[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <4E5D210D.3040608@oracle.com>
Date: Tue, 30 Aug 2011 10:42:37 -0700
From: Sunil Mushran <sunil.mushran@...cle.com>
To: Marco Stornelli <marco.stornelli@...il.com>
CC: Zach Brown <zab@...bo.net>, Dave Chinner <david@...morbit.com>,
Andreas Dilger <adilger@...ger.ca>,
Christoph Hellwig <hch@...radead.org>,
Josef Bacik <josef@...hat.com>, linux-fsdevel@...r.kernel.org,
linux-kernel@...r.kernel.org, linux-btrfs@...r.kernel.org,
xfs@....sgi.com, viro@...iv.linux.org.uk, dchinner@...hat.com
Subject: Re: [PATCH] xfstests 255: add a seek_data/seek_hole tester
On 08/27/2011 01:30 AM, Marco Stornelli wrote:
> Il 26/08/2011 16:41, Zach Brown ha scritto:
>>>> Hole: a range of the file that contains no data or is made up
>>>> entirely of NULL (zero) data. Holes include preallocated ranges of
>>>> files that have not had actual data written to them.
>>
>>> No for me. A hole is made up of zero data? It's a strange definition
>>> for me.
>>
>> It's a very natural definition for me. It mirrors the behaviour of
>> read() of sparse data inside i_size that file system authors already
>> have to consider.
>>
>> It's also a reminder for people that this interface is about avoiding
>> reading zeros. Systems that track contents can do this for files that
>> had tons of zeros written. The data is there but the app is
>> specifically asking us to skip it by using SEEK_DATA.
>>
>> - z
>>
>
> I think we need to consider a hole and "data not present/not written yet" as two different cases even they are related. For example, if I do an fallocate without keep size option, then I do a read, I have the same behavior of sparse data inside i_size, but the blocks are allocated so no sparse data in this case. Simply there are no difference from app point of view.
Exactly. That's why seek_hole should identify them alike, if possible.
But that should not be a requirement because the sole aim here is
to improve performance. Identifying a hole as data is not the end of
the world. In some cases it may be more efficient. We just have to
ensure that we don't identify data as a hole.
BTW, we still have the fiemap interface that allows users to identify
unwritten extents, etc. Use that if you want that kind of detail.
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
Powered by blists - more mailing lists