Skip to contentSkip to conversation

drm/sysfb: simpledrm: Various improvements

PATCH Superseded Thomas Zimmermann 3 participants 23 messages Jun 22 – Jun 23

Subsystems (4)

v2 of 3 Superseded Latest is v3 →

lore
Share & tools
lore

Thomas Zimmermann

author maintainer
#0 lore raw

v2 cover drm/sysfb: simpledrm: Various improvements

Patches 1 to 5 improve validation of the values provided by the device firmware.

Patch 6 improves support for panel rotation. There's a rotation property in the DT panel node. Read the value and set the connector's orientation property accordingly.

v2:
- further improve input validation (Sashiko)

Thomas Zimmermann (6):
drm/sysfb: simpledrm: Improve framebuffer-size validation drm/sysfb: simpledrm: Improve panel-size validation drm/sysfb: simpledrm: Inline simplefb_get_validated_int() drm/sysfb: simpledrm: Improve stride validation drm/sysfb: simpledrm: Validate mmap size against framebuffer size drm/sysfb: simpledrm: Read panel orientation from DT node

 drivers/gpu/drm/sysfb/simpledrm.c | 133 ++++++++++++++++++++++++------
 1 file changed, 106 insertions(+), 27 deletions(-)

base-commit: fc59f76558703febba8056be87d1c97d14f7485e

Signature · 1 line
2.54.0
author maintainer
#1 lore raw

PATCH v2 1/6 drm/sysfb: simpledrm: Improve framebuffer-size validation

Validate the framebuffer size from the firmware against the limitations of struct drm_display_mode. The type only stores sizes in 16-bit fields. Fail probing on errors.

v2:
- remove unused function simplefb_get_validated_int0() (Sashiko)

Fixes 11e8f5fd223b ("drm: Add simpledrm driver")
Cc <[email protected]> # v5.14+
---
 drivers/gpu/drm/sysfb/simpledrm.c | 15 ++++-----------
 1 file changed, 4 insertions(+), 11 deletions(-)
drivers/gpu/drm/sysfb/simpledrm.c +4 -11
index fc168920f2c6..15dcafa9d524 100644
@@ -48,13 +48,6 @@ simplefb_get_validated_int(struct drm_device *dev, const char *name,
 	return drm_sysfb_get_validated_int(dev, name, value, INT_MAX);
 }
 
-static int
-simplefb_get_validated_int0(struct drm_device *dev, const char *name,
-			    uint32_t value)
-{
-	return drm_sysfb_get_validated_int0(dev, name, value, INT_MAX);
-}
-
 static const struct drm_format_info *
 simplefb_get_validated_format(struct drm_device *dev, const char *format_name)
 {
@@ -88,14 +81,14 @@ static int
 simplefb_get_width_pd(struct drm_device *dev,
 		      const struct simplefb_platform_data *pd)
 {
-	return simplefb_get_validated_int0(dev, "width", pd->width);
+	return drm_sysfb_get_validated_int0(dev, "width", pd->width, U16_MAX);
 }
 
 static int
 simplefb_get_height_pd(struct drm_device *dev,
 		       const struct simplefb_platform_data *pd)
 {
-	return simplefb_get_validated_int0(dev, "height", pd->height);
+	return drm_sysfb_get_validated_int0(dev, "height", pd->height, U16_MAX);
 }
 
 static int
@@ -144,7 +137,7 @@ simplefb_get_width_of(struct drm_device *dev, struct device_node *of_node)
 
 	if (ret)
 		return ret;
-	return simplefb_get_validated_int0(dev, "width", width);
+	return drm_sysfb_get_validated_int0(dev, "width", width, U16_MAX);
 }
 
 static int
@@ -155,7 +148,7 @@ simplefb_get_height_of(struct drm_device *dev, struct device_node *of_node)
 
 	if (ret)
 		return ret;
-	return simplefb_get_validated_int0(dev, "height", height);
+	return drm_sysfb_get_validated_int0(dev, "height", height, U16_MAX);
 }
 
 static int
Signature · 1 line
2.54.0
#2 lore raw

Re: [PATCH v2 1/6] drm/sysfb: simpledrm: Improve framebuffer-size validation

Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:

Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:

Pre-existing issues:
- [High] The simpledrm driver fails to correctly validate that the framebuffer dimensions fit within the mapped memory resource size due to an inverted condition. --

commit 1c3adcb617c4a3add398f3f24ada5131889b5d6a
Author: Thomas Zimmermann <[email protected]>

drm/sysfb: simpledrm: Improve framebuffer-size validation

Validate the framebuffer size from the firmware against the limitations of struct drm_display_mode. The type only stores sizes in 16-bit fields. Fail probing on errors.

Cc <[email protected]> # v5.14+
Quoted 4 lines
drivers/gpu/drm/sysfb/simpledrm.c +0 -0
index fc168920f2c64..15dcafa9d524a 100644

[ ... ]

drivers/gpu/drm/sysfb/simpledrm.c simplefb_get_height_of(struct drm_device *dev, struct device_node *of_node) · line 154

Quoted 9 lines
drivers/gpu/drm/sysfb/simpledrm.c +1 -1
@@ -155,7 +148,7 @@ simplefb_get_height_of(struct drm_device *dev, struct device_node *of_node)
 
 	if (ret)
 		return ret;
-	return simplefb_get_validated_int0(dev, "height", height);
+	return drm_sysfb_get_validated_int0(dev, "height", height, U16_MAX);
 }
 
 static int

[Severity: High]
This isn't a bug introduced by this patch, but commit 'drm/sysfb:
simpledrm: Validate mmap size against framebuffer size' in this series attempted to fix the missing validation and failed due to an inverted condition.

Does simpledrm_device_create() fail to correctly validate that the framebuffer dimensions fit within the mapped memory resource size?

Looking at simpledrm_device_create(), the check appears to be inverted in both the system memory and I/O memory fallback paths:

drivers/gpu/drm/sysfb/simpledrm.c:simpledrm_device_create() {
        ...
		if (resource_size(mem) > size) {
			drm_err(dev,
				"framebuffer size of %llu exceeds memory range %pr\n",
				size, mem);
			return ERR_PTR(-EINVAL);
		}
        ...
}

Because this check returns -EINVAL when resource_size(mem) > size, it incorrectly rejects valid configurations where the physical memory region is larger than the required framebuffer.

More importantly, it allows invalid configurations where size > resource_size(mem). In this case, devm_memremap() maps the smaller resource_size(mem), but drm_fb_memcpy() will access size bytes during plane updates, leading to an out-of-bounds kernel memory access past the mapped area.

Should this check be size > resource_size(mem) instead?

Signature · 1 line
Sashiko AI review · https://sashiko.dev/#/patchset/[email protected]?part=1
Reviewed-by on v2 1/6
#3 lore raw

Re: [PATCH v2 1/6] drm/sysfb: simpledrm: Improve framebuffer-size validation

In reply to Thomas Zimmermann
Quoted 7 lines · Thomas Zimmermann

Validate the framebuffer size from the firmware against the limitations of struct drm_display_mode. The type only stores sizes in 16-bit fields. Fail probing on errors.

v2:
- remove unused function simplefb_get_validated_int0() (Sashiko)

Quoted 6 lines
Fixes 11e8f5fd223b ("drm: Add simpledrm driver")
Cc <[email protected]> # v5.14+
---
 drivers/gpu/drm/sysfb/simpledrm.c | 15 ++++-----------
 1 file changed, 4 insertions(+), 11 deletions(-)
author maintainer
#4 lore raw

PATCH v2 2/6 drm/sysfb: simpledrm: Improve panel-size validation

In reply to Thomas Zimmermann

Validate the panel size from the device-tree node against the limitations of struct drm_display_mode. The type only stores sizes in 16-bit fields. Fail transparently on errors; do not warn.

v2:
- only use initialized values in debugging output (Sashiko)

Fixes 2a6d731a8f16 ("drm/simpledrm: Allow physical width and height configuration via panel node")
Cc Rayyan Ansari <[email protected]>
Cc <[email protected]> # v6.4+
---
 drivers/gpu/drm/sysfb/simpledrm.c | 40 ++++++++++++++++++++++++++++---
 1 file changed, 37 insertions(+), 3 deletions(-)
drivers/gpu/drm/sysfb/simpledrm.c +37 -3
index 15dcafa9d524..fa2121f81def 100644
@@ -193,6 +193,40 @@ simplefb_get_memory_of(struct drm_device *dev, struct device_node *of_node)
 	return res;
 }
 
+static u16
+__simplefb_get_panel_size_mm_of(struct drm_device *dev, struct device_node *of_panel_node,
+				const char *name)
+{
+	int ret;
+	u32 value;
+
+	ret = of_property_read_u32(of_panel_node, name, &value);
+	if (ret) {
+		drm_dbg(dev, "simplefb: cannot parse panel %s: error %d\n",
+			name, ret);
+		return 0; /* not an error, simply ignore */
+	}
+	if (value > U16_MAX) {
+		drm_dbg(dev, "simplefb: panel %s of %u exceeds maximum value\n",
+			name, value);
+		return 0; /* not an error, simply ignore */
+	}
+
+	return value;
+}
+
+static u16
+simplefb_get_panel_width_mm_of(struct drm_device *dev, struct device_node *of_panel_node)
+{
+	return __simplefb_get_panel_size_mm_of(dev, of_panel_node, "width-mm");
+}
+
+static u16
+simplefb_get_panel_height_mm_of(struct drm_device *dev, struct device_node *of_panel_node)
+{
+	return __simplefb_get_panel_size_mm_of(dev, of_panel_node, "height-mm");
+}
+
 /*
  * Simple Framebuffer device
  */
@@ -594,7 +628,7 @@ static struct simpledrm_device *simpledrm_device_create(struct drm_driver *drv,
 	struct drm_sysfb_device *sysfb;
 	struct drm_device *dev;
 	int width, height, stride;
-	int width_mm = 0, height_mm = 0;
+	u16 width_mm = 0, height_mm = 0;
 	struct device_node *panel_node;
 	const struct drm_format_info *format;
 	struct resource *res, *mem = NULL;
@@ -658,8 +692,8 @@ static struct simpledrm_device *simpledrm_device_create(struct drm_driver *drv,
 			return ERR_CAST(mem);
 		panel_node = of_parse_phandle(of_node, "panel", 0);
 		if (panel_node) {
-			simplefb_read_u32_of(dev, panel_node, "width-mm", &width_mm);
-			simplefb_read_u32_of(dev, panel_node, "height-mm", &height_mm);
+			width_mm = simplefb_get_panel_width_mm_of(dev, panel_node);
+			height_mm = simplefb_get_panel_height_mm_of(dev, panel_node);
 			of_node_put(panel_node);
 		}
 	} else {
Signature · 1 line
2.54.0
Reviewed-by on v2 2/6
#5 lore raw

Re: [PATCH v2 2/6] drm/sysfb: simpledrm: Improve panel-size validation

Quoted 30 lines · Thomas Zimmermann

Validate the panel size from the device-tree node against the limitations of struct drm_display_mode. The type only stores sizes in 16-bit fields. Fail transparently on errors; do not warn.

v2:
- only use initialized values in debugging output (Sashiko)

Fixes 2a6d731a8f16 ("drm/simpledrm: Allow physical width and height configuration via panel node")
Cc Rayyan Ansari <[email protected]>
Cc <[email protected]> # v6.4+
---
 drivers/gpu/drm/sysfb/simpledrm.c | 40 ++++++++++++++++++++++++++++---
 1 file changed, 37 insertions(+), 3 deletions(-)
drivers/gpu/drm/sysfb/simpledrm.c +7 -0
index 15dcafa9d524..fa2121f81def 100644
@@ -193,6 +193,40 @@ simplefb_get_memory_of(struct drm_device *dev, struct device_node *of_node)
 	return res;
 }
 
+static u16
+__simplefb_get_panel_size_mm_of(struct drm_device *dev, struct device_node *of_panel_node,
+				const char *name)
+{
+	int ret;
+	u32 value;
+

drivers/gpu/drm/sysfb/simpledrm.c simplefb_get_memory_of(struct drm_device *dev, struct device_node *of_node) · line 212

Quoted 10 lines
drivers/gpu/drm/sysfb/simpledrm.c +10 -0
+	ret = of_property_read_u32(of_panel_node, name, &value);
+	if (ret) {
+		drm_dbg(dev, "simplefb: cannot parse panel %s: error %d\n",
+			name, ret);
+		return 0; /* not an error, simply ignore */
+	}
+	if (value > U16_MAX) {
+		drm_dbg(dev, "simplefb: panel %s of %u exceeds maximum value\n",
+			name, value);
+		return 0; /* not an error, simply ignore */

I wonder if it's perhaps better to move this comment to the function scope and explain why this can be ignored. I didn't know and had to go look at drm_sysfb_mode() to see that if these are 0, it'll compute the physical dimensions based on a default of 96 DPI.

Quoted 30 lines
drivers/gpu/drm/sysfb/simpledrm.c +18 -1
+	}
+
+	return value;
+}
+
+static u16
+simplefb_get_panel_width_mm_of(struct drm_device *dev, struct device_node *of_panel_node)
+{
+	return __simplefb_get_panel_size_mm_of(dev, of_panel_node, "width-mm");
+}
+
+static u16
+simplefb_get_panel_height_mm_of(struct drm_device *dev, struct device_node *of_panel_node)
+{
+	return __simplefb_get_panel_size_mm_of(dev, of_panel_node, "height-mm");
+}
+
 /*
  * Simple Framebuffer device
  */
@@ -594,7 +628,7 @@ static struct simpledrm_device *simpledrm_device_create(struct drm_driver *drv,
 	struct drm_sysfb_device *sysfb;
 	struct drm_device *dev;
 	int width, height, stride;
-	int width_mm = 0, height_mm = 0;
+	u16 width_mm = 0, height_mm = 0;
 	struct device_node *panel_node;
 	const struct drm_format_info *format;
 	struct resource *res, *mem = NULL;

drivers/gpu/drm/sysfb/simpledrm.c static struct simpledrm_device *simpledrm_device_create(struct drm_driver *drv, · line 699

Quoted 10 lines
drivers/gpu/drm/sysfb/simpledrm.c +2 -2
@@ -658,8 +692,8 @@ static struct simpledrm_device *simpledrm_device_create(struct drm_driver *drv,
 			return ERR_CAST(mem);
 		panel_node = of_parse_phandle(of_node, "panel", 0);
 		if (panel_node) {
-			simplefb_read_u32_of(dev, panel_node, "width-mm", &width_mm);
-			simplefb_read_u32_of(dev, panel_node, "height-mm", &height_mm);
+			width_mm = simplefb_get_panel_width_mm_of(dev, panel_node);
+			height_mm = simplefb_get_panel_height_mm_of(dev, panel_node);
 			of_node_put(panel_node);
 		}
 	} else {

The drm_sysfb_mode() function that width_mm and height_mm get passed into accepts them as unsigned int, so maybe that should be changed as well for more consistency?

In either case, since they all end up in the u16 in the struct, it's obviously correct to check for the range when parsing, so:

author maintainer
#6 lore raw

Re: [PATCH v2 2/6] drm/sysfb: simpledrm: Improve panel-size validation

Hi

Am 22.06.26 um 18:05 schrieb Thierry Reding:

Quoted 37 lines · Thomas Zimmermann
Quoted 36 lines

Validate the panel size from the device-tree node against the limitations of struct drm_display_mode. The type only stores sizes in 16-bit fields. Fail transparently on errors; do not warn.

v2:
- only use initialized values in debugging output (Sashiko)

Fixes 2a6d731a8f16 ("drm/simpledrm: Allow physical width and height configuration via panel node")
Cc Rayyan Ansari <[email protected]>
Cc <[email protected]> # v6.4+
---
  drivers/gpu/drm/sysfb/simpledrm.c | 40 ++++++++++++++++++++++++++++---
  1 file changed, 37 insertions(+), 3 deletions(-)
drivers/gpu/drm/sysfb/simpledrm.c +13 -0
index 15dcafa9d524..fa2121f81def 100644
@@ -193,6 +193,40 @@ simplefb_get_memory_of(struct drm_device *dev, struct device_node *of_node)
  	return res;
  }
 
+static u16
+__simplefb_get_panel_size_mm_of(struct drm_device *dev, struct device_node *of_panel_node,
+				const char *name)
+{
+	int ret;
+	u32 value;
+
+	ret = of_property_read_u32(of_panel_node, name, &value);
+	if (ret) {
+		drm_dbg(dev, "simplefb: cannot parse panel %s: error %d\n",
+			name, ret);
+		return 0; /* not an error, simply ignore */
+	}
Quoted 8 lines

drivers/gpu/drm/sysfb/simpledrm.c simplefb_get_memory_of(struct drm_device *dev, struct device_node *of_node) · line 212

Quoted 4 lines
drivers/gpu/drm/sysfb/simpledrm.c +4 -0
+	if (value > U16_MAX) {
+		drm_dbg(dev, "simplefb: panel %s of %u exceeds maximum value\n",
+			name, value);
+		return 0; /* not an error, simply ignore */

I wonder if it's perhaps better to move this comment to the function scope and explain why this can be ignored. I didn't know and had to go look at drm_sysfb_mode() to see that if these are 0, it'll compute the physical dimensions based on a default of 96 DPI.

Makes sense.

Quoted 36 lines
Quoted 36 lines
drivers/gpu/drm/sysfb/simpledrm.c +18 -3
+	}
+
+	return value;
+}
+
+static u16
+simplefb_get_panel_width_mm_of(struct drm_device *dev, struct device_node *of_panel_node)
+{
+	return __simplefb_get_panel_size_mm_of(dev, of_panel_node, "width-mm");
+}
+
+static u16
+simplefb_get_panel_height_mm_of(struct drm_device *dev, struct device_node *of_panel_node)
+{
+	return __simplefb_get_panel_size_mm_of(dev, of_panel_node, "height-mm");
+}
+
  /*
   * Simple Framebuffer device
   */
@@ -594,7 +628,7 @@ static struct simpledrm_device *simpledrm_device_create(struct drm_driver *drv,
  	struct drm_sysfb_device *sysfb;
  	struct drm_device *dev;
  	int width, height, stride;
-	int width_mm = 0, height_mm = 0;
+	u16 width_mm = 0, height_mm = 0;
  	struct device_node *panel_node;
  	const struct drm_format_info *format;
  	struct resource *res, *mem = NULL;
@@ -658,8 +692,8 @@ static struct simpledrm_device *simpledrm_device_create(struct drm_driver *drv,
  			return ERR_CAST(mem);
  		panel_node = of_parse_phandle(of_node, "panel", 0);
  		if (panel_node) {
-			simplefb_read_u32_of(dev, panel_node, "width-mm", &width_mm);
-			simplefb_read_u32_of(dev, panel_node, "height-mm", &height_mm);
Quoted 8 lines

drivers/gpu/drm/sysfb/simpledrm.c static struct simpledrm_device *simpledrm_device_create(struct drm_driver *drv, · line 699

Quoted 5 lines
drivers/gpu/drm/sysfb/simpledrm.c +2 -0
+			width_mm = simplefb_get_panel_width_mm_of(dev, panel_node);
+			height_mm = simplefb_get_panel_height_mm_of(dev, panel_node);
  			of_node_put(panel_node);
  		}
  	} else {

The drm_sysfb_mode() function that width_mm and height_mm get passed into accepts them as unsigned int, so maybe that should be changed as well for more consistency?

Also makes sense.

Best regards
Thomas

Quoted 5 lines

In either case, since they all end up in the u16 in the struct, it's obviously correct to check for the range when parsing, so:

Signature · 6 lines
--
Thomas Zimmermann
Graphics Driver Developer
SUSE Software Solutions Germany GmbH
Frankenstr. 146, 90461 Nürnberg, Germany, www.suse.com
GF: Jochen Jaser, Andrew McDonald, Werner Knoblich, (HRB 36809, AG Nürnberg)
author maintainer
#7 lore raw

PATCH v2 3/6 drm/sysfb: simpledrm: Inline simplefb_get_validated_int()

In reply to Thomas Zimmermann

The helper simplefb_get_validated_int() is only a wrapper around drm_sysfb_get_validated_int(). Inline the function into its callers.

---
 drivers/gpu/drm/sysfb/simpledrm.c | 11 ++---------
 1 file changed, 2 insertions(+), 9 deletions(-)
drivers/gpu/drm/sysfb/simpledrm.c +2 -9
index fa2121f81def..709922ea656c 100644
@@ -41,13 +41,6 @@
  * Helpers for simplefb
  */
 
-static int
-simplefb_get_validated_int(struct drm_device *dev, const char *name,
-			   uint32_t value)
-{
-	return drm_sysfb_get_validated_int(dev, name, value, INT_MAX);
-}
-
 static const struct drm_format_info *
 simplefb_get_validated_format(struct drm_device *dev, const char *format_name)
 {
@@ -95,7 +88,7 @@ static int
 simplefb_get_stride_pd(struct drm_device *dev,
 		       const struct simplefb_platform_data *pd)
 {
-	return simplefb_get_validated_int(dev, "stride", pd->stride);
+	return drm_sysfb_get_validated_int(dev, "stride", pd->stride, INT_MAX);
 }
 
 static const struct drm_format_info *
@@ -159,7 +152,7 @@ simplefb_get_stride_of(struct drm_device *dev, struct device_node *of_node)
 
 	if (ret)
 		return ret;
-	return simplefb_get_validated_int(dev, "stride", stride);
+	return drm_sysfb_get_validated_int(dev, "stride", stride, INT_MAX);
 }
 
 static const struct drm_format_info *
Signature · 1 line
2.54.0
Reviewed-by on v2 3/6
#8 lore raw

Re: [PATCH v2 3/6] drm/sysfb: simpledrm: Inline simplefb_get_validated_int()

Quoted 7 lines · Thomas Zimmermann

The helper simplefb_get_validated_int() is only a wrapper around drm_sysfb_get_validated_int(). Inline the function into its callers.

---
 drivers/gpu/drm/sysfb/simpledrm.c | 11 ++---------
 1 file changed, 2 insertions(+), 9 deletions(-)
author maintainer
#9 lore raw

PATCH v2 4/6 drm/sysfb: simpledrm: Improve stride validation

In reply to Thomas Zimmermann

Validate the computed stride against the maximum value INT_MAX.

Fixes 7bfa5c7b28d6 ("drm/simpledrm: Compute linestride with drm_format_info_min_pitch()")
Cc <[email protected]> # v6.1+
---
 drivers/gpu/drm/sysfb/simpledrm.c | 10 ++++++++--
 1 file changed, 8 insertions(+), 2 deletions(-)
drivers/gpu/drm/sysfb/simpledrm.c +8 -2
index 709922ea656c..76b9a3f5c4ef 100644
@@ -694,9 +694,15 @@ static struct simpledrm_device *simpledrm_device_create(struct drm_driver *drv,
 		return ERR_PTR(-ENODEV);
 	}
 	if (!stride) {
-		stride = drm_format_info_min_pitch(format, 0, width);
-		if (drm_WARN_ON(dev, !stride))
+		u64 pitch = drm_format_info_min_pitch(format, 0, width);
+
+		if (drm_WARN_ON(dev, !pitch)) {
+			return ERR_PTR(-EINVAL); /* driver bug */
+		} else if (pitch > INT_MAX) {
+			drm_warn(dev, "stride of %llu exceeds maximum\n", pitch);
 			return ERR_PTR(-EINVAL);
+		}
+		stride = pitch;
 	}
 
 	sysfb->fb_mode = drm_sysfb_mode(width, height, width_mm, height_mm);
Signature · 1 line
2.54.0
Reviewed-by on v2 4/6
#10 lore raw

Re: [PATCH v2 4/6] drm/sysfb: simpledrm: Improve stride validation

Quoted 8 lines · Thomas Zimmermann

Validate the computed stride against the maximum value INT_MAX.

Fixes 7bfa5c7b28d6 ("drm/simpledrm: Compute linestride with drm_format_info_min_pitch()")
Cc <[email protected]> # v6.1+
---
 drivers/gpu/drm/sysfb/simpledrm.c | 10 ++++++++--
 1 file changed, 8 insertions(+), 2 deletions(-)

Ugh... I have a hard time imagining that we'd ever exceed INT_MAX here, but I guess better safe than sorry?

author maintainer
#11 lore raw

Re: [PATCH v2 4/6] drm/sysfb: simpledrm: Improve stride validation

Hi

Am 22.06.26 um 18:10 schrieb Thierry Reding:

Quoted 11 lines · Thomas Zimmermann
Quoted 8 lines

Validate the computed stride against the maximum value INT_MAX.

Fixes 7bfa5c7b28d6 ("drm/simpledrm: Compute linestride with drm_format_info_min_pitch()")
Cc <[email protected]> # v6.1+
---
  drivers/gpu/drm/sysfb/simpledrm.c | 10 ++++++++--
  1 file changed, 8 insertions(+), 2 deletions(-)

Ugh... I have a hard time imagining that we'd ever exceed INT_MAX here, but I guess better safe than sorry?

Yeah. The idea of these patches is to harden the driver against all kind of nonsense the firmware might give it. With patch 1 that limits the width to 2^16, I don't think it's possible to exceed a pitch of INT_MAX, but let's check anyway.

Best regards
Thomas

Quoted 2 lines
Signature · 6 lines
--
Thomas Zimmermann
Graphics Driver Developer
SUSE Software Solutions Germany GmbH
Frankenstr. 146, 90461 Nürnberg, Germany, www.suse.com
GF: Jochen Jaser, Andrew McDonald, Werner Knoblich, (HRB 36809, AG Nürnberg)
author maintainer
#12 lore raw

PATCH v2 5/6 drm/sysfb: simpledrm: Validate mmap size against framebuffer size

In reply to Thomas Zimmermann

The size of the mmap'ed framebuffer could be smaller than the minimum required framebuffer size. Validate the resource size against the framebuffer size.

Buggy firmware that triggers this check should be fixed up with a quirk on a case-by-case base.

---
 drivers/gpu/drm/sysfb/simpledrm.c | 25 +++++++++++++++++++++++++
 1 file changed, 25 insertions(+)
drivers/gpu/drm/sysfb/simpledrm.c +25 -0
index 76b9a3f5c4ef..a04f0c852ea4 100644
@@ -6,6 +6,7 @@
 #include <linux/of_address.h>
 #include <linux/of_clk.h>
 #include <linux/of_reserved_mem.h>
+#include <linux/overflow.h>
 #include <linux/platform_data/simplefb.h>
 #include <linux/platform_device.h>
 #include <linux/pm.h>
@@ -624,6 +625,7 @@ static struct simpledrm_device *simpledrm_device_create(struct drm_driver *drv,
 	u16 width_mm = 0, height_mm = 0;
 	struct device_node *panel_node;
 	const struct drm_format_info *format;
+	u64 size;
 	struct resource *res, *mem = NULL;
 	struct drm_plane *primary_plane;
 	struct drm_crtc *crtc;
@@ -704,6 +706,15 @@ static struct simpledrm_device *simpledrm_device_create(struct drm_driver *drv,
 		}
 		stride = pitch;
 	}
+	if (check_mul_overflow(height, stride, &size)) {
+		drm_err(dev, "framebuffer size exceeds maximum\n");
+		return ERR_PTR(-EINVAL);
+	}
+	size = ALIGN(size, PAGE_SIZE);
+	if (size < PAGE_SIZE) {
+		drm_err(dev, "framebuffer alignment exceeds maximum\n");
+		return ERR_PTR(-EINVAL);
+	}
 
 	sysfb->fb_mode = drm_sysfb_mode(width, height, width_mm, height_mm);
 	sysfb->fb_format = format;
@@ -729,6 +740,13 @@ static struct simpledrm_device *simpledrm_device_create(struct drm_driver *drv,
 
 		drm_dbg(dev, "using system memory framebuffer at %pr\n", mem);
 
+		if (resource_size(mem) > size) {
+			drm_err(dev,
+				"framebuffer size of %llu exceeds memory range %pr\n",
+				size, mem);
+			return ERR_PTR(-EINVAL);
+		}
+
 		screen_base = devm_memremap(dev->dev, mem->start, resource_size(mem), MEMREMAP_WC);
 		if (IS_ERR(screen_base))
 			return screen_base;
@@ -762,6 +780,13 @@ static struct simpledrm_device *simpledrm_device_create(struct drm_driver *drv,
 			mem = res;
 		}
 
+		if (resource_size(mem) > size) {
+			drm_err(dev,
+				"framebuffer size of %llu exceeds memory range %pr\n",
+				size, mem);
+			return ERR_PTR(-EINVAL);
+		}
+
 		screen_base = devm_ioremap_wc(&pdev->dev, mem->start, resource_size(mem));
 		if (!screen_base)
 			return ERR_PTR(-ENOMEM);
Signature · 1 line
2.54.0
#13 lore raw

Re: [PATCH v2 5/6] drm/sysfb: simpledrm: Validate mmap size against framebuffer size

Thank you for your contribution! Sashiko AI review found 2 potential issue(s) to consider: - [High] The validation check comparing the memory resource size against the required framebuffer size is…

Thank you for your contribution! Sashiko AI review found 2 potential issue(s) to consider:
- [High] The validation check comparing the memory resource size against the required framebuffer size is inverted, causing valid device tree configurations to fail probing and bypassing protection for undersized framebuffers.
- [Low] The condition if (size < PAGE_SIZE) functions as a check for a zero-sized framebuffer rather than an alignment overflow guard, resulting in a misleading error message.
--

commit fe687b775049a45634b1930d3d958d911916b07a
Author: Thomas Zimmermann <[email protected]>

drm/sysfb: simpledrm: Validate mmap size against framebuffer size

The size of the mmap'ed framebuffer could be smaller than the minimum required framebuffer size. Validate the resource size against the framebuffer size.

Buggy firmware that triggers this check should be fixed up with a quirk on a case-by-case base.

Quoted 4 lines
drivers/gpu/drm/sysfb/simpledrm.c +0 -0
index 76b9a3f5c4efa..a04f0c852ea48 100644

[ ... ]

drivers/gpu/drm/sysfb/simpledrm.c static struct simpledrm_device *simpledrm_device_create(struct drm_driver *drv, · line 715

Quoted 11 lines
drivers/gpu/drm/sysfb/simpledrm.c +7 -0
@@ -704,6 +706,15 @@ static struct simpledrm_device *simpledrm_device_create(struct drm_driver *drv,
 		}
 		stride = pitch;
 	}
+	if (check_mul_overflow(height, stride, &size)) {
+		drm_err(dev, "framebuffer size exceeds maximum\n");
+		return ERR_PTR(-EINVAL);
+	}
+	size = ALIGN(size, PAGE_SIZE);
+	if (size < PAGE_SIZE) {
+		drm_err(dev, "framebuffer alignment exceeds maximum\n");

[Severity: Low]
Does this code accurately report an alignment overflow?

Because both height and stride are bounded by INT_MAX, their product (size) cannot exceed ~2^62, meaning the ALIGN operation will not overflow a 64-bit integer.

The condition size < PAGE_SIZE will only be true if size was exactly 0 before alignment. While rejecting a zero-sized framebuffer is correct, the error message 'framebuffer alignment exceeds maximum' might be misleading for developers debugging initialization failures.

Quoted 5 lines
drivers/gpu/drm/sysfb/simpledrm.c +2 -0
+		return ERR_PTR(-EINVAL);
+	}
 
 	sysfb->fb_mode = drm_sysfb_mode(width, height, width_mm, height_mm);
 	sysfb->fb_format = format;

[ ... ]

drivers/gpu/drm/sysfb/simpledrm.c static struct simpledrm_device *simpledrm_device_create(struct drm_driver *drv, · line 745

Quoted 7 lines
drivers/gpu/drm/sysfb/simpledrm.c +3 -0
@@ -729,6 +740,13 @@ static struct simpledrm_device *simpledrm_device_create(struct drm_driver *drv,
 
 		drm_dbg(dev, "using system memory framebuffer at %pr\n", mem);
 
+		if (resource_size(mem) > size) {
+			drm_err(dev,
+				"framebuffer size of %llu exceeds memory range %pr\n",

[Severity: High]
Is the validation check here inverted?

The condition evaluates to true when the provided memory region is larger than the required framebuffer size, which is a valid configuration when the reserved-memory region is larger than the exact resolution requires.

This logic inversion causes the driver to incorrectly log 'framebuffer size... exceeds memory range' and abort probing for valid configurations, while allowing memory regions that are actually too small.

Will this bypass the intended validation and potentially lead to out-of-bounds memory accesses if the framebuffer is not fully backed by the memory resource?

Quoted 7 lines
drivers/gpu/drm/sysfb/simpledrm.c +4 -0
+				size, mem);
+			return ERR_PTR(-EINVAL);
+		}
+
 		screen_base = devm_memremap(dev->dev, mem->start, resource_size(mem), MEMREMAP_WC);
 		if (IS_ERR(screen_base))
 			return screen_base;

[ ... ]

drivers/gpu/drm/sysfb/simpledrm.c static struct simpledrm_device *simpledrm_device_create(struct drm_driver *drv, · line 785

Quoted 7 lines
drivers/gpu/drm/sysfb/simpledrm.c +3 -0
@@ -762,6 +780,13 @@ static struct simpledrm_device *simpledrm_device_create(struct drm_driver *drv,
 			mem = res;
 		}
 
+		if (resource_size(mem) > size) {
+			drm_err(dev,
+				"framebuffer size of %llu exceeds memory range %pr\n",

[Severity: High]
Is this check also inverted for the I/O memory path?

Similar to the system memory path above, this will reject valid framebuffers that fit within the memory range and allow undersized ones.

Quoted 5 lines
drivers/gpu/drm/sysfb/simpledrm.c +4 -0
+				size, mem);
+			return ERR_PTR(-EINVAL);
+		}
+
 		screen_base = devm_ioremap_wc(&pdev->dev, mem->start, resource_size(mem));
Signature · 1 line
Sashiko AI review · https://sashiko.dev/#/patchset/[email protected]?part=5
author maintainer
#14 lore raw

Re: [PATCH v2 5/6] drm/sysfb: simpledrm: Validate mmap size against framebuffer size

Am 22.06.26 um 15:33 schrieb [email protected]:

Quoted 37 lines

Thank you for your contribution! Sashiko AI review found 2 potential issue(s) to consider:
- [High] The validation check comparing the memory resource size against the required framebuffer size is inverted, causing valid device tree configurations to fail probing and bypassing protection for undersized framebuffers.
- [Low] The condition if (size < PAGE_SIZE) functions as a check for a zero-sized framebuffer rather than an alignment overflow guard, resulting in a misleading error message.
--

commit fe687b775049a45634b1930d3d958d911916b07a
Author: Thomas Zimmermann <[email protected]>

drm/sysfb: simpledrm: Validate mmap size against framebuffer size

The size of the mmap'ed framebuffer could be smaller than the minimum required framebuffer size. Validate the resource size against the framebuffer size.

Buggy firmware that triggers this check should be fixed up with a quirk on a case-by-case base.

Quoted 4 lines
drivers/gpu/drm/sysfb/simpledrm.c +0 -0
index 76b9a3f5c4efa..a04f0c852ea48 100644

[ ... ]

drivers/gpu/drm/sysfb/simpledrm.c static struct simpledrm_device *simpledrm_device_create(struct drm_driver *drv, · line 715

Quoted 11 lines
drivers/gpu/drm/sysfb/simpledrm.c +7 -0
@@ -704,6 +706,15 @@ static struct simpledrm_device *simpledrm_device_create(struct drm_driver *drv,
  		}
  		stride = pitch;
  	}
+	if (check_mul_overflow(height, stride, &size)) {
+		drm_err(dev, "framebuffer size exceeds maximum\n");
+		return ERR_PTR(-EINVAL);
+	}
+	size = ALIGN(size, PAGE_SIZE);
+	if (size < PAGE_SIZE) {
+		drm_err(dev, "framebuffer alignment exceeds maximum\n");

[Severity: Low]
Does this code accurately report an alignment overflow?

Quoted 3 lines

Because both height and stride are bounded by INT_MAX, their product (size) cannot exceed ~2^62, meaning the ALIGN operation will not overflow a 64-bit integer.

This protects against overflowing in the ALIGN() call. While we cannot trigger this problem now, it is still the safer thing to do this.

Quoted 5 lines

The condition size < PAGE_SIZE will only be true if size was exactly 0 before alignment. While rejecting a zero-sized framebuffer is correct, the error message 'framebuffer alignment exceeds maximum' might be misleading for developers debugging initialization failures.

Neither height nor stride can be 0 IIRC. We should not see size being 0.

Quoted 18 lines
Quoted 6 lines
drivers/gpu/drm/sysfb/simpledrm.c +2 -0
+		return ERR_PTR(-EINVAL);
+	}
 
  	sysfb->fb_mode = drm_sysfb_mode(width, height, width_mm, height_mm);
  	sysfb->fb_format = format;

[ ... ]

drivers/gpu/drm/sysfb/simpledrm.c static struct simpledrm_device *simpledrm_device_create(struct drm_driver *drv, · line 745

Quoted 7 lines
drivers/gpu/drm/sysfb/simpledrm.c +3 -0
@@ -729,6 +740,13 @@ static struct simpledrm_device *simpledrm_device_create(struct drm_driver *drv,
 
  		drm_dbg(dev, "using system memory framebuffer at %pr\n", mem);
 
+		if (resource_size(mem) > size) {
+			drm_err(dev,
+				"framebuffer size of %llu exceeds memory range %pr\n",

[Severity: High]
Is the validation check here inverted?

Quoted 3 lines

The condition evaluates to true when the provided memory region is larger than the required framebuffer size, which is a valid configuration when the reserved-memory region is larger than the exact resolution requires.

Arghh, when I tested with bochs, I hit exactly the case of mem == size, so it worked. :/

Quoted 35 lines

This logic inversion causes the driver to incorrectly log 'framebuffer size... exceeds memory range' and abort probing for valid configurations, while allowing memory regions that are actually too small.

Will this bypass the intended validation and potentially lead to out-of-bounds memory accesses if the framebuffer is not fully backed by the memory resource?

Quoted 7 lines
drivers/gpu/drm/sysfb/simpledrm.c +4 -0
+				size, mem);
+			return ERR_PTR(-EINVAL);
+		}
+
  		screen_base = devm_memremap(dev->dev, mem->start, resource_size(mem), MEMREMAP_WC);
  		if (IS_ERR(screen_base))
  			return screen_base;

[ ... ]

drivers/gpu/drm/sysfb/simpledrm.c static struct simpledrm_device *simpledrm_device_create(struct drm_driver *drv, · line 785

Quoted 7 lines
drivers/gpu/drm/sysfb/simpledrm.c +3 -0
@@ -762,6 +780,13 @@ static struct simpledrm_device *simpledrm_device_create(struct drm_driver *drv,
  			mem = res;
  		}
 
+		if (resource_size(mem) > size) {
+			drm_err(dev,
+				"framebuffer size of %llu exceeds memory range %pr\n",

[Severity: High]
Is this check also inverted for the I/O memory path?

Similar to the system memory path above, this will reject valid framebuffers that fit within the memory range and allow undersized ones.

Quoted 5 lines
drivers/gpu/drm/sysfb/simpledrm.c +4 -0
+				size, mem);
+			return ERR_PTR(-EINVAL);
+		}
+
  		screen_base = devm_ioremap_wc(&pdev->dev, mem->start, resource_size(mem));
Signature · 6 lines
--
Thomas Zimmermann
Graphics Driver Developer
SUSE Software Solutions Germany GmbH
Frankenstr. 146, 90461 Nürnberg, Germany, www.suse.com
GF: Jochen Jaser, Andrew McDonald, Werner Knoblich, (HRB 36809, AG Nürnberg)
Reviewed-by on v2 5/6
#15 lore raw

Re: [PATCH v2 5/6] drm/sysfb: simpledrm: Validate mmap size against framebuffer size

In reply to Thomas Zimmermann
Quoted 34 lines · Thomas Zimmermann

The size of the mmap'ed framebuffer could be smaller than the minimum required framebuffer size. Validate the resource size against the framebuffer size.

Buggy firmware that triggers this check should be fixed up with a quirk on a case-by-case base.

---
 drivers/gpu/drm/sysfb/simpledrm.c | 25 +++++++++++++++++++++++++
 1 file changed, 25 insertions(+)
drivers/gpu/drm/sysfb/simpledrm.c +2 -0
index 76b9a3f5c4ef..a04f0c852ea4 100644
@@ -6,6 +6,7 @@
 #include <linux/of_address.h>
 #include <linux/of_clk.h>
 #include <linux/of_reserved_mem.h>
+#include <linux/overflow.h>
 #include <linux/platform_data/simplefb.h>
 #include <linux/platform_device.h>
 #include <linux/pm.h>
@@ -624,6 +625,7 @@ static struct simpledrm_device *simpledrm_device_create(struct drm_driver *drv,
 	u16 width_mm = 0, height_mm = 0;
 	struct device_node *panel_node;
 	const struct drm_format_info *format;
+	u64 size;
 	struct resource *res, *mem = NULL;
 	struct drm_plane *primary_plane;
 	struct drm_crtc *crtc;

drivers/gpu/drm/sysfb/simpledrm.c static struct simpledrm_device *simpledrm_device_create(struct drm_driver *drv, · line 715

Quoted 10 lines
drivers/gpu/drm/sysfb/simpledrm.c +7 -0
@@ -704,6 +706,15 @@ static struct simpledrm_device *simpledrm_device_create(struct drm_driver *drv,
 		}
 		stride = pitch;
 	}
+	if (check_mul_overflow(height, stride, &size)) {
+		drm_err(dev, "framebuffer size exceeds maximum\n");
+		return ERR_PTR(-EINVAL);
+	}
+	size = ALIGN(size, PAGE_SIZE);
+	if (size < PAGE_SIZE) {
+		drm_err(dev, "framebuffer alignment exceeds maximum\n");

That error message doesn't make sense to me. Maybe "framebuffer alignment below minimum", or something along those lines?

With that fixed:

author maintainer
#16 lore raw

Re: [PATCH v2 5/6] drm/sysfb: simpledrm: Validate mmap size against framebuffer size

Hi

Am 22.06.26 um 18:13 schrieb Thierry Reding:

Quoted 39 lines · Thomas Zimmermann
Quoted 38 lines

The size of the mmap'ed framebuffer could be smaller than the minimum required framebuffer size. Validate the resource size against the framebuffer size.

Buggy firmware that triggers this check should be fixed up with a quirk on a case-by-case base.

---
  drivers/gpu/drm/sysfb/simpledrm.c | 25 +++++++++++++++++++++++++
  1 file changed, 25 insertions(+)
drivers/gpu/drm/sysfb/simpledrm.c +3 -0
index 76b9a3f5c4ef..a04f0c852ea4 100644
@@ -6,6 +6,7 @@
  #include <linux/of_address.h>
  #include <linux/of_clk.h>
  #include <linux/of_reserved_mem.h>
+#include <linux/overflow.h>
  #include <linux/platform_data/simplefb.h>
  #include <linux/platform_device.h>
  #include <linux/pm.h>
@@ -624,6 +625,7 @@ static struct simpledrm_device *simpledrm_device_create(struct drm_driver *drv,
  	u16 width_mm = 0, height_mm = 0;
  	struct device_node *panel_node;
  	const struct drm_format_info *format;
+	u64 size;
  	struct resource *res, *mem = NULL;
  	struct drm_plane *primary_plane;
  	struct drm_crtc *crtc;
@@ -704,6 +706,15 @@ static struct simpledrm_device *simpledrm_device_create(struct drm_driver *drv,
  		}
  		stride = pitch;
  	}
+	if (check_mul_overflow(height, stride, &size)) {
Quoted 8 lines

drivers/gpu/drm/sysfb/simpledrm.c static struct simpledrm_device *simpledrm_device_create(struct drm_driver *drv, · line 715

Quoted 6 lines
drivers/gpu/drm/sysfb/simpledrm.c +6 -0
+		drm_err(dev, "framebuffer size exceeds maximum\n");
+		return ERR_PTR(-EINVAL);
+	}
+	size = ALIGN(size, PAGE_SIZE);
+	if (size < PAGE_SIZE) {
+		drm_err(dev, "framebuffer alignment exceeds maximum\n");

That error message doesn't make sense to me. Maybe "framebuffer alignment below minimum", or something along those lines?

This tests that the align operation did not overflow size. The result would then be 0. With the earlier patches limiting the height and stride and size being of u64, it's not possible any longer, I think. But testing it doesn't really cost us much.

What do you think of "aligned framebuffer size exceeds maximum"?

Best regards
Thomas

Quoted 4 lines

With that fixed:

Signature · 6 lines
--
Thomas Zimmermann
Graphics Driver Developer
SUSE Software Solutions Germany GmbH
Frankenstr. 146, 90461 Nürnberg, Germany, www.suse.com
GF: Jochen Jaser, Andrew McDonald, Werner Knoblich, (HRB 36809, AG Nürnberg)
#17 lore raw

Re: [PATCH v2 5/6] drm/sysfb: simpledrm: Validate mmap size against framebuffer size

Quoted 51 lines · Thomas Zimmermann

Hi

Am 22.06.26 um 18:13 schrieb Thierry Reding:

Quoted 39 lines · Thomas Zimmermann
Quoted 38 lines

The size of the mmap'ed framebuffer could be smaller than the minimum required framebuffer size. Validate the resource size against the framebuffer size.

Buggy firmware that triggers this check should be fixed up with a quirk on a case-by-case base.

---
  drivers/gpu/drm/sysfb/simpledrm.c | 25 +++++++++++++++++++++++++
  1 file changed, 25 insertions(+)
drivers/gpu/drm/sysfb/simpledrm.c +3 -0
index 76b9a3f5c4ef..a04f0c852ea4 100644
@@ -6,6 +6,7 @@
  #include <linux/of_address.h>
  #include <linux/of_clk.h>
  #include <linux/of_reserved_mem.h>
+#include <linux/overflow.h>
  #include <linux/platform_data/simplefb.h>
  #include <linux/platform_device.h>
  #include <linux/pm.h>
@@ -624,6 +625,7 @@ static struct simpledrm_device *simpledrm_device_create(struct drm_driver *drv,
  	u16 width_mm = 0, height_mm = 0;
  	struct device_node *panel_node;
  	const struct drm_format_info *format;
+	u64 size;
  	struct resource *res, *mem = NULL;
  	struct drm_plane *primary_plane;
  	struct drm_crtc *crtc;
@@ -704,6 +706,15 @@ static struct simpledrm_device *simpledrm_device_create(struct drm_driver *drv,
  		}
  		stride = pitch;
  	}
+	if (check_mul_overflow(height, stride, &size)) {
Quoted 8 lines

drivers/gpu/drm/sysfb/simpledrm.c static struct simpledrm_device *simpledrm_device_create(struct drm_driver *drv, · line 715

Quoted 6 lines
drivers/gpu/drm/sysfb/simpledrm.c +6 -0
+		drm_err(dev, "framebuffer size exceeds maximum\n");
+		return ERR_PTR(-EINVAL);
+	}
+	size = ALIGN(size, PAGE_SIZE);
+	if (size < PAGE_SIZE) {
+		drm_err(dev, "framebuffer alignment exceeds maximum\n");

That error message doesn't make sense to me. Maybe "framebuffer alignment below minimum", or something along those lines?

Quoted 6 lines

This tests that the align operation did not overflow size. The result would then be 0. With the earlier patches limiting the height and stride and size being of u64, it's not possible any longer, I think. But testing it doesn't really cost us much.

What do you think of "aligned framebuffer size exceeds maximum"?

Ah, I misunderstood what the code was doing. Yes, I think that message makes it a bit clearer what the error is.

Thierry

author maintainer
#18 lore raw

PATCH v2 6/6 drm/sysfb: simpledrm: Read panel orientation from DT node

In reply to Thomas Zimmermann

A DeviceTree panel node can specify the panel's rotation in steps of 90 degrees. Set the DRM connector orientation from the field. Do nothing if the field is missing or invalid.

---
 drivers/gpu/drm/sysfb/simpledrm.c | 32 +++++++++++++++++++++++++++++--
 1 file changed, 30 insertions(+), 2 deletions(-)
drivers/gpu/drm/sysfb/simpledrm.c +30 -2
index a04f0c852ea4..b585d17ca817 100644
@@ -221,6 +221,33 @@ simplefb_get_panel_height_mm_of(struct drm_device *dev, struct device_node *of_p
 	return __simplefb_get_panel_size_mm_of(dev, of_panel_node, "height-mm");
 }
 
+static enum drm_panel_orientation
+simplefb_get_panel_orientation_of(struct drm_device *dev, struct device_node *of_panel_node)
+{
+	int ret;
+	u32 rotation;
+
+	ret = of_property_read_u32(of_panel_node, "rotation", &rotation);
+	if (ret) {
+		drm_dbg(dev, "simplefb: cannot parse panel rotation: error %d\n", ret);
+		return DRM_MODE_PANEL_ORIENTATION_UNKNOWN;
+	}
+
+	switch (rotation) {
+	case 0:
+		return DRM_MODE_PANEL_ORIENTATION_NORMAL;
+	case 90:
+		return DRM_MODE_PANEL_ORIENTATION_RIGHT_UP;
+	case 180:
+		return DRM_MODE_PANEL_ORIENTATION_BOTTOM_UP;
+	case 270:
+		return DRM_MODE_PANEL_ORIENTATION_LEFT_UP;
+	default:
+		drm_dbg(dev, "simplefb: cannot parse panel rotation: value %u\n", rotation);
+		return DRM_MODE_PANEL_ORIENTATION_UNKNOWN;
+	}
+}
+
 /*
  * Simple Framebuffer device
  */
@@ -624,6 +651,7 @@ static struct simpledrm_device *simpledrm_device_create(struct drm_driver *drv,
 	int width, height, stride;
 	u16 width_mm = 0, height_mm = 0;
 	struct device_node *panel_node;
+	enum drm_panel_orientation orientation = DRM_MODE_PANEL_ORIENTATION_UNKNOWN;
 	const struct drm_format_info *format;
 	u64 size;
 	struct resource *res, *mem = NULL;
@@ -689,6 +717,7 @@ static struct simpledrm_device *simpledrm_device_create(struct drm_driver *drv,
 		if (panel_node) {
 			width_mm = simplefb_get_panel_width_mm_of(dev, panel_node);
 			height_mm = simplefb_get_panel_height_mm_of(dev, panel_node);
+			orientation = simplefb_get_panel_orientation_of(dev, panel_node);
 			of_node_put(panel_node);
 		}
 	} else {
@@ -853,8 +882,7 @@ static struct simpledrm_device *simpledrm_device_create(struct drm_driver *drv,
 	if (ret)
 		return ERR_PTR(ret);
 	drm_connector_helper_add(connector, &simpledrm_connector_helper_funcs);
-	drm_connector_set_panel_orientation_with_quirk(connector,
-						       DRM_MODE_PANEL_ORIENTATION_UNKNOWN,
+	drm_connector_set_panel_orientation_with_quirk(connector, orientation,
 						       width, height);
 
 	ret = drm_connector_attach_encoder(connector, encoder);
Signature · 1 line
2.54.0
#19 lore raw

Re: [PATCH v2 6/6] drm/sysfb: simpledrm: Read panel orientation from DT node

Thank you for your contribution! Sashiko AI review found 2 potential issue(s) to consider: - [High] Blindly applying physical panel rotation to a logical framebuffer can cause double rotation if the…

Thank you for your contribution! Sashiko AI review found 2 potential issue(s) to consider:
- [High] Blindly applying physical panel rotation to a logical framebuffer can cause double rotation if the firmware already applied hardware rotation.
- [Low] The patch duplicates the existing of_drm_get_panel_orientation() helper and introduces noisy debug logging for normal conditions. --

commit a3e8de9befee02f75cbf1e8ca7ff150f4438c5f5
Author: Thomas Zimmermann <[email protected]>

drm/sysfb: simpledrm: Read panel orientation from DT node

A DeviceTree panel node can specify the panel's rotation in steps of 90 degrees. Set the DRM connector orientation from the field. Do nothing if the field is missing or invalid.

Quoted 7 lines
drivers/gpu/drm/sysfb/simpledrm.c +0 -0
index a04f0c852ea48..b585d17ca8177 100644
@@ -221,6 +221,33 @@ simplefb_get_panel_height_mm_of(struct drm_device *dev, struct device_node *of_p
 	return __simplefb_get_panel_size_mm_of(dev, of_panel_node, "height-mm");
 }

drivers/gpu/drm/sysfb/simpledrm.c simplefb_get_panel_height_mm_of(struct drm_device *dev, struct device_node *of_p · line 232

Quoted 10 lines
drivers/gpu/drm/sysfb/simpledrm.c +9 -0
 
+static enum drm_panel_orientation
+simplefb_get_panel_orientation_of(struct drm_device *dev, struct device_node *of_panel_node)
+{
+	int ret;
+	u32 rotation;
+
+	ret = of_property_read_u32(of_panel_node, "rotation", &rotation);
+	if (ret) {
+		drm_dbg(dev, "simplefb: cannot parse panel rotation: error %d\n", ret);

[Severity: Low]
Is it intentional to log a debug message here when the property is missing?

Since most device tree panel nodes do not include a rotation property, of_property_read_u32() will return -EINVAL and spam the DRM debug logs with "cannot parse panel rotation: error -22" on almost every boot when debug logging is enabled.

Also, does this function reimplement the parsing logic from of_drm_get_panel_orientation() which explicitly treats missing properties silently?

Quoted 6 lines
drivers/gpu/drm/sysfb/simpledrm.c +6 -0
+		return DRM_MODE_PANEL_ORIENTATION_UNKNOWN;
+	}
+
+	switch (rotation) {
+	case 0:
+		return DRM_MODE_PANEL_ORIENTATION_NORMAL;

[ ... ]

drivers/gpu/drm/sysfb/simpledrm.c static struct simpledrm_device *simpledrm_device_create(struct drm_driver *drv, · line 720

Quoted 5 lines
drivers/gpu/drm/sysfb/simpledrm.c +1 -0
@@ -689,6 +717,7 @@ static struct simpledrm_device *simpledrm_device_create(struct drm_driver *drv,
 		if (panel_node) {
 			width_mm = simplefb_get_panel_width_mm_of(dev, panel_node);
 			height_mm = simplefb_get_panel_height_mm_of(dev, panel_node);
+			orientation = simplefb_get_panel_orientation_of(dev, panel_node);

[Severity: High]
Does unconditionally applying the physical panel rotation to the logical framebuffer risk double rotation?

The simpledrm driver operates on a logical framebuffer pre-configured by the firmware. If the firmware has already configured the display controller to perform hardware rotation (e.g., exposing a 1920x1080 memory buffer for a physically 1080x1920 panel mounted sideways), the simplefb node describes an already-upright logical framebuffer.

If simpledrm blindly applies the physical rotation from the panel node to the DRM connector, userspace compositors might apply a software rotation on top of the firmware's hardware rotation, causing the final image to be rendered sideways or distorted.

Is there a way to verify if the firmware already applied hardware rotation before mapping physical panel properties here?

Quoted 3 lines
drivers/gpu/drm/sysfb/simpledrm.c +0 -0
 			of_node_put(panel_node);
 		}
 	} else {
Signature · 1 line
Sashiko AI review · https://sashiko.dev/#/patchset/[email protected]?part=6
#20 lore raw

Re: [PATCH v2 6/6] drm/sysfb: simpledrm: Read panel orientation from DT node

In reply to Thomas Zimmermann
Quoted 16 lines · Thomas Zimmermann

A DeviceTree panel node can specify the panel's rotation in steps of 90 degrees. Set the DRM connector orientation from the field. Do nothing if the field is missing or invalid.

---
 drivers/gpu/drm/sysfb/simpledrm.c | 32 +++++++++++++++++++++++++++++--
 1 file changed, 30 insertions(+), 2 deletions(-)
drivers/gpu/drm/sysfb/simpledrm.c +0 -0
index a04f0c852ea4..b585d17ca817 100644
@@ -221,6 +221,33 @@ simplefb_get_panel_height_mm_of(struct drm_device *dev, struct device_node *of_p
 	return __simplefb_get_panel_size_mm_of(dev, of_panel_node, "height-mm");
 }

drivers/gpu/drm/sysfb/simpledrm.c simplefb_get_panel_height_mm_of(struct drm_device *dev, struct device_node *of_p · line 232

Quoted 10 lines
drivers/gpu/drm/sysfb/simpledrm.c +9 -0
 
+static enum drm_panel_orientation
+simplefb_get_panel_orientation_of(struct drm_device *dev, struct device_node *of_panel_node)
+{
+	int ret;
+	u32 rotation;
+
+	ret = of_property_read_u32(of_panel_node, "rotation", &rotation);
+	if (ret) {
+		drm_dbg(dev, "simplefb: cannot parse panel rotation: error %d\n", ret);

Why is this debug level? Maybe we don't want an error if the property is absent, but if it really can't be parsed, why not print out a proper error?

Quoted 4 lines
drivers/gpu/drm/sysfb/simpledrm.c +4 -0
+		return DRM_MODE_PANEL_ORIENTATION_UNKNOWN;
+	}
+
+	switch (rotation) {

drivers/gpu/drm/sysfb/simpledrm.c simplefb_get_panel_height_mm_of(struct drm_device *dev, struct device_node *of_p · line 246

Quoted 10 lines
drivers/gpu/drm/sysfb/simpledrm.c +10 -0
+	case 0:
+		return DRM_MODE_PANEL_ORIENTATION_NORMAL;
+	case 90:
+		return DRM_MODE_PANEL_ORIENTATION_RIGHT_UP;
+	case 180:
+		return DRM_MODE_PANEL_ORIENTATION_BOTTOM_UP;
+	case 270:
+		return DRM_MODE_PANEL_ORIENTATION_LEFT_UP;
+	default:
+		drm_dbg(dev, "simplefb: cannot parse panel rotation: value %u\n", rotation);

Same here. Seems like we should be verbose when this contains an invalid value. DT checkers should catch these already, so if the invalid value makes it here, maybe we should sound the alarm.

Thierry

author maintainer
#21 lore raw

Re: [PATCH v2 6/6] drm/sysfb: simpledrm: Read panel orientation from DT node

Hi

Am 22.06.26 um 18:17 schrieb Thierry Reding:

Quoted 40 lines · Thomas Zimmermann
Quoted 16 lines

A DeviceTree panel node can specify the panel's rotation in steps of 90 degrees. Set the DRM connector orientation from the field. Do nothing if the field is missing or invalid.

---
  drivers/gpu/drm/sysfb/simpledrm.c | 32 +++++++++++++++++++++++++++++--
  1 file changed, 30 insertions(+), 2 deletions(-)
drivers/gpu/drm/sysfb/simpledrm.c +0 -0
index a04f0c852ea4..b585d17ca817 100644
@@ -221,6 +221,33 @@ simplefb_get_panel_height_mm_of(struct drm_device *dev, struct device_node *of_p
  	return __simplefb_get_panel_size_mm_of(dev, of_panel_node, "height-mm");
  }

drivers/gpu/drm/sysfb/simpledrm.c simplefb_get_panel_height_mm_of(struct drm_device *dev, struct device_node *of_p · line 232

Quoted 10 lines
drivers/gpu/drm/sysfb/simpledrm.c +9 -0
 
+static enum drm_panel_orientation
+simplefb_get_panel_orientation_of(struct drm_device *dev, struct device_node *of_panel_node)
+{
+	int ret;
+	u32 rotation;
+
+	ret = of_property_read_u32(of_panel_node, "rotation", &rotation);
+	if (ret) {
+		drm_dbg(dev, "simplefb: cannot parse panel rotation: error %d\n", ret);

Why is this debug level? Maybe we don't want an error if the property is absent, but if it really can't be parsed, why not print out a proper error?

Quoted 9 lines
drivers/gpu/drm/sysfb/simpledrm.c +9 -0
+		return DRM_MODE_PANEL_ORIENTATION_UNKNOWN;
+	}
+
+	switch (rotation) {
+	case 0:
+		return DRM_MODE_PANEL_ORIENTATION_NORMAL;
+	case 90:
+		return DRM_MODE_PANEL_ORIENTATION_RIGHT_UP;
+	case 180:
Quoted 8 lines

drivers/gpu/drm/sysfb/simpledrm.c simplefb_get_panel_height_mm_of(struct drm_device *dev, struct device_node *of_p · line 246

Quoted 5 lines
drivers/gpu/drm/sysfb/simpledrm.c +5 -0
+		return DRM_MODE_PANEL_ORIENTATION_BOTTOM_UP;
+	case 270:
+		return DRM_MODE_PANEL_ORIENTATION_LEFT_UP;
+	default:
+		drm_dbg(dev, "simplefb: cannot parse panel rotation: value %u\n", rotation);

Same here. Seems like we should be verbose when this contains an invalid value. DT checkers should catch these already, so if the invalid value makes it here, maybe we should sound the alarm.

Ok. The review bot told me about of_drm_get_panel_orientation(). I intent to reimplement this patch on top of that helper.

Best regards
Thomas

Quoted 2 lines

Thierry

Signature · 6 lines
--
Thomas Zimmermann
Graphics Driver Developer
SUSE Software Solutions Germany GmbH
Frankenstr. 146, 90461 Nürnberg, Germany, www.suse.com
GF: Jochen Jaser, Andrew McDonald, Werner Knoblich, (HRB 36809, AG Nürnberg)
#22 lore raw

Re: [PATCH v2 6/6] drm/sysfb: simpledrm: Read panel orientation from DT node

Quoted 51 lines · Thomas Zimmermann

Hi

Am 22.06.26 um 18:17 schrieb Thierry Reding:

Quoted 39 lines · Thomas Zimmermann
Quoted 15 lines

A DeviceTree panel node can specify the panel's rotation in steps of 90 degrees. Set the DRM connector orientation from the field. Do nothing if the field is missing or invalid.

---
  drivers/gpu/drm/sysfb/simpledrm.c | 32 +++++++++++++++++++++++++++++--
  1 file changed, 30 insertions(+), 2 deletions(-)
drivers/gpu/drm/sysfb/simpledrm.c +0 -0
index a04f0c852ea4..b585d17ca817 100644
@@ -221,6 +221,33 @@ simplefb_get_panel_height_mm_of(struct drm_device *dev, struct device_node *of_p
  	return __simplefb_get_panel_size_mm_of(dev, of_panel_node, "height-mm");

drivers/gpu/drm/sysfb/simpledrm.c simplefb_get_panel_height_mm_of(struct drm_device *dev, struct device_node *of_p · line 231

Quoted 10 lines
drivers/gpu/drm/sysfb/simpledrm.c +9 -0
  }
+static enum drm_panel_orientation
+simplefb_get_panel_orientation_of(struct drm_device *dev, struct device_node *of_panel_node)
+{
+	int ret;
+	u32 rotation;
+
+	ret = of_property_read_u32(of_panel_node, "rotation", &rotation);
+	if (ret) {
+		drm_dbg(dev, "simplefb: cannot parse panel rotation: error %d\n", ret);

Why is this debug level? Maybe we don't want an error if the property is absent, but if it really can't be parsed, why not print out a proper error?

Quoted 9 lines
drivers/gpu/drm/sysfb/simpledrm.c +9 -0
+		return DRM_MODE_PANEL_ORIENTATION_UNKNOWN;
+	}
+
+	switch (rotation) {
+	case 0:
+		return DRM_MODE_PANEL_ORIENTATION_NORMAL;
+	case 90:
+		return DRM_MODE_PANEL_ORIENTATION_RIGHT_UP;
+	case 180:
Quoted 8 lines

drivers/gpu/drm/sysfb/simpledrm.c simplefb_get_panel_height_mm_of(struct drm_device *dev, struct device_node *of_p · line 246

Quoted 5 lines
drivers/gpu/drm/sysfb/simpledrm.c +5 -0
+		return DRM_MODE_PANEL_ORIENTATION_BOTTOM_UP;
+	case 270:
+		return DRM_MODE_PANEL_ORIENTATION_LEFT_UP;
+	default:
+		drm_dbg(dev, "simplefb: cannot parse panel rotation: value %u\n", rotation);

Same here. Seems like we should be verbose when this contains an invalid value. DT checkers should catch these already, so if the invalid value makes it here, maybe we should sound the alarm.

Quoted 2 lines

Ok. The review bot told me about of_drm_get_panel_orientation(). I intent to reimplement this patch on top of that helper.

Makes sense, yeah.

Thierry