Linux filesystem development
 help / color / mirror / Atom feed
From: Al Viro <viro@zeniv.linux.org.uk>
To: Christoph Hellwig <hch@lst.de>
Cc: linux-fsdevel@vger.kernel.org
Subject: Re: [RFC] weird use of dget_parent() in xfs_filestream_get_parent()
Date: Tue, 25 Mar 2025 05:50:56 +0000	[thread overview]
Message-ID: <20250325055056.GK2023217@ZenIV> (raw)
In-Reply-To: <20250324215215.GJ2023217@ZenIV>

On Mon, Mar 24, 2025 at 09:52:15PM +0000, Al Viro wrote:
> Function in question:
> 
> static struct xfs_inode *
> xfs_filestream_get_parent(
>         struct xfs_inode        *ip)
> {
>         struct inode            *inode = VFS_I(ip), *dir = NULL;
>         struct dentry           *dentry, *parent;
> 
>         dentry = d_find_alias(inode);
>         if (!dentry)
>                 goto out;
> 
>         parent = dget_parent(dentry);
>         if (!parent)
>                 goto out_dput;
> 
>         dir = igrab(d_inode(parent));
>         dput(parent);
> 
> out_dput:
>         dput(dentry);
> out:
>         return dir ? XFS_I(dir) : NULL;
> }
> 
> 1) dget_parent(dentry) never returns NULL; if you have IS_ROOT(dentry) you
> get an equivalent of dget(dentry).  What do you want returned in that
> case?
> 
> 2) why bother with dget_parent() in the first place?  What's wrong with
> 	spin_lock(&dentry->d_lock); // stabilizes ->d_parent
>         dir = igrab(dentry->d_parent->d_inode);
> 	spin_unlock(&dentry->d_lock);
> 
> or, if you intended NULL for root, 
> 	spin_lock(&dentry->d_lock); // stabilizes ->d_parent
> 	if (!IS_ROOT(dentry))
> 		dir = igrab(dentry->d_parent->d_inode);
> 	spin_unlock(&dentry->d_lock);
> 
> 
> Is there anything subtle I'm missing here?

For (2) there is - ->i_lock nests outside of ->d_lock, so igrab() under
any ->d_lock is not valid.

For (1) we provably can't get NULL - there are two returns in dget_parent():
                if (!read_seqcount_retry(&dentry->d_seq, seq))
                        return ret;
and
        spin_unlock(&ret->d_lock);
        return ret;
and neither can return a NULL - we'd oops immediately prior to getting
there.

The same goes for xrep_findparent_from_dcache() - in
        parent = dget_parent(dentry);
	if (!parent)
		goto out_dput;

	ASSERT(parent->d_sb == sc->ip->i_mount->m_super);

	pip = igrab(d_inode(parent));
	dput(parent);
we need to use dget_parent(), but we really can't get a NULL from
it.

  reply	other threads:[~2025-03-25  5:50 UTC|newest]

Thread overview: 3+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2025-03-24 21:52 [RFC] weird use of dget_parent() in xfs_filestream_get_parent() Al Viro
2025-03-25  5:50 ` Al Viro [this message]
2025-04-08  6:26 ` Christoph Hellwig

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=20250325055056.GK2023217@ZenIV \
    --to=viro@zeniv.linux.org.uk \
    --cc=hch@lst.de \
    --cc=linux-fsdevel@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