linux-kernel - Re: [PATCH] usb: dwc3: gadget: Avoid race between dwc3 interrupt handler and irq thread handler

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: <87o9zwauwu.fsf@linux.intel.com>
Date:   Wed, 28 Dec 2016 18:19:13 +0200
From:   Felipe Balbi <balbi@...nel.org>
To:     Janusz Dziedzic <janusz.dziedzic@...il.com>
Cc:     Lu Baolu <baolu.lu@...ux.intel.com>,
        Baolin Wang <baolin.wang@...aro.org>,
        Greg KH <gregkh@...uxfoundation.org>,
        USB <linux-usb@...r.kernel.org>,
        LKML <linux-kernel@...r.kernel.org>,
        Linaro Kernel Mailman List <linaro-kernel@...ts.linaro.org>,
        Mark Brown <broonie@...nel.org>
Subject: Re: [PATCH] usb: dwc3: gadget: Avoid race between dwc3 interrupt handler and irq thread handler


Hi,

Janusz Dziedzic <janusz.dziedzic@...il.com> writes:
>>>> On some platfroms(like x86 platform), when one core is running the USB gadget
>>>> irq thread handler by dwc3_thread_interrupt(), meanwhile another core also can
>>>> respond other interrupts from dwc3 controller and modify the event buffer by
>>>> dwc3_interrupt() function, that will cause getting the wrong event count in
>>>> irq thread handler to make the USB function abnormal.
>>>>
>>>> We should add spin_lock/unlock() in dwc3_check_event_buf() to avoid this race.
>>>
>>> Why not spin_lock_irq ones? This lock seems to be used in both
>>> normal and interrupt threads. Or, I missed anything?
>>
>> this is top half handler. Interrupts are already disabled.
>>
> BTW,
> We don't use spin_lock in top half handler.
> Maybe we should/can switch all spin_lock_irqsave() to simple
> spin_lock() in the thread/callbacks?

in theory, yes we've masked all interrupts from this controller for the
duration of the thread handler. However this breaks networking
gadgets. I can only guess network stack has a hard requirement to run
with IRQs disabled.

> Or there is a reason to use irqsave() version?

see above :-)

-- 
balbi