[PATCH 0/3] btrfs: minor block group cleanups

[PATCH 0/3] btrfs: minor block group cleanups

[Anand Jain]

Minor cleanup patches.
1/3 cleansup the return value of btrfs_return_cluster_to_free_space().
2/3 and 3/3 together cleanups block group reference count.

2/3 and 3/3 are related. However 1/3 is independent to the rest of the
patches in this set.

Anand Jain (3):
  btrfs: fix btrfs_return_cluster_to_free_space() return
  btrfs: rename btrfs_block_group::count
  btrfs: use helper btrfs_get_block_group

 fs/btrfs/block-group.c      |  8 ++++----
 fs/btrfs/block-group.h      |  2 +-
 fs/btrfs/free-space-cache.c | 17 +++++++----------
 fs/btrfs/free-space-cache.h |  2 +-
 4 files changed, 13 insertions(+), 16 deletions(-)

-- 
2.25.1

[PATCH 1/3] btrfs: fix btrfs_return_cluster_to_free_space() return

[Anand Jain]

__btrfs_return_cluster_to_free_space() returns only 0. And all its parent
functions don't need the return value either so make this a void function.

Further, as none of the callers of btrfs_return_cluster_to_free_space() is
actually using the return from this function, make this function also
return void.

Signed-off-by: Anand Jain <anand.jain@oracle.com>
---
 fs/btrfs/free-space-cache.c | 13 +++++--------
 fs/btrfs/free-space-cache.h |  2 +-
 2 files changed, 6 insertions(+), 9 deletions(-)

diff --git a/fs/btrfs/free-space-cache.c b/fs/btrfs/free-space-cache.c
index 55955bd424d7..1bf08c855edb 100644
--- a/fs/btrfs/free-space-cache.c
+++ b/fs/btrfs/free-space-cache.c
@@ -2703,7 +2703,7 @@ void btrfs_init_free_space_ctl(struct btrfs_block_group *block_group)
  * pointed to by the cluster, someone else raced in and freed the
  * cluster already.  In that case, we just return without changing anything
  */
-static int
+static void
 __btrfs_return_cluster_to_free_space(
 			     struct btrfs_block_group *block_group,
 			     struct btrfs_free_cluster *cluster)
@@ -2756,7 +2756,6 @@ __btrfs_return_cluster_to_free_space(
 out:
 	spin_unlock(&cluster->lock);
 	btrfs_put_block_group(block_group);
-	return 0;
 }
 
 static void __btrfs_remove_free_space_cache_locked(
@@ -2907,12 +2906,11 @@ u64 btrfs_find_space_for_alloc(struct btrfs_block_group *block_group,
  * Otherwise, it'll get a reference on the block group pointed to by the
  * cluster and remove the cluster from it.
  */
-int btrfs_return_cluster_to_free_space(
+void btrfs_return_cluster_to_free_space(
 			       struct btrfs_block_group *block_group,
 			       struct btrfs_free_cluster *cluster)
 {
 	struct btrfs_free_space_ctl *ctl;
-	int ret;
 
 	/* first, get a safe pointer to the block group */
 	spin_lock(&cluster->lock);
@@ -2920,12 +2918,12 @@ int btrfs_return_cluster_to_free_space(
 		block_group = cluster->block_group;
 		if (!block_group) {
 			spin_unlock(&cluster->lock);
-			return 0;
+			return;
 		}
 	} else if (cluster->block_group != block_group) {
 		/* someone else has already freed it don't redo their work */
 		spin_unlock(&cluster->lock);
-		return 0;
+		return;
 	}
 	atomic_inc(&block_group->count);
 	spin_unlock(&cluster->lock);
@@ -2934,14 +2932,13 @@ int btrfs_return_cluster_to_free_space(
 
 	/* now return any extents the cluster had on it */
 	spin_lock(&ctl->tree_lock);
-	ret = __btrfs_return_cluster_to_free_space(block_group, cluster);
+	__btrfs_return_cluster_to_free_space(block_group, cluster);
 	spin_unlock(&ctl->tree_lock);
 
 	btrfs_discard_queue_work(&block_group->fs_info->discard_ctl, block_group);
 
 	/* finally drop our ref */
 	btrfs_put_block_group(block_group);
-	return ret;
 }
 
 static u64 btrfs_alloc_from_bitmap(struct btrfs_block_group *block_group,
diff --git a/fs/btrfs/free-space-cache.h b/fs/btrfs/free-space-cache.h
index 2e0a8077aa74..e3d5e0ad8f8e 100644
--- a/fs/btrfs/free-space-cache.h
+++ b/fs/btrfs/free-space-cache.h
@@ -136,7 +136,7 @@ void btrfs_init_free_cluster(struct btrfs_free_cluster *cluster);
 u64 btrfs_alloc_from_cluster(struct btrfs_block_group *block_group,
 			     struct btrfs_free_cluster *cluster, u64 bytes,
 			     u64 min_start, u64 *max_extent_size);
-int btrfs_return_cluster_to_free_space(
+void btrfs_return_cluster_to_free_space(
 			       struct btrfs_block_group *block_group,
 			       struct btrfs_free_cluster *cluster);
 int btrfs_trim_block_group(struct btrfs_block_group *block_group,
-- 
2.25.1

[PATCH 2/3] btrfs: rename btrfs_block_group::count

[Anand Jain]

The name 'count' is a very commonly used name. It is often difficult to
review the code to check if there is any leak. So rename it to
'bg_count', which is unique enough.

This patch also serves as a preparatory patch to either make sure
btrfs_get_block_group() is used instead of open coded the same or just
open code every where as btrfs_get_block_group() is a one-liner.

Signed-off-by: Anand Jain <anand.jain@oracle.com>
---
 fs/btrfs/block-group.c      | 8 ++++----
 fs/btrfs/block-group.h      | 2 +-
 fs/btrfs/free-space-cache.c | 4 ++--
 3 files changed, 7 insertions(+), 7 deletions(-)

diff --git a/fs/btrfs/block-group.c b/fs/btrfs/block-group.c
index 31ca2cfb7e3e..8111f6750063 100644
--- a/fs/btrfs/block-group.c
+++ b/fs/btrfs/block-group.c
@@ -118,12 +118,12 @@ u64 btrfs_get_alloc_profile(struct btrfs_fs_info *fs_info, u64 orig_flags)
 
 void btrfs_get_block_group(struct btrfs_block_group *cache)
 {
-	atomic_inc(&cache->count);
+	atomic_inc(&cache->bg_count);
 }
 
 void btrfs_put_block_group(struct btrfs_block_group *cache)
 {
-	if (atomic_dec_and_test(&cache->count)) {
+	if (atomic_dec_and_test(&cache->bg_count)) {
 		WARN_ON(cache->pinned > 0);
 		WARN_ON(cache->reserved > 0);
 
@@ -1805,7 +1805,7 @@ static struct btrfs_block_group *btrfs_create_block_group_cache(
 
 	cache->discard_index = BTRFS_DISCARD_INDEX_UNUSED;
 
-	atomic_set(&cache->count, 1);
+	atomic_set(&cache->bg_count, 1);
 	spin_lock_init(&cache->lock);
 	init_rwsem(&cache->data_rwsem);
 	INIT_LIST_HEAD(&cache->list);
@@ -3379,7 +3379,7 @@ int btrfs_free_block_groups(struct btrfs_fs_info *info)
 		ASSERT(list_empty(&block_group->dirty_list));
 		ASSERT(list_empty(&block_group->io_list));
 		ASSERT(list_empty(&block_group->bg_list));
-		ASSERT(atomic_read(&block_group->count) == 1);
+		ASSERT(atomic_read(&block_group->bg_count) == 1);
 		btrfs_put_block_group(block_group);
 
 		spin_lock(&info->block_group_cache_lock);
diff --git a/fs/btrfs/block-group.h b/fs/btrfs/block-group.h
index b6ee70a039c7..f0ef8be08747 100644
--- a/fs/btrfs/block-group.h
+++ b/fs/btrfs/block-group.h
@@ -115,7 +115,7 @@ struct btrfs_block_group {
 	struct list_head list;
 
 	/* Usage count */
-	atomic_t count;
+	atomic_t bg_count;
 
 	/*
 	 * List of struct btrfs_free_clusters for this block group.
diff --git a/fs/btrfs/free-space-cache.c b/fs/btrfs/free-space-cache.c
index 1bf08c855edb..169b1117c1a3 100644
--- a/fs/btrfs/free-space-cache.c
+++ b/fs/btrfs/free-space-cache.c
@@ -2925,7 +2925,7 @@ void btrfs_return_cluster_to_free_space(
 		spin_unlock(&cluster->lock);
 		return;
 	}
-	atomic_inc(&block_group->count);
+	atomic_inc(&block_group->bg_count);
 	spin_unlock(&cluster->lock);
 
 	ctl = block_group->free_space_ctl;
@@ -3355,7 +3355,7 @@ int btrfs_find_space_cluster(struct btrfs_block_group *block_group,
 		list_del_init(&entry->list);
 
 	if (!ret) {
-		atomic_inc(&block_group->count);
+		atomic_inc(&block_group->bg_count);
 		list_add_tail(&cluster->block_group_list,
 			      &block_group->cluster_list);
 		cluster->block_group = block_group;
-- 
2.25.1

[PATCH 3/3] btrfs: use helper btrfs_get_block_group

[Anand Jain]

Use the helper function where it is open coded to increment the
block_group reference count 'bg_count' so that it is less confusing
while reviewing the code. As btrfs_get_block_group() is a one-liner
we could have open-coded it, but its partner function
btrfs_put_block_group() isn't one-liner which does the free part in it.

Signed-off-by: Anand Jain <anand.jain@oracle.com>
---
 fs/btrfs/free-space-cache.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/fs/btrfs/free-space-cache.c b/fs/btrfs/free-space-cache.c
index 169b1117c1a3..867204b48ccf 100644
--- a/fs/btrfs/free-space-cache.c
+++ b/fs/btrfs/free-space-cache.c
@@ -2925,7 +2925,7 @@ void btrfs_return_cluster_to_free_space(
 		spin_unlock(&cluster->lock);
 		return;
 	}
-	atomic_inc(&block_group->bg_count);
+	btrfs_get_block_group(block_group);
 	spin_unlock(&cluster->lock);
 
 	ctl = block_group->free_space_ctl;
@@ -3355,7 +3355,7 @@ int btrfs_find_space_cluster(struct btrfs_block_group *block_group,
 		list_del_init(&entry->list);
 
 	if (!ret) {
-		atomic_inc(&block_group->bg_count);
+		btrfs_get_block_group(block_group);
 		list_add_tail(&cluster->block_group_list,
 			      &block_group->cluster_list);
 		cluster->block_group = block_group;
-- 
2.25.1

Re: [PATCH 1/3] btrfs: fix btrfs_return_cluster_to_free_space() return

[Nikolay Borisov]

On 3.06.20 г. 13:10 ч., Anand Jain wrote:
> __btrfs_return_cluster_to_free_space() returns only 0. And all its parent
> functions don't need the return value either so make this a void function.
> 
> Further, as none of the callers of btrfs_return_cluster_to_free_space() is
> actually using the return from this function, make this function also
> return void.
> 
> Signed-off-by: Anand Jain <anand.jain@oracle.com>

Reviewed-by: Nikolay Borisov <nborisov@suse.com>

Re: [PATCH 2/3] btrfs: rename btrfs_block_group::count

[Filipe Manana]

On Wed, Jun 3, 2020 at 11:13 AM Anand Jain <anand.jain@oracle.com> wrote:
>
> The name 'count' is a very commonly used name. It is often difficult to
> review the code to check if there is any leak. So rename it to
> 'bg_count', which is unique enough.

Hum? Seriously?

count to bg_count?
It's a member of the block group structure, adding a bg_ prefix adds
no value at all, we know the count is related to the block group.
I could somewhat understand if you named it 'refcount' instead.

Thanks.

>
> This patch also serves as a preparatory patch to either make sure
> btrfs_get_block_group() is used instead of open coded the same or just
> open code every where as btrfs_get_block_group() is a one-liner.
>
> Signed-off-by: Anand Jain <anand.jain@oracle.com>
> ---
>  fs/btrfs/block-group.c      | 8 ++++----
>  fs/btrfs/block-group.h      | 2 +-
>  fs/btrfs/free-space-cache.c | 4 ++--
>  3 files changed, 7 insertions(+), 7 deletions(-)
>
> diff --git a/fs/btrfs/block-group.c b/fs/btrfs/block-group.c
> index 31ca2cfb7e3e..8111f6750063 100644
> --- a/fs/btrfs/block-group.c
> +++ b/fs/btrfs/block-group.c
> @@ -118,12 +118,12 @@ u64 btrfs_get_alloc_profile(struct btrfs_fs_info *fs_info, u64 orig_flags)
>
>  void btrfs_get_block_group(struct btrfs_block_group *cache)
>  {
> -       atomic_inc(&cache->count);
> +       atomic_inc(&cache->bg_count);
>  }
>
>  void btrfs_put_block_group(struct btrfs_block_group *cache)
>  {
> -       if (atomic_dec_and_test(&cache->count)) {
> +       if (atomic_dec_and_test(&cache->bg_count)) {
>                 WARN_ON(cache->pinned > 0);
>                 WARN_ON(cache->reserved > 0);
>
> @@ -1805,7 +1805,7 @@ static struct btrfs_block_group *btrfs_create_block_group_cache(
>
>         cache->discard_index = BTRFS_DISCARD_INDEX_UNUSED;
>
> -       atomic_set(&cache->count, 1);
> +       atomic_set(&cache->bg_count, 1);
>         spin_lock_init(&cache->lock);
>         init_rwsem(&cache->data_rwsem);
>         INIT_LIST_HEAD(&cache->list);
> @@ -3379,7 +3379,7 @@ int btrfs_free_block_groups(struct btrfs_fs_info *info)
>                 ASSERT(list_empty(&block_group->dirty_list));
>                 ASSERT(list_empty(&block_group->io_list));
>                 ASSERT(list_empty(&block_group->bg_list));
> -               ASSERT(atomic_read(&block_group->count) == 1);
> +               ASSERT(atomic_read(&block_group->bg_count) == 1);
>                 btrfs_put_block_group(block_group);
>
>                 spin_lock(&info->block_group_cache_lock);
> diff --git a/fs/btrfs/block-group.h b/fs/btrfs/block-group.h
> index b6ee70a039c7..f0ef8be08747 100644
> --- a/fs/btrfs/block-group.h
> +++ b/fs/btrfs/block-group.h
> @@ -115,7 +115,7 @@ struct btrfs_block_group {
>         struct list_head list;
>
>         /* Usage count */
> -       atomic_t count;
> +       atomic_t bg_count;
>
>         /*
>          * List of struct btrfs_free_clusters for this block group.
> diff --git a/fs/btrfs/free-space-cache.c b/fs/btrfs/free-space-cache.c
> index 1bf08c855edb..169b1117c1a3 100644
> --- a/fs/btrfs/free-space-cache.c
> +++ b/fs/btrfs/free-space-cache.c
> @@ -2925,7 +2925,7 @@ void btrfs_return_cluster_to_free_space(
>                 spin_unlock(&cluster->lock);
>                 return;
>         }
> -       atomic_inc(&block_group->count);
> +       atomic_inc(&block_group->bg_count);
>         spin_unlock(&cluster->lock);
>
>         ctl = block_group->free_space_ctl;
> @@ -3355,7 +3355,7 @@ int btrfs_find_space_cluster(struct btrfs_block_group *block_group,
>                 list_del_init(&entry->list);
>
>         if (!ret) {
> -               atomic_inc(&block_group->count);
> +               atomic_inc(&block_group->bg_count);
>                 list_add_tail(&cluster->block_group_list,
>                               &block_group->cluster_list);
>                 cluster->block_group = block_group;
> --
> 2.25.1
>


-- 
Filipe David Manana,

“Whether you think you can, or you think you can't — you're right.”

Re: [PATCH 2/3] btrfs: rename btrfs_block_group::count

[Nikolay Borisov]

On 3.06.20 г. 13:10 ч., Anand Jain wrote:
> The name 'count' is a very commonly used name. It is often difficult to
> review the code to check if there is any leak. So rename it to
> 'bg_count', which is unique enough.
> 
> This patch also serves as a preparatory patch to either make sure
> btrfs_get_block_group() is used instead of open coded the same or just
> open code every where as btrfs_get_block_group() is a one-liner.
> 
> Signed-off-by: Anand Jain <anand.jain@oracle.com>

This introduces unnecessary churn, instead if all uses happen through
the btrfs_get_block_group/put apis you can simply grep for those. All
other uses are RO. Also bg_count is not very descriptive given this is
really a refcount. Open-coding probably would be fine if this type is
switched to refcount_t rather than an atomic_t.

Re: [PATCH 3/3] btrfs: use helper btrfs_get_block_group

[Nikolay Borisov]

On 3.06.20 г. 13:10 ч., Anand Jain wrote:
> Use the helper function where it is open coded to increment the
> block_group reference count 'bg_count' so that it is less confusing
> while reviewing the code. As btrfs_get_block_group() is a one-liner
> we could have open-coded it, but its partner function
> btrfs_put_block_group() isn't one-liner which does the free part in it.
> 
> Signed-off-by: Anand Jain <anand.jain@oracle.com>


Reviewed-by: Nikolay Borisov <nborisov@suse.com>

Re: [PATCH 1/3] btrfs: fix btrfs_return_cluster_to_free_space() return

[Johannes Thumshirn]

Looks good,
Reviewed-by: Johannes Thumshirn <johannes.thumshirn@wdc.com>

Re: [PATCH 3/3] btrfs: use helper btrfs_get_block_group

[Johannes Thumshirn]

Looks good,
Reviewed-by: Johannes Thumshirn <johannes.thumshirn@wdc.com>

Re: [PATCH 2/3] btrfs: rename btrfs_block_group::count

[Anand Jain]

On 3/6/20 6:19 pm, Filipe Manana wrote:
> On Wed, Jun 3, 2020 at 11:13 AM Anand Jain <anand.jain@oracle.com> wrote:
>>
>> The name 'count' is a very commonly used name. It is often difficult to
>> review the code to check if there is any leak. So rename it to
>> 'bg_count', which is unique enough.
> 
> Hum? Seriously?
> 
> count to bg_count?
> It's a member of the block group structure, adding a bg_ prefix adds
> no value at all, we know the count is related to the block group.
> I could somewhat understand if you named it 'refcount' instead.


Oh right. Now, bg_count does not make sense to me as well.

Something like bg_refcount is better so that it is easy to search
where all its been used. IMO.

Are we ok with btrfs_block_group::bg_refcount instead ?

Thanks,
Anand


> Thanks.

Re: [PATCH 2/3] btrfs: rename btrfs_block_group::count

[Filipe Manana]

On Wed, Jun 3, 2020 at 11:43 AM Anand Jain <anand.jain@oracle.com> wrote:
>
>
>
> On 3/6/20 6:19 pm, Filipe Manana wrote:
> > On Wed, Jun 3, 2020 at 11:13 AM Anand Jain <anand.jain@oracle.com> wrote:
> >>
> >> The name 'count' is a very commonly used name. It is often difficult to
> >> review the code to check if there is any leak. So rename it to
> >> 'bg_count', which is unique enough.
> >
> > Hum? Seriously?
> >
> > count to bg_count?
> > It's a member of the block group structure, adding a bg_ prefix adds
> > no value at all, we know the count is related to the block group.
> > I could somewhat understand if you named it 'refcount' instead.
>
>
> Oh right. Now, bg_count does not make sense to me as well.
>
> Something like bg_refcount is better so that it is easy to search
> where all its been used. IMO.
>
> Are we ok with btrfs_block_group::bg_refcount instead ?

Why rename at all?
It's pretty obvious it's a reference count. Count/counter is not an
unusual name to use for a reference count, it even has get and put
helpers...

>
> Thanks,
> Anand
>
>
> > Thanks.



-- 
Filipe David Manana,

“Whether you think you can, or you think you can't — you're right.”

Re: [PATCH 2/3] btrfs: rename btrfs_block_group::count

[Anand Jain]

> 
> Why rename at all?
> It's pretty obvious it's a reference count. Count/counter is not an
> unusual name to use for a reference count, it even has get and put
> helpers...

I understand as the member count is in btrfs_block_group so prefix bg_ 
looks redundant, but prefix bg_ is still useful. With out the prefix, 
the grep ">count" *.c didn't help straight away to verify where is 
block_group count used. At two places we missed using the get helper 
which the patch 3/3 fixed.

It's ok to drop this rename patch if you find it's not inline with the 
rest of the codes.

Re: [PATCH 2/3] btrfs: rename btrfs_block_group::count

[Filipe Manana]

On Wed, Jun 3, 2020 at 12:38 PM Anand Jain <anandsuveer@gmail.com> wrote:
>
>
> >
> > Why rename at all?
> > It's pretty obvious it's a reference count. Count/counter is not an
> > unusual name to use for a reference count, it even has get and put
> > helpers...
>
> I understand as the member count is in btrfs_block_group so prefix bg_
> looks redundant, but prefix bg_ is still useful. With out the prefix,
> the grep ">count" *.c didn't help straight away to verify where is

Following that logic, we would add the bg_ prefix to most members of
block_group...
It has members named 'start', 'length', 'flags', etc. Those are
generic too, many other structures have members with exactly those
names...

Don't take it wrong, but I don't find it useful at all.

Thanks.

> block_group count used. At two places we missed using the get helper
> which the patch 3/3 fixed.
>
> It's ok to drop this rename patch if you find it's not inline with the
> rest of the codes.



-- 
Filipe David Manana,

“Whether you think you can, or you think you can't — you're right.”

Re: [PATCH 0/3] btrfs: minor block group cleanups

[David Sterba]

On Wed, Jun 03, 2020 at 06:10:17PM +0800, Anand Jain wrote:
> Minor cleanup patches.
> 1/3 cleansup the return value of btrfs_return_cluster_to_free_space().
> 2/3 and 3/3 together cleanups block group reference count.
> 
> 2/3 and 3/3 are related. However 1/3 is independent to the rest of the
> patches in this set.
> 
> Anand Jain (3):
>   btrfs: fix btrfs_return_cluster_to_free_space() return
>   btrfs: rename btrfs_block_group::count
>   btrfs: use helper btrfs_get_block_group

1 and 3 applied but I don't recommend to spend time on cleaning up
free-space-cache.[ch] as it's going to be phased out.