[PATCHv2] serial: 8250: Add Quark X1000 to 8250 PCI driver

[PATCHv2] serial: 8250: Add Quark X1000 to 8250 PCI driver

[Bryan O'Donoghue]

Adds PCI identifier for the X1000
Adds clocking and register size and register shift

Signed-off-by: Bryan O'Donoghue <pure.logic@nexus-software.ie>
---
 drivers/tty/serial/8250/8250_pci.c | 33 +++++++++++++++++++++++++++++++++
 1 file changed, 33 insertions(+)

diff --git a/drivers/tty/serial/8250/8250_pci.c b/drivers/tty/serial/8250/8250_pci.c
index 61830b1..5e72959 100644
--- a/drivers/tty/serial/8250/8250_pci.c
+++ b/drivers/tty/serial/8250/8250_pci.c
@@ -1481,6 +1481,19 @@ byt_serial_setup(struct serial_private *priv,
 	return ret;
 }
 
+#define PCI_DEVICE_ID_INTEL_QRK_UART	0x0936
+
+static int
+pci_intel_qrk_setup(struct serial_private *priv,
+		const struct pciserial_board *board,
+		struct uart_8250_port *port, int idx)
+{
+	unsigned int bar, offset = board->first_offset;
+
+	bar = FL_GET_BASE(board->flags);
+
+	return setup_port(priv, port, bar, offset, board->reg_shift);
+}
 static int
 pci_omegapci_setup(struct serial_private *priv,
 		      const struct pciserial_board *board,
@@ -1898,6 +1911,13 @@ static struct pci_serial_quirk pci_serial_quirks[] __refdata = {
 		.subdevice	= PCI_ANY_ID,
 		.setup		= byt_serial_setup,
 	},
+	{
+		.vendor		= PCI_VENDOR_ID_INTEL,
+		.device		= PCI_DEVICE_ID_INTEL_QRK_UART,
+		.subvendor	= PCI_ANY_ID,
+		.subdevice	= PCI_ANY_ID,
+		.setup		= pci_intel_qrk_setup,
+	},
 	/*
 	 * ITE
 	 */
@@ -2740,6 +2760,7 @@ enum pci_board_num_t {
 	pbn_ADDIDATA_PCIe_8_3906250,
 	pbn_ce4100_1_115200,
 	pbn_byt,
+	pbn_qrk,
 	pbn_omegapci,
 	pbn_NETMOS9900_2s_115200,
 	pbn_brcm_trumanage,
@@ -3490,6 +3511,12 @@ static struct pciserial_board pci_boards[] = {
 		.uart_offset	= 0x80,
 		.reg_shift      = 2,
 	},
+	[pbn_qrk] = {
+		.flags		= FL_BASE0,
+		.num_ports	= 1,
+		.base_baud	= 2764800,
+		.reg_shift	= 2,
+	},
 	[pbn_omegapci] = {
 		.flags		= FL_BASE0,
 		.num_ports	= 8,
@@ -5192,6 +5219,12 @@ static struct pci_device_id serial_pci_tbl[] = {
 		pbn_byt },
 
 	/*
+	 * Intel Quark x1000
+	 */
+	{	PCI_VENDOR_ID_INTEL, PCI_DEVICE_ID_INTEL_QRK_UART,
+		PCI_ANY_ID, PCI_ANY_ID, 0, 0,
+		pbn_qrk },
+	/*
 	 * Cronyx Omega PCI
 	 */
 	{	PCI_VENDOR_ID_PLX, PCI_DEVICE_ID_PLX_CRONYX_OMEGA,
-- 
1.9.1

Re: [PATCHv2] serial: 8250: Add Quark X1000 to 8250 PCI driver

[Heikki Krogerus]

On Fri, Sep 19, 2014 at 08:43:49AM +0100, Bryan O'Donoghue wrote:
> Adds PCI identifier for the X1000
> Adds clocking and register size and register shift
> 
> Signed-off-by: Bryan O'Donoghue <pure.logic@nexus-software.ie>
> ---
>  drivers/tty/serial/8250/8250_pci.c | 33 +++++++++++++++++++++++++++++++++
>  1 file changed, 33 insertions(+)

The UART on Quark SoC has an integrated DMA engine and Kean Ho (CC'd)
is working with support for it. The plan is to add completely separate
probe driver for the Quark UART. Dealing with the integrated DMA
engine is too much for a quirk in 8250_pci.c.

So we can take this now and when Kean Ho is ready he needs to
basically revert it, or you could already introduce the new probe
driver, 8250_quark.c, and Kean Ho can then add the DMA engine handling
later to it.

IMO we should add the 8250_quark.c already now.


> diff --git a/drivers/tty/serial/8250/8250_pci.c b/drivers/tty/serial/8250/8250_pci.c
> index 61830b1..5e72959 100644
> --- a/drivers/tty/serial/8250/8250_pci.c
> +++ b/drivers/tty/serial/8250/8250_pci.c
> @@ -1481,6 +1481,19 @@ byt_serial_setup(struct serial_private *priv,
>  	return ret;
>  }
>  
> +#define PCI_DEVICE_ID_INTEL_QRK_UART	0x0936
> +
> +static int
> +pci_intel_qrk_setup(struct serial_private *priv,
> +		const struct pciserial_board *board,
> +		struct uart_8250_port *port, int idx)
> +{
> +	unsigned int bar, offset = board->first_offset;
> +
> +	bar = FL_GET_BASE(board->flags);
> +
> +	return setup_port(priv, port, bar, offset, board->reg_shift);
> +}
>  static int
>  pci_omegapci_setup(struct serial_private *priv,
>  		      const struct pciserial_board *board,
> @@ -1898,6 +1911,13 @@ static struct pci_serial_quirk pci_serial_quirks[] __refdata = {
>  		.subdevice	= PCI_ANY_ID,
>  		.setup		= byt_serial_setup,
>  	},
> +	{
> +		.vendor		= PCI_VENDOR_ID_INTEL,
> +		.device		= PCI_DEVICE_ID_INTEL_QRK_UART,
> +		.subvendor	= PCI_ANY_ID,
> +		.subdevice	= PCI_ANY_ID,
> +		.setup		= pci_intel_qrk_setup,
> +	},
>  	/*
>  	 * ITE
>  	 */
> @@ -2740,6 +2760,7 @@ enum pci_board_num_t {
>  	pbn_ADDIDATA_PCIe_8_3906250,
>  	pbn_ce4100_1_115200,
>  	pbn_byt,
> +	pbn_qrk,
>  	pbn_omegapci,
>  	pbn_NETMOS9900_2s_115200,
>  	pbn_brcm_trumanage,
> @@ -3490,6 +3511,12 @@ static struct pciserial_board pci_boards[] = {
>  		.uart_offset	= 0x80,
>  		.reg_shift      = 2,
>  	},
> +	[pbn_qrk] = {
> +		.flags		= FL_BASE0,
> +		.num_ports	= 1,
> +		.base_baud	= 2764800,
> +		.reg_shift	= 2,
> +	},
>  	[pbn_omegapci] = {
>  		.flags		= FL_BASE0,
>  		.num_ports	= 8,
> @@ -5192,6 +5219,12 @@ static struct pci_device_id serial_pci_tbl[] = {
>  		pbn_byt },
>  
>  	/*
> +	 * Intel Quark x1000
> +	 */
> +	{	PCI_VENDOR_ID_INTEL, PCI_DEVICE_ID_INTEL_QRK_UART,
> +		PCI_ANY_ID, PCI_ANY_ID, 0, 0,
> +		pbn_qrk },
> +	/*
>  	 * Cronyx Omega PCI
>  	 */
>  	{	PCI_VENDOR_ID_PLX, PCI_DEVICE_ID_PLX_CRONYX_OMEGA,
> -- 
> 1.9.1

Cheers,

-- 
heikki

Re: [PATCHv2] serial: 8250: Add Quark X1000 to 8250 PCI driver

[Andy Shevchenko]

On Fri, Sep 19, 2014 at 1:58 PM, Heikki Krogerus
<heikki.krogerus@linux.intel.com> wrote:
> On Fri, Sep 19, 2014 at 08:43:49AM +0100, Bryan O'Donoghue wrote:
>> Adds PCI identifier for the X1000
>> Adds clocking and register size and register shift
>>
>> Signed-off-by: Bryan O'Donoghue <pure.logic@nexus-software.ie>
>> ---
>>  drivers/tty/serial/8250/8250_pci.c | 33 +++++++++++++++++++++++++++++++++
>>  1 file changed, 33 insertions(+)
>
> The UART on Quark SoC has an integrated DMA engine and Kean Ho (CC'd)
> is working with support for it. The plan is to add completely separate
> probe driver for the Quark UART. Dealing with the integrated DMA
> engine is too much for a quirk in 8250_pci.c.
>
> So we can take this now and when Kean Ho is ready he needs to
> basically revert it, or you could already introduce the new probe
> driver, 8250_quark.c, and Kean Ho can then add the DMA engine handling
> later to it.
>
> IMO we should add the 8250_quark.c already now.

I'm working on providing dw_dma_probe()/_remove() for the kernel
users. Besides that the serial setup should look like a BYT version
(of course, the device would be different which does DMA). I would try
not to forget to Cc'ed you when I send patch series to the dmaengine
mailing list.

>
>
>> diff --git a/drivers/tty/serial/8250/8250_pci.c b/drivers/tty/serial/8250/8250_pci.c
>> index 61830b1..5e72959 100644
>> --- a/drivers/tty/serial/8250/8250_pci.c
>> +++ b/drivers/tty/serial/8250/8250_pci.c
>> @@ -1481,6 +1481,19 @@ byt_serial_setup(struct serial_private *priv,
>>       return ret;
>>  }
>>
>> +#define PCI_DEVICE_ID_INTEL_QRK_UART 0x0936
>> +
>> +static int
>> +pci_intel_qrk_setup(struct serial_private *priv,
>> +             const struct pciserial_board *board,
>> +             struct uart_8250_port *port, int idx)
>> +{
>> +     unsigned int bar, offset = board->first_offset;
>> +
>> +     bar = FL_GET_BASE(board->flags);
>> +
>> +     return setup_port(priv, port, bar, offset, board->reg_shift);
>> +}
>>  static int
>>  pci_omegapci_setup(struct serial_private *priv,
>>                     const struct pciserial_board *board,
>> @@ -1898,6 +1911,13 @@ static struct pci_serial_quirk pci_serial_quirks[] __refdata = {
>>               .subdevice      = PCI_ANY_ID,
>>               .setup          = byt_serial_setup,
>>       },
>> +     {
>> +             .vendor         = PCI_VENDOR_ID_INTEL,
>> +             .device         = PCI_DEVICE_ID_INTEL_QRK_UART,
>> +             .subvendor      = PCI_ANY_ID,
>> +             .subdevice      = PCI_ANY_ID,
>> +             .setup          = pci_intel_qrk_setup,
>> +     },
>>       /*
>>        * ITE
>>        */
>> @@ -2740,6 +2760,7 @@ enum pci_board_num_t {
>>       pbn_ADDIDATA_PCIe_8_3906250,
>>       pbn_ce4100_1_115200,
>>       pbn_byt,
>> +     pbn_qrk,
>>       pbn_omegapci,
>>       pbn_NETMOS9900_2s_115200,
>>       pbn_brcm_trumanage,
>> @@ -3490,6 +3511,12 @@ static struct pciserial_board pci_boards[] = {
>>               .uart_offset    = 0x80,
>>               .reg_shift      = 2,
>>       },
>> +     [pbn_qrk] = {
>> +             .flags          = FL_BASE0,
>> +             .num_ports      = 1,
>> +             .base_baud      = 2764800,
>> +             .reg_shift      = 2,
>> +     },
>>       [pbn_omegapci] = {
>>               .flags          = FL_BASE0,
>>               .num_ports      = 8,
>> @@ -5192,6 +5219,12 @@ static struct pci_device_id serial_pci_tbl[] = {
>>               pbn_byt },
>>
>>       /*
>> +      * Intel Quark x1000
>> +      */
>> +     {       PCI_VENDOR_ID_INTEL, PCI_DEVICE_ID_INTEL_QRK_UART,
>> +             PCI_ANY_ID, PCI_ANY_ID, 0, 0,
>> +             pbn_qrk },
>> +     /*
>>        * Cronyx Omega PCI
>>        */
>>       {       PCI_VENDOR_ID_PLX, PCI_DEVICE_ID_PLX_CRONYX_OMEGA,
>> --
>> 1.9.1
>
> Cheers,
>
> --
> heikki
> --
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at  http://www.tux.org/lkml/



-- 
With Best Regards,
Andy Shevchenko

Re: [PATCHv2] serial: 8250: Add Quark X1000 to 8250 PCI driver

[Bryan O'Donoghue]

On 19/09/14 11:58, Heikki Krogerus wrote:
> On Fri, Sep 19, 2014 at 08:43:49AM +0100, Bryan O'Donoghue wrote:
>> Adds PCI identifier for the X1000
>> Adds clocking and register size and register shift
>>
>> Signed-off-by: Bryan O'Donoghue <pure.logic@nexus-software.ie>
>> ---
>>   drivers/tty/serial/8250/8250_pci.c | 33 +++++++++++++++++++++++++++++++++
>>   1 file changed, 33 insertions(+)
>
> The UART on Quark SoC has an integrated DMA engine and Kean Ho (CC'd)
> is working with support for it. The plan is to add completely separate
> probe driver for the Quark UART. Dealing with the integrated DMA
> engine is too much for a quirk in 8250_pci.c.
>
> So we can take this now and when Kean Ho is ready he needs to
> basically revert it, or you could already introduce the new probe
> driver, 8250_quark.c, and Kean Ho can then add the DMA engine handling
> later to it.
>
> IMO we should add the 8250_quark.c already now.

Sounds great - is the 8250_quark.c code ready ?

If so then fantastic I agree with you.

If not then, lets do as you suggest Heikki and apply the simple patch 
above in the interim.


Best,
BOD

RE: [PATCHv2] serial: 8250: Add Quark X1000 to 8250 PCI driver

[Chew, Kean Ho]

[-- Attachment #1: Type: text/plain, Size: 473 bytes --]

> > So we can take this now and when Kean Ho is ready he needs to
> > basically revert it, or you could already introduce the new probe
> > driver, 8250_quark.c, and Kean Ho can then add the DMA engine handling
> > later to it.
> >
> > IMO we should add the 8250_quark.c already now.
> Sounds great - is the 8250_quark.c code ready ? 

I'm still working on the DMA portion and requires more time.
If the DMA section can be replaced by a placeholder, I should be good.

-KH

[-- Attachment #2: smime.p7s --]
[-- Type: application/pkcs7-signature, Size: 6697 bytes --]

Re: [PATCHv2] serial: 8250: Add Quark X1000 to 8250 PCI driver

[Bryan O'Donoghue]

On 19/09/14 13:06, Chew, Kean Ho wrote:
>>> So we can take this now and when Kean Ho is ready he needs to
>>> basically revert it, or you could already introduce the new probe
>>> driver, 8250_quark.c, and Kean Ho can then add the DMA engine handling
>>> later to it.
>>>
>>> IMO we should add the 8250_quark.c already now.
>> Sounds great - is the 8250_quark.c code ready ?
>
> I'm still working on the DMA portion and requires more time.
> If the DMA section can be replaced by a placeholder, I should be good.

OK so.

Can you submit that soon, or should the simple PCI enabling code to be 
added ?


--
BOD

RE: [PATCHv2] serial: 8250: Add Quark X1000 to 8250 PCI driver

[Chew, Kean Ho]

> Can you submit that soon
Will do.

Re: [PATCHv2] serial: 8250: Add Quark X1000 to 8250 PCI driver

[Bryan O'Donoghue]

On 19/09/14 14:10, Chew, Kean Ho wrote:
>> Can you submit that soon
> Will do.
>
Hi Ho.

Do you have an idea when ?

I guess if we're saying the simple patch which works right now - should 
be rejected in favour of alternative code coming - then at a minimum 
could you please give an indication of when that alternative code will 
be available ?

Otherwise I think it makes sense to enable the hardware now - using the 
bog-standard means of doing that - just like BYT and others :)

-- 
BOD

RE: [PATCHv2] serial: 8250: Add Quark X1000 to 8250 PCI driver

[Chew, Kean Ho]

> Hi Ho.
> 
> Do you have an idea when ?
> 
> I guess if we're saying the simple patch which works right now - should be
> rejected in favour of alternative code coming - then at a minimum could you
> please give an indication of when that alternative code will be available ?
> 
> Otherwise I think it makes sense to enable the hardware now - using the
> bog-standard means of doing that - just like BYT and others :)

Make sense. I go with Hekki's direction. 

Re: [PATCHv2] serial: 8250: Add Quark X1000 to 8250 PCI driver

[Bryan O'Donoghue]

On 19/09/14 15:31, Chew, Kean Ho wrote:
>> Hi Ho.
>>
>> Do you have an idea when ?
>>
>> I guess if we're saying the simple patch which works right now - should be
>> rejected in favour of alternative code coming - then at a minimum could you
>> please give an indication of when that alternative code will be available ?
>>
>> Otherwise I think it makes sense to enable the hardware now - using the
>> bog-standard means of doing that - just like BYT and others :)
>
> Make sense. I go with Hekki's direction.
>

Great.

So Hekki, are OK with the suggested path here then ? I think myself and 
Ho are pretty much in agreement.

Apply the patch I've submitted to 8250_pci.c - and then do whatever mods 
are appropriate later to enable whatever you're doing quark_8250.c.

I think, it'd be a win from a Galileo enabling point of view.


Best,
Bryan

Re: [PATCHv2] serial: 8250: Add Quark X1000 to 8250 PCI driver

[Heikki Krogerus]

On Fri, Sep 19, 2014 at 08:43:49AM +0100, Bryan O'Donoghue wrote:
> Adds PCI identifier for the X1000
> Adds clocking and register size and register shift

I would be more interested in the details about the Quart UART, like
how many ports you got, maybe the clock frequency etc. then about the
actual steps the you are taking in the patch.


> Signed-off-by: Bryan O'Donoghue <pure.logic@nexus-software.ie>
> ---
>  drivers/tty/serial/8250/8250_pci.c | 33 +++++++++++++++++++++++++++++++++
>  1 file changed, 33 insertions(+)
> 
> diff --git a/drivers/tty/serial/8250/8250_pci.c b/drivers/tty/serial/8250/8250_pci.c
> index 61830b1..5e72959 100644
> --- a/drivers/tty/serial/8250/8250_pci.c
> +++ b/drivers/tty/serial/8250/8250_pci.c
> @@ -1481,6 +1481,19 @@ byt_serial_setup(struct serial_private *priv,
>  	return ret;
>  }
>  
> +#define PCI_DEVICE_ID_INTEL_QRK_UART	0x0936
> +
> +static int
> +pci_intel_qrk_setup(struct serial_private *priv,
> +		const struct pciserial_board *board,
> +		struct uart_8250_port *port, int idx)
> +{
> +	unsigned int bar, offset = board->first_offset;
> +
> +	bar = FL_GET_BASE(board->flags);
> +
> +	return setup_port(priv, port, bar, offset, board->reg_shift);
> +}

You don't need to add this. Just use pci_default_setup.


Thanks,

-- 
heikki