Re: [PATCH 6/9] xfs: implement the GETFSMAP ioctl

["Darrick J. Wong" <darrick.wong@oracle.com>]

On Wed, Mar 01, 2017 at 11:57:08AM -0500, Brian Foster wrote:
> On Tue, Feb 28, 2017 at 10:46:28AM -0800, Darrick J. Wong wrote:
> > From: Darrick J. Wong <darrick.wong@oracle.com>
> > 
> > Introduce a new ioctl that uses the reverse mapping btree to return
> > information about the physical layout of the filesystem.
> > 
> > Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
> > ---
> > v2: improve comments and refactor common code
> > ---
> 
> Mostly looks good, just a few notes on potential cleanups..
> 
> >  fs/xfs/Makefile        |    1 
> >  fs/xfs/libxfs/xfs_fs.h |   13 +
> >  fs/xfs/xfs_fsmap.c     |  819 ++++++++++++++++++++++++++++++++++++++++++++++++
> >  fs/xfs/xfs_fsmap.h     |   53 +++
> >  fs/xfs/xfs_ioctl.c     |   84 +++++
> >  fs/xfs/xfs_ioctl32.c   |    2 
> >  fs/xfs/xfs_trace.c     |    1 
> >  fs/xfs/xfs_trace.h     |   84 +++++
> >  fs/xfs/xfs_trans.c     |   22 +
> >  fs/xfs/xfs_trans.h     |    2 
> >  10 files changed, 1081 insertions(+)
> >  create mode 100644 fs/xfs/xfs_fsmap.c
> >  create mode 100644 fs/xfs/xfs_fsmap.h
> > 
> > 
> > diff --git a/fs/xfs/Makefile b/fs/xfs/Makefile
> > index c7515d4..0e7ee30 100644
> > --- a/fs/xfs/Makefile
> > +++ b/fs/xfs/Makefile
> > @@ -80,6 +80,7 @@ xfs-y				+= xfs_aops.o \
> >  				   xfs_extent_busy.o \
> >  				   xfs_file.o \
> >  				   xfs_filestream.o \
> > +				   xfs_fsmap.o \
> >  				   xfs_fsops.o \
> >  				   xfs_globals.o \
> >  				   xfs_icache.o \
> > diff --git a/fs/xfs/libxfs/xfs_fs.h b/fs/xfs/libxfs/xfs_fs.h
> > index b72dc82..095bdf0 100644
> > --- a/fs/xfs/libxfs/xfs_fs.h
> > +++ b/fs/xfs/libxfs/xfs_fs.h
> > @@ -92,6 +92,18 @@ struct getbmapx {
> >  #define BMV_OF_LAST		0x4	/* segment is the last in the file */
> >  #define BMV_OF_SHARED		0x8	/* segment shared with another file */
> >  
> > +/*	fmr_owner special values for FS_IOC_GETFSMAP */
> > +#define XFS_FMR_OWN_FREE	FMR_OWN_FREE      /* free space */
> > +#define XFS_FMR_OWN_UNKNOWN	FMR_OWN_UNKNOWN   /* unknown owner */
> > +#define XFS_FMR_OWN_FS		FMR_OWNER('X', 1) /* static fs metadata */
> > +#define XFS_FMR_OWN_LOG		FMR_OWNER('X', 2) /* journalling log */
> > +#define XFS_FMR_OWN_AG		FMR_OWNER('X', 3) /* per-AG metadata */
> > +#define XFS_FMR_OWN_INOBT	FMR_OWNER('X', 4) /* inode btree blocks */
> > +#define XFS_FMR_OWN_INODES	FMR_OWNER('X', 5) /* inodes */
> > +#define XFS_FMR_OWN_REFC	FMR_OWNER('X', 6) /* refcount tree */
> > +#define XFS_FMR_OWN_COW		FMR_OWNER('X', 7) /* cow staging */
> > +#define XFS_FMR_OWN_DEFECTIVE	FMR_OWNER('X', 8) /* bad blocks */
> > +
> >  /*
> >   * Structure for XFS_IOC_FSSETDM.
> >   * For use by backup and restore programs to set the XFS on-disk inode
> > @@ -502,6 +514,7 @@ typedef struct xfs_swapext
> >  #define XFS_IOC_GETBMAPX	_IOWR('X', 56, struct getbmap)
> >  #define XFS_IOC_ZERO_RANGE	_IOW ('X', 57, struct xfs_flock64)
> >  #define XFS_IOC_FREE_EOFBLOCKS	_IOR ('X', 58, struct xfs_fs_eofblocks)
> > +/*	XFS_IOC_GETFSMAP ------ hoisted 59         */
> >  
> >  /*
> >   * ioctl commands that replace IRIX syssgi()'s
> > diff --git a/fs/xfs/xfs_fsmap.c b/fs/xfs/xfs_fsmap.c
> > new file mode 100644
> > index 0000000..2d0fa2a
> > --- /dev/null
> > +++ b/fs/xfs/xfs_fsmap.c
> > @@ -0,0 +1,819 @@
> ...
> > +/* Compare a record against our starting point */
> > +static bool
> > +xfs_getfsmap_rec_before_low_key(
> > +	struct xfs_getfsmap_info	*info,
> > +	struct xfs_rmap_irec		*rec)
> > +{
> 
> I didn't catch this the first time around, but this is really just an
> rmap record comparison function that looks like could be made more
> generic. E.g., disregard the "low key" semantics in the name and just
> pass two xfs_rmap_irec structures..?

Hmm.  I bet you're right, and we can factor rmap_compare out of
xfs_repair as an added bonus.

> > +	uint64_t			x, y;
> > +
> > +	if (rec->rm_startblock < info->low.rm_startblock)
> > +		return true;
> > +	if (rec->rm_startblock > info->low.rm_startblock)
> > +		return false;
> > +
> > +	if (rec->rm_owner < info->low.rm_owner)
> > +		return true;
> > +	if (rec->rm_owner > info->low.rm_owner)
> > +		return false;
> > +
> > +	/*
> > +	 * Separate data and attr rmaps into non-overlapping parts of
> > +	 * the 2^64 offset space to simplify the comparison logic.  The
> > +	 * on-disk rmapbt code already has bit packing helpers that do
> > +	 * this, so reuse them here.
> > +	 */
> > +	x = xfs_rmap_irec_offset_pack(rec);
> > +	y = xfs_rmap_irec_offset_pack(&info->low);
> > +	if (x < y)
> > +		return true;
> > +	return false;
> > +}
> > +
> ...
> > +/* Report any gap at the end of the rmap records. */
> > +STATIC int
> > +xfs_getfsmap_datadev_rmapbt_end(
> > +	struct xfs_btree_cur		*cur,
> > +	struct xfs_getfsmap_info	*info,
> > +	void				*priv)
> > +{
> > +	return xfs_getfsmap_datadev_helper(cur, &info->high, info);
> > +}
> > +
> > +/* Actually query the rmap btree. */
> > +STATIC int
> > +xfs_getfsmap_datadev_rmapbt_query(
> > +	struct xfs_trans		*tp,
> > +	struct xfs_getfsmap_info	*info,
> > +	struct xfs_btree_cur		**curpp,
> > +	void				*priv)
> > +{
> > +	*curpp = xfs_rmapbt_init_cursor(tp->t_mountp, tp, info->agf_bp,
> > +			info->agno);
> > +	return xfs_rmap_query_range(*curpp, &info->low, &info->high,
> > +			xfs_getfsmap_datadev_helper, info);
> > +}
> 
> Hmm.. I haven't got through the end of the latest series yet, but do we
> really need separate query and end handlers here? So far, it looks like
> these could be combined via info->last. For example, something like:
>
> xfs_getfsmap_datadev_rmapbt_query()
> {
> 	if (!*curpp)
> 		*curpp = init_cursor(...);
> 	/* report any gap between the last record we saw and the high key */
> 	if (info->last)
> 		return xfs_getfsmap_datadev_helper(...);
> 	/* query the rmapbt */
> 	return xfs_rmap_query_range(...);
> }
> 
> Thoughts?

That would work.  Though it should never be the case that info->last ==
true at the same time *curpp == NULL, since we're either entering an AG
for the first time (and therefore need to create a cursor) or reporting
a gap at eofs, in which case we reuse the not-yet-deleted cursor from
the last AG.

IOWs, the above can be simplified to:

{
	if (info->last)
		return xfs_getfsmap_datadev_helper(...);
	*curpp = init_cursor(...);
	return xfs_rmap_query_range(...);
}

...modulo comments and other tidying.

> > +/* Execute a getfsmap query against the regular data device rmapbt. */
> > +STATIC int
> > +xfs_getfsmap_datadev_rmapbt(
> > +	struct xfs_trans		*tp,
> > +	struct xfs_fsmap		*keys,
> > +	struct xfs_getfsmap_info	*info)
> > +{
> > +	info->missing_owner = XFS_FMR_OWN_FREE;
> > +	return __xfs_getfsmap_datadev(tp, keys, info,
> > +			xfs_getfsmap_datadev_rmapbt_query,
> > +			xfs_getfsmap_datadev_rmapbt_end, NULL);
> > +}
> > +
> ...
> > +#define XFS_GETFSMAP_DEVS	2
> > +/*
> > + * Get filesystem's extents as described in head, and format for
> > + * output.  Calls formatter to fill the user's buffer until all
> > + * extents are mapped, until the passed-in head->fmh_count slots have
> > + * been filled, or until the formatter short-circuits the loop, if it
> > + * is tracking filled-in extents on its own.
> > + *
> > + * Key to Confusion
> > + * ----------------
> > + * There are multiple levels of keys and counters at work here:
> > + * xfs_fsmap_head.fmh_keys	-- low and high fsmap keys passed in;
> > + * 				   these reflect fs-wide sector addrs.
> > + * xfs_getfsmap_info.rkey_low	-- pointer to fmh_keys[0].
> 
> Thanks for the comment...
> 
> info.rkey_low is set below (in xfs_getfsmap()) and then otherwise
> appears to be unused now that the next_daddr fixup bits are gone.

Heh, yep, rkey_low can go away completely now.  Thanks for the review
and helping me see the forest through the trees. :)

--D

> 
> Brian
> 
> > + * dkeys			-- fmh_keys used to query each device;
> > + * 				   these are fmh_keys but w/ the low key
> > + * 				   bumped up by fmr_length.
> > + * xfs_getfsmap_info.next_daddr	-- next disk addr we expect to see; this
> > + *				   is how we detect gaps in the fsmap
> > +				   records and report them.
> > + * xfs_getfsmap_info.low/high	-- per-AG low/high keys computed from
> > + * 				   dkeys; used to query the metadata.
> > + */
> > +int
> > +xfs_getfsmap(
> > +	struct xfs_mount		*mp,
> > +	struct xfs_fsmap_head		*head,
> > +	xfs_fsmap_format_t		formatter,
> > +	void				*arg)
> > +{
> > +	struct xfs_trans		*tp = NULL;
> > +	struct xfs_fsmap		dkeys[2];	/* per-dev keys */
> > +	struct xfs_getfsmap_dev		handlers[XFS_GETFSMAP_DEVS];
> > +	struct xfs_getfsmap_info	info = {0};
> > +	int				i;
> > +	int				error = 0;
> > +
> > +	if (!xfs_sb_version_hasrmapbt(&mp->m_sb))
> > +		return -EOPNOTSUPP;
> > +	if (head->fmh_iflags & ~FMH_IF_VALID)
> > +		return -EINVAL;
> > +	if (!xfs_getfsmap_is_valid_device(mp, &head->fmh_keys[0]) ||
> > +	    !xfs_getfsmap_is_valid_device(mp, &head->fmh_keys[1]))
> > +		return -EINVAL;
> > +
> > +	head->fmh_entries = 0;
> > +
> > +	/* Set up our device handlers. */
> > +	memset(handlers, 0, sizeof(handlers));
> > +	handlers[0].dev = new_encode_dev(mp->m_ddev_targp->bt_dev);
> > +	handlers[0].fn = xfs_getfsmap_datadev_rmapbt;
> > +	if (mp->m_logdev_targp != mp->m_ddev_targp) {
> > +		handlers[1].dev = new_encode_dev(mp->m_logdev_targp->bt_dev);
> > +		handlers[1].fn = xfs_getfsmap_logdev;
> > +	}
> > +
> > +	xfs_sort(handlers, XFS_GETFSMAP_DEVS, sizeof(struct xfs_getfsmap_dev),
> > +			xfs_getfsmap_dev_compare);
> > +
> > +	/*
> > +	 * To continue where we left off, we allow userspace to use the
> > +	 * last mapping from a previous call as the low key of the next.
> > +	 * This is identified by a non-zero length in the low key. We
> > +	 * have to increment the low key in this scenario to ensure we
> > +	 * don't return the same mapping again, and instead return the
> > +	 * very next mapping.
> > +	 *
> > +	 * If the low key mapping refers to file data, the same physical
> > +	 * blocks could be mapped to several other files/offsets.
> > +	 * According to rmapbt record ordering, the minimal next
> > +	 * possible record for the block range is the next starting
> > +	 * offset in the same inode. Therefore, bump the file offset to
> > +	 * continue the search appropriately.  For all other low key
> > +	 * mapping types (attr blocks, metadata), bump the physical
> > +	 * offset as there can be no other mapping for the same physical
> > +	 * block range.
> > +	 */
> > +	dkeys[0] = head->fmh_keys[0];
> > +	if (dkeys[0].fmr_flags & (FMR_OF_SPECIAL_OWNER | FMR_OF_EXTENT_MAP)) {
> > +		dkeys[0].fmr_physical += dkeys[0].fmr_length;
> > +		dkeys[0].fmr_owner = 0;
> > +		if (dkeys[0].fmr_offset)
> > +			return -EINVAL;
> > +	} else
> > +		dkeys[0].fmr_offset += dkeys[0].fmr_length;
> > +	dkeys[0].fmr_length = 0;
> > +	memset(&dkeys[1], 0xFF, sizeof(struct xfs_fsmap));
> > +
> > +	if (!xfs_getfsmap_check_keys(dkeys, &head->fmh_keys[1]))
> > +		return -EINVAL;
> > +
> > +	info.next_daddr = head->fmh_keys[0].fmr_physical +
> > +			  head->fmh_keys[0].fmr_length;
> > +	info.rkey_low = &head->fmh_keys[0];
> > +	info.formatter = formatter;
> > +	info.format_arg = arg;
> > +	info.head = head;
> > +
> > +	/* For each device we support... */
> > +	for (i = 0; i < XFS_GETFSMAP_DEVS; i++) {
> > +		/* Is this device within the range the user asked for? */
> > +		if (!handlers[i].fn)
> > +			continue;
> > +		if (head->fmh_keys[0].fmr_device > handlers[i].dev)
> > +			continue;
> > +		if (head->fmh_keys[1].fmr_device < handlers[i].dev)
> > +			break;
> > +
> > +		/*
> > +		 * If this device number matches the high key, we have
> > +		 * to pass the high key to the handler to limit the
> > +		 * query results.  If the device number exceeds the
> > +		 * low key, zero out the low key so that we get
> > +		 * everything from the beginning.
> > +		 */
> > +		if (handlers[i].dev == head->fmh_keys[1].fmr_device)
> > +			dkeys[1] = head->fmh_keys[1];
> > +		if (handlers[i].dev > head->fmh_keys[0].fmr_device)
> > +			memset(&dkeys[0], 0, sizeof(struct xfs_fsmap));
> > +
> > +		error = xfs_trans_alloc_empty(mp, &tp);
> > +		if (error)
> > +			break;
> > +
> > +		info.dev = handlers[i].dev;
> > +		info.last = false;
> > +		info.agno = NULLAGNUMBER;
> > +		error = handlers[i].fn(tp, dkeys, &info);
> > +		if (error)
> > +			break;
> > +		xfs_trans_cancel(tp);
> > +		tp = NULL;
> > +		info.next_daddr = 0;
> > +	}
> > +
> > +	if (tp)
> > +		xfs_trans_cancel(tp);
> > +	head->fmh_oflags = FMH_OF_DEV_T;
> > +	return error;
> > +}
> > diff --git a/fs/xfs/xfs_fsmap.h b/fs/xfs/xfs_fsmap.h
> > new file mode 100644
> > index 0000000..0b9bf82
> > --- /dev/null
> > +++ b/fs/xfs/xfs_fsmap.h
> > @@ -0,0 +1,53 @@
> > +/*
> > + * Copyright (C) 2017 Oracle.  All Rights Reserved.
> > + *
> > + * Author: Darrick J. Wong <darrick.wong@oracle.com>
> > + *
> > + * This program is free software; you can redistribute it and/or
> > + * modify it under the terms of the GNU General Public License
> > + * as published by the Free Software Foundation; either version 2
> > + * of the License, or (at your option) any later version.
> > + *
> > + * This program is distributed in the hope that it would be useful,
> > + * but WITHOUT ANY WARRANTY; without even the implied warranty of
> > + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
> > + * GNU General Public License for more details.
> > + *
> > + * You should have received a copy of the GNU General Public License
> > + * along with this program; if not, write the Free Software Foundation,
> > + * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301, USA.
> > + */
> > +#ifndef __XFS_FSMAP_H__
> > +#define __XFS_FSMAP_H__
> > +
> > +struct fsmap;
> > +
> > +/* internal fsmap representation */
> > +struct xfs_fsmap {
> > +	dev_t		fmr_device;	/* device id */
> > +	uint32_t	fmr_flags;	/* mapping flags */
> > +	uint64_t	fmr_physical;	/* device offset of segment */
> > +	uint64_t	fmr_owner;	/* owner id */
> > +	xfs_fileoff_t	fmr_offset;	/* file offset of segment */
> > +	xfs_filblks_t	fmr_length;	/* length of segment, blocks */
> > +};
> > +
> > +struct xfs_fsmap_head {
> > +	uint32_t	fmh_iflags;	/* control flags */
> > +	uint32_t	fmh_oflags;	/* output flags */
> > +	unsigned int	fmh_count;	/* # of entries in array incl. input */
> > +	unsigned int	fmh_entries;	/* # of entries filled in (output). */
> > +
> > +	struct xfs_fsmap fmh_keys[2];	/* low and high keys */
> > +};
> > +
> > +void xfs_fsmap_from_internal(struct fsmap *dest, struct xfs_fsmap *src);
> > +void xfs_fsmap_to_internal(struct xfs_fsmap *dest, struct fsmap *src);
> > +
> > +/* fsmap to userspace formatter - copy to user & advance pointer */
> > +typedef int (*xfs_fsmap_format_t)(struct xfs_fsmap *, void *);
> > +
> > +int xfs_getfsmap(struct xfs_mount *mp, struct xfs_fsmap_head *head,
> > +		xfs_fsmap_format_t formatter, void *arg);
> > +
> > +#endif /* __XFS_FSMAP_H__ */
> > diff --git a/fs/xfs/xfs_ioctl.c b/fs/xfs/xfs_ioctl.c
> > index c67cfb4..52f635a 100644
> > --- a/fs/xfs/xfs_ioctl.c
> > +++ b/fs/xfs/xfs_ioctl.c
> > @@ -41,6 +41,9 @@
> >  #include "xfs_trans.h"
> >  #include "xfs_pnfs.h"
> >  #include "xfs_acl.h"
> > +#include "xfs_btree.h"
> > +#include <linux/fsmap.h>
> > +#include "xfs_fsmap.h"
> >  
> >  #include <linux/capability.h>
> >  #include <linux/dcache.h>
> > @@ -1607,6 +1610,84 @@ xfs_ioc_getbmapx(
> >  	return 0;
> >  }
> >  
> > +struct getfsmap_info {
> > +	struct xfs_mount	*mp;
> > +	struct fsmap __user	*data;
> > +	__u32			last_flags;
> > +};
> > +
> > +STATIC int
> > +xfs_getfsmap_format(struct xfs_fsmap *xfm, void *priv)
> > +{
> > +	struct getfsmap_info	*info = priv;
> > +	struct fsmap		fm;
> > +
> > +	trace_xfs_getfsmap_mapping(info->mp, xfm);
> > +
> > +	info->last_flags = xfm->fmr_flags;
> > +	xfs_fsmap_from_internal(&fm, xfm);
> > +	if (copy_to_user(info->data, &fm, sizeof(struct fsmap)))
> > +		return -EFAULT;
> > +
> > +	info->data++;
> > +	return 0;
> > +}
> > +
> > +STATIC int
> > +xfs_ioc_getfsmap(
> > +	struct xfs_inode	*ip,
> > +	void			__user *arg)
> > +{
> > +	struct getfsmap_info	info;
> > +	struct xfs_fsmap_head	xhead = {0};
> > +	struct fsmap_head	head;
> > +	bool			aborted = false;
> > +	int			error;
> > +
> > +	if (copy_from_user(&head, arg, sizeof(struct fsmap_head)))
> > +		return -EFAULT;
> > +	if (memchr_inv(head.fmh_reserved, 0, sizeof(head.fmh_reserved)) ||
> > +	    memchr_inv(head.fmh_keys[0].fmr_reserved, 0,
> > +		       sizeof(head.fmh_keys[0].fmr_reserved)) ||
> > +	    memchr_inv(head.fmh_keys[1].fmr_reserved, 0,
> > +		       sizeof(head.fmh_keys[1].fmr_reserved)))
> > +		return -EINVAL;
> > +
> > +	xhead.fmh_iflags = head.fmh_iflags;
> > +	xhead.fmh_count = head.fmh_count;
> > +	xfs_fsmap_to_internal(&xhead.fmh_keys[0], &head.fmh_keys[0]);
> > +	xfs_fsmap_to_internal(&xhead.fmh_keys[1], &head.fmh_keys[1]);
> > +
> > +	trace_xfs_getfsmap_low_key(ip->i_mount, &xhead.fmh_keys[0]);
> > +	trace_xfs_getfsmap_high_key(ip->i_mount, &xhead.fmh_keys[1]);
> > +
> > +	info.mp = ip->i_mount;
> > +	info.data = ((__force struct fsmap_head *)arg)->fmh_recs;
> > +	error = xfs_getfsmap(ip->i_mount, &xhead, xfs_getfsmap_format, &info);
> > +	if (error == XFS_BTREE_QUERY_RANGE_ABORT) {
> > +		error = 0;
> > +		aborted = true;
> > +	} else if (error)
> > +		return error;
> > +
> > +	/* If we didn't abort, set the "last" flag in the last fmx */
> > +	if (!aborted && xhead.fmh_entries) {
> > +		info.data--;
> > +		info.last_flags |= FMR_OF_LAST;
> > +		if (copy_to_user(&info.data->fmr_flags, &info.last_flags,
> > +				sizeof(info.last_flags)))
> > +			return -EFAULT;
> > +	}
> > +
> > +	/* copy back header */
> > +	head.fmh_entries = xhead.fmh_entries;
> > +	head.fmh_oflags = xhead.fmh_oflags;
> > +	if (copy_to_user(arg, &head, sizeof(struct fsmap_head)))
> > +		return -EFAULT;
> > +
> > +	return 0;
> > +}
> > +
> >  int
> >  xfs_ioc_swapext(
> >  	xfs_swapext_t	*sxp)
> > @@ -1787,6 +1868,9 @@ xfs_file_ioctl(
> >  	case XFS_IOC_GETBMAPX:
> >  		return xfs_ioc_getbmapx(ip, arg);
> >  
> > +	case FS_IOC_GETFSMAP:
> > +		return xfs_ioc_getfsmap(ip, arg);
> > +
> >  	case XFS_IOC_FD_TO_HANDLE:
> >  	case XFS_IOC_PATH_TO_HANDLE:
> >  	case XFS_IOC_PATH_TO_FSHANDLE: {
> > diff --git a/fs/xfs/xfs_ioctl32.c b/fs/xfs/xfs_ioctl32.c
> > index 7c49938..fa0bc4d 100644
> > --- a/fs/xfs/xfs_ioctl32.c
> > +++ b/fs/xfs/xfs_ioctl32.c
> > @@ -20,6 +20,7 @@
> >  #include <linux/mount.h>
> >  #include <linux/slab.h>
> >  #include <linux/uaccess.h>
> > +#include <linux/fsmap.h>
> >  #include "xfs.h"
> >  #include "xfs_fs.h"
> >  #include "xfs_format.h"
> > @@ -554,6 +555,7 @@ xfs_file_compat_ioctl(
> >  	case XFS_IOC_GOINGDOWN:
> >  	case XFS_IOC_ERROR_INJECTION:
> >  	case XFS_IOC_ERROR_CLEARALL:
> > +	case FS_IOC_GETFSMAP:
> >  		return xfs_file_ioctl(filp, cmd, p);
> >  #ifndef BROKEN_X86_ALIGNMENT
> >  	/* These are handled fine if no alignment issues */
> > diff --git a/fs/xfs/xfs_trace.c b/fs/xfs/xfs_trace.c
> > index 7f17ae6..5d95fe3 100644
> > --- a/fs/xfs/xfs_trace.c
> > +++ b/fs/xfs/xfs_trace.c
> > @@ -47,6 +47,7 @@
> >  #include "xfs_inode_item.h"
> >  #include "xfs_bmap_btree.h"
> >  #include "xfs_filestream.h"
> > +#include "xfs_fsmap.h"
> >  
> >  /*
> >   * We include this last to have the helpers above available for the trace
> > diff --git a/fs/xfs/xfs_trace.h b/fs/xfs/xfs_trace.h
> > index d3d11905..ef666e6 100644
> > --- a/fs/xfs/xfs_trace.h
> > +++ b/fs/xfs/xfs_trace.h
> > @@ -40,6 +40,8 @@ struct xfs_inode_log_format;
> >  struct xfs_bmbt_irec;
> >  struct xfs_btree_cur;
> >  struct xfs_refcount_irec;
> > +struct xfs_fsmap;
> > +struct xfs_rmap_irec;
> >  
> >  DECLARE_EVENT_CLASS(xfs_attr_list_class,
> >  	TP_PROTO(struct xfs_attr_list_context *ctx),
> > @@ -3270,6 +3272,88 @@ DEFINE_INODE_IREC_EVENT(xfs_swap_extent_rmap_remap);
> >  DEFINE_INODE_IREC_EVENT(xfs_swap_extent_rmap_remap_piece);
> >  DEFINE_INODE_ERROR_EVENT(xfs_swap_extent_rmap_error);
> >  
> > +/* fsmap traces */
> > +DECLARE_EVENT_CLASS(xfs_fsmap_class,
> > +	TP_PROTO(struct xfs_mount *mp, u32 keydev, xfs_agnumber_t agno,
> > +		 struct xfs_rmap_irec *rmap),
> > +	TP_ARGS(mp, keydev, agno, rmap),
> > +	TP_STRUCT__entry(
> > +		__field(dev_t, dev)
> > +		__field(dev_t, keydev)
> > +		__field(xfs_agnumber_t, agno)
> > +		__field(xfs_fsblock_t, bno)
> > +		__field(xfs_filblks_t, len)
> > +		__field(__uint64_t, owner)
> > +		__field(__uint64_t, offset)
> > +		__field(unsigned int, flags)
> > +	),
> > +	TP_fast_assign(
> > +		__entry->dev = mp->m_super->s_dev;
> > +		__entry->keydev = new_decode_dev(keydev);
> > +		__entry->agno = agno;
> > +		__entry->bno = rmap->rm_startblock;
> > +		__entry->len = rmap->rm_blockcount;
> > +		__entry->owner = rmap->rm_owner;
> > +		__entry->offset = rmap->rm_offset;
> > +		__entry->flags = rmap->rm_flags;
> > +	),
> > +	TP_printk("dev %d:%d keydev %d:%d agno %u bno %llu len %llu owner %lld offset %llu flags 0x%x\n",
> > +		  MAJOR(__entry->dev), MINOR(__entry->dev),
> > +		  MAJOR(__entry->keydev), MINOR(__entry->keydev),
> > +		  __entry->agno,
> > +		  __entry->bno,
> > +		  __entry->len,
> > +		  __entry->owner,
> > +		  __entry->offset,
> > +		  __entry->flags)
> > +)
> > +#define DEFINE_FSMAP_EVENT(name) \
> > +DEFINE_EVENT(xfs_fsmap_class, name, \
> > +	TP_PROTO(struct xfs_mount *mp, u32 keydev, xfs_agnumber_t agno, \
> > +		 struct xfs_rmap_irec *rmap), \
> > +	TP_ARGS(mp, keydev, agno, rmap))
> > +DEFINE_FSMAP_EVENT(xfs_fsmap_low_key);
> > +DEFINE_FSMAP_EVENT(xfs_fsmap_high_key);
> > +DEFINE_FSMAP_EVENT(xfs_fsmap_mapping);
> > +
> > +DECLARE_EVENT_CLASS(xfs_getfsmap_class,
> > +	TP_PROTO(struct xfs_mount *mp, struct xfs_fsmap *fsmap),
> > +	TP_ARGS(mp, fsmap),
> > +	TP_STRUCT__entry(
> > +		__field(dev_t, dev)
> > +		__field(dev_t, keydev)
> > +		__field(xfs_daddr_t, block)
> > +		__field(xfs_daddr_t, len)
> > +		__field(__uint64_t, owner)
> > +		__field(__uint64_t, offset)
> > +		__field(__uint64_t, flags)
> > +	),
> > +	TP_fast_assign(
> > +		__entry->dev = mp->m_super->s_dev;
> > +		__entry->keydev = new_decode_dev(fsmap->fmr_device);
> > +		__entry->block = fsmap->fmr_physical;
> > +		__entry->len = fsmap->fmr_length;
> > +		__entry->owner = fsmap->fmr_owner;
> > +		__entry->offset = fsmap->fmr_offset;
> > +		__entry->flags = fsmap->fmr_flags;
> > +	),
> > +	TP_printk("dev %d:%d keydev %d:%d block %llu len %llu owner %lld offset %llu flags 0x%llx\n",
> > +		  MAJOR(__entry->dev), MINOR(__entry->dev),
> > +		  MAJOR(__entry->keydev), MINOR(__entry->keydev),
> > +		  __entry->block,
> > +		  __entry->len,
> > +		  __entry->owner,
> > +		  __entry->offset,
> > +		  __entry->flags)
> > +)
> > +#define DEFINE_GETFSMAP_EVENT(name) \
> > +DEFINE_EVENT(xfs_getfsmap_class, name, \
> > +	TP_PROTO(struct xfs_mount *mp, struct xfs_fsmap *fsmap), \
> > +	TP_ARGS(mp, fsmap))
> > +DEFINE_GETFSMAP_EVENT(xfs_getfsmap_low_key);
> > +DEFINE_GETFSMAP_EVENT(xfs_getfsmap_high_key);
> > +DEFINE_GETFSMAP_EVENT(xfs_getfsmap_mapping);
> > +
> >  #endif /* _TRACE_XFS_H */
> >  
> >  #undef TRACE_INCLUDE_PATH
> > diff --git a/fs/xfs/xfs_trans.c b/fs/xfs/xfs_trans.c
> > index 70f42ea..a280e12 100644
> > --- a/fs/xfs/xfs_trans.c
> > +++ b/fs/xfs/xfs_trans.c
> > @@ -263,6 +263,28 @@ xfs_trans_alloc(
> >  }
> >  
> >  /*
> > + * Create an empty transaction with no reservation.  This is a defensive
> > + * mechanism for routines that query metadata without actually modifying
> > + * them -- if the metadata being queried is somehow cross-linked (think a
> > + * btree block pointer that points higher in the tree), we risk deadlock.
> > + * However, blocks grabbed as part of a transaction can be re-grabbed.
> > + * The verifiers will notice the corrupt block and the operation will fail
> > + * back to userspace without deadlocking.
> > + *
> > + * Note the zero-length reservation; this transaction MUST be cancelled
> > + * without any dirty data.
> > + */
> > +int
> > +xfs_trans_alloc_empty(
> > +	struct xfs_mount		*mp,
> > +	struct xfs_trans		**tpp)
> > +{
> > +	struct xfs_trans_res		resv = {0};
> > +
> > +	return xfs_trans_alloc(mp, &resv, 0, 0, XFS_TRANS_NO_WRITECOUNT, tpp);
> > +}
> > +
> > +/*
> >   * Record the indicated change to the given field for application
> >   * to the file system's superblock when the transaction commits.
> >   * For now, just store the change in the transaction structure.
> > diff --git a/fs/xfs/xfs_trans.h b/fs/xfs/xfs_trans.h
> > index 61b7fbd..98024cb 100644
> > --- a/fs/xfs/xfs_trans.h
> > +++ b/fs/xfs/xfs_trans.h
> > @@ -159,6 +159,8 @@ typedef struct xfs_trans {
> >  int		xfs_trans_alloc(struct xfs_mount *mp, struct xfs_trans_res *resp,
> >  			uint blocks, uint rtextents, uint flags,
> >  			struct xfs_trans **tpp);
> > +int		xfs_trans_alloc_empty(struct xfs_mount *mp,
> > +			struct xfs_trans **tpp);
> >  void		xfs_trans_mod_sb(xfs_trans_t *, uint, int64_t);
> >  
> >  struct xfs_buf	*xfs_trans_get_buf_map(struct xfs_trans *tp,
> > 
> > --
> > To unsubscribe from this list: send the line "unsubscribe linux-xfs" in
> > the body of a message to majordomo@vger.kernel.org
> > More majordomo info at  http://vger.kernel.org/majordomo-info.html