On Thu, May 11, 2023, Krishna Kurapati PSSNV wrote: > > > On 5/11/2023 7:19 AM, Thinh Nguyen wrote: > > On Thu, May 11, 2023, Krishna Kurapati PSSNV wrote: > > > > > > > > > On 5/11/2023 2:59 AM, Thinh Nguyen wrote: > > > > Hi, > > > > > > > > On Wed, May 10, 2023, Krishna Kurapati PSSNV wrote: > > > > > Hi Team, > > > > > > > > > > If we set maximum speed to "super-speed" for a SSP capable controller, I > > > > > see device mode working in SS but host mode still works in SSP. From the > > > > > yaml documentation of max-speed, I see the following: > > > > > > > > > > maximum-speed: > > > > > description: > > > > > Tells USB controllers we want to work up to a certain speed. Incase > > > > > this isn't passed via DT, USB controllers should default to their maximum HW > > > > > capability. > > > > > > > > > > Shouldn't this maximum speed get limited to gen1 even for host mode ? > > > > > > > > The description is ambiguous here. However, we only ever use this > > > > property to limit the speed for device mode (from what I see in the > > > > other drivers of the kernel). Should we ever need to limit this for host > > > > mode, IMHO, we should introduce a new property for that and clarify this > > > > property. > > > > > > > Hi Thinh, > > > > > > Sorry for not being clear. The following is the version and ID of SM8550 > > > controller: (usb31 controller) > > > > > > VER_NUM: 0x3139302A > > > VER_TYPE: 0x67612A2A > > > USB3_PRI_USB31_IP_NAME (0xA60C120) = 0x33313130 > > > > > > We need to force this to gen-1 to fallback to SS even for host mode too as > > > customers are facing compliance issues on SSP. > > > > > > > I meant the documentation for maximum-speed property isn't clear for > > cases such as DRD usb controller where the controller can operate in > > different modes. Your description is pretty clear. > > > > > > > > > > > > If so, can we set LLUCTL Force_Gen1 bit if (Controller is Gen2 && max_speed > > > > > = SuperSpeed) somewhere after GCTL core soft reset is done in set_mode ? > > > > > > > > BTW, that's for dwc_usb31. For dwc_usb32, we use a different register to > > > > limit different eSS speeds and lanecount. > > > > > > > > To limit to highspeed for xhci, we can disable usb3 ports (clear the > > > > PORTSC.PP). I don't think we can force limit to full or low speed for > > > > dwc3 host mode. > > > > > > > > > > From your explanation, can I send a patch for with the following: > > > > > > 1. Update description for max speed > > > 2. Add force_gen1 property and if it is set, check the version (usb31) and > > > set LLUCTL.Force_Gen1 bit. > > > > > > Later if someone wants to add code for usb32 as well, they can reuse the > > > same property and extend the code for the same. > > > > > > Let me know if this would be acceptable. > > > > > > > I'm thinking to introduce a generic "host-maximum-speed" property > > instead of a "force_gen1" that's specific to dwc3 controller only. > > > > So that we can update the documentation for "maximum-speed" as follow: > > > > * The "maximum-speed" applies to both host and device modes. If > > "host-maximum-speed" is specified, then the host may have a different > > constraint. > > > > Or > > > > * The "maximum-speed" only applies for device mode. The > > "host-maximum-speed" is for host mode. > > > > Thanks, > > Thinh > > Hi Thinh, > > If we do so, we need to update the code for supporting other speeds as > well (changes in xhci would be needed I believe). Wanted to limit this to > dwc3 only and so was suggesting to add force_gen1 property. > I agree that it may get unnecessarily complicated. How about let's assume "maximum-speed" apply for both modes. Should there be a case where we need to specify different speeds for different mode, then we can introduce a new property. Thanks, Thinh
