public inbox for linux-btrfs@vger.kernel.org
 help / color / mirror / Atom feed
From: Filipe Manana <fdmanana@kernel.org>
To: Nikolay Borisov <nborisov@suse.com>
Cc: linux-btrfs@vger.kernel.org
Subject: Re: [PATCH 07/12] btrfs: improve batch deletion of delayed dir index items
Date: Thu, 2 Jun 2022 09:55:02 +0100	[thread overview]
Message-ID: <20220602085502.GA3331364@falcondesktop> (raw)
In-Reply-To: <2cc71b46-7e53-d1a1-327f-39b28b18687e@suse.com>

On Thu, Jun 02, 2022 at 11:24:53AM +0300, Nikolay Borisov wrote:
> 
> 
> On 31.05.22 г. 18:06 ч., fdmanana@kernel.org wrote:
> > From: Filipe Manana <fdmanana@suse.com>
> > 
> 
> <snip>
> 
> Reviewed-by: Nikolay Borisov <nborisov@suse.com>
> 
> One small idea, see further down below though.

That's a rather different thing from what the patch intends, which
is just to improve the deletion of items in the btree.

Send that as a separate patch with a proper changelog.

Thanks.

> 
> > 
> > Signed-off-by: Filipe Manana <fdmanana@suse.com>
> > ---
> >   fs/btrfs/delayed-inode.c | 60 +++++++++++++++++-----------------------
> >   1 file changed, 25 insertions(+), 35 deletions(-)
> > 
> > diff --git a/fs/btrfs/delayed-inode.c b/fs/btrfs/delayed-inode.c
> > index 0125586fd565..74c806d3ab2a 100644
> > --- a/fs/btrfs/delayed-inode.c
> > +++ b/fs/btrfs/delayed-inode.c
> > @@ -791,68 +791,58 @@ static int btrfs_batch_delete_items(struct btrfs_trans_handle *trans,
> >   {
> 
> <snip>
> 
> > +	while (slot < last_slot) {
> > +		struct btrfs_key key;
> > -		curr = next;
> >   		next = __btrfs_next_delayed_item(curr);
> >   		if (!next)
> >   			break;
> > -		if (!btrfs_is_continuous_delayed_item(curr, next))
> > +		slot++;
> > +		btrfs_item_key_to_cpu(leaf, &key, slot);
> > +		if (btrfs_comp_cpu_keys(&next->key, &key) != 0)
> >   			break;
> > -
> > -		i++;
> > -		if (i > last_item)
> > -			break;
> > -		btrfs_item_key_to_cpu(leaf, &key, i);
> > +		nitems++;
> > +		curr = next;
> > +		list_add_tail(&curr->tree_list, &batch_list);
> >   	}
> > -	/*
> > -	 * Our caller always gives us a path pointing to an existing item, so
> > -	 * this can not happen.
> > -	 */
> > -	ASSERT(nitems >= 1);
> > -	if (nitems < 1)
> > -		return -ENOENT;
> > -
> >   	ret = btrfs_del_items(trans, root, path, path->slots[0], nitems);
> >   	if (ret)
> > -		goto out;
> > +		return ret;
> > -	list_for_each_entry_safe(curr, next, &head, tree_list) {
> > +	list_for_each_entry_safe(curr, next, &batch_list, tree_list) {
> >   		btrfs_delayed_item_release_metadata(root, curr);
> >   		list_del(&curr->tree_list);
> >   		btrfs_release_delayed_item(curr);
> >   	}
> 
> nit: This function could be further optimized
> with the following snippet.
> 
> diff --git a/fs/btrfs/delayed-inode.c b/fs/btrfs/delayed-inode.c
> index fedbc45b71a2..950f5a2a9950 100644
> --- a/fs/btrfs/delayed-inode.c
> +++ b/fs/btrfs/delayed-inode.c
> @@ -805,6 +805,7 @@ static int btrfs_batch_delete_items(struct btrfs_trans_handle *trans,
>         LIST_HEAD(batch_list);
>         int nitems, slot, last_slot;
>         int ret;
> +       u64 bytes_reserved = item->bytes_reserved;
>         ASSERT(leaf != NULL);
> @@ -840,6 +841,7 @@ static int btrfs_batch_delete_items(struct btrfs_trans_handle *trans,
>                         break;
>                 nitems++;
>                 curr = next;
> +               bytes_reserved += curr->bytes_reserved;
>                 list_add_tail(&curr->tree_list, &batch_list);
>         }
> @@ -847,8 +849,11 @@ static int btrfs_batch_delete_items(struct btrfs_trans_handle *trans,
>         if (ret)
>                 return ret;
> +       btrfs_block_rsv_release(root->fs_info, &root->fs_info->delayed_block_rsv,
> +                               bytes_reserved, NULL);
> +
>         list_for_each_entry_safe(curr, next, &batch_list, tree_list) {
> -               btrfs_delayed_item_release_metadata(root, curr);
> +               //btrfs_delayed_item_release_metadata(root, curr);
>                 list_del(&curr->tree_list);
>                 btrfs_release_delayed_item(curr);
>         }
> 
> 
> Running your series VS this diff on top produces:
> 
> Before this change:
> @block_rsv_release: 1004
> @btrfs_delete_delayed_items_total_time: 14602
> @delete_batches: 505
> 
> After:
> @block_rsv_release: 510
> @btrfs_delete_delayed_items_total_time: 13643
> @delete_batches: 507
> 
> 
> > -out:
> > -	return ret;
> > +	return 0;
> >   }
> >   static int btrfs_delete_delayed_items(struct btrfs_trans_handle *trans,

  reply	other threads:[~2022-06-02  8:55 UTC|newest]

Thread overview: 34+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2022-05-31 15:06 [PATCH 00/12] btrfs: some improvements and cleanups around delayed items fdmanana
2022-05-31 15:06 ` [PATCH 01/12] btrfs: balance btree dirty pages and delayed items after a rename fdmanana
2022-05-31 15:16   ` Nikolay Borisov
2022-05-31 23:13   ` Anand Jain
2022-05-31 15:06 ` [PATCH 02/12] btrfs: free the path earlier when creating a new inode fdmanana
2022-05-31 15:21   ` Nikolay Borisov
2022-05-31 23:22   ` Anand Jain
2022-06-01  9:34     ` Filipe Manana
2022-06-01 11:11       ` Anand Jain
2022-06-01 11:51         ` David Sterba
2022-05-31 15:06 ` [PATCH 03/12] btrfs: balance btree dirty pages and delayed items after clone and dedupe fdmanana
2022-06-01  0:54   ` Anand Jain
2022-05-31 15:06 ` [PATCH 04/12] btrfs: add assertions when deleting batches of delayed items fdmanana
2022-06-01  1:34   ` Anand Jain
2022-05-31 15:06 ` [PATCH 05/12] btrfs: deal with deletion errors when deleting " fdmanana
2022-06-01  1:44   ` Anand Jain
2022-05-31 15:06 ` [PATCH 06/12] btrfs: refactor the delayed item deletion entry point fdmanana
2022-05-31 15:06 ` [PATCH 07/12] btrfs: improve batch deletion of delayed dir index items fdmanana
2022-06-02  8:24   ` Nikolay Borisov
2022-06-02  8:55     ` Filipe Manana [this message]
2022-05-31 15:06 ` [PATCH 08/12] btrfs: assert that delayed item is a dir index item when adding it fdmanana
2022-05-31 15:06 ` [PATCH 09/12] btrfs: improve batch insertion of delayed dir index items fdmanana
2022-05-31 15:06 ` [PATCH 10/12] btrfs: do not BUG_ON() on failure to reserve metadata for delayed item fdmanana
2022-05-31 15:06 ` [PATCH 11/12] btrfs: set delayed item type when initializing it fdmanana
2022-05-31 15:06 ` [PATCH 12/12] btrfs: reduce amount of reserved metadata for delayed item insertion fdmanana
2022-06-08 15:23   ` [btrfs] 62bd8124e2: WARNING:at_fs/btrfs/block-rsv.c:#btrfs_release_global_block_rsv[btrfs] kernel test robot
2022-06-09  9:46     ` Filipe Manana
2022-06-10  1:26       ` Oliver Sang
2022-06-12 14:36         ` Oliver Sang
2022-06-13 10:50           ` Filipe Manana
2022-06-16  2:42             ` Oliver Sang
2022-06-17 10:32               ` Filipe Manana
2022-06-01 18:35 ` [PATCH 00/12] btrfs: some improvements and cleanups around delayed items David Sterba
2022-06-02  9:34 ` Nikolay Borisov

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=20220602085502.GA3331364@falcondesktop \
    --to=fdmanana@kernel.org \
    --cc=linux-btrfs@vger.kernel.org \
    --cc=nborisov@suse.com \
    /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 a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox