Re: [PATCH v6 04/13] usb: dwc3: Add splitdisable quirk for Hisilicon Kirin Soc

[Thinh Nguyen <Thinh.Nguyen@synopsys.com>]

Felipe Balbi wrote:
> Hi,
>
> Mauro Carvalho Chehab <mchehab+huawei@kernel.org> writes:
>>>> I tested here, together with the Hikey 970 phy RFC patches I sent
>>>> last week.
>>>>
>>>> Without this patch, the USB HID driver receives -EPROTO from
>>>> submitted URBs, causing it to enter into an endless reset cycle
>>>> on every 500 ms, at the hid_io_error() logic.  
>>>> Tested-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
>>>>
>>>> If you prefer, I can re-submit this one with my SOB.  
>>> Please do, but since you're changing device tree, I need Rob's acked-by.
>> Ok, I'll do that.
> thanks
>
>>>> Em Sat, 20 Apr 2019 14:40:10 +0800
>>>> Yu Chen <chenyu56@huawei.com> escreveu:
>>>>  
>>>>> SPLIT_BOUNDARY_DISABLE should be set for DesignWare USB3 DRD Core
>>>>> of Hisilicon Kirin Soc when dwc3 core act as host.  
>>> is this Kirin-specific or is this something that we should do a revision
>>> check? 
>> I've no idea. I don't have any datasheets from this device.
> I see
>
>>> Why does it affect only Hikey kirin? 
>> As John Stultz didn't re-submit this one (and looking at the DT
>> between Kirin 960 and 970 from the original Kernel 4.9 official
>> drivers), I suspect that only Kirin 970 requires this quirk.
>>
>> It could well be due to some Dwc3 revision, but it could also be due
>> to some differences at the USB part of the SoC, as there are a
> the reason I ask is that if it's caused by dwc3 revision, then we don't
> need the extra dt property, we can rely on a revision check. If it's
> something that can't be detected in runtime, then we need a property.
>
>> few other things different between hikey 960 and 970: it has a
>> different PHY driver, and there are also some differences at the
>> USB HUB which is connected into it.
>>
>> On both devices, the USB physical ports are actually connected
>> into a HUB. In the case of Hikey 970, the hub seems to be a
>> TI TUSB8041 4-Port Hub:
>> 	
>> 	$ lsusb
>> 	Bus 002 Device 002: ID 0451:8140 Texas Instruments, Inc. TUSB8041 4-Port Hub
>> 	Bus 002 Device 001: ID 1d6b:0003 Linux Foundation 3.0 root hub
>> 	Bus 001 Device 004: ID 090c:1000 Silicon Motion, Inc. - Taiwan (formerly Feiya Technology Corp.) Flash Drive
>> 	Bus 001 Device 003: ID 413c:301a Dell Computer Corp. Dell MS116 Optical Mouse
>> 	Bus 001 Device 002: ID 0451:8142 Texas Instruments, Inc. TUSB8041 4-Port Hub
>> 	Bus 001 Device 001: ID 1d6b:0002 Linux Foundation 2.0 root hub
>>
>>> What's the dwc3 revision on
>>> that SoC (grep SNPSID /sys/kernel/debugfs/*dwc3/regdump)?
>> 	GSNPSID = 0x33313130
> This isn't even listed as a known revision in dwc3/core.h. Thinh, could
> the issue being described here caused by a known Erratum with this
> particular revision?

If you have a STAR issue number, then I may be able to look it up.

Btw, GNSPSID register is the ID for DWC_usb31 IP. For DWC_usb31 and
DWC_usb32, we need to read the version number and version type registers
(offset 0xc1a0 and 0xc1a4 respectively).

We need to add these registers to the register dump.

BR,
Thinh

>
>>>>> +		reg = dwc3_readl(dwc->regs, DWC3_GUCTL3);
>>>>> +		reg |= DWC3_GUCTL3_SPLITDISABLE;
>>>>> +		dwc3_writel(dwc->regs, DWC3_GUCTL3, reg);
>>>>> +	}
>>>>> +}
>>>>> +#else
>>>>> +#define dwc3_complete NULL
>>>>>  #endif /* CONFIG_PM_SLEEP */
>>>>>  
>>>>>  static const struct dev_pm_ops dwc3_dev_pm_ops = {
>>>>>  	SET_SYSTEM_SLEEP_PM_OPS(dwc3_suspend, dwc3_resume)
>>>>> +	.complete = dwc3_complete,  
>>> why is this done on complete? Why can't it be done at the end of
>>> dwc3_resume()?
>> Again, no idea. I didn't actually tried to suspend/resume.
>>
>> Maybe the original author can shed a light on it.
> yeah, would be nice :-)
>