Re: [Ocfs2-devel] [PATCH v3 06/10] fsdax: Add dax_iomap_cow_copy() for dax_iomap_zero

From: "ruansy.fnst@fujitsu.com" <ruansy.fnst@fujitsu.com>
To: Ritesh Harjani <ritesh.list@gmail.com>
Cc: "jack@suse.cz" <jack@suse.cz>,
	"darrick.wong@oracle.com" <darrick.wong@oracle.com>,
	"linux-nvdimm@lists.01.org" <linux-nvdimm@lists.01.org>,
	"david@fromorbit.com" <david@fromorbit.com>,
	"linux-kernel@vger.kernel.org" <linux-kernel@vger.kernel.org>,
	"linux-xfs@vger.kernel.org" <linux-xfs@vger.kernel.org>,
	"ocfs2-devel@oss.oracle.com" <ocfs2-devel@oss.oracle.com>,
	"viro@zeniv.linux.org.uk" <viro@zeniv.linux.org.uk>,
	"linux-fsdevel@vger.kernel.org" <linux-fsdevel@vger.kernel.org>,
	"dan.j.williams@intel.com" <dan.j.williams@intel.com>,
	"linux-btrfs@vger.kernel.org" <linux-btrfs@vger.kernel.org>
Subject: Re: [Ocfs2-devel] [PATCH v3 06/10] fsdax: Add dax_iomap_cow_copy() for dax_iomap_zero
Date: Thu, 1 Apr 2021 07:00:55 +0000	[thread overview]
Message-ID: <OSBPR01MB29206BCD28F769EA87E1609BF47B9@OSBPR01MB2920.jpnprd01.prod.outlook.com> (raw)
In-Reply-To: <20210401064506.47pz6u2gegp6s2ky@riteshh-domain>

> -----Original Message-----
> From: Ritesh Harjani <ritesh.list@gmail.com>
> Sent: Thursday, April 1, 2021 2:45 PM
> Subject: Re: [PATCH v3 06/10] fsdax: Add dax_iomap_cow_copy() for
> dax_iomap_zero
> 
> On 21/03/19 09:52AM, Shiyang Ruan wrote:
> > Punch hole on a reflinked file needs dax_copy_edge() too.  Otherwise,
> > data in not aligned area will be not correct.  So, add the srcmap to
> > dax_iomap_zero() and replace memset() as dax_copy_edge().
> >
> > Signed-off-by: Shiyang Ruan <ruansy.fnst@fujitsu.com>
> > ---
> >  fs/dax.c               | 9 +++++++--
> >  fs/iomap/buffered-io.c | 2 +-
> >  include/linux/dax.h    | 3 ++-
> >  3 files changed, 10 insertions(+), 4 deletions(-)
> >
> > diff --git a/fs/dax.c b/fs/dax.c
> > index cfe513eb111e..348297b38f76 100644
> > --- a/fs/dax.c
> > +++ b/fs/dax.c
> > @@ -1174,7 +1174,8 @@ static vm_fault_t dax_pmd_load_hole(struct
> > xa_state *xas, struct vm_fault *vmf,  }  #endif /* CONFIG_FS_DAX_PMD
> > */
> >
> > -s64 dax_iomap_zero(loff_t pos, u64 length, struct iomap *iomap)
> > +s64 dax_iomap_zero(loff_t pos, u64 length, struct iomap *iomap,
> > +		struct iomap *srcmap)
> 
> Do we know why does dax_iomap_zero() operates on PAGE_SIZE range?
> IIUC, dax_zero_page_range() can take nr_pages as a parameter. But we still
> always use one page at a time. Why is that?

I think we can handle more than one page here.  The length can be more than one PAGE_SIZE.

> 
> >  {
> >  	sector_t sector = iomap_sector(iomap, pos & PAGE_MASK);
> >  	pgoff_t pgoff;
> > @@ -1204,7 +1205,11 @@ s64 dax_iomap_zero(loff_t pos, u64 length, struct
> iomap *iomap)
> >  	}
> >
> >  	if (!page_aligned) {
> > -		memset(kaddr + offset, 0, size);
> > +		if (iomap->addr != srcmap->addr)
> > +			dax_iomap_cow_copy(offset, size, PAGE_SIZE, srcmap,
> > +					   kaddr, true);
> > +		else
> > +			memset(kaddr + offset, 0, size);
> >  		dax_flush(iomap->dax_dev, kaddr + offset, size);
> >  	}
> >  	dax_read_unlock(id);
> >
> 
> Maybe the above could be simplified to this?
> 
> 	if (page_aligned) {
> 		rc = dax_zero_page_range(iomap->dax_dev, pgoff, 1);
> 	} else {
> 		rc = dax_direct_access(iomap->dax_dev, pgoff, 1, &kaddr, NULL);

This looks good.  Need add check for rc here.

> 		if (iomap->addr != srcmap->addr)
> 			dax_iomap_cow_copy(offset, size, PAGE_SIZE, srcmap,
> 					   kaddr, true);
> 		else
> 			memset(kaddr + offset, 0, size);
> 		dax_flush(iomap->dax_dev, kaddr + offset, size);
> 	}
> 
> 	dax_read_unlock(id);
> 	return rc < 0 ? rc : size;
> 
> Other than that looks good.
> Feel free to add.
> Reviewed-by: Ritesh Harjani <ritesh.list@gmail.com>
> 

--
Thanks,
Ruan Shiyang.

_______________________________________________
Ocfs2-devel mailing list
Ocfs2-devel@oss.oracle.com
https://oss.oracle.com/mailman/listinfo/ocfs2-devel