xfs
[Top] [All Lists]

Re: [PATCHv2 5/5] xfs: kill off l_sectbb_mask

To: Alex Elder <aelder@xxxxxxx>
Subject: Re: [PATCHv2 5/5] xfs: kill off l_sectbb_mask
From: Christoph Hellwig <hch@xxxxxxxxxxxxx>
Date: Sun, 18 Apr 2010 13:21:21 -0400
Cc: xfs@xxxxxxxxxxx
In-reply-to: <201004162054.o3GKs54f025204@xxxxxxxxxxxxxxxxxxxxxx>
References: <201004162054.o3GKs54f025204@xxxxxxxxxxxxxxxxxxxxxx>
User-agent: Mutt/1.5.19 (2009-01-05)
On Fri, Apr 16, 2010 at 03:54:05PM -0500, Alex Elder wrote:
> -     uint                    l_sectBBsize;   /* sector size in BBs */
> -     uint                    l_sectbb_mask;  /* sector size (in BBs)
> -                                              * alignment mask */
> +     uint                    l_sectBBsize;   /* sector size in BBs (2^n) */

The 2^n comment doesn't make too much sense to me.

But in general looks good, nice cleanup.


Signed-off-by: Christoph Hellwig <hch@xxxxxx>

> @@ -128,14 +132,14 @@ xlog_align(
>       int             nbblks,
>       xfs_buf_t       *bp)
>  {
> +     xfs_daddr_t     offset;
>       xfs_caddr_t     ptr;
>  
> -     if (log->l_sectBBsize == 1)
> -             return XFS_BUF_PTR(bp);
> +     offset = blk_no & ((xfs_daddr_t) log->l_sectBBsize - 1);
> +     ptr = XFS_BUF_PTR(bp) + BBTOB(offset);
> +
> +     ASSERT(ptr + BBTOB(nbblks) <= XFS_BUF_PTR(bp) + XFS_BUF_SIZE(bp));

And Dave's version of this assert definitively makes a lot more sense,
this is one while evolved from the previous one is rather ugly.

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