Skip to contentSkip to conversation

drm/amdgpu: fix a signedness bug in __verify_ras_table_checksum()

PATCH In review Dan Carpenter 2 participants 6 messages Jul 3, 2021 – Jul 6, 2021

Subsystems (3)
lore
Share & tools
lore
#0 lore raw

PATCH 1/4 drm/amdgpu: fix a signedness bug in __verify_ras_table_checksum()

If amdgpu_eeprom_read() returns a negative error code then the error handling checks:

	if (res < buf_size) {

The problem is that "buf_size" is a u32 so negative values are type promoted to a high positive values and the condition is false. Fix this by changing the type of "buf_size" to int.

Fixes 79beb6114014 ("drm/amdgpu: Optimize EEPROM RAS table I/O")

---
It's hard for me to tell the exact upper bound that "buf_size" can be, but if it's over USHRT_MAX then we are well toasted.

 drivers/gpu/drm/amd/amdgpu/amdgpu_ras_eeprom.c | 3 +--
 1 file changed, 1 insertion(+), 2 deletions(-)
drivers/gpu/drm/amd/amdgpu/amdgpu_ras_eeprom.c +1 -2
index fc70620369e4..f07a456506ef 100644
@@ -978,9 +978,8 @@ const struct file_operations amdgpu_ras_debugfs_eeprom_table_ops = {
 static int __verify_ras_table_checksum(struct amdgpu_ras_eeprom_control *control)
 {
 	struct amdgpu_device *adev = to_amdgpu_device(control);
-	int res;
+	int buf_size, res;
 	u8  csum, *buf, *pp;
-	u32 buf_size;
 
 	buf_size = RAS_TABLE_HEADER_SIZE +
 		control->ras_num_recs * RAS_TABLE_RECORD_SIZE;
Signature · 1 line
2.30.2
#1 lore raw

PATCH 2/4 drm/amdgpu: Fix signedness bug in __amdgpu_eeprom_xfer()

The i2c_transfer() function returns negatives or else the number of messages transferred. This code does not work because ARRAY_SIZE() is type size_t and so that means negative values of "r" are type promoted to high positive values which are greater than the ARRAY_SIZE().

Fix this by changing the < to != which works regardless of type promotion.

Fixes 6cda0af81a50 ("drm/amdgpu: Fixes to the AMDGPU EEPROM driver")
---
 drivers/gpu/drm/amd/amdgpu/amdgpu_eeprom.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/gpu/drm/amd/amdgpu/amdgpu_eeprom.c +1 -1
index 4c3c65a5acae..4d9eb0137f8c 100644
@@ -147,7 +147,7 @@ static int __amdgpu_eeprom_xfer(struct i2c_adapter *i2c_adap, u32 eeprom_addr,
 		/* This constitutes a START-STOP transaction.
 		 */
 		r = i2c_transfer(i2c_adap, msgs, ARRAY_SIZE(msgs));
-		if (r < ARRAY_SIZE(msgs))
+		if (r != ARRAY_SIZE(msgs))
 			break;
 
 		if (!read) {
Signature · 6 lines
2.30.2

_______________________________________________
amd-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/amd-gfx
#2 lore raw

PATCH 3/4 drm/amdgpu: unlock on error in amdgpu_ras_debugfs_table_read()

In reply to Dan Carpenter

This error path needs to unlock before returning. While we're at it, the correct error code from copy_to_user() failure is -EFAULT, not -EINVAL.

Fixes 9b790694a031 (""drm/amdgpu: RAS EEPROM table is now in debugfs)
---
 drivers/gpu/drm/amd/amdgpu/amdgpu_ras_eeprom.c | 8 +++++---
 1 file changed, 5 insertions(+), 3 deletions(-)
drivers/gpu/drm/amd/amdgpu/amdgpu_ras_eeprom.c +5 -3
index f07a456506ef..3e33e85d8dbc 100644
@@ -821,7 +821,7 @@ static ssize_t amdgpu_ras_debugfs_table_read(struct file *f, char __user *buf,
 	struct amdgpu_ras *ras = amdgpu_ras_get_context(adev);
 	struct amdgpu_ras_eeprom_control *control = &ras->eeprom_control;
 	const size_t orig_size = size;
-	int res = -EINVAL;
+	int res = -EFAULT;
 	size_t data_len;
 
 	mutex_lock(&control->ras_tbl_mutex);
@@ -912,8 +912,10 @@ static ssize_t amdgpu_ras_debugfs_table_read(struct file *f, char __user *buf,
 				 record.retired_page);
 
 			data_len = min_t(size_t, rec_hdr_fmt_size - r, size);
-			if (copy_to_user(buf, &data[r], data_len))
-				return -EINVAL;
+			if (copy_to_user(buf, &data[r], data_len)) {
+				res = -EFAULT;
+				goto Out;
+			}
 			buf += data_len;
 			size -= data_len;
 			*pos += data_len;
Signature · 6 lines
2.30.2

_______________________________________________
amd-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/amd-gfx
#3 lore raw

PATCH 4/4 drm/amdgpu: return -EFAULT if copy_to_user() fails

In reply to Dan Carpenter

If copy_to_user() fails then this should return -EFAULT instead of -EINVAL.

Fixes 9b790694a031 ("drm/amdgpu: RAS EEPROM table is now in debugfs")
---
 drivers/gpu/drm/amd/amdgpu/amdgpu_ras_eeprom.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
drivers/gpu/drm/amd/amdgpu/amdgpu_ras_eeprom.c +2 -2
index 3e33e85d8dbc..d2e5b2567bc1 100644
@@ -771,7 +771,7 @@ amdgpu_ras_debugfs_eeprom_size_read(struct file *f, char __user *buf,
 	res = min_t(size_t, res, size);
 
 	if (copy_to_user(buf, &data[*pos], res))
-		return -EINVAL;
+		return -EFAULT;
 
 	*pos += res;
 
@@ -950,7 +950,7 @@ amdgpu_ras_debugfs_eeprom_table_read(struct file *f, char __user *buf,
 		res = min_t(size_t, res, size);
 
 		if (copy_to_user(buf, &data[*pos], res))
-			return -EINVAL;
+			return -EFAULT;
 
 		*pos += res;
 
Signature · 1 line
2.30.2
Reviewed-by on v1 1/4
#4 lore raw
In reply to Dan Carpenter

Series is,

Regards,
Luben

Quoted 33 lines · Dan Carpenter

If amdgpu_eeprom_read() returns a negative error code then the error handling checks:

	if (res < buf_size) {

The problem is that "buf_size" is a u32 so negative values are type promoted to a high positive values and the condition is false. Fix this by changing the type of "buf_size" to int.

Fixes 79beb6114014 ("drm/amdgpu: Optimize EEPROM RAS table I/O")

---
It's hard for me to tell the exact upper bound that "buf_size" can be, but if it's over USHRT_MAX then we are well toasted.

 drivers/gpu/drm/amd/amdgpu/amdgpu_ras_eeprom.c | 3 +--
 1 file changed, 1 insertion(+), 2 deletions(-)
drivers/gpu/drm/amd/amdgpu/amdgpu_ras_eeprom.c +1 -2
index fc70620369e4..f07a456506ef 100644
@@ -978,9 +978,8 @@ const struct file_operations amdgpu_ras_debugfs_eeprom_table_ops = {
 static int __verify_ras_table_checksum(struct amdgpu_ras_eeprom_control *control)
 {
 	struct amdgpu_device *adev = to_amdgpu_device(control);
-	int res;
+	int buf_size, res;
 	u8  csum, *buf, *pp;
-	u32 buf_size;
 
 	buf_size = RAS_TABLE_HEADER_SIZE +
 		control->ras_num_recs * RAS_TABLE_RECORD_SIZE;
#5 lore raw

Alex,

I think we should pull these through amd-staging-drm-next.

Regards,
Luben

Quoted 38 lines · Luben Tuikov

Series is,

Regards,
Luben

Quoted 31 lines · Dan Carpenter

If amdgpu_eeprom_read() returns a negative error code then the error handling checks:

	if (res < buf_size) {

The problem is that "buf_size" is a u32 so negative values are type promoted to a high positive values and the condition is false. Fix this by changing the type of "buf_size" to int.

Fixes 79beb6114014 ("drm/amdgpu: Optimize EEPROM RAS table I/O")

---
It's hard for me to tell the exact upper bound that "buf_size" can be, but if it's over USHRT_MAX then we are well toasted.

 drivers/gpu/drm/amd/amdgpu/amdgpu_ras_eeprom.c | 3 +--
 1 file changed, 1 insertion(+), 2 deletions(-)
drivers/gpu/drm/amd/amdgpu/amdgpu_ras_eeprom.c +1 -2
index fc70620369e4..f07a456506ef 100644
@@ -978,9 +978,8 @@ const struct file_operations amdgpu_ras_debugfs_eeprom_table_ops = {
 static int __verify_ras_table_checksum(struct amdgpu_ras_eeprom_control *control)
 {
 	struct amdgpu_device *adev = to_amdgpu_device(control);
-	int res;
+	int buf_size, res;
 	u8  csum, *buf, *pp;
-	u32 buf_size;
 
Quoted 2 lines

drivers/gpu/drm/amd/amdgpu/amdgpu_ras_eeprom.c const struct file_operations amdgpu_ras_debugfs_eeprom_table_ops = { · line 985

Quoted 2 lines
drivers/gpu/drm/amd/amdgpu/amdgpu_ras_eeprom.c +0 -0
 	buf_size = RAS_TABLE_HEADER_SIZE +
 		control->ras_num_recs * RAS_TABLE_RECORD_SIZE;

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