[Top] [All Lists]

Re: spurious -ENOSPC on XFS

To: Christoph Hellwig <hch@xxxxxxxxxxxxx>
Subject: Re: spurious -ENOSPC on XFS
From: Lachlan McIlroy <lachlan@xxxxxxx>
Date: Tue, 13 Jan 2009 16:58:01 +1100
Cc: Mikulas Patocka <mpatocka@xxxxxxxxxx>, linux-kernel@xxxxxxxxxxxxxxx, xfs@xxxxxxxxxxx
In-reply-to: <20090112151133.GA24852@xxxxxxxxxxxxx>
Organization: SGI
References: <Pine.LNX.4.64.0901120509550.11089@xxxxxxxxxxxxxxxxxxxxxxxxxxx> <20090112151133.GA24852@xxxxxxxxxxxxx>
Reply-to: lachlan@xxxxxxx
User-agent: Thunderbird (X11/20081209)
Christoph Hellwig wrote:
On Mon, Jan 12, 2009 at 06:14:36AM -0500, Mikulas Patocka wrote:

I discovered a bug in XFS in delayed allocation.

When you take a small partition (52MB in my case) and copy many small files on it (source code) that barely fits there, you get -ENOSPC. Then sync the partition, some free space pops up, click "retry" in MC an the copy continues. They you get again -ENOSPC, you must sync, click "retry" and go on. And so on few times until the source code finally fits on the XFS partition.

This misbehavior is apparently caused by delayed allocation, delayed allocation does not exactly know how much space will be occupied by data, so it makes some upper bound guess. Because free space count is only a guess, not the actual data being consumed, XFS should not return -ENOSPC on behalf of it. When the free space overflows, XFS should sync itself, retry allocation and only return -ENOSPC if it fails the second time, after the sync.
This sounds like a problem with speculative allocation - delayed allocations
beyond eof.  Even if we write a small file, say 4k, a 64k chunk of delayed
allocation will be credited to the file.  If we extend the file we use more
of that chunk instead of allocating more.  When the file is closed and the
last inode reference is dropped the excess delayed allocations are released
back to freespace - this must be happening during the sync.

This looks a lot like: http://oss.sgi.com/bugzilla/show_bug.cgi?id=724
It's on my short-term todo list to turn the testcase in that entry
into a proper xfsqa testcase and followup on the investigation by
Dave and Eric.

xfs mailing list

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