Skip to contentSkip to conversation

drm/dp_mst: Lower down debug info level when receive NAK

PATCH In review Wayne Lin 2 participants 3 messages Apr 28, 2022 – Apr 29, 2022

Subsystems (2)
lore
Share & tools
lore

Wayne Lin

author
#0 lore raw

PATCH 1/1 drm/dp_mst: Lower down debug info level when receive NAK

[Why]
It's reasonable that we receive NAK while doing DP_REMOTE_DPCD_READ. Downstream device might reply NAK with the reason and source should react accordingly.

e.g.
1. When downstream device can't handle corresponding message in time, it then replies NAK as reason been set as DEFER.
2. When multi-function branch-sink device doesn't enumerate virtual DP peer devices for those multi-function down facing ports. Without virtual DPCD, branch device might reply NAK with reason as BAD_PARAM indicating this port can't do aux DPCD read.

It's expected result. Not an error.

[How]
Use drm_dbg_kms() to replace drm_err() when receive NAK.

Changes since v1:
* drm_dp_mst_topology.c file path changed. Folder was rename from 'dp' to 'display'

---
 drivers/gpu/drm/display/drm_dp_mst_topology.c | 3 +--
 1 file changed, 1 insertion(+), 2 deletions(-)
drivers/gpu/drm/display/drm_dp_mst_topology.c +1 -2
index 8526aae75c6d..f27aa0b95bea 100644
@@ -3557,9 +3557,8 @@ static int drm_dp_send_dpcd_read(struct drm_dp_mst_topology_mgr *mgr,
 	if (ret < 0)
 		goto fail_free;
 
-	/* DPCD read should never be NACKed */
 	if (txmsg->reply.reply_type == 1) {
-		drm_err(mgr->dev, "mstb %p port %d: DPCD read on addr 0x%x for %d bytes NAKed\n",
+		drm_dbg_kms(mgr->dev, "mstb %p port %d: DPCD read on addr 0x%x for %d bytes NAKed\n",
 			mstb, port->port_num, offset, size);
 		ret = -EIO;
 		goto fail_free;
Signature · 1 line
2.36.0
Reviewed-by on v1 1/1
#1 lore raw

Will push to drm-misc-next in a bit

Quoted 47 lines · Wayne Lin

[Why]
It's reasonable that we receive NAK while doing DP_REMOTE_DPCD_READ. Downstream device might reply NAK with the reason and source should react accordingly.

e.g.
1. When downstream device can't handle corresponding message in time, it then replies NAK as reason been set as DEFER.
2. When multi-function branch-sink device doesn't enumerate virtual DP peer devices for those multi-function down facing ports. Without virtual DPCD, branch device might reply NAK with reason as BAD_PARAM indicating this port can't do aux DPCD read.

It's expected result. Not an error.

[How]
Use drm_dbg_kms() to replace drm_err() when receive NAK.

Changes since v1:
* drm_dp_mst_topology.c file path changed. Folder was rename from 'dp' to 'display'

---
 drivers/gpu/drm/display/drm_dp_mst_topology.c | 3 +--

 1 file changed, 1 insertion(+), 2 deletions(-)
drivers/gpu/drm/display/drm_dp_mst_topology.c +0 -0
index 8526aae75c6d..f27aa0b95bea 100644
@@ -3557,9 +3557,8 @@ static int drm_dp_send_dpcd_read(struct
 drm_dp_mst_topology_mgr *mgr,
         if (ret < 0)
                 goto fail_free;


-       /* DPCD read should never be NACKed */
        if (txmsg->reply.reply_type == 1) {
-               drm_err(mgr->dev, "mstb %p port %d: DPCD read on addr 0x%x for %d bytes NAKed\n",
+               drm_dbg_kms(mgr->dev, "mstb %p port %d: DPCD read on addr 0x%x for %d bytes NAKed\n",
                        mstb, port->port_num, offset, size); ret = -EIO;
                goto fail_free;

Signature · 3 lines
Cheers,
 Lyude Paul (she/her)
 Software Engineer at Red Hat
author Reviewed-by Lyude Paul
#2 lore raw

[Public]

Thank you Lyude!

Regards,
Wayne
________________________________________
From: Lyude Paul <[email protected]>
Sent: Friday, April 29, 2022 00:45
To: Lin, Wayne; [email protected]
Cc: [email protected]; Wentland, Harry; Zuo, Jerry; Siqueira, Rodrigo
Subject: Re: [PATCH] drm/dp_mst: Lower down debug info level when receive NAK

Will push to drm-misc-next in a bit

Quoted 44 lines · Wayne Lin

[Why]
It's reasonable that we receive NAK while doing DP_REMOTE_DPCD_READ. Downstream device might reply NAK with the reason and source should react accordingly.

e.g.
1. When downstream device can't handle corresponding message in time, it then replies NAK as reason been set as DEFER.
2. When multi-function branch-sink device doesn't enumerate virtual DP peer devices for those multi-function down facing ports. Without virtual DPCD, branch device might reply NAK with reason as BAD_PARAM indicating this port can't do aux DPCD read.

It's expected result. Not an error.

[How]
Use drm_dbg_kms() to replace drm_err() when receive NAK.

Changes since v1:
* drm_dp_mst_topology.c file path changed. Folder was rename from 'dp' to 'display'

---
 drivers/gpu/drm/display/drm_dp_mst_topology.c | 3 +--
 1 file changed, 1 insertion(+), 2 deletions(-)
drivers/gpu/drm/display/drm_dp_mst_topology.c +1 -2
index 8526aae75c6d..f27aa0b95bea 100644
@@ -3557,9 +3557,8 @@ static int drm_dp_send_dpcd_read(struct
 drm_dp_mst_topology_mgr *mgr,
        if (ret < 0)
                goto fail_free;
 
-       /* DPCD read should never be NACKed */
        if (txmsg->reply.reply_type == 1) {
-               drm_err(mgr->dev, "mstb %p port %d: DPCD read on addr 0x%x
 for %d bytes NAKed\n",
+               drm_dbg_kms(mgr->dev, "mstb %p port %d: DPCD read on addr
 0x%x for %d bytes NAKed\n",
Quoted 3 lines
                        mstb, port->port_num, offset, size);
                ret = -EIO;
                goto fail_free;

--
Cheers,
Lyude Paul (she/her)
Software Engineer at Red Hat