LVM2 topology support
v2 cover LVM2 topology support
LVM2 topology support to properly align the PV's pe_start using the associated 'alignment_offset' exposed in sysfs.
Provides manual configuration with pvcreate --dataalignmentoffset and auto config control via 'devices/data_alignment_offset_detection' in lvm.conf.
Documentation was added to the lvm.conf and pvcreate man pages.
Testing with pvcreate + vgextend + vgreduce + vgextend + vgreduce ... proved to maintain the pe_start from the initial pvcreate.
Also added another method for auto detection of --dataalignment using the topology information that is exposed in sysfs. Controlled via 'devices/data_alignment_detection' in lvm2.conf.
Mike Snitzer (4):
cmdline: support for bytes and sectors
pvcreate: --dataalignmentoffset
Add devices/data_alignment_offset_detection to lvm.conf. Add devices/data_alignment_detection to lvm.conf.
WHATS_NEW | 5 ++ doc/example.conf | 19 ++++++- lib/config/defaults.h | 2 + lib/device/device.c | 98 ++++++++++++++++++++++++++++++++++++++ lib/device/device.h | 9 ++++ lib/format1/format1.c | 1 + lib/format_pool/format_pool.c | 1 + lib/format_text/archiver.c | 2 +- lib/format_text/format-text.c | 24 ++++++++- lib/metadata/metadata-exported.h | 1 + lib/metadata/metadata.c | 36 +++++++++++++- lib/metadata/metadata.h | 1 + man/lvcreate.8.in | 6 +- man/lvextend.8.in | 4 +- man/lvm.conf.5.in | 22 +++++++- man/lvreduce.8.in | 4 +- man/lvresize.8.in | 4 +- man/pvcreate.8.in | 18 ++++++- man/vgchange.8.in | 4 +- man/vgcreate.8.in | 4 +- tools/args.h | 1 + tools/commands.h | 33 +++++++------ tools/lvmcmdline.c | 20 +++++-- tools/pvcreate.c | 23 ++++++++- tools/vgconvert.c | 2 +- 25 files changed, 292 insertions(+), 52 deletions(-)
PATCH v2 1/4 cmdline: support for bytes and sectors
Allow specifying commandline sizes in terms of bytes and sectors.
Update the man pages to uniformly document size units.
--- WHATS_NEW | 2 ++ man/lvcreate.8.in | 6 +++--- man/lvextend.8.in | 4 ++-- man/lvreduce.8.in | 4 ++-- man/lvresize.8.in | 4 ++-- man/vgchange.8.in | 4 ++-- man/vgcreate.8.in | 4 ++-- tools/commands.h | 26 +++++++++++++------------- tools/lvmcmdline.c | 20 ++++++++++++++------ 9 files changed, 42 insertions(+), 32 deletions(-)
WHATS_NEW +2 -0
@@ -1,5 +1,7 @@ Version 2.02.48 - =============================== + Update the man pages to uniformly document size units. + Allow specifying commandline sizes in terms of bytes and sectors. Explicitly request fallback to default major number in device mapper. Ignore suspended devices during repair. Call vgreduce --removemissing automatically to clean up bad PVs in dmeventd.
man/lvcreate.8.in +3 -3
@@ -9,7 +9,7 @@ lvcreate \- create a logical volume in an existing volume group [\-h|\-?|\-\-help] [\-i|\-\-stripes Stripes [\-I|\-\-stripesize StripeSize]] {\-l|\-\-extents LogicalExtentsNumber[%{VG|PVS|FREE}] | - \-L|\-\-size LogicalVolumeSize[kKmMgGtT]} + \-L|\-\-size LogicalVolumeSize[sSbBkKmMgGtTpPeE]} [\-M|\-\-persistent y|n] [\-\-minor minor] [\-m|\-\-mirrors Mirrors [\-\-nosync] [\-\-mirrorlog {disk|core}] [\-\-corelog] [\-R|\-\-regionsize MirrorLogRegionSize]]
@@ -23,7 +23,7 @@ VolumeGroupName [PhysicalVolumePath...] .br .B lvcreate {\-l|\-\-extents LogicalExtentsNumber[%{VG|FREE}] | - \-L|\-\-size LogicalVolumeSize[kKmMgGtT]} + \-L|\-\-size LogicalVolumeSize[sSbBkKmMgGtTpPeE]} [\-c|\-\-chunksize ChunkSize] \-n|\-\-name SnapshotLogicalVolumeName {{\-s|\-\-snapshot}
@@ -76,7 +76,7 @@ free space in the Volume Group with the suffix %FREE, or of the remaining free space for the specified PhysicalVolume(s) with the suffix %PVS, .TP -.I \-L, \-\-size LogicalVolumeSize[kKmMgGtTpPeE] +.I \-L, \-\-size LogicalVolumeSize[sSbBkKmMgGtTpPeE] Gives the size to allocate for the new logical volume. A size suffix of K for kilobytes, M for megabytes, G for gigabytes, T for terabytes, P for petabytes
man/lvextend.8.in +2 -2
@@ -7,7 +7,7 @@ lvextend \- extend the size of a logical volume [\-A|\-\-autobackup y|n] [\-d|\-\-debug] [\-h|\-?|\-\-help] [\-i|\-\-stripes Stripes [\-I|\-\-stripesize StripeSize]] {\-l|\-\-extents [+]LogicalExtentsNumber[%{VG|LV|PVS|FREE}] | -\-L|\-\-size [+]LogicalVolumeSize[kKmMgGtT]} +\-L|\-\-size [+]LogicalVolumeSize[sSbBkKmMgGtTpPeE]} [\-t|\-\-test] [\-v|\-\-verbose] LogicalVolumePath [PhysicalVolumePath...] .SH DESCRIPTION
@@ -32,7 +32,7 @@ free space for the specified PhysicalVolume(s) with the suffix %PVS, or as a percentage of the remaining free space in the Volume Group with the suffix %FREE. .TP -.I \-L, \-\-size [+]LogicalVolumeSize[kKmMgGtTpPeE] +.I \-L, \-\-size [+]LogicalVolumeSize[sSbBkKmMgGtTpPeE] Extend or set the logical volume size in units of megabytes. A size suffix of M for megabytes, G for gigabytes, T for terabytes, P for petabytes
man/lvreduce.8.in +2 -2
@@ -6,7 +6,7 @@ lvreduce \- reduce the size of a logical volume [\-A|\-\-autobackup y|n] [\-d|\-\-debug] [\-f|\-\-force] [\-h|\-?|\-\-help] {\-l|\-\-extents [\-]LogicalExtentsNumber[%{VG|LV|FREE}] | -\-L|\-\-size [\-]LogicalVolumeSize[kKmMgGtT]} +\-L|\-\-size [\-]LogicalVolumeSize[sSbBkKmMgGtTpPeE]} [\-t|\-\-test] [\-v|\-\-verbose] LogicalVolume[Path] .SH DESCRIPTION
@@ -46,7 +46,7 @@ in the Volume Group with the suffix %VG or relative to the existing size of the Logical Volume with the suffix %LV or as a percentage of the remaining free space in the Volume Group with the suffix %FREE. .TP -.I \-L, \-\-size [\-]LogicalVolumeSize[kKmMgGtTpPeE] +.I \-L, \-\-size [\-]LogicalVolumeSize[sSbBkKmMgGtTpPeE] Reduce or set the logical volume size in units of megabyte by default. A size suffix of k for kilobyte, m for megabyte, g for gigabytes, t for terabytes, p for petabytes
man/lvresize.8.in +2 -2
@@ -7,7 +7,7 @@ lvresize \- resize a logical volume [\-A|\-\-autobackup y|n] [\-d|\-\-debug] [\-h|\-?|\-\-help] [\-i|\-\-stripes Stripes [\-I|\-\-stripesize StripeSize]] {\-l|\-\-extents [+]LogicalExtentsNumber[%{VG|LV|PVS|FREE}] | -\-L|\-\-size [+]LogicalVolumeSize[kKmMgGtT]} +\-L|\-\-size [+]LogicalVolumeSize[sSbBkKmMgGtTpPeE]} [\-t|\-\-test] [\-v|\-\-verbose] LogicalVolumePath [PhysicalVolumePath...] .SH DESCRIPTION
@@ -36,7 +36,7 @@ the remaining free space of the PhysicalVolumes on the command line with the suffix %PVS, or as a percentage of the remaining free space in the Volume Group with the suffix %FREE. .TP -.I \-L, \-\-size [+|-]LogicalVolumeSize[kKmMgGtTpPeE] +.I \-L, \-\-size [+|-]LogicalVolumeSize[sSbBkKmMgGtTpPeE] Change or set the logical volume size in units of megabytes. A size suffix of M for megabytes, G for gigabytes, T for terabytes, P for petabytes
man/vgchange.8.in +2 -2
@@ -24,7 +24,7 @@ vgchange \- change attributes of a volume group .IR MaxPhysicalVolumes ] .RB [ \-P | \-\-partial] .RB [ \-s | \-\-physicalextentsize -.IR PhysicalExtentSize [ \fBkKmMgGtT\fR ]] +.IR PhysicalExtentSize [ \fBsSbBkKmMgGtTpPeE\fR ]] .RB [ \-\-refresh] .RB [ -t | \-\-test] .RB [ \-v | \-\-verbose]
@@ -108,7 +108,7 @@ for tool performance reasons, you should consider some use of \fB--metadatacopies 0\fP as described in \fBpvcreate(8)\fP. .TP -.BR \-s ", " \-\-physicalextentsize " " \fIPhysicalExtentSize\fR[\fBkKmMgGtT\fR] +.BR \-s ", " \-\-physicalextentsize " " \fIPhysicalExtentSize\fR[\fBsSbBkKmMgGtTpPeE\fR] Changes the physical extent size on physical volumes of this volume group. A size suffix (k for kilobytes up to t for terabytes) is optional, megabytes is the default if no suffix is present.
man/vgcreate.8.in +2 -2
@@ -17,7 +17,7 @@ vgcreate \- create a volume group .RB [ -p | \-\-maxphysicalvolumes .IR MaxPhysicalVolumes ] .RB [ \-s | \-\-physicalextentsize -.IR PhysicalExtentSize [ \fBkKmMgGtT\fR ]] +.IR PhysicalExtentSize [ \fBsSbBkKmMgGtTpPeE\fR ]] .RB [ \-t | \-\-test ] .RB [ \-v | \-\-verbose ] .RB [ \-\-version ]
@@ -66,7 +66,7 @@ for tool performance reasons, you should consider some use of \fB--metadatacopies 0\fP as described in \fBpvcreate(8)\fP. .TP -.BR \-s ", " \-\-physicalextentsize " " \fIPhysicalExtentSize\fR[\fBkKmMgGtT\fR] +.BR \-s ", " \-\-physicalextentsize " " \fIPhysicalExtentSize\fR[\fBsSbBkKmMgGtTpPeE\fR] Sets the physical extent size on physical volumes of this volume group. A size suffix (k for kilobytes up to t for terabytes) is optional, megabytes is the default if no suffix is present.
tools/commands.h +13 -13
@@ -19,7 +19,7 @@ xx(e2fsadm,"e2fsadm " "[-d|--debug] " "[-h|--help] " "[-n|--nofsck]" "\n" "\t{[-l|--extents] [+|-]LogicalExtentsNumber |" "\n" -"\t [-L|--size] [+|-]LogicalVolumeSize[kKmMgGtTpPeE]}" "\n" +"\t [-L|--size] [+|-]LogicalVolumeSize[sSbBkKmMgGtTpPeE]}" "\n" "\t[-t|--test] " "\n" "\t[-v|--verbose] " "\n" "\t[--version] " "\n"
@@ -131,7 +131,7 @@ xx(lvcreate,"\t[-h|-?|--help]\n" "\t[-i|--stripes Stripes [-I|--stripesize StripeSize]]\n" "\t{-l|--extents LogicalExtentsNumber |\n" -"\t -L|--size LogicalVolumeSize[kKmMgGtTpPeE]}\n" +"\t -L|--size LogicalVolumeSize[sSbBkKmMgGtTpPeE]}\n" "\t[-M|--persistent {y|n}] [--major major] [--minor minor]\n" "\t[-m|--mirrors Mirrors [--nosync] [{--mirrorlog {disk|core}|--corelog}]]\n" "\t[-n|--name LogicalVolumeName]\n"
@@ -157,7 +157,7 @@ xx(lvcreate,"\t[-h|-?|--help]\n" "\t[-i|--stripes Stripes [-I|--stripesize StripeSize]]\n" "\t{-l|--extents LogicalExtentsNumber[%{VG|LV|PVS|FREE}] |\n" -"\t -L|--size LogicalVolumeSize[kKmMgGtTpPeE]}\n" +"\t -L|--size LogicalVolumeSize[sSbBkKmMgGtTpPeE]}\n" "\t[-M|--persistent {y|n}] [--major major] [--minor minor]\n" "\t[-n|--name LogicalVolumeName]\n" "\t[-p|--permission {r|rw}]\n"
@@ -226,7 +226,7 @@ xx(lvextend,"\t[-h|--help]\n" "\t[-i|--stripes Stripes [-I|--stripesize StripeSize]]\n" "\t{-l|--extents [+]LogicalExtentsNumber[%{VG|PVS|FREE}] |\n" -"\t -L|--size [+]LogicalVolumeSize[kKmMgGtTpPeE]}\n" +"\t -L|--size [+]LogicalVolumeSize[sSbBkKmMgGtTpPeE]}\n" "\t[-m|--mirrors Mirrors]\n" "\t[-n|--nofsck]\n" "\t[-r|--resizefs]\n"
@@ -296,7 +296,7 @@ xx(lvreduce,"\t[-f|--force]\n" "\t[-h|--help]\n" "\t{-l|--extents [-]LogicalExtentsNumber[%{VG|LV|FREE}] |\n" -"\t -L|--size [-]LogicalVolumeSize[kKmMgGtTpPeE]}\n" +"\t -L|--size [-]LogicalVolumeSize[sSbBkKmMgGtTpPeE]}\n" "\t[-n|--nofsck]\n" "\t[-r|--resizefs]\n" "\t[-t|--test]\n"
@@ -349,7 +349,7 @@ xx(lvresize,"\t[-h|--help]\n" "\t[-i|--stripes Stripes [-I|--stripesize StripeSize]]\n" "\t{-l|--extents [+|-]LogicalExtentsNumber[%{VG|LV|PVS|FREE}] |\n" -"\t -L|--size [+|-]LogicalVolumeSize[kKmMgGtTpPeE]}\n" +"\t -L|--size [+|-]LogicalVolumeSize[sSbBkKmMgGtTpPeE]}\n" "\t[-n|--nofsck]\n" "\t[-r|--resizefs]\n" "\t[-t|--test]\n"
@@ -434,7 +434,7 @@ xx(pvresize,"pvresize " "\n" "\t[-d|--debug]" "\n" "\t[-h|-?|--help] " "\n" -"\t[--setphysicalvolumesize PhysicalVolumeSize[kKmMgGtTpPeE]" "\n" +"\t[--setphysicalvolumesize PhysicalVolumeSize[sSbBkKmMgGtTpPeE]" "\n" "\t[-t|--test] " "\n" "\t[-v|--verbose] " "\n" "\t[--version] " "\n"
@@ -466,9 +466,9 @@ xx(pvcreate,"\t[--labelsector sector] " "\n" "\t[-M|--metadatatype 1|2]" "\n" "\t[--metadatacopies #copies]" "\n" -"\t[--metadatasize MetadataSize[kKmMgGtTpPeE]]" "\n" -"\t[--dataalignment Alignment[kKmMgGtTpPeE]]" "\n" -"\t[--setphysicalvolumesize PhysicalVolumeSize[kKmMgGtTpPeE]" "\n" +"\t[--metadatasize MetadataSize[sSbBkKmMgGtTpPeE]]" "\n" +"\t[--dataalignment Alignment[sSbBkKmMgGtTpPeE]]" "\n" +"\t[--setphysicalvolumesize PhysicalVolumeSize[sSbBkKmMgGtTpPeE]" "\n" "\t[-t|--test] " "\n" "\t[-u|--uuid uuid] " "\n" "\t[-v|--verbose] " "\n"
@@ -681,7 +681,7 @@ xx(vgchange,"\t -x|--resizeable {y|n} |" "\n" "\t -l|--logicalvolume MaxLogicalVolumes |" "\n" "\t -p|--maxphysicalvolumes MaxPhysicalVolumes |" "\n" -"\t -s|--physicalextentsize PhysicalExtentSize[kKmMgGtTpPeE] |" "\n" +"\t -s|--physicalextentsize PhysicalExtentSize[sSbBkKmMgGtTpPeE] |" "\n" "\t --addtag Tag |\n" "\t --deltag Tag}\n" "\t[VolumeGroupName...]\n",
@@ -711,7 +711,7 @@ xx(vgconvert,"\t[--labelsector sector] " "\n" "\t[-M|--metadatatype 1|2]" "\n" "\t[--metadatacopies #copies]" "\n" -"\t[--metadatasize MetadataSize[kKmMgGtTpPeE]]" "\n" +"\t[--metadatasize MetadataSize[sSbBkKmMgGtTpPeE]]" "\n" "\t[-t|--test] " "\n" "\t[-v|--verbose] " "\n" "\t[--version] " "\n"
@@ -733,7 +733,7 @@ xx(vgcreate,"\t[-l|--maxlogicalvolumes MaxLogicalVolumes]" "\n" "\t[-M|--metadatatype 1|2] " "\n" "\t[-p|--maxphysicalvolumes MaxPhysicalVolumes] " "\n" -"\t[-s|--physicalextentsize PhysicalExtentSize[kKmMgGtTpPeE]] " "\n" +"\t[-s|--physicalextentsize PhysicalExtentSize[sSbBkKmMgGtTpPeE]] " "\n" "\t[-t|--test] " "\n" "\t[-v|--verbose]" "\n" "\t[--version] " "\n"
tools/lvmcmdline.c +14 -6
@@ -239,7 +239,7 @@ static int _size_arg(struct cmd_context *cmd __attribute((unused)), struct arg * {char *ptr;int i; -static const char *suffixes ="kmgtpe" ; +static const char *suffixes ="kmgtpebs" ;char *val;double v;
@@ -274,11 +274,19 @@ static int _size_arg(struct cmd_context *cmd __attribute((unused)), struct arg *if (i <0 )return 0 ; - -while (i-- >0 ) - v *=1024 ; - - v *=2 ; +else if (i ==7 ) { +/* sectors */ + v = v; + }else if (i ==6 ) { +/* bytes */ +if (v <512 ) +return 0 ; + v /=512 ; + }else { +while (i-- >0 ) + v *=1024 ; + v *=2 ; + } }else v *= factor;
Signature · 1 line
1.6.2.2
PATCH v2 2/4 pvcreate: --dataalignmentoffset
In reply to Mike SnitzerImplement pvcreate --dataalignmentoffset to pad offset of the start of the aligned data area (pe_start).
Mike Snitzer <[email protected]>
--- WHATS_NEW | 1 + lib/format1/format1.c | 1 + lib/format_pool/format_pool.c | 1 + lib/format_text/archiver.c | 2 +- lib/format_text/format-text.c | 24 +++++++++++++++++++++--- lib/metadata/metadata-exported.h | 1 + lib/metadata/metadata.c | 10 ++++++++-- lib/metadata/metadata.h | 1 + man/pvcreate.8.in | 18 ++++++++++++++++-- tools/args.h | 1 + tools/commands.h | 7 ++++--- tools/pvcreate.c | 23 +++++++++++++++++++++-- tools/vgconvert.c | 2 +- 13 files changed, 78 insertions(+), 14 deletions(-)
WHATS_NEW +1 -0
@@ -1,5 +1,6 @@ Version 2.02.48 - =============================== + Implement pvcreate --dataalignmentoffset to pad offset of pe_start. Update the man pages to uniformly document size units. Allow specifying commandline sizes in terms of bytes and sectors. Explicitly request fallback to default major number in device mapper.
lib/format1/format1.c +1 -0
@@ -296,6 +296,7 @@ static int _format1_pv_setup(const struct format_type *fmt, uint64_t pe_start, uint32_t extent_count, uint32_t extent_size, unsigned long data_alignment __attribute((unused)), + unsigned long data_alignment_offset __attribute((unused)), int pvmetadatacopies __attribute((unused)), uint64_t pvmetadatasize __attribute((unused)), struct dm_list *mdas __attribute((unused)), struct physical_volume *pv, struct volume_group *vg __attribute((unused)))
lib/format_pool/format_pool.c +1 -0
@@ -192,6 +192,7 @@ static int _pool_pv_setup(const struct format_type *fmt __attribute((unused)), uint32_t extent_count __attribute((unused)), uint32_t extent_size __attribute((unused)), unsigned long data_alignment __attribute((unused)), + unsigned long data_alignment_offset __attribute((unused)), int pvmetadatacopies __attribute((unused)), uint64_t pvmetadatasize __attribute((unused)), struct dm_list *mdas __attribute((unused)),
lib/format_text/archiver.c +1 -1
@@ -316,7 +316,7 @@ int backup_restore_vg(struct cmd_context *cmd, struct volume_group *vg)return 0 ; }if (!vg->fid->fmt->ops-> -pv_setup (vg->fid->fmt,UINT64_C (0 ),0 ,0 ,0 ,0UL , +pv_setup (vg->fid->fmt,UINT64_C (0 ),0 ,0 ,0 ,0 ,0UL ,UINT64_C (0 ), &vg->fid->metadata_areas, pv, vg)) {log_error ("Format-specific setup for %s failed" ,pv_dev_name (pv));
lib/format_text/format-text.c +21 -3
@@ -1175,12 +1175,13 @@ static int _text_scan(const struct format_type *fmt)Always have an mda between end-of-label and pe_align() boundary */ static int _mda_setup (const struct format_type *fmt,uint64_t pe_start,uint64_t pe_end, +unsigned long data_alignment_offset,int pvmetadatacopies,uint64_t pvmetadatasize,struct dm_list *mdas,struct physical_volume *pv,struct volume_group *vg __attribute((unused))) { -uint64_t mda_adjustment, disk_size, alignment; +uint64_t mda_adjustment, adjustment, disk_size, alignment;uint64_t start1, mda_size1;/* First area - start of disk */ uint64_t start2, mda_size2;/* Second area - end of disk */ uint64_t wipe_size =8 <<SECTOR_SHIFT ;
@@ -1251,6 +1252,22 @@ static int _mda_setup(const struct format_type *fmt,return 0 ; } +if (data_alignment_offset && !pe_start && !pe_end) { +/* +* proactively adjust pv->pe_start to allow for +* proper padding with data_alignment_offset +* - doing so in _text_pv_write is too late given +* its lack of data_alignment_offset info +*/ + pv->pe_start = (start1 + mda_size1) >>SECTOR_SHIFT ; +/* round up to pe_align boundry */ + adjustment = pv->pe_start % pv->pe_align; +if (adjustment) + pv->pe_start += (pv->pe_align - adjustment); + + pv->pe_start += data_alignment_offset; + } +if (pvmetadatacopies ==1 )return 1 ; }else
@@ -1598,6 +1615,7 @@ static struct metadata_area_ops _metadata_text_raw_ops = {static int _text_pv_setup (const struct format_type *fmt,uint64_t pe_start,uint32_t extent_count,uint32_t extent_size,unsigned long data_alignment, +unsigned long data_alignment_offset,int pvmetadatacopies,uint64_t pvmetadatasize,struct dm_list *mdas,struct physical_volume *pv,struct volume_group *vg)
@@ -1707,8 +1725,8 @@ static int _text_pv_setup(const struct format_type *fmt,if (extent_count) pe_end = pe_start + extent_count * extent_size -1 ; -if (!_mda_setup (fmt, pe_start, pe_end, pvmetadatacopies, - pvmetadatasize, mdas, pv, vg)) +if (!_mda_setup (fmt, pe_start, pe_end, data_alignment_offset, + pvmetadatacopies, pvmetadatasize, mdas, pv, vg)) return_0; }
lib/metadata/metadata-exported.h +1 -0
@@ -426,6 +426,7 @@ pv_t *pv_create(const struct cmd_context *cmd,struct id *id, uint64_t size, unsigned long data_alignment, + unsigned long data_alignment_offset, uint64_t pe_start, uint32_t existing_extent_count, uint32_t existing_extent_size,
lib/metadata/metadata.c +8 -2
@@ -48,6 +48,7 @@ static struct physical_volume *_pv_create(const struct format_type *fmt,struct device *dev, struct id *id, uint64_t size, unsigned long data_alignment, + unsigned long data_alignment_offset, uint64_t pe_start, uint32_t existing_extent_count, uint32_t existing_extent_size,
@@ -157,7 +158,7 @@ int add_pv_to_vg(struct volume_group *vg, const char *pv_name, pv->pe_alloc_count =0 ;if (!fid->fmt->ops->pv_setup (fid->fmt,UINT64_C (0 ),0 , - vg->extent_size,0 ,0UL ,UINT64_C (0 ), + vg->extent_size,0 ,0 ,0UL ,UINT64_C (0 ), &fid->metadata_areas, pv, vg)) {log_error ("Format-specific setup of physical volume '%s' " "failed." , pv_name);
@@ -787,6 +788,7 @@ int vg_split_mdas(struct cmd_context *cmd __attribute((unused)),* @id: PV UUID to use for initialization * @size: size of the PV in sectors * @data_alignment: requested alignment of data +* @data_alignment_offset: requested offset to aligned data * @pe_start: physical extent start * @existing_extent_count * @existing_extent_size
@@ -805,13 +807,15 @@ pv_t *pv_create(const struct cmd_context *cmd, struct device *dev, struct id *id, uint64_t size, unsigned long data_alignment, + unsigned long data_alignment_offset, uint64_t pe_start, uint32_t existing_extent_count, uint32_t existing_extent_size, int pvmetadatacopies, uint64_t pvmetadatasize, struct dm_list *mdas) { - return _pv_create(cmd->fmt, dev, id, size, data_alignment, pe_start, + return _pv_create(cmd->fmt, dev, id, size, + data_alignment, data_alignment_offset, pe_start, existing_extent_count, existing_extent_size, pvmetadatacopies,
@@ -856,6 +860,7 @@ static struct physical_volume *_pv_create(const struct format_type *fmt,struct device *dev, struct id *id, uint64_t size, unsigned long data_alignment, + unsigned long data_alignment_offset, uint64_t pe_start, uint32_t existing_extent_count, uint32_t existing_extent_size,
@@ -907,6 +912,7 @@ static struct physical_volume *_pv_create(const struct format_type *fmt,if (!fmt->ops->pv_setup (fmt, pe_start, existing_extent_count, existing_extent_size, data_alignment, + data_alignment_offset, pvmetadatacopies, pvmetadatasize, mdas, pv,NULL )) { log_error("%s: Format-specific setup of physical volume "
lib/metadata/metadata.h +1 -0
@@ -213,6 +213,7 @@ struct format_handler {int (*pv_setup) (const struct format_type * fmt,uint64_t pe_start,uint32_t extent_count,uint32_t extent_size,unsigned long data_alignment, +unsigned long data_alignment_offset,int pvmetadatacopies,uint64_t pvmetadatasize,struct dm_list * mdas,struct physical_volume * pv,struct volume_group * vg);
man/pvcreate.8.in +16 -2
@@ -14,6 +14,7 @@ pvcreate \- initialize a disk or partition for use by LVM .RB [ \-\-metadatacopies #copies ] .RB [ \-\-metadatasize size ] .RB [ \-\-dataalignment alignment ] +.RB [ \-\-dataalignmentoffset offset ] .RB [ \-\-restorefile file ] .RB [ \-\-setphysicalvolumesize size ] .RB [ \-u | \-\-uuid uuid ]
@@ -97,7 +98,12 @@ the Volume Group with \fBvgcreate\fP. .sp To see the location of the first Physical Extent of an existing Physical Volume use \fBpvs -o +pe_start\fP . It will be a multiple of the requested -\fBdata_alignment\fP. +\fBalignment\fP plus the \fBalignment_offset\fP from +\fBdata_alignment_offset_detection\fP (if enabled in \fBlvm.conf\fP) or +\fB--dataalignmentoffset\fP. +.TP +.BR \-\-dataalignmentoffset " alignment_offset" +The offset of the start of the aligned data area will include \fBalignment_offset\fP. .TP .BR \-\-metadatacopies " copies" The number of metadata areas to set aside on each PV. Currently
@@ -128,13 +134,21 @@ in the source). Use with care. .TP .BR \-\-setphysicalvolumesize " size" Overrides the automatically-detected size of the PV. Use with care. -.SH Example +.SH EXAMPLES Initialize partition #4 on the third SCSI disk and the entire fifth SCSI disk for later use by LVM: .sp .B pvcreate /dev/sdc4 /dev/sde .sp +If the 2nd SCSI disk is a 4KB sector drive that compensates for windows +partitioning (sector 7 is the lowest aligned logical block, the 4KB +sectors start at LBA -1, and consequently sector 63 is aligned on a 4KB +boundary) manually account for this when initializing for use by LVM: +.sp +.B pvcreate --dataalignmentoffset 7S /dev/sdb +.sp .SH SEE ALSO +.BR lvm.conf (5), .BR lvm (8), .BR vgcreate (8), .BR vgextend (8),
tools/args.h +1 -0
@@ -59,6 +59,7 @@ arg(nameprefixes_ARG, '\0', "nameprefixes", NULL, 0) arg(unquoted_ARG, '\0', "unquoted", NULL, 0) arg(rows_ARG, '\0', "rows", NULL, 0) arg(dataalignment_ARG, '\0', "dataalignment", size_kb_arg, 0) +arg(dataalignmentoffset_ARG, '\0', "dataalignmentoffset", size_kb_arg, 0) arg(virtualoriginsize_ARG, '\0', "virtualoriginsize", size_mb_arg, 0)arg (virtualsize_ARG,'\0' ,"virtualsize" , size_mb_arg,0 )
tools/commands.h +4 -3
@@ -468,6 +468,7 @@ xx(pvcreate,"\t[--metadatacopies #copies]" "\n" "\t[--metadatasize MetadataSize[sSbBkKmMgGtTpPeE]]" "\n" "\t[--dataalignment Alignment[sSbBkKmMgGtTpPeE]]" "\n" +"\t[--dataalignmentoffset AlignmentOffset[sSbBkKmMgGtTpPeE]]" "\n" "\t[--setphysicalvolumesize PhysicalVolumeSize[sSbBkKmMgGtTpPeE]" "\n" "\t[-t|--test] " "\n" "\t[-u|--uuid uuid] " "\n"
@@ -477,9 +478,9 @@ xx(pvcreate, "\t[--version] " "\n" "\tPhysicalVolume [PhysicalVolume...]\n", - dataalignment_ARG, force_ARG, test_ARG, labelsector_ARG, metadatatype_ARG, - metadatacopies_ARG, metadatasize_ARG, physicalvolumesize_ARG, - restorefile_ARG, uuidstr_ARG, yes_ARG, zero_ARG) + dataalignment_ARG, dataalignmentoffset_ARG, force_ARG, test_ARG, + labelsector_ARG, metadatatype_ARG, metadatacopies_ARG, metadatasize_ARG, + physicalvolumesize_ARG, restorefile_ARG, uuidstr_ARG, yes_ARG, zero_ARG) xx(pvdata, "Display the on-disk metadata for physical volume(s)",
tools/pvcreate.c +21 -2
@@ -20,6 +20,7 @@ struct pvcreate_params {int zero;uint64_t size;uint64_t data_alignment; +uint64_t data_alignment_offset;int pvmetadatacopies;uint64_t pvmetadatasize;int64_t labelsector;
@@ -203,8 +204,8 @@ static int pvcreate_single(struct cmd_context *cmd, const char *pv_name,dm_list_init (&mdas);if (!(pv =pv_create (cmd, dev, pp->idp, pp->size, - pp->data_alignment, pp->pe_start, - pp->extent_count, pp->extent_size, + pp->data_alignment, pp->data_alignment_offset, + pp->pe_start, pp->extent_count, pp->extent_size, pp->pvmetadatacopies, pp->pvmetadatasize,&mdas))) {log_error ("Failed to setup physical volume \"%s\"" , pv_name);
@@ -377,6 +378,24 @@ static int pvcreate_validate_params(struct cmd_context *cmd, pp->data_alignment =0 ; } +if (arg_sign_value (cmd, dataalignmentoffset_ARG,0 ) ==SIGN_MINUS ) { +log_error ("Physical volume data alignment offset may not be negative" ); +return 0 ; + } + pp->data_alignment_offset =arg_uint64_value (cmd, dataalignmentoffset_ARG,UINT64_C (0 )); + +if (pp->data_alignment_offset >ULONG_MAX ) { +log_error ("Physical volume data alignment offset is too big." ); +return 0 ; + } + +if (pp->data_alignment_offset && pp->pe_start) { +log_warn ("WARNING: Ignoring data alignment offset %" PRIu64 +" incompatible with --restorefile value (%" + PRIu64")." , pp->data_alignment_offset, pp->pe_start); + pp->data_alignment_offset =0 ; + } +if (arg_sign_value (cmd, metadatasize_ARG,0 ) ==SIGN_MINUS ) {log_error ("Metadata size may not be negative" );return 0 ;
tools/vgconvert.c +1 -1
@@ -120,7 +120,7 @@ static int vgconvert_single(struct cmd_context *cmd, const char *vg_name,dm_list_init (&mdas);if (!(pv =pv_create (cmd,pv_dev (existing_pv), - &existing_pv->id, size,0 , + &existing_pv->id, size,0 ,0 , pe_start,pv_pe_count (existing_pv),pv_pe_size (existing_pv), pvmetadatacopies, pvmetadatasize, &mdas))) {
Signature · 1 line
1.6.2.2
PATCH v2 3/4 Add devices/data_alignment_offset_detection to lvm.conf.
In reply to Mike SnitzerIf the pvcreate --dataalignmentoffset option is not specified the offset of the start of a PV's aligned data area will be padded with the associated 'alignment_offset' exposed in sysfs (unless devices/data_alignment_offset_detection is disabled in lvm.conf).
--- WHATS_NEW | 1 + doc/example.conf | 5 +++ lib/config/defaults.h | 1 + lib/device/device.c | 72 +++++++++++++++++++++++++++++++++++++++++++++++ lib/device/device.h | 3 ++ lib/metadata/metadata.c | 8 +++++ man/lvm.conf.5.in | 9 +++++- 7 files changed, 98 insertions(+), 1 deletions(-)
WHATS_NEW +1 -0
@@ -1,5 +1,6 @@ Version 2.02.48 - =============================== + Add devices/data_alignment_offset_detection to lvm.conf. Implement pvcreate --dataalignmentoffset to pad offset of pe_start. Update the man pages to uniformly document size units. Allow specifying commandline sizes in terms of bytes and sectors.
doc/example.conf +5 -0
@@ -104,6 +104,11 @@ devices { # Set to 0 for the default alignment of 64KB or page size, if larger. data_alignment = 0 + # By default, the offset of the start of a PV's aligned data area + # will be padded with the 'alignment_offset' exposed in sysfs. + # 1 enables; 0 disables. + data_alignment_offset_detection = 1 + # If, while scanning the system for PVs, LVM2 encounters a device-mapper # device that has its I/O suspended, it waits for it to become accessible. # Set this to 1 to skip such devices. This should only be needed
lib/config/defaults.h +1 -0
@@ -34,6 +34,7 @@#define DEFAULT_MD_COMPONENT_DETECTION 1#define DEFAULT_MD_CHUNK_ALIGNMENT 1#define DEFAULT_IGNORE_SUSPENDED_DEVICES 1 +#define DEFAULT_DATA_ALIGNMENT_OFFSET_DETECTION 1#define DEFAULT_LOCK_DIR "/var/lock/lvm"#define DEFAULT_LOCKING_LIB "liblvm2clusterlock.so"
lib/device/device.c +72 -0
75 lines not on this page · raw message
lib/device/device.h +3 -0
@@ -100,4 +100,7 @@ unsigned long dev_md_chunk_size(const char *sysfs_dir, struct device *dev);int is_partitioned_dev (struct device *dev); +unsigned long dev_alignment_offset (const char *sysfs_dir, +struct device *dev); +#endif
lib/metadata/metadata.c +8 -0
@@ -910,6 +910,14 @@ static struct physical_volume *_pv_create(const struct format_type *fmt, pv->fmt = fmt; pv->vg_name = fmt->orphan_vg_name; +if (!pe_start && !data_alignment_offset && +find_config_tree_bool (pv->fmt->cmd, +"devices/data_alignment_offset_detection" , +DEFAULT_DATA_ALIGNMENT_OFFSET_DETECTION )) { + data_alignment_offset = +dev_alignment_offset (pv->fmt->cmd->sysfs_dir, pv->dev); + } + if (!fmt->ops->pv_setup(fmt, pe_start, existing_extent_count, existing_extent_size, data_alignment, data_alignment_offset,
man/lvm.conf.5.in +8 -1
@@ -142,10 +142,17 @@ when creating a new Physical Volume using the \fBlvm2\fP format. If a Physical Volume is placed directly upon an md device and \fBmd_chunk_alignment\fP is enabled this parameter is ignored. Set to 0 to use the default alignment of 64KB or the page size, if larger. +.IP +\fBdata_alignment_offset_detection\fP \(em If set to 1, and your kernel +provides topology information in sysfs for the Physical Volume, the +start of the aligned data area of the Physical Volume will be padded +with the alignment_offset exposed in sysfs. .sp To see the location of the first Physical Extent of an existing Physical Volume use \fBpvs -o +pe_start\fP . It will be a multiple of the requested -\fBdata_alignment\fP. +\fBdata_alignment\fP plus the alignment_offset from +\fBdata_alignment_offset_detection\fP (if enabled) or the pvcreate +commandline. .TP \fBlog\fP \(em Default log settings .IP
Signature · 1 line
1.6.2.2
PATCH v2 4/4 Add devices/data_alignment_detection to lvm.conf.
In reply to Mike SnitzerAdds 'data_alignment_detection' config option to the devices section of lvm.conf. If your kernel provides topology information in sysfs for the Physical Volume, the start of data area will be aligned on a multiple of the ?minimum_io_size? or ?optimal_io_size? exposed in sysfs.
minimum_io_size is used if optimal_io_size is undefined (0). If both md_chunk_alignment and data_alignment_detection are enabled the result of data_alignment_detection is used.
--- WHATS_NEW | 1 + doc/example.conf | 14 +++++++++++--- lib/config/defaults.h | 1 + lib/device/device.c | 26 ++++++++++++++++++++++++++ lib/device/device.h | 6 ++++++ lib/metadata/metadata.c | 18 ++++++++++++++++++ man/lvm.conf.5.in | 13 +++++++++++-- 7 files changed, 74 insertions(+), 5 deletions(-)
WHATS_NEW +1 -0
@@ -1,5 +1,6 @@ Version 2.02.48 - =============================== + Add devices/data_alignment_detection to lvm.conf. Add devices/data_alignment_offset_detection to lvm.conf. Implement pvcreate --dataalignmentoffset to pad offset of pe_start. Update the man pages to uniformly document size units.
doc/example.conf +11 -3
@@ -94,13 +94,21 @@ devices { md_component_detection = 1 # By default, if a PV is placed directly upon an md device, LVM2 - # will align its data blocks with the the chunk_size exposed in sysfs. + # will align its data blocks with the chunk_size exposed in sysfs. # 1 enables; 0 disables. md_chunk_alignment = 1 + # By default, the start of a PV's data area will be aligned with + # the 'minimum_io_size' or 'optimal_io_size' exposed in sysfs. + # minimum_io_size is used if optimal_io_size is undefined (0). + # This offers a superset of what md_chunk_alignment detects + # (aka minimum_io_size) and is therefore preferred over it. + # 1 enables; 0 disables. + data_alignment_detection = 1 + # Alignment (in KB) of start of data area when creating a new PV. - # If a PV is placed directly upon an md device and md_chunk_alignment is - # enabled this parameter is ignored. + # If a PV is placed directly upon an md device and md_chunk_alignment or + # data_alignment_detection is enabled this parameter is ignored. # Set to 0 for the default alignment of 64KB or page size, if larger. data_alignment = 0
lib/config/defaults.h +1 -0
@@ -35,6 +35,7 @@#define DEFAULT_MD_CHUNK_ALIGNMENT 1#define DEFAULT_IGNORE_SUSPENDED_DEVICES 1#define DEFAULT_DATA_ALIGNMENT_OFFSET_DETECTION 1 +#define DEFAULT_DATA_ALIGNMENT_DETECTION 1#define DEFAULT_LOCK_DIR "/var/lock/lvm"#define DEFAULT_LOCKING_LIB "liblvm2clusterlock.so"
lib/device/device.c +26 -0
@@ -341,6 +341,20 @@ unsigned long dev_alignment_offset(const char *sysfs_dir, sysfs_dir, dev); } +unsigned long dev_minimum_io_size (const char *sysfs_dir, +struct device *dev) +{ +return __dev_topology_attribute ("minimum_io_size" , + sysfs_dir, dev); +} + +unsigned long dev_optimal_io_size (const char *sysfs_dir, +struct device *dev) +{ +return __dev_topology_attribute ("optimal_io_size" , + sysfs_dir, dev); +} +#else unsigned long dev_alignment_offset(const char *sysfs_dir,
@@ -349,4 +363,16 @@ unsigned long dev_alignment_offset(const char *sysfs_dir, return 0UL; } +unsigned long dev_minimum_io_size (const char *sysfs_dir, +struct device *dev) +{ +return 0UL ; +} + +unsigned long dev_optimal_io_size (const char *sysfs_dir, +struct device *dev) +{ +return 0UL ; +} +#endif
lib/device/device.h +6 -0
@@ -103,4 +103,10 @@ int is_partitioned_dev(struct device *dev);unsigned long dev_alignment_offset (const char *sysfs_dir,struct device *dev); +unsigned long dev_minimum_io_size (const char *sysfs_dir, +struct device *dev); + +unsigned long dev_optimal_io_size (const char *sysfs_dir, +struct device *dev); +#endif
lib/metadata/metadata.c +18 -0
@@ -90,6 +90,24 @@ unsigned long set_pe_align(struct physical_volume *pv, unsigned long data_alignmdev_md_chunk_size (pv->fmt->cmd->sysfs_dir, pv->dev)); +/* +* Align to topology's minimum_io_size or optimal_io_size if present +* - minimum_io_size offers the smallest request the device can perform +* without incurring a read-modify-write penalty (e.g. MD's chunk size) +* - optimal_io_size offers the device's preferred unit of receiving I/O +* (e.g. MD's stripe width) +*/ +if (find_config_tree_bool (pv->fmt->cmd,"devices/data_alignment_detection" , +DEFAULT_DATA_ALIGNMENT_DETECTION )) { + pv->pe_align =MAX (pv->pe_align, +dev_minimum_io_size (pv->fmt->cmd->sysfs_dir, + pv->dev)); + + pv->pe_align =MAX (pv->pe_align, +dev_optimal_io_size (pv->fmt->cmd->sysfs_dir, + pv->dev)); + } +log_very_verbose ("%s: Setting PE alignment to %lu sectors." ,dev_name (pv->dev), pv->pe_align);
man/lvm.conf.5.in +11 -2
@@ -134,13 +134,22 @@ superblocks. This doesn't always work satisfactorily e.g. if a device has been reused without wiping the md superblocks first. .IP \fBmd_chunk_alignment\fP \(em If set to 1, and a Physical Volume is placed -directly upon an md device, LVM2 will align its data blocks with the the +directly upon an md device, LVM2 will align its data blocks with the chunk_size exposed in sysfs. .IP +\fBdata_alignment_detection\fP \(em If set to 1, and your kernel provides +topology information in sysfs for the Physical Volume, the start of data +area will be aligned on a multiple of the ?minimum_io_size? or +?optimal_io_size? exposed in sysfs. minimum_io_size is used if +optimal_io_size is undefined (0). If both \fBmd_chunk_alignment\fP and +\fBdata_alignment_detection\fP are enabled the result of +\fBdata_alignment_detection\fP is used. +.IP \fBdata_alignment\fP \(em Default alignment (in KB) of start of data area when creating a new Physical Volume using the \fBlvm2\fP format. If a Physical Volume is placed directly upon an md device and -\fBmd_chunk_alignment\fP is enabled this parameter is ignored. +\fBmd_chunk_alignment\fP or \fBdata_alignment_detection\fP is enabled +this parameter is ignored. Set to 0 to use the default alignment of 64KB or the page size, if larger. .IP \fBdata_alignment_offset_detection\fP \(em If set to 1, and your kernel
Signature · 1 line
1.6.2.2