Readahead clenups

PATCH In review David Sterba 3 participants 16 messages Mar 15, 2017 – Mar 31, 2017 Btrfs file system

lore

Share & tools

lore

David Sterba

author maintainer

2017-03-15 16:03 UTC #0 lore raw

cover Readahead clenups

I spotted some GFP_NOFS uses in readahead and converted them to GFP_KERNEL with a few cleanups along the way.

David Sterba (7):
btrfs: preallocate radix tree node for readahead
btrfs: use simpler readahead zone lookups
btrfs: preallocate radix tree node for global readahead tree
btrfs: remove redundant parameter from btree_readahead_hook
btrfs: remove redundant parameter from reada_find_zone
btrfs: remove redundant parameter from reada_start_machine_dev
btrfs: remove local blocksize variable in reada_find_extent

 fs/btrfs/ctree.h   |  3 +-
 fs/btrfs/disk-io.c |  6 ++--
 fs/btrfs/reada.c   | 89 ++++++++++++++++++++++++++++--------------------------
 fs/btrfs/volumes.c |  2 +-
 4 files changed, 51 insertions(+), 49 deletions(-)

Signature · 1 line

2.12.0

David Sterba

author maintainer

2017-03-15 16:03 UTC #1 lore raw

PATCH 1/7 btrfs: preallocate radix tree node for readahead

We can preallocate the node so insertion does not have to do that under the lock. The GFP flags for the per-device radix tree are initialized to
GFP_NOFS & ~__GFP_DIRECT_RECLAIM
but we can use GFP_KERNEL, same as an allocation above anyway, but also because readahead is optional and not on any critical writeout path.

Signed-off-by David Sterba <[email protected]>

---
 fs/btrfs/reada.c   | 7 +++++++
 fs/btrfs/volumes.c | 2 +-
 2 files changed, 8 insertions(+), 1 deletion(-)

fs/btrfs/reada.c +7 -0

index e88bca87f5d2..fdae8ca79401 100644
@@ -270,6 +270,12 @@ static struct reada_zone *reada_find_zone(struct btrfs_fs_info *fs_info,
 	if (!zone)
 		return NULL;
 
+	ret = radix_tree_preload(GFP_KERNEL);
+	if (ret) {
+		kfree(zone);
+		return NULL;
+	}
+
 	zone->start = start;
 	zone->end = end;
 	INIT_LIST_HEAD(&zone->list);

@@ -299,6 +305,7 @@ static struct reada_zone *reada_find_zone(struct btrfs_fs_info *fs_info,
 			zone = NULL;
 	}
 	spin_unlock(&fs_info->reada_lock);
+	radix_tree_preload_end();
 
 	return zone;
 }

fs/btrfs/volumes.c +1 -1

index 73d56eef5e60..f158b8657ae3 100644
@@ -247,7 +247,7 @@ static struct btrfs_device *__alloc_device(void)
 	atomic_set(&dev->reada_in_flight, 0);
 	atomic_set(&dev->dev_stats_ccnt, 0);
 	btrfs_device_data_ordered_init(dev);
-	INIT_RADIX_TREE(&dev->reada_zones, GFP_NOFS & ~__GFP_DIRECT_RECLAIM);
+	INIT_RADIX_TREE(&dev->reada_zones, GFP_KERNEL);
 	INIT_RADIX_TREE(&dev->reada_extents, GFP_NOFS & ~__GFP_DIRECT_RECLAIM);
 
 	return dev;

Signature · 1 line

2.12.0

Liu Bo

Reviewed-by on v1 1/7

2017-03-20 20:41 UTC #2 lore raw

Re: [PATCH 1/7] btrfs: preallocate radix tree node for readahead

Quoted 6 lines · David Sterba

We can preallocate the node so insertion does not have to do that under the lock. The GFP flags for the per-device radix tree are initialized to
GFP_NOFS & ~__GFP_DIRECT_RECLAIM
but we can use GFP_KERNEL, same as an allocation above anyway, but also because readahead is optional and not on any critical writeout path.

Reviewed-by Liu Bo <[email protected]>

Thanks,

-liubo

Quoted 51 lines

Signed-off-by David Sterba <[email protected]>

---
 fs/btrfs/reada.c   | 7 +++++++
 fs/btrfs/volumes.c | 2 +-
 2 files changed, 8 insertions(+), 1 deletion(-)

fs/btrfs/reada.c +7 -0

index e88bca87f5d2..fdae8ca79401 100644
@@ -270,6 +270,12 @@ static struct reada_zone *reada_find_zone(struct btrfs_fs_info *fs_info,
 	if (!zone)
 		return NULL;
 
+	ret = radix_tree_preload(GFP_KERNEL);
+	if (ret) {
+		kfree(zone);
+		return NULL;
+	}
+
 	zone->start = start;
 	zone->end = end;
 	INIT_LIST_HEAD(&zone->list);

@@ -299,6 +305,7 @@ static struct reada_zone *reada_find_zone(struct btrfs_fs_info *fs_info,
 			zone = NULL;
 	}
 	spin_unlock(&fs_info->reada_lock);
+	radix_tree_preload_end();
 
 	return zone;
 }

fs/btrfs/volumes.c +1 -1

index 73d56eef5e60..f158b8657ae3 100644
@@ -247,7 +247,7 @@ static struct btrfs_device *__alloc_device(void)
 	atomic_set(&dev->reada_in_flight, 0);
 	atomic_set(&dev->dev_stats_ccnt, 0);
 	btrfs_device_data_ordered_init(dev);
-	INIT_RADIX_TREE(&dev->reada_zones, GFP_NOFS & ~__GFP_DIRECT_RECLAIM);
+	INIT_RADIX_TREE(&dev->reada_zones, GFP_KERNEL);
 	INIT_RADIX_TREE(&dev->reada_extents, GFP_NOFS & ~__GFP_DIRECT_RECLAIM);
 
 	return dev;

Signature · 6 lines

2.12.0

--
To unsubscribe from this list: send the line "unsubscribe linux-btrfs" in
the body of a message to [email protected]
More majordomo info at  http://vger.kernel.org/majordomo-info.html

David Sterba

author maintainer

2017-03-15 16:03 UTC #3 lore raw

PATCH 2/7 btrfs: use simpler readahead zone lookups

In reply to David Sterba

No point using radix_tree_gang_lookup if we're looking up just one slot.

Signed-off-by David Sterba <[email protected]>

---
 fs/btrfs/reada.c | 52 ++++++++++++++++++++++------------------------------
 1 file changed, 22 insertions(+), 30 deletions(-)

fs/btrfs/reada.c +22 ~~-30~~

index fdae8ca79401..dd78af5d265d 100644
@@ -246,11 +246,9 @@ static struct reada_zone *reada_find_zone(struct btrfs_fs_info *fs_info,
 	u64 end;
 	int i;
 
-	zone = NULL;
 	spin_lock(&fs_info->reada_lock);
-	ret = radix_tree_gang_lookup(&dev->reada_zones, (void **)&zone,
-				     logical >> PAGE_SHIFT, 1);
-	if (ret == 1 && logical >= zone->start && logical <= zone->end) {
+	zone = radix_tree_lookup(&dev->reada_zones, logical >> PAGE_SHIFT);
+	if (zone && logical >= zone->start && logical <= zone->end) {
 		kref_get(&zone->refcnt);
 		spin_unlock(&fs_info->reada_lock);
 		return zone;

@@ -297,9 +295,9 @@ static struct reada_zone *reada_find_zone(struct btrfs_fs_info *fs_info,
 
 	if (ret == -EEXIST) {
 		kfree(zone);
-		ret = radix_tree_gang_lookup(&dev->reada_zones, (void **)&zone,
-					     logical >> PAGE_SHIFT, 1);
-		if (ret == 1 && logical >= zone->start && logical <= zone->end)
+		zone = radix_tree_lookup(&dev->reada_zones,
+				logical >> PAGE_SHIFT);
+		if (zone && logical >= zone->start && logical <= zone->end)
 			kref_get(&zone->refcnt);
 		else
 			zone = NULL;

@@ -604,7 +602,6 @@ static int reada_pick_zone(struct btrfs_device *dev)
 	u64 top_elems = 0;
 	u64 top_locked_elems = 0;
 	unsigned long index = 0;
-	int ret;
 
 	if (dev->reada_curr_zone) {
 		reada_peer_zones_set_lock(dev->reada_curr_zone, 0);

@@ -615,9 +612,8 @@ static int reada_pick_zone(struct btrfs_device *dev)
 	while (1) {
 		struct reada_zone *zone;
 
-		ret = radix_tree_gang_lookup(&dev->reada_zones,
-					     (void **)&zone, index, 1);
-		if (ret == 0)
+		zone = radix_tree_lookup(&dev->reada_zones, index);
+		if (!zone)
 			break;
 		index = (zone->end >> PAGE_SHIFT) + 1;
 		if (zone->locked) {

@@ -669,19 +665,18 @@ static int reada_start_machine_dev(struct btrfs_fs_info *fs_info,
 	 * a contiguous block of extents, we could also coagulate them or use
 	 * plugging to speed things up
 	 */
-	ret = radix_tree_gang_lookup(&dev->reada_extents, (void **)&re,
-				     dev->reada_next >> PAGE_SHIFT, 1);
-	if (ret == 0 || re->logical > dev->reada_curr_zone->end) {
+	re = radix_tree_lookup(&dev->reada_extents,
+			dev->reada_next >> PAGE_SHIFT);
+	if (!re || re->logical > dev->reada_curr_zone->end) {
 		ret = reada_pick_zone(dev);
 		if (!ret) {
 			spin_unlock(&fs_info->reada_lock);
 			return 0;
 		}
-		re = NULL;
-		ret = radix_tree_gang_lookup(&dev->reada_extents, (void **)&re,
-					dev->reada_next >> PAGE_SHIFT, 1);
+		re = radix_tree_lookup(&dev->reada_extents,
+				dev->reada_next >> PAGE_SHIFT);
 	}
-	if (ret == 0) {
+	if (!re) {
 		spin_unlock(&fs_info->reada_lock);
 		return 0;
 	}

@@ -809,7 +804,6 @@ static void dump_devs(struct btrfs_fs_info *fs_info, int all)
 	struct btrfs_device *device;
 	struct btrfs_fs_devices *fs_devices = fs_info->fs_devices;
 	unsigned long index;
-	int ret;
 	int i;
 	int j;
 	int cnt;

@@ -821,9 +815,9 @@ static void dump_devs(struct btrfs_fs_info *fs_info, int all)
 		index = 0;
 		while (1) {
 			struct reada_zone *zone;
-			ret = radix_tree_gang_lookup(&device->reada_zones,
-						     (void **)&zone, index, 1);
-			if (ret == 0)
+
+			zone = radix_tree_lookup(&device->reada_zones, index);
+			if (!zone)
 				break;
 			pr_debug("  zone %llu-%llu elems %llu locked %d devs",
 				    zone->start, zone->end, zone->elems,

@@ -841,11 +835,10 @@ static void dump_devs(struct btrfs_fs_info *fs_info, int all)
 		cnt = 0;
 		index = 0;
 		while (all) {
-			struct reada_extent *re = NULL;
+			struct reada_extent *re;
 
-			ret = radix_tree_gang_lookup(&device->reada_extents,
-						     (void **)&re, index, 1);
-			if (ret == 0)
+			re = radix_tree_lookup(&device->reada_extents, index);
+			if (!re)
 				break;
 			pr_debug("  re: logical %llu size %u empty %d scheduled %d",
 				re->logical, fs_info->nodesize,

@@ -870,11 +863,10 @@ static void dump_devs(struct btrfs_fs_info *fs_info, int all)
 	index = 0;
 	cnt = 0;
 	while (all) {
-		struct reada_extent *re = NULL;
+		struct reada_extent *re;
 
-		ret = radix_tree_gang_lookup(&fs_info->reada_tree, (void **)&re,
-					     index, 1);
-		if (ret == 0)
+		re = radix_tree_lookup(&fs_info->reada_tree, index);
+		if (!re)
 			break;
 		if (!re->scheduled) {
 			index = (re->logical >> PAGE_SHIFT) + 1;

Signature · 1 line

2.12.0

Liu Bo

Reviewed-by on v1 2/7

2017-03-20 20:45 UTC #4 lore raw

Re: [PATCH 2/7] btrfs: use simpler readahead zone lookups

Quoted 2 lines · David Sterba

No point using radix_tree_gang_lookup if we're looking up just one slot.

Reviewed-by Liu Bo <[email protected]>

Thanks,

-liubo

Quoted 140 lines

Signed-off-by David Sterba <[email protected]>

---
 fs/btrfs/reada.c | 52 ++++++++++++++++++++++------------------------------
 1 file changed, 22 insertions(+), 30 deletions(-)

fs/btrfs/reada.c +22 ~~-30~~

index fdae8ca79401..dd78af5d265d 100644
@@ -246,11 +246,9 @@ static struct reada_zone *reada_find_zone(struct btrfs_fs_info *fs_info,
 	u64 end;
 	int i;
 
-	zone = NULL;
 	spin_lock(&fs_info->reada_lock);
-	ret = radix_tree_gang_lookup(&dev->reada_zones, (void **)&zone,
-				     logical >> PAGE_SHIFT, 1);
-	if (ret == 1 && logical >= zone->start && logical <= zone->end) {
+	zone = radix_tree_lookup(&dev->reada_zones, logical >> PAGE_SHIFT);
+	if (zone && logical >= zone->start && logical <= zone->end) {
 		kref_get(&zone->refcnt);
 		spin_unlock(&fs_info->reada_lock);
 		return zone;

@@ -297,9 +295,9 @@ static struct reada_zone *reada_find_zone(struct btrfs_fs_info *fs_info,
 
 	if (ret == -EEXIST) {
 		kfree(zone);
-		ret = radix_tree_gang_lookup(&dev->reada_zones, (void **)&zone,
-					     logical >> PAGE_SHIFT, 1);
-		if (ret == 1 && logical >= zone->start && logical <= zone->end)
+		zone = radix_tree_lookup(&dev->reada_zones,
+				logical >> PAGE_SHIFT);
+		if (zone && logical >= zone->start && logical <= zone->end)
 			kref_get(&zone->refcnt);
 		else
 			zone = NULL;

@@ -604,7 +602,6 @@ static int reada_pick_zone(struct btrfs_device *dev)
 	u64 top_elems = 0;
 	u64 top_locked_elems = 0;
 	unsigned long index = 0;
-	int ret;
 
 	if (dev->reada_curr_zone) {
 		reada_peer_zones_set_lock(dev->reada_curr_zone, 0);

@@ -615,9 +612,8 @@ static int reada_pick_zone(struct btrfs_device *dev)
 	while (1) {
 		struct reada_zone *zone;
 
-		ret = radix_tree_gang_lookup(&dev->reada_zones,
-					     (void **)&zone, index, 1);
-		if (ret == 0)
+		zone = radix_tree_lookup(&dev->reada_zones, index);
+		if (!zone)
 			break;
 		index = (zone->end >> PAGE_SHIFT) + 1;
 		if (zone->locked) {

@@ -669,19 +665,18 @@ static int reada_start_machine_dev(struct btrfs_fs_info *fs_info,
 	 * a contiguous block of extents, we could also coagulate them or use
 	 * plugging to speed things up
 	 */
-	ret = radix_tree_gang_lookup(&dev->reada_extents, (void **)&re,
-				     dev->reada_next >> PAGE_SHIFT, 1);
-	if (ret == 0 || re->logical > dev->reada_curr_zone->end) {
+	re = radix_tree_lookup(&dev->reada_extents,
+			dev->reada_next >> PAGE_SHIFT);
+	if (!re || re->logical > dev->reada_curr_zone->end) {
 		ret = reada_pick_zone(dev);
 		if (!ret) {
 			spin_unlock(&fs_info->reada_lock);
 			return 0;
 		}
-		re = NULL;
-		ret = radix_tree_gang_lookup(&dev->reada_extents, (void **)&re,
-					dev->reada_next >> PAGE_SHIFT, 1);
+		re = radix_tree_lookup(&dev->reada_extents,
+				dev->reada_next >> PAGE_SHIFT);
 	}
-	if (ret == 0) {
+	if (!re) {
 		spin_unlock(&fs_info->reada_lock);
 		return 0;
 	}

@@ -809,7 +804,6 @@ static void dump_devs(struct btrfs_fs_info *fs_info, int all)
 	struct btrfs_device *device;
 	struct btrfs_fs_devices *fs_devices = fs_info->fs_devices;
 	unsigned long index;
-	int ret;
 	int i;
 	int j;
 	int cnt;

@@ -821,9 +815,9 @@ static void dump_devs(struct btrfs_fs_info *fs_info, int all)
 		index = 0;
 		while (1) {
 			struct reada_zone *zone;
-			ret = radix_tree_gang_lookup(&device->reada_zones,
-						     (void **)&zone, index, 1);
-			if (ret == 0)
+
+			zone = radix_tree_lookup(&device->reada_zones, index);
+			if (!zone)
 				break;
 			pr_debug("  zone %llu-%llu elems %llu locked %d devs",
 				    zone->start, zone->end, zone->elems,

@@ -841,11 +835,10 @@ static void dump_devs(struct btrfs_fs_info *fs_info, int all)
 		cnt = 0;
 		index = 0;
 		while (all) {
-			struct reada_extent *re = NULL;
+			struct reada_extent *re;
 
-			ret = radix_tree_gang_lookup(&device->reada_extents,
-						     (void **)&re, index, 1);
-			if (ret == 0)
+			re = radix_tree_lookup(&device->reada_extents, index);
+			if (!re)
 				break;
 			pr_debug("  re: logical %llu size %u empty %d scheduled %d",
 				re->logical, fs_info->nodesize,

@@ -870,11 +863,10 @@ static void dump_devs(struct btrfs_fs_info *fs_info, int all)
 	index = 0;
 	cnt = 0;
 	while (all) {
-		struct reada_extent *re = NULL;
+		struct reada_extent *re;
 
-		ret = radix_tree_gang_lookup(&fs_info->reada_tree, (void **)&re,
-					     index, 1);
-		if (ret == 0)
+		re = radix_tree_lookup(&fs_info->reada_tree, index);
+		if (!re)
 			break;
 		if (!re->scheduled) {
 			index = (re->logical >> PAGE_SHIFT) + 1;

Signature · 6 lines

2.12.0

--
To unsubscribe from this list: send the line "unsubscribe linux-btrfs" in
the body of a message to [email protected]
More majordomo info at  http://vger.kernel.org/majordomo-info.html

David Sterba

2017-03-31 17:34 UTC #5 lore raw

Re: [PATCH 2/7] btrfs: use simpler readahead zone lookups

In reply to David Sterba

Quoted 3 lines · David Sterba

No point using radix_tree_gang_lookup if we're looking up just one slot.
Signed-off-by David Sterba <[email protected]>

I've bisected to this patch, causes a hang in btrfs/011. I'll revert it for until I find out the cause.

David Sterba

author maintainer

2017-03-15 16:03 UTC #6 lore raw

PATCH 3/7 btrfs: preallocate radix tree node for global readahead tree

In reply to David Sterba

We can preallocate the node so insertion does not have to do that under the lock. The GFP flags for the global radix tree are initialized to
GFP_NOFS & ~__GFP_DIRECT_RECLAIM
but we can use GFP_KERNEL, because readahead is optional and not on any critical writeout path.

Signed-off-by David Sterba <[email protected]>

---
 fs/btrfs/disk-io.c | 2 +-
 fs/btrfs/reada.c   | 7 +++++++
 2 files changed, 8 insertions(+), 1 deletion(-)

fs/btrfs/disk-io.c +1 -1

index 08b74daf35d0..1d4c30327247 100644
@@ -2693,7 +2693,7 @@ int open_ctree(struct super_block *sb,
 	fs_info->commit_interval = BTRFS_DEFAULT_COMMIT_INTERVAL;
 	fs_info->avg_delayed_ref_runtime = NSEC_PER_SEC >> 6; /* div by 64 */
 	/* readahead state */
-	INIT_RADIX_TREE(&fs_info->reada_tree, GFP_NOFS & ~__GFP_DIRECT_RECLAIM);
+	INIT_RADIX_TREE(&fs_info->reada_tree, GFP_KERNEL);
 	spin_lock_init(&fs_info->reada_lock);
 
 	fs_info->thread_pool_size = min_t(unsigned long,

fs/btrfs/reada.c +7 -0

index dd78af5d265d..4c5a9b241cab 100644
@@ -391,6 +391,10 @@ static struct reada_extent *reada_find_extent(struct btrfs_fs_info *fs_info,
 		goto error;
 	}
 
+	ret = radix_tree_preload(GFP_KERNEL);
+	if (ret)
+		goto error;
+
 	/* insert extent in reada_tree + all per-device trees, all or nothing */
 	btrfs_dev_replace_lock(&fs_info->dev_replace, 0);
 	spin_lock(&fs_info->reada_lock);

@@ -400,13 +404,16 @@ static struct reada_extent *reada_find_extent(struct btrfs_fs_info *fs_info,
 		re_exist->refcnt++;
 		spin_unlock(&fs_info->reada_lock);
 		btrfs_dev_replace_unlock(&fs_info->dev_replace, 0);
+		radix_tree_preload_end();
 		goto error;
 	}
 	if (ret) {
 		spin_unlock(&fs_info->reada_lock);
 		btrfs_dev_replace_unlock(&fs_info->dev_replace, 0);
+		radix_tree_preload_end();
 		goto error;
 	}
+	radix_tree_preload_end();
 	prev_dev = NULL;
 	dev_replace_is_ongoing = btrfs_dev_replace_is_ongoing(
 			&fs_info->dev_replace);

Signature · 1 line

2.12.0

Liu Bo

Reviewed-by on v1 3/7

2017-03-20 21:00 UTC #7 lore raw

Re: [PATCH 3/7] btrfs: preallocate radix tree node for global readahead tree

Quoted 5 lines · David Sterba

We can preallocate the node so insertion does not have to do that under the lock. The GFP flags for the global radix tree are initialized to
GFP_NOFS & ~__GFP_DIRECT_RECLAIM
but we can use GFP_KERNEL, because readahead is optional and not on any critical writeout path.

Reviewed-by Liu Bo <[email protected]>

Thanks,

-liubo

Quoted 59 lines

Signed-off-by David Sterba <[email protected]>

---
 fs/btrfs/disk-io.c | 2 +-
 fs/btrfs/reada.c   | 7 +++++++
 2 files changed, 8 insertions(+), 1 deletion(-)

fs/btrfs/disk-io.c +1 -1

index 08b74daf35d0..1d4c30327247 100644
@@ -2693,7 +2693,7 @@ int open_ctree(struct super_block *sb,
 	fs_info->commit_interval = BTRFS_DEFAULT_COMMIT_INTERVAL;
 	fs_info->avg_delayed_ref_runtime = NSEC_PER_SEC >> 6; /* div by 64 */
 	/* readahead state */
-	INIT_RADIX_TREE(&fs_info->reada_tree, GFP_NOFS & ~__GFP_DIRECT_RECLAIM);
+	INIT_RADIX_TREE(&fs_info->reada_tree, GFP_KERNEL);
 	spin_lock_init(&fs_info->reada_lock);
 
 	fs_info->thread_pool_size = min_t(unsigned long,

fs/btrfs/reada.c +7 -0

index dd78af5d265d..4c5a9b241cab 100644
@@ -391,6 +391,10 @@ static struct reada_extent *reada_find_extent(struct btrfs_fs_info *fs_info,
 		goto error;
 	}
 
+	ret = radix_tree_preload(GFP_KERNEL);
+	if (ret)
+		goto error;
+
 	/* insert extent in reada_tree + all per-device trees, all or nothing */
 	btrfs_dev_replace_lock(&fs_info->dev_replace, 0);
 	spin_lock(&fs_info->reada_lock);

@@ -400,13 +404,16 @@ static struct reada_extent *reada_find_extent(struct btrfs_fs_info *fs_info,
 		re_exist->refcnt++;
 		spin_unlock(&fs_info->reada_lock);
 		btrfs_dev_replace_unlock(&fs_info->dev_replace, 0);
+		radix_tree_preload_end();
 		goto error;
 	}
 	if (ret) {
 		spin_unlock(&fs_info->reada_lock);
 		btrfs_dev_replace_unlock(&fs_info->dev_replace, 0);
+		radix_tree_preload_end();
 		goto error;
 	}
+	radix_tree_preload_end();
 	prev_dev = NULL;
 	dev_replace_is_ongoing = btrfs_dev_replace_is_ongoing(
 			&fs_info->dev_replace);

Signature · 6 lines

2.12.0

--
To unsubscribe from this list: send the line "unsubscribe linux-btrfs" in
the body of a message to [email protected]
More majordomo info at  http://vger.kernel.org/majordomo-info.html

David Sterba

author maintainer

2017-03-15 16:03 UTC #8 lore raw

PATCH 4/7 btrfs: remove redundant parameter from btree_readahead_hook

In reply to David Sterba

We can read fs_info from eb.

Signed-off-by David Sterba <[email protected]>

---
 fs/btrfs/ctree.h   | 3 +--
 fs/btrfs/disk-io.c | 4 ++--
 fs/btrfs/reada.c   | 4 ++--
 3 files changed, 5 insertions(+), 6 deletions(-)

fs/btrfs/ctree.h +1 -2

index 29b7fc28c607..173fac68323a 100644
@@ -3671,8 +3671,7 @@ struct reada_control *btrfs_reada_add(struct btrfs_root *root,
 			      struct btrfs_key *start, struct btrfs_key *end);
 int btrfs_reada_wait(void *handle);
 void btrfs_reada_detach(void *handle);
-int btree_readahead_hook(struct btrfs_fs_info *fs_info,
-			 struct extent_buffer *eb, int err);
+int btree_readahead_hook(struct extent_buffer *eb, int err);
 
 static inline int is_fstree(u64 rootid)
 {

fs/btrfs/disk-io.c +2 -2

index 1d4c30327247..995b28179af9 100644
@@ -762,7 +762,7 @@ static int btree_readpage_end_io_hook(struct btrfs_io_bio *io_bio,
 err:
 	if (reads_done &&
 	    test_and_clear_bit(EXTENT_BUFFER_READAHEAD, &eb->bflags))
-		btree_readahead_hook(fs_info, eb, ret);
+		btree_readahead_hook(eb, ret);
 
 	if (ret) {
 		/*

@@ -787,7 +787,7 @@ static int btree_io_failed_hook(struct page *page, int failed_mirror)
 	eb->read_mirror = failed_mirror;
 	atomic_dec(&eb->io_pages);
 	if (test_and_clear_bit(EXTENT_BUFFER_READAHEAD, &eb->bflags))
-		btree_readahead_hook(eb->fs_info, eb, -EIO);
+		btree_readahead_hook(eb, -EIO);
 	return -EIO;	/* we fixed nothing */
 }

fs/btrfs/reada.c +2 -2

index 4c5a9b241cab..5edf7328f67d 100644
@@ -209,9 +209,9 @@ static void __readahead_hook(struct btrfs_fs_info *fs_info,
 	return;
 }
 
-int btree_readahead_hook(struct btrfs_fs_info *fs_info,
-			 struct extent_buffer *eb, int err)
+int btree_readahead_hook(struct extent_buffer *eb, int err)
 {
+	struct btrfs_fs_info *fs_info = eb->fs_info;
 	int ret = 0;
 	struct reada_extent *re;

Signature · 1 line

2.12.0

Liu Bo

Reviewed-by on v1 4/7

2017-03-20 21:00 UTC #9 lore raw

Re: [PATCH 4/7] btrfs: remove redundant parameter from btree_readahead_hook

Quoted 1 line · David Sterba

We can read fs_info from eb.

Reviewed-by Liu Bo <[email protected]>

Thanks,

-liubo

Quoted 67 lines

Signed-off-by David Sterba <[email protected]>

---
 fs/btrfs/ctree.h   | 3 +--
 fs/btrfs/disk-io.c | 4 ++--
 fs/btrfs/reada.c   | 4 ++--
 3 files changed, 5 insertions(+), 6 deletions(-)

fs/btrfs/ctree.h +1 -2

index 29b7fc28c607..173fac68323a 100644
@@ -3671,8 +3671,7 @@ struct reada_control *btrfs_reada_add(struct btrfs_root *root,
 			      struct btrfs_key *start, struct btrfs_key *end);
 int btrfs_reada_wait(void *handle);
 void btrfs_reada_detach(void *handle);
-int btree_readahead_hook(struct btrfs_fs_info *fs_info,
-			 struct extent_buffer *eb, int err);
+int btree_readahead_hook(struct extent_buffer *eb, int err);
 
 static inline int is_fstree(u64 rootid)
 {

fs/btrfs/disk-io.c +2 -2

index 1d4c30327247..995b28179af9 100644
@@ -762,7 +762,7 @@ static int btree_readpage_end_io_hook(struct btrfs_io_bio *io_bio,
 err:
 	if (reads_done &&
 	    test_and_clear_bit(EXTENT_BUFFER_READAHEAD, &eb->bflags))
-		btree_readahead_hook(fs_info, eb, ret);
+		btree_readahead_hook(eb, ret);
 
 	if (ret) {
 		/*

@@ -787,7 +787,7 @@ static int btree_io_failed_hook(struct page *page, int failed_mirror)
 	eb->read_mirror = failed_mirror;
 	atomic_dec(&eb->io_pages);
 	if (test_and_clear_bit(EXTENT_BUFFER_READAHEAD, &eb->bflags))
-		btree_readahead_hook(eb->fs_info, eb, -EIO);
+		btree_readahead_hook(eb, -EIO);
 	return -EIO;	/* we fixed nothing */
 }

fs/btrfs/reada.c +2 -2

index 4c5a9b241cab..5edf7328f67d 100644
@@ -209,9 +209,9 @@ static void __readahead_hook(struct btrfs_fs_info *fs_info,
 	return;
 }
 
-int btree_readahead_hook(struct btrfs_fs_info *fs_info,
-			 struct extent_buffer *eb, int err)
+int btree_readahead_hook(struct extent_buffer *eb, int err)
 {
+	struct btrfs_fs_info *fs_info = eb->fs_info;
 	int ret = 0;
 	struct reada_extent *re;

Signature · 6 lines

2.12.0

--
To unsubscribe from this list: send the line "unsubscribe linux-btrfs" in
the body of a message to [email protected]
More majordomo info at  http://vger.kernel.org/majordomo-info.html

David Sterba

author maintainer

2017-03-15 16:03 UTC #10 lore raw

PATCH 5/7 btrfs: remove redundant parameter from reada_find_zone

In reply to David Sterba

We can read fs_info from dev.

Signed-off-by David Sterba <[email protected]>

---
 fs/btrfs/reada.c | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

fs/btrfs/reada.c +3 -3

index 5edf7328f67d..c1fc79cd4b2a 100644
@@ -235,10 +235,10 @@ int btree_readahead_hook(struct extent_buffer *eb, int err)
 	return ret;
 }
 
-static struct reada_zone *reada_find_zone(struct btrfs_fs_info *fs_info,
-					  struct btrfs_device *dev, u64 logical,
+static struct reada_zone *reada_find_zone(struct btrfs_device *dev, u64 logical,
 					  struct btrfs_bio *bbio)
 {
+	struct btrfs_fs_info *fs_info = dev->fs_info;
 	int ret;
 	struct reada_zone *zone;
 	struct btrfs_block_group_cache *cache = NULL;

@@ -372,7 +372,7 @@ static struct reada_extent *reada_find_extent(struct btrfs_fs_info *fs_info,
 		 if (!dev->bdev)
 			continue;
 
-		zone = reada_find_zone(fs_info, dev, logical, bbio);
+		zone = reada_find_zone(dev, logical, bbio);
 		if (!zone)
 			continue;

Signature · 1 line

2.12.0

Liu Bo

Reviewed-by on v1 5/7

2017-03-20 21:10 UTC #11 lore raw

Re: [PATCH 5/7] btrfs: remove redundant parameter from reada_find_zone

Quoted 2 lines · David Sterba

We can read fs_info from dev.

Reviewed-by Liu Bo <[email protected]>

Thanks,

-liubo

Quoted 38 lines

Signed-off-by David Sterba <[email protected]>

---
 fs/btrfs/reada.c | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

fs/btrfs/reada.c +3 -3

index 5edf7328f67d..c1fc79cd4b2a 100644
@@ -235,10 +235,10 @@ int btree_readahead_hook(struct extent_buffer *eb, int err)
 	return ret;
 }
 
-static struct reada_zone *reada_find_zone(struct btrfs_fs_info *fs_info,
-					  struct btrfs_device *dev, u64 logical,
+static struct reada_zone *reada_find_zone(struct btrfs_device *dev, u64 logical,
 					  struct btrfs_bio *bbio)
 {
+	struct btrfs_fs_info *fs_info = dev->fs_info;
 	int ret;
 	struct reada_zone *zone;
 	struct btrfs_block_group_cache *cache = NULL;

@@ -372,7 +372,7 @@ static struct reada_extent *reada_find_extent(struct btrfs_fs_info *fs_info,
 		 if (!dev->bdev)
 			continue;
 
-		zone = reada_find_zone(fs_info, dev, logical, bbio);
+		zone = reada_find_zone(dev, logical, bbio);
 		if (!zone)
 			continue;

Signature · 6 lines

2.12.0

--
To unsubscribe from this list: send the line "unsubscribe linux-btrfs" in
the body of a message to [email protected]
More majordomo info at  http://vger.kernel.org/majordomo-info.html

David Sterba

author maintainer

2017-03-15 16:03 UTC #12 lore raw

PATCH 6/7 btrfs: remove redundant parameter from reada_start_machine_dev

In reply to David Sterba

We can read fs_info from dev.

Signed-off-by David Sterba <[email protected]>

---
 fs/btrfs/reada.c | 7 +++----
 1 file changed, 3 insertions(+), 4 deletions(-)

fs/btrfs/reada.c +3 -4

index c1fc79cd4b2a..91df381a60ce 100644
@@ -649,9 +649,9 @@ static int reada_pick_zone(struct btrfs_device *dev)
 	return 1;
 }
 
-static int reada_start_machine_dev(struct btrfs_fs_info *fs_info,
-				   struct btrfs_device *dev)
+static int reada_start_machine_dev(struct btrfs_device *dev)
 {
+	struct btrfs_fs_info *fs_info = dev->fs_info;
 	struct reada_extent *re = NULL;
 	int mirror_num = 0;
 	struct extent_buffer *eb = NULL;

@@ -763,8 +763,7 @@ static void __reada_start_machine(struct btrfs_fs_info *fs_info)
 		list_for_each_entry(device, &fs_devices->devices, dev_list) {
 			if (atomic_read(&device->reada_in_flight) <
 			    MAX_IN_FLIGHT)
-				enqueued += reada_start_machine_dev(fs_info,
-								    device);
+				enqueued += reada_start_machine_dev(device);
 		}
 		mutex_unlock(&fs_devices->device_list_mutex);
 		total += enqueued;

Signature · 1 line

2.12.0

Liu Bo

Reviewed-by on v1 6/7

2017-03-20 22:11 UTC #13 lore raw

Re: [PATCH 6/7] btrfs: remove redundant parameter from reada_start_machine_dev

Quoted 2 lines · David Sterba

We can read fs_info from dev.

Reviewed-by Liu Bo <[email protected]>

Thanks,

-liubo

Quoted 38 lines

Signed-off-by David Sterba <[email protected]>

---
 fs/btrfs/reada.c | 7 +++----
 1 file changed, 3 insertions(+), 4 deletions(-)

fs/btrfs/reada.c +3 -4

index c1fc79cd4b2a..91df381a60ce 100644
@@ -649,9 +649,9 @@ static int reada_pick_zone(struct btrfs_device *dev)
 	return 1;
 }
 
-static int reada_start_machine_dev(struct btrfs_fs_info *fs_info,
-				   struct btrfs_device *dev)
+static int reada_start_machine_dev(struct btrfs_device *dev)
 {
+	struct btrfs_fs_info *fs_info = dev->fs_info;
 	struct reada_extent *re = NULL;
 	int mirror_num = 0;
 	struct extent_buffer *eb = NULL;

@@ -763,8 +763,7 @@ static void __reada_start_machine(struct btrfs_fs_info *fs_info)
 		list_for_each_entry(device, &fs_devices->devices, dev_list) {
 			if (atomic_read(&device->reada_in_flight) <
 			    MAX_IN_FLIGHT)
-				enqueued += reada_start_machine_dev(fs_info,
-								    device);
+				enqueued += reada_start_machine_dev(device);
 		}
 		mutex_unlock(&fs_devices->device_list_mutex);
 		total += enqueued;

Signature · 6 lines

2.12.0

--
To unsubscribe from this list: send the line "unsubscribe linux-btrfs" in
the body of a message to [email protected]
More majordomo info at  http://vger.kernel.org/majordomo-info.html

David Sterba

author maintainer

2017-03-15 16:03 UTC #14 lore raw

PATCH 7/7 btrfs: remove local blocksize variable in reada_find_extent

In reply to David Sterba

The name is misleading and the local variable serves no purpose.

Signed-off-by David Sterba <[email protected]>

---
 fs/btrfs/reada.c | 6 ++----
 1 file changed, 2 insertions(+), 4 deletions(-)

fs/btrfs/reada.c +2 -4

index 91df381a60ce..64425c3fe4f5 100644
@@ -318,7 +318,6 @@ static struct reada_extent *reada_find_extent(struct btrfs_fs_info *fs_info,
 	struct btrfs_bio *bbio = NULL;
 	struct btrfs_device *dev;
 	struct btrfs_device *prev_dev;
-	u32 blocksize;
 	u64 length;
 	int real_stripes;
 	int nzones = 0;

@@ -339,7 +338,6 @@ static struct reada_extent *reada_find_extent(struct btrfs_fs_info *fs_info,
 	if (!re)
 		return NULL;
 
-	blocksize = fs_info->nodesize;
 	re->logical = logical;
 	re->top = *top;
 	INIT_LIST_HEAD(&re->extctl);

@@ -349,10 +347,10 @@ static struct reada_extent *reada_find_extent(struct btrfs_fs_info *fs_info,
 	/*
 	 * map block
 	 */
-	length = blocksize;
+	length = fs_info->nodesize;
 	ret = btrfs_map_block(fs_info, BTRFS_MAP_GET_READ_MIRRORS, logical,
 			&length, &bbio, 0);
-	if (ret || !bbio || length < blocksize)
+	if (ret || !bbio || length < fs_info->nodesize)
 		goto error;
 
 	if (bbio->num_stripes > BTRFS_MAX_MIRRORS) {

Signature · 1 line

2.12.0

Liu Bo

Reviewed-by on v1 7/7

2017-03-20 21:10 UTC #15 lore raw

Re: [PATCH 7/7] btrfs: remove local blocksize variable in reada_find_extent

Quoted 1 line · David Sterba

The name is misleading and the local variable serves no purpose.

Reviewed-by Liu Bo <[email protected]>

Thanks,

-liubo

Quoted 46 lines

Signed-off-by David Sterba <[email protected]>

---
 fs/btrfs/reada.c | 6 ++----
 1 file changed, 2 insertions(+), 4 deletions(-)

fs/btrfs/reada.c +2 -4

index 91df381a60ce..64425c3fe4f5 100644
@@ -318,7 +318,6 @@ static struct reada_extent *reada_find_extent(struct btrfs_fs_info *fs_info,
 	struct btrfs_bio *bbio = NULL;
 	struct btrfs_device *dev;
 	struct btrfs_device *prev_dev;
-	u32 blocksize;
 	u64 length;
 	int real_stripes;
 	int nzones = 0;

@@ -339,7 +338,6 @@ static struct reada_extent *reada_find_extent(struct btrfs_fs_info *fs_info,
 	if (!re)
 		return NULL;
 
-	blocksize = fs_info->nodesize;
 	re->logical = logical;
 	re->top = *top;
 	INIT_LIST_HEAD(&re->extctl);

@@ -349,10 +347,10 @@ static struct reada_extent *reada_find_extent(struct btrfs_fs_info *fs_info,
 	/*
 	 * map block
 	 */
-	length = blocksize;
+	length = fs_info->nodesize;
 	ret = btrfs_map_block(fs_info, BTRFS_MAP_GET_READ_MIRRORS, logical,
 			&length, &bbio, 0);
-	if (ret || !bbio || length < blocksize)
+	if (ret || !bbio || length < fs_info->nodesize)
 		goto error;
 
 	if (bbio->num_stripes > BTRFS_MAX_MIRRORS) {

Signature · 6 lines

2.12.0

--
To unsubscribe from this list: send the line "unsubscribe linux-btrfs" in
the body of a message to [email protected]
More majordomo info at  http://vger.kernel.org/majordomo-info.html