Message-ID: <4C16308E.1030305@oracle.com>
Date: Mon, 14 Jun 2010 21:37:18 +0800
From: Tao Ma <tao.ma@oracle.com>
Organization: Oracle Corporation
User-Agent: Thunderbird 2.0.0.24 (Windows/20100228)
MIME-Version: 1.0
To: Dave Chinner <david@fromorbit.com>
CC: xfs@oss.sgi.com, linux-kernel@vger.kernel.org, sandeen@sandeen.net,
       Alex Elder <aelder@sgi.com>, Christoph Hellwig <hch@lst.de>
Subject: Re: [PATCH v2] xfs: Make fiemap works with sparse file.
References: <1276308495-14267-1-git-send-email-tao.ma@oracle.com> <20100614002705.GA6590@dastard> <20100614122912.GD6590@dastard>
In-Reply-To: <20100614122912.GD6590@dastard>
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit
Sender: linux-kernel-owner@vger.kernel.org
Content-Length: 1924
Lines: 46

Hi Dave,

Dave Chinner wrote:
> On Mon, Jun 14, 2010 at 10:27:06AM +1000, Dave Chinner wrote:
>   
>> On Sat, Jun 12, 2010 at 10:08:15AM +0800, Tao Ma wrote:
>>     
>>> The reason is that in xfs_getbmap we will
>>> calculate holes and set it in 'out', while out is malloced by
>>> bmv_count(fi_extent_max+1) which didn't consider holes. So in the
>>> worst case, if 'out' vector looks like
>>> [hole, extent, hole, extent, hole, ... hole, extent, hole],
>>> we will only return half of fi_extent_max extents.
>>>       
>> Right, it's not broken, we simply return less than fi_extent_mex
>> extents when there are holes. I don't see that as a problem as
>> applications have to handle that case anyway, and....
>>
>>     
>>> So in xfs_vn_fiemap, we should consider this worst case. If the
>>> user wants fi_extent_max extents, we need a 'out' with size of
>>> 2 *fi_extent_max + 2(one more the header).
>>>       
>> That's rather dangerous, I think. It relies on other code to catch
>> the buffer overrun that this sets up for fragmented, non-sparse
>> files. Personally I'd much prefer to return fewer extents for sparse
>> files than to add a landmine like this into the kernel code....
>>     
>
> I just had a thought - if you want to avoid holes being reported to
> fiemap, then add a BMV_IF_NO_HOLES flag to xfs_getbmap() and skip
> holes in the mappin gloop when this flag is set. That will make
> fiemap fill in the full number of extents without hacking the
> extent count...
>   
yeah, that should work and I will try to generate a patch for it.
I am not quite familiar with xfs, so please be kind to me if I make some 
stupid mistake in the patch. ;)

Regards,
Tao
--
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/