Re: [PATCH v10 00/30] usb: dwc3: part 2, make the resynced code work

[Quentin Schulz <quentin.schulz@cherry.de>]

Hi Anders,

On 11/4/26 6:51 PM, Quentin Schulz wrote:
> Hi Anders,
> 
> On 11/3/26 2:45 PM, Anders Roxell wrote:
>> On 2026-10-30 15:28, Quentin Schulz wrote:
>>> Hi Andrew,
>>>
>>> On 10/28/26 3:15 PM, Anders Roxell wrote:
>>>> Hi,
>>>>
>>
>> [ ... ]
>>
>>> - Q7 USB P0 (lower) with Sandisk Extreme Pro USB stick,
>>> - Q7 USB P2 (upper) with Sandisk Extreme Pro USB stick,
>>>
>>> RK3399 Puma behaves the same in main and with your series applied.
>>> RK3588 Tiger has some regressions.
>>>
>>> Q7 USB P2 (upper) doesn't detect the Sandisk Ultra Flair USB stick 
>>> anymore.
>>> It does on main branch, but on the USB2 controller only.
>>>
>>> Q7 USB P0 (lower) has issues handling the Kingston USB stick. 1st try:
>>
>> Hi Quentin,
>>
>> Thanks for testing both boards.
>>
>> About the OTG port: no need to spend more time on it.
>>
>> v10 had two bugs in the dwc3 core. It never read the core version, so
>> every version check was false. It also cleared the GSBUSCFG0 REQINFO
>> bits on every controller, and main does not touch them. v11 fixes both.
>> My best guess for the Kingston write on P0 is REQINFO, but I have no
>> Tiger to test it on.
>>
>> Did the other sticks pass with v10? That is the Ultra Flair and the
>> Extreme Pro on P0, and the Kingston and the Extreme Pro on P2. I want
>> to know if it is only one stick on each port.
>>
> 
> - Q7 USB P0 (lower) with Sandisk Ultra Flair USB stick,
> - Q7 USB P2 (upper) with Sandisk Ultra Flair USB stick,
> - Q7 USB P0 (lower) with Kingston USB stick,
> - Q7 USB P2 (upper) with Kingston USB stick,
> - Q7 USB P0 (lower) with Sandisk Extreme Pro USB stick,
> - Q7 USB P2 (upper) with Sandisk Extreme Pro USB stick,
> 
> was what I tested, on both v10 and main, on both RK3588 Tiger and RK3399 
> Puma. I only reported the ones that didn't work, so yes to your question 
> above.
> 
>> Can you run the same tests with v11? It is on [1], on top of main
>> ab4dc5aacf95.
>>
> 
> With v11 checked out, I only tested the scenarios with regressions I 
> reported in the previous mail, that is only, on RK3588 Tiger only, with 
> Kingston USB stick on P0 and Ultra Flair on P2.
> 
> 2/2 tries were successful (regression gone) for the Kingston stick on P0.
> 
> 5/5 tries for Ultra Flair on P2 (detected on the USB2 controller same as 
> on main) were successful (regression gone). I got one try for Ultra 
> Flair on P2 that reported the stick on the USB3 controller, funnily 
> enough, the only try when I got a "cannot reset port 1!?". All other 
> times, the USB stick was attached to the USB2 controller (like on main).
> 
> I would like to run all test scenarios again and see if v11 introduces 
> other regressions compared to v10, but this takes me about 2h 
> uninterrupted and I'm not sure I will have time tomorrow or Friday as 
> there are other fires to put out (still have to review the OpenSSL 
> provider thing again, and I'm now finally impacted by it myself as I 
> updated to Fedora 45 with OpenSSL 4 this morning).
> 

For the v11 (commit 6030c4d33099 ("usb: dwc3: enable SUSPHY again after 
xhci is registered")):

Tested-by: Quentin Schulz <quentin.schulz@cherry.de> RK3399 Puma 
usbdrd_dwc3_1
Tested-by: Quentin Schulz <quentin.schulz@cherry.de> RK3588 Tiger 
usb_host1_xhci + usb_host2_xhci

I've just remembered I've now got a Pine64 Quartz64 (RK3566) with two 
DWC3, will try it next Monday.

I should have access to a Pine64 Rock64 (RK3328) starting 2027 where I 
could test the DWC3 controller in there, if I don't forget (90% I will) 
I can then report regressions, if any.

Not having access to any RK3528 device with a DWC3 (MangoPi 28k, NanoPi 
R28S, NanoPi Zero 2, Radxa E20C, Radxa Rock 2A/2F), or to any RK3576 
device with a DWC3 (Anbernic RG Vita Pro, ArmSom Sige5, Khadas Edge 2L, 
NanoPi M5/R76S, Radxa Rock 4D), I cannot test those. I'm assuming Alexey 
may be able to help for the RK3576 with the FlipperOne device (or others 
since he has quite the collection of RK3576 boards IIRC). For RK3528, 
maybe Jonas can help.

I would like to give the v11 a try on RK3566 next Monday before giving a 
definitive go for Rockchip platforms, but this looks good to me.

Cheers,
Quentin