Re: [PATCH v3] xen-blk(front|back): Handle large physical sector disks

[Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>]

On Tue, May 28, 2013 at 02:55:22PM +0200, Stefan Bader wrote:
> On 28.05.2013 14:47, Konrad Rzeszutek Wilk wrote:
> > On Wed, May 15, 2013 at 05:02:35PM +0200, Stefan Bader wrote:
> >> Changed the xenbus_gather but kept the xenbus_dev_error (without being fatal) as
> >> that seems closer to the style of the other calls.
> > 
> > Hm, could you rebase this on top of stable/for-jens-3.10 please?
> > 
> > I get this:
> > 
> > patching file drivers/block/xen-blkback/xenbus.c
> > Hunk #1 FAILED at 704.
> > 1 out of 1 hunk FAILED -- saving rejects to file drivers/block/xen-blkback/xenbus.c.rej
> > patching file drivers/block/xen-blkfront.c
> > Hunk #1 FAILED at 542.
> > patch: **** malformed patch at line 133: 16
> > 
> > 
> 
> I rather suspect another TB fail. Lets try the same on-top-of-Linus and xen-git
> as attachements...

So better, but stable/for-jens-3.10 has some extra changes.

patch -p1 < /tmp/0001-xen-blk-Use-physical-sector-size-for-setup.patch 
patching file drivers/block/xen-blkback/xenbus.c
Hunk #1 succeeded at 782 (offset 78 lines).
patching file drivers/block/xen-blkfront.c
Hunk #1 FAILED at 542.
Hunk #2 succeeded at 629 (offset 65 lines).
Hunk #3 succeeded at 736 (offset 68 lines).
Hunk #4 FAILED at 736.
Hunk #5 succeeded at 1698 (offset 301 lines).
Hunk #6 succeeded at 1748 (offset 308 lines).
Hunk #7 succeeded at 1811 with fuzz 2 (offset 315 lines).
2 out of 7 hunks FAILED -- saving rejects to file drivers/block/xen-blkfront.c.rej

> 
> -Stefan
> 

> From e1bc2a0027c20679bfc81beb3f2e888b283a44f2 Mon Sep 17 00:00:00 2001
> From: Stefan Bader <stefan.bader@canonical.com>
> Date: Mon, 13 May 2013 16:28:15 +0200
> Subject: [PATCH] xen/blk: Use physical sector size for setup
> 
> Currently xen-blkback passes the logical sector size over xenbus and
> xen-blkfront sets up the paravirt disk with that logical block size.
> But newer drives usually have the logical sector size set to 512 for
> compatibility reasons and would show the actual sector size only in
> physical sector size.
> This results in the device being partitioned and accessed in dom0 with
> the correct sector size, but the guest thinks 512 bytes is the correct
> block size. And that results in poor performance.
> 
> To fix this, blkback gets modified to pass also physical-sector-size
> over xenbus and blkfront to use both values to set up the paravirt
> disk. I did not just change the passed in sector-size because I am
> not sure having a bigger logical sector size than the physical one
> is valid (and that would happen if a newer dom0 kernel hits an older
> domU kernel). Also this way a domU set up before should still be
> accessible (just some tools might detect the unaligned setup).
> 
> [v2: Make xenbus write failure non-fatal]
> [v3: Use xenbus_scanf instead of xenbus_gather]
> 
> Signed-off-by: Stefan Bader <stefan.bader@canonical.com>
> ---
>  drivers/block/xen-blkback/xenbus.c |    5 +++++
>  drivers/block/xen-blkfront.c       |   23 +++++++++++++++++++----
>  2 files changed, 24 insertions(+), 4 deletions(-)
> 
> diff --git a/drivers/block/xen-blkback/xenbus.c b/drivers/block/xen-blkback/xenbus.c
> index 8bfd1bc..25463e3 100644
> --- a/drivers/block/xen-blkback/xenbus.c
> +++ b/drivers/block/xen-blkback/xenbus.c
> @@ -704,6 +704,11 @@ again:
>  				 dev->nodename);
>  		goto abort;
>  	}
> +	err = xenbus_printf(xbt, dev->nodename, "physical-sector-size", "%u",
> +			    bdev_physical_block_size(be->blkif->vbd.bdev));
> +	if (err)
> +		xenbus_dev_error(dev, err, "writing %s/physical-sector-size",
> +				 dev->nodename);
>  
>  	err = xenbus_transaction_end(xbt, 0);
>  	if (err == -EAGAIN)
> diff --git a/drivers/block/xen-blkfront.c b/drivers/block/xen-blkfront.c
> index d89ef86..33b5618 100644
> --- a/drivers/block/xen-blkfront.c
> +++ b/drivers/block/xen-blkfront.c
> @@ -542,7 +542,8 @@ wait:
>  		flush_requests(info);
>  }
>  
> -static int xlvbd_init_blk_queue(struct gendisk *gd, u16 sector_size)
> +static int xlvbd_init_blk_queue(struct gendisk *gd, u16 sector_size,
> +				unsigned int physical_sector_size)
>  {
>  	struct request_queue *rq;
>  	struct blkfront_info *info = gd->private_data;
> @@ -564,6 +565,7 @@ static int xlvbd_init_blk_queue(struct gendisk *gd, u16 sector_size)
>  
>  	/* Hard sector size and max sectors impersonate the equiv. hardware. */
>  	blk_queue_logical_block_size(rq, sector_size);
> +	blk_queue_physical_block_size(rq, physical_sector_size);
>  	blk_queue_max_hw_sectors(rq, 512);
>  
>  	/* Each segment in a request is up to an aligned page in size. */
> @@ -667,7 +669,8 @@ static char *encode_disk_name(char *ptr, unsigned int n)
>  
>  static int xlvbd_alloc_gendisk(blkif_sector_t capacity,
>  			       struct blkfront_info *info,
> -			       u16 vdisk_info, u16 sector_size)
> +			       u16 vdisk_info, u16 sector_size,
> +			       unsigned int physical_sector_size)
>  {
>  	struct gendisk *gd;
>  	int nr_minors = 1;
> @@ -734,7 +737,7 @@ static int xlvbd_alloc_gendisk(blkif_sector_t capacity,
>  	gd->driverfs_dev = &(info->xbdev->dev);
>  	set_capacity(gd, capacity);
>  
> -	if (xlvbd_init_blk_queue(gd, sector_size)) {
> +	if (xlvbd_init_blk_queue(gd, sector_size, physical_sector_size)) {
>  		del_gendisk(gd);
>  		goto release;
>  	}
> @@ -1395,6 +1398,7 @@ static void blkfront_connect(struct blkfront_info *info)
>  {
>  	unsigned long long sectors;
>  	unsigned long sector_size;
> +	unsigned int physical_sector_size;
>  	unsigned int binfo;
>  	int err;
>  	int barrier, flush, discard, persistent;
> @@ -1437,6 +1441,16 @@ static void blkfront_connect(struct blkfront_info *info)
>  		return;
>  	}
>  
> +	/*
> +	 * physcial-sector-size is a newer field, so old backends may not
> +	 * provide this. Assume physical sector size to be the same as
> +	 * sector_size in that case.
> +	 */
> +	err = xenbus_scanf(XBT_NIL, info->xbdev->otherend,
> +			   "physical-sector-size", "%u", &physical_sector_size);
> +	if (err != 1)
> +		physical_sector_size = sector_size;
> +
>  	info->feature_flush = 0;
>  	info->flush_op = 0;
>  
> @@ -1483,7 +1497,8 @@ static void blkfront_connect(struct blkfront_info *info)
>  	else
>  		info->feature_persistent = persistent;
>  
> -	err = xlvbd_alloc_gendisk(sectors, info, binfo, sector_size);
> +	err = xlvbd_alloc_gendisk(sectors, info, binfo, sector_size,
> +				  physical_sector_size);
>  	if (err) {
>  		xenbus_dev_fatal(info->xbdev, err, "xlvbd_add at %s",
>  				 info->xbdev->otherend);
> -- 
> 1.7.9.5
> 

> From 8d1023ce11b9067346e9794d95b2876d98484f43 Mon Sep 17 00:00:00 2001
> From: Stefan Bader <stefan.bader@canonical.com>
> Date: Wed, 22 May 2013 15:11:18 +0200
> Subject: [PATCH] blkif.h: Document the physical-sector-size extension
> 
> Signed-off-by: Stefan Bader <stefan.bader@canonical.com>
> ---
>  xen/include/public/io/blkif.h |   14 ++++++++++----
>  1 file changed, 10 insertions(+), 4 deletions(-)
> 
> diff --git a/xen/include/public/io/blkif.h b/xen/include/public/io/blkif.h
> index 97b423b..f7c3366 100644
> --- a/xen/include/public/io/blkif.h
> +++ b/xen/include/public/io/blkif.h
> @@ -208,12 +208,17 @@
>   * sector-size
>   *      Values:         <uint32_t>
>   *
> - *      The native sector size, in bytes, of the backend device.
> + *      The logical sector size, in bytes, of the backend device.
> + *
> + * physical-sector-size
> + *      Values:         <unsigned int>
> + *
> + *      The physical sector size, in bytes, of the backend device.
>   *
>   * sectors
>   *      Values:         <uint64_t>
>   *
> - *      The size of the backend device, expressed in units of its native
> + *      The size of the backend device, expressed in units of its logical
>   *      sector size ("sector-size").
>   *
>   *****************************************************************************
> @@ -473,8 +478,9 @@
>   * NB. first_sect and last_sect in blkif_request_segment, as well as
>   * sector_number in blkif_request, are always expressed in 512-byte units.
>   * However they must be properly aligned to the real sector size of the
> - * physical disk, which is reported in the "sector-size" node in the backend
> - * xenbus info. Also the xenbus "sectors" node is expressed in 512-byte units.
> + * physical disk, which is reported in the "physical-sector-size" node in
> + * the backend xenbus info. Also the xenbus "sectors" node is expressed in
> + * 512-byte units.
>   */
>  struct blkif_request_segment {
>      grant_ref_t gref;        /* reference to I/O buffer frame        */
> -- 
> 1.7.9.5
>