All of lore.kernel.org
 help / color / mirror / Atom feed
From: Qu Wenruo <quwenruo.btrfs@gmx.com>
To: Josef Bacik <josef@toxicpanda.com>,
	kernel-team@fb.com, linux-btrfs@vger.kernel.org
Subject: Re: [PATCH 6/7] btrfs: hold a ref on the root->reloc_root
Date: Tue, 3 Mar 2020 09:12:21 +0800	[thread overview]
Message-ID: <1b3545b5-ab19-6f0d-7dd3-d80fe20a865e@gmx.com> (raw)
In-Reply-To: <20200302184757.44176-7-josef@toxicpanda.com>


[-- Attachment #1.1: Type: text/plain, Size: 4904 bytes --]



On 2020/3/3 上午2:47, Josef Bacik wrote:
> We previously were relying on root->reloc_root to be cleaned up by the
> drop snapshot, or the error handling.  However if btrfs_drop_snapshot()
> failed it wouldn't drop the ref for the root.  Also we sort of depend on
> the right thing to happen with moving reloc roots between lists and the
> fs root they belong to, which makes it hard to figure out who owns the
> reference.
> 
> Fix this by explicitly holding a reference on the reloc root for
> roo->reloc_root.  This means that we hold two references on reloc roots,
> one for whichever reloc_roots list it's attached to, and the
> root->reloc_root we're on.
> 
> This makes it easier to reason out who owns a reference on the root, and
> when it needs to be dropped.
> 
> Signed-off-by: Josef Bacik <josef@toxicpanda.com>

A small question inlined below, despite that,

Reviewed-by: Qu Wenruo <wqu@suse.com>

> ---
>  fs/btrfs/relocation.c | 44 ++++++++++++++++++++++++++++++++-----------
>  1 file changed, 33 insertions(+), 11 deletions(-)
> 
> diff --git a/fs/btrfs/relocation.c b/fs/btrfs/relocation.c
> index acd21c156378..c8ff28930677 100644
> --- a/fs/btrfs/relocation.c
> +++ b/fs/btrfs/relocation.c
> @@ -1384,6 +1384,7 @@ static void __del_reloc_root(struct btrfs_root *root)
>  	struct rb_node *rb_node;
>  	struct mapping_node *node = NULL;
>  	struct reloc_control *rc = fs_info->reloc_ctl;
> +	bool put_ref = false;
>  
>  	if (rc && root->node) {
>  		spin_lock(&rc->reloc_root_tree.lock);
> @@ -1400,8 +1401,13 @@ static void __del_reloc_root(struct btrfs_root *root)
>  	}
>  
>  	spin_lock(&fs_info->trans_lock);
> -	list_del_init(&root->root_list);
> +	if (!list_empty(&root->root_list)) {

Can we make the ref of reloc root completely free from the list operation?
It still looks like a compromise between fully ref counted reloc root
and original non-ref counted one.

Thanks,
Qu

> +		put_ref = true;
> +		list_del_init(&root->root_list);
> +	}
>  	spin_unlock(&fs_info->trans_lock);
> +	if (put_ref)
> +		btrfs_put_root(root);
>  	kfree(node);
>  }
>  
> @@ -1555,7 +1561,7 @@ int btrfs_init_reloc_root(struct btrfs_trans_handle *trans,
>  
>  	ret = __add_reloc_root(reloc_root);
>  	BUG_ON(ret < 0);
> -	root->reloc_root = reloc_root;
> +	root->reloc_root = btrfs_grab_root(reloc_root);
>  	return 0;
>  }
>  
> @@ -1576,6 +1582,13 @@ int btrfs_update_reloc_root(struct btrfs_trans_handle *trans,
>  	reloc_root = root->reloc_root;
>  	root_item = &reloc_root->root_item;
>  
> +	/*
> +	 * We are probably ok here, but __del_reloc_root() will drop its ref of
> +	 * the root.  We have the ref fro root->reloc_root, but just in case

s/fro/for/

> +	 * hold it while we update the reloc root.
> +	 */
> +	btrfs_grab_root(reloc_root);
> +
>  	/* root->reloc_root will stay until current relocation finished */
>  	if (fs_info->reloc_ctl->merge_reloc_tree &&
>  	    btrfs_root_refs(root_item) == 0) {
> @@ -1597,7 +1610,7 @@ int btrfs_update_reloc_root(struct btrfs_trans_handle *trans,
>  	ret = btrfs_update_root(trans, fs_info->tree_root,
>  				&reloc_root->root_key, root_item);
>  	BUG_ON(ret);
> -
> +	btrfs_put_root(reloc_root);
>  out:
>  	return 0;
>  }
> @@ -2297,19 +2310,28 @@ static int clean_dirty_subvols(struct reloc_control *rc)
>  			 */
>  			smp_wmb();
>  			clear_bit(BTRFS_ROOT_DEAD_RELOC_TREE, &root->state);
> -
>  			if (reloc_root) {
> -
> +				/*
> +				 * btrfs_drop_snapshot drops our ref we hold for
> +				 * ->reloc_root.  If it fails however we must
> +				 * drop the ref ourselves.
> +				 */
>  				ret2 = btrfs_drop_snapshot(reloc_root, NULL, 0, 1);
> -				if (ret2 < 0 && !ret)
> -					ret = ret2;
> +				if (ret2 < 0) {
> +					btrfs_put_root(reloc_root);
> +					if (!ret)
> +						ret = ret2;
> +				}
>  			}
>  			btrfs_put_root(root);
>  		} else {
>  			/* Orphan reloc tree, just clean it up */
>  			ret2 = btrfs_drop_snapshot(root, NULL, 0, 1);
> -			if (ret2 < 0 && !ret)
> -				ret = ret2;
> +			if (ret2 < 0) {
> +				btrfs_put_root(root);
> +				if (!ret)
> +					ret = ret2;
> +			}
>  		}
>  	}
>  	return ret;
> @@ -4687,7 +4709,7 @@ int btrfs_recover_relocation(struct btrfs_root *root)
>  
>  		err = __add_reloc_root(reloc_root);
>  		BUG_ON(err < 0); /* -ENOMEM or logic error */
> -		fs_root->reloc_root = reloc_root;
> +		fs_root->reloc_root = btrfs_grab_root(reloc_root);
>  		btrfs_put_root(fs_root);
>  	}
>  
> @@ -4912,7 +4934,7 @@ int btrfs_reloc_post_snapshot(struct btrfs_trans_handle *trans,
>  
>  	ret = __add_reloc_root(reloc_root);
>  	BUG_ON(ret < 0);
> -	new_root->reloc_root = reloc_root;
> +	new_root->reloc_root = btrfs_grab_root(reloc_root);
>  
>  	if (rc->create_reloc_tree)
>  		ret = clone_backref_node(trans, rc, root, reloc_root);
> 


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 488 bytes --]

  reply	other threads:[~2020-03-03  1:12 UTC|newest]

Thread overview: 26+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2020-03-02 18:47 [PATCH 0/7] relocation error handling fixes Josef Bacik
2020-03-02 18:47 ` [PATCH 1/7] btrfs: drop block from cache on error in relocation Josef Bacik
2020-03-03  0:59   ` Qu Wenruo
2020-03-03 14:59   ` David Sterba
2020-03-03 20:27     ` Josef Bacik
2020-03-02 18:47 ` [PATCH 2/7] btrfs: unset reloc control if we fail to recover Josef Bacik
2020-03-03  0:58   ` Qu Wenruo
2020-03-03  1:03     ` Josef Bacik
2020-03-03  1:18       ` Qu Wenruo
2020-03-03 15:21     ` David Sterba
2020-03-02 18:47 ` [PATCH 3/7] btrfs: splice rc->reloc_roots onto reloc roots in recover Josef Bacik
2020-03-03  1:02   ` Qu Wenruo
2020-03-02 18:47 ` [PATCH 4/7] btrfs: run clean_dirty_subvols if we fail to start a trans Josef Bacik
2020-03-03  1:04   ` Qu Wenruo
2020-03-03 15:32   ` David Sterba
2020-03-02 18:47 ` [PATCH 5/7] btrfs: clear BTRFS_ROOT_DEAD_RELOC_TREE before dropping the reloc root Josef Bacik
2020-03-02 19:31   ` David Sterba
2020-03-02 19:51     ` Josef Bacik
2020-03-03 15:34       ` David Sterba
2020-03-03  0:31   ` Qu Wenruo
2020-03-02 18:47 ` [PATCH 6/7] btrfs: hold a ref on the root->reloc_root Josef Bacik
2020-03-03  1:12   ` Qu Wenruo [this message]
2020-03-03  1:14     ` Josef Bacik
2020-03-03 15:51   ` David Sterba
2020-03-02 18:47 ` [PATCH 7/7] btrfs: remove a BUG_ON() from merge_reloc_roots() Josef Bacik
2020-03-03  1:17   ` Qu Wenruo

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=1b3545b5-ab19-6f0d-7dd3-d80fe20a865e@gmx.com \
    --to=quwenruo.btrfs@gmx.com \
    --cc=josef@toxicpanda.com \
    --cc=kernel-team@fb.com \
    --cc=linux-btrfs@vger.kernel.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.