Re: usb: dwc2: USB on Agilex platform broken after commit 50fb0c128b6e

[Dinh Nguyen <dinguyen@kernel.org>]

On 12/14/21 9:50 AM, Dinh Nguyen wrote:
> 
> 
> On 12/14/21 12:13 AM, Minas Harutyunyan wrote:
>> Hi Dinh,
>>
>> On 12/14/2021 12:40 AM, Dinh Nguyen wrote:
>>>
>>>
>>> On 12/13/21 10:13 AM, Dinh Nguyen wrote:
>>>> Hi Minas,
>>>>
>>>> Thanks for replying!
>>>>
>>>> On 12/13/21 3:47 AM, Minas Harutyunyan wrote:
>>>>> Hi Dinh,
>>>>>
>>>>> On 12/9/2021 1:44 AM, Dinh Nguyen wrote:
>>>>>> Hi Arthur,
>>>>>>
>>>>>> The usb driver is getting this error:
>>>>>>
>>>>>> [    1.539856] Hardware name: SoCFPGA Agilex SoCDK (DT)
>>>>>> [    1.544799] Call trace:
>>>>>> [    1.547233]  dump_backtrace+0x0/0x1c0
>>>>>> [    1.550897]  show_stack+0x18/0x68
>>>>>> [    1.554202]  dump_stack+0xd8/0x134
>>>>>> [    1.557597]  __report_bad_irq+0x38/0xf0
>>>>>> [    1.561423]  note_interrupt+0x31c/0x3e8
>>>>>> [    1.565244]  handle_irq_event_percpu+0x60/0x90
>>>>>> [    1.569671]  handle_irq_event+0x48/0xe0
>>>>>> [    1.573491]  handle_fasteoi_irq+0xb0/0x138
>>>>>> [    1.577574]  generic_handle_irq+0x30/0x48
>>>>>> [    1.581567]  __handle_domain_irq+0x60/0xb8
>>>>>> [    1.585646]  gic_handle_irq+0xb4/0xd0
>>>>>> [    1.589300]  el1_irq+0xb8/0x150
>>>>>> [    1.592428]  arch_cpu_idle+0x18/0x28
>>>>>> [    1.595992]  default_idle_call+0x24/0x5c
>>>>>> [    1.599899]  do_idle+0x1f4/0x290
>>>>>> [    1.603118]  cpu_startup_entry+0x28/0x68
>>>>>> [    1.607026]  rest_init+0xd8/0xe8
>>>>>> [    1.610242]  arch_call_rest_init+0x10/0x1c
>>>>>> [    1.614329]  start_kernel+0x50c/0x544
>>>>>> [    1.617980] handlers:
>>>>>> [    1.620241] [<00000000ae601aac>] dwc2_handle_common_intr
>>>>>> [    1.625541] [<00000000255d793a>] dwc2_hsotg_irq
>>>>>> [    1.630056] [<000000000bc68ab4>] usb_hcd_irq
>>>>>> [    1.634312] Disabling IRQ #33
>>>>>>
>>>>>> After a git-bisect, the result is this commit:
>>>>>>
>>>>>>
>>>>>> commit 50fb0c128b6e8df62200ba85c05ecd6b1396476d
>>>>>> Author: Artur Petrosyan<Arthur.Petrosyan@synopsys.com>
>>>>>> Date:   Tue Apr 13 11:37:15 2021 +0400
>>>>>>
>>>>>>        usb: dwc2: Add clock gating entering flow by system suspend
>>>>>>        If not hibernation nor partial power down are supported,
>>>>>>        clock gating is used to save power.
>>>>>>        Adds a new flow of entering clock gating when PC is
>>>>>>        suspended.
>>>>>>        Signed-off-by: Artur Petrosyan<Arthur.Petrosyan@synopsys.com>
>>>>>> Link:https://urldefense.com/v3/__https://lore.kernel.org/r/20210413073716.30C13A0094@mailhost.synopsys.com__;!!A4F2R9G_pg!IjGS70cMXB9miMnPdcoGriUpDvARm7ScwrlQXAiH9qN-ZnpcaiM_gD-KWp___MVyvEPm_djJ$ 
>>>>>>
>>>>>>
>>>>>>        Signed-off-by: Greg Kroah-Hartman<gregkh@linuxfoundation.org>
>>>>>>
>>>>>> The DWC2 USB version on the Agilex platform is
>>>>>> 0x4F54330A
>>>>>>
>>>>> Could you please provide HW configuration registers values 
>>>>> (GHWCFG1-4).
>>>>>
>>>>
>>>> [    0.987779] dwc2 ffb00000.usb: GHWCFG1    @0xFFFF800012940044 :
>>>> 0x00000000
>>>> [    0.994644] dwc2 ffb00000.usb: GHWCFG2    @0xFFFF800012940048 :
>>>> 0x238FFC90
>>>> [    1.001498] dwc2 ffb00000.usb: GHWCFG3    @0xFFFF80001294004C :
>>>> 0x1F8002E8
>>>> [    1.008350] dwc2 ffb00000.usb: GHWCFG4    @0xFFFF800012940050 :
>>>> 0xFE0F0020
>>>>
>>>>
>>>> I did a little debugging, and noticed that the GINTSTS_PRTINT is
>>>> firing, so the dwc2_handle_common_intr() is getting called
>>>> continuously. There's doesn't seem to be way to handle the
>>>> GINTSTS_PRTINT bit, because there's a check for device_mode().
>>>>
>>>
>>> It looks like the LPMMode bit is disabled in the GHWCFG3. Could that be
>>> the reason why?
>>>
>>> Dinh
>>
>> Could you please disable clock gating by adding to
>> dwc2_set_default_params() function follow line:
>> p->no_clock_gating = true;
>> like in dwc2_set_s3c6400_params() function.
>>
> 
> That did not fix the issue. However, if I remove the line:
> 
> +               /* After entering suspend, hardware is not accessible */
> +               clear_bit(HCD_FLAG_HW_ACCESSIBLE, &hcd->flags);
> 
> Then the issue gets fixed.
> 

I don't see _dwc2_hcd_resume() getting called.

Dinh