Document the new GETFSMAP ioctl that returns the physical layout of a (disk-based) filesystem. Signed-off-by: Darrick J. Wong <darrick.wong@xxxxxxxxxx> --- man2/ioctl_getfsmap.2 | 362 +++++++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 362 insertions(+) create mode 100644 man2/ioctl_getfsmap.2 diff --git a/man2/ioctl_getfsmap.2 b/man2/ioctl_getfsmap.2 new file mode 100644 index 0000000..ef9daef --- /dev/null +++ b/man2/ioctl_getfsmap.2 @@ -0,0 +1,362 @@ +.\" Copyright (c) 2017, Oracle. All rights reserved. +.\" +.\" %%%LICENSE_START(GPLv2+_DOC_FULL) +.\" This is free documentation; 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. +.\" +.\" The GNU General Public License's references to "object code" +.\" and "executables" are to be interpreted as the output of any +.\" document formatting or typesetting system, including +.\" intermediate and printed output. +.\" +.\" This manual is distributed in the hope that it will 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 manual; if not, see +.\" <http://www.gnu.org/licenses/>. +.\" %%%LICENSE_END +.TH IOCTL-GETFSMAP 2 2017-02-10 "Linux" "Linux Programmer's Manual" +.SH NAME +ioctl_getfsmap \- retrieve the physical layout of the filesystem +.SH SYNOPSIS +.br +.B #include <sys/ioctl.h> +.br +.B #include <linux/fs.h> +.br +.B #include <linux/fsmap.h> +.sp +.BI "int ioctl(int " fd ", FS_IOC_GETFSMAP, struct fsmap_head * " arg ); +.SH DESCRIPTION +This +.BR ioctl (2) +retrieves physical extent mappings for a filesystem. +This information can be used to discover which files are mapped to a physical +block, examine free space, or find known bad blocks, among other things. + +The sole argument to this ioctl should be a pointer to a single +.BR "struct fsmap_head" ":" +.in +4n +.nf + +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 */ +}; + +.fi +.in +The two +.I fmh_keys +array elements specify the lowest and highest reverse-mapping +keys, respectively, for which userspace would like physical mapping +information. +A reverse mapping key consists of the tuple (device, block, owner, offset). +The owner and offset fields are part of the key because some filesystems +support sharing physical blocks between multiple files and +therefore may return multiple mappings for a given physical block. +.PP +Filesystem mappings are copied into the +.I fmh_recs +array, which immediately follows the header data. +.SS Fields of struct fsmap_head +.PP +The +.I fmh_iflags +field is a bitmask passed to the kernel to alter the output. +There are no flags defined, so this value must be zero. + +.PP +The +.I fmh_oflags +field is a bitmask of flags that concern all output mappings. +If +.B FMH_OF_DEV_T +is set, then the +.I fmr_device +field represents a +.B dev_t +structure containing the major and minor numbers of the block device. + +.PP +The +.I fmh_count +field contains the number of elements in the array being passed to the +kernel. +If this value is 0, +.I fmh_entries +will be set to the number of records that would have been returned had +the array been large enough; +no mapping information will be returned. + +.PP +The +.I fmh_entries +field contains the number of elements in the +.I fmh_recs +array that contain useful information. + +.PP +The +.I fmh_reserved +fields must be set to zero. + +.SS Keys +.PP +The two key records in +.B fsmap_head.fmh_keys +specify the lowest and highest extent records in the keyspace that the caller +wants returned. +A filesystem that can share blocks between files likely requires the tuple +.RI "(" "device" ", " "physical" ", " "owner" ", " "offset" ", " "flags" ")" +to uniquely index any filesystem mapping record. +Classic non-sharing filesystems might be able to identify any record with only +.RI "(" "device" ", " "physical" ", " "flags" ")." +For example, if the low key is set to (8:0, 36864, 0, 0, 0), the filesystem will +only return records for extents starting at or above 36KiB on disk. +If the high key is set to (8:0, 1048576, 0, 0, 0), only records below 1MiB will +be returned. +The format of +.B fmr_device +in the keys must match the format of the same field in the output records, +as defined below. +By convention, the field +.B fsmap_head.fmh_keys[0] +must contain the low key and +.B fsmap_head.fmh_keys[1] +must contain the high key for the request. +.PP +For convenience, if +.B fmr_length +is set in the low key, it will be added to +.IR fmr_block " or " fmr_offset +as appropriate. +The caller can take advantage of this subtlety to set up subsequent calls +by copying +.B fsmap_head.fmh_recs[fsmap_head.fmh_entries - 1] +into the low key. +The function +.B fsmap_advance +provides this functionality. + +.SS Fields of struct fsmap +.PP +The +.I fmr_device +field uniquely identifies the underlying storage device. +If the +.B FMH_OF_DEV_T +flag is set in the header's +.I fmh_oflags +field, this field contains a +.B dev_t +from which major and minor numbers can be extracted. +If the flag is not set, this field contains a value that must be unique +for each unique storage device. + +.PP +The +.I fmr_physical +field contains the disk address of the extent in bytes. + +.PP +The +.I fmr_owner +field contains the owner of the extent. +This is an inode number unless +.B FMR_OF_SPECIAL_OWNER +is set in the +.I fmr_flags +field, in which case the value is determined by the filesystem. +See the section below about special owner values for more details. + +.PP +The +.I fmr_offset +field contains the logical address in the mapping record in bytes. +This field has no meaning if the +.BR FMR_OF_SPECIAL_OWNER " or " FMR_OF_EXTENT_MAP +flags are set in +.IR fmr_flags "." + +.PP +The +.I fmr_length +field contains the length of the extent in bytes. + +.PP +The +.I fmr_flags +field is a bitmask of extent state flags. +The bits are: +.RS 0.4i +.TP +.B FMR_OF_PREALLOC +The extent is allocated but not yet written. +.TP +.B FMR_OF_ATTR_FORK +This extent contains extended attribute data. +.TP +.B FMR_OF_EXTENT_MAP +This extent contains extent map information for the owner. +.TP +.B FMR_OF_SHARED +Parts of this extent may be shared. +.TP +.B FMR_OF_SPECIAL_OWNER +The +.I fmr_owner +field contains a special value instead of an inode number. +.TP +.B FMR_OF_LAST +This is the last record in the filesystem. +.RE + +.PP +The +.I fmr_reserved +field will be set to zero. + +.SS Special Owner Values +The following special owner values are generic to all filesystems: +.RS 0.4i +.TP +.B FMR_OWN_FREE +Free space. +.TP +.B FMR_OWN_UNKNOWN +This extent is in use but its owner is not known. +.TP +.B FMR_OWN_METADATA +This extent is filesystem metadata. +.RE + +XFS can return the following special owner values: +.RS 0.4i +.TP +.B XFS_FMR_OWN_FREE +Free space. +.TP +.B XFS_FMR_OWN_UNKNOWN +This extent is in use but its owner is not known. +.TP +.B XFS_FMR_OWN_FS +Static filesystem metadata which exists at a fixed address. +These are the AG superblock, the AGF, the AGFL, and the AGI headers. +.TP +.B XFS_FMR_OWN_LOG +The filesystem journal. +.TP +.B XFS_FMR_OWN_AG +Allocation group metadata, such as the free space btrees and the +reverse mapping btrees. +.TP +.B XFS_FMR_OWN_INOBT +The inode and free inode btrees. +.TP +.B XFS_FMR_OWN_INODES +Inode records. +.TP +.B XFS_FMR_OWN_REFC +Reference count information. +.TP +.B XFS_FMR_OWN_COW +This extent is being used to stage a copy-on-write. +.TP +.B XFS_FMR_OWN_DEFECTIVE: +This extent has been marked defective either by the filesystem or the +underlying device. +.RE + +ext4 can return the following special owner values: +.RS 0.4i +.TP +.B EXT4_FMR_OWN_FREE +Free space. +.TP +.B EXT4_FMR_OWN_UNKNOWN +This extent is in use but its owner is not known. +.TP +.B EXT4_FMR_OWN_FS +Static filesystem metadata which exists at a fixed address. +This is the superblock and the group descriptors. +.TP +.B EXT4_FMR_OWN_LOG +The filesystem journal. +.TP +.B EXT4_FMR_OWN_INODES +Inode records. +.TP +.B EXT4_FMR_OWN_BLKBM +Block bitmap. +.TP +.B EXT4_FMR_OWN_INOBM +Inode bitmap. +.RE + +.SH RETURN VALUE +On error, \-1 is returned, and +.I errno +is set to indicate the error. +.PP +.SH ERRORS +Error codes can be one of, but are not limited to, the following: +.TP +.B EINVAL +The array is not long enough, or a non-zero value was passed in one of the +fields that must be zero. +.TP +.B EFAULT +The pointer passed in was not mapped to a valid memory address. +.TP +.B EBADF +.IR fd +is not open for reading. +.TP +.B EPERM +This query is not allowed. +.TP +.B EOPNOTSUPP +The filesystem does not support this command. +.TP +.B EUCLEAN +The filesystem metadata is corrupt and needs repair. +.TP +.B EBADMSG +The filesystem has detected a checksum error in the metadata. +.TP +.B ENOMEM +Insufficient memory to process the request. + +.SH EXAMPLE +.TP +Please see io/fsmap.c in the xfsprogs distribution for a sample program. + +.SH CONFORMING TO +This API is Linux-specific. +Not all filesystems support it. +.fi +.in +.SH SEE ALSO +.BR ioctl (2)
