* Kishon Vijay Abraham I <kishon@xxxxxx> [150805 07:10]: > On Wednesday 05 August 2015 01:31 PM, Tony Lindgren wrote: > > > > We don't have syscon-otghs and to me it seems we need a PHY driver > > as I pointed out at: > > If *syscon-otghs* is not present, then it'll fall-back to using the *ctrl-module*. OK great. > > > > https://lkml.org/lkml/2015/6/24/231 > > Maybe I should have explained this in the previous thread. The *otghs* register > that we are trying to access here does _not_ belong to the PHY. It acts as > mailbox register from MUSB glue (TI integration layer) to MUSB core. That's why > it's programmed in the TI glue layer (omap2430.c). > > Even when we were using the older API [omap_control_usb_set_mode()], we first > call omap_musb_mailbox from the PHY drivers (phy-twl4030-usb.c, > phy-twl6030-usb.c) and then omap_musb_mailbox in the TI glue writes to the > control module instead of PHY drivers directly calling omap_control_usb_set_mode(). Hmm looking at "Table 18-204. CONTROL_USBOTGHS_CONTROL" it seems to mention "transceiver" for quite a few bitfields :) Probably what that register does is control a PHY over ULPI. So from Linux kernel point of view we're best off treating it as a PHY. It seems it should have a minimal PHY driver similar to what we have for dm816x control module in drivers/phy/phy-dm816x-usb.c. For reference, here is the register bitfields pasted from 4460 TRM: Table 18-204. CONTROL_USBOTGHS_CONTROL, p3972 Physical Address 0x4A00 233C BIT NAME DESCIPTION 8 DISCHRGVBUS ... OTG transceiver does (not) discharge VBUS ... 7 CHRGVBUS ... OTG transceiver does (not) charge VBUS ... 6 IDPULLUP ... OTG transceiver does (not) drive VBUS ... 4 IDDIG ... OTG transceiver does (not) apply a pullup to ID ... 3 SESSEND ... VBUS voltage is above/below VB_SESS_END ... 2 VBUSVALID ... VBUS is above the threshold ... 1 BVALID ... VBUS voltage is above/below VB_SESS_VLD ... 0 AVALID ... BUS voltage is above/below VA_SESS_VLD ... So how about just adding ONTROL_USBOTGHS_CONTROL support to the existing drivers/phy/phy-omap-usb2.c instead? It seems that it should allow us to completely get rid of the custom mailbox stuff for MUSB 2430 support? > > So let's sort that issue first. It also seems this just completely > > breaks the MUSB support? > > Why do you think so? If *syscon-otghs* is not present in dt, then it'll > fall-back to using the *ctrl-module* and everything should work seamlessly. OK that's good to hear. IMO drivers/phy/phy-omap4.c or similar should manage the syscon-otghs syscon register, not MUSB driver. Regards, Tony -- To unsubscribe from this list: send the line "unsubscribe devicetree" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
