Hi Shimoda-San, Thanks for the feedback. > Subject: RE: [PATCH 04/12] usb: host: ehci-r8a77470: Add EHCI support for > Renesas RZ/G1C devices > > Hi Biju-san, > > Thank you for the patch! > > > From: Biju Das, Sent: Monday, April 1, 2019 9:54 PM > <snip> > > +/******* USB2.0 Host registers (original offset is +0x100) *******/ > > +#define USB2_INT_ENABLE 0x100 > > +#define USB2_SPD_RSM_TIMSET 0x20c > > +#define USB2_OC_TIMSET 0x210 > > + > > +#define USB2_INT_ENABLE_USBH_INTB_EN BIT(2) > > +#define USB2_INT_ENABLE_USBH_INTA_EN BIT(1) > > +#define USB2_INT_ENABLE_INIT > (USB2_INT_ENABLE_USBH_INTB_EN | \ > > + USB2_INT_ENABLE_USBH_INTA_EN) > > + > > +#define USB2_SPD_RSM_TIMSET_INIT 0x014e029b > > +#define USB2_OC_TIMSET_INIT 0x000209ab > > I should have notice this codes at the previous patch though, I'm thinking we > can reuse the R-Car Gen3 codes because these values are similar with the > phy-rcar-gen3-usb2.c like below: > ---------------------------------------------------------------------- > /* INT_ENABLE */ > #define USB2_INT_ENABLE_UCOM_INTEN BIT(3) > #define USB2_INT_ENABLE_USBH_INTB_EN BIT(2) > #define USB2_INT_ENABLE_USBH_INTA_EN BIT(1) > #define USB2_INT_ENABLE_INIT > (USB2_INT_ENABLE_UCOM_INTEN | \ > USB2_INT_ENABLE_USBH_INTB_EN > | \ > USB2_INT_ENABLE_USBH_INTA_EN) > <snip> > /* SPD_RSM_TIMSET */ > #define USB2_SPD_RSM_TIMSET_INIT 0x014e029b > > /* OC_TIMSET */ > #define USB2_OC_TIMSET_INIT 0x000209ab > ---------------------------------------------------------------------- > > If so, we can use ehci-platform.c and ohci-platform.c as-is. In this case, the > "phy: renesas: phy-rcar-gen2: Add support for r8a77470" driver should be > initialized at first to release PLLRESET. And then, the phy-rcar-gen3-usb2 > driver would be initialized as second step. I think we can set these resisters at > phy_init timing because the following patch did so: > > https://patchwork.kernel.org/patch/10655855/ > > What do you think? Good catch. This solution will avoid overlapping regions as mentioned by Rob in the below patch and also will avoid modifying ehci-platform.c as suggested by Alen. https://patchwork.kernel.org/patch/10655853/ I will send V3 based on this solution. Please let me know, if you think otherwise. Regards, Biju
