RE: [PATCH] ARC: Add PCIe support for ARC HSDK platform

[Alexey Brodkin]

Hi Vineet,

> -----Original Message-----
> From: Vineet Gupta
> Sent: Monday, June 18, 2018 9:11 PM
> To: Alexey Brodkin <abrodkin@xxxxxxxxxxxx>; gustavo.pimentel@xxxxxxxxxxxx
> Cc: robh@xxxxxxxxxx; linux-kernel@xxxxxxxxxxxxxxx; Eugeniy.Paltsev@xxxxxxxxxxxx; sboyd@xxxxxxxxxxxxxx; linux-snps-
> arc@xxxxxxxxxxxxxxxxxxx
> Subject: Re: [PATCH] ARC: Add PCIe support for ARC HSDK platform
> 
> On 06/18/2018 08:20 AM, Alexey Brodkin wrote:
> >> +static void __init hsdk_enable_gpio_intc_wire(void)
> >> +{
> >> +	u32 val = GPIO_HAPS_INT;
> >> +
> >> +	iowrite32(0xffffffff, (void __iomem *) GPIO_INTMASK);
> >> +	iowrite32(~val, (void __iomem *) GPIO_INTMASK);
> >> +	iowrite32(0x00000000, (void __iomem *) GPIO_INTTYPE_LEVEL);
> >> +	iowrite32(0xffffffff, (void __iomem *) GPIO_INT_POLARITY);
> >> +	iowrite32(val, (void __iomem *) GPIO_INTEN);
> >> +}
> > I would suggest to have a map of really used lines and enable all of them
> > instead of adding one-by-one occasionally.
> 
> More importantly, adding any code in this file is an absolute abomination and only
> desired if this is a platform specific hack which can't be added in the generic
> driver and/or specified via the Device Tree. Here it seems like we are enabling
> some gpio lines which likely could be done via the gpio driver paths ?

See my comment much higher - what we're doing here is pretty much the same as
With AXS10x - we implement wires which connect input interrupt lines to IDU.
Thus I'm not sure if it worth messing with DW APB GPIO INTC driver which as I mentioned
As well won't work as it is and we'll need to patch it to support multiple outputs mode.

-Alexey