Re: [Xenomai-core] [PATCH V2] rtcan: add rtcan_plx_pci driver

["P.B.Cheblakov" <P.B.Cheblakov@domain.hid>]

Wolfgang Grandegger wrote:
> On 05/31/2010 09:43 AM, Gilles Chanteperdrix wrote:
>   
>> Wolfgang Grandegger wrote:
>>     
>>> On 05/30/2010 03:12 PM, Gilles Chanteperdrix wrote:
>>>       
>>>> Wolfgang Grandegger wrote:
>>>>         
>>>>> From my point of view, yes. Gilles could you please commit this patch?
>>>>>           
>>>> No, it seems the patch causes a build issue:
>>>>
>>>> http://sisyphus.hd.free.fr/~gilles/bx/
>>>>         
>>> Oops, obviously the Kconfig entry is invalid:
>>>
>>> +config XENO_DRIVERS_CAN_SJA1000_ESD_PCI (DEPRECATED)
>>>
>>> The "(DEPRECATED)" is causing the trouble, I assume.
>>>
>>>  	depends on XENO_DRIVERS_CAN_SJA1000 && PCI
>>>  	tristate "ESD PCI Cards"
>>>
>>> Maybe you want to add the "(DEPRECATED)" here.
>>>
>>>  	help
>>> @@ -61,6 +76,9 @@ config XENO_DRIVERS_CAN_SJA1000_ESD_PCI
>>>  	CAN-PCI/266, CAN-PMC/266 (PMC), CAN-CPCI/200 (CompactPCI),
>>>  	CAN-PCIe2000 (PCI Express) and CAN-PCI104/200 (PCI104)
>>>  	from the esd electronic system design gmbh (http://www.esd.eu).
>>> +	
>>>
>>> Trailing white space!
>>>
>>> +	This driver is deprecated. It's functionality is now provided by
>>> +	"PLX90xx PCI-bridge based Cards" driver.
>>>
>>> Pavel, could you please fix that.
>>>       
>> Ok. If these are the only differences between the patch which was
>> actually tested and the patch you posted, simply tell us, I can fix it
>> in my tree.
>>     
>
> The attached patch fixes the remaining issues, including some compiler
> warnings :-(.
>
> Thanks,
>
> Wolfgang.
>
> ---
>  ksrc/drivers/can/sja1000/Kconfig         |   10 ++++------
>  ksrc/drivers/can/sja1000/rtcan_plx_pci.c |    8 ++++----
>  2 files changed, 8 insertions(+), 10 deletions(-)
>   
[snip]
> Index: xenomai-2.5/ksrc/drivers/can/sja1000/rtcan_plx_pci.c
> ===================================================================
> --- xenomai-2.5.orig/ksrc/drivers/can/sja1000/rtcan_plx_pci.c
> +++ xenomai-2.5/ksrc/drivers/can/sja1000/rtcan_plx_pci.c
> @@ -288,12 +288,12 @@ static DEFINE_PCI_DEVICE_TABLE(plx_pci_t
>  };
>  MODULE_DEVICE_TABLE(pci, plx_pci_tbl);
>  
> -static u8 plx_pci_read_reg(const struct rtcan_device *dev, int port)
> +static u8 plx_pci_read_reg(struct rtcan_device *dev, int port)
>  {
>  	return ioread8((void* __iomem)dev->base_addr + port);
>  }
>  
> -static void plx_pci_write_reg(const struct rtcan_device *dev, int port, u8 val)
> +static void plx_pci_write_reg(struct rtcan_device *dev, int port, u8 val)
>  {
>  	iowrite8(val, (void* __iomem)dev->base_addr + port);
>  }
> @@ -303,7 +303,7 @@ static void plx_pci_write_reg(const stru
>   * by trying to switch 'em from the Basic mode into the PeliCAN mode.
>   * Also check states of some registers in reset mode.
>   */
> -static inline int plx_pci_check_sja1000(const struct rtcan_device *dev)
> +static inline int plx_pci_check_sja1000(struct rtcan_device *dev)
>  {
>  	int flag = 0;
>  
> @@ -507,7 +507,7 @@ static int __devinit plx_pci_add_card(st
>  		}
>  
>  		strncpy(dev->name, RTCAN_DEV_NAME, IFNAMSIZ);
> -		dev->board_name = ci->name;
> +		dev->board_name = (char *)ci->name;
>  
>  		card->rtcan_dev[i] = dev;
>  		chip = card->rtcan_dev[i]->priv;
>   

Wolfgang, I purposely emphasize diferences between Socketcan API and 
RT-CAN API. I think, that API in Socketcan is slightly better (e.g. 
'const' in function declaration).
I'm sorry that I haven't said it previously.

Pavel