Skip to contentSkip to conversation

gpu: radeon: fix a potential NULL-pointer dereference

PATCH In review Kangjie Lu 2 participants 4 messages Mar 23, 2019 – Mar 26, 2019

Subsystems (2)
lore
Share & tools
lore

Kangjie Lu

author
#0 lore raw

PATCH 1/1 gpu: radeon: fix a potential NULL-pointer dereference

In case alloc_workqueue fails, the fix frees memory and returns to avoid potential NULL pointer dereference.

---
 drivers/gpu/drm/radeon/radeon_display.c | 5 +++++
 1 file changed, 5 insertions(+)
drivers/gpu/drm/radeon/radeon_display.c +5 -0
index aa898c699101..a31305755a77 100644
@@ -678,6 +678,11 @@ static void radeon_crtc_init(struct drm_device *dev, int index)
 	drm_mode_crtc_set_gamma_size(&radeon_crtc->base, 256);
 	radeon_crtc->crtc_id = index;
 	radeon_crtc->flip_queue = alloc_workqueue("radeon-crtc", WQ_HIGHPRI, 0);
+	if (!radeon_crtc->flip_queue) {
+		DRM_ERROR("failed to allocate the flip queue\n");
+		kfree(radeon_crtc);
+		return;
+	}
 	rdev->mode_info.crtcs[index] = radeon_crtc;
 
 	if (rdev->family >= CHIP_BONAIRE) {
Signature · 1 line
2.17.1
#1 lore raw

Hi Kangjie,

thanks for your patch.

Quoted 13 lines · Kangjie Lu

In case alloc_workqueue fails, the fix frees memory and returns to avoid potential NULL pointer dereference.

---
 drivers/gpu/drm/radeon/radeon_display.c | 5 +++++
 1 file changed, 5 insertions(+)
drivers/gpu/drm/radeon/radeon_display.c +0 -0
index aa898c699101..a31305755a77 100644

drivers/gpu/drm/radeon/radeon_display.c static void radeon_crtc_init(struct drm_device *dev, int index) · line 683

Quoted 6 lines
drivers/gpu/drm/radeon/radeon_display.c +3 -0
@@ -678,6 +678,11 @@ static void radeon_crtc_init(struct drm_device *dev, int index)
 	drm_mode_crtc_set_gamma_size(&radeon_crtc->base, 256);
 	radeon_crtc->crtc_id = index;
 	radeon_crtc->flip_queue = alloc_workqueue("radeon-crtc", WQ_HIGHPRI, 0);
+	if (!radeon_crtc->flip_queue) {
+		DRM_ERROR("failed to allocate the flip queue\n");
+		kfree(radeon_crtc);

This would leak some memory referenced by struct drm_crtc. To solve this, I suggest calling radeon_crtc_destroy here and making that cope with radeon_crtc->flip_queue being NULL.

Also, I'm not sure all driver code can handle some CRTCs not initializing. Given that, and as alloc_workqueue presumably only fails if the system is essentially out of memory anyway, it's probably better for radeon_crtc_init to return -ENOMEM in this case and for radeon_modeset_init to propagate that, which will prevent the driver as a whole from initializing.

Signature · 6 lines
Earthling Michel Dänzer               |              https://www.amd.com
Libre software enthusiast             |             Mesa and X developer
_______________________________________________
amd-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/amd-gfx
author
#2 lore raw

PATCH v2 1/1 gpu: radeon: fix a potential NULL-pointer dereference

In case alloc_workqueue fails, the fix frees memory and returns -ENOMEM to avoid potential NULL pointer dereference.

---
v2: use radeon_crtc_destroy to properly clean up resources as suggested by Michel Dänzer <[email protected]>

---
 drivers/gpu/drm/radeon/radeon_display.c | 15 ++++++++++++---
 1 file changed, 12 insertions(+), 3 deletions(-)
drivers/gpu/drm/radeon/radeon_display.c +12 -3
index aa898c699101..3c6d3289316b 100644
@@ -663,7 +663,7 @@ static const struct drm_crtc_funcs radeon_crtc_funcs = {
 	.page_flip_target = radeon_crtc_page_flip_target,
 };
 
-static void radeon_crtc_init(struct drm_device *dev, int index)
+static int radeon_crtc_init(struct drm_device *dev, int index)
 {
 	struct radeon_device *rdev = dev->dev_private;
 	struct radeon_crtc *radeon_crtc;
@@ -671,13 +671,18 @@ static void radeon_crtc_init(struct drm_device *dev, int index)
 
 	radeon_crtc = kzalloc(sizeof(struct radeon_crtc) + (RADEONFB_CONN_LIMIT * sizeof(struct drm_connector *)), GFP_KERNEL);
 	if (radeon_crtc == NULL)
-		return;
+		return -ENOMEM;
 
 	drm_crtc_init(dev, &radeon_crtc->base, &radeon_crtc_funcs);
 
 	drm_mode_crtc_set_gamma_size(&radeon_crtc->base, 256);
 	radeon_crtc->crtc_id = index;
 	radeon_crtc->flip_queue = alloc_workqueue("radeon-crtc", WQ_HIGHPRI, 0);
+	if (!radeon_crtc->flip_queue) {
+		DRM_ERROR("failed to allocate the flip queue\n");
+		radeon_crtc_destroy(&radeon_crtc->base);
+		return -ENOMEM;
+	}
 	rdev->mode_info.crtcs[index] = radeon_crtc;
 
 	if (rdev->family >= CHIP_BONAIRE) {
@@ -706,6 +711,8 @@ static void radeon_crtc_init(struct drm_device *dev, int index)
 		radeon_atombios_init_crtc(dev, radeon_crtc);
 	else
 		radeon_legacy_init_crtc(dev, radeon_crtc);
+
+	return 0;
 }
 
 static const char *encoder_names[38] = {
@@ -1612,7 +1619,9 @@ int radeon_modeset_init(struct radeon_device *rdev)
 
 	/* allocate crtcs */
 	for (i = 0; i < rdev->num_crtc; i++) {
-		radeon_crtc_init(rdev->ddev, i);
+		ret = radeon_crtc_init(rdev->ddev, i);
+		if (ret)
+			return ret;
 	}
 
 	/* okay we should have all the bios connectors */
Signature · 1 line
2.17.1
#3 lore raw
Quoted 17 lines · Kangjie Lu

In case alloc_workqueue fails, the fix frees memory and returns -ENOMEM to avoid potential NULL pointer dereference.

---
v2: use radeon_crtc_destroy to properly clean up resources as suggested by Michel Dänzer <[email protected]>

[...]

drivers/gpu/drm/radeon/radeon_display.c +1 -1
@@ -671,13 +671,18 @@ static void radeon_crtc_init(struct drm_device *dev, int index)
 
 	radeon_crtc = kzalloc(sizeof(struct radeon_crtc) + (RADEONFB_CONN_LIMIT * sizeof(struct drm_connector *)), GFP_KERNEL);
 	if (radeon_crtc == NULL)
-		return;
+		return -ENOMEM;
 

drivers/gpu/drm/radeon/radeon_display.c static void radeon_crtc_init(struct drm_device *dev, int index) · line 685

Quoted 10 lines
drivers/gpu/drm/radeon/radeon_display.c +5 -0
 	drm_crtc_init(dev, &radeon_crtc->base, &radeon_crtc_funcs);
 
 	drm_mode_crtc_set_gamma_size(&radeon_crtc->base, 256);
 	radeon_crtc->crtc_id = index;
 	radeon_crtc->flip_queue = alloc_workqueue("radeon-crtc", WQ_HIGHPRI, 0);
+	if (!radeon_crtc->flip_queue) {
+		DRM_ERROR("failed to allocate the flip queue\n");
+		radeon_crtc_destroy(&radeon_crtc->base);
+		return -ENOMEM;
+	}

radeon_crtc_destroy currently unconditionally calls:

	destroy_workqueue(radeon_crtc->flip_queue);

AFAICT destroy_workqueue will blow up if NULL is passed to it, so radeon_crtc_destroy needs to check for that.

Signature · 6 lines
Earthling Michel Dänzer               |              https://www.amd.com
Libre software enthusiast             |             Mesa and X developer
_______________________________________________
amd-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/amd-gfx