Skip to contentSkip to conversation

drm/simple_kms_helper: Fix NULL pointer dereference with no active CRTC

PATCH In review Oleksandr Andrushchenko 3 participants 10 messages Feb 13, 2018 – Mar 5, 2018

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

PATCH 1/1 drm/simple_kms_helper: Fix NULL pointer dereference with no active CRTC

From: Oleksandr Andrushchenko <[email protected]>

It is possible that drm_simple_kms_plane_atomic_check called with no CRTC set, e.g. when user-space application sets CRTC_ID/FB_ID to 0 before doing any actual drawing. This leads to NULL pointer dereference because in this case new CRTC state is NULL and must be checked before accessing.

---
 drivers/gpu/drm/drm_simple_kms_helper.c | 6 ++++--
 1 file changed, 4 insertions(+), 2 deletions(-)
drivers/gpu/drm/drm_simple_kms_helper.c +4 -2
index 9ca8a4a59b74..a05eca9cec8b 100644
@@ -121,8 +121,10 @@ static int drm_simple_kms_plane_atomic_check(struct drm_plane *plane,
 	pipe = container_of(plane, struct drm_simple_display_pipe, plane);
 	crtc_state = drm_atomic_get_new_crtc_state(plane_state->state,
 						   &pipe->crtc);
-	if (!crtc_state->enable)
-		return 0; /* nothing to check when disabling or disabled */
+
+	if (!crtc_state || !crtc_state->enable)
+		/* nothing to check when disabling or disabled or no CRTC set */
+		return 0;
 
 	if (crtc_state->enable)
 		drm_mode_get_hv_timing(&crtc_state->mode,
Signature · 6 lines
2.7.4

_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel
#1 lore raw
Quoted 20 lines · Oleksandr Andrushchenko

From: Oleksandr Andrushchenko <[email protected]>

It is possible that drm_simple_kms_plane_atomic_check called with no CRTC set, e.g. when user-space application sets CRTC_ID/FB_ID to 0 before doing any actual drawing. This leads to NULL pointer dereference because in this case new CRTC state is NULL and must be checked before accessing.

---
 drivers/gpu/drm/drm_simple_kms_helper.c | 6 ++++--
 1 file changed, 4 insertions(+), 2 deletions(-)
drivers/gpu/drm/drm_simple_kms_helper.c +0 -0
index 9ca8a4a59b74..a05eca9cec8b 100644
@@ -121,8 +121,10 @@ static int drm_simple_kms_plane_atomic_check(struct drm_plane *plane,
 	pipe = container_of(plane, struct drm_simple_display_pipe, plane);
 	crtc_state = drm_atomic_get_new_crtc_state(plane_state->state,

drivers/gpu/drm/drm_simple_kms_helper.c static int drm_simple_kms_plane_atomic_check(struct drm_plane *plane, · line 130

Quoted 10 lines
drivers/gpu/drm/drm_simple_kms_helper.c +4 -2
 						   &pipe->crtc);
-	if (!crtc_state->enable)
-		return 0; /* nothing to check when disabling or disabled */
+
+	if (!crtc_state || !crtc_state->enable)
+		/* nothing to check when disabling or disabled or no CRTC set */
+		return 0;
 
 	if (crtc_state->enable)
 		drm_mode_get_hv_timing(&crtc_state->mode,

Hm, this is a bit annoying, since the can_position = false parameter to drm_atomic_helper_check_plane_state is supposed to catch all this. Would moving all the checks after the call to that helper, and gating them on plane_state->visible also work?

We'd need to add a guarantee to drm_atomic_helper_check_plane_state that it can cope with crtc_state == NULL, but I think that's a good idea anyway. Atm it shouldn't end up looking at the crtc_state pointer in that case.

Otherwise we'll just go with your fix, but it feels all a bit too fragile, hence why I want to explore more robust options a bit.
-Daniel

Signature · 7 lines
Daniel Vetter
Software Engineer, Intel Corporation
http://blog.ffwll.ch
_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel
#2 lore raw
Quoted 29 lines · Daniel Vetter
Quoted 28 lines · Oleksandr Andrushchenko

From: Oleksandr Andrushchenko <[email protected]>

It is possible that drm_simple_kms_plane_atomic_check called with no CRTC set, e.g. when user-space application sets CRTC_ID/FB_ID to 0 before doing any actual drawing. This leads to NULL pointer dereference because in this case new CRTC state is NULL and must be checked before accessing.

---
  drivers/gpu/drm/drm_simple_kms_helper.c | 6 ++++--
  1 file changed, 4 insertions(+), 2 deletions(-)
drivers/gpu/drm/drm_simple_kms_helper.c +4 -2
index 9ca8a4a59b74..a05eca9cec8b 100644
@@ -121,8 +121,10 @@ static int drm_simple_kms_plane_atomic_check(struct drm_plane *plane,
  	pipe = container_of(plane, struct drm_simple_display_pipe, plane);
  	crtc_state = drm_atomic_get_new_crtc_state(plane_state->state,
  						   &pipe->crtc);
-	if (!crtc_state->enable)
-		return 0; /* nothing to check when disabling or disabled */
+
+	if (!crtc_state || !crtc_state->enable)
+		/* nothing to check when disabling or disabled or no CRTC set */
+		return 0;
 
Quoted 6 lines

drivers/gpu/drm/drm_simple_kms_helper.c static int drm_simple_kms_plane_atomic_check(struct drm_plane *plane, · line 130

Quoted 2 lines
drivers/gpu/drm/drm_simple_kms_helper.c +0 -0
  	if (crtc_state->enable)
  		drm_mode_get_hv_timing(&crtc_state->mode,

Hm, this is a bit annoying, since the can_position = false parameter to drm_atomic_helper_check_plane_state is supposed to catch all this. Would moving all the checks after the call to that helper, and gating them on plane_state->visible also work?

Yes, it does work if this is what you mean:

drivers/gpu/drm/drm_simple_kms_helper.c +0 -0
index a05eca9cec8b..c48858bb2823 100644
@@ -122,14 +122,6 @@ static int drm_simple_kms_plane_atomic_check(struct drm_plane *plane,

        crtc_state = drm_atomic_get_new_crtc_state(plane_state->state, &pipe->crtc);

-       if (!crtc_state || !crtc_state->enable)
-               /* nothing to check when disabling or disabled or no CRTC set */ -               return 0;
-
-       if (crtc_state->enable)
-               drm_mode_get_hv_timing(&crtc_state->mode,
-                                      &clip.x2, &clip.y2);
-
        ret = drm_atomic_helper_check_plane_state(plane_state, crtc_state, &clip, DRM_PLANE_HELPER_NO_SCALING,
@@ -138,6 +130,13 @@ static int drm_simple_kms_plane_atomic_check(struct drm_plane *plane, if (ret)
                return ret;

+       if (!plane_state->visible || !crtc_state->enable)
+               return 0; /* nothing to check when disabling or disabled */ +
+       if (plane_state->visible && crtc_state->enable)
+               drm_mode_get_hv_timing(&crtc_state->mode,
+                                      &clip.x2, &clip.y2);
+
        if (!plane_state->visible)
                return -EINVAL;

Quoted 4 lines

We'd need to add a guarantee to drm_atomic_helper_check_plane_state that it can cope with crtc_state == NULL, but I think that's a good idea anyway. Atm it shouldn't end up looking at the crtc_state pointer in that case.

It doesn't look at it at the moment

Quoted 2 lines

Otherwise we'll just go with your fix, but it feels all a bit too fragile, hence why I want to explore more robust options a bit.

At list with the change above it passes my test which failed before. Although I cannot confirm it works for others, but it certainly does for me.

Quoted 1 line

-Daniel

Do you want me to send v1 with the code above?

Thank you,
Oleksandr
_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel

#3 lore raw
Quoted 35 lines · Oleksandr Andrushchenko
Quoted 26 lines · Daniel Vetter
Quoted 25 lines · Oleksandr Andrushchenko

From: Oleksandr Andrushchenko <[email protected]>

It is possible that drm_simple_kms_plane_atomic_check called with no CRTC set, e.g. when user-space application sets CRTC_ID/FB_ID to 0 before doing any actual drawing. This leads to NULL pointer dereference because in this case new CRTC state is NULL and must be checked before accessing.

---
  drivers/gpu/drm/drm_simple_kms_helper.c | 6 ++++--
  1 file changed, 4 insertions(+), 2 deletions(-)
drivers/gpu/drm/drm_simple_kms_helper.c +2 -2
index 9ca8a4a59b74..a05eca9cec8b 100644
@@ -121,8 +121,10 @@ static int drm_simple_kms_plane_atomic_check(struct drm_plane *plane,
  	pipe = container_of(plane, struct drm_simple_display_pipe, plane);
  	crtc_state = drm_atomic_get_new_crtc_state(plane_state->state,
  						   &pipe->crtc);
-	if (!crtc_state->enable)
-		return 0; /* nothing to check when disabling or disabled */
+
+	if (!crtc_state || !crtc_state->enable)
Quoted 8 lines

drivers/gpu/drm/drm_simple_kms_helper.c static int drm_simple_kms_plane_atomic_check(struct drm_plane *plane, · line 129

Quoted 4 lines
drivers/gpu/drm/drm_simple_kms_helper.c +2 -0
+		/* nothing to check when disabling or disabled or no CRTC set */
+		return 0;
  	if (crtc_state->enable)
  		drm_mode_get_hv_timing(&crtc_state->mode,

Hm, this is a bit annoying, since the can_position = false parameter to drm_atomic_helper_check_plane_state is supposed to catch all this. Would moving all the checks after the call to that helper, and gating them on plane_state->visible also work?

Quoted 1 line

Yes, it does work if this is what you mean:

I wasn't sure, thanks for figuring this out!

Quoted 23 lines
drivers/gpu/drm/drm_simple_kms_helper.c +0 -2
index a05eca9cec8b..c48858bb2823 100644
@@ -122,14 +122,6 @@ static int drm_simple_kms_plane_atomic_check(struct
 drm_plane *plane,
         crtc_state = drm_atomic_get_new_crtc_state(plane_state->state,
 &pipe->crtc);
 
-       if (!crtc_state || !crtc_state->enable)
-               /* nothing to check when disabling or disabled or no CRTC

set */
-               return 0;

drivers/gpu/drm/drm_simple_kms_helper.c +0 -5
-
-       if (crtc_state->enable)
-               drm_mode_get_hv_timing(&crtc_state->mode,
-                                      &clip.x2, &clip.y2);
-

        ret = drm_atomic_helper_check_plane_state(plane_state, crtc_state, &clip, DRM_PLANE_HELPER_NO_SCALING,

drivers/gpu/drm/drm_simple_kms_helper.c static int drm_simple_kms_plane_atomic_check(struct · line 135

Quoted 6 lines
drivers/gpu/drm/drm_simple_kms_helper.c +2 -0
@@ -138,6 +130,13 @@ static int drm_simple_kms_plane_atomic_check(struct
 drm_plane *plane,
         if (ret)
                 return ret;
 
+       if (!plane_state->visible || !crtc_state->enable)
+               return 0; /* nothing to check when disabling or disabled */
if (!plane_state->visible) {
	WARN_ON(crtc_state->enabled);
	return 0;
}

The helper call above should guarantee this.

drivers/gpu/drm/drm_simple_kms_helper.c static int drm_simple_kms_plane_atomic_check(struct · line 137

Quoted 2 lines
drivers/gpu/drm/drm_simple_kms_helper.c +2 -0
+
+       if (plane_state->visible && crtc_state->enable)

Similar here.

drivers/gpu/drm/drm_simple_kms_helper.c static int drm_simple_kms_plane_atomic_check(struct · line 142

Quoted 5 lines
drivers/gpu/drm/drm_simple_kms_helper.c +3 -0
+               drm_mode_get_hv_timing(&crtc_state->mode,
+                                      &clip.x2, &clip.y2);
+
         if (!plane_state->visible)
                 return -EINVAL;

This can now be removed, the plane helper takes care of checking for plane_state->visible != crtc_state->enable. Please also remove.

Quoted 4 lines
Quoted 4 lines

We'd need to add a guarantee to drm_atomic_helper_check_plane_state that it can cope with crtc_state == NULL, but I think that's a good idea anyway. Atm it shouldn't end up looking at the crtc_state pointer in that case.

Quoted 8 lines

It doesn't look at it at the moment

Quoted 2 lines

Otherwise we'll just go with your fix, but it feels all a bit too fragile, hence why I want to explore more robust options a bit.

At list with the change above it passes my test which failed before. Although I cannot confirm it works for others, but it certainly does for me.

Quoted 1 line

-Daniel

Do you want me to send v1 with the code above?

Yes please, with my above cleanup suggestions.

Thanks, Daniel

Signature · 7 lines
Daniel Vetter
Software Engineer, Intel Corporation
http://blog.ffwll.ch
_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel
#4 lore raw
Quoted 66 lines · Daniel Vetter
Quoted 35 lines · Oleksandr Andrushchenko
Quoted 26 lines · Daniel Vetter
Quoted 25 lines · Oleksandr Andrushchenko

From: Oleksandr Andrushchenko <[email protected]>

It is possible that drm_simple_kms_plane_atomic_check called with no CRTC set, e.g. when user-space application sets CRTC_ID/FB_ID to 0 before doing any actual drawing. This leads to NULL pointer dereference because in this case new CRTC state is NULL and must be checked before accessing.

---
   drivers/gpu/drm/drm_simple_kms_helper.c | 6 ++++--
   1 file changed, 4 insertions(+), 2 deletions(-)
drivers/gpu/drm/drm_simple_kms_helper.c +2 -2
index 9ca8a4a59b74..a05eca9cec8b 100644
@@ -121,8 +121,10 @@ static int drm_simple_kms_plane_atomic_check(struct drm_plane *plane,
   	pipe = container_of(plane, struct drm_simple_display_pipe, plane);
   	crtc_state = drm_atomic_get_new_crtc_state(plane_state->state,
   						   &pipe->crtc);
-	if (!crtc_state->enable)
-		return 0; /* nothing to check when disabling or disabled */
+
+	if (!crtc_state || !crtc_state->enable)
Quoted 8 lines

drivers/gpu/drm/drm_simple_kms_helper.c static int drm_simple_kms_plane_atomic_check(struct drm_plane *plane, · line 129

Quoted 4 lines
drivers/gpu/drm/drm_simple_kms_helper.c +2 -0
+		/* nothing to check when disabling or disabled or no CRTC set */
+		return 0;
   	if (crtc_state->enable)
   		drm_mode_get_hv_timing(&crtc_state->mode,

Hm, this is a bit annoying, since the can_position = false parameter to drm_atomic_helper_check_plane_state is supposed to catch all this. Would moving all the checks after the call to that helper, and gating them on plane_state->visible also work?

Quoted 1 line

Yes, it does work if this is what you mean:

I wasn't sure, thanks for figuring this out!

Quoted 27 lines
drivers/gpu/drm/drm_simple_kms_helper.c +0 -2
index a05eca9cec8b..c48858bb2823 100644
@@ -122,14 +122,6 @@ static int drm_simple_kms_plane_atomic_check(struct
 drm_plane *plane,
         crtc_state = drm_atomic_get_new_crtc_state(plane_state->state,
 &pipe->crtc);
 
-       if (!crtc_state || !crtc_state->enable)
-               /* nothing to check when disabling or disabled or no CRTC

set */
-               return 0;

drivers/gpu/drm/drm_simple_kms_helper.c +0 -5
-
-       if (crtc_state->enable)
-               drm_mode_get_hv_timing(&crtc_state->mode,
-                                      &clip.x2, &clip.y2);
-
         ret = drm_atomic_helper_check_plane_state(plane_state, crtc_state,
                                                   &clip,

DRM_PLANE_HELPER_NO_SCALING,

drivers/gpu/drm/drm_simple_kms_helper.c +0 -0
@@ -138,6 +130,13 @@ static int drm_simple_kms_plane_atomic_check(struct
 drm_plane *plane,
         if (ret)
                 return ret;
 
Quoted 8 lines

drivers/gpu/drm/drm_simple_kms_helper.c static int drm_simple_kms_plane_atomic_check(struct · line 135

Quoted 2 lines
drivers/gpu/drm/drm_simple_kms_helper.c +2 -0
+       if (!plane_state->visible || !crtc_state->enable)
+               return 0; /* nothing to check when disabling or disabled */
if (!plane_state->visible) {
	WARN_ON(crtc_state->enabled);
	return 0;
}

The helper call above should guarantee this.

Yes, but I still see cases when crtc_state is NULL, thus making crtc_state->enable to fail

Quoted 17 lines

drivers/gpu/drm/drm_simple_kms_helper.c static int drm_simple_kms_plane_atomic_check(struct · line 137

Quoted 2 lines
drivers/gpu/drm/drm_simple_kms_helper.c +2 -0
+
+       if (plane_state->visible && crtc_state->enable)

Similar here.

drivers/gpu/drm/drm_simple_kms_helper.c static int drm_simple_kms_plane_atomic_check(struct · line 142

Quoted 5 lines
drivers/gpu/drm/drm_simple_kms_helper.c +3 -0
+               drm_mode_get_hv_timing(&crtc_state->mode,
+                                      &clip.x2, &clip.y2);
+
         if (!plane_state->visible)
                 return -EINVAL;

This can now be removed, the plane helper takes care of checking for plane_state->visible != crtc_state->enable. Please also remove.

Quoted 5 lines
Quoted 4 lines

We'd need to add a guarantee to drm_atomic_helper_check_plane_state that it can cope with crtc_state == NULL, but I think that's a good idea anyway. Atm it shouldn't end up looking at the crtc_state pointer in that case.

It doesn't look at it at the moment

Quoted 8 lines
Quoted 7 lines
Quoted 2 lines

Otherwise we'll just go with your fix, but it feels all a bit too fragile, hence why I want to explore more robust options a bit.

At list with the change above it passes my test which failed before. Although I cannot confirm it works for others, but it certainly does for me.

Quoted 1 line

-Daniel

Do you want me to send v1 with the code above?

Yes please, with my above cleanup suggestions.

Please see the patch under test attached (I believe it is what you mean, with the only change that
    if (!plane_state->visible) {
        *if (crtc_state)*
            WARN_ON(crtc_state->enable);
        return 0;
    }
check is used).

Whith this patch + additional logs I have:

[   18.939204] [drm:drm_ioctl [drm]] pid=2105, dev=0xe200, auth=1, DRM_IOCTL_MODE_ATOMIC
[...]
[   18.939681] [drm:drm_atomic_set_crtc_for_plane [drm]] Link plane state 00000000c302cbbf to [NOCRTC]
[   18.939822] [drm:drm_atomic_set_fb_for_plane [drm]] Set [NOFB] for plane state 00000000c302cbbf
[   18.939963] [drm:drm_atomic_print_state [drm]] checking 000000000bc224e7
[   18.939988] vdispl vdispl.0: [drm] plane[29]: plane-0
[   18.940003] vdispl vdispl.0: [drm]   crtc=(null)
[   18.940018] vdispl vdispl.0: [drm]   fb=0
[   18.940032] vdispl vdispl.0: [drm]   crtc-pos=0x0+0+0
[   18.940048] vdispl vdispl.0: [drm] src-pos=0.000000x0.000000+0.000000+0.000000
[   18.940067] vdispl vdispl.0: [drm]   rotation=1
[   18.940199] [drm:drm_atomic_check_only [drm]] checking 000000000bc224e7
[   18.940226] ================================= plane_state->visible 0 crtc_state           (null)
[...]
[   18.978146] [drm:drm_atomic_set_crtc_for_plane [drm]] Link plane state 000000006bd50580 to [CRTC:30:crtc-0]
[   18.978292] [drm:drm_atomic_set_fb_for_plane [drm]] Set [FB:35] for plane state 000000006bd50580
[   18.978993] [drm:drm_atomic_set_mode_prop_for_crtc [drm]] Set [MODE:1024x768] for CRTC state 00000000e5a28f6a
[   18.979425] [drm:drm_atomic_check_only [drm]] checking 000000000bc224e7
[   18.979540] [drm:drm_atomic_helper_check_modeset [drm_kms_helper]] [CRTC:30:crtc-0] mode changed
[   18.979632] [drm:drm_atomic_helper_check_modeset [drm_kms_helper]] [CRTC:30:crtc-0] enable changed
[   18.979708] [drm:drm_atomic_helper_check_modeset [drm_kms_helper]] [CRTC:30:crtc-0] active changed
[   18.979792] [drm:drm_atomic_helper_check_modeset [drm_kms_helper]] Updating routing for [CONNECTOR:28:Virtual-1]
[   18.979877] [drm:drm_atomic_helper_check_modeset [drm_kms_helper]] [CONNECTOR:28:Virtual-1] using [ENCODER:31:None-31] on [CRTC:30:crtc-0] [   18.979960] [drm:drm_atomic_helper_check_modeset [drm_kms_helper]] [CRTC:30:crtc-0] needs all connectors, enable: y, active: y [   18.980139] [drm:drm_atomic_add_affected_connectors [drm]] Adding all current connectors for [CRTC:30:crtc-0] to 000000000bc224e7 [   18.980184] ================================= plane_state->visible 0 crtc_state 00000000e5a28f6a
[   18.980205] crtc_state->enable 1

*[   19.022608] WARNING: CPU: 1 PID: 2105 at drivers/gpu/drm/drm_simple_kms_helper.c:137 drm_simple_kms_plane_atomic_check+0xdc/0xf8 [drm_kms_helper]*

[...]

[   19.113601] ================================= plane_state->visible 0 crtc_state 00000000e5a28f6a
[   19.113623] crtc_state->enable 1
[   19.113792] WARNING: CPU: 1 PID: 2105 at drivers/gpu/drm/drm_simple_kms_helper.c:137 drm_simple_kms_plane_atomic_check+0xdc/0xf8 [drm_kms_helper]

[...]

And finally

[   19.340249] ================================= plane_state->visible 0 crtc_state 0000000036a1b1f5 [   19.340271] crtc_state->enable 0

So, it seems that crtc_state can still be NULL if "!plane_state->visible" making NULL pointer dereference, so we need a check for that.
Yet, !plane_state->visible && crtc_state->enable makes WARN_ON to fire always. So, probably we may want removing it.

Quoted 2 lines

Thanks, Daniel

Thank you,
Oleksandr

Quoted 1 line

From dbcce708b237740158a2c16029c56a579324f269 Mon Sep 17 00:00:00 2001

From: Oleksandr Andrushchenko <[email protected]>
Date: Tue, 13 Feb 2018 10:32:20 +0200
Subject: [PATCH] drm/simple_kms_helper: Fix NULL pointer dereference with no
active CRTC

It is possible that drm_simple_kms_plane_atomic_check called with no CRTC set, e.g. when user-space application sets CRTC_ID/FB_ID to 0 before doing any actual drawing. This leads to NULL pointer dereference because in this case new CRTC state is NULL and must be checked before accessing.

---
 drivers/gpu/drm/drm_simple_kms_helper.c | 15 +++++++--------
 1 file changed, 7 insertions(+), 8 deletions(-)
drivers/gpu/drm/drm_simple_kms_helper.c +7 -8
index 9ca8a4a59b74..f54711ff9767 100644
@@ -121,12 +121,6 @@ static int drm_simple_kms_plane_atomic_check(struct drm_plane *plane,
 	pipe = container_of(plane, struct drm_simple_display_pipe, plane);
 	crtc_state = drm_atomic_get_new_crtc_state(plane_state->state,
 						   &pipe->crtc);
-	if (!crtc_state->enable)
-		return 0; /* nothing to check when disabling or disabled */
-
-	if (crtc_state->enable)
-		drm_mode_get_hv_timing(&crtc_state->mode,
-				       &clip.x2, &clip.y2);
 
 	ret = drm_atomic_helper_check_plane_state(plane_state, crtc_state,
 						  &clip,
@@ -136,8 +130,13 @@ static int drm_simple_kms_plane_atomic_check(struct drm_plane *plane,
 	if (ret)
 		return ret;
 
-	if (!plane_state->visible)
-		return -EINVAL;
+	if (!plane_state->visible) {
+		if (crtc_state)
+			WARN_ON(crtc_state->enable);
+		return 0;
+	}
+
+	drm_mode_get_hv_timing(&crtc_state->mode, &clip.x2, &clip.y2);
 
 	if (!pipe->funcs || !pipe->funcs->check)
 		return 0;
Signature · 1 line
2.7.4

_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel

Reviewed-by
#5 lore raw
Quoted 74 lines · Oleksandr Andrushchenko
Quoted 73 lines · Daniel Vetter
Quoted 35 lines · Oleksandr Andrushchenko
Quoted 26 lines · Daniel Vetter
Quoted 25 lines · Oleksandr Andrushchenko

From: Oleksandr Andrushchenko <[email protected]>

It is possible that drm_simple_kms_plane_atomic_check called with no CRTC set, e.g. when user-space application sets CRTC_ID/FB_ID to 0 before doing any actual drawing. This leads to NULL pointer dereference because in this case new CRTC state is NULL and must be checked before accessing.

---
   drivers/gpu/drm/drm_simple_kms_helper.c | 6 ++++--
   1 file changed, 4 insertions(+), 2 deletions(-)
drivers/gpu/drm/drm_simple_kms_helper.c +2 -2
index 9ca8a4a59b74..a05eca9cec8b 100644
@@ -121,8 +121,10 @@ static int drm_simple_kms_plane_atomic_check(struct drm_plane *plane,
   	pipe = container_of(plane, struct drm_simple_display_pipe, plane);
   	crtc_state = drm_atomic_get_new_crtc_state(plane_state->state,
   						   &pipe->crtc);
-	if (!crtc_state->enable)
-		return 0; /* nothing to check when disabling or disabled */
+
+	if (!crtc_state || !crtc_state->enable)
Quoted 8 lines

drivers/gpu/drm/drm_simple_kms_helper.c static int drm_simple_kms_plane_atomic_check(struct drm_plane *plane, · line 129

Quoted 4 lines
drivers/gpu/drm/drm_simple_kms_helper.c +2 -0
+		/* nothing to check when disabling or disabled or no CRTC set */
+		return 0;
   	if (crtc_state->enable)
   		drm_mode_get_hv_timing(&crtc_state->mode,

Hm, this is a bit annoying, since the can_position = false parameter to drm_atomic_helper_check_plane_state is supposed to catch all this. Would moving all the checks after the call to that helper, and gating them on plane_state->visible also work?

Quoted 1 line

Yes, it does work if this is what you mean:

I wasn't sure, thanks for figuring this out!

Quoted 23 lines
drivers/gpu/drm/drm_simple_kms_helper.c +0 -2
index a05eca9cec8b..c48858bb2823 100644
@@ -122,14 +122,6 @@ static int drm_simple_kms_plane_atomic_check(struct
 drm_plane *plane,
         crtc_state = drm_atomic_get_new_crtc_state(plane_state->state,
 &pipe->crtc);
 
-       if (!crtc_state || !crtc_state->enable)
-               /* nothing to check when disabling or disabled or no CRTC

set */
-               return 0;

drivers/gpu/drm/drm_simple_kms_helper.c +0 -5
-
-       if (crtc_state->enable)
-               drm_mode_get_hv_timing(&crtc_state->mode,
-                                      &clip.x2, &clip.y2);
-
         ret = drm_atomic_helper_check_plane_state(plane_state, crtc_state,
                                                   &clip,

DRM_PLANE_HELPER_NO_SCALING,

drivers/gpu/drm/drm_simple_kms_helper.c static int drm_simple_kms_plane_atomic_check(struct · line 135

Quoted 6 lines
drivers/gpu/drm/drm_simple_kms_helper.c +2 -0
@@ -138,6 +130,13 @@ static int drm_simple_kms_plane_atomic_check(struct
 drm_plane *plane,
         if (ret)
                 return ret;
 
+       if (!plane_state->visible || !crtc_state->enable)
+               return 0; /* nothing to check when disabling or disabled */
if (!plane_state->visible) {
	WARN_ON(crtc_state->enabled);
	return 0;
}
Quoted 3 lines
Quoted 1 line

The helper call above should guarantee this.

Yes, but I still see cases when crtc_state is NULL, thus making crtc_state->enable to fail

Right, when the plane is completely off there's no CRTC state. Correct check should be

	WARN_ON(crtc_state && crtc_state->enabled);
Quoted 148 lines
Quoted 17 lines

drivers/gpu/drm/drm_simple_kms_helper.c static int drm_simple_kms_plane_atomic_check(struct · line 137

Quoted 2 lines
drivers/gpu/drm/drm_simple_kms_helper.c +2 -0
+
+       if (plane_state->visible && crtc_state->enable)

Similar here.

drivers/gpu/drm/drm_simple_kms_helper.c static int drm_simple_kms_plane_atomic_check(struct · line 142

Quoted 5 lines
drivers/gpu/drm/drm_simple_kms_helper.c +3 -0
+               drm_mode_get_hv_timing(&crtc_state->mode,
+                                      &clip.x2, &clip.y2);
+
         if (!plane_state->visible)
                 return -EINVAL;

This can now be removed, the plane helper takes care of checking for plane_state->visible != crtc_state->enable. Please also remove.

Quoted 5 lines
Quoted 4 lines

We'd need to add a guarantee to drm_atomic_helper_check_plane_state that it can cope with crtc_state == NULL, but I think that's a good idea anyway. Atm it shouldn't end up looking at the crtc_state pointer in that case.

It doesn't look at it at the moment

Quoted 8 lines
Quoted 7 lines
Quoted 2 lines

Otherwise we'll just go with your fix, but it feels all a bit too fragile, hence why I want to explore more robust options a bit.

At list with the change above it passes my test which failed before. Although I cannot confirm it works for others, but it certainly does for me.

Quoted 1 line

-Daniel

Do you want me to send v1 with the code above?

Yes please, with my above cleanup suggestions.

Please see the patch under test attached (I believe it is what you mean, with the only change that
    if (!plane_state->visible) {
        *if (crtc_state)*
            WARN_ON(crtc_state->enable);
        return 0;
    }
check is used).

Whith this patch + additional logs I have:

[   18.939204] [drm:drm_ioctl [drm]] pid=2105, dev=0xe200, auth=1, DRM_IOCTL_MODE_ATOMIC
[...]
[   18.939681] [drm:drm_atomic_set_crtc_for_plane [drm]] Link plane state 00000000c302cbbf to [NOCRTC]
[   18.939822] [drm:drm_atomic_set_fb_for_plane [drm]] Set [NOFB] for plane state 00000000c302cbbf
[   18.939963] [drm:drm_atomic_print_state [drm]] checking 000000000bc224e7 [   18.939988] vdispl vdispl.0: [drm] plane[29]: plane-0
[   18.940003] vdispl vdispl.0: [drm]   crtc=(null)
[   18.940018] vdispl vdispl.0: [drm]   fb=0
[   18.940032] vdispl vdispl.0: [drm]   crtc-pos=0x0+0+0
[   18.940048] vdispl vdispl.0: [drm] src-pos=0.000000x0.000000+0.000000+0.000000
[   18.940067] vdispl vdispl.0: [drm]   rotation=1
[   18.940199] [drm:drm_atomic_check_only [drm]] checking 000000000bc224e7 [   18.940226] ================================= plane_state->visible 0 crtc_state           (null)
[...]
[   18.978146] [drm:drm_atomic_set_crtc_for_plane [drm]] Link plane state 000000006bd50580 to [CRTC:30:crtc-0]
[   18.978292] [drm:drm_atomic_set_fb_for_plane [drm]] Set [FB:35] for plane state 000000006bd50580
[   18.978993] [drm:drm_atomic_set_mode_prop_for_crtc [drm]] Set [MODE:1024x768] for CRTC state 00000000e5a28f6a
[   18.979425] [drm:drm_atomic_check_only [drm]] checking 000000000bc224e7 [   18.979540] [drm:drm_atomic_helper_check_modeset [drm_kms_helper]] [CRTC:30:crtc-0] mode changed
[   18.979632] [drm:drm_atomic_helper_check_modeset [drm_kms_helper]] [CRTC:30:crtc-0] enable changed
[   18.979708] [drm:drm_atomic_helper_check_modeset [drm_kms_helper]] [CRTC:30:crtc-0] active changed
[   18.979792] [drm:drm_atomic_helper_check_modeset [drm_kms_helper]] Updating routing for [CONNECTOR:28:Virtual-1]
[   18.979877] [drm:drm_atomic_helper_check_modeset [drm_kms_helper]] [CONNECTOR:28:Virtual-1] using [ENCODER:31:None-31] on [CRTC:30:crtc-0] [   18.979960] [drm:drm_atomic_helper_check_modeset [drm_kms_helper]] [CRTC:30:crtc-0] needs all connectors, enable: y, active: y
[   18.980139] [drm:drm_atomic_add_affected_connectors [drm]] Adding all current connectors for [CRTC:30:crtc-0] to 000000000bc224e7
[   18.980184] ================================= plane_state->visible 0 crtc_state 00000000e5a28f6a
[   18.980205] crtc_state->enable 1

*[   19.022608] WARNING: CPU: 1 PID: 2105 at drivers/gpu/drm/drm_simple_kms_helper.c:137 drm_simple_kms_plane_atomic_check+0xdc/0xf8 [drm_kms_helper]*

[...]

[   19.113601] ================================= plane_state->visible 0 crtc_state 00000000e5a28f6a
[   19.113623] crtc_state->enable 1
[   19.113792] WARNING: CPU: 1 PID: 2105 at drivers/gpu/drm/drm_simple_kms_helper.c:137 drm_simple_kms_plane_atomic_check+0xdc/0xf8 [drm_kms_helper]

[...]

And finally

[   19.340249] ================================= plane_state->visible 0 crtc_state 0000000036a1b1f5
[   19.340271] crtc_state->enable 0

So, it seems that crtc_state can still be NULL if "!plane_state->visible" making
NULL pointer dereference, so we need a check for that.
Yet, !plane_state->visible && crtc_state->enable makes WARN_ON to fire always. So, probably we may want removing it.

Quoted 2 lines

Thanks, Daniel

Thank you,
Oleksandr

From dbcce708b237740158a2c16029c56a579324f269 Mon Sep 17 00:00:00 2001
From: Oleksandr Andrushchenko <[email protected]>
Date: Tue, 13 Feb 2018 10:32:20 +0200
Subject: [PATCH] drm/simple_kms_helper: Fix NULL pointer dereference with no
active CRTC

It is possible that drm_simple_kms_plane_atomic_check called with no CRTC set, e.g. when user-space application sets CRTC_ID/FB_ID to 0 before doing any actual drawing. This leads to NULL pointer dereference because in this case new CRTC state is NULL and must be checked before accessing.

---
 drivers/gpu/drm/drm_simple_kms_helper.c | 15 +++++++--------
 1 file changed, 7 insertions(+), 8 deletions(-)
drivers/gpu/drm/drm_simple_kms_helper.c +0 -6
index 9ca8a4a59b74..f54711ff9767 100644
@@ -121,12 +121,6 @@ static int drm_simple_kms_plane_atomic_check(struct drm_plane *plane,
 	pipe = container_of(plane, struct drm_simple_display_pipe, plane);
 	crtc_state = drm_atomic_get_new_crtc_state(plane_state->state,
 						   &pipe->crtc);
-	if (!crtc_state->enable)
-		return 0; /* nothing to check when disabling or disabled */
-
-	if (crtc_state->enable)
-		drm_mode_get_hv_timing(&crtc_state->mode,
-				       &clip.x2, &clip.y2);
 
 	ret = drm_atomic_helper_check_plane_state(plane_state, crtc_state,
 						  &clip,
@@ -136,8 +130,13 @@ static int drm_simple_kms_plane_atomic_check(struct drm_plane *plane,
 	if (ret)
 		return ret;

drivers/gpu/drm/drm_simple_kms_helper.c static int drm_simple_kms_plane_atomic_check(struct drm_plane *plane, · line 139

Quoted 10 lines
drivers/gpu/drm/drm_simple_kms_helper.c +7 -2
 
-	if (!plane_state->visible)
-		return -EINVAL;
+	if (!plane_state->visible) {
+		if (crtc_state)
+			WARN_ON(crtc_state->enable);
+		return 0;
+	}
+
+	drm_mode_get_hv_timing(&crtc_state->mode, &clip.x2, &clip.y2);

lgtm. With or without the bikeshed to pull the crtc_state check into the WARN_ON.

Please resubmit as a stand-alone patch, patchwork can't pull patches out of attachments :-/
-Daniel

Quoted 11 lines
drivers/gpu/drm/drm_simple_kms_helper.c +0 -0
 	if (!pipe->funcs || !pipe->funcs->check)
 		return 0;
Signature · 7 lines
2.7.4


_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel
Signature · 7 lines
Daniel Vetter
Software Engineer, Intel Corporation
http://blog.ffwll.ch
_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel
#6 lore raw
Quoted 75 lines · Daniel Vetter
Quoted 74 lines · Oleksandr Andrushchenko
Quoted 73 lines · Daniel Vetter
Quoted 35 lines · Oleksandr Andrushchenko
Quoted 26 lines · Daniel Vetter
Quoted 25 lines · Oleksandr Andrushchenko

From: Oleksandr Andrushchenko <[email protected]>

It is possible that drm_simple_kms_plane_atomic_check called with no CRTC set, e.g. when user-space application sets CRTC_ID/FB_ID to 0 before doing any actual drawing. This leads to NULL pointer dereference because in this case new CRTC state is NULL and must be checked before accessing.

---
    drivers/gpu/drm/drm_simple_kms_helper.c | 6 ++++--
    1 file changed, 4 insertions(+), 2 deletions(-)
drivers/gpu/drm/drm_simple_kms_helper.c +2 -2
index 9ca8a4a59b74..a05eca9cec8b 100644
@@ -121,8 +121,10 @@ static int drm_simple_kms_plane_atomic_check(struct drm_plane *plane,
    	pipe = container_of(plane, struct drm_simple_display_pipe, plane);
    	crtc_state = drm_atomic_get_new_crtc_state(plane_state->state,
    						   &pipe->crtc);
-	if (!crtc_state->enable)
-		return 0; /* nothing to check when disabling or disabled */
+
+	if (!crtc_state || !crtc_state->enable)
Quoted 8 lines

drivers/gpu/drm/drm_simple_kms_helper.c static int drm_simple_kms_plane_atomic_check(struct drm_plane *plane, · line 129

Quoted 4 lines
drivers/gpu/drm/drm_simple_kms_helper.c +2 -0
+		/* nothing to check when disabling or disabled or no CRTC set */
+		return 0;
    	if (crtc_state->enable)
    		drm_mode_get_hv_timing(&crtc_state->mode,

Hm, this is a bit annoying, since the can_position = false parameter to drm_atomic_helper_check_plane_state is supposed to catch all this. Would moving all the checks after the call to that helper, and gating them on plane_state->visible also work?

Quoted 1 line

Yes, it does work if this is what you mean:

I wasn't sure, thanks for figuring this out!

Quoted 23 lines
drivers/gpu/drm/drm_simple_kms_helper.c +0 -2
index a05eca9cec8b..c48858bb2823 100644
@@ -122,14 +122,6 @@ static int drm_simple_kms_plane_atomic_check(struct
 drm_plane *plane,
          crtc_state = drm_atomic_get_new_crtc_state(plane_state->state,
 &pipe->crtc);
 
-       if (!crtc_state || !crtc_state->enable)
-               /* nothing to check when disabling or disabled or no CRTC

set */
-               return 0;

drivers/gpu/drm/drm_simple_kms_helper.c +0 -5
-
-       if (crtc_state->enable)
-               drm_mode_get_hv_timing(&crtc_state->mode,
-                                      &clip.x2, &clip.y2);
-
          ret = drm_atomic_helper_check_plane_state(plane_state, crtc_state,
                                                    &clip,

DRM_PLANE_HELPER_NO_SCALING,

drivers/gpu/drm/drm_simple_kms_helper.c static int drm_simple_kms_plane_atomic_check(struct · line 135

Quoted 6 lines
drivers/gpu/drm/drm_simple_kms_helper.c +2 -0
@@ -138,6 +130,13 @@ static int drm_simple_kms_plane_atomic_check(struct
 drm_plane *plane,
          if (ret)
                  return ret;
 
+       if (!plane_state->visible || !crtc_state->enable)
+               return 0; /* nothing to check when disabling or disabled */
if (!plane_state->visible) {
	WARN_ON(crtc_state->enabled);
	return 0;
}
Quoted 7 lines
Quoted 3 lines
Quoted 1 line

The helper call above should guarantee this.

Yes, but I still see cases when crtc_state is NULL, thus making crtc_state->enable to fail

Right, when the plane is completely off there's no CRTC state. Correct check should be

	WARN_ON(crtc_state && crtc_state->enabled);

ok, will update with this additional check

Quoted 160 lines
Quoted 147 lines
Quoted 18 lines

drivers/gpu/drm/drm_simple_kms_helper.c static int drm_simple_kms_plane_atomic_check(struct · line 137

Quoted 3 lines
drivers/gpu/drm/drm_simple_kms_helper.c +2 -0
+
+       if (plane_state->visible && crtc_state->enable)

Similar here.

drivers/gpu/drm/drm_simple_kms_helper.c static int drm_simple_kms_plane_atomic_check(struct · line 142

Quoted 5 lines
drivers/gpu/drm/drm_simple_kms_helper.c +3 -0
+               drm_mode_get_hv_timing(&crtc_state->mode,
+                                      &clip.x2, &clip.y2);
+
          if (!plane_state->visible)
                  return -EINVAL;

This can now be removed, the plane helper takes care of checking for plane_state->visible != crtc_state->enable. Please also remove.

Quoted 5 lines
Quoted 4 lines

We'd need to add a guarantee to drm_atomic_helper_check_plane_state that it can cope with crtc_state == NULL, but I think that's a good idea anyway. Atm it shouldn't end up looking at the crtc_state pointer in that case.

It doesn't look at it at the moment

Quoted 8 lines
Quoted 7 lines
Quoted 2 lines

Otherwise we'll just go with your fix, but it feels all a bit too fragile, hence why I want to explore more robust options a bit.

At list with the change above it passes my test which failed before. Although I cannot confirm it works for others, but it certainly does for me.

Quoted 1 line

-Daniel

Do you want me to send v1 with the code above?

Yes please, with my above cleanup suggestions.

Please see the patch under test attached (I believe it is what you mean, with the only change that
    if (!plane_state->visible) {
        *if (crtc_state)*
            WARN_ON(crtc_state->enable);
        return 0;
    }
check is used).

Whith this patch + additional logs I have:

[   18.939204] [drm:drm_ioctl [drm]] pid=2105, dev=0xe200, auth=1, DRM_IOCTL_MODE_ATOMIC
[...]
[   18.939681] [drm:drm_atomic_set_crtc_for_plane [drm]] Link plane state 00000000c302cbbf to [NOCRTC]
[   18.939822] [drm:drm_atomic_set_fb_for_plane [drm]] Set [NOFB] for plane state 00000000c302cbbf
[   18.939963] [drm:drm_atomic_print_state [drm]] checking 000000000bc224e7 [   18.939988] vdispl vdispl.0: [drm] plane[29]: plane-0
[   18.940003] vdispl vdispl.0: [drm]   crtc=(null)
[   18.940018] vdispl vdispl.0: [drm]   fb=0
[   18.940032] vdispl vdispl.0: [drm]   crtc-pos=0x0+0+0
[   18.940048] vdispl vdispl.0: [drm] src-pos=0.000000x0.000000+0.000000+0.000000
[   18.940067] vdispl vdispl.0: [drm]   rotation=1
[   18.940199] [drm:drm_atomic_check_only [drm]] checking 000000000bc224e7 [   18.940226] ================================= plane_state->visible 0 crtc_state           (null)
[...]
[   18.978146] [drm:drm_atomic_set_crtc_for_plane [drm]] Link plane state 000000006bd50580 to [CRTC:30:crtc-0]
[   18.978292] [drm:drm_atomic_set_fb_for_plane [drm]] Set [FB:35] for plane state 000000006bd50580
[   18.978993] [drm:drm_atomic_set_mode_prop_for_crtc [drm]] Set [MODE:1024x768] for CRTC state 00000000e5a28f6a
[   18.979425] [drm:drm_atomic_check_only [drm]] checking 000000000bc224e7 [   18.979540] [drm:drm_atomic_helper_check_modeset [drm_kms_helper]] [CRTC:30:crtc-0] mode changed
[   18.979632] [drm:drm_atomic_helper_check_modeset [drm_kms_helper]] [CRTC:30:crtc-0] enable changed
[   18.979708] [drm:drm_atomic_helper_check_modeset [drm_kms_helper]] [CRTC:30:crtc-0] active changed
[   18.979792] [drm:drm_atomic_helper_check_modeset [drm_kms_helper]] Updating routing for [CONNECTOR:28:Virtual-1]
[   18.979877] [drm:drm_atomic_helper_check_modeset [drm_kms_helper]] [CONNECTOR:28:Virtual-1] using [ENCODER:31:None-31] on [CRTC:30:crtc-0] [   18.979960] [drm:drm_atomic_helper_check_modeset [drm_kms_helper]] [CRTC:30:crtc-0] needs all connectors, enable: y, active: y
[   18.980139] [drm:drm_atomic_add_affected_connectors [drm]] Adding all current connectors for [CRTC:30:crtc-0] to 000000000bc224e7
[   18.980184] ================================= plane_state->visible 0 crtc_state 00000000e5a28f6a
[   18.980205] crtc_state->enable 1

*[   19.022608] WARNING: CPU: 1 PID: 2105 at drivers/gpu/drm/drm_simple_kms_helper.c:137 drm_simple_kms_plane_atomic_check+0xdc/0xf8 [drm_kms_helper]*

[...]

[   19.113601] ================================= plane_state->visible 0 crtc_state 00000000e5a28f6a
[   19.113623] crtc_state->enable 1
[   19.113792] WARNING: CPU: 1 PID: 2105 at drivers/gpu/drm/drm_simple_kms_helper.c:137 drm_simple_kms_plane_atomic_check+0xdc/0xf8 [drm_kms_helper]

[...]

And finally

[   19.340249] ================================= plane_state->visible 0 crtc_state 0000000036a1b1f5
[   19.340271] crtc_state->enable 0

So, it seems that crtc_state can still be NULL if "!plane_state->visible" making
NULL pointer dereference, so we need a check for that.
Yet, !plane_state->visible && crtc_state->enable makes WARN_ON to fire always. So, probably we may want removing it.

Quoted 1 line

Thanks, Daniel

Thank you,
Oleksandr
From dbcce708b237740158a2c16029c56a579324f269 Mon Sep 17 00:00:00 2001
From: Oleksandr Andrushchenko <[email protected]>
Date: Tue, 13 Feb 2018 10:32:20 +0200
Subject: [PATCH] drm/simple_kms_helper: Fix NULL pointer dereference with no
active CRTC

It is possible that drm_simple_kms_plane_atomic_check called with no CRTC set, e.g. when user-space application sets CRTC_ID/FB_ID to 0 before doing any actual drawing. This leads to NULL pointer dereference because in this case new CRTC state is NULL and must be checked before accessing.

---
  drivers/gpu/drm/drm_simple_kms_helper.c | 15 +++++++--------
  1 file changed, 7 insertions(+), 8 deletions(-)
drivers/gpu/drm/drm_simple_kms_helper.c +0 -6
index 9ca8a4a59b74..f54711ff9767 100644
@@ -121,12 +121,6 @@ static int drm_simple_kms_plane_atomic_check(struct drm_plane *plane,
  	pipe = container_of(plane, struct drm_simple_display_pipe, plane);
  	crtc_state = drm_atomic_get_new_crtc_state(plane_state->state,
  						   &pipe->crtc);
-	if (!crtc_state->enable)
-		return 0; /* nothing to check when disabling or disabled */
-
-	if (crtc_state->enable)
-		drm_mode_get_hv_timing(&crtc_state->mode,
-				       &clip.x2, &clip.y2);
 
  	ret = drm_atomic_helper_check_plane_state(plane_state, crtc_state,
  						  &clip,
@@ -136,8 +130,13 @@ static int drm_simple_kms_plane_atomic_check(struct drm_plane *plane,
  	if (ret)
  		return ret;

drivers/gpu/drm/drm_simple_kms_helper.c static int drm_simple_kms_plane_atomic_check(struct drm_plane *plane, · line 139

Quoted 10 lines
drivers/gpu/drm/drm_simple_kms_helper.c +7 -2
 
-	if (!plane_state->visible)
-		return -EINVAL;
+	if (!plane_state->visible) {
+		if (crtc_state)
+			WARN_ON(crtc_state->enable);
+		return 0;
+	}
+
+	drm_mode_get_hv_timing(&crtc_state->mode, &clip.x2, &clip.y2);

lgtm. With or without the bikeshed to pull the crtc_state check into the WARN_ON.

Quoted 1 line

Thank you

Quoted 3 lines

Please resubmit as a stand-alone patch, patchwork can't pull patches out of attachments :-/

oh, that was for demonstration purpose only, so we are on the same page and see the patch we are discussing ;)

Quoted 13 lines

-Daniel

Quoted 10 lines
drivers/gpu/drm/drm_simple_kms_helper.c +0 -0
  	if (!pipe->funcs || !pipe->funcs->check)
  		return 0;
Signature · 6 lines
2.7.4

_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel

_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel

#7 lore raw
Quoted 83 lines · Oleksandr Andrushchenko
Quoted 82 lines · Daniel Vetter
Quoted 81 lines · Oleksandr Andrushchenko
Quoted 80 lines · Daniel Vetter
Quoted 38 lines · Oleksandr Andrushchenko
Quoted 29 lines · Daniel Vetter
Quoted 28 lines · Oleksandr Andrushchenko

From: Oleksandr Andrushchenko <[email protected]>

It is possible that drm_simple_kms_plane_atomic_check called with no CRTC set, e.g. when user-space application sets CRTC_ID/FB_ID to 0 before doing any actual drawing. This leads to NULL pointer dereference because in this case new CRTC state is NULL and must be checked before accessing.

---
    drivers/gpu/drm/drm_simple_kms_helper.c | 6 ++++--

    1 file changed, 4 insertions(+), 2 deletions(-)
drivers/gpu/drm/drm_simple_kms_helper.c +0 -2
index 9ca8a4a59b74..a05eca9cec8b 100644
@@ -121,8 +121,10 @@ static int 
 drm_simple_kms_plane_atomic_check(struct drm_plane *plane,
         pipe = container_of(plane, struct drm_simple_display_pipe, plane);
         crtc_state = drm_atomic_get_new_crtc_state(plane_state->state,
                                &pipe->crtc);
-    if (!crtc_state->enable)
-        return 0; /* nothing to check when disabling or 

disabled */
+
+    if (!crtc_state || !crtc_state->enable)
+        /* nothing to check when disabling or disabled or no

Quoted 8 lines
Quoted 4 lines

CRTC set */
+        return 0;
        if (crtc_state->enable)
drm_mode_get_hv_timing(&crtc_state->mode,

Hm, this is a bit annoying, since the can_position = false parameter to drm_atomic_helper_check_plane_state is supposed to catch all this. Would moving all the checks after the call to that helper, and gating them on plane_state->visible also work?

Quoted 1 line

Yes, it does work if this is what you mean:

I wasn't sure, thanks for figuring this out!

Quoted 30 lines
drivers/gpu/drm/drm_simple_kms_helper.c +0 -0
index a05eca9cec8b..c48858bb2823 100644
@@ -122,14 +122,6 @@ static int 

drm_simple_kms_plane_atomic_check(struct
drm_plane *plane,
          crtc_state = drm_atomic_get_new_crtc_state(plane_state->state, &pipe->crtc);

-       if (!crtc_state || !crtc_state->enable)
-               /* nothing to check when disabling or disabled or no CRTC
set */
-               return 0;

drivers/gpu/drm/drm_simple_kms_helper.c +0 -5
-
-       if (crtc_state->enable)
- drm_mode_get_hv_timing(&crtc_state->mode,
-                                      &clip.x2, &clip.y2);
-

          ret = drm_atomic_helper_check_plane_state(plane_state, crtc_state, &clip,
DRM_PLANE_HELPER_NO_SCALING,

drivers/gpu/drm/drm_simple_kms_helper.c +0 -0
@@ -138,6 +130,13 @@ static int 

drm_simple_kms_plane_atomic_check(struct
drm_plane *plane,
          if (ret)
                  return ret;

Quoted 3 lines

+       if (!plane_state->visible || !crtc_state->enable) +               return 0; /* nothing to check when disabling or disabled */

if (!plane_state->visible) {
    WARN_ON(crtc_state->enabled);
    return 0;
}

Quoted 8 lines
Quoted 7 lines
Quoted 3 lines
Quoted 1 line

The helper call above should guarantee this.

Yes, but I still see cases when crtc_state is NULL, thus making crtc_state->enable to fail

Right, when the plane is completely off there's no CRTC state. Correct check should be

    WARN_ON(crtc_state && crtc_state->enabled);

ok, will update with this additional check

huh, this indeed solves the NULL pointer dereference, but floods a lot with every page flip I have, e.g. !plane_state->visible == true and crtc_state is not NULL and crtc_state->enable == true, thus firing WARN_ON.
Is this something wrong with my use-case/driver or it is still legal to have such a configuration and leave it without WARN_ON and just return 0?

Quoted 160 lines
Quoted 160 lines
Quoted 147 lines
Quoted 18 lines
Quoted 3 lines

+
+       if (plane_state->visible && crtc_state->enable)

Similar here.

Quoted 5 lines

+ drm_mode_get_hv_timing(&crtc_state->mode,
+                                      &clip.x2, &clip.y2); +
          if (!plane_state->visible)
                  return -EINVAL;

This can now be removed, the plane helper takes care of checking for plane_state->visible != crtc_state->enable. Please also remove.

Quoted 5 lines
Quoted 4 lines

We'd need to add a guarantee to drm_atomic_helper_check_plane_state that it can cope with crtc_state == NULL, but I think that's a good idea anyway. Atm it shouldn't end up looking at the crtc_state pointer in that case.

It doesn't look at it at the moment

Quoted 8 lines
Quoted 7 lines
Quoted 2 lines

Otherwise we'll just go with your fix, but it feels all a bit too fragile, hence why I want to explore more robust options a bit.

At list with the change above it passes my test which failed before. Although I cannot confirm it works for others, but it certainly does for me.

Quoted 1 line

-Daniel

Do you want me to send v1 with the code above?

Yes please, with my above cleanup suggestions.

Please see the patch under test attached (I believe it is what you mean, with the only change that
     if (!plane_state->visible) {
         *if (crtc_state)*
             WARN_ON(crtc_state->enable);
         return 0;
     }
check is used).

Whith this patch + additional logs I have:

[   18.939204] [drm:drm_ioctl [drm]] pid=2105, dev=0xe200, auth=1, DRM_IOCTL_MODE_ATOMIC
[...]
[   18.939681] [drm:drm_atomic_set_crtc_for_plane [drm]] Link plane state 00000000c302cbbf to [NOCRTC]
[   18.939822] [drm:drm_atomic_set_fb_for_plane [drm]] Set [NOFB] for plane state 00000000c302cbbf
[   18.939963] [drm:drm_atomic_print_state [drm]] checking 000000000bc224e7 [   18.939988] vdispl vdispl.0: [drm] plane[29]: plane-0
[   18.940003] vdispl vdispl.0: [drm]   crtc=(null)
[   18.940018] vdispl vdispl.0: [drm]   fb=0
[   18.940032] vdispl vdispl.0: [drm]   crtc-pos=0x0+0+0
[   18.940048] vdispl vdispl.0: [drm] src-pos=0.000000x0.000000+0.000000+0.000000
[   18.940067] vdispl vdispl.0: [drm]   rotation=1
[   18.940199] [drm:drm_atomic_check_only [drm]] checking 000000000bc224e7 [   18.940226] ================================= plane_state->visible 0 crtc_state           (null)
[...]
[   18.978146] [drm:drm_atomic_set_crtc_for_plane [drm]] Link plane state 000000006bd50580 to [CRTC:30:crtc-0]
[   18.978292] [drm:drm_atomic_set_fb_for_plane [drm]] Set [FB:35] for plane state 000000006bd50580
[   18.978993] [drm:drm_atomic_set_mode_prop_for_crtc [drm]] Set [MODE:1024x768] for CRTC state 00000000e5a28f6a
[   18.979425] [drm:drm_atomic_check_only [drm]] checking 000000000bc224e7 [   18.979540] [drm:drm_atomic_helper_check_modeset [drm_kms_helper]] [CRTC:30:crtc-0] mode changed
[   18.979632] [drm:drm_atomic_helper_check_modeset [drm_kms_helper]] [CRTC:30:crtc-0] enable changed
[   18.979708] [drm:drm_atomic_helper_check_modeset [drm_kms_helper]] [CRTC:30:crtc-0] active changed
[   18.979792] [drm:drm_atomic_helper_check_modeset [drm_kms_helper]] Updating routing for [CONNECTOR:28:Virtual-1]
[   18.979877] [drm:drm_atomic_helper_check_modeset [drm_kms_helper]] [CONNECTOR:28:Virtual-1] using [ENCODER:31:None-31] on [CRTC:30:crtc-0] [   18.979960] [drm:drm_atomic_helper_check_modeset [drm_kms_helper]] [CRTC:30:crtc-0] needs all connectors, enable: y, active: y
[   18.980139] [drm:drm_atomic_add_affected_connectors [drm]] Adding all current connectors for [CRTC:30:crtc-0] to 000000000bc224e7
[   18.980184] ================================= plane_state->visible 0 crtc_state 00000000e5a28f6a
[   18.980205] crtc_state->enable 1

*[   19.022608] WARNING: CPU: 1 PID: 2105 at drivers/gpu/drm/drm_simple_kms_helper.c:137 drm_simple_kms_plane_atomic_check+0xdc/0xf8 [drm_kms_helper]*

[...]

[   19.113601] ================================= plane_state->visible 0 crtc_state 00000000e5a28f6a
[   19.113623] crtc_state->enable 1
[   19.113792] WARNING: CPU: 1 PID: 2105 at drivers/gpu/drm/drm_simple_kms_helper.c:137 drm_simple_kms_plane_atomic_check+0xdc/0xf8 [drm_kms_helper]

[...]

And finally

[   19.340249] ================================= plane_state->visible 0 crtc_state 0000000036a1b1f5
[   19.340271] crtc_state->enable 0

So, it seems that crtc_state can still be NULL if "!plane_state->visible" making
NULL pointer dereference, so we need a check for that.
Yet, !plane_state->visible && crtc_state->enable makes WARN_ON to fire always. So, probably we may want removing it.

Quoted 1 line

Thanks, Daniel

Thank you,
Oleksandr
 From dbcce708b237740158a2c16029c56a579324f269 Mon Sep 17 00:00:00 2001
From: Oleksandr Andrushchenko <[email protected]>
Date: Tue, 13 Feb 2018 10:32:20 +0200
Subject: [PATCH] drm/simple_kms_helper: Fix NULL pointer dereference with no active CRTC

It is possible that drm_simple_kms_plane_atomic_check called with no CRTC set, e.g. when user-space application sets CRTC_ID/FB_ID to 0 before doing any actual drawing. This leads to NULL pointer dereference because in this case new CRTC state is NULL and must be checked before accessing.

---
  drivers/gpu/drm/drm_simple_kms_helper.c | 15 +++++++--------

  1 file changed, 7 insertions(+), 8 deletions(-)
drivers/gpu/drm/drm_simple_kms_helper.c +0 -6
index 9ca8a4a59b74..f54711ff9767 100644
@@ -121,12 +121,6 @@ static int 
 drm_simple_kms_plane_atomic_check(struct drm_plane *plane,
       pipe = container_of(plane, struct drm_simple_display_pipe, plane);
       crtc_state = drm_atomic_get_new_crtc_state(plane_state->state,
                              &pipe->crtc);
-    if (!crtc_state->enable)
-        return 0; /* nothing to check when disabling or disabled */
-
-    if (crtc_state->enable)
-        drm_mode_get_hv_timing(&crtc_state->mode,
-                       &clip.x2, &clip.y2);
         ret = drm_atomic_helper_check_plane_state(plane_state, crtc_state,
                             &clip,
@@ -136,8 +130,13 @@ static int 
 drm_simple_kms_plane_atomic_check(struct drm_plane *plane,
       if (ret)

drivers/gpu/drm/drm_simple_kms_helper.c static int · line 140

Quoted 10 lines
drivers/gpu/drm/drm_simple_kms_helper.c +7 -1
           return ret;
   -    if (!plane_state->visible)
-        return -EINVAL;
+    if (!plane_state->visible) {
+        if (crtc_state)
+            WARN_ON(crtc_state->enable);
+        return 0;
+    }
+
+    drm_mode_get_hv_timing(&crtc_state->mode, &clip.x2, &clip.y2);

lgtm. With or without the bikeshed to pull the crtc_state check into the WARN_ON.

Quoted 20 lines
Quoted 1 line

Thank you

Quoted 3 lines

Please resubmit as a stand-alone patch, patchwork can't pull patches out of attachments :-/

oh, that was for demonstration purpose only, so we are on the same page and see the patch we are discussing ;)

Quoted 12 lines

-Daniel

Quoted 9 lines
drivers/gpu/drm/drm_simple_kms_helper.c +0 -0
         if (!pipe->funcs || !pipe->funcs->check)
           return 0;
Signature · 6 lines
2.7.4

_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel

_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel

#8 lore raw
Quoted 106 lines · Oleksandr Andrushchenko
Quoted 105 lines · Oleksandr Andrushchenko
Quoted 104 lines · Daniel Vetter
Quoted 96 lines · Oleksandr Andrushchenko
Quoted 95 lines · Daniel Vetter
Quoted 49 lines · Oleksandr Andrushchenko
Quoted 40 lines · Daniel Vetter
Quoted 38 lines · Oleksandr Andrushchenko

From: Oleksandr Andrushchenko <[email protected]>

It is possible that drm_simple_kms_plane_atomic_check called with no CRTC set, e.g. when user-space
application sets CRTC_ID/FB_ID
to 0 before doing any actual drawing. This leads to NULL pointer dereference because in this case new CRTC state is NULL and must be checked before accessing.

Signed-off-by: Oleksandr Andrushchenko <[email protected]>
---
    drivers/gpu/drm/drm_simple_kms_helper.c | 6 ++++--

    1 file changed, 4 insertions(+), 2 deletions(-)

diff --git
a/drivers/gpu/drm/drm_simple_kms_helper.c
b/drivers/gpu/drm/drm_simple_kms_helper.c
index 9ca8a4a59b74..a05eca9cec8b 100644

drivers/gpu/drm/drm_simple_kms_helper.c +0 -0
@@ -121,8 +121,10 @@ static int

drm_simple_kms_plane_atomic_check(struct
drm_plane *plane,
        pipe = container_of(plane, struct
drm_simple_display_pipe, plane);
        crtc_state =
drm_atomic_get_new_crtc_state(plane_state->state,
                               &pipe->crtc);
-    if (!crtc_state->enable)
-        return 0; /* nothing to check when
disabling or disabled */
+
+    if (!crtc_state || !crtc_state->enable)
+        /* nothing to check when disabling or
disabled or no CRTC set */
+        return 0;
        if (crtc_state->enable)

Quoted 8 lines
Quoted 1 line

drm_mode_get_hv_timing(&crtc_state->mode,

Hm, this is a bit annoying, since the can_position = false parameter to drm_atomic_helper_check_plane_state is supposed to catch all this. Would
moving all the checks after the call to that helper, and gating them on
plane_state->visible also work?

Quoted 1 line

Yes, it does work if this is what you mean:

I wasn't sure, thanks for figuring this out!

Quoted 33 lines
drivers/gpu/drm/drm_simple_kms_helper.c +0 -0
index a05eca9cec8b..c48858bb2823 100644
@@ -122,14 +122,6 @@ static int

drm_simple_kms_plane_atomic_check(struct
drm_plane *plane,
          crtc_state =
drm_atomic_get_new_crtc_state(plane_state->state,
&pipe->crtc);

-       if (!crtc_state || !crtc_state->enable) -               /* nothing to check when disabling or disabled or no CRTC
set */
-               return 0;

drivers/gpu/drm/drm_simple_kms_helper.c +0 -5
-
-       if (crtc_state->enable)
- drm_mode_get_hv_timing(&crtc_state->mode,
-                                      &clip.x2, &clip.y2);
-

          ret =
drm_atomic_helper_check_plane_state(plane_state,
crtc_state,
&clip,
DRM_PLANE_HELPER_NO_SCALING,

drivers/gpu/drm/drm_simple_kms_helper.c +0 -0
@@ -138,6 +130,13 @@ static int

drm_simple_kms_plane_atomic_check(struct
drm_plane *plane,
          if (ret)
                  return ret;

Quoted 3 lines

+       if (!plane_state->visible || !crtc_state->enable) +               return 0; /* nothing to check when disabling or disabled */

if (!plane_state->visible) {
    WARN_ON(crtc_state->enabled);
    return 0;
}

Quoted 3 lines
Quoted 1 line

The helper call above should guarantee this.

Yes, but I still see cases when crtc_state is NULL, thus making crtc_state->enable to fail

Right, when the plane is completely off there's no CRTC state. Correct check should be

    WARN_ON(crtc_state && crtc_state->enabled);

Quoted 8 lines
Quoted 1 line

ok, will update with this additional check

huh, this indeed solves the NULL pointer dereference, but floods a lot with every page flip I have, e.g. !plane_state->visible == true and crtc_state is not NULL and crtc_state->enable == true, thus firing WARN_ON.
Is this something wrong with my use-case/driver or it is still legal to have such a configuration and leave it without WARN_ON and just return 0?

1 week of vacation later I have to admit that this WARN_ON is completely bogus :-)

Sorry for all the confusion, pls leave it out.
-Daniel

Quoted 203 lines
Quoted 178 lines
Quoted 178 lines
Quoted 165 lines
Quoted 23 lines
Quoted 3 lines

+
+       if (plane_state->visible && crtc_state->enable)

Similar here.

Quoted 5 lines

+ drm_mode_get_hv_timing(&crtc_state->mode,
+                                      &clip.x2, &clip.y2); +
          if (!plane_state->visible)
                  return -EINVAL;

This can now be removed, the plane helper takes care of checking for plane_state->visible != crtc_state->enable. Please also remove.

Quoted 10 lines
Quoted 6 lines

We'd need to add a guarantee to drm_atomic_helper_check_plane_state that
it can cope with crtc_state == NULL, but I think that's a good idea anyway. Atm it shouldn't end up looking at the crtc_state pointer in that
case.

It doesn't look at it at the moment

Quoted 3 lines

Otherwise we'll just go with your fix, but it feels all a bit too fragile,
hence why I want to explore more robust options a bit.

Quoted 6 lines
Quoted 5 lines

At list with the change above it passes my test which failed before. Although I cannot confirm it works for others, but it certainly does for me.

Quoted 1 line

-Daniel

Do you want me to send v1 with the code above?

Yes please, with my above cleanup suggestions.

Please see the patch under test attached (I believe it is what you mean,
with the only change that
     if (!plane_state->visible) {
         *if (crtc_state)* WARN_ON(crtc_state->enable);
         return 0;
     }
check is used).

Whith this patch + additional logs I have:

[   18.939204] [drm:drm_ioctl [drm]] pid=2105, dev=0xe200, auth=1, DRM_IOCTL_MODE_ATOMIC
[...]
[   18.939681] [drm:drm_atomic_set_crtc_for_plane [drm]] Link plane state
00000000c302cbbf to [NOCRTC]
[   18.939822] [drm:drm_atomic_set_fb_for_plane [drm]] Set [NOFB] for plane
state 00000000c302cbbf
[   18.939963] [drm:drm_atomic_print_state [drm]] checking 000000000bc224e7
[   18.939988] vdispl vdispl.0: [drm] plane[29]: plane-0
[   18.940003] vdispl vdispl.0: [drm]   crtc=(null)
[   18.940018] vdispl vdispl.0: [drm]   fb=0
[   18.940032] vdispl vdispl.0: [drm]   crtc-pos=0x0+0+0
[   18.940048] vdispl vdispl.0: [drm] src-pos=0.000000x0.000000+0.000000+0.000000
[   18.940067] vdispl vdispl.0: [drm]   rotation=1
[   18.940199] [drm:drm_atomic_check_only [drm]] checking 000000000bc224e7
[   18.940226] ================================= plane_state->visible 0 crtc_state           (null)
[...]
[   18.978146] [drm:drm_atomic_set_crtc_for_plane [drm]] Link plane state
000000006bd50580 to [CRTC:30:crtc-0]
[   18.978292] [drm:drm_atomic_set_fb_for_plane [drm]] Set [FB:35] for plane
state 000000006bd50580
[   18.978993] [drm:drm_atomic_set_mode_prop_for_crtc [drm]] Set [MODE:1024x768] for CRTC state 00000000e5a28f6a
[   18.979425] [drm:drm_atomic_check_only [drm]] checking 000000000bc224e7
[   18.979540] [drm:drm_atomic_helper_check_modeset [drm_kms_helper]] [CRTC:30:crtc-0] mode changed
[   18.979632] [drm:drm_atomic_helper_check_modeset [drm_kms_helper]] [CRTC:30:crtc-0] enable changed
[   18.979708] [drm:drm_atomic_helper_check_modeset [drm_kms_helper]] [CRTC:30:crtc-0] active changed
[   18.979792] [drm:drm_atomic_helper_check_modeset [drm_kms_helper]] Updating routing for [CONNECTOR:28:Virtual-1]
[   18.979877] [drm:drm_atomic_helper_check_modeset [drm_kms_helper]] [CONNECTOR:28:Virtual-1] using [ENCODER:31:None-31] on [CRTC:30:crtc-0] [   18.979960] [drm:drm_atomic_helper_check_modeset [drm_kms_helper]] [CRTC:30:crtc-0] needs all connectors, enable: y, active: y [   18.980139] [drm:drm_atomic_add_affected_connectors [drm]] Adding all
current connectors for [CRTC:30:crtc-0] to 000000000bc224e7 [   18.980184] ================================= plane_state->visible 0 crtc_state 00000000e5a28f6a
[   18.980205] crtc_state->enable 1

*[   19.022608] WARNING: CPU: 1 PID: 2105 at drivers/gpu/drm/drm_simple_kms_helper.c:137 drm_simple_kms_plane_atomic_check+0xdc/0xf8 [drm_kms_helper]*

[...]

[   19.113601] ================================= plane_state->visible 0 crtc_state 00000000e5a28f6a
[   19.113623] crtc_state->enable 1
[   19.113792] WARNING: CPU: 1 PID: 2105 at drivers/gpu/drm/drm_simple_kms_helper.c:137 drm_simple_kms_plane_atomic_check+0xdc/0xf8 [drm_kms_helper]

[...]

And finally

[   19.340249] ================================= plane_state->visible 0 crtc_state 0000000036a1b1f5
[   19.340271] crtc_state->enable 0

So, it seems that crtc_state can still be NULL if "!plane_state->visible"
making
NULL pointer dereference, so we need a check for that.
Yet, !plane_state->visible && crtc_state->enable makes WARN_ON to fire always. So, probably we may want removing it.

Quoted 1 line

Thanks, Daniel

Thank you,
Oleksandr
 From dbcce708b237740158a2c16029c56a579324f269 Mon Sep 17 00:00:00 2001
From: Oleksandr Andrushchenko <[email protected]>
Date: Tue, 13 Feb 2018 10:32:20 +0200
Subject: [PATCH] drm/simple_kms_helper: Fix NULL pointer dereference with no
  active CRTC

It is possible that drm_simple_kms_plane_atomic_check called with no CRTC set, e.g. when user-space application sets CRTC_ID/FB_ID to 0 before doing any actual drawing. This leads to NULL pointer dereference because in this case new CRTC state is NULL and must be checked before accessing.

Signed-off-by: Oleksandr Andrushchenko <[email protected]>
---
  drivers/gpu/drm/drm_simple_kms_helper.c | 15 +++++++--------

  1 file changed, 7 insertions(+), 8 deletions(-)
drivers/gpu/drm/drm_simple_kms_helper.c +0 -6
index 9ca8a4a59b74..f54711ff9767 100644
@@ -121,12 +121,6 @@ static int
 drm_simple_kms_plane_atomic_check(struct drm_plane *plane,
       pipe = container_of(plane, struct drm_simple_display_pipe,
 plane);
       crtc_state = drm_atomic_get_new_crtc_state(plane_state->state,
                              &pipe->crtc);
-    if (!crtc_state->enable)
-        return 0; /* nothing to check when disabling or disabled */
-
-    if (crtc_state->enable)
-        drm_mode_get_hv_timing(&crtc_state->mode,
-                       &clip.x2, &clip.y2);
         ret = drm_atomic_helper_check_plane_state(plane_state,

crtc_state,
                            &clip,

drivers/gpu/drm/drm_simple_kms_helper.c +0 -0
@@ -136,8 +130,13 @@ static int
 drm_simple_kms_plane_atomic_check(struct drm_plane *plane,
       if (ret)

drivers/gpu/drm/drm_simple_kms_helper.c static int · line 140

Quoted 10 lines
drivers/gpu/drm/drm_simple_kms_helper.c +7 -1
           return ret;
   -    if (!plane_state->visible)
-        return -EINVAL;
+    if (!plane_state->visible) {
+        if (crtc_state)
+            WARN_ON(crtc_state->enable);
+        return 0;
+    }
+
+    drm_mode_get_hv_timing(&crtc_state->mode, &clip.x2, &clip.y2);

lgtm. With or without the bikeshed to pull the crtc_state check into the WARN_ON.

Quoted 20 lines
Quoted 1 line

Thank you

Quoted 3 lines

Please resubmit as a stand-alone patch, patchwork can't pull patches out of attachments :-/

oh, that was for demonstration purpose only, so we are on the same page and see the patch we are discussing ;)

Quoted 12 lines

-Daniel

Quoted 9 lines
drivers/gpu/drm/drm_simple_kms_helper.c +0 -0
         if (!pipe->funcs || !pipe->funcs->check)
           return 0;
Signature · 6 lines
2.7.4

_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel

_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel

Signature · 3 lines
Daniel Vetter
Software Engineer, Intel Corporation
http://blog.ffwll.ch
#9 lore raw
Quoted 115 lines · Daniel Vetter
Quoted 106 lines · Oleksandr Andrushchenko
Quoted 105 lines · Oleksandr Andrushchenko
Quoted 104 lines · Daniel Vetter
Quoted 96 lines · Oleksandr Andrushchenko
Quoted 95 lines · Daniel Vetter
Quoted 49 lines · Oleksandr Andrushchenko
Quoted 40 lines · Daniel Vetter
Quoted 38 lines · Oleksandr Andrushchenko

From: Oleksandr Andrushchenko <[email protected]>

It is possible that drm_simple_kms_plane_atomic_check called with no CRTC set, e.g. when user-space
application sets CRTC_ID/FB_ID
to 0 before doing any actual drawing. This leads to NULL pointer dereference because in this case new CRTC state is NULL and must be checked before accessing.

Signed-off-by: Oleksandr Andrushchenko
<[email protected]>

---
     drivers/gpu/drm/drm_simple_kms_helper.c | 6 ++++--
     1 file changed, 4 insertions(+), 2 deletions(-)

diff --git
a/drivers/gpu/drm/drm_simple_kms_helper.c
b/drivers/gpu/drm/drm_simple_kms_helper.c
index 9ca8a4a59b74..a05eca9cec8b 100644

drivers/gpu/drm/drm_simple_kms_helper.c +0 -0
@@ -121,8 +121,10 @@ static int

drm_simple_kms_plane_atomic_check(struct
drm_plane *plane,
        pipe = container_of(plane, struct
drm_simple_display_pipe, plane);
        crtc_state =
drm_atomic_get_new_crtc_state(plane_state->state,
                               &pipe->crtc);
-    if (!crtc_state->enable)
-        return 0; /* nothing to check when
disabling or disabled */
+
+    if (!crtc_state || !crtc_state->enable)
+        /* nothing to check when disabling or
disabled or no CRTC set */
+        return 0;
        if (crtc_state->enable)

Quoted 8 lines
Quoted 1 line

drm_mode_get_hv_timing(&crtc_state->mode,

Hm, this is a bit annoying, since the can_position = false parameter to drm_atomic_helper_check_plane_state is supposed to catch all this. Would
moving all the checks after the call to that helper, and gating them on
plane_state->visible also work?

Quoted 1 line

Yes, it does work if this is what you mean:

I wasn't sure, thanks for figuring this out!

Quoted 33 lines
drivers/gpu/drm/drm_simple_kms_helper.c +0 -0
index a05eca9cec8b..c48858bb2823 100644
@@ -122,14 +122,6 @@ static int

drm_simple_kms_plane_atomic_check(struct
drm_plane *plane,
          crtc_state =
drm_atomic_get_new_crtc_state(plane_state->state,
&pipe->crtc);

-       if (!crtc_state || !crtc_state->enable) -               /* nothing to check when disabling or disabled or no CRTC
set */
-               return 0;

drivers/gpu/drm/drm_simple_kms_helper.c +0 -5
-
-       if (crtc_state->enable)
- drm_mode_get_hv_timing(&crtc_state->mode,
-                                      &clip.x2, &clip.y2);
-
           ret =

drm_atomic_helper_check_plane_state(plane_state,
crtc_state,
&clip,
DRM_PLANE_HELPER_NO_SCALING,

drivers/gpu/drm/drm_simple_kms_helper.c +0 -0
@@ -138,6 +130,13 @@ static int

drm_simple_kms_plane_atomic_check(struct
drm_plane *plane,
          if (ret)
                  return ret;

Quoted 3 lines

+       if (!plane_state->visible || !crtc_state->enable) +               return 0; /* nothing to check when disabling or disabled */

if (!plane_state->visible) {
    WARN_ON(crtc_state->enabled);
    return 0;
}

Quoted 3 lines
Quoted 1 line

The helper call above should guarantee this.

Yes, but I still see cases when crtc_state is NULL, thus making crtc_state->enable to fail

Right, when the plane is completely off there's no CRTC state. Correct check should be

    WARN_ON(crtc_state && crtc_state->enabled);

Quoted 8 lines
Quoted 1 line

ok, will update with this additional check

huh, this indeed solves the NULL pointer dereference, but floods a lot with every page flip I have, e.g. !plane_state->visible == true and crtc_state is not NULL and crtc_state->enable == true, thus firing WARN_ON.
Is this something wrong with my use-case/driver or it is still legal to have such a configuration and leave it without WARN_ON and just return 0?

Quoted 2 lines

1 week of vacation later I have to admit that this WARN_ON is completely bogus :-)

np ;)

Quoted 201 lines

Sorry for all the confusion, pls leave it out.
-Daniel

Quoted 198 lines
Quoted 177 lines
Quoted 177 lines
Quoted 164 lines
Quoted 22 lines
Quoted 2 lines

+
+       if (plane_state->visible && crtc_state->enable)

Similar here.

Quoted 5 lines

+ drm_mode_get_hv_timing(&crtc_state->mode,
+                                      &clip.x2, &clip.y2); +
          if (!plane_state->visible)
                  return -EINVAL;

This can now be removed, the plane helper takes care of checking for plane_state->visible != crtc_state->enable. Please also remove.

Quoted 10 lines
Quoted 6 lines

We'd need to add a guarantee to drm_atomic_helper_check_plane_state that
it can cope with crtc_state == NULL, but I think that's a good idea anyway. Atm it shouldn't end up looking at the crtc_state pointer in that
case.

It doesn't look at it at the moment

Quoted 3 lines

Otherwise we'll just go with your fix, but it feels all a bit too fragile,
hence why I want to explore more robust options a bit.

Quoted 6 lines
Quoted 5 lines

At list with the change above it passes my test which failed before. Although I cannot confirm it works for others, but it certainly does for me.

Quoted 1 line

-Daniel

Do you want me to send v1 with the code above?

Yes please, with my above cleanup suggestions.

Please see the patch under test attached (I believe it is what you mean,
with the only change that
     if (!plane_state->visible) {
         *if (crtc_state)* WARN_ON(crtc_state->enable);
         return 0;
     }
check is used).

Whith this patch + additional logs I have:

[   18.939204] [drm:drm_ioctl [drm]] pid=2105, dev=0xe200, auth=1, DRM_IOCTL_MODE_ATOMIC
[...]
[   18.939681] [drm:drm_atomic_set_crtc_for_plane [drm]] Link plane state
00000000c302cbbf to [NOCRTC]
[   18.939822] [drm:drm_atomic_set_fb_for_plane [drm]] Set [NOFB] for plane
state 00000000c302cbbf
[   18.939963] [drm:drm_atomic_print_state [drm]] checking 000000000bc224e7
[   18.939988] vdispl vdispl.0: [drm] plane[29]: plane-0
[   18.940003] vdispl vdispl.0: [drm]   crtc=(null)
[   18.940018] vdispl vdispl.0: [drm]   fb=0
[   18.940032] vdispl vdispl.0: [drm]   crtc-pos=0x0+0+0
[   18.940048] vdispl vdispl.0: [drm] src-pos=0.000000x0.000000+0.000000+0.000000
[   18.940067] vdispl vdispl.0: [drm]   rotation=1
[   18.940199] [drm:drm_atomic_check_only [drm]] checking 000000000bc224e7
[   18.940226] ================================= plane_state->visible 0 crtc_state           (null)
[...]
[   18.978146] [drm:drm_atomic_set_crtc_for_plane [drm]] Link plane state
000000006bd50580 to [CRTC:30:crtc-0]
[   18.978292] [drm:drm_atomic_set_fb_for_plane [drm]] Set [FB:35] for plane
state 000000006bd50580
[   18.978993] [drm:drm_atomic_set_mode_prop_for_crtc [drm]] Set [MODE:1024x768] for CRTC state 00000000e5a28f6a
[   18.979425] [drm:drm_atomic_check_only [drm]] checking 000000000bc224e7
[   18.979540] [drm:drm_atomic_helper_check_modeset [drm_kms_helper]] [CRTC:30:crtc-0] mode changed
[   18.979632] [drm:drm_atomic_helper_check_modeset [drm_kms_helper]] [CRTC:30:crtc-0] enable changed
[   18.979708] [drm:drm_atomic_helper_check_modeset [drm_kms_helper]] [CRTC:30:crtc-0] active changed
[   18.979792] [drm:drm_atomic_helper_check_modeset [drm_kms_helper]] Updating routing for [CONNECTOR:28:Virtual-1]
[   18.979877] [drm:drm_atomic_helper_check_modeset [drm_kms_helper]] [CONNECTOR:28:Virtual-1] using [ENCODER:31:None-31] on [CRTC:30:crtc-0] [   18.979960] [drm:drm_atomic_helper_check_modeset [drm_kms_helper]] [CRTC:30:crtc-0] needs all connectors, enable: y, active: y [   18.980139] [drm:drm_atomic_add_affected_connectors [drm]] Adding all
current connectors for [CRTC:30:crtc-0] to 000000000bc224e7 [   18.980184] ================================= plane_state->visible 0 crtc_state 00000000e5a28f6a
[   18.980205] crtc_state->enable 1

*[   19.022608] WARNING: CPU: 1 PID: 2105 at drivers/gpu/drm/drm_simple_kms_helper.c:137 drm_simple_kms_plane_atomic_check+0xdc/0xf8 [drm_kms_helper]*

[...]

[   19.113601] ================================= plane_state->visible 0 crtc_state 00000000e5a28f6a
[   19.113623] crtc_state->enable 1
[   19.113792] WARNING: CPU: 1 PID: 2105 at drivers/gpu/drm/drm_simple_kms_helper.c:137 drm_simple_kms_plane_atomic_check+0xdc/0xf8 [drm_kms_helper]

[...]

And finally

[   19.340249] ================================= plane_state->visible 0 crtc_state 0000000036a1b1f5
[   19.340271] crtc_state->enable 0

So, it seems that crtc_state can still be NULL if "!plane_state->visible"
making
NULL pointer dereference, so we need a check for that.
Yet, !plane_state->visible && crtc_state->enable makes WARN_ON to fire always. So, probably we may want removing it.

Quoted 1 line

Thanks, Daniel

Thank you,
Oleksandr
 From dbcce708b237740158a2c16029c56a579324f269 Mon Sep 17 00:00:00 2001
From: Oleksandr Andrushchenko <[email protected]>
Date: Tue, 13 Feb 2018 10:32:20 +0200
Subject: [PATCH] drm/simple_kms_helper: Fix NULL pointer dereference with no
  active CRTC

It is possible that drm_simple_kms_plane_atomic_check called with no CRTC set, e.g. when user-space application sets CRTC_ID/FB_ID to 0 before doing any actual drawing. This leads to NULL pointer dereference because in this case new CRTC state is NULL and must be checked before accessing.

Signed-off-by: Oleksandr Andrushchenko
<[email protected]>

---
   drivers/gpu/drm/drm_simple_kms_helper.c | 15 +++++++--------
   1 file changed, 7 insertions(+), 8 deletions(-)
drivers/gpu/drm/drm_simple_kms_helper.c +0 -6
index 9ca8a4a59b74..f54711ff9767 100644
@@ -121,12 +121,6 @@ static int
 drm_simple_kms_plane_atomic_check(struct drm_plane *plane,
       pipe = container_of(plane, struct drm_simple_display_pipe,
 plane);
       crtc_state = drm_atomic_get_new_crtc_state(plane_state->state,
                              &pipe->crtc);
-    if (!crtc_state->enable)
-        return 0; /* nothing to check when disabling or disabled */
-
-    if (crtc_state->enable)
-        drm_mode_get_hv_timing(&crtc_state->mode,
-                       &clip.x2, &clip.y2);
         ret = drm_atomic_helper_check_plane_state(plane_state,

crtc_state,
                            &clip,

drivers/gpu/drm/drm_simple_kms_helper.c +0 -0
@@ -136,8 +130,13 @@ static int
 drm_simple_kms_plane_atomic_check(struct drm_plane *plane,
       if (ret)

drivers/gpu/drm/drm_simple_kms_helper.c static int · line 140

Quoted 10 lines
drivers/gpu/drm/drm_simple_kms_helper.c +7 -1
           return ret;
   -    if (!plane_state->visible)
-        return -EINVAL;
+    if (!plane_state->visible) {
+        if (crtc_state)
+            WARN_ON(crtc_state->enable);
+        return 0;
+    }
+
+    drm_mode_get_hv_timing(&crtc_state->mode, &clip.x2, &clip.y2);

lgtm. With or without the bikeshed to pull the crtc_state check into the WARN_ON.

Quoted 17 lines
Quoted 1 line

Thank you

Quoted 2 lines

Please resubmit as a stand-alone patch, patchwork can't pull patches out of attachments :-/

oh, that was for demonstration purpose only, so we are on the same page and see the patch we are discussing ;)

Quoted 11 lines

-Daniel

Quoted 9 lines
drivers/gpu/drm/drm_simple_kms_helper.c +0 -0
         if (!pipe->funcs || !pipe->funcs->check)
           return 0;
Signature · 6 lines
2.7.4

_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel

_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel