From: liubo <liubo2009@cn.fujitsu.com>
To: Josef Bacik <josef@redhat.com>
Cc: linux-btrfs@vger.kernel.org
Subject: Re: [PATCH] Btrfs: avoid taking the chunk_mutex in do_chunk_alloc V2
Date: Wed, 13 Apr 2011 09:19:39 +0800 [thread overview]
Message-ID: <4DA4FA2B.7060704@cn.fujitsu.com> (raw)
In-Reply-To: <1302612902-14113-1-git-send-email-josef@redhat.com>
On 04/12/2011 08:55 PM, Josef Bacik wrote:
> Everytime we try to allocate disk space we try and see if we can pre-emptively
> allocate a chunk, but in the common case we don't allocate anything, so there is
> no sense in taking the chunk_mutex at all. So instead if we are allocating a
> chunk, mark it in the space_info so we don't get two people trying to allocate
> at the same time. Thanks,
>
Reviewed-by: Liu Bo <liubo2009@cn.fujitsu.com>
> Signed-off-by: Josef Bacik <josef@redhat.com>
> ---
> V1->V2: Return in the case where we don't need to allocate a chunk instead of
> going to out.
>
> fs/btrfs/ctree.h | 5 +++--
> fs/btrfs/extent-tree.c | 29 ++++++++++++++++++++++++-----
> 2 files changed, 27 insertions(+), 7 deletions(-)
>
> diff --git a/fs/btrfs/ctree.h b/fs/btrfs/ctree.h
> index 0d00a07..a566780 100644
> --- a/fs/btrfs/ctree.h
> +++ b/fs/btrfs/ctree.h
> @@ -740,10 +740,11 @@ struct btrfs_space_info {
> */
> unsigned long reservation_progress;
>
> - int full; /* indicates that we cannot allocate any more
> + int full:1; /* indicates that we cannot allocate any more
> chunks for this space */
> - int force_alloc; /* set if we need to force a chunk alloc for
> + int force_alloc:1; /* set if we need to force a chunk alloc for
> this space */
> + int chunk_alloc:1; /* set if we are allocating a chunk */
>
> struct list_head list;
>
> diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c
> index f619c3c..362cc9b 100644
> --- a/fs/btrfs/extent-tree.c
> +++ b/fs/btrfs/extent-tree.c
> @@ -3020,6 +3020,7 @@ static int update_space_info(struct btrfs_fs_info *info, u64 flags,
> found->bytes_may_use = 0;
> found->full = 0;
> found->force_alloc = 0;
> + found->chunk_alloc = 0;
> *space_info = found;
> list_add_rcu(&found->list, &info->space_info);
> atomic_set(&found->caching_threads, 0);
> @@ -3273,10 +3274,9 @@ static int do_chunk_alloc(struct btrfs_trans_handle *trans,
> {
> struct btrfs_space_info *space_info;
> struct btrfs_fs_info *fs_info = extent_root->fs_info;
> + int wait_for_alloc = 0;
> int ret = 0;
>
> - mutex_lock(&fs_info->chunk_mutex);
> -
> flags = btrfs_reduce_alloc_profile(extent_root, flags);
>
> space_info = __find_space_info(extent_root->fs_info, flags);
> @@ -3287,21 +3287,40 @@ static int do_chunk_alloc(struct btrfs_trans_handle *trans,
> }
> BUG_ON(!space_info);
>
> +again:
> spin_lock(&space_info->lock);
> if (space_info->force_alloc)
> force = 1;
> if (space_info->full) {
> spin_unlock(&space_info->lock);
> - goto out;
> + return 0;
> }
>
> if (!force && !should_alloc_chunk(extent_root, space_info,
> alloc_bytes)) {
> spin_unlock(&space_info->lock);
> - goto out;
> + return 0;
> + } else if (space_info->chunk_alloc) {
> + wait_for_alloc = 1;
> + } else {
> + space_info->chunk_alloc = 1;
> }
> spin_unlock(&space_info->lock);
>
> + mutex_lock(&fs_info->chunk_mutex);
> +
> + /*
> + * The chunk_mutex is held throughout the entirety of a chunk
> + * allocation, so once we've acquired the chunk_mutex we know that the
> + * other guy is done and we need to recheck and see if we should
> + * allocate.
> + */
> + if (wait_for_alloc) {
> + mutex_unlock(&fs_info->chunk_mutex);
> + wait_for_alloc = 0;
> + goto again;
> + }
> +
> /*
> * If we have mixed data/metadata chunks we want to make sure we keep
> * allocating mixed chunks instead of individual chunks.
> @@ -3327,9 +3346,9 @@ static int do_chunk_alloc(struct btrfs_trans_handle *trans,
> space_info->full = 1;
> else
> ret = 1;
> + space_info->chunk_alloc = 0;
> space_info->force_alloc = 0;
> spin_unlock(&space_info->lock);
> -out:
> mutex_unlock(&extent_root->fs_info->chunk_mutex);
> return ret;
> }
prev parent reply other threads:[~2011-04-13 1:19 UTC|newest]
Thread overview: 2+ messages / expand[flat|nested] mbox.gz Atom feed top
2011-04-12 12:55 [PATCH] Btrfs: avoid taking the chunk_mutex in do_chunk_alloc V2 Josef Bacik
2011-04-13 1:19 ` liubo [this message]
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=4DA4FA2B.7060704@cn.fujitsu.com \
--to=liubo2009@cn.fujitsu.com \
--cc=josef@redhat.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 a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).