xfs
[Top] [All Lists]

Re: [PATCH] always set a/c/mtime through ->setattr

To: hch@xxxxxx
Subject: Re: [PATCH] always set a/c/mtime through ->setattr
From: Miklos Szeredi <miklos@xxxxxxxxxx>
Date: Thu, 22 May 2008 20:10:50 +0200
Cc: miklos@xxxxxxxxxx, hch@xxxxxx, viro@xxxxxxxxxxxxxxxxxx, linux-fsdevel@xxxxxxxxxxxxxxx, xfs@xxxxxxxxxxx, Artem.Bityutskiy@xxxxxxxxx
In-reply-to: <20080520083351.GA14826@xxxxxx> (message from Christoph Hellwig on Tue, 20 May 2008 10:33:51 +0200)
References: <20080520060838.GA6436@xxxxxx> <E1JyMSu-0001au-96@xxxxxxxxxxxxxxxxxxx> <20080520083351.GA14826@xxxxxx>
Sender: xfs-bounce@xxxxxxxxxxx
> All major disk or in-memory filesystems except for XFS just pass down
> ATTR_*TIME requests to inode_setattr which is not more than just
> dirtying the inode.  NFS and CIFS set S_NOCMTIME so they're not affected
> by this at all.

I've checked, and relatively few filesystems set S_NOCMTIME:

 cifs, fuse, nfs

But there are quite a few others which don't call inode_setattr (which
means that the unchanged time optimization is lost), or which do
something possibly slow in their ->setattr():

 adfs, 9p, afs, coda, gfs2 ...

just to name a few at the start of the alphabet.

So it looks to me as this could cause some unintended performance
regressions in these filesystems.

Miklos


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