Skip to contentSkip to conversation

Error handling corner case found during audits

PATCH Applied Alan Cox 3 participants 3 messages Apr 29, 2008 – May 6, 2008

Subsystems (2)
lore
Share & tools
lore

Alan Cox

author
#0 lore raw

Not sure what should happen here.

drivers/net/e1000e/ethtool.c +2 -0
diff -u --new-file --recursive --exclude-from /usr/src/exclude linux.vanilla-2.6.25-mm1/drivers/net/e1000e/ethtool.c linux-2.6.25-mm1/drivers/net/e1000e/ethtool.c
@@ -494,6 +494,8 @@
 		for (i = 0; i < last_word - first_word + 1; i++) {
 			ret_val = e1000_read_nvm(hw, first_word + i, 1,
 						      &eeprom_buff[i]);
+			/* ERROR: This path leaves eeprom_buf containing
+			   old kernel bytes we then byteswap/return */
 			if (ret_val)
 				break;
 		}
#1 lore raw
Quoted 7 lines · Alan Cox

Not sure what should happen here.

diff -u --new-file --recursive --exclude-from /usr/src/exclude linux.vanilla-2.6.25-mm1/drivers/net/e1000e/ethtool.c linux-2.6.25-mm1/drivers/net/e1000e/ethtool.c

drivers/net/e1000e/ethtool.c line 501

Quoted 8 lines
drivers/net/e1000e/ethtool.c +2 -0
@@ -494,6 +494,8 @@
 		for (i = 0; i < last_word - first_word + 1; i++) {
 			ret_val = e1000_read_nvm(hw, first_word + i, 1,
 						      &eeprom_buff[i]);
+			/* ERROR: This path leaves eeprom_buf containing
+			   old kernel bytes we then byteswap/return */
 			if (ret_val)
 				break;
 		}

either we fill the buffer with 0xff (the determined value for "empty eeprom"), or just kzalloc the buffer instead. This should be enough of a warning for the user that something is really wrong.

Auke

---

e1000e: don't return half-read eeprom on error

On a read error, e1000e might have returned uninitialized block of eeprom data back to userspace. The convention is that 0xff is "empty", so mark the entire eeprom as empty in case of an error.

---

drivers/net/e1000e/ethtool.c +5 -1
index b1b784a..8b04a42 100644
@@ -510,8 +510,12 @@ static int e1000_get_eeprom(struct net_device *netdev,
 		for (i = 0; i < last_word - first_word + 1; i++) {
 			ret_val = e1000_read_nvm(hw, first_word + i, 1,
 						      &eeprom_buff[i]);
-			if (ret_val)
+			if (ret_val) {
+				/* a read error occurred, throw away the
+				 * result */
+				memset(eeprom_buff, 0xff, sizeof(eeprom_buff));
 				break;
+			}
 		}
 	}
 
Applied
#2 lore raw
Quoted 41 lines · Auke
Quoted 7 lines · Alan Cox

Not sure what should happen here.

diff -u --new-file --recursive --exclude-from /usr/src/exclude linux.vanilla-2.6.25-mm1/drivers/net/e1000e/ethtool.c linux-2.6.25-mm1/drivers/net/e1000e/ethtool.c

drivers/net/e1000e/ethtool.c line 501

Quoted 8 lines
drivers/net/e1000e/ethtool.c +2 -0
@@ -494,6 +494,8 @@
 		for (i = 0; i < last_word - first_word + 1; i++) {
 			ret_val = e1000_read_nvm(hw, first_word + i, 1,
 						      &eeprom_buff[i]);
+			/* ERROR: This path leaves eeprom_buf containing
+			   old kernel bytes we then byteswap/return */
 			if (ret_val)
 				break;
 		}

either we fill the buffer with 0xff (the determined value for "empty eeprom"), or just kzalloc the buffer instead. This should be enough of a warning for the user that something is really wrong.

Auke

---

e1000e: don't return half-read eeprom on error

On a read error, e1000e might have returned uninitialized block of eeprom data back to userspace. The convention is that 0xff is "empty", so mark the entire eeprom as empty in case of an error.

---

drivers/net/e1000e/ethtool.c +0 -0
index b1b784a..8b04a42 100644
@@ -510,8 +510,12 @@ static int e1000_get_eeprom(struct net_device *netdev,
 		for (i = 0; i < last_word - first_word + 1; i++) {
 			ret_val = e1000_read_nvm(hw, first_word + i, 1,

drivers/net/e1000e/ethtool.c static int e1000_get_eeprom(struct net_device *netdev, · line 520

Quoted 10 lines
drivers/net/e1000e/ethtool.c +5 -1
 						      &eeprom_buff[i]);
-			if (ret_val)
+			if (ret_val) {
+				/* a read error occurred, throw away the
+				 * result */
+				memset(eeprom_buff, 0xff, sizeof(eeprom_buff));
 				break;
+			}
 		}
 	}

applied