Re: [PATCH v3] drm/mediatek: dp: Fix suspend/resume training failure

["CK Hu (胡俊光)" <ck.hu@mediatek.com>]

On Mon, 2025-08-11 at 17:28 +0800, CK Hu wrote:
> On Sun, 2025-08-10 at 14:17 +0800, Haru Zheng wrote:
> > External email : Please do not click links or open attachments until you have verified the sender or the content.
> > 
> > 
> > When suspending and resuming DisplayPort via Type-C,
> > link training will be fail.
> > 
> > This patch backports the software IRQ handling for DP,
> > as eDP uses hardware IRQ while DP uses software IRQ.
> 
> This patch just modify the flow of DP, not eDP, so do not mention eDP here.
> For DP, I don't know why hardware IRQ would fail.
> HDMI support suspend and resume and it does not software IRQ.
> Please describe the hardware behavior when resume and point out why hardware would not trigger hardware hot plug interrupt.

In addition, resume flow is similar to boot up flow.

In boot up flow, the step is:
1. Plug in cable
2. Power on DP hardware.
3. DP HDP hardware interrupt happen.

So, in resume flow:
1. Plug in cable
2. Power on DP hardware
3. DP HDP hardware interrupt happen. (disappear?)

The flow is the same, why resume has no hardware interrupt?

Regards,
CK

> 
> > Additionally, cable_plugged_in is flipped in
> > mtk_dp_hpd_event to ensure correct hotplug detection
> > during resume.
> > 
> > These changes fix the DP training failure after suspend/resume.
> > 
> > Fixes: f70ac097a2cf ("drm/mediatek: Add MT8195 Embedded DisplayPort driver")
> > Signed-off-by: Haru Zheng <towwy321@gmail.com>
> > 
> > ---