xfs
[Top] [All Lists]

Re: [PATCH 9/9] Clean up open coded inode dirty checks

To: Joe Perches <joe@xxxxxxxxxxx>
Subject: Re: [PATCH 9/9] Clean up open coded inode dirty checks
From: Jan Engelhardt <jengelh@xxxxxxxxxxxxxxx>
Date: Fri, 23 Nov 2007 21:16:05 +0100 (CET)
Cc: Christoph Hellwig <hch@xxxxxxxxxxxxx>, David Chinner <dgc@xxxxxxx>, xfs-oss <xfs@xxxxxxxxxxx>, lkml <linux-kernel@xxxxxxxxxxxxxxx>
In-reply-to: <1195847251.4930.21.camel@localhost>
References: <20071122004422.GO114266761@xxxxxxx> <20071123180239.GA13229@xxxxxxxxxxxxx> <Pine.LNX.4.64.0711231914300.12552@xxxxxxxxxxxxxxxxxxxxxxxxx> <1195847251.4930.21.camel@localhost>
Sender: xfs-bounce@xxxxxxxxxxx
On Nov 23 2007 11:47, Joe Perches wrote:
>On Fri, 2007-11-23 at 19:16 +0100, Jan Engelhardt wrote:
>> static inline bool xfs_inode_clean(const struct xfs_inode *ip)
>> {
>>      if (ip->i_itemp == NULL)
>>              return true;
>>      if (!(ip->i_itemp->ili_format.ilf_fields & XFS_ILOG_ALL) &&
>>          ip->i_update_core == NULL)
>>              return true;
>>      return false;
>> }
>
>Your code changed the test.

See - the previous cryptic constructs could not even be decoded ;-)

>xfs_inode.i_update_core is an unsigned char.
>
>I believe reordering the tests to avoid a possibly
>unnecessary dereference is better.
>
>       if (ip->i_update_core)
>               return false;
>       if (!ip->i_itemp)
>               return true;
>       return ip->i_itemp->ili_format.ilf_fields & XFS_ILOG_ALL;

Yeah, something like that.

Note: the function SHOULD return bool for this, to quash the
ilf_fields & XFS_ILOG_ALL into 0/1.


<Prev in Thread] Current Thread [Next in Thread>