Linux filesystem development
 help / color / mirror / Atom feed
From: Jan Kara <jack@suse.cz>
To: Christian Brauner <brauner@kernel.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>,
	NeilBrown <neilb@suse.de>, Peter Zijlstra <peterz@infradead.org>,
	Ingo Molnar <mingo@redhat.com>, Jeff Layton <jlayton@kernel.org>,
	Jan Kara <jack@suse.cz>,
	linux-fsdevel@vger.kernel.org
Subject: Re: [PATCH v3 1/6] fs: add i_state helpers
Date: Thu, 5 Sep 2024 18:01:44 +0200	[thread overview]
Message-ID: <20240905160144.w5evqmfzztdxedba@quack3> (raw)
In-Reply-To: <20240823-work-i_state-v3-1-5cd5fd207a57@kernel.org>

On Fri 23-08-24 14:47:35, Christian Brauner wrote:
> The i_state member is an unsigned long so that it can be used with the
> wait bit infrastructure which expects unsigned long. This wastes 4 bytes
> which we're unlikely to ever use. Switch to using the var event wait
> mechanism using the address of the bit. Thanks to Linus for the address
> idea.
> 
> Signed-off-by: Christian Brauner <brauner@kernel.org>

Looks good. Interesting trick with the wakeup address. I'm not sure we
100% need it since e.g. __I_NEW waiters are unlikely to combine with
__I_SYNC waiters so sharing a waitqueue should not cause big issues but I
guess we'll deal with that if we run out of waiting bits with the current
scheme. Feel free to add:

Reviewed-by: Jan Kara <jack@suse.cz>

								Honza

> ---
>  fs/inode.c         | 11 +++++++++++
>  include/linux/fs.h | 15 +++++++++++++++
>  2 files changed, 26 insertions(+)
> 
> diff --git a/fs/inode.c b/fs/inode.c
> index 154f8689457f..877c64a1bf63 100644
> --- a/fs/inode.c
> +++ b/fs/inode.c
> @@ -472,6 +472,17 @@ static void __inode_add_lru(struct inode *inode, bool rotate)
>  		inode->i_state |= I_REFERENCED;
>  }
>  
> +struct wait_queue_head *inode_bit_waitqueue(struct wait_bit_queue_entry *wqe,
> +					    struct inode *inode, u32 bit)
> +{
> +        void *bit_address;
> +
> +        bit_address = inode_state_wait_address(inode, bit);
> +        init_wait_var_entry(wqe, bit_address, 0);
> +        return __var_waitqueue(bit_address);
> +}
> +EXPORT_SYMBOL(inode_bit_waitqueue);
> +
>  /*
>   * Add inode to LRU if needed (inode is unused and clean).
>   *
> diff --git a/include/linux/fs.h b/include/linux/fs.h
> index 23e7d46b818a..1d895b8cb801 100644
> --- a/include/linux/fs.h
> +++ b/include/linux/fs.h
> @@ -744,6 +744,21 @@ struct inode {
>  	void			*i_private; /* fs or device private pointer */
>  } __randomize_layout;
>  
> +/*
> + * Get bit address from inode->i_state to use with wait_var_event()
> + * infrastructre.
> + */
> +#define inode_state_wait_address(inode, bit) ((char *)&(inode)->i_state + (bit))
> +
> +struct wait_queue_head *inode_bit_waitqueue(struct wait_bit_queue_entry *wqe,
> +					    struct inode *inode, u32 bit);
> +
> +static inline void inode_wake_up_bit(struct inode *inode, u32 bit)
> +{
> +	/* Caller is responsible for correct memory barriers. */
> +	wake_up_var(inode_state_wait_address(inode, bit));
> +}
> +
>  struct timespec64 timestamp_truncate(struct timespec64 t, struct inode *inode);
>  
>  static inline unsigned int i_blocksize(const struct inode *node)
> 
> -- 
> 2.43.0
> 
-- 
Jan Kara <jack@suse.com>
SUSE Labs, CR

  reply	other threads:[~2024-09-05 16:01 UTC|newest]

Thread overview: 49+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2024-08-21 15:47 [PATCH RFC v2 0/6] inode: turn i_state into u32 Christian Brauner
2024-08-21 15:47 ` [PATCH RFC v2 1/6] fs: add i_state helpers Christian Brauner
2024-08-21 22:12   ` NeilBrown
2024-08-21 22:47     ` Linus Torvalds
2024-08-21 23:34     ` Dave Chinner
2024-08-23  0:08       ` NeilBrown
2024-08-22  8:27     ` Christian Brauner
2024-08-22  8:37       ` Linus Torvalds
2024-08-23  0:14       ` NeilBrown
2024-08-23  2:52         ` Linus Torvalds
2024-08-23  3:05           ` Linus Torvalds
2024-08-23  3:44             ` Linus Torvalds
2024-08-23  5:01           ` NeilBrown
2024-08-23 12:47           ` [PATCH v3 0/6] inode: turn i_state into u32 Christian Brauner
2024-08-23 12:47             ` [PATCH v3 1/6] fs: add i_state helpers Christian Brauner
2024-09-05 16:01               ` Jan Kara [this message]
2024-08-23 12:47             ` [PATCH v3 2/6] fs: reorder i_state bits Christian Brauner
2024-09-05 16:02               ` Jan Kara
2024-08-23 12:47             ` [PATCH v3 3/6] inode: port __I_SYNC to var event Christian Brauner
2024-09-05 16:02               ` Jan Kara
2024-08-23 12:47             ` [PATCH v3 4/6] inode: port __I_NEW " Christian Brauner
2024-09-06 13:30               ` Jan Kara
2024-08-23 12:47             ` [PATCH v3 5/6] inode: port __I_LRU_ISOLATING " Christian Brauner
2024-09-09  7:35               ` Jan Kara
2024-08-23 12:47             ` [PATCH v3 6/6] inode: make i_state a u32 Christian Brauner
2024-09-09  7:35               ` Jan Kara
2024-08-23 15:06             ` [PATCH v3 0/6] inode: turn i_state into u32 Josef Bacik
2024-08-21 22:28   ` [PATCH RFC v2 1/6] fs: add i_state helpers Dave Chinner
2024-08-21 22:53     ` Linus Torvalds
2024-08-21 15:47 ` [PATCH RFC v2 2/6] fs: reorder i_state bits Christian Brauner
2024-08-21 15:47 ` [PATCH RFC v2 3/6] writeback: port __I_SYNC to var event Christian Brauner
2024-08-21 15:47 ` [PATCH RFC v2 4/6] inode: port __I_NEW " Christian Brauner
2024-08-23  0:31   ` NeilBrown
2024-08-23  8:20     ` Christian Brauner
2024-08-23 11:07       ` Christian Brauner
2024-08-21 15:47 ` [PATCH RFC v2 5/6] inode: port __I_LRU_ISOLATING " Christian Brauner
2024-08-21 19:41   ` Jeff Layton
2024-08-22  8:53     ` Christian Brauner
2024-08-22  9:48       ` Mateusz Guzik
2024-08-22 11:10         ` Christian Brauner
2024-08-22 12:46           ` Mateusz Guzik
2024-08-23  0:36   ` NeilBrown
2024-08-23  2:24     ` Linus Torvalds
2024-08-23  8:29       ` Christian Brauner
2024-08-21 15:47 ` [PATCH RFC v2 6/6] inode: make i_state a u32 Christian Brauner
2024-08-21 21:03   ` Andreas Dilger
2024-08-22  8:31     ` Christian Brauner
2024-08-21 19:15 ` [PATCH RFC v2 0/6] inode: turn i_state into u32 Josef Bacik
2024-08-21 19:42 ` Jeff Layton

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=20240905160144.w5evqmfzztdxedba@quack3 \
    --to=jack@suse.cz \
    --cc=brauner@kernel.org \
    --cc=jlayton@kernel.org \
    --cc=linux-fsdevel@vger.kernel.org \
    --cc=mingo@redhat.com \
    --cc=neilb@suse.de \
    --cc=peterz@infradead.org \
    --cc=torvalds@linux-foundation.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