linux-kernel - Re: [PATCH 18/19] drm/msm/dp: Move link training to atomic

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: <a085fb45-91e2-4827-b8e9-8af90796cc49@oss.qualcomm.com>
Date: Fri, 1 Aug 2025 16:58:55 -0700
From: Jessica Zhang <jessica.zhang@....qualcomm.com>
To: Dmitry Baryshkov <dmitry.baryshkov@....qualcomm.com>
Cc: Rob Clark <robin.clark@....qualcomm.com>,
        Abhinav Kumar <abhinav.kumar@...ux.dev>,
        Dmitry Baryshkov
 <lumag@...nel.org>, Sean Paul <sean@...rly.run>,
        Marijn Suijten <marijn.suijten@...ainline.org>,
        David Airlie <airlied@...il.com>, Simona Vetter <simona@...ll.ch>,
        linux-arm-msm@...r.kernel.org, dri-devel@...ts.freedesktop.org,
        freedreno@...ts.freedesktop.org, linux-kernel@...r.kernel.org,
        Yongxing Mou <quic_yongmou@...cinc.com>
Subject: Re: [PATCH 18/19] drm/msm/dp: Move link training to atomic_enable()



On 7/14/2025 4:54 AM, Dmitry Baryshkov wrote:
> On Fri, Jul 11, 2025 at 05:58:23PM -0700, Jessica Zhang wrote:
>> Currently, the DP link training is being done during HPD. Move
>> link training to atomic_enable() in accordance with the atomic_enable()
>> documentation.
>>
>> In addition, don't disable the link until atomic_post_disable() (as part
>> of the dp_ctrl_off[_link_stream]() helpers).
>>
>> Since the link training is moved to a later part of the enable sequence,
>> change the bridge detect() to return true when the display is physically
>> connected instead of when the link is ready.
> 
> These two parts should be patch #2 in the series.
> 
>>
>> Finally, call the plug/unplug handlers directly in hpd_notify() instead
>> of queueing them in the event thread so that they aren't preempted by
>> other events.
>>
>> Signed-off-by: Jessica Zhang <jessica.zhang@....qualcomm.com>
>> ---
>>   drivers/gpu/drm/msm/dp/dp_display.c | 15 ++++++++-------
>>   drivers/gpu/drm/msm/dp/dp_drm.c     |  6 +++---
>>   2 files changed, 11 insertions(+), 10 deletions(-)
>>
>> diff --git a/drivers/gpu/drm/msm/dp/dp_display.c b/drivers/gpu/drm/msm/dp/dp_display.c
>> index 87f2750a99ca..32e1ee40c2c3 100644
>> --- a/drivers/gpu/drm/msm/dp/dp_display.c
>> +++ b/drivers/gpu/drm/msm/dp/dp_display.c
>> @@ -410,11 +410,6 @@ static int msm_dp_display_process_hpd_high(struct msm_dp_display_private *dp)
>>   	msm_dp_link_psm_config(dp->link, &dp->panel->link_info, false);
>>   
>>   	msm_dp_link_reset_phy_params_vx_px(dp->link);
>> -	rc = msm_dp_ctrl_on_link(dp->ctrl);
>> -	if (rc) {
>> -		DRM_ERROR("failed to complete DP link training\n");
>> -		goto end;
>> -	}
>>   
>>   	msm_dp_add_event(dp, EV_USER_NOTIFICATION, true, 0);
>>   
>> @@ -1561,6 +1556,12 @@ void msm_dp_bridge_atomic_enable(struct drm_bridge *drm_bridge,
>>   		force_link_train = true;
>>   	}
>>   
>> +	rc = msm_dp_ctrl_on_link(msm_dp_display->ctrl);
>> +	if (rc) {
>> +		DRM_ERROR("Failed link training (rc=%d)\n", rc);
>> +		dp->connector->state->link_status = DRM_LINK_STATUS_BAD;
>> +	}
>> +
>>   	msm_dp_display_enable(msm_dp_display, force_link_train);
>>   
>>   	rc = msm_dp_display_post_enable(dp);
>> @@ -1706,7 +1707,7 @@ void msm_dp_bridge_hpd_notify(struct drm_bridge *bridge,
>>   		return;
>>   
>>   	if (!msm_dp_display->link_ready && status == connector_status_connected)
>> -		msm_dp_add_event(dp, EV_HPD_PLUG_INT, 0, 0);
>> +		msm_dp_hpd_plug_handle(dp, 0);
>>   	else if (msm_dp_display->link_ready && status == connector_status_disconnected)
>> -		msm_dp_add_event(dp, EV_HPD_UNPLUG_INT, 0, 0);
>> +		msm_dp_hpd_unplug_handle(dp, 0);
> 
> This chunk should be separated from this patch. I'd ask to drop
> EV_HPD_PLUG_INT / EV_HPD_UNPLUG_INT completely and call DRM functions
> all over the place instead. You can do it in a single patch, which comes
> after this one.

Hi Dmitry,

Sure I can split this into a separate patch.

Is the goal here to remove the event queue entirely?

I can drop EV_USER_NOTIFICATION, but I'm not sure if I can completely 
drop EV_HPD_[UN]PLUG_INT entirely without major refactor of the 
plug/unplug handlers since they are used for the HPD IRQ handling.

Thanks,

Jessica Zhang

> 
>>   }
>