linux-xfs.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: "Darrick J. Wong" <darrick.wong@oracle.com>
To: Eric Sandeen <sandeen@sandeen.net>
Cc: sandeen@redhat.com, linux-xfs@vger.kernel.org
Subject: Re: [PATCH 0.9/9] xfs: introduce the XFS_IOC_GETFSMAP ioctl
Date: Fri, 26 May 2017 14:41:42 -0700	[thread overview]
Message-ID: <20170526214142.GO4521@birch.djwong.org> (raw)
In-Reply-To: <9d94c89f-f730-955c-bde3-eb9264737303@sandeen.net>

On Fri, May 26, 2017 at 04:20:49PM -0500, Eric Sandeen wrote:
> 
> 
> On 5/8/17 2:47 PM, 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>
> 
> Ok, howzabout this:
> 
> Changes from your version:
> 
> - remove libxfs stuff, I synced that separately already
> - rename sys_fsmap just fsmap (it's not a syscall)
> - change meaning of "have_fsmap" - it now means we have it on the
>   system, not that this platform (linux) might support it.*

I'll have a look... though afaict this removes anything that would
prevent us from building the fsmap stuff on a non-linux platform.
Seeing as fsmap is linux-only, we might as well restrict it on the
platforms we still support.

Darwin?  (I think Jan Tulak added this, though OSX doen't do XFS...)
BSD?  (Wasn't XFS removed from their kernel a few years ago?)

> - remove OVERRIDE_GETFSMAP - we aren't overriding anything now,
>   we are using our own definition only if the system doesn't have it

Ok.

> - #include <linux/fs.h> before fsmap.h in autoconf or it'll /never/
>   build and be detected

Ok.

> * we didn't bother with this for i.e. fiemap or anything else,
>   afaict, and I'd like to just kill off the other platforms anyway.
>   (I guess they didn't carry along their own copy, though, but meh,
>   I have no idea if they even build at this point)
> 
> See what you think?
> 
> 
> diff --git a/configure.ac b/configure.ac
> index aa102e4..9534986 100644
> --- a/configure.ac
> +++ b/configure.ac
> @@ -142,6 +142,7 @@ AC_HAVE_READDIR
>  AC_HAVE_FSETXATTR
>  AC_HAVE_MREMAP
>  AC_NEED_INTERNAL_FSXATTR
> +AC_HAVE_GETFSMAP
>  
>  if test "$enable_blkid" = yes; then
>  AC_HAVE_BLKID_TOPO
> diff --git a/include/builddefs.in b/include/builddefs.in
> index 4d6bb2d..675bedf 100644
> --- a/include/builddefs.in
> +++ b/include/builddefs.in
> @@ -112,6 +112,7 @@ HAVE_FLS = @have_fls@
>  HAVE_FSETXATTR = @have_fsetxattr@
>  HAVE_MREMAP = @have_mremap@
>  NEED_INTERNAL_FSXATTR = @need_internal_fsxattr@
> +HAVE_GETFSMAP = @have_getfsmap@
>  
>  GCCFLAGS = -funsigned-char -fno-strict-aliasing -Wall
>  #	   -Wbitwise -Wno-transparent-union -Wno-old-initializer -Wno-decl
> @@ -150,6 +151,9 @@ endif
>  ifeq ($(NEED_INTERNAL_FSXATTR),yes)
>  PCFLAGS+= -DOVERRIDE_SYSTEM_FSXATTR
>  endif
> +ifeq ($(HAVE_FSMAP),yes)
> +PCFLAGS+= -DHAVE_FSMAP
> +endif
>  
>  
>  GCFLAGS = $(DEBUG) \
> diff --git a/include/linux.h b/include/linux.h
> index 6a676ca..38969d1 100644
> --- a/include/linux.h
> +++ b/include/linux.h
> @@ -222,4 +222,107 @@ struct fsxattr {
>  #define FS_XFLAG_COWEXTSIZE	0x00010000	/* CoW extent size allocator hint */
>  #endif
>  
> +#ifndef HAVE_GETFSMAP
> +/*
> + *	Structure for FS_IOC_GETFSMAP.
> + *
> + *	The memory layout for this call are the scalar values defined in
> + *	struct fsmap_head, followed by two struct fsmap that describe
> + *	the lower and upper bound of mappings to return, followed by an
> + *	array of struct fsmap mappings.
> + *
> + *	fmh_iflags control the output of the call, whereas fmh_oflags report
> + *	on the overall record output.  fmh_count should be set to the
> + *	length of the fmh_recs array, and fmh_entries will be set to the
> + *	number of entries filled out during each call.  If fmh_count is
> + *	zero, the number of reverse mappings will be returned in
> + *	fmh_entries, though no mappings will be returned.  fmh_reserved
> + *	must be set to zero.
> + *
> + *	The two elements in the fmh_keys array are used to constrain the
> + *	output.  The first element in the array should represent the
> + *	lowest disk mapping ("low key") that the user wants to learn
> + *	about.  If this value is all zeroes, the filesystem will return
> + *	the first entry it knows about.  For a subsequent call, the
> + *	contents of fsmap_head.fmh_recs[fsmap_head.fmh_count - 1] should be
> + *	copied into fmh_keys[0] to have the kernel start where it left off.
> + *
> + *	The second element in the fmh_keys array should represent the
> + *	highest disk mapping ("high key") that the user wants to learn
> + *	about.  If this value is all ones, the filesystem will not stop
> + *	until it runs out of mapping to return or runs out of space in
> + *	fmh_recs.
> + *
> + *	fmr_device can be either a 32-bit cookie representing a device, or
> + *	a 32-bit dev_t if the FMH_OF_DEV_T flag is set.  fmr_physical,
> + *	fmr_offset, and fmr_length are expressed in units of bytes.
> + *	fmr_owner is either an inode number, or a special value if
> + *	FMR_OF_SPECIAL_OWNER is set in fmr_flags.
> + */
> +struct fsmap {
> +	__u32		fmr_device;	/* device id */
> +	__u32		fmr_flags;	/* mapping flags */
> +	__u64		fmr_physical;	/* device offset of segment */
> +	__u64		fmr_owner;	/* owner id */
> +	__u64		fmr_offset;	/* file offset of segment */
> +	__u64		fmr_length;	/* length of segment */
> +	__u64		fmr_reserved[3];	/* must be zero */
> +};
> +
> +struct fsmap_head {
> +	__u32		fmh_iflags;	/* control flags */
> +	__u32		fmh_oflags;	/* output flags */
> +	__u32		fmh_count;	/* # of entries in array incl. input */
> +	__u32		fmh_entries;	/* # of entries filled in (output). */
> +	__u64		fmh_reserved[6];	/* must be zero */
> +
> +	struct fsmap	fmh_keys[2];	/* low and high keys for the mapping search */
> +	struct fsmap	fmh_recs[];	/* returned records */
> +};
> +
> +/* Size of an fsmap_head with room for nr records. */
> +static inline size_t
> +fsmap_sizeof(
> +	unsigned int	nr)
> +{
> +	return sizeof(struct fsmap_head) + nr * sizeof(struct fsmap);
> +}
> +
> +/* Start the next fsmap query at the end of the current query results. */
> +static inline void
> +fsmap_advance(
> +	struct fsmap_head	*head)
> +{
> +	head->fmh_keys[0] = head->fmh_recs[head->fmh_entries - 1];
> +}
> +
> +/*	fmh_iflags values - set by XFS_IOC_GETFSMAP caller in the header. */
> +/* no flags defined yet */
> +#define FMH_IF_VALID		0
> +
> +/*	fmh_oflags values - returned in the header segment only. */
> +#define FMH_OF_DEV_T		0x1	/* fmr_device values will be dev_t */
> +
> +/*	fmr_flags values - returned for each non-header segment */
> +#define FMR_OF_PREALLOC		0x1	/* segment = unwritten pre-allocation */
> +#define FMR_OF_ATTR_FORK	0x2	/* segment = attribute fork */
> +#define FMR_OF_EXTENT_MAP	0x4	/* segment = extent map */
> +#define FMR_OF_SHARED		0x8	/* segment = shared with another file */
> +#define FMR_OF_SPECIAL_OWNER	0x10	/* owner is a special value */
> +#define FMR_OF_LAST		0x20	/* segment is the last in the FS */
> +
> +/* Each FS gets to define its own special owner codes. */
> +#define FMR_OWNER(type, code)	(((__u64)type << 32) | \
> +				 ((__u64)code & 0xFFFFFFFFULL))
> +#define FMR_OWNER_TYPE(owner)	((__u32)((__u64)owner >> 32))
> +#define FMR_OWNER_CODE(owner)	((__u32)(((__u64)owner & 0xFFFFFFFFULL)))
> +#define FMR_OWN_FREE		FMR_OWNER(0, 1) /* free space */
> +#define FMR_OWN_UNKNOWN		FMR_OWNER(0, 2) /* unknown owner */
> +#define FMR_OWN_METADATA	FMR_OWNER(0, 3) /* metadata */
> +
> +#define FS_IOC_GETFSMAP		_IOWR('X', 59, struct fsmap_head)
> +
> +#define HAVE_GETFSMAP
> +#endif /* HAVE_GETFSMAP */
> +
>  #endif	/* __XFS_LINUX_H__ */
> diff --git a/m4/package_libcdev.m4 b/m4/package_libcdev.m4
> index bc3b4ce..fa5b639 100644
> --- a/m4/package_libcdev.m4
> +++ b/m4/package_libcdev.m4
> @@ -277,3 +277,23 @@ AC_DEFUN([AC_NEED_INTERNAL_FSXATTR],
>      )
>      AC_SUBST(need_internal_fsxattr)
>    ])
> +
> +#
> +# Check if we have a FS_IOC_GETFSMAP ioctl (Linux)
> +#
> +AC_DEFUN([AC_HAVE_GETFSMAP],
> +  [ AC_MSG_CHECKING([for GETFSMAP])
> +    AC_TRY_LINK([
> +#define _GNU_SOURCE
> +#include <sys/syscall.h>
> +#include <unistd.h>
> +#include <linux/fs.h>
> +#include <linux/fsmap.h>
> +    ], [
> +         unsigned long x = FS_IOC_GETFSMAP;
> +         struct fsmap_head fh;
> +    ], have_getfsmap=yes
> +       AC_MSG_RESULT(yes),
> +       AC_MSG_RESULT(no))
> +    AC_SUBST(have_getfsmap)
> +  ])
> 

  reply	other threads:[~2017-05-26 21:41 UTC|newest]

Thread overview: 40+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2017-05-07 15:56 [PATCH v7 0/9] xfsprogs 4.12: GETFSMAP support Darrick J. Wong
2017-05-07 15:56 ` [PATCH 1/9] xfs_io: support the new getfsmap ioctl Darrick J. Wong
2017-05-08 21:01   ` Eric Sandeen
2017-05-15 19:18     ` Darrick J. Wong
2017-05-15 19:30       ` Eric Sandeen
2017-05-07 15:56 ` [PATCH 2/9] xfs_repair: replace rmap_compare with libxfs version Darrick J. Wong
2017-05-07 15:56 ` [PATCH 3/9] xfs_spaceman: space management tool Darrick J. Wong
2017-05-27  1:34   ` Eric Sandeen
2017-05-30 17:37     ` Darrick J. Wong
2017-05-30 18:17       ` Eric Sandeen
2017-05-30 18:47         ` Darrick J. Wong
2017-06-02 19:44           ` Darrick J. Wong
2017-05-07 15:56 ` [PATCH 4/9] xfs_spaceman: add FITRIM support Darrick J. Wong
2017-05-27  0:27   ` Eric Sandeen
2017-05-30 18:24     ` Darrick J. Wong
2017-05-07 15:56 ` [PATCH 5/9] xfs_spaceman: add new speculative prealloc control Darrick J. Wong
2017-05-27  1:45   ` Eric Sandeen
2017-05-30 18:34     ` Darrick J. Wong
2017-05-07 15:56 ` [PATCH 6/9] xfs_spaceman: AG state control Darrick J. Wong
2017-05-26 23:06   ` Eric Sandeen
2017-05-30 18:30     ` Darrick J. Wong
2017-05-07 15:57 ` [PATCH 7/9] xfs_spaceman: Free space mapping command Darrick J. Wong
2017-05-27  1:57   ` Eric Sandeen
2017-05-30 18:40     ` Darrick J. Wong
2017-05-30 18:56       ` Eric Sandeen
2017-05-30 19:19         ` Darrick J. Wong
2017-05-07 15:57 ` [PATCH 8/9] xfs_spaceman: add a man page Darrick J. Wong
2017-05-07 15:57 ` [PATCH 9/9] xfs_spaceman: add group summary mode Darrick J. Wong
2017-05-08 19:47 ` [PATCH 0.9/9] xfs: introduce the XFS_IOC_GETFSMAP ioctl Darrick J. Wong
2017-05-10 14:46   ` Eric Sandeen
2017-05-10 17:03     ` Darrick J. Wong
2017-05-12 22:29   ` Eric Sandeen
2017-05-12 23:05     ` Darrick J. Wong
2017-05-12 23:11       ` Eric Sandeen
2017-05-26 21:20   ` Eric Sandeen
2017-05-26 21:41     ` Darrick J. Wong [this message]
2017-05-26 22:12       ` Eric Sandeen
2017-05-30 18:44         ` Darrick J. Wong
2017-05-30 18:47           ` Eric Sandeen
2017-05-30 18:59             ` Eric Sandeen

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20170526214142.GO4521@birch.djwong.org \
    --to=darrick.wong@oracle.com \
    --cc=linux-xfs@vger.kernel.org \
    --cc=sandeen@redhat.com \
    --cc=sandeen@sandeen.net \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).