linux-kernel - [PATCH 3/3] drm: zynqmp_dp: Retrain link after HPD if necessary

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] [day] [month] [year] [list]

Message-Id: <20251024191707.2310589-4-sean.anderson@linux.dev>
Date: Fri, 24 Oct 2025 15:17:07 -0400
From: Sean Anderson <sean.anderson@...ux.dev>
To: Laurent Pinchart <laurent.pinchart@...asonboard.com>,
	Tomi Valkeinen <tomi.valkeinen@...asonboard.com>,
	Maarten Lankhorst <maarten.lankhorst@...ux.intel.com>,
	Maxime Ripard <mripard@...nel.org>,
	Thomas Zimmermann <tzimmermann@...e.de>,
	dri-devel@...ts.freedesktop.org
Cc: linux-kernel@...r.kernel.org,
	David Airlie <airlied@...il.com>,
	Hyun Kwon <hyun.kwon@...inx.com>,
	Simona Vetter <simona@...ll.ch>,
	Michal Simek <michal.simek@....com>,
	linux-arm-kernel@...ts.infradead.org,
	Sean Anderson <sean.anderson@...ux.dev>
Subject: [PATCH 3/3] drm: zynqmp_dp: Retrain link after HPD if necessary

The section 5.1.4 of the v1.2 DisplayPort standard says

> The Source device shall respond to Hot Plug event/Hot Re-plug event by
> first reading DPCD Link/Sink Device Status registers at DPCD 00200h
> through 00205h.... If the link is unstable or lost, the Source device
> then reads the DPCD Receiver Capabilities registers at DPCD 00000h
> through 0000Fh to determine the appropriate information needed to
> train the link. The Source device shall then initiate link training.

However, zynqmp_dp_hpd_work_func does not check the link status. This
may prevent the sink from detecting the source if, for example, the user
disconnects the cable and then reconnects it. I encountered this problem
when testing a mini DP connector (although I had no problem when using a
full-size connector with the existing driver).

Follow the spec by checking the link status after a HPD event and
retraining if necessary.

Fixes: d76271d22694 ("drm: xlnx: DRM/KMS driver for Xilinx ZynqMP DisplayPort Subsystem")
Signed-off-by: Sean Anderson <sean.anderson@...ux.dev>
---

 drivers/gpu/drm/xlnx/zynqmp_dp.c | 37 ++++++++++++++++++++------------
 1 file changed, 23 insertions(+), 14 deletions(-)

diff --git a/drivers/gpu/drm/xlnx/zynqmp_dp.c b/drivers/gpu/drm/xlnx/zynqmp_dp.c
index caf2e0ce3644..a90bc0e406f6 100644
--- a/drivers/gpu/drm/xlnx/zynqmp_dp.c
+++ b/drivers/gpu/drm/xlnx/zynqmp_dp.c
@@ -1677,6 +1677,24 @@ static int zynqmp_dp_bridge_atomic_check(struct drm_bridge *bridge,
 	return 0;
 }
 
+static bool zynqmp_hpd_needs_retain(struct zynqmp_dp *dp)
+{
+	u8 status[DP_LINK_STATUS_SIZE + 2];
+	int err;
+
+	err = drm_dp_dpcd_read(&dp->aux, DP_SINK_COUNT, status,
+			       DP_LINK_STATUS_SIZE + 2);
+	if (err < 0) {
+		dev_dbg_ratelimited(dp->dev,
+				    "could not read sink status: %d\n", err);
+		return false;
+	}
+
+	return status[4] & DP_LINK_STATUS_UPDATED ||
+	       !drm_dp_clock_recovery_ok(&status[2], dp->mode.lane_cnt) ||
+	       !drm_dp_channel_eq_ok(&status[2], dp->mode.lane_cnt);
+}
+
 static enum drm_connector_status __zynqmp_dp_bridge_detect(struct zynqmp_dp *dp)
 {
 	struct zynqmp_dp_link_config *link_config = &dp->link_config;
@@ -1698,6 +1716,9 @@ static enum drm_connector_status __zynqmp_dp_bridge_detect(struct zynqmp_dp *dp)
 
 	if (state & ZYNQMP_DP_INTERRUPT_SIGNAL_STATE_HPD) {
 		WRITE_ONCE(dp->status, connector_status_connected);
+		if (!zynqmp_hpd_needs_retain(dp))
+			return connector_status_connected;
+
 		ret = drm_dp_dpcd_read(&dp->aux, 0x0, dp->dpcd,
 				       sizeof(dp->dpcd));
 		if (ret < 0) {
@@ -2335,25 +2356,13 @@ static void zynqmp_dp_hpd_irq_work_func(struct work_struct *work)
 {
 	struct zynqmp_dp *dp = container_of(work, struct zynqmp_dp,
 					    hpd_irq_work);
-	u8 status[DP_LINK_STATUS_SIZE + 2];
-	int err;
 
 	guard(mutex)(&dp->lock);
 	if (dp->ignore_hpd)
 		return;
 
-	err = drm_dp_dpcd_read(&dp->aux, DP_SINK_COUNT, status,
-			       DP_LINK_STATUS_SIZE + 2);
-	if (err < 0) {
-		dev_dbg_ratelimited(dp->dev,
-				    "could not read sink status: %d\n", err);
-	} else {
-		if (status[4] & DP_LINK_STATUS_UPDATED ||
-		    !drm_dp_clock_recovery_ok(&status[2], dp->mode.lane_cnt) ||
-		    !drm_dp_channel_eq_ok(&status[2], dp->mode.lane_cnt)) {
-			zynqmp_dp_train_loop(dp);
-		}
-	}
+	if (zynqmp_hpd_needs_retain(dp))
+		zynqmp_dp_train_loop(dp);
 }
 
 static irqreturn_t zynqmp_dp_irq_handler(int irq, void *data)
-- 
2.35.1.1320.gc452695387.dirty