Skip to contentSkip to conversation

auxdisplay: ht16k33: don't access uninitialized data

PATCH In review Arnd Bergmann 3 participants 3 messages Mar 28, 2017 – Mar 29, 2017

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

PATCH 1/1 auxdisplay: ht16k33: don't access uninitialized data

gcc-7.0.1 points out that we copy uninitialized data from the stack into a per-device structure:

drivers/auxdisplay/ht16k33.c: In function 'ht16k33_keypad_irq_thread':
arch/x86/include/asm/string_32.h:78:16: error: 'new_state' may be used uninitialized in this function [-Werror=maybe-uninitialized] arch/x86/include/asm/string_32.h:79:22: error: '*((void *)&new_state+4)' may be used uninitialized in this function [-Werror=maybe-uninitialized]

The access is harmless because we never read the data, but we are better off not doing this, so this changes the code to only copy the data that was actually initialized. To make sure we don't overflow the stack with an incorrect DT, we also need to add a sanity checkin the probe function.

---
 drivers/auxdisplay/ht16k33.c | 8 +++++++-
 1 file changed, 7 insertions(+), 1 deletion(-)
drivers/auxdisplay/ht16k33.c +7 -1
index f66b45b235b0..ba6370974574 100644
@@ -278,7 +278,7 @@ static bool ht16k33_keypad_scan(struct ht16k33_keypad *keypad)
 		}
 	}
 	input_sync(keypad->dev);
-	memcpy(keypad->last_key_state, new_state, sizeof(new_state));
+	memcpy(keypad->last_key_state, new_state, sizeof(u16) * keypad->cols);
 
 	return pressed;
 }
@@ -353,6 +353,12 @@ static int ht16k33_keypad_probe(struct i2c_client *client,
 	err = matrix_keypad_parse_of_params(&client->dev, &rows, &cols);
 	if (err)
 		return err;
+	if (rows > HT16K33_MATRIX_KEYPAD_MAX_ROWS ||
+	    cols > HT16K33_MATRIX_KEYPAD_MAX_COLS) {
+		dev_err(&client->dev, "%u rows or %u cols out of range in DT\n",
+			rows, cols);
+		return -ERANGE;
+	}
 
 	keypad->rows = rows;
 	keypad->cols = cols;
Signature · 1 line
2.9.0
Reviewed-by on v1 1/1
#1 lore raw
Quoted 7 lines · Arnd Bergmann

gcc-7.0.1 points out that we copy uninitialized data from the stack into a per-device structure:

drivers/auxdisplay/ht16k33.c: In function 'ht16k33_keypad_irq_thread':
arch/x86/include/asm/string_32.h:78:16: error: 'new_state' may be used uninitialized in this function [-Werror=maybe-uninitialized] arch/x86/include/asm/string_32.h:79:22: error: '*((void *)&new_state+4)' may be used uninitialized in this function [-Werror=maybe-uninitialized]

Quoted 7 lines

The access is harmless because we never read the data, but we are better off not doing this, so this changes the code to only copy the data that was actually initialized. To make sure we don't overflow the stack with an incorrect DT, we also need to add a sanity checkin the probe function.

Quoted 33 lines
---
 drivers/auxdisplay/ht16k33.c | 8 +++++++-
 1 file changed, 7 insertions(+), 1 deletion(-)
drivers/auxdisplay/ht16k33.c +7 -1
index f66b45b235b0..ba6370974574 100644
@@ -278,7 +278,7 @@ static bool ht16k33_keypad_scan(struct ht16k33_keypad *keypad)
 		}
 	}
 	input_sync(keypad->dev);
-	memcpy(keypad->last_key_state, new_state, sizeof(new_state));
+	memcpy(keypad->last_key_state, new_state, sizeof(u16) * keypad->cols);
 
 	return pressed;
 }
@@ -353,6 +353,12 @@ static int ht16k33_keypad_probe(struct i2c_client *client,
 	err = matrix_keypad_parse_of_params(&client->dev, &rows, &cols);
 	if (err)
 		return err;
+	if (rows > HT16K33_MATRIX_KEYPAD_MAX_ROWS ||
+	    cols > HT16K33_MATRIX_KEYPAD_MAX_COLS) {
+		dev_err(&client->dev, "%u rows or %u cols out of range in DT\n",
+			rows, cols);
+		return -ERANGE;
+	}
 
 	keypad->rows = rows;
 	keypad->cols = cols;
Signature · 1 line
2.9.0
Signature · 1 line
Dmitry
maintainer Reviewed-by on v1 1/1
#2 lore raw
In reply to Arnd Bergmann
Quoted 7 lines · On Tue, 28 Mar 2017 12:11:49 +0200 Arnd Bergmann <[email protected]> wrote:

gcc-7.0.1 points out that we copy uninitialized data from the stack into a per-device structure:

drivers/auxdisplay/ht16k33.c: In function 'ht16k33_keypad_irq_thread':
arch/x86/include/asm/string_32.h:78:16: error: 'new_state' may be used uninitialized in this function [-Werror=maybe-uninitialized] arch/x86/include/asm/string_32.h:79:22: error: '*((void *)&new_state+4)' may be used uninitialized in this function [-Werror=maybe-uninitialized]

Quoted 7 lines

The access is harmless because we never read the data, but we are better off not doing this, so this changes the code to only copy the data that was actually initialized. To make sure we don't overflow the stack with an incorrect DT, we also need to add a sanity checkin the probe function.

Quoted 30 lines
---
 drivers/auxdisplay/ht16k33.c | 8 +++++++-
 1 file changed, 7 insertions(+), 1 deletion(-)
drivers/auxdisplay/ht16k33.c +7 -1
index f66b45b235b0..ba6370974574 100644
@@ -278,7 +278,7 @@ static bool ht16k33_keypad_scan(struct ht16k33_keypad *keypad)
 		}
 	}
 	input_sync(keypad->dev);
-	memcpy(keypad->last_key_state, new_state, sizeof(new_state));
+	memcpy(keypad->last_key_state, new_state, sizeof(u16) * keypad->cols);
 
 	return pressed;
 }
@@ -353,6 +353,12 @@ static int ht16k33_keypad_probe(struct i2c_client *client,
 	err = matrix_keypad_parse_of_params(&client->dev, &rows, &cols);
 	if (err)
 		return err;
+	if (rows > HT16K33_MATRIX_KEYPAD_MAX_ROWS ||
+	    cols > HT16K33_MATRIX_KEYPAD_MAX_COLS) {
+		dev_err(&client->dev, "%u rows or %u cols out of range in DT\n",
+			rows, cols);
+		return -ERANGE;
+	}
 
 	keypad->rows = rows;
 	keypad->cols = cols;