xfs
[Top] [All Lists]

Re: 2.6.32.3 x86_64 - XFS hangs, all I/O to D-state bug (again)

To: Justin Piszcz <jpiszcz@xxxxxxxxxxxxxxx>
Subject: Re: 2.6.32.3 x86_64 - XFS hangs, all I/O to D-state bug (again)
From: Christoph Hellwig <hch@xxxxxxxxxxxxx>
Date: Thu, 11 Feb 2010 09:56:32 -0500
Cc: linux-kernel@xxxxxxxxxxxxxxx, linux-raid@xxxxxxxxxxxxxxx, xfs@xxxxxxxxxxx, Alan Piszcz <ap@xxxxxxxxxxxxx>
In-reply-to: <20100211145123.GA13114@xxxxxxxxxxxxx>
References: <alpine.DEB.2.00.1002110903080.25392@xxxxxxxxxxxxxxxx> <20100211145123.GA13114@xxxxxxxxxxxxx>
User-agent: Mutt/1.5.19 (2009-01-05)
On Thu, Feb 11, 2010 at 09:51:23AM -0500, Christoph Hellwig wrote:
> Can you check if this still happens with the patch below?

Or the 2.6.32 variant of it:

Index: xfs/fs/xfs/linux-2.6/xfs_buf.c
===================================================================
--- xfs.orig/fs/xfs/linux-2.6/xfs_buf.c 2010-02-11 15:52:14.442003504 +0100
+++ xfs/fs/xfs/linux-2.6/xfs_buf.c      2010-02-11 15:52:33.172024387 +0100
@@ -1182,8 +1182,7 @@ _xfs_buf_ioapply(
                bp->b_flags &= ~_XBF_RUN_QUEUES;
                rw = (bp->b_flags & XBF_WRITE) ? WRITE_SYNC : READ_SYNC;
        } else {
-               rw = (bp->b_flags & XBF_WRITE) ? WRITE :
-                    (bp->b_flags & XBF_READ_AHEAD) ? READA : READ;
+               rw = (bp->b_flags & XBF_WRITE) ? WRITE : READ;
        }
 
        /* Special code path for reading a sub page size buffer in --

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