[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <alpine.LFD.2.00.1209140919580.2455@dhcp-196-88.bos.redhat.com>
Date: Fri, 14 Sep 2012 09:21:19 -0400 (EDT)
From: Lukáš Czerner <lczerner@...hat.com>
To: Lukáš Czerner <lczerner@...hat.com>
cc: Christoph Hellwig <hch@...radead.org>,
Andrew Morton <akpm@...ux-foundation.org>,
linux-fsdevel@...r.kernel.org, linux-ext4@...r.kernel.org,
tytso@....edu, hughd@...gle.com, linux-mm@...ck.org
Subject: Re: [PATCH 01/15 v2] mm: add invalidatepage_range address space
operation
On Wed, 5 Sep 2012, Lukáš Czerner wrote:
> Date: Wed, 5 Sep 2012 12:42:54 -0400 (EDT)
> From: Lukáš Czerner <lczerner@...hat.com>
> To: Christoph Hellwig <hch@...radead.org>
> Cc: Luk?? Czerner <lczerner@...hat.com>,
> Andrew Morton <akpm@...ux-foundation.org>, linux-fsdevel@...r.kernel.org,
> linux-ext4@...r.kernel.org, tytso@....edu, hughd@...gle.com,
> linux-mm@...ck.org
> Subject: Re: [PATCH 01/15 v2] mm: add invalidatepage_range address space
> operation
>
> On Wed, 5 Sep 2012, Christoph Hellwig wrote:
>
> > Date: Wed, 5 Sep 2012 11:56:48 -0400
> > From: Christoph Hellwig <hch@...radead.org>
> > To: Luk?? Czerner <lczerner@...hat.com>
> > Cc: Andrew Morton <akpm@...ux-foundation.org>, linux-fsdevel@...r.kernel.org,
> > linux-ext4@...r.kernel.org, tytso@....edu, hughd@...gle.com,
> > linux-mm@...ck.org
> > Subject: Re: [PATCH 01/15 v2] mm: add invalidatepage_range address space
> > operation
> >
> > On Wed, Sep 05, 2012 at 10:36:00AM -0400, Luk?? Czerner wrote:
> > > However if we would want to keep ->invalidatepage_range() and
> > > ->invalidatepage() completely separate then we would have to have
> > > separate truncate_inode_pages_range() and truncate_pagecache_range()
> > > as well for the separation to actually matter. And IMO this would be
> > > much worse...
> >
> > What's the problem with simply changing the ->invalidatepage prototype
> > to always pass the range and updating all instances for it?
> >
>
> The problem is that it would require me to implement this
> functionality for _all_ the file systems, because it is not just
> about changing the prototype, but also changing the implementation to
> be able to handle unaligned end of the range. This change would
> involve 20 file systems.
>
> It is not impossible though... so if people think that it's the
> right way to go, then I guess it can be done.
>
> -Lukas
Are there still any objections or comments about this ?
-Lukas
Powered by blists - more mailing lists