Hi, On 4/16/2020 1:46 PM, Marek Vasut wrote: > On 4/15/20 5:10 PM, Minas Harutyunyan wrote: >> Hi, > > Hi, > >> On 4/15/2020 6:55 PM, Marek Vasut wrote: >>> On 4/15/20 10:37 AM, Minas Harutyunyan wrote: >>>> Hi, >>>> >>>> On 4/14/2020 4:18 PM, Marek Vasut wrote: >>>>> On 4/14/20 9:14 AM, Minas Harutyunyan wrote: >>>> >>>>> >>>>>> 2. debug log >>>>> >>>>> What kind of debug log ? >>>>> >>>> Drivers debug log starting from dwc2 loading and connector connecting to >>>> device, where issue seen. >>> >>> How do I enable the one you need ? >>> >> To enable debug prints from dwc2 use 'make menuconfig': >> >> Symbol: USB_DWC2_DEBUG [=y] > > Oh, this one, OK. There you go. > > Note that there is no further output when I plug in the USB microB cable > from a PC. > 1. Are you sure that you provided full debug log for dwc2? Actually first debug prints of dwc2 should be: [72891.088917] dwc2 dwc2.1.auto: mapped PA ef100000 to VA 00000000ca45f000 [72891.088924] dwc2 dwc2.1.auto: registering common handler for irq16 [72891.088963] dwc2 dwc2.1.auto: Core Release: 4.00a (snpsid=4f54400a) [72891.088974] dwc2 dwc2.1.auto: Forcing mode to host [72891.100285] dwc2 dwc2.1.auto: Forcing mode to device [72891.112307] dwc2 dwc2.1.auto: Clearing force mode bits [72891.215275] dwc2 dwc2.1.auto: NonPeriodic TXFIFO size: 1024 [72891.215278] dwc2 dwc2.1.auto: RXFIFO size: 2048 ... 2. Does your driver built for 'dual role device' or 'device only'? 3. First observation from register dumps: core in SOFT disconnect state, which can be reason of observed issue. Could you please update your patch and add only core_connect function call: + dwc2_hsotg_core_connect(hsotg); + return 0; error4: Or instead of patch, if you have any utility in system which will allow from command line clear soft disconnect bit from DCTL register? Offset of DCTL register is 0x804. SoftDiscon bit number is 1. Clear this bit and then plug cable. Unfortunately, currently I'm working from home and can't remotely connect/disconnect cable in lab to try to reproduce observed by you issue. Thanks, Minas
