linux-kernel - Re: [PATCH 1/2] usbip: Remove repeated setting of hcd->state in vhci_bus

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 for Android: free password hash cracker in your pocket

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

Message-ID: <0ad76300-9b7f-f368-0533-1f88de649d10@kernel.org>
Date:   Wed, 8 May 2019 11:28:05 -0600
From:   shuah <shuah@...nel.org>
To:     Suwan Kim <suwan.kim027@...il.com>
Cc:     valentina.manea.m@...il.com, gregkh@...uxfoundation.org,
        linux-usb@...r.kernel.org, linux-kernel@...r.kernel.org,
        shuah <shuah@...nel.org>
Subject: Re: [PATCH 1/2] usbip: Remove repeated setting of hcd->state in
 vhci_bus_suspend()

On 5/7/19 9:49 AM, Suwan Kim wrote:
> Hi Shuah,
> 
> On Mon, May 06, 2019 at 09:13:02AM -0600, shuah wrote:
>> On 5/6/19 6:55 AM, Suwan Kim wrote:
>>> When hcd suspends execution, hcd_bus_suspend() calls vhci_bus_suspend()
>>> which sets hcd->state as HC_STATE_SUSPENDED. But after calling
>>> vhci_bus_suspend(), hcd_bus_suspend() also sets hcd->state as
>>> HC_STATE_SUSPENDED.
>>> So, setting hcd->state in vhci_hcd_suspend() is unnecessary.
>>>
>>> Signed-off-by: Suwan Kim <suwan.kim027@...il.com>
>>> ---
>>>    drivers/usb/usbip/vhci_hcd.c | 4 ----
>>>    1 file changed, 4 deletions(-)
>>>
>>> diff --git a/drivers/usb/usbip/vhci_hcd.c b/drivers/usb/usbip/vhci_hcd.c
>>> index 667d9c0ec905..e6f378d00fb6 100644
>>> --- a/drivers/usb/usbip/vhci_hcd.c
>>> +++ b/drivers/usb/usbip/vhci_hcd.c
>>> @@ -1238,10 +1238,6 @@ static int vhci_bus_suspend(struct usb_hcd *hcd)
>>>    	dev_dbg(&hcd->self.root_hub->dev, "%s\n", __func__);
>>> -	spin_lock_irqsave(&vhci->lock, flags);
>>> -	hcd->state = HC_STATE_SUSPENDED;
>>> -	spin_unlock_irqrestore(&vhci->lock, flags);
>>> -
>>>    	return 0;
>>>    }
>>>
>>
>> Tell me more about why you think this change is needed? How did you test
>> this change?
> 
> I think that host controller specific functions, vhci_bus_resume()
> or vhci_bus_suspend() in the case of vhci, usually process host
> controller specific data (struct vhci_hcd) not a generic data
> (struct usb_hcd). The generic data is usually processed by generic HCD
> layer. But vhci_bus_resume() and vhci_bus_suspend() set generic data
> (hcd->state) and moreover this variable is set in generic HCD layer
> once again(hcd_bus_resume() and hcd_bus_suspend()).
> 
> So, i think host controller specific functions (vhci_bus_resume()
> and vhci_bus_suspend()) don't need to set the generic data that is
> "hcd->state = HC_STATE_RUNNING or HC_STATE_SUSPEND".
> 

It depends. In this case, vhci_hcd is virtual driver and maintains
port status for devices that are remote. It checks hcd->state in
vhci_hub_status().

It updates the hcd->state in suspend with vhci->lock hold and checks
status from vhci_hub_status(). Removing updating hcd->state from 
vhci_bus_suspend()will open a window where vhci_hub_status() might
find it wrong state.

Same thing for HC_STATE_RUNNING. There are few other drivers that
do the same for similar reasons. xhci_hcd, and dummy_hcd and a few
more. This is the same comment on your patch that removes
HC_STATE_RUNNING update from resume.

> For test, I loaded vhci-hcd module, suspended and resumed my computer
> and checked hcd->state variable. There is no difference compared with
> not modified version because my patch just removes repeated and
> unnecessary part.
> 

This won't fully test the condition since there are remote devices
attached and triggering suspend while attach/detach is in progress.

thanks,
-- Shuah