[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <64ad7661-4551-7b00-604b-6e15da23a1c7@quicinc.com>
Date: Fri, 23 Aug 2024 14:55:47 +0530
From: Prashanth K <quic_prashk@...cinc.com>
To: Michael Nazzareno Trimarchi <michael@...rulasolutions.com>
CC: 胡连勤 <hulianqin@...o.com>,
"gregkh@...uxfoundation.org" <gregkh@...uxfoundation.org>,
"quic_jjohnson@...cinc.com" <quic_jjohnson@...cinc.com>,
"linux-usb@...r.kernel.org" <linux-usb@...r.kernel.org>,
"linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>,
opensource.kernel <opensource.kernel@...o.com>,
"akpm@...ux-foundation.org"
<akpm@...ux-foundation.org>
Subject: Re: [PATCH v6] usb: gadget: u_serial: Add null pointer check in
gs_read_complete & gs_write_complete
On 23-08-24 01:27 pm, Michael Nazzareno Trimarchi wrote:
> Hi Prashanth
>
> On Fri, Aug 23, 2024 at 9:34 AM Prashanth K <quic_prashk@...cinc.com> wrote:
>>
>>
>>
>> On 23-08-24 12:28 pm, Michael Nazzareno Trimarchi wrote:
>>> Hi
>>>
>>> On Fri, Aug 23, 2024 at 8:40 AM 胡连勤 <hulianqin@...o.com> wrote:
>>>>
>>>> Hello linux community expert:
>>>>
>>>>>> Fixes: c1dca562be8a ("usb gadget: split out serial core")
>>>>>> Cc: stable@...r.kernel.org
>>>>>> Signed-off-by: Lianqin Hu <hulianqin@...o.com>
>>>>>> ---
>>>>>> v6:
>>>>>> - Update the commit text
>>>>>> - Add the Fixes tag
>>>>>> - CC stable kernel
>>>>>> - Add serial_port_lock protection when checking port pointer
>>>>>> - Optimize code comments
>>>>>> - Delete log printing
>>>>
>>>>> You need to list ALL of the versions here, I seem to have missed v4 and
>>>>> v5 somewhere so I don't know what changed there.
>>>>
>>
>> [...]
>>>>> nested spinlocks, why? Did you run this with lockdep enabled to verify you aren't hitting a different bug now?
>>>> Because there is a competition relationship between this function and the gserial_disconnect function,
>>>> the gserial_disconnect function first obtains serial_port_lock and then obtains port->port_lock.
>>>> The purpose of nesting is to ensure that when gs_read_complete is executed, it can be successfully executed after obtaining serial_port_lock.
>>>> gserial_disconnect(..)
>>>> {
>>>> struct gs_port *port = gser->ioport;
>>>> ...
>>>> spin_lock_irqsave(&serial_port_lock, flags);
>>>> spin_lock(&port->port_lock);
>>>> ...
>>>> gser->ioport = NULL; ---> port = NULL;
>>>> ...
>>>> spin_unlock(&port->port_lock);
>>>> spin_unlock_irqrestore(&serial_port_lock, flags);
>>>> }
>>>>
>>>> After enabling the lockdep function (CONFIG_DEBUG_LOCK_ALLOC=y), there is no lockdep-related warning information.
>>>>
>>>>> And why is one irqsave and one not? That feels odd, it might be right, but you need to document here why the difference.
>>>> After the gs_read_complete function is executed, spin_unlock_irqrestore is used to restore the previous state,
>>>
>>> 胡连勤 this is not a common locking pattern that is the reason that
>>> should be properly described.
>> This pattern was already used on gser_suspend/resume callbacks, this was
>> done because the lock was stored under port (and port itself was
>> becoming null), hence we added a static spinlock to mitigate it.
>>>
> I see now, but why don't disable the endpoint before disconnecting?
>
> /* disable endpoints, aborting down any active I/O */
> usb_ep_disable(gser->out);
> usb_ep_disable(gser->in);
>
> Michael
>
Not sure about this case, I think generally we need stop IO before
disabling EP, otherwise TX/RX functions may queue requests while EP is
getting disabled, thats why i think port is removed before ep_disable.
Moreover these callbacks (complete/suspend/resume etc) comes from UDC
and can be async, so better to use locks to prevent these kind of races.
Regards,
Prashanth K
Powered by blists - more mailing lists