Received-SPF: pass (google.com: domain of linux-kernel+bounces-197090-linux.lists.archive=gmail.com@vger.kernel.org designates 2604:1380:45d1:ec00::1 as permitted sender) client-ip=2604:1380:45d1:ec00::1;
Date: Fri, 31 May 2024 11:44:57 -0400
From: Brian Foster <bfoster@redhat.com>
To: Christoph Hellwig <hch@infradead.org>
Cc: "Darrick J. Wong" <djwong@kernel.org>,
	Zhang Yi <yi.zhang@huaweicloud.com>, linux-xfs@vger.kernel.org,
	linux-fsdevel@vger.kernel.org, linux-kernel@vger.kernel.org,
	brauner@kernel.org, david@fromorbit.com, chandanbabu@kernel.org,
	jack@suse.cz, willy@infradead.org, yi.zhang@huawei.com,
	chengzhihao1@huawei.com, yukuai3@huawei.com
Subject: Re: [RFC PATCH v4 1/8] iomap: zeroing needs to be pagecache aware
Message-ID: <ZlnweWTV4Y5STK-q@bfoster>
References: <20240529095206.2568162-1-yi.zhang@huaweicloud.com>
 <20240529095206.2568162-2-yi.zhang@huaweicloud.com>
 <ZlnMfSJcm5k6Dg_e@infradead.org>
 <20240531140358.GF52987@frogsfrogsfrogs>
 <ZlnZMiBJ6Fapor5G@infradead.org>
Precedence: bulk
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <ZlnZMiBJ6Fapor5G@infradead.org>

On Fri, May 31, 2024 at 07:05:38AM -0700, Christoph Hellwig wrote:
> On Fri, May 31, 2024 at 07:03:58AM -0700, Darrick J. Wong wrote:
> > > +			/*
> > > +			 * XXX: It would be nice if we could get the offset of
> > > +			 * the next entry in the pagecache so that we don't have
> > > +			 * to iterate one page at a time here.
> > > +			 */
> > > +			offset = offset_in_page(pos);
> > > +			if (bytes > PAGE_SIZE - offset)
> > > +				bytes = PAGE_SIZE - offset;
> > 
> > Why is it PAGE_SIZE here and not folio_size() like below?
> > 
> > (I know you're just copying the existing code; I'm merely wondering if
> > this is some minor bug.)
> 
> See the comment just above :)
> 
> 

FWIW, something like the following is pretty slow with the current
implementation on a quick test:

  xfs_io -fc "falloc -k 0 1t" -c "pwrite 1000g 4k" <file>

.. so I'd think you'd want some kind of data seek or something to more
efficiently process the range.

Brian