Date: Tue, 29 Sep 2009 16:57:58 -0700
From: Andrew Morton <akpm@linux-foundation.org>
To: Sage Weil <sage@newdream.net>
Cc: linux-fsdevel@vger.kernel.org, linux-kernel@vger.kernel.org,
       yehuda@newdream.net, sage@newdream.net
Subject: Re: [PATCH 03/21] ceph: client types
Message-Id: <20090929165758.1b7df6c9.akpm@linux-foundation.org>
In-Reply-To: <1253641129-28434-4-git-send-email-sage@newdream.net>
References: <1253641129-28434-1-git-send-email-sage@newdream.net>
	<1253641129-28434-2-git-send-email-sage@newdream.net>
	<1253641129-28434-3-git-send-email-sage@newdream.net>
	<1253641129-28434-4-git-send-email-sage@newdream.net>
Mime-Version: 1.0
Content-Type: text/plain; charset=US-ASCII
Content-Transfer-Encoding: 7bit
Sender: linux-kernel-owner@vger.kernel.org
Content-Length: 1955
Lines: 72

On Tue, 22 Sep 2009 10:38:31 -0700
Sage Weil <sage@newdream.net> wrote:

> We first define constants, types, and prototypes for the kernel client
> proper.
> 
> A few subsystems are defined separately later: the MDS, OSD, and
> monitor clients, and the messaging layer.
> 
>
> ...
>
> +static inline bool ceph_i_test(struct inode *inode, unsigned mask)
> +{
> +	struct ceph_inode_info *ci = ceph_inode(inode);
> +	bool r;
> +
> +	spin_lock(&inode->i_lock);
> +	r = (ci->i_ceph_flags & mask) == mask;
> +	spin_unlock(&inode->i_lock);
> +	return r;
> +}

I'm not sure that the locking in ceph_i_test() makes much sense.  The
condition we just checked could change one femtosecond after we dropped
the lock, so why take the lock?

> +
> +/* find a specific frag @f */
> +static inline struct ceph_inode_frag *
> +__ceph_find_frag(struct ceph_inode_info *ci, u32 f)
> +{
> +	struct rb_node *n = ci->i_fragtree.rb_node;
> +
> +	while (n) {
> +		struct ceph_inode_frag *frag =
> +			rb_entry(n, struct ceph_inode_frag, node);
> +		int c = frag_compare(f, frag->frag);
> +		if (c < 0)
> +			n = n->rb_left;
> +		else if (c > 0)
> +			n = n->rb_right;
> +		else
> +			return frag;
> +	}
> +	return NULL;
> +}

Please review the entire fs and verify that all inlining decisions were
appropriate.  Unless this function has a single call site, this
decision wasn't appropriate.

And if it _does_ have a single callsite, it should be defined in the
relevant .c file, not in .h

> +/*
> + * choose fragment for value @v.  copy frag content to pfrag, if leaf
> + * exists
> + */
> +extern u32 ceph_choose_frag(struct ceph_inode_info *ci, u32 v,
> +			    struct ceph_inode_frag *pfrag,
> +			    int *found);
> +
>
> ...
>

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/