linux-kernel - Re: [PATCH v1 2/2] drm/msm/dp: add mutex to protect internal_hpd against race condition between different threads

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: <9d405506-3083-3827-990b-ff2151bf6024@quicinc.com>
Date:   Wed, 10 May 2023 16:30:38 -0700
From:   Abhinav Kumar <quic_abhinavk@...cinc.com>
To:     Kuogee Hsieh <quic_khsieh@...cinc.com>,
        Stephen Boyd <swboyd@...omium.org>, <agross@...nel.org>,
        <airlied@...il.com>, <andersson@...nel.org>, <daniel@...ll.ch>,
        <dianders@...omium.org>, <dmitry.baryshkov@...aro.org>,
        <dri-devel@...ts.freedesktop.org>, <robdclark@...il.com>,
        <sean@...rly.run>, <vkoul@...nel.org>
CC:     <quic_jesszhan@...cinc.com>, <quic_sbillaka@...cinc.com>,
        <marijn.suijten@...ainline.org>, <freedreno@...ts.freedesktop.org>,
        <linux-arm-msm@...r.kernel.org>, <linux-kernel@...r.kernel.org>
Subject: Re: [PATCH v1 2/2] drm/msm/dp: add mutex to protect internal_hpd
 against race condition between different threads

Hi Stephen

On 5/10/2023 4:19 PM, Kuogee Hsieh wrote:
> internal_hpd is referenced at both plug and unplug handle.
> 
> The majority purpose of  mutext is try to serialize internal_hpd between 
> dp_bridge_hpd_disable() and either plug or unplug handle.
> 
> 
> On 5/10/2023 4:11 PM, Abhinav Kumar wrote:
>>
>>
>> On 5/10/2023 3:46 PM, Stephen Boyd wrote:
>>> Quoting Kuogee Hsieh (2023-05-10 13:31:05)
>>>> Intrenal_hpd is referenced by event thread but set by drm bridge 
>>>> callback
>>>> context. Add mutex to protect internal_hpd to avoid conflicts between
>>>> threads.
>>>>
>>>> Signed-off-by: Kuogee Hsieh <quic_khsieh@...cinc.com>
>>>> ---
>>>
>>> This patch looks completely unnecessary. How can dp_bridge_hpd_enable()
>>> be called at the same time that dp_bridge_hpd_disable() is called or
>>> dp_bridge_hpd_notify() is called? Isn't there locking or ordering at a
>>> higher layer?
>>
>> Ack. We can drop this patch because we are protected by 
>> bridge->hpd_mutex in drm_bridge_hpd_enable() / drm_bridge_hpd_disable 
>> () and drm_bridge_hpd_notify().

I understood now, so what kuogee is referring to is that this 
event_mutex protection is to not protect those 3 calls from each other 
(since they are already protected as we saw above) but because 
dp_hpd_plug_handle/dp_hpd_unplug_handle still uses 
dp_display.internal_hpd to re-enable the hot-plug interrupt, this is 
making sure that flow is protected as well.