[Top] [All Lists]


To: Nathan Scott <nathans@xxxxxxx>
Subject: Re: XFS Writes: IOLOCK_EXCL
From: Steve Lord <lord@xxxxxxx>
Date: Thu, 26 Feb 2004 07:27:02 -0600
Cc: Alex Wun <alexwun@xxxxxxxxxxxxx>, linux-xfs@xxxxxxxxxxx
In-reply-to: <20040226040651.GC1177@frodo>
References: <Pine.LNX.4.44.0402240914430.12803-100000@xxxxxxxxxxxxxxxxxxxxxx> <01fa01c3fafa$7c3482d0$9002a8c0@xxxxxxxxxxxxx> <20040225062907.GB1832@frodo> <403CAA00.20502@xxxxxxx> <20040226040651.GC1177@frodo>
Sender: linux-xfs-bounce@xxxxxxxxxxx
User-agent: Mozilla Thunderbird 0.5 (X11/20040208)
Nathan Scott wrote:

I think its always valid to do this (after auditing the callers),
but doesn't seem to do "general speedup" - the "usual suspects"
benchmarks don't show any improvements anyway.  Alex, what was
your workload?  (you didn't get back to me yet).


I missed the i_sem in nfs, did not look high enough up the stack.
Possibly the double flush in there was catching some extra data,
which considering we do this under the i_sem is somewhat unlikely.

The change would however allow reads to proceed across flushes
as the fsync_datawait which matters is now the one in the
fsync caller which is not holding any lock a read cares about.


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