Date: Tue, 16 Mar 2004 15:39:00 -0800
From: Andrew Morton <akpm@osdl.org>
To: mason@suse.com, daniel@osdl.org, linux-kernel@vger.kernel.org,
       linux-aio@kvack.org
Subject: Re: 2.6.4-mm2
Message-Id: <20040316153900.1e845ba2.akpm@osdl.org>
In-Reply-To: <20040316152843.667a623d.akpm@osdl.org>
References: <20040314172809.31bd72f7.akpm@osdl.org>
	<1079461971.23783.5.camel@ibm-c.pdx.osdl.net>
	<1079474312.4186.927.camel@watt.suse.com>
	<20040316152106.22053934.akpm@osdl.org>
	<20040316152843.667a623d.akpm@osdl.org>
Mime-Version: 1.0
Content-Type: text/plain; charset=US-ASCII
Content-Transfer-Encoding: 7bit
Sender: linux-kernel-owner@vger.kernel.org
Content-Length: 1182
Lines: 28

Andrew Morton <akpm@osdl.org> wrote:
>
> Andrew Morton <akpm@osdl.org> wrote:
> >
> > I'm thinking that the right thing to do here is to change submit_bh()
> > callers and ll_rw_block() to run set_page_writeback(bh->b_page) when they
> > start the buffer writeout and to do the run-around-the-buffer_heads thing
> > at I/O completion.
> 
> A page may have a mix of writeback and dirty+non-writeback buffers.  It
> appears that the page-level writeback code will handle this correctly.  But
> it requires that the page lock be held when we run set_page_writeback(), so
> that tears that.  hmm.

I'll work this out yet.

We change the PageWriteback() predicate to go in and see if any of the
buffer_heads are under I/O too.  And we change wait_on_page_writeback() to
also wait on any buffer_heads.

That means two new silly address_space ops.  Or a new page flag which means
"the thing at ->private is buffer_heads".  Probably the latter.

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/