linux-kernel - Re: [PATCH] usb: ohci-at91: Fix the unhandle interrupt when resume

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <8569ced1-ed6d-18b3-5223-a8bd923f864b@gmail.com>
Date:   Sun, 25 Jun 2023 12:01:12 +0800
From:   Guiting Shen <aarongt.shen@...il.com>
To:     Alan Stern <stern@...land.harvard.edu>
Cc:     gregkh@...uxfoundation.org, nicolas.ferre@...rochip.com,
        alexandre.belloni@...tlin.com, claudiu.beznea@...rochip.com,
        linux-usb@...r.kernel.org, linux-arm-kernel@...ts.infradead.org,
        linux-kernel@...r.kernel.org
Subject: Re: [PATCH] usb: ohci-at91: Fix the unhandle interrupt when resume

On Fri, Jun 23, 2023 at 23:52:53PM GMT+8, Alan Stern wrote:
> On Fri, Jun 23, 2023 at 11:44:04AM +0800, Guiting Shen wrote:
>>
>>
>> On Thu,Jun 22,2023 at 22:29:43PM GMT+8, Alan Stern wrote:
>>> On Thu, Jun 22, 2023 at 10:57:39AM +0800, Guiting Shen wrote:
>>>> The ohci_hcd_at91_drv_suspend() sets ohci->rh_state to OHCI_RH_HALTED when
>>>> suspend which will let the ohci_irq() skip the interrupt after resume. And
>>>> nobody to handle this interrupt.
>>>>
>>>> Set the ohci->rh_state to OHCI_RH_SUSPEND instead of OHCI_RH_HALTED when
>>>> suspend to fix it.
>>>>
>>>> Signed-off-by: Guiting Shen <aarongt.shen@...il.com>
>>>> ---
>>>>   drivers/usb/host/ohci-at91.c | 2 +-
>>>>   1 file changed, 1 insertion(+), 1 deletion(-)
>>>>
>>>> diff --git a/drivers/usb/host/ohci-at91.c b/drivers/usb/host/ohci-at91.c
>>>> index b9ce8d80f20b..7a970e573668 100644
>>>> --- a/drivers/usb/host/ohci-at91.c
>>>> +++ b/drivers/usb/host/ohci-at91.c
>>>> @@ -645,7 +645,7 @@ ohci_hcd_at91_drv_suspend(struct device *dev)
>>>>   	 * REVISIT: some boards will be able to turn VBUS off...
>>>>   	 */
>>>>   	if (!ohci_at91->wakeup) {
>>>> -		ohci->rh_state = OHCI_RH_HALTED;
>>>> +		ohci->rh_state = OHCI_RH_SUSPENDED;
>>>
>>> It looks like this change ignores the comment immediately above it
>>> (just before the start of this hunk).
>>>
>>> If you want to find a way to handle IRQs better after the controller
>>> resumes, maybe you should change the resume routine instead of the
>>> suspend routine.
>>>
>>> Alan Stern
>>
>> The comment which was added with commit-id 0365ee0a8f745 may be outdated
>> because ohci_suspend() and ohci_at91_port_suspend() is used to suspend
>> instead of setting ohci->rh_state to OHCI_RH_HALTED.
> 
> The comment says nothing about ohci->rh_state; it talks about the 
> integrated transceivers and the 48 MHz clock.  I don't see why you would 
> think the comment is outdated.
> 

The comment says the reason to discard connection state by reset, but I
don't see any reset operation at ohci-at91 suspend/resume routine in
source code.And ohci_suspend() disable irq emission and mark HW
unaccessible  maybe do the same effect as set ohci->rh_state to
OHCI_RH_HALTED to discard connection state which I think the comment is
outdated.

>> What's more, I found that only ohci-at91 driver to set the ohci->rh_state
>> which may be unnessory because the ohci_suspend() disable irq emission and
>> mark HW unaccessible and ohci_at91_port_suspend() suspend the controller.
>>
>> Is it really need to set ohci->rh_state in ohci_hcd_at91_drv_suspend()?
>>
>> It maybe confused to set ohci->rh_state to OHCI_RH_SUSPEND in resume
>> routine.
> 
> I'm not really sure what that assignment was intended to accomplish, but 
> maybe it was meant to force a reset when the controller resumes.
> 
> You could get the same result by leaving ohci->rh_state set to 
> OHCI_RH_SUSPENDED but changing ohci_hcd_at91_drv_resume().  Instead of 
> calling ohci_resume(hcd, false), have it call:
> 
> 	ohci_resume(hcd, !ohci_at91->wakeup);
> 
> That way, if the wakeup flag is clear and the clock was stopped, 
> ohci_resume() will call ohci_usb_reset().  You should also add a comment 
> explaining the reason.
> 
> I can't test this because I don't have the AT91 hardware.
> 

It works by your methods to force a reset in my sama5d3 soc hardware.
And I found that the ohci->rh_state was already OHCI_RH_SUSPEND before
set it OHCI_RH_HALTED in suspend.So the question is whether to set it
OHCI_RH_HALETED in ohci-at91 suspend routine.

It also works by comment the ohci->rh_state = OHCI_RH_HALTED in suspend
routine.But I think it is safer to use your methods.

Do you have any suggestion on it before I send v2 patch latter.

-- 
Regards,
Guiting Shen