2022-06-11 08:49:18

by Greg Kroah-Hartman

[permalink] [raw]
Subject: Re: Regression in 5.18.3 from "xhci: Set HCD flag to defer primary roothub registration"

On Sat, Jun 11, 2022 at 10:49:23AM +0400, Christian Hewitt wrote:
> Commit 6c64a664e1cff339ec698d803fa8cbb9af5d95ce “xhci: Set HCD flag to defer
> primary roothub registration” added for Linux 5.18.3 caused a regression on
> some Amlogic S912 devices (original user forum report with an Android TV box
> and confirmed with a Khadas VIM2 board). I do not see issues with older S905
> (WeTeK Play2) or newer S922X (Odroid N2+) devices running the same kernel.
> There are no kernel splats or error messages but lsusb shows no output and
> nothing works. Simple revert restores the previous good working behaviour.
>
> dmesg with commit http://ix.io/3ZTv
> dmesg with revert http://ix.io/3ZTw
>
> I’m not a coder so will need to be fed instructions to assist with debugging
> the issue further if you don't have access to an Amlogic S912 device. I can
> share devices online for remote poking around if needed.

Does 5.19-rc1 also have this problem?

thanks,

greg k-h


2022-06-11 14:45:02

by Christian Hewitt

[permalink] [raw]
Subject: Re: Regression in 5.18.3 from "xhci: Set HCD flag to defer primary roothub registration"


> On 11 Jun 2022, at 11:32 am, Greg Kroah-Hartman <[email protected]> wrote:
>
> On Sat, Jun 11, 2022 at 10:49:23AM +0400, Christian Hewitt wrote:
>> Commit 6c64a664e1cff339ec698d803fa8cbb9af5d95ce “xhci: Set HCD flag to defer
>> primary roothub registration” added for Linux 5.18.3 caused a regression on
>> some Amlogic S912 devices (original user forum report with an Android TV box
>> and confirmed with a Khadas VIM2 board). I do not see issues with older S905
>> (WeTeK Play2) or newer S922X (Odroid N2+) devices running the same kernel.
>> There are no kernel splats or error messages but lsusb shows no output and
>> nothing works. Simple revert restores the previous good working behaviour.
>>
>> dmesg with commit http://ix.io/3ZTv
>> dmesg with revert http://ix.io/3ZTw
>>
>> I’m not a coder so will need to be fed instructions to assist with debugging
>> the issue further if you don't have access to an Amlogic S912 device. I can
>> share devices online for remote poking around if needed.
>
> Does 5.19-rc1 also have this problem?

No, and in the process of fixing up my normal patchset to test 5.19-rc1 I’ve
spotted some other USB patches picked from lists. Once dropped from 5.18.3 usb
works again. So my bad, sorry for the noise!

Christian