[U-Boot] [PATCH 3/4][v2]include:configs:ls1021atwr: Enable USB IP support

[Ramneek Mehresh]

> -----Original Message-----
> From: Marek Vasut [mailto:marex at denx.de]
> Sent: Friday, April 24, 2015 9:00 AM
> To: Mehresh Ramneek-B31383
> Cc: u-boot at lists.denx.de
> Subject: Re: [PATCH 3/4][v2]include:configs:ls1021atwr: Enable USB IP
> support
> 
> On Thursday, April 23, 2015 at 07:03:14 PM, Ramneek Mehresh wrote:
> > Enable USB IP support for both EHCI and XHCI for ls1021atwr platform
> >
> > Signed-off-by: Ramneek Mehresh <ramneek.mehresh at freescale.com>
> > ---
> >  include/configs/ls1021atwr.h | 36
> > ++++++++++++++++++++++++++++++++++++
> >  include/linux/usb/xhci-fsl.h |  5 +++++
> >  2 files changed, 41 insertions(+)
> >
> > diff --git a/include/configs/ls1021atwr.h
> > b/include/configs/ls1021atwr.h index a13876b..f208638 100644
> > --- a/include/configs/ls1021atwr.h
> > +++ b/include/configs/ls1021atwr.h
> > @@ -28,6 +28,42 @@
> >  #define CONFIG_SYS_INIT_RAM_SIZE	OCRAM_SIZE
> >
> >  /*
> > + * USB
> > + */
> > +
> > +/* EHCI Support - disbaled by default as
> > + * there is no signal coming out of soc on
> > + * this board for this controller. However,
> > + * the silicon still has this controller,
> > + * and anyone can use this controller by
> > + * taking signals out on their board.
> > + */
> 
> Multiline comment again ;-)
> 
> > +/*#define CONFIG_HAS_FSL_DR_USB*/
> > +
> > +#ifdef CONFIG_HAS_FSL_DR_USB
> > +#define CONFIG_USB_EHCI
> > +#define CONFIG_USB_EHCI_FSL
> > +#define CONFIG_EHCI_HCD_INIT_AFTER_RESET #endif
> > +
> > +/* XHCI Support - enabled by default */ #define
> > +CONFIG_HAS_FSL_XHCI_USB
> > +
> > +#ifdef CONFIG_HAS_FSL_XHCI_USB
> > +#define CONFIG_USB_XHCI_FSL
> > +#define CONFIG_USB_XHCI_DWC3
> > +#define CONFIG_USB_XHCI
> > +#define CONFIG_USB_MAX_CONTROLLER_COUNT        1
> > +#define CONFIG_SYS_USB_XHCI_MAX_ROOT_PORTS     2
> > +#endif
> > +
> > +#if defined(CONFIG_HAS_FSL_DR_USB) ||
> > +defined(CONFIG_HAS_FSL_XHCI_USB) #define CONFIG_CMD_USB
> #define
> > +CONFIG_USB_STORAGE #define CONFIG_CMD_EXT2 #endif
> > +
> > +/*
> >   * Generic Timer Definitions
> >   */
> >  #define GENERIC_TIMER_CLK		12500000
> > diff --git a/include/linux/usb/xhci-fsl.h
> > b/include/linux/usb/xhci-fsl.h index 8eaab2c..329abf7 100644
> > --- a/include/linux/usb/xhci-fsl.h
> > +++ b/include/linux/usb/xhci-fsl.h
> > @@ -46,6 +46,11 @@
> >  #define USBOTGSS_IRQ_SET_1_OEVT_EN		BIT(16)
> >  #define USBOTGSS_IRQ_SET_1_DMADISABLECLR_EN	BIT(17)
> >
> > +#ifdef CONFIG_LS102XA
> > +#define CONFIG_SYS_FSL_XHCI_USB1_ADDR
> > +CONFIG_SYS_LS102XA_XHCI_USB1_ADDR #define
> > +CONFIG_SYS_FSL_XHCI_USB2_ADDR 0
> 
> Do you need to define this _bogus_ address at all? If so, then the driver
> which depends on this is broken. Why can't you just define a list of addresses
> instead ? You would be able to use ARRAY_SIZE() in the driver to determine
> how many controllers there are then. This is how it would look like:
> 
> #define CONFIG_FOO_BAR_ADDRS { USB1_ADDR, USB2_ADDR, ...,
> USBn_ADDR }
> 
> In the driver, there'd be:
> 
> <type> addrs[] = CONFIG_FOO_BAR_ADDRS;
> 
I agree to use an array for defining list of controller addresses. However, the no.
of controller(s) to be initialized on a particular platform is determined by CONFIG_USB_MAX_CONTROLLER_COUNT
used in usb_init() function. This macro is defined in each platform file, and defines the index argument passed on to
xhci_hcd_init(). There may be some platform on which we can have more than one controller in soc, but only
one is used (exposed via external connector). Hence,  CONFIG_USB_MAX_CONTROLLER_COUNT is defined by
platform header file.
Hence, I can assign controller address on the basis of address:
struct fsl_xhci *ctx = &fsl_xhci;
ctx->hcd = addrs[index];

> > +#endif
> > +
> >  struct fsl_xhci {
> >  	struct xhci_hccr *hcd;
> >  	struct dwc3 *dwc3_reg;