsa_uart_pops;
> > + uap->fixed_baud = 115200;
> > +
> > + snprintf(uap->type, sizeof(uap->type), "SBSA");
>
> This looks sort of heavy-handed.
>
> Is this the only possible configuration of the device in the ACPI case?
>

As far as I can tell yes, but ARM haven't actually published a document
to state that as fact.

This does replace the platform_probe that Russel was unhappy about for
the ACPI case.

Graeme

2015-12-02 00:55:42

by Rafael J. Wysocki

[permalink] [raw]
Subject: Re: [PATCH 3/3] serial: amba-pl011: add ACPI support to AMBA probe

On Tuesday, December 01, 2015 12:21:00 PM Graeme Gregory wrote:
> On Tue, Dec 01, 2015 at 03:21:47AM +0100, Rafael J. Wysocki wrote:
> > On Wednesday, September 30, 2015 11:38:50 AM Graeme Gregory wrote:
> > > In ACPI this device is only defined in SBSA mode so
> > > if we are probing from ACPI use this mode.

[cut]

> > > + uap->clk = devm_clk_get(&dev->dev, NULL);
> > > + if (IS_ERR(uap->clk))
> > > + return PTR_ERR(uap->clk);
> > > +
> > > + uap->vendor = vendor;
> > > + uap->lcrh_rx = vendor->lcrh_rx;
> > > + uap->lcrh_tx = vendor->lcrh_tx;
> > > + uap->fifosize = vendor->get_fifosize(dev);
> > > + uap->port.ops = &amba_pl011_pops;
> > > + snprintf(uap->type, sizeof(uap->type), "PL011 rev%u",
> > > + amba_rev(dev));
> > > + } else {
> > > + uap->vendor = &vendor_sbsa;
> > > + uap->fifosize = 32;
> > > + uap->port.ops = &sbsa_uart_pops;
> > > + uap->fixed_baud = 115200;
> > > +
> > > + snprintf(uap->type, sizeof(uap->type), "SBSA");
> >
> > This looks sort of heavy-handed.
> >
> > Is this the only possible configuration of the device in the ACPI case?
> >
>
> As far as I can tell yes, but ARM haven't actually published a document
> to state that as fact.

At least a comment explaining what kind of information this is based on would
be useful. Otherwise one has to wonder where this is coming from.

> This does replace the platform_probe that Russel was unhappy about for
> the ACPI case.

I can't really comment on that.

Thanks,
Rafael