> -----Original Message----- > From: Felipe Balbi [mailto:balbi@xxxxxx] > Sent: Tuesday, November 10, 2015 9:10 AM > To: John Youn <John.Youn@xxxxxxxxxxxx>; McCauley, Ben > <Ben.McCauley@xxxxxxxxxx>; John Youn <John.Youn@xxxxxxxxxxxx>; > gregkh@xxxxxxxxxxxxxxxxxxx > Cc: linux-usb@xxxxxxxxxxxxxxx; Schroeder, Jay <Jay.Schroeder@xxxxxxxxxx> > Subject: Re: [PATCH] USB: DCW3: GADGET: Set Correct Max Speed > > > Hi, > > John Youn <John.Youn@xxxxxxxxxxxx> writes: > > On 11/9/2015 1:09 PM, Felipe Balbi wrote: > >> > >> Hi, > >> > >> "McCauley, Ben" <Ben.McCauley@xxxxxxxxxx> writes: > >>>> "McCauley, Ben" <Ben.McCauley@xxxxxxxxxx> writes: > >>>>>>> Ben is talking about setting the gadget->max_speed, which is > >>>>>>> hard-coded in dwc3 to USB_SPEED_SUPER. > >>>>>>> > >>>>>>> For this core, the maximum supported speed may not actually be > >>>>>>> super speed. Many customers choose to use the 3.0 core in > >>>>>>> 2.0-only mode with a HS phy. > >>>>>> > >>>>>> yeah, I got all that. But we still have the problem of > >>>>>> metastability for anything <2.20a. For those cores, we just > >>>>>> cannot mess around with any of these settings (STAR 9000525659, if > you wanna check). > >>>>>> > >>>>>>> And with the 3.1 core, it my be integrated in a system that is > >>>>>>> HS-only, SS (gen1), or SSP (gen2) capable. So the > >>>>>>> gadget->max_speed needs to be set correctly based on some > >>>>>>> user-defined or detected value. > >>>>>>> > >>>>>>> We have a similar change in our local tree for 3.1, except we > >>>>>>> set it based on GHWPARAMS3.SSPHY_INTERFACE. But I think it > makes > >>>> sense > >>>>>>> to > >>>>>> set > >>>>>>> it from dwc->maximum_speed as well, for testing or to override it. > >>>>>> > >>>>>> My other fear is about people setting this wrongly and later > >>>>>> complaining that it's not working. I would very much prefer your > >>>>>> patch which detects whether SSPHY_INTERFACE is around, however > >>>>>> that > >>>> won't always be correct. > >>>>>> > >>>>>> I just noticed that in our USB2-only SoC (AM437x), > >>>>>> SSPHY_INTERFACE is set to > >>>>>> 1 even though this SoC does _not_ have any USB3 capabilities. > >>>>>> > >>>>>> GHWPARAMS3 = 0x10420085 > >>>>>> > >>>>>> I actually tested this problem as listed above and connected my > >>>>>> USB2-only SoC to a windows box (both USB2 and USB3 hosts) and > did > >>>>>> not see any such error messages. Any ideas why you have problems > >>>>>> an I don't > >>>> ? > >>>>> > >>>>> Using an actual 3.0 device I have been able to generate the > >>>>> following message "This device can perform faster", when > >>>>> connecting to one of the 2.0 ports on the same PC. It might have > >>>>> something to do with the organization of the USB hubs In my > >>>>> computer making windows think that a 3.0 device was connected to a > >>>>> 2.0 when the reality is swapped when connecting the gadget. > >>>> > >>>> if you're connecting a USB 3.0 device to a USB 2.0 port, windows is > >>>> doing the right thing. It's telling you "if you want to use the > >>>> full potential of this device, you should connect it to one of your > available USB 3.0 ports". > >>>> > >>>> Neither Windows nor dwc3 are wrong in this case. > >>>> > >>>>>> Maybe, all we have to do is avoid adding a SuperSpeed USB > >>>>>> Capability descriptor if we didn't negotiate for superspeed. All > >>>>>> the rest can remain the > >>>>>> same: > >>>>>> > >>>>>> diff --git a/drivers/usb/gadget/composite.c > >>>>>> b/drivers/usb/gadget/composite.c index > 8b14c2a13ac5..672ae6bfcad8 > >>>>>> 100644 > >>>>>> --- a/drivers/usb/gadget/composite.c > >>>>>> +++ b/drivers/usb/gadget/composite.c > >>>>>> @@ -547,9 +547,8 @@ static int count_configs(struct > >>>>>> usb_composite_dev *cdev, unsigned type) static int > >>>>>> bos_desc(struct usb_composite_dev > >>>>>> *cdev) { > >>>>>> struct usb_ext_cap_descriptor *usb_ext; > >>>>>> - struct usb_ss_cap_descriptor *ss_cap; > >>>>>> - struct usb_dcd_config_params dcd_config_params; > >>>>>> struct usb_bos_descriptor *bos = cdev->req->buf; > >>>>>> + struct usb_gadget *gadget = cdev->gadget; > >>>>>> > >>>>>> bos->bLength = USB_DT_BOS_SIZE; > >>>>>> bos->bDescriptorType = USB_DT_BOS; @@ -569,33 +568,38 @@ > >>>>>> static int bos_desc(struct usb_composite_dev > >>>>>> *cdev) > >>>>>> usb_ext->bDevCapabilityType = USB_CAP_TYPE_EXT; > >>>>>> usb_ext->bmAttributes = cpu_to_le32(USB_LPM_SUPPORT | > >>>>>> USB_BESL_SUPPORT); > >>>>>> > >>>>>> - /* > >>>>>> - * The Superspeed USB Capability descriptor shall be > implemented by > >>>>>> all > >>>>>> - * SuperSpeed devices. > >>>>>> - */ > >>>>>> - ss_cap = cdev->req->buf + le16_to_cpu(bos->wTotalLength); > >>>>>> - bos->bNumDeviceCaps++; > >>>>>> - le16_add_cpu(&bos->wTotalLength, > USB_DT_USB_SS_CAP_SIZE); > >>>>>> - ss_cap->bLength = USB_DT_USB_SS_CAP_SIZE; > >>>>>> - ss_cap->bDescriptorType = USB_DT_DEVICE_CAPABILITY; > >>>>>> - ss_cap->bDevCapabilityType = USB_SS_CAP_TYPE; > >>>>>> - ss_cap->bmAttributes = 0; /* LTM is not supported yet */ > >>>>>> - ss_cap->wSpeedSupported = > >>>>>> cpu_to_le16(USB_LOW_SPEED_OPERATION | > >>>>>> + if (gadget->speed >= USB_SPEED_SUPER) { > >>>>>> + struct usb_ss_cap_descriptor *ss_cap; > >>>>>> + struct usb_dcd_config_params dcd_config_params; > >>>>>> + > >>>>>> + /* > >>>>>> + * The Superspeed USB Capability descriptor shall > >>>>>> + be > >>>>>> implemented > >>>>>> + * by all SuperSpeed devices. > >>>>>> + */ > >>>>>> + ss_cap = cdev->req->buf + le16_to_cpu(bos- > >wTotalLength); > >>>>>> + bos->bNumDeviceCaps++; > >>>>>> + le16_add_cpu(&bos->wTotalLength, > >>>>>> USB_DT_USB_SS_CAP_SIZE); > >>>>>> + ss_cap->bLength = USB_DT_USB_SS_CAP_SIZE; > >>>>>> + ss_cap->bDescriptorType = USB_DT_DEVICE_CAPABILITY; > >>>>>> + ss_cap->bDevCapabilityType = USB_SS_CAP_TYPE; > >>>>>> + ss_cap->bmAttributes = 0; /* LTM is not supported yet */ > >>>>>> + ss_cap->wSpeedSupported = > >>>>>> cpu_to_le16(USB_LOW_SPEED_OPERATION | > >>>>>> USB_FULL_SPEED_OPERATION | > >>>>>> USB_HIGH_SPEED_OPERATION | > >>>>>> USB_5GBPS_OPERATION); > >>>>>> - ss_cap->bFunctionalitySupport = USB_LOW_SPEED_OPERATION; > >>>>>> + ss_cap->bFunctionalitySupport = > >>>>>> USB_LOW_SPEED_OPERATION; > >>>>>> > >>>>>> - /* Get Controller configuration */ > >>>>>> - if (cdev->gadget->ops->get_config_params) > >>>>>> - cdev->gadget->ops- > >>>>>>> get_config_params(&dcd_config_params); > >>>>>> - else { > >>>>>> + /* Get Controller configuration */ > >>>>>> dcd_config_params.bU1devExitLat = > >>>>>> USB_DEFAULT_U1_DEV_EXIT_LAT; > >>>>>> dcd_config_params.bU2DevExitLat = > >>>>>> cpu_to_le16(USB_DEFAULT_U2_DEV_EXIT_LAT); > >>>>>> + > >>>>>> + if (cdev->gadget->ops->get_config_params) > >>>>>> + cdev->gadget->ops- > >>>>>>> get_config_params(&dcd_config_params); > >>>>>> + > >>>>>> + ss_cap->bU1devExitLat = dcd_config_params.bU1devExitLat; > >>>>>> + ss_cap->bU2DevExitLat = > >>>>>> + dcd_config_params.bU2DevExitLat; > >>>>>> } > >>>>>> - ss_cap->bU1devExitLat = dcd_config_params.bU1devExitLat; > >>>>>> - ss_cap->bU2DevExitLat = dcd_config_params.bU2DevExitLat; > >>>>>> > >>>>>> return le16_to_cpu(bos->wTotalLength); } > >>>>>> > >>>>>> Can you see if this solves the issue you're having ? > >>>>>> > >>>>> > >>>>> I can confirm that this does resolve Windows reporting a that a > >>>>> 3.0 device was connected to a non 3.0 port. > >>>>> > >>>>> I am concerned that though this fixes the issue, it could result > >>>>> in > >>>>> 3.0 devices not reporting themselves as a 3.0 device when > >>>>> connected to a 2.0 port. > >>>> > >>>> right, I was thinking about the same thing after I sent this. But > >>>> can you really describe the problem you're having ? Is your device > >>>> SuperSpeed capable or not ? Are you connecting it to a Highspeed port > or Superspeed port ? > >>>> > >>> > >>> The device is not SuperSpeed capable and the issue comes up when > >>> connecting the device to a > >> > >> a couple emails back you stated: > >> > >> 'Using an actual 3.0 device I have been able to generate the > >> following message "This device can perform faster", when connecting > >> to one of the 2.0 ports on the same PC. It might have something to do > >> with the organization of the USB hubs In my computer making windows > >> think that a 3.0 device was connected to a 2.0 when the reality is swapped > when connecting the gadget.' > >> > >> and now you say again that the device is _not_ Superspeed capable. > >> Please make up your mind otherwise I really cannot help you. > >> > >>> SuperSpeed port. The crux of the issue is that I have a coworker > >>> laptop which when connected to the device the device receives a U2 > >>> command multiple times and after cycling through suspend and resume > >>> a couple times finalizes in the suspend state. So that reality is > >>> that there are two issues, > >> > >> what runs on this desktop ? Linux ? Windows ? Which version ? (Linux 4.0 > ? > >> Windows 7 ?). What's the model of that laptop (although this is less > important). > >> > >>> 1. PC reports 3.0 when the device does not support 3.0. This results > >>> in a confusing user experience and worth resolving. > >> > >> The real thing is that we still want to report our BOS descriptor, > >> just skip Superspeed capability descriptor if this instance of dwc3 > >> really can't do superspeed due to missing PIPE3 PHY or whatever. > >> However, we can't use negotiated speed for that because we still want > >> that warning for the case where the device _is_ superspeed capable > >> (i.e. dwc3 with superspeed PHY) but is connected to USB 2.0 port. > >> > >>> 2. Device initialized in suspend resulting in the g_hid module not > >>> functioning. The first patch fixed this as a side effect. This is > >>> the actual issue I am working to fix. This does not happen on all SS > >>> capable computers. I have only found the one model so far. > >> > >> Wich SoC are you using ? Based on the fact that you pointed me to > >> omapzoom.org, I'll assume you're using something from the DRA7x > >> family, am I right ? I'll think of a solution for the mainline > >> kernel, but I'll ask again that you get in touch with your FAE for support > with your v3.14 kernel. > >> > >> It seems like we need a new USB_SPEED_HIGH_LPM added so we have > means > >> to tell composite.c that we don't support Superspeed but support > highspeed _with_ LPM. > >> > >> Meanwhile, I guess the quickest "solution" _is_ your patch, but I > >> need a very large FIXME comment explaining that we would be better > >> off with a new USB_SPEED_HIGH_LPM because of the whole bcdUSB > 0x0210 > >> thing. Also, when adding the comment, I need a dev_info() message to > >> show up on all cores <2.20a which makes it clear that we're changing > >> max_speed, but we can't change DCFG programming because that would > cause the metastability error. > >> > >> regards > >> > > > > I think we need a way to set gadget->max_speed in any case. Especially > > with 3.1 where it might be HS, SS, or SSP. I think setting via the > > maximum_speed param seems the best option. Perhaps as an override in > > case the GHWPARAMS3 is wrong. > > right, this is what I tried to say :-) We just need the comment explaining the > situation and adding a FIXME note for the 2.1 thing. > > > If the PHY is restricted to HS-only then it shouldn't matter that > > DCFG.speed is programmed to SUPERSPEED. So the workaround for > > controllers < 2.20a can remain as-is. > > that's correct. > > > The LPM in high speed is a separate issue that exists even now. We > > address this issue by overriding bcdUSB to 0x0201 for our HS product, > > as specified in USB LPM ECN section 3. > > > > I think a possible solution to this is to use the > > gadget->get_config_params(). We can add a field in the > > usb_dcd_config_params struct to say whether the gadget is LPM capable. > > > > Then if gadget->max_speed == HIGH_SPEED, and the device is LPM > > capable, we can return bcdUSB = 0x201. > > > > Any thoughts on that? > > interesting, let me think about that. > > Ben, can you respin your patch to also add the comment explaining the > situation so we don't loose track of it ? Then I can apply once -rc1 is out. > > thanks > > -- > Balbi The max speed was always being set to USB_SUPER_SPEED even when the phy was only capable of HIGH_SPEED. This uses the value set for the phy to set the max for the gadget. It resolves an issue with Window PCs where they report that using a USB3.0 port would result in better performance even when connecting through a 3.0 port. A follow on patch will be needed to add USB_SPEED_HIGH_LPM to support High Speed devices which also support LPM. Signed-off-by: McCauley, Ben <ben.mccauley@xxxxxxxxxx> --- diff --git a/drivers/usb/dwc3/gadget.c b/drivers/usb/dwc3/gadget.c index 1eaf31c..342be2f 100644 --- a/drivers/usb/dwc3/gadget.c +++ b/drivers/usb/dwc3/gadget.c @@ -2899,7 +2899,15 @@ } dwc->gadget.ops = &dwc3_gadget_ops; - dwc->gadget.max_speed = USB_SPEED_SUPER; + /* + * FIXME A new enumeration USB_SPEED_HIGH_LPM needs to be added + * to support USB 2.10. Setting max_speed to dwc->maximum_speed + * disables LPM support for devices which do not support + * SuperSpeed but are LPM capable. + */ + if (dwc->maximum_speed < USB_SPEED_SUPER) + dev_info(dwc->dev, "SuperSpeed not supported.\n"); + dwc->gadget.max_speed = dwc->maximum_speed; dwc->gadget.speed = USB_SPEED_UNKNOWN; dwc->gadget.sg_supported = true; dwc->gadget.name = "dwc3-gadget" ________________________________ CONFIDENTIALITY NOTICE: This email and any attachments are for the sole use of the intended recipient(s) and contain information that may be confidential and/or legally privileged. If you have received this email in error, please notify the sender by reply email and delete the message. Any disclosure, copying, distribution or use of this communication (including attachments) by someone other than the intended recipient is prohibited. Thank you. -- To unsubscribe from this list: send the line "unsubscribe linux-usb" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
