Re: [PATCH 1/8] btrfs: delayed-ref: Introduce better documented delayed ref structures

[Qu Wenruo <quwenruo.btrfs@gmx.com>]

On 2018/12/10 下午5:48, Nikolay Borisov wrote:
> 
> 
> On 6.12.18 г. 8:58 ч., Qu Wenruo wrote:
>> Current delayed ref interface has several problems:
>> - Longer and longer parameter lists
>>   bytenr
>>   num_bytes
>>   parent
>>   ref_root
>>   owner
>>   offset
>>   for_reloc << Only qgroup code cares.
>>
>> - Different interpretation for the same parameter
>>   Above @owner for data ref is ino owning this extent,
>>   while for tree ref, it's level. They are even in different size range.
>>   For level we only need 0~8, while for ino it's
>>   BTRFS_FIRST_FREE_OBJECTID~BTRFS_LAST_FREE_OBJECTID.
>>
>>   And @offset doesn't even makes sense for tree ref.
>>
>>   Such parameter reuse may look clever as an hidden union, but it
>>   destroys code readability.
>>
>> To solve both problems, we introduce a new structure, btrfs_ref to solve
>> them:
>>
>> - Structure instead of long parameter list
>>   This makes later expansion easier, and better documented.
>>
>> - Use btrfs_ref::type to distinguish data and tree ref
>>
>> - Use proper union to store data/tree ref specific structures.
>>
>> - Use separate functions to fill data/tree ref data, with a common generic
>>   function to fill common bytenr/num_bytes members.
>>
>> All parameters will find its place in btrfs_ref, and an extra member,
>> real_root, inspired by ref-verify code, is newly introduced for later
>> qgroup code, to record which tree is triggered this extent modification.
>>
>> This patch doesn't touch any code, but provides the basis for incoming
>> refactors.
>>
>> Signed-off-by: Qu Wenruo <wqu@suse.com>
>> ---
>>  fs/btrfs/delayed-ref.h | 109 +++++++++++++++++++++++++++++++++++++++++
>>  1 file changed, 109 insertions(+)
>>
>> diff --git a/fs/btrfs/delayed-ref.h b/fs/btrfs/delayed-ref.h
>> index d8fa12d3f2cc..e36d6b05d85e 100644
>> --- a/fs/btrfs/delayed-ref.h
>> +++ b/fs/btrfs/delayed-ref.h
>> @@ -176,6 +176,81 @@ struct btrfs_delayed_ref_root {
>>  	u64 qgroup_to_skip;
>>  };
>>  
>> +enum btrfs_ref_type {
>> +	BTRFS_REF_NOT_SET = 0,
> 
> No need for explicit initialisation when your enums start from 0.

Forgot that.

> 
>> +	BTRFS_REF_DATA,
>> +	BTRFS_REF_METADATA,
>> +	BTRFS_REF_LAST,
>> +};
>> +
>> +struct btrfs_data_ref {
>> +	/*
>> +	 * For EXTENT_DATA_REF
>> +	 *
>> +	 * @ref_root:	current owner of the extent.
>> +	 * 		may differ from btrfs_ref::real_root.
> 
> Here 'owner' is a bit ambiguous. Isn't this the root of the owner

Well, the @ref_root is a little confusing.
Here it could mean "the current btrfs_header_owner(eb)".

I don't really know the correct term here.

> 
>> +	 * @ino: 	inode number of the owner. 
> 
> And the owner is really the owning inode?

Yes, the owning inode.

> 
>> +	 * @offset:	*CALCULATED* offset. Not EXTENT_DATA key offset.
> 
> What does calculated mean here?

Because btrfs data backref instead of using the offset of the owning
EXTENT_DATA key, it uses key->offset - extent_offset.

> 
>> +	 *
>> +	 */
> 
> Ugh, this is ugly, why not have a single /* */ line above each member
> and document them like that?

Because the complexity of @ref_root can't really go one line.

> 
>> +	u64 ref_root;
>> +	u64 ino;
>> +	u64 offset;
>> +};
>> +
>> +struct btrfs_tree_ref {
>> +	/* Common for all sub types and skinny combination */
>> +	int level;
> 
> Like you've done here. Mention that this is the level in the tree that
> this reference is located at.
> 
>> +
>> +	/*
>> +	 * For TREE_BLOCK_REF (skinny metadata, either inline or keyed)
>> +	 *
>> +	 * root here may differ from btrfs_ref::real_root.
>> +	 */
> 
> Make it refer to the documentation of the generic btrfs_ref structure
> because if someone reads this comment and doesn't read the one in
> btrfs_ref then it's not clear under what conditions they might differ.
> 
>> +	u64 root;
>> +
>> +	/* For non-skinny metadata, no special member needed */
>> +};
>> +
>> +struct btrfs_ref {
>> +	enum btrfs_ref_type type;
>> +	int action;
>> +
>> +	/*
>> +	 * Use full backref(SHARED_BLOCK_REF or SHARED_DATA_REF) for this
>> +	 * extent and its children.
>> +	 * Set for reloc trees.
>> +	 */
>> +	unsigned int use_fullback:1;
> 
> 'fullback' is too terse, use_backreferences or something conveying more
> information?

@use_backreferences looks even stranger to me.

Here the point is, if this bit set, all backref will use
SHARED_BLOCK_REF or SHARED_DATA_REF subtype, and just leave a pointer to
its parent.

Any good idea for explaining this?

> Also please use explicit bool type:
> 
> bool xxxx:1

Is this valid? Haven't seen such usage in kernel code IIRC.

> 
>> +
>> +	/*
>> +	 * Whether this extent should go through qgroup record.
>> +	 * Normally false, but for certain case like delayed subtree scan,
>> +	 * this can hugely reduce qgroup overhead.
>> +	 */
>> +	unsigned int skip_qgroup:1;
> 
> again, explicit bool type please.
> 
>> +
>> +	/*
>> +	 * Who owns this reference modification, optional.
>> +	 *
>> +	 * One example:
>> +	 * When creating reloc tree for source fs, it will increase tree block
>> +	 * ref for children tree blocks.
>> +	 * In that case, btrfs_ref::real_root = reloc tree,
>> +	 * while btrfs_ref::tree_ref::root = fs tree.
>> +	 */
>> +	u64 real_root;
>> +	u64 bytenr;
>> +	u64 len;
>> +
>> +	/* Common @parent for SHARED_DATA_REF/SHARED_BLOCK_REF */
> 
> OK, it's common but what does it really contain? Either document it or
> rename it to parent_block or something.
> 
>> +	u64 parent;
>> +	union {
>> +		struct btrfs_data_ref data_ref;
>> +		struct btrfs_tree_ref tree_ref;
>> +	};
>> +};
>> +
>>  extern struct kmem_cache *btrfs_delayed_ref_head_cachep;
>>  extern struct kmem_cache *btrfs_delayed_tree_ref_cachep;
>>  extern struct kmem_cache *btrfs_delayed_data_ref_cachep;
>> @@ -184,6 +259,40 @@ extern struct kmem_cache *btrfs_delayed_extent_op_cachep;
>>  int __init btrfs_delayed_ref_init(void);
>>  void __cold btrfs_delayed_ref_exit(void);
>>  
>> +static inline void btrfs_init_generic_ref(struct btrfs_ref *generic_ref,
>> +				int action, u64 bytenr, u64 len, u64 real_root,
>> +				u64 parent)
>> +{
>> +	generic_ref->action = action;
> 
> IMO it makes sense in this series to have a patch which converts the
> action defines to an enum and subsequently modify functions/structs to
> actually be of enum type.

Makes sense.

Thanks,
Qu

> 
>> +	generic_ref->bytenr = bytenr;
>> +	generic_ref->len = len;
>> +	generic_ref->real_root = real_root;
>> +	generic_ref->parent = parent;
>> +}
>> +
>> +static inline void btrfs_init_tree_ref(struct btrfs_ref *generic_ref,
>> +				int level, u64 root)
>> +{
>> +	/* If @real_root not set, use @root as fallback */
>> +	if (!generic_ref->real_root)
>> +		generic_ref->real_root = root;
>> +	generic_ref->tree_ref.level = level;
>> +	generic_ref->tree_ref.root = root;
>> +	generic_ref->type = BTRFS_REF_METADATA;
>> +}
>> +
>> +static inline void btrfs_init_data_ref(struct btrfs_ref *generic_ref,
>> +				u64 ref_root, u64 ino, u64 offset)
>> +{
>> +	/* If @real_root not set, use @root as fallback */
>> +	if (!generic_ref->real_root)
>> +		generic_ref->real_root = ref_root;
>> +	generic_ref->data_ref.ref_root = ref_root;
>> +	generic_ref->data_ref.ino = ino;
>> +	generic_ref->data_ref.offset = offset;
>> +	generic_ref->type = BTRFS_REF_DATA;
>> +}
>> +
>>  static inline struct btrfs_delayed_extent_op *
>>  btrfs_alloc_delayed_extent_op(void)
>>  {
>>