Skip to contentSkip to conversation

fix reversed logic in device_area_is_valid

PATCH In review Mikulas Patocka 2 participants 5 messages Jul 27, 2009 – Jul 30, 2009 Device mapper

lore
Share & tools
lore

Mikulas Patocka

author maintainer
#0 lore raw

PATCH 1/1 fix reversed logic in device_area_is_valid

Hi

This patch fixes a bug introduced in 2.6.31-rc1. Targets that define iterate_devices and don't contain any devices (such as dm-loop in fsio) don't work.

Mikulas

---

Fix reverse logic in device_area_is_valid

->iterate_devices method calls the callback for every underlying device in the target. If the callback returns non-zero, iterate_devices exits and returns this value. If the callback returns zero for all the devices, iterate_devices returns zero.

The logic to check for invalid device areas was reversed. device_area_is_valid returned 0 on error and 1 on success.

Thus:
- error was returned only if all the devices vere errorneous. If some of them returned 1 (success), dm_calculate_queue_limits understood this as success.
- if the target had no device (the example is dm-loop target in FSIO mode), iterate_devices returned error straight away and dm_calculate_queue_limits understood this as an error.

This patch reverses the logic so that 0 means success and 1 means error.

---
 drivers/md/dm-table.c |   16 ++++++++--------
 1 file changed, 8 insertions(+), 8 deletions(-)
drivers/md/dm-table.c +8 -8
Index: linux-2.6.31-rc3-devel/drivers/md/dm-table.c
===================================================================
@@ -357,16 +357,16 @@ static int device_area_is_valid(struct d
 	char b[BDEVNAME_SIZE];
 
 	if (!dev_size)
-		return 1;
+		return 0;
 
 	if ((start >= dev_size) || (start + ti->len > dev_size)) {
 		DMWARN("%s: %s too small for target",
 		       dm_device_name(ti->table->md), bdevname(bdev, b));
-		return 0;
+		return 1;
 	}
 
 	if (logical_block_size_sectors <= 1)
-		return 1;
+		return 0;
 
 	if (start & (logical_block_size_sectors - 1)) {
 		DMWARN("%s: start=%llu not aligned to h/w "
@@ -374,7 +374,7 @@ static int device_area_is_valid(struct d
 		       dm_device_name(ti->table->md),
 		       (unsigned long long)start,
 		       limits->logical_block_size, bdevname(bdev, b));
-		return 0;
+		return 1;
 	}
 
 	if (ti->len & (logical_block_size_sectors - 1)) {
@@ -383,10 +383,10 @@ static int device_area_is_valid(struct d
 		       dm_device_name(ti->table->md),
 		       (unsigned long long)ti->len,
 		       limits->logical_block_size, bdevname(bdev, b));
-		return 0;
+		return 1;
 	}
 
-	return 1;
+	return 0;
 }
 
 /*
@@ -1005,8 +1005,8 @@ int dm_calculate_queue_limits(struct dm_
 		 * Check each device area is consistent with the target's
 		 * overall queue limits.
 		 */
-		if (!ti->type->iterate_devices(ti, device_area_is_valid,
-					       &ti_limits))
+		if (ti->type->iterate_devices(ti, device_area_is_valid,
+					      &ti_limits))
 			return -EINVAL;
 
 combine_limits:
#1 lore raw
Quoted 27 lines · Mikulas Patocka

Hi

This patch fixes a bug introduced in 2.6.31-rc1. Targets that define iterate_devices and don't contain any devices (such as dm-loop in fsio) don't work.

Mikulas

---

Fix reverse logic in device_area_is_valid

->iterate_devices method calls the callback for every underlying device in the target. If the callback returns non-zero, iterate_devices exits and returns this value. If the callback returns zero for all the devices, iterate_devices returns zero.

The logic to check for invalid device areas was reversed. device_area_is_valid returned 0 on error and 1 on success.

Thus:
- error was returned only if all the devices vere errorneous. If some of them returned 1 (success), dm_calculate_queue_limits understood this as success.
- if the target had no device (the example is dm-loop target in FSIO mode), iterate_devices returned error straight away and dm_calculate_queue_limits understood this as an error.

Quoted 3 lines

This patch reverses the logic so that 0 means success and 1 means error.

Nice catch, interesting that this was somehow missed. All targets do treat a non-zero return from .iterate_devices as an error.

We should probably rename device_area_is_valid to device_area_invalid.

Mike

author maintainer
#2 lore raw
Quoted 32 lines · Mike Snitzer
Quoted 29 lines · Mikulas Patocka

Hi

This patch fixes a bug introduced in 2.6.31-rc1. Targets that define iterate_devices and don't contain any devices (such as dm-loop in fsio) don't work.

Mikulas

---

Fix reverse logic in device_area_is_valid

->iterate_devices method calls the callback for every underlying device in the target. If the callback returns non-zero, iterate_devices exits and returns this value. If the callback returns zero for all the devices, iterate_devices returns zero.

The logic to check for invalid device areas was reversed. device_area_is_valid returned 0 on error and 1 on success.

Thus:
- error was returned only if all the devices vere errorneous. If some of them returned 1 (success), dm_calculate_queue_limits understood this as success.
- if the target had no device (the example is dm-loop target in FSIO mode), iterate_devices returned error straight away and dm_calculate_queue_limits understood this as an error.

This patch reverses the logic so that 0 means success and 1 means error.

Quoted 8 lines
Quoted 1 line

Nice catch, interesting that this was somehow missed. All targets do treat a non-zero return from .iterate_devices as an error.

We should probably rename device_area_is_valid to device_area_invalid.

Mike

Yes, you can rename it to that.

Mikulas

#3 lore raw

PATCH 1/1 Rename device_area_is_valid and reverse logic accordingly

Quoted 39 lines · Mikulas Patocka
Quoted 37 lines · Mike Snitzer
Quoted 27 lines · Mikulas Patocka

Hi

This patch fixes a bug introduced in 2.6.31-rc1. Targets that define iterate_devices and don't contain any devices (such as dm-loop in fsio) don't work.

Mikulas

---

Fix reverse logic in device_area_is_valid

->iterate_devices method calls the callback for every underlying device in the target. If the callback returns non-zero, iterate_devices exits and returns this value. If the callback returns zero for all the devices, iterate_devices returns zero.

The logic to check for invalid device areas was reversed. device_area_is_valid returned 0 on error and 1 on success.

Thus:
- error was returned only if all the devices vere errorneous. If some of them returned 1 (success), dm_calculate_queue_limits understood this as success.
- if the target had no device (the example is dm-loop target in FSIO mode), iterate_devices returned error straight away and dm_calculate_queue_limits understood this as an error.

Quoted 3 lines

This patch reverses the logic so that 0 means success and 1 means error.

Nice catch, interesting that this was somehow missed. All targets do treat a non-zero return from .iterate_devices as an error.

Quoted 7 lines
Quoted 3 lines

We should probably rename device_area_is_valid to device_area_invalid.

Mike

Yes, you can rename it to that.

Mikulas

---

From: Mikulas Patocka <[email protected]>

Rename device_area_is_valid and reverse logic accordingly

->iterate_devices method calls the callback for every underlying device in the target. If the callback returns non-zero iterate_devices exits immediately and the caller should understand that an error occurred.

The logic to check for valid device areas was reversed relative to proper use with iterate_devices. device_area_is_valid returned 0 on error and 1 on success. Thus error (from iterate_devices) was returned only if all the devices were invalid.

This patch renames device_area_is_valid to device_area_is_invalid and reverses the logic so that 0 means valid and 1 means invalid (to match iterate_devices).

---
 drivers/md/dm-table.c |   22 +++++++++++-----------
 1 files changed, 11 insertions(+), 11 deletions(-)
drivers/md/dm-table.c +11 -11
index d952b34..aa60526 100644
@@ -343,10 +343,10 @@ static void close_dev(struct dm_dev_internal *d, struct mapped_device *md)
 }
 
 /*
- * If possible, this checks an area of a destination device is valid.
+ * If possible, this checks an area of a destination device is invalid.
  */
-static int device_area_is_valid(struct dm_target *ti, struct dm_dev *dev,
-				sector_t start, sector_t len, void *data)
+static int device_area_is_invalid(struct dm_target *ti, struct dm_dev *dev,
+				  sector_t start, sector_t len, void *data)
 {
 	struct queue_limits *limits = data;
 	struct block_device *bdev = dev->bdev;
@@ -357,16 +357,16 @@ static int device_area_is_valid(struct dm_target *ti, struct dm_dev *dev,
 	char b[BDEVNAME_SIZE];
 
 	if (!dev_size)
-		return 1;
+		return 0;
 
 	if ((start >= dev_size) || (start + len > dev_size)) {
 		DMWARN("%s: %s too small for target",
 		       dm_device_name(ti->table->md), bdevname(bdev, b));
-		return 0;
+		return 1;
 	}
 
 	if (logical_block_size_sectors <= 1)
-		return 1;
+		return 0;
 
 	if (start & (logical_block_size_sectors - 1)) {
 		DMWARN("%s: start=%llu not aligned to h/w "
@@ -374,7 +374,7 @@ static int device_area_is_valid(struct dm_target *ti, struct dm_dev *dev,
 		       dm_device_name(ti->table->md),
 		       (unsigned long long)start,
 		       limits->logical_block_size, bdevname(bdev, b));
-		return 0;
+		return 1;
 	}
 
 	if (len & (logical_block_size_sectors - 1)) {
@@ -383,10 +383,10 @@ static int device_area_is_valid(struct dm_target *ti, struct dm_dev *dev,
 		       dm_device_name(ti->table->md),
 		       (unsigned long long)len,
 		       limits->logical_block_size, bdevname(bdev, b));
-		return 0;
+		return 1;
 	}
 
-	return 1;
+	return 0;
 }
 
 /*
@@ -1000,8 +1000,8 @@ int dm_calculate_queue_limits(struct dm_table *table,
 		 * Check each device area is consistent with the target's
 		 * overall queue limits.
 		 */
-		if (!ti->type->iterate_devices(ti, device_area_is_valid,
-					       &ti_limits))
+		if (ti->type->iterate_devices(ti, device_area_is_invalid,
+					      &ti_limits))
 			return -EINVAL;
 
 combine_limits:
author maintainer Acked-by on v1 1/1
#4 lore raw

Re: [PATCH] Rename device_area_is_valid and reverse logic accordingly

Quoted 88 lines

---

From: Mikulas Patocka <[email protected]>

Rename device_area_is_valid and reverse logic accordingly

->iterate_devices method calls the callback for every underlying device in the target. If the callback returns non-zero iterate_devices exits immediately and the caller should understand that an error occurred.

The logic to check for valid device areas was reversed relative to proper use with iterate_devices. device_area_is_valid returned 0 on error and 1 on success. Thus error (from iterate_devices) was returned only if all the devices were invalid.

This patch renames device_area_is_valid to device_area_is_invalid and reverses the logic so that 0 means valid and 1 means invalid (to match iterate_devices).

---
 drivers/md/dm-table.c |   22 +++++++++++-----------
 1 files changed, 11 insertions(+), 11 deletions(-)
drivers/md/dm-table.c +9 -9
index d952b34..aa60526 100644
@@ -343,10 +343,10 @@ static void close_dev(struct dm_dev_internal *d, struct mapped_device *md)
 }
 
 /*
- * If possible, this checks an area of a destination device is valid.
+ * If possible, this checks an area of a destination device is invalid.
  */
-static int device_area_is_valid(struct dm_target *ti, struct dm_dev *dev,
-				sector_t start, sector_t len, void *data)
+static int device_area_is_invalid(struct dm_target *ti, struct dm_dev *dev,
+				  sector_t start, sector_t len, void *data)
 {
 	struct queue_limits *limits = data;
 	struct block_device *bdev = dev->bdev;
@@ -357,16 +357,16 @@ static int device_area_is_valid(struct dm_target *ti, struct dm_dev *dev,
 	char b[BDEVNAME_SIZE];
 
 	if (!dev_size)
-		return 1;
+		return 0;
 
 	if ((start >= dev_size) || (start + len > dev_size)) {
 		DMWARN("%s: %s too small for target",
 		       dm_device_name(ti->table->md), bdevname(bdev, b));
-		return 0;
+		return 1;
 	}
 
 	if (logical_block_size_sectors <= 1)
-		return 1;
+		return 0;
 
 	if (start & (logical_block_size_sectors - 1)) {
 		DMWARN("%s: start=%llu not aligned to h/w "
@@ -374,7 +374,7 @@ static int device_area_is_valid(struct dm_target *ti, struct dm_dev *dev,
 		       dm_device_name(ti->table->md),
 		       (unsigned long long)start,
 		       limits->logical_block_size, bdevname(bdev, b));
-		return 0;
+		return 1;
 	}
 
 	if (len & (logical_block_size_sectors - 1)) {
@@ -383,10 +383,10 @@ static int device_area_is_valid(struct dm_target *ti, struct dm_dev *dev,
 		       dm_device_name(ti->table->md),
 		       (unsigned long long)len,
 		       limits->logical_block_size, bdevname(bdev, b));
-		return 0;
+		return 1;
 	}
 
-	return 1;
+	return 0;
 }
 
 /*

drivers/md/dm-table.c int dm_calculate_queue_limits(struct dm_table *table, · line 1007

Quoted 10 lines
drivers/md/dm-table.c +2 -2
@@ -1000,8 +1000,8 @@ int dm_calculate_queue_limits(struct dm_table *table,
 		 * Check each device area is consistent with the target's
 		 * overall queue limits.
 		 */
-		if (!ti->type->iterate_devices(ti, device_area_is_valid,
-					       &ti_limits))
+		if (ti->type->iterate_devices(ti, device_area_is_invalid,
+					      &ti_limits))
 			return -EINVAL;
 
 combine_limits: