From: Al Viro <viro@ZenIV.linux.org.uk>
To: Dave Chinner <david@fromorbit.com>
Cc: linux-fsdevel@vger.kernel.org, linux-kernel@vger.kernel.org
Subject: Re: [PATCH 1/8] fs: protect inode->i_state with inode->i_lock
Date: Fri, 29 Oct 2010 05:29:19 +0100 [thread overview]
Message-ID: <20101029042919.GM19804@ZenIV.linux.org.uk> (raw)
In-Reply-To: <1288322620-8566-2-git-send-email-david@fromorbit.com>
On Fri, Oct 29, 2010 at 02:23:33PM +1100, Dave Chinner wrote:
> @@ -293,9 +293,11 @@ static void inode_wait_for_writeback(struct inode *inode)
>
> wqh = bit_waitqueue(&inode->i_state, __I_SYNC);
> while (inode->i_state & I_SYNC) {
> + spin_unlock(&inode->i_lock);
Minor annoyance, but... Let's replace spaces with tab in that while.
> + spin_unlock(&inode->i_lock);
> +
> pages_skipped = wbc->pages_skipped;
> writeback_single_inode(inode, wbc);
Might make sense to lift locking i_lock into callers of
writeback_single_inode() (it has to grab the damn thing as soon as it's
called) and collapse it with spin_unlock() here. Separate patch.
> + spin_unlock(&inode->i_lock);
> /*
> * If the inode was already on b_dirty/b_io/b_more_io, don't
> * reposition it (that would break b_dirty time-ordering).
I'm not sure there's any point in dropping it here, actually.
> + spin_lock(&inode->i_lock);
> inode->i_state = I_FREEING | I_CLEAR;
> + spin_unlock(&inode->i_lock);
_Probably_ not needed here; we already had I_FREEING set by the time
we'd called that and no other thread will modify ->i_state of that
inode at that point. Check for I_CLEAR will be later in the same
thread, so no barriers are needed. Separate patch.
> @@ -552,8 +568,6 @@ int invalidate_inodes(struct super_block *sb)
> */
> list_move(&inode->i_lru, &dispose);
> list_del_init(&inode->i_wb_list);
> - if (!(inode->i_state & (I_DIRTY | I_SYNC)))
> - percpu_counter_dec(&nr_inodes_unused);
> }
> spin_unlock(&inode_lock);
Ho-hum...
I'm not sure we need that list_del_init() here, actually, seeing that
I_FREEING is already set... For later patch, anyway.
> @@ -917,10 +946,12 @@ static struct inode *get_new_inode_fast(struct super_block *sb,
> /* We released the lock, so.. */
> old = find_inode_fast(sb, head, ino);
> if (!old) {
> + spin_lock(&inode->i_lock);
> inode->i_ino = ino;
> + inode->i_state = I_NEW;
> + spin_unlock(&inode->i_lock);
Almost certainly not needed; nobody can find this inode at that point.
> + * wake_up_bit(&inode->i_state, __I_NEW) after removing from the hash list
> + * will DTRT.
Add that i_lock is not regained.
next prev parent reply other threads:[~2010-10-29 4:29 UTC|newest]
Thread overview: 16+ messages / expand[flat|nested] mbox.gz Atom feed top
2010-10-29 3:23 fs: break out inode operations from inode_lock V3 Dave Chinner
2010-10-29 3:23 ` [PATCH 1/8] fs: protect inode->i_state with inode->i_lock Dave Chinner
2010-10-29 4:29 ` Al Viro [this message]
2010-10-29 3:23 ` [PATCH 2/8] fs: factor inode disposal Dave Chinner
2010-10-29 4:45 ` Al Viro
2010-10-29 3:23 ` [PATCH 3/8] fs: Lock the inode LRU list separately Dave Chinner
2010-10-29 3:23 ` [PATCH 4/8] fs: remove inode_lock from iput_final and prune_icache Dave Chinner
2010-10-29 5:14 ` Al Viro
2010-10-29 3:23 ` [PATCH 5/8] fs: move i_sb_list out from under inode_lock Dave Chinner
2010-10-29 3:23 ` [PATCH 6/8] fs: move i_wb_list " Dave Chinner
2010-10-29 3:23 ` [PATCH 7/8] fs: rename inode_lock to inode_hash_lock Dave Chinner
2010-10-29 3:23 ` [PATCH 8/8] fs: Clean up documentation references to inode_lock Dave Chinner
2010-10-29 5:17 ` Al Viro
2010-10-29 7:46 ` Christoph Hellwig
-- strict thread matches above, loose matches on Subject: below --
2010-10-29 8:59 fs: break out inode operations from inode_lock V4 Dave Chinner
2010-10-29 8:59 ` [PATCH 1/8] fs: protect inode->i_state with inode->i_lock Dave Chinner
2011-03-22 11:23 vfs: inode lock breakup Dave Chinner
2011-03-22 11:23 ` [PATCH 1/8] fs: protect inode->i_state with inode->i_lock Dave Chinner
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=20101029042919.GM19804@ZenIV.linux.org.uk \
--to=viro@zeniv.linux.org.uk \
--cc=david@fromorbit.com \
--cc=linux-fsdevel@vger.kernel.org \
--cc=linux-kernel@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).