December 2016 - lvm2-commits - Fedora mailing-lists

master - man: fix name of 'write_time' field in dmstats.8.in
by Bryn Reeves 26 Dec '16

26 Dec '16

Gitweb: http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=e75f0b7c771ac6b3… Commit: e75f0b7c771ac6b3a2925af92a9f3098fa8381a7 Parent: 96a1943fb8d208aa84c386cf48f2b62038b09e2e Author: Bryn M. Reeves <bmr(a)redhat.com> AuthorDate: Sun Dec 25 17:36:35 2016 +0000 Committer: Bryn M. Reeves <bmr(a)redhat.com> CommitterDate: Sun Dec 25 17:36:35 2016 +0000 man: fix name of 'write_time' field in dmstats.8.in --- man/dmstats.8.in | 2 +- 1 files changed, 1 insertions(+), 1 deletions(-) diff --git a/man/dmstats.8.in b/man/dmstats.8.in index fbc7811..c3ede01 100644 --- a/man/dmstats.8.in +++ b/man/dmstats.8.in @@ -878,7 +878,7 @@ Count of writes merged this interval. .B write_sector_count Count of 512 byte sectors written this interval. .TP -.B write_nsecs +.B write_time Accumulated duration of all write requests (ns). .TP .B in_progress_count

1 0

master - tests: update test
by Zdenek Kabelac 23 Dec '16

23 Dec '16

Gitweb: http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=96a1943fb8d208aa… Commit: 96a1943fb8d208aa84c386cf48f2b62038b09e2e Parent: 14902d173917e814061a80f16ea54435abba1f91 Author: Zdenek Kabelac <zkabelac(a)redhat.com> AuthorDate: Fri Dec 23 12:52:45 2016 +0100 Committer: Zdenek Kabelac <zkabelac(a)redhat.com> CommitterDate: Fri Dec 23 13:16:35 2016 +0100 tests: update test lvm2 now correctly reports thin_id after action of merged thin, but before physical metadata update as we know the merge has happened. --- test/shell/thin-vglock.sh | 15 ++++++++++++--- 1 files changed, 12 insertions(+), 3 deletions(-) diff --git a/test/shell/thin-vglock.sh b/test/shell/thin-vglock.sh index e59636d..40f3adf 100644 --- a/test/shell/thin-vglock.sh +++ b/test/shell/thin-vglock.sh @@ -1,5 +1,5 @@ #!/bin/sh -# Copyright (C) 2014 Red Hat, Inc. All rights reserved. +# Copyright (C) 2014-2016 Red Hat, Inc. All rights reserved. # # This copyrighted material is made available to anyone wishing to use, # modify, copy, or redistribute it subject to the terms and conditions @@ -37,15 +37,24 @@ check lv_field $vg/snap thin_id "3" lvconvert --merge $vg/snap umount mnt + +check lv_field $vg/$lv1 thin_id "1" +check lv_field $vg/pool transaction_id "3" + vgchange -an $vg # Check reboot case vgchange -ay --sysinit $vg -# Metadata are still not updated (--poll n) -check lv_field $vg/$lv1 thin_id "1" + +# Check correct thin_id is shown after activation +# even when metadata were not yet physically modified. +# Merge take its place during activation, +# but pool transaction_id still needs metadata update. +check lv_field $vg/$lv1 thin_id "3" check lv_field $vg/pool transaction_id "3" # Check the metadata are updated after refresh +# vgchange --refresh $vg check lv_field $vg/$lv1 thin_id "3" check lv_field $vg/pool transaction_id "4"

1 0

master - validation: temporarily let pass linear with chunk_size
by Zdenek Kabelac 23 Dec '16

23 Dec '16

Gitweb: http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=14902d173917e814… Commit: 14902d173917e814061a80f16ea54435abba1f91 Parent: 95d68f1d0e16f553f4f12046ceb7b6ff8d251336 Author: Zdenek Kabelac <zkabelac(a)redhat.com> AuthorDate: Fri Dec 23 12:40:40 2016 +0100 Committer: Zdenek Kabelac <zkabelac(a)redhat.com> CommitterDate: Fri Dec 23 13:16:06 2016 +0100 validation: temporarily let pass linear with chunk_size Old pool format seems to be setting chunk_size. For now let validation pass with this. --- lib/metadata/merge.c | 2 ++ 1 files changed, 2 insertions(+), 0 deletions(-) diff --git a/lib/metadata/merge.c b/lib/metadata/merge.c index a1c9455..59c0885 100644 --- a/lib/metadata/merge.c +++ b/lib/metadata/merge.c @@ -440,6 +440,8 @@ static void _check_lv_segment(struct logical_volume *lv, struct lv_segment *seg, } if (!seg_is_pool(seg) && + /* FIXME: format_pool/import_export.c _add_linear_seg() sets chunk_size */ + !seg_is_linear(seg) && !seg_is_snapshot(seg)) { if (seg->chunk_size) seg_error("sets chunk_size");

1 0

master - lvchange: allow a transiently failed RaidLV to be refreshed
by Heinz Mauelshagen 23 Dec '16

23 Dec '16

Gitweb: http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=95d68f1d0e16f553… Commit: 95d68f1d0e16f553f4f12046ceb7b6ff8d251336 Parent: 62be9c8de430a054d5de9b652949f58a684a0cf6 Author: Heinz Mauelshagen <heinzm(a)redhat.com> AuthorDate: Fri Dec 23 03:35:13 2016 +0100 Committer: Heinz Mauelshagen <heinzm(a)redhat.com> CommitterDate: Fri Dec 23 03:41:32 2016 +0100 lvchange: allow a transiently failed RaidLV to be refreshed Add to commits 87117c2b2546 and 0b8bf73a63d8 to avoid refreshing two times altogether, thus avoiding issues related to clustered, remotely activated RaidLV. Avoid need to repeat "lvchange --refresh RaidLV" two times as a workaround to refresh a RaidLV. Fix handles removal of temporary *-missing-* devices created for any missing segments in RAID SubLVs during activation. Because the kernel dm-raid target isn't able to handle transiently failing devices properly we need "[dm-devel][PATCH] dm raid: fix transient device failure processing" as well. test: add lvchange-raid-transient-failures.sh and enhance lvconvert-raid.sh Resolves: rhbz1025322 Related: rhbz1265191 Related: rhbz1399844 Related: rhbz1404425 --- lib/activate/activate.c | 75 ++++++++++++++++++++++++ lib/activate/activate.h | 2 + lib/metadata/lv_manip.c | 34 +++-------- test/shell/lvchange-raid-transient-failures.sh | 69 ++++++++++++++++++++++ test/shell/lvconvert-raid.sh | 19 +++++- 5 files changed, 171 insertions(+), 28 deletions(-) diff --git a/lib/activate/activate.c b/lib/activate/activate.c index b7009e6..742d838 100644 --- a/lib/activate/activate.c +++ b/lib/activate/activate.c @@ -358,6 +358,10 @@ int lv_mknodes(struct cmd_context *cmd, const struct logical_volume *lv) { return 1; } +int lv_deactivate_any_missing_subdevs(const struct logical_volume *lv) +{ + return 1; +} int pv_uses_vg(struct physical_volume *pv, struct volume_group *vg) { @@ -2573,6 +2577,77 @@ int lv_mknodes(struct cmd_context *cmd, const struct logical_volume *lv) return r; } +/* Remove any existing, closed mapped device by @name */ +static int _remove_dm_dev_by_name(const char *name) +{ + int r = 0; + struct dm_task *dmt; + struct dm_info info; + + if (!(dmt = dm_task_create(DM_DEVICE_INFO))) + return_0; + + /* Check, if the device exists. */ + if (dm_task_set_name(dmt, name) && dm_task_run(dmt) && dm_task_get_info(dmt, &info)) { + dm_task_destroy(dmt); + + /* Ignore non-existing or open dm devices */ + if (!info.exists || info.open_count) + return 1; + + if (!(dmt = dm_task_create(DM_DEVICE_REMOVE))) + return_0; + + if (dm_task_set_name(dmt, name)) + r = dm_task_run(dmt); + } + + dm_task_destroy(dmt); + + return r; +} + +/* Work all segments of @lv removing any existing, closed "*-missing_N_0" sub devices. */ +static int _lv_remove_any_missing_subdevs(struct logical_volume *lv) +{ + if (lv) { + uint32_t seg_no = 0; + char name[257]; + struct lv_segment *seg; + + dm_list_iterate_items(seg, &lv->segments) { + if (seg->area_count != 1) + return_0; + if (dm_snprintf(name, sizeof(name), "%s-%s-missing_%u_0", seg->lv->vg->name, seg->lv->name, seg_no) < 0) + return 0; + if (!_remove_dm_dev_by_name(name)) + return 0; + + seg_no++; + } + } + + return 1; +} + +/* Remove any "*-missing_*" sub devices added by the activation layer for an rmate/rimage missing PV mapping */ +int lv_deactivate_any_missing_subdevs(const struct logical_volume *lv) +{ + uint32_t s; + struct lv_segment *seg = first_seg(lv); + + for (s = 0; s < seg->area_count; s++) { + if (seg_type(seg, s) == AREA_LV && + !_lv_remove_any_missing_subdevs(seg_lv(seg, s))) + return 0; + if (seg->meta_areas && seg_metatype(seg, s) == AREA_LV && + !_lv_remove_any_missing_subdevs(seg_metalv(seg, s))) + return 0; + } + + return 1; +} + /* * Does PV use VG somewhere in its construction? * Returns 1 on failure. diff --git a/lib/activate/activate.h b/lib/activate/activate.h index db8d997..85c1521 100644 --- a/lib/activate/activate.h +++ b/lib/activate/activate.h @@ -124,6 +124,8 @@ int lv_deactivate(struct cmd_context *cmd, const char *lvid_s, const struct logi int lv_mknodes(struct cmd_context *cmd, const struct logical_volume *lv); +int lv_deactivate_any_missing_subdevs(const struct logical_volume *lv); + /* * Returns 1 if info structure has been populated, else 0 on failure. * When lvinfo* is NULL, it returns 1 if the device is locally active, 0 otherwise. diff --git a/lib/metadata/lv_manip.c b/lib/metadata/lv_manip.c index 3862f11..e5808ec 100644 --- a/lib/metadata/lv_manip.c +++ b/lib/metadata/lv_manip.c @@ -1419,35 +1419,19 @@ static int _lv_refresh_suspend_resume(const struct logical_volume *lv) int lv_refresh_suspend_resume(const struct logical_volume *lv) { + if (!_lv_refresh_suspend_resume(lv)) + return 0; + /* - * FIXME: - * - * in case of RAID, refresh the SubLVs before - * refreshing the top-level one in order to cope - * with transient failures of SubLVs. + * Remove any transiently activated error + * devices which arean't used any more. */ - if (lv_is_raid(lv)) { - if (vg_is_clustered(lv->vg) && - lv_is_active_remotely(lv)) { - if (!_lv_refresh_suspend_resume(lv)) - return 0; - } else { - uint32_t s; - struct lv_segment *seg = first_seg(lv); - - for (s = 0; s < seg->area_count; s++) { - if (seg_type(seg, s) == AREA_LV && - !_lv_refresh_suspend_resume(seg_lv(seg, s))) - return 0; - if (seg->meta_areas && - seg_metatype(seg, s) == AREA_LV && - !_lv_refresh_suspend_resume(seg_metalv(seg, s))) - return 0; - } - } + if (lv_is_raid(lv) && !lv_deactivate_any_missing_subdevs(lv)) { + log_error("Failed to remove temporary SubLVs from %s", display_lvname(lv)); + return 0; } - return _lv_refresh_suspend_resume(lv); + return 1; } /* diff --git a/test/shell/lvchange-raid-transient-failures.sh b/test/shell/lvchange-raid-transient-failures.sh new file mode 100644 index 0000000..844f217 --- /dev/null +++ b/test/shell/lvchange-raid-transient-failures.sh @@ -0,0 +1,69 @@ +#!/bin/sh +# Copyright (C) 2016 Red Hat, Inc. All rights reserved. +# +# This copyrighted material is made available to anyone wishing to use, +# modify, copy, or redistribute it subject to the terms and conditions +# of the GNU General Public License v.2. +# +# You should have received a copy of the GNU General Public License +# along with this program; if not, write to the Free Software Foundation, +# Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + +SKIP_WITH_LVMLOCKD=1 +SKIP_WITH_LVMPOLLD=1 + +. lib/inittest + +aux have_raid 1 10 1 || skip +aux prepare_vg 6 + +# +# FIXME: add multi-segment leg tests +# + +function _check_raid +{ + local vg=$1 + shift + local lv=$1 + shift + local fail=$1 + shift + local good=$1 + shift + local devs=$* + + aux wait_for_sync $vg $lv + aux disable_dev --error --silent $devs + mkfs.ext4 "$DM_DEV_DIR/$vg/$lv" + fsck.ext4 -fn "$DM_DEV_DIR/$vg/$lv" + check raid_leg_status $vg $lv "$fail" + aux enable_dev --silent $devs + lvs -a -o +devices $vg | tee out + not grep unknown out + lvchange --refresh $vg/$lv + fsck.ext4 -fn "$DM_DEV_DIR/$vg/$lv" + aux wait_for_sync $vg $lv + fsck.ext4 -fn "$DM_DEV_DIR/$vg/$lv" + check raid_leg_status $vg $lv "$good" +} + +# raid1 with transiently failing devices +lv=4way +lvcreate -aey --type raid1 -m 3 --ignoremonitoring -L 1 -n $lv $vg +_check_raid $vg $lv "ADAD" "AAAA" $dev2 $dev4 +lvremove -y $vg/$lv + +# raid6 with transiently failing devices +lv=6way +lvcreate -aey --type raid6 -i 4 --ignoremonitoring -L 1 -n $lv $vg +_check_raid $vg $lv "ADADAA" "AAAAAA" $dev2 $dev4 +lvremove -y $vg/$lv + +# raid10 with transiently failing devices +lv=6way +lvcreate -aey --type raid10 -i 3 -m 1 --ignoremonitoring -L 1 -n $lv $vg +_check_raid $vg $lv "ADADDA" "AAAAAA" $dev2 $dev4 $dev5 +lvremove -y $vg/$lv + +vgremove -f $vg diff --git a/test/shell/lvconvert-raid.sh b/test/shell/lvconvert-raid.sh index 25bc4a8..8538c41 100644 --- a/test/shell/lvconvert-raid.sh +++ b/test/shell/lvconvert-raid.sh @@ -32,7 +32,8 @@ get_image_pvs() { aux have_raid 1 3 0 || skip aux prepare_pvs 9 -vgcreate -s 256k $vg $(cat DEVICES) +# vgcreate -s 256k $vg $(cat DEVICES) +vgcreate -s 2m $vg $(cat DEVICES) ########################################### # RAID1 convert tests @@ -135,15 +136,27 @@ lvconvert --yes --splitmirrors 1 --name $lv2 $vg/$lv1 "$dev2" lvremove -ff $vg ########################################### -# RAID1 split + trackchanges / merge +# RAID1 split + trackchanges / merge with content check ########################################### # 3-way to 2-way/linear -lvcreate --type raid1 -m 2 -l 2 -n $lv1 $vg +lvcreate --type raid1 -m 2 -l 1 -n $lv1 $vg +mkfs.ext4 "$DM_DEV_DIR/$vg/$lv1" +fsck.ext4 -fn "$DM_DEV_DIR/$vg/$lv1" aux wait_for_sync $vg $lv1 +fsck.ext4 -fn "$DM_DEV_DIR/$vg/$lv1" lvconvert --splitmirrors 1 --trackchanges $vg/$lv1 check lv_exists $vg $lv1 check linear $vg ${lv1}_rimage_2 +fsck.ext4 -fn "$DM_DEV_DIR/mapper/$vg-${lv1}_rimage_2" +dd of="$DM_DEV_DIR/$vg/$lv1" if=/dev/zero bs=512 oflag=direct count=`blockdev --getsz "$DM_DEV_DIR/$vg/$lv1"` +not fsck.ext4 -fn "$DM_DEV_DIR/$vg/$lv1" +fsck.ext4 -fn "$DM_DEV_DIR/mapper/$vg-${lv1}_rimage_2" +# FIXME: needed on tiny loop but not on real block backend ? +lvchange --refresh $vg/$lv1 lvconvert --merge $vg/${lv1}_rimage_2 +aux wait_for_sync $vg $lv1 +lvconvert --splitmirrors 1 --trackchanges $vg/$lv1 +not fsck.ext4 -fn "$DM_DEV_DIR/mapper/$vg-${lv1}_rimage_2" # FIXME: ensure no residual devices lvremove -ff $vg

1 0

master - tests: use hold_device_open
by Zdenek Kabelac 23 Dec '16

23 Dec '16

Gitweb: http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=62be9c8de430a054… Commit: 62be9c8de430a054d5de9b652949f58a684a0cf6 Parent: e1943fc07f81e1c824757ccdac45c1e43a57af28 Author: Zdenek Kabelac <zkabelac(a)redhat.com> AuthorDate: Thu Dec 22 23:31:22 2016 +0100 Committer: Zdenek Kabelac <zkabelac(a)redhat.com> CommitterDate: Thu Dec 22 23:37:07 2016 +0100 tests: use hold_device_open --- test/shell/snapshot-merge-stack.sh | 4 +--- test/shell/snapshot-usage.sh | 15 +-------------- test/shell/thin-autoumount-dmeventd.sh | 6 +++--- 3 files changed, 5 insertions(+), 20 deletions(-) diff --git a/test/shell/snapshot-merge-stack.sh b/test/shell/snapshot-merge-stack.sh index bfafe71..2fdfaa4 100644 --- a/test/shell/snapshot-merge-stack.sh +++ b/test/shell/snapshot-merge-stack.sh @@ -29,9 +29,7 @@ snap_and_merge() { sync lvs -a $vg - # keep device open to prevent instant merge - sleep 20 < "$DM_DEV_DIR/$vg/$lv1" & - SLEEP_PID=$! + SLEEP_PID=$(aux hold_device_open $vg $lv1 20) # initiate background merge lvconvert -b --merge $vg/$lv2 diff --git a/test/shell/snapshot-usage.sh b/test/shell/snapshot-usage.sh index 042b4bf..a780664 100644 --- a/test/shell/snapshot-usage.sh +++ b/test/shell/snapshot-usage.sh @@ -24,16 +24,6 @@ fill() { die "Snapshot does not fit $1" } -# Wait until device is opened -wait_for_open_() { - for i in $(seq 1 50) ; do - test $(dmsetup info --noheadings -c -o open $1) -ne 0 && return - sleep 0.1 - done - - die "$1 expected to be openned, but it's not!" -} - cleanup_tail() { test -z "$SLEEP_PID" || kill $SLEEP_PID || true @@ -125,10 +115,7 @@ lvchange -ay $vg1 check lv_field $vg1/$lv1 lv_active "$CHECK_ACTIVE" # Test removal of opened (but unmounted) snapshot (device busy) for a while -sleep 120 < "$DM_DEV_DIR/$vg1/$lv1" & -SLEEP_PID=$! - -wait_for_open_ "$vg1-$lv1" +SLEEP_PID=$(aux hold_device_open $vg1 $lv1 60) # Opened virtual snapshot device is not removable # it should retry device removal for a few seconds diff --git a/test/shell/thin-autoumount-dmeventd.sh b/test/shell/thin-autoumount-dmeventd.sh index 3729e18..40dd75f 100644 --- a/test/shell/thin-autoumount-dmeventd.sh +++ b/test/shell/thin-autoumount-dmeventd.sh @@ -72,18 +72,18 @@ touch "$mntusedir/file$$" sync # Running 'keeper' process sleep holds the block device still in use -sleep 60 < "$mntusedir/file$$" & +sleep 60 < "$mntusedir/file$$" >/dev/null 2>&1 & PID_SLEEP=$! lvs -a $vg # Fill pool above 95% (to cause 'forced lazy umount) dd if=/dev/zero of="$mntdir/file$$" bs=256K count=20 conv=fdatasync -sync + lvs -a $vg # Could loop here for a few secs so dmeventd can do some work # In the worst case check only happens every 10 seconds :( -# With low water mark it should react way faster +# With low water mark it quickly discovers overflow and umounts $vg/$lv1 for i in $(seq 1 12) ; do is_lv_opened_ "$vg/$lv1" || break test $i -lt 12 || die "$mntdir should have been unmounted by dmeventd!"

1 0

master - tests: add device holding function
by Zdenek Kabelac 23 Dec '16

23 Dec '16

Gitweb: http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=e1943fc07f81e1c8… Commit: e1943fc07f81e1c824757ccdac45c1e43a57af28 Parent: 1053d46aff11a8e592ca91ba56d6400f4693b9a5 Author: Zdenek Kabelac <zkabelac(a)redhat.com> AuthorDate: Thu Dec 22 22:21:09 2016 +0100 Committer: Zdenek Kabelac <zkabelac(a)redhat.com> CommitterDate: Thu Dec 22 23:37:07 2016 +0100 tests: add device holding function Hold device open with sleep and wait till sleep really opens given devices. --- test/lib/aux.sh | 21 +++++++++++++++++++++ 1 files changed, 21 insertions(+), 0 deletions(-) diff --git a/test/lib/aux.sh b/test/lib/aux.sh index a195883..174e05e 100644 --- a/test/lib/aux.sh +++ b/test/lib/aux.sh @@ -1491,6 +1491,27 @@ wait_pvmove_lv_ready() { fi } +# Holds device open with sleep which automatically expires after given timeout +# Prints PID of running holding sleep process in background +hold_device_open() { + local vgname=$1 + local lvname=$2 + local sec=${3:-20} # default 20sec + + sleep $sec < "$DM_DEV_DIR/$vgname/$lvname" >/dev/null 2>&1 & + SLEEP_PID=$! + # wait till device is openned + for i in $(seq 1 50) ; do + if test "$(dmsetup info --noheadings -c -o open $vgname-$lvname)" -ne 0 ; then + echo "$SLEEP_PID" + return + fi + sleep .1 + done + + die "$vgname-$lvname expected to be openned, but it's not!" +} + # return total memory size in kB units total_mem() { while IFS=":" read -r a b ; do

1 0

master - tests: workaround failure on fc23
by Zdenek Kabelac 23 Dec '16

23 Dec '16

Gitweb: http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=1053d46aff11a8e5… Commit: 1053d46aff11a8e592ca91ba56d6400f4693b9a5 Parent: dd19b5698553b4eb0214825a6dcb88a5330b4097 Author: Zdenek Kabelac <zkabelac(a)redhat.com> AuthorDate: Tue Dec 20 16:29:23 2016 +0100 Committer: Zdenek Kabelac <zkabelac(a)redhat.com> CommitterDate: Thu Dec 22 23:37:07 2016 +0100 tests: workaround failure on fc23 --- test/shell/lvconvert-thin-external-cache.sh | 6 ++++++ 1 files changed, 6 insertions(+), 0 deletions(-) diff --git a/test/shell/lvconvert-thin-external-cache.sh b/test/shell/lvconvert-thin-external-cache.sh index 5fc1ff8..dae65a0 100644 --- a/test/shell/lvconvert-thin-external-cache.sh +++ b/test/shell/lvconvert-thin-external-cache.sh @@ -50,6 +50,12 @@ not lvconvert --thin --thinpool $vg/tpool $vg/$lv1 # Switch to 'writethrough' - this should be supported lvchange --cachemode writethrough $vg/$lv1 +# FIXME +# systemd on fc23 'strikes-in' and unmounts mnt +# ATM the reason is unclear (bug in systemd, bad udev rules?) +# as a workaround mount again and 'WARN' test +should not mount "$DM_DEV_DIR/$vg/$lv1" mnt + lvconvert --thin $vg/$lv1 --originname extorg --thinpool $vg/tpool # check cache exist as extorg-real

1 0

master - thin: refresh status when error processing fails
by Zdenek Kabelac 23 Dec '16

23 Dec '16

Gitweb: http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=dd19b5698553b4eb… Commit: dd19b5698553b4eb0214825a6dcb88a5330b4097 Parent: 77997c7673bfca56f51ae4eb55a50bc76e40fe79 Author: Zdenek Kabelac <zkabelac(a)redhat.com> AuthorDate: Thu Dec 22 23:28:04 2016 +0100 Committer: Zdenek Kabelac <zkabelac(a)redhat.com> CommitterDate: Thu Dec 22 23:37:07 2016 +0100 thin: refresh status when error processing fails When thin-pool processes event and 'lvextend --use-policies' fails rather capture up-to-date new info as the fullness percentage may have jumped noticable. This way we could use 'more' correct numbers when checking for thresholds. --- WHATS_NEW_DM | 1 + daemons/dmeventd/plugins/thin/dmeventd_thin.c | 26 +++++++++++++++++++++++++ 2 files changed, 27 insertions(+), 0 deletions(-) diff --git a/WHATS_NEW_DM b/WHATS_NEW_DM index 3f9eeac..67aed5e 100644 --- a/WHATS_NEW_DM +++ b/WHATS_NEW_DM @@ -1,5 +1,6 @@ Version 1.02.138 - ===================================== + Thin dmeventd plugin reacts faster on lvextend failure path with umount. Add dm_stats_bind_from_fd() to bind a stats handle from a file descriptor. Do not try call callback when reverting activation on error path. Fix file mapping for extents with physically adjacent extents. diff --git a/daemons/dmeventd/plugins/thin/dmeventd_thin.c b/daemons/dmeventd/plugins/thin/dmeventd_thin.c index 0c26baf..65de465 100644 --- a/daemons/dmeventd/plugins/thin/dmeventd_thin.c +++ b/daemons/dmeventd/plugins/thin/dmeventd_thin.c @@ -328,6 +328,7 @@ void process_event(struct dm_task *dmt, char *params; int needs_policy = 0; int needs_umount = 0; + struct dm_task *new_dmt = NULL; #if THIN_DEBUG log_debug("Watch for tp-data:%.2f%% tp-metadata:%.2f%%.", @@ -346,6 +347,28 @@ void process_event(struct dm_task *dmt, goto out; stack; + + /* + * Rather update oldish status + * since after 'command' processing + * percentage info could have changed a lot. + * If we would get above UMOUNT_THRESH + * we would wait for next sigalarm. + */ + if (!(new_dmt = dm_task_create(DM_DEVICE_STATUS))) + goto_out; + + if (!dm_task_set_uuid(new_dmt, dm_task_get_uuid(dmt))) + goto_out; + + /* Non-blocking status read */ + if (!dm_task_no_flush(new_dmt)) + log_warn("WARNING: Can't set no_flush for dm status."); + + if (!dm_task_run(new_dmt)) + goto_out; + + dmt = new_dmt; } dm_get_next_target(dmt, next, &start, &length, &target_type, &params); @@ -433,6 +456,9 @@ out: device, state->fails); pthread_kill(pthread_self(), SIGALRM); } + + if (new_dmt) + dm_task_destroy(new_dmt); } int register_device(const char *device,

1 0

master - report: show proper info for merging origin
by Zdenek Kabelac 23 Dec '16

23 Dec '16

Gitweb: http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=77997c7673bfca56… Commit: 77997c7673bfca56f51ae4eb55a50bc76e40fe79 Parent: 2aee4769b43a08eed549bd324664af5f406a25b2 Author: Zdenek Kabelac <zkabelac(a)redhat.com> AuthorDate: Thu Dec 22 19:51:35 2016 +0100 Committer: Zdenek Kabelac <zkabelac(a)redhat.com> CommitterDate: Thu Dec 22 23:37:07 2016 +0100 report: show proper info for merging origin When there is 'merging' of an origin in progress, but metadata stil do provide both origin and snapshot, we should show data from merged snapshot. This is important mainly for thin case, where there was a window, where i.e. 'lvs -o+device_id' would report information about 'already gone' origin thin LV. This race window is usually hard to trigger but can be ocasionally hit. Usually shortly after activation, but before polling process manages to update metadata after merge. --- WHATS_NEW | 1 + tools/reporter.c | 58 ++++++++++++++++++++++++++++++++++++++++++++++++++++++ 2 files changed, 59 insertions(+), 0 deletions(-) diff --git a/WHATS_NEW b/WHATS_NEW index dcfd85a..e5bbfe0 100644 --- a/WHATS_NEW +++ b/WHATS_NEW @@ -1,5 +1,6 @@ Version 2.02.169 - ===================================== + Report thin LV date for merged LV when the merge is in progress. Detect if snapshot merge really started before polling for progress. Checking LV for merging origin requires also it has merged snapshot. Extend validation of metadata processing. diff --git a/tools/reporter.c b/tools/reporter.c index fc87c06..6c5996b 100644 --- a/tools/reporter.c +++ b/tools/reporter.c @@ -113,6 +113,40 @@ static int _do_info_and_status(struct cmd_context *cmd, return 1; } +/* Check if this is really merging origin. + * In such case, origin is gone, and user should see + * only data from merged snapshot. Important for thin. */ +static int _check_merging_origin(const struct logical_volume *lv, + struct lv_with_info_and_seg_status *status, + int *merged) +{ + uint32_t device_id; + + *merged = 0; + + switch (status->seg_status.type) { + case SEG_STATUS_THIN: + /* Get 'device_id' from active dm-table */ + if (!lv_thin_device_id(lv, &device_id)) + return_0; + + if (lv->snapshot->device_id != device_id) + return 1; + break; + case SEG_STATUS_SNAPSHOT: + break; + default: + return 1; + } + + /* Origin is gone */ + log_debug_activation("Merge is progress, reporting merged LV %s.", + display_lvname(lv->snapshot->lv)); + *merged = 1; + + return 1; +} + static int _do_lvs_with_info_and_status_single(struct cmd_context *cmd, const struct logical_volume *lv, int do_info, int do_status, @@ -123,10 +157,22 @@ static int _do_lvs_with_info_and_status_single(struct cmd_context *cmd, .seg_status.type = SEG_STATUS_NONE }; int r = ECMD_FAILED; + int merged; + + if (lv_is_merging_origin(lv)) + /* Status is need to know which LV should be shown */ + do_status = 1; if (!_do_info_and_status(cmd, first_seg(lv), &status, do_info, do_status)) goto_out; + if (lv_is_merging_origin(lv)) { + if (!_check_merging_origin(lv, &status, &merged)) + goto_out; + if (merged) + lv = lv->snapshot->lv; + } + if (!report_object(sh ? : handle->custom_handle, sh != NULL, lv->vg, lv, NULL, NULL, NULL, &status, NULL)) goto out; @@ -173,10 +219,22 @@ static int _do_segs_with_info_and_status_single(struct cmd_context *cmd, .seg_status.type = SEG_STATUS_NONE }; int r = ECMD_FAILED; + int merged; + + if (lv_is_merging_origin(seg->lv)) + /* Status is need to know which LV should be shown */ + do_status = 1; if (!_do_info_and_status(cmd, seg, &status, do_info, do_status)) goto_out; + if (lv_is_merging_origin(seg->lv)) { + if (!_check_merging_origin(seg->lv, &status, &merged)) + goto_out; + if (merged) + seg = seg->lv->snapshot; + } + if (!report_object(sh ? : handle->custom_handle, sh != NULL, seg->lv->vg, seg->lv, NULL, seg, NULL, &status, NULL)) goto_out;

1 0

master - snapshot: validate merge has started
by Zdenek Kabelac 23 Dec '16

23 Dec '16

Gitweb: http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=2aee4769b43a08ee… Commit: 2aee4769b43a08eed549bd324664af5f406a25b2 Parent: 95e3dd5fb1297f6b1aa23cbedad1ab3dc14343a7 Author: Zdenek Kabelac <zkabelac(a)redhat.com> AuthorDate: Thu Dec 22 19:46:02 2016 +0100 Committer: Zdenek Kabelac <zkabelac(a)redhat.com> CommitterDate: Thu Dec 22 23:37:07 2016 +0100 snapshot: validate merge has started Before starting polling process, validate the merge has actually started so there is not pointless invoke of lvmpolld. This also fixes reported message from command, so user has correct info whether merging has already started or if it's delayed for next activation. --- WHATS_NEW | 1 + tools/lvconvert.c | 9 +++++++-- 2 files changed, 8 insertions(+), 2 deletions(-) diff --git a/WHATS_NEW b/WHATS_NEW index 8fefe6a..dcfd85a 100644 --- a/WHATS_NEW +++ b/WHATS_NEW @@ -1,5 +1,6 @@ Version 2.02.169 - ===================================== + Detect if snapshot merge really started before polling for progress. Checking LV for merging origin requires also it has merged snapshot. Extend validation of metadata processing. Enable usage of cached volumes as snapshot origin LV. diff --git a/tools/lvconvert.c b/tools/lvconvert.c index 0f0f0ad..ba8396f 100644 --- a/tools/lvconvert.c +++ b/tools/lvconvert.c @@ -2497,8 +2497,13 @@ static int _lvconvert_merge_old_snapshot(struct cmd_context *cmd, if (!lv_update_and_reload(origin)) return_0; - lp->need_polling = 1; - lp->lv_to_poll = origin; + if (lv_has_target_type(origin->vg->vgmem, origin, NULL, + TARGET_NAME_SNAPSHOT_MERGE)) { + lp->need_polling = 1; + lp->lv_to_poll = origin; + } else + /* Race during table reload prevented merging */ + merge_on_activate = 1; } if (merge_on_activate)

1 0

master - lv: more exact check for merging origin
by Zdenek Kabelac 23 Dec '16

23 Dec '16

Gitweb: http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=95e3dd5fb1297f6b… Commit: 95e3dd5fb1297f6b1aa23cbedad1ab3dc14343a7 Parent: 9491ab41cd61501039027fa5d7cfe89f78504132 Author: Zdenek Kabelac <zkabelac(a)redhat.com> AuthorDate: Thu Dec 22 21:15:31 2016 +0100 Committer: Zdenek Kabelac <zkabelac(a)redhat.com> CommitterDate: Thu Dec 22 23:37:07 2016 +0100 lv: more exact check for merging origin Merging origin has 'MERGE_LV' and should also have its merging snapshot. --- WHATS_NEW | 1 + lib/metadata/metadata-exported.h | 2 +- 2 files changed, 2 insertions(+), 1 deletions(-) diff --git a/WHATS_NEW b/WHATS_NEW index 6ab385d..8fefe6a 100644 --- a/WHATS_NEW +++ b/WHATS_NEW @@ -1,5 +1,6 @@ Version 2.02.169 - ===================================== + Checking LV for merging origin requires also it has merged snapshot. Extend validation of metadata processing. Enable usage of cached volumes as snapshot origin LV. Fix displayed lv name when splitting snapshot (2.02.146). diff --git a/lib/metadata/metadata-exported.h b/lib/metadata/metadata-exported.h index aa40c6c..4e05885 100644 --- a/lib/metadata/metadata-exported.h +++ b/lib/metadata/metadata-exported.h @@ -198,7 +198,7 @@ #define lv_is_partial(lv) (((lv)->status & PARTIAL_LV) ? 1 : 0) #define lv_is_virtual(lv) (((lv)->status & VIRTUAL) ? 1 : 0) #define lv_is_merging(lv) (((lv)->status & MERGING) ? 1 : 0) -#define lv_is_merging_origin(lv) (lv_is_merging(lv)) +#define lv_is_merging_origin(lv) (lv_is_merging(lv) && (lv)->snapshot) #define lv_is_snapshot(lv) (((lv)->status & SNAPSHOT) ? 1 : 0) #define lv_is_converting(lv) (((lv)->status & CONVERTING) ? 1 : 0) #define lv_is_external_origin(lv) (((lv)->external_count > 0) ? 1 : 0)

1 0

master - validation: rework segment validation
by Zdenek Kabelac 23 Dec '16

23 Dec '16

Gitweb: http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=9491ab41cd615010… Commit: 9491ab41cd61501039027fa5d7cfe89f78504132 Parent: eacff5c189d0cb3c6ac17a8f0af0fe440f12c4b7 Author: Zdenek Kabelac <zkabelac(a)redhat.com> AuthorDate: Tue Dec 20 15:59:11 2016 +0100 Committer: Zdenek Kabelac <zkabelac(a)redhat.com> CommitterDate: Thu Dec 22 23:37:07 2016 +0100 validation: rework segment validation Move individual segment validation to a separate function executed for 'complete_vg'. Move some 'extra' validation bits from 'raid' validation to global segtype validation (so extending existing normal validation) TODO: still some test are left to be moved. Reduce some duplication in validation process - there are still some left thought so still room for improving overal speed. --- WHATS_NEW | 1 + lib/metadata/merge.c | 527 ++++++++++++++++++++++---------------------------- 2 files changed, 233 insertions(+), 295 deletions(-) diff --git a/WHATS_NEW b/WHATS_NEW index fa8f5c3..6ab385d 100644 --- a/WHATS_NEW +++ b/WHATS_NEW @@ -1,5 +1,6 @@ Version 2.02.169 - ===================================== + Extend validation of metadata processing. Enable usage of cached volumes as snapshot origin LV. Fix displayed lv name when splitting snapshot (2.02.146). Warn about command not making metadata backup just once per command. diff --git a/lib/metadata/merge.c b/lib/metadata/merge.c index 947513b..a1c9455 100644 --- a/lib/metadata/merge.c +++ b/lib/metadata/merge.c @@ -71,6 +71,13 @@ int lv_merge_segments(struct logical_volume *lv) if (error_count++ > ERROR_MAX) \ goto out +#define seg_error(msg) { \ + log_error("LV %s, segment %u invalid: %s for %s segment.", \ + seg->lv->name, seg_count, (msg), lvseg_name(seg)); \ + if ((*error_count)++ > ERROR_MAX) \ + return; \ + } + /* * RAID segment property checks. * @@ -188,44 +195,10 @@ static void _check_non_raid_seg_members(struct lv_segment *seg, int *error_count { if (seg->origin) /* snap and thin */ raid_seg_error("non-zero origin LV"); - if (seg->indirect_origin) /* thin */ - raid_seg_error("non-zero indirect_origin LV"); - if (seg->merge_lv) /* thin */ - raid_seg_error("non-zero merge LV"); if (seg->cow) /* snap */ raid_seg_error("non-zero cow LV"); if (!dm_list_empty(&seg->origin_list)) /* snap */ raid_seg_error("non-zero origin_list"); - if (seg->log_lv) - raid_seg_error("non-zero log LV"); - if (seg->segtype_private) - raid_seg_error("non-zero segtype_private"); - /* thin members */ - if (seg->metadata_lv) - raid_seg_error("non-zero metadata LV"); - if (seg->transaction_id) - raid_seg_error("non-zero transaction_id"); - if (seg->zero_new_blocks) - raid_seg_error("non-zero zero_new_blocks"); - if (seg->discards) - raid_seg_error("non-zero discards"); - if (!dm_list_empty(&seg->thin_messages)) - raid_seg_error("non-zero thin_messages list"); - if (seg->external_lv) - raid_seg_error("non-zero external LV"); - if (seg->pool_lv) - raid_seg_error("non-zero pool LV"); - if (seg->device_id) - raid_seg_error("non-zero device_id"); - /* cache members */ - if (seg->cache_mode) - raid_seg_error("non-zero cache_mode"); - if (seg->policy_name) - raid_seg_error("non-zero policy_name"); - if (seg->policy_settings) - raid_seg_error("non-zero policy_settings"); - if (seg->cleaner_policy) - raid_seg_error("non-zero cleaner_policy"); /* replicator members (deprecated) */ if (seg->replicator) raid_seg_error("non-zero replicator"); @@ -249,9 +222,6 @@ static void _check_raid_seg(struct lv_segment *seg, int *error_count) uint32_t area_len, s; /* General checks applying to all RAIDs */ - if (!seg_is_raid(seg)) - raid_seg_error("erroneous RAID check"); - if (!seg->area_count) raid_seg_error("zero area count"); @@ -276,9 +246,6 @@ static void _check_raid_seg(struct lv_segment *seg, int *error_count) return; } - if (seg->chunk_size) - raid_seg_error_val("non-zero chunk_size", seg->chunk_size); - /* FIXME: should we check any non-RAID segment struct members at all? */ _check_non_raid_seg_members(seg, error_count); @@ -329,6 +296,167 @@ static void _check_raid_seg(struct lv_segment *seg, int *error_count) } /* END: RAID segment property checks. */ +static void _check_lv_segment(struct logical_volume *lv, struct lv_segment *seg, + unsigned seg_count, int *error_count) +{ + struct lv_segment *seg2; + + if (lv_is_mirror_image(lv) && + (!(seg2 = find_mirror_seg(seg)) || !seg_is_mirrored(seg2))) + seg_error("mirror image is not mirrored"); + + if (seg_is_cache(seg)) { + if (!lv_is_cache(lv)) + seg_error("is not flagged as cache LV"); + + if (!seg->pool_lv) { + seg_error("is missing cache pool LV"); + } else if (!lv_is_cache_pool(seg->pool_lv)) + seg_error("is not referencing cache pool LV"); + } else { /* !cache */ + if (seg->cleaner_policy) + seg_error("sets cleaner_policy"); + } + + if (seg_is_cache_pool(seg)) { + if (!dm_list_empty(&seg->lv->segs_using_this_lv)) { + switch (seg->cache_mode) { + case CACHE_MODE_WRITETHROUGH: + case CACHE_MODE_WRITEBACK: + case CACHE_MODE_PASSTHROUGH: + break; + default: + seg_error("has invalid cache's feature flag") + } + if (!seg->policy_name) + seg_error("is missing cache policy name"); + } + } else { /* !cache_pool */ + if (seg->cache_mode) + seg_error("sets cache mode"); + if (seg->policy_name) + seg_error("sets policy name"); + if (seg->policy_settings) + seg_error("sets policy settings"); + } + + if (!seg_can_error_when_full(seg) && lv_is_error_when_full(lv)) + seg_error("does not support flag ERROR_WHEN_FULL."); + + if (seg_is_mirrored(seg)) { + /* Check mirror log - which is attached to the mirrored seg */ + if (seg->log_lv) { + if (!lv_is_mirror_log(seg->log_lv)) + seg_error("log LV is not a mirror log"); + + if (!(seg2 = first_seg(seg->log_lv)) || (find_mirror_seg(seg2) != seg)) + seg_error("log LV does not point back to mirror segment"); + } + } else { /* !mirrored */ + if (seg->log_lv) { + if (lv_is_raid_image(lv)) + seg_error("log LV is not a mirror log or a RAID image"); + } + } + + if (seg_is_raid(seg)) + _check_raid_seg(seg, error_count); + + if (seg_is_pool(seg)) { + if ((seg->area_count != 1) || (seg_type(seg, 0) != AREA_LV)) { + seg_error("is missing a pool data LV"); + } else if (!(seg2 = first_seg(seg_lv(seg, 0))) || (find_pool_seg(seg2) != seg)) + seg_error("data LV does not refer back to pool LV"); + + if (!seg->metadata_lv) { + seg_error("is missing a pool metadata LV"); + } else if (!(seg2 = first_seg(seg->metadata_lv)) || (find_pool_seg(seg2) != seg)) + seg_error("metadata LV does not refer back to pool LV"); + + if (!validate_pool_chunk_size(lv->vg->cmd, seg->segtype, seg->chunk_size)) + seg_error("has invalid chunk size."); + } else { /* !thin_pool && !cache_pool */ + if (seg->metadata_lv) + seg_error("must not have pool metadata LV set"); + } + + if (seg_is_thin_pool(seg)) { + if (!lv_is_thin_pool(lv)) + seg_error("is not flagged as thin pool LV"); + + if (lv_is_thin_volume(lv)) + seg_error("is a thin volume that must not contain thin pool segment"); + } else { /* !thin_pool */ + if (seg->zero_new_blocks) + seg_error("sets zero_new_blocks"); + if (seg->discards) + seg_error("sets discards"); + if (!dm_list_empty(&seg->thin_messages)) + seg_error("sets thin_messages list"); + } + + if (seg_is_thin_volume(seg)) { + if (!lv_is_thin_volume(lv)) + seg_error("is not flagged as thin volume LV"); + + if (lv_is_thin_pool(lv)) + seg_error("is a thin pool that must not contain thin volume segment"); + + if (!seg->pool_lv) { + seg_error("is missing thin pool LV"); + } else if (!lv_is_thin_pool(seg->pool_lv)) + seg_error("is not referencing thin pool LV"); + + if (seg->device_id > DM_THIN_MAX_DEVICE_ID) + seg_error("has too large device id"); + + if (seg->external_lv && + !lv_is_external_origin(seg->external_lv)) + seg_error("external LV is not flagged as a external origin LV"); + + if (seg->merge_lv) { + if (!lv_is_thin_volume(seg->merge_lv)) + seg_error("merge LV is not flagged as a thin LV"); + + if (!lv_is_merging_origin(seg->merge_lv)) + seg_error("merge LV is not flagged as merging"); + } + } else { /* !thin */ + if (seg->device_id) + seg_error("sets device_id"); + if (seg->external_lv) + seg_error("sets external LV"); + if (seg->merge_lv) + seg_error("sets merge LV"); + if (seg->indirect_origin) + seg_error("sets indirect_origin LV"); + } + + /* Some multi-seg vars excluded here */ + if (!seg_is_cache(seg) && + !seg_is_thin_volume(seg)) { + if (seg->pool_lv) + seg_error("sets pool LV"); + } + + if (!seg_is_pool(seg) && + !seg_is_snapshot(seg)) { + if (seg->chunk_size) + seg_error("sets chunk_size"); + } + + if (!seg_is_thin_pool(seg) && + !seg_is_thin_volume(seg)) { + if (seg->transaction_id) + seg_error("sets transaction_id"); + } + + if (!seg_unknown(seg)) { + if (seg->segtype_private) + seg_error("set segtype_private"); + } +} + /* * Verify that an LV's segments are consecutive, complete and don't overlap. */ @@ -336,7 +464,7 @@ int check_lv_segments(struct logical_volume *lv, int complete_vg) { struct lv_segment *seg, *seg2; uint32_t le = 0; - unsigned seg_count = 0, seg_found; + unsigned seg_count = 0, seg_found, external_lv_found = 0; uint32_t area_multiplier, s; struct seg_list *sl; struct glv_list *glvl; @@ -344,80 +472,15 @@ int check_lv_segments(struct logical_volume *lv, int complete_vg) struct replicator_site *rsite; struct replicator_device *rdev; - /* Check LV flags match first segment type */ - if (complete_vg) { - if (lv_is_thin_volume(lv)) { - if (dm_list_size(&lv->segments) != 1) { - log_error("LV %s is thin volume without exactly one segment.", - lv->name); - inc_error_count; - } else if (!seg_is_thin_volume(first_seg(lv))) { - log_error("LV %s is thin volume without first thin volume segment.", - lv->name); - inc_error_count; - } - } - - if (lv_is_thin_pool(lv)) { - if (dm_list_size(&lv->segments) != 1) { - log_error("LV %s is thin pool volume without exactly one segment.", - lv->name); - inc_error_count; - } else if (!seg_is_thin_pool(first_seg(lv))) { - log_error("LV %s is thin pool without first thin pool segment.", - lv->name); - inc_error_count; - } - } + dm_list_iterate_items(seg, &lv->segments) { + seg_count++; - if (lv_is_pool_data(lv) && - (!(seg2 = first_seg(lv)) || !(seg2 = find_pool_seg(seg2)) || - seg2->area_count != 1 || seg_type(seg2, 0) != AREA_LV || - seg_lv(seg2, 0) != lv)) { - log_error("LV %s: segment 1 pool data LV does not point back to same LV", - lv->name); + if (seg->lv != lv) { + log_error("LV %s invalid: segment %u is referencing different LV.", + lv->name, seg_count); inc_error_count; } - if (lv_is_pool_metadata(lv)) { - if (!(seg2 = first_seg(lv)) || !(seg2 = find_pool_seg(seg2)) || - seg2->metadata_lv != lv) { - log_error("LV %s: segment 1 pool metadata LV does not point back to same LV", - lv->name); - inc_error_count; - } - if (lv_is_thin_pool_metadata(lv) && - !strstr(lv->name, "_tmeta")) { - log_error("LV %s: thin pool metadata LV does not use _tmeta", - lv->name); - inc_error_count; - } else if (lv_is_cache_pool_metadata(lv) && - !strstr(lv->name, "_cmeta")) { - log_error("LV %s: cache pool metadata LV does not use _cmeta", - lv->name); - inc_error_count; - } - } - - if (lv_is_external_origin(lv)) { - seg_found = 0; - dm_list_iterate_items(sl, &lv->segs_using_this_lv) - if (sl->seg->external_lv == lv) - seg_found++; - if (seg_found != lv->external_count) { - log_error("LV %s: external origin count does not match.", - lv->name); - inc_error_count; - } - } - } - - dm_list_iterate_items(seg, &lv->segments) { - seg_count++; - - if (complete_vg && seg_is_raid(seg)) - _check_raid_seg(seg, &error_count); - if (seg->le != le) { log_error("LV %s invalid: segment %u should begin at " "LE %" PRIu32 " (found %" PRIu32 ").", @@ -435,186 +498,6 @@ int check_lv_segments(struct logical_volume *lv, int complete_vg) inc_error_count; } - if (lv_is_error_when_full(lv) && - !seg_can_error_when_full(seg)) { - log_error("LV %s: segment %u (%s) does not support flag " - "ERROR_WHEN_FULL.", lv->name, seg_count, seg->segtype->name); - inc_error_count; - } - - if (complete_vg && seg->log_lv && - !seg_is_mirrored(seg) && lv_is_raid_image(lv)) { - log_error("LV %s: segment %u log LV %s is not a " - "mirror log or a RAID image", - lv->name, seg_count, seg->log_lv->name); - inc_error_count; - } - - /* - * Check mirror log - which is attached to the mirrored seg - */ - if (complete_vg && seg->log_lv && seg_is_mirrored(seg)) { - if (!lv_is_mirror_log(seg->log_lv)) { - log_error("LV %s: segment %u log LV %s is not " - "a mirror log", - lv->name, seg_count, seg->log_lv->name); - inc_error_count; - } - - if (!(seg2 = first_seg(seg->log_lv)) || - find_mirror_seg(seg2) != seg) { - log_error("LV %s: segment %u log LV does not " - "point back to mirror segment", - lv->name, seg_count); - inc_error_count; - } - } - - if (complete_vg && lv_is_mirror_image(lv)) { - if (!(seg2 = find_mirror_seg(seg)) || - !seg_is_mirrored(seg2)) { - log_error("LV %s: segment %u mirror image " - "is not mirrored", - lv->name, seg_count); - inc_error_count; - } - } - - /* Check the various thin segment types */ - if (complete_vg) { - if (seg_is_thin_pool(seg)) { - if (!lv_is_thin_pool(lv)) { - log_error("LV %s is missing thin pool flag for segment %u", - lv->name, seg_count); - inc_error_count; - } - - if (lv_is_thin_volume(lv)) { - log_error("LV %s is a thin volume that must not contain thin pool segment %u", - lv->name, seg_count); - inc_error_count; - } - - } - if (seg_is_cache_pool(seg) && - !dm_list_empty(&seg->lv->segs_using_this_lv)) { - switch (seg->cache_mode) { - case CACHE_MODE_WRITETHROUGH: - case CACHE_MODE_WRITEBACK: - case CACHE_MODE_PASSTHROUGH: - break; - default: - log_error("LV %s has invalid cache's feature flag.", - lv->name); - inc_error_count; - } - if (!seg->policy_name) { - log_error("LV %s is missing cache policy name.", lv->name); - inc_error_count; - } - } - if (seg_is_pool(seg)) { - if (seg->area_count != 1 || - seg_type(seg, 0) != AREA_LV) { - log_error("LV %s: %s segment %u is missing a pool data LV", - lv->name, seg->segtype->name, seg_count); - inc_error_count; - } else if (!(seg2 = first_seg(seg_lv(seg, 0))) || find_pool_seg(seg2) != seg) { - log_error("LV %s: %s segment %u data LV does not refer back to pool LV", - lv->name, seg->segtype->name, seg_count); - inc_error_count; - } - - if (!seg->metadata_lv) { - log_error("LV %s: %s segment %u is missing a pool metadata LV", - lv->name, seg->segtype->name, seg_count); - inc_error_count; - } else if (!(seg2 = first_seg(seg->metadata_lv)) || - find_pool_seg(seg2) != seg) { - log_error("LV %s: %s segment %u metadata LV does not refer back to pool LV", - lv->name, seg->segtype->name, seg_count); - inc_error_count; - } - - if (!validate_pool_chunk_size(lv->vg->cmd, seg->segtype, seg->chunk_size)) { - log_error("LV %s: %s segment %u has invalid chunk size %u.", - lv->name, seg->segtype->name, seg_count, seg->chunk_size); - inc_error_count; - } - } else { - if (seg->metadata_lv) { - log_error("LV %s: segment %u must not have pool metadata LV set", - lv->name, seg_count); - inc_error_count; - } - } - - if (seg_is_thin_volume(seg)) { - if (!lv_is_thin_volume(lv)) { - log_error("LV %s is missing thin volume flag for segment %u", - lv->name, seg_count); - inc_error_count; - } - - if (lv_is_thin_pool(lv)) { - log_error("LV %s is a thin pool that must not contain thin volume segment %u", - lv->name, seg_count); - inc_error_count; - } - - if (!seg->pool_lv) { - log_error("LV %s: segment %u is missing thin pool LV", - lv->name, seg_count); - inc_error_count; - } else if (!lv_is_thin_pool(seg->pool_lv)) { - log_error("LV %s: thin volume segment %u pool LV is not flagged as a pool LV", - lv->name, seg_count); - inc_error_count; - } - - if (seg->device_id > DM_THIN_MAX_DEVICE_ID) { - log_error("LV %s: thin volume segment %u has too large device id %u", - lv->name, seg_count, seg->device_id); - inc_error_count; - } - if (seg->external_lv && (seg->external_lv->status & LVM_WRITE)) { - log_error("LV %s: external origin %s is writable.", - lv->name, seg->external_lv->name); - inc_error_count; - } - - if (seg->merge_lv) { - if (!lv_is_thin_volume(seg->merge_lv)) { - log_error("LV %s: thin volume segment %u merging LV %s is not flagged as a thin LV", - lv->name, seg_count, seg->merge_lv->name); - inc_error_count; - } - if (!lv_is_merging_origin(seg->merge_lv)) { - log_error("LV %s: merging LV %s is not flagged as merging.", - lv->name, seg->merge_lv->name); - inc_error_count; - } - } - } else if (seg_is_cache(seg)) { - if (!lv_is_cache(lv)) { - log_error("LV %s is missing cache flag for segment %u", - lv->name, seg_count); - inc_error_count; - } - if (!seg->pool_lv) { - log_error("LV %s: segment %u is missing cache_pool LV", - lv->name, seg_count); - inc_error_count; - } - } else { - if (seg->pool_lv) { - log_error("LV %s: segment %u must not have pool LV set", - lv->name, seg_count); - inc_error_count; - } - } - } - if (seg_is_snapshot(seg)) { if (seg->cow && seg->cow == seg->origin) { log_error("LV %s: segment %u has same LV %s for " @@ -627,6 +510,9 @@ int check_lv_segments(struct logical_volume *lv, int complete_vg) if (seg_is_replicator(seg) && !check_replicator_segment(seg)) inc_error_count; + if (complete_vg) + _check_lv_segment(lv, seg, seg_count, &error_count); + for (s = 0; s < seg->area_count; s++) { if (seg_type(seg, s) == AREA_UNASSIGNED) { log_error("LV %s: segment %u has unassigned " @@ -708,6 +594,12 @@ int check_lv_segments(struct logical_volume *lv, int complete_vg) le += seg->len; } + if (le != lv->le_count) { + log_error("LV %s: inconsistent LE count %u != %u", + lv->name, le, lv->le_count); + inc_error_count; + } + dm_list_iterate_items(sl, &lv->segs_using_this_lv) { seg = sl->seg; seg_found = 0; @@ -768,6 +660,10 @@ int check_lv_segments(struct logical_volume *lv, int complete_vg) lv->name); inc_error_count; } + + /* Validation of external origin counter */ + if (seg->external_lv == lv) + external_lv_found++; } dm_list_iterate_items(glvl, &lv->indirect_glvs) { @@ -789,10 +685,51 @@ int check_lv_segments(struct logical_volume *lv, int complete_vg) } } - if (le != lv->le_count) { - log_error("LV %s: inconsistent LE count %u != %u", - lv->name, le, lv->le_count); - inc_error_count; + /* Check LV flags match first segment type */ + if (complete_vg) { + if ((seg_count != 1) && + (lv_is_cache(lv) || + lv_is_cache_pool(lv) || + lv_is_raid(lv) || + lv_is_snapshot(lv) || + lv_is_thin_pool(lv) || + lv_is_thin_volume(lv))) { + log_error("LV %s must have exactly one segment.", + lv->name); + inc_error_count; + } + + if (lv_is_pool_data(lv) && + (!(seg2 = first_seg(lv)) || !(seg2 = find_pool_seg(seg2)) || + seg2->area_count != 1 || seg_type(seg2, 0) != AREA_LV || + seg_lv(seg2, 0) != lv)) { + log_error("LV %s: segment 1 pool data LV does not point back to same LV", + lv->name); + inc_error_count; + } + + if (lv_is_thin_pool_metadata(lv) && !strstr(lv->name, "_tmeta")) { + log_error("LV %s: thin pool metadata LV does not use _tmeta.", + lv->name); + inc_error_count; + } else if (lv_is_cache_pool_metadata(lv) && !strstr(lv->name, "_cmeta")) { + log_error("LV %s: cache pool metadata LV does not use _cmeta.", + lv->name); + inc_error_count; + } + + if (lv_is_external_origin(lv)) { + if (lv->external_count != external_lv_found) { + log_error("LV %s: external origin count does not match.", + lv->name); + inc_error_count; + } + if (lv->status & LVM_WRITE) { + log_error("LV %s: external origin cant't be writable.", + lv->name); + inc_error_count; + } + } } out:

1 0

master - lvmdbustest: Print messages if timeout value > 10%
by tasleson 21 Dec '16

21 Dec '16

Gitweb: http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=eacff5c189d0cb3c… Commit: eacff5c189d0cb3c6ac17a8f0af0fe440f12c4b7 Parent: a7e1f973ccd1abfc0f868f0b9c2382c2034e1633 Author: Tony Asleson <tasleson(a)redhat.com> AuthorDate: Wed Dec 14 15:32:08 2016 -0600 Committer: Tony Asleson <tasleson(a)redhat.com> CommitterDate: Tue Dec 20 11:06:57 2016 -0600 lvmdbustest: Print messages if timeout value > 10% We will dump some informational messages if the time to return when we specify a timeout exceeds 10% of requested. --- test/dbus/testlib.py | 25 ++++++++++++++++++++++--- 1 files changed, 22 insertions(+), 3 deletions(-) diff --git a/test/dbus/testlib.py b/test/dbus/testlib.py index 89e73d2..34ff33a 100644 --- a/test/dbus/testlib.py +++ b/test/dbus/testlib.py @@ -16,6 +16,7 @@ from collections import OrderedDict import dbus import os import sys +import time BUS_NAME = os.getenv('LVM_DBUS_NAME', 'com.redhat.lvmdbus1') BASE_INTERFACE = 'com.redhat.lvmdbus1' @@ -188,10 +189,15 @@ class RemoteInterface(object): def __init__( self, dbus_object, interface, introspect, - properties=None): + properties=None, timelimit=-1): self.dbus_object = dbus_object self.interface = interface self.introspect = introspect + self.tmo = 0 + + if timelimit >= 0: + self.tmo = float(timelimit) + self.tmo *= 1.10 self.dbus_interface = dbus.Interface(self.dbus_object, self.interface) self._set_props(properties) @@ -203,7 +209,19 @@ class RemoteInterface(object): return functools.partial(self, item) def _wrapper(self, _method_name, *args, **kwargs): + + # Lets see how long a method takes to execute, in call cases we should + # return something when the time limit has been reached. + start = time.time() result = getattr(self.dbus_interface, _method_name)(*args, **kwargs) + end = time.time() + + diff = end - start + + if self.tmo > 0.0: + if diff > self.tmo: + std_err_print("\n Time exceeded: %f > %f %s" % + (diff, self.tmo, _method_name)) if self.introspect: if 'RETURN_VALUE' in self.introspect[ @@ -236,13 +254,14 @@ class ClientProxy(object): short_name = ClientProxy._intf_short_name(interface) self.short_interface_names.append(short_name) ro = RemoteInterface(self.dbus_object, interface, introspect, - properties) + properties, timelimit=self.tmo) setattr(self, short_name, ro) def __init__(self, bus, object_path, interface_prop_hash=None, - interfaces=None): + interfaces=None, timelimit=-1): self.object_path = object_path self.short_interface_names = [] + self.tmo = timelimit self.dbus_object = bus.get_object( BUS_NAME, self.object_path, introspect=False)

1 0

master - lvmdbusd: Use timeout_add instead
by tasleson 21 Dec '16

21 Dec '16

Gitweb: http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=a7e1f973ccd1abfc… Commit: a7e1f973ccd1abfc0f868f0b9c2382c2034e1633 Parent: 75568294be01f74ddadbf2c9bacf5d2fce219a44 Author: Tony Asleson <tasleson(a)redhat.com> AuthorDate: Wed Dec 14 15:30:01 2016 -0600 Committer: Tony Asleson <tasleson(a)redhat.com> CommitterDate: Tue Dec 20 11:06:57 2016 -0600 lvmdbusd: Use timeout_add instead The function timeout_add_seconds has quite a bit of variability. Using timeout_add which specifies the timeout in ms instead of seconds. Testing shows that this is much more consistent which should improve clients that are using shorter timeouts for the API and the connection. --- daemons/lvmdbusd/request.py | 15 +++++++++------ 1 files changed, 9 insertions(+), 6 deletions(-) diff --git a/daemons/lvmdbusd/request.py b/daemons/lvmdbusd/request.py index 7c5da46..78392de 100644 --- a/daemons/lvmdbusd/request.py +++ b/daemons/lvmdbusd/request.py @@ -19,7 +19,6 @@ from .utils import log_error, mt_async_result class RequestEntry(object): def __init__(self, tmo, method, arguments, cb, cb_error, return_tuple=True, job_state=None): - self.tmo = tmo self.method = method self.arguments = arguments self.cb = cb @@ -35,23 +34,27 @@ class RequestEntry(object): self._return_tuple = return_tuple self._job_state = job_state - if self.tmo < 0: + if tmo < 0: # Client is willing to block forever pass elif tmo == 0: self._return_job() else: - self.timer_id = GLib.timeout_add_seconds( - tmo, RequestEntry._request_timeout, self) + # Note: using 990 instead of 1000 for second to ms conversion to + # account for overhead. Goal is to return just before the + # timeout amount has expired. Better to be a little early than + # late. + self.timer_id = GLib.timeout_add( + tmo * 990, RequestEntry._request_timeout, self) @staticmethod def _request_timeout(r): """ Method which gets called when the timer runs out! :param r: RequestEntry which timed out - :return: Nothing + :return: Result of timer_expired """ - r.timer_expired() + return r.timer_expired() def _return_job(self): # Return job is only called when we create a request object or when

1 0

master - lvmdbusd: Use cfg.reload() instead of dbo.refresh
by tasleson 21 Dec '16

21 Dec '16

Gitweb: http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=75568294be01f74d… Commit: 75568294be01f74ddadbf2c9bacf5d2fce219a44 Parent: 6fe6e8053a136af45174f153c00d404995af6e1c Author: Tony Asleson <tasleson(a)redhat.com> AuthorDate: Mon Dec 12 16:15:12 2016 -0600 Committer: Tony Asleson <tasleson(a)redhat.com> CommitterDate: Tue Dec 20 11:06:57 2016 -0600 lvmdbusd: Use cfg.reload() instead of dbo.refresh We want to update the data and send out any signals as needed, not just update the in memory database. --- daemons/lvmdbusd/lv.py | 4 ++-- daemons/lvmdbusd/vg.py | 4 ++-- 2 files changed, 4 insertions(+), 4 deletions(-) diff --git a/daemons/lvmdbusd/lv.py b/daemons/lvmdbusd/lv.py index 79fff87..fa54322 100644 --- a/daemons/lvmdbusd/lv.py +++ b/daemons/lvmdbusd/lv.py @@ -618,7 +618,7 @@ class Lv(LvCommon): rc, out, err = cmdhandler.activate_deactivate( 'lvchange', lv_name, activate, control_flags, options) if rc == 0: - dbo.refresh() + cfg.load() return '/' else: raise dbus.exceptions.DBusException( @@ -667,7 +667,7 @@ class Lv(LvCommon): rc, out, err = cmdhandler.lv_tag( lv_name, tags_add, tags_del, tag_options) if rc == 0: - dbo.refresh() + cfg.load() return '/' else: raise dbus.exceptions.DBusException( diff --git a/daemons/lvmdbusd/vg.py b/daemons/lvmdbusd/vg.py index d84bb75..9b1df4e 100644 --- a/daemons/lvmdbusd/vg.py +++ b/daemons/lvmdbusd/vg.py @@ -727,7 +727,7 @@ class Vg(AutomatedProperties): rc, out, err = cmdhandler.vg_tag( vg_name, tags_add, tags_del, tag_options) if rc == 0: - dbo.refresh() + cfg.load() return '/' else: raise dbus.exceptions.DBusException( @@ -780,7 +780,7 @@ class Vg(AutomatedProperties): if dbo: rc, out, err = method(vg_name, value, options) if rc == 0: - dbo.refresh() + cfg.load() return '/' else: raise dbus.exceptions.DBusException(

1 0

master - lvmdbusd: Remove un-needed main thread execution
by tasleson 21 Dec '16

21 Dec '16

Gitweb: http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=6fe6e8053a136af4… Commit: 6fe6e8053a136af45174f153c00d404995af6e1c Parent: f47da0ad234118f4a7a5e64b7a8480f6afd8e1de Author: Tony Asleson <tasleson(a)redhat.com> AuthorDate: Mon Dec 12 16:13:27 2016 -0600 Committer: Tony Asleson <tasleson(a)redhat.com> CommitterDate: Tue Dec 20 11:06:57 2016 -0600 lvmdbusd: Remove un-needed main thread execution --- daemons/lvmdbusd/request.py | 7 +++++-- 1 files changed, 5 insertions(+), 2 deletions(-) diff --git a/daemons/lvmdbusd/request.py b/daemons/lvmdbusd/request.py index a2c2ac9..7c5da46 100644 --- a/daemons/lvmdbusd/request.py +++ b/daemons/lvmdbusd/request.py @@ -54,12 +54,15 @@ class RequestEntry(object): r.timer_expired() def _return_job(self): + # Return job is only called when we create a request object or when + # we pop a timer. In both cases we are running in the correct context + # and do not need to schedule the call back in main context. self._job = Job(self, self._job_state) cfg.om.register_object(self._job, True) if self._return_tuple: - mt_async_result(self.cb, ('/', self._job.dbus_object_path())) + self.cb(('/', self._job.dbus_object_path())) else: - mt_async_result(self.cb, self._job.dbus_object_path()) + self.cb(self._job.dbus_object_path()) def run_cmd(self): try:

1 0

master - tests: usage of cached volume for snapshot
by Zdenek Kabelac 19 Dec '16

19 Dec '16

Gitweb: http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=f47da0ad234118f4… Commit: f47da0ad234118f4a7a5e64b7a8480f6afd8e1de Parent: 0c56eb8f438a17a8c98470f9f6b2222235a7bf59 Author: Zdenek Kabelac <zkabelac(a)redhat.com> AuthorDate: Mon Dec 19 14:05:16 2016 +0100 Committer: Zdenek Kabelac <zkabelac(a)redhat.com> CommitterDate: Mon Dec 19 14:41:43 2016 +0100 tests: usage of cached volume for snapshot --- test/shell/lvconvert-cache-snapshot.sh | 63 ++++++++++++++++++++++++++++++++ test/shell/lvconvert-cache.sh | 1 - test/shell/lvcreate-cache-snapshot.sh | 63 ++++++++++++++++++++++++++++++++ 3 files changed, 126 insertions(+), 1 deletions(-) diff --git a/test/shell/lvconvert-cache-snapshot.sh b/test/shell/lvconvert-cache-snapshot.sh new file mode 100644 index 0000000..8f0ab53 --- /dev/null +++ b/test/shell/lvconvert-cache-snapshot.sh @@ -0,0 +1,63 @@ +#!/bin/sh + +# Copyright (C) 2016 Red Hat, Inc. All rights reserved. +# +# This copyrighted material is made available to anyone wishing to use, +# modify, copy, or redistribute it subject to the terms and conditions +# of the GNU General Public License v.2. +# +# You should have received a copy of the GNU General Public License +# along with this program; if not, write to the Free Software Foundation, +# Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + +# Test various supported conversion of snapshot of cached volume + +SKIP_WITH_LVMLOCKD=1 +SKIP_WITH_LVMPOLLD=1 + +. lib/inittest + +aux have_cache 1 3 0 || skip + +aux prepare_vg 1 + + +# Prepare cached LV +lvcreate -aey -L1 -n $lv1 $vg +lvcreate -H -L2 -n cpool $vg/$lv1 + +# Prepare snapshot 'cow' LV +lvcreate -L3 -n cow $vg + +# Can't use 'cached' cow volume +not lvconvert -s cow $vg/$lv1 + +# Use cached LV with 'striped' cow volume +lvconvert -y -s $vg/$lv1 cow +check lv_field $vg/cow segtype linear +check lv_field $vg/$lv1 segtype cache + +# Drop cache while being in-use origin +lvconvert --splitcache $vg/$lv1 +check lv_field $vg/$lv1 segtype linear + +# Cache existing origin +lvconvert -y --cache $vg/$lv1 --cachepool $vg/cpool +check lv_field $vg/$lv1 segtype cache + +# Cannot split from 'origin' (being cached LV) +not lvconvert -y --splitsnapshot $vg/$lv1 + +lvchange --cachemode writeback $vg/$lv1 +check lv_field $vg/$lv1 cache_mode "writeback" +check grep_dmsetup status ${vg}-${lv1}-real "writeback" + +lvchange --cachemode writethrough $vg/$lv1 +check lv_field $vg/$lv1 cache_mode "writethrough" +check grep_dmsetup status ${vg}-${lv1}-real "writethrough" + +# Split 'cow' from cached origin +lvconvert -y --splitsnapshot $vg/cow +get lv_field $vg/cow attr | grep "^-wi" + +vgremove -f $vg diff --git a/test/shell/lvconvert-cache.sh b/test/shell/lvconvert-cache.sh index 76116ea..31fe4e8 100644 --- a/test/shell/lvconvert-cache.sh +++ b/test/shell/lvconvert-cache.sh @@ -23,7 +23,6 @@ aux prepare_vg 5 80 lvcreate --type cache-pool -an -v -L 2 -n cpool $vg lvcreate -H -L 4 -n corigin --cachepool $vg/cpool -fail lvcreate -s -L2 $vg/corigin fail lvcreate -s -L2 $vg/cpool fail lvcreate -s -L2 $vg/cpool_cdata fail lvcreate -s -L2 $vg/cpool_cmeta diff --git a/test/shell/lvcreate-cache-snapshot.sh b/test/shell/lvcreate-cache-snapshot.sh new file mode 100644 index 0000000..6b6ef2c --- /dev/null +++ b/test/shell/lvcreate-cache-snapshot.sh @@ -0,0 +1,63 @@ +#!/bin/sh +# Copyright (C) 2016 Red Hat, Inc. All rights reserved. +# +# This copyrighted material is made available to anyone wishing to use, +# modify, copy, or redistribute it subject to the terms and conditions +# of the GNU General Public License v.2. +# +# You should have received a copy of the GNU General Public License +# along with this program; if not, write to the Free Software Foundation, +# Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + +# Exercise creation of snapshot of cached LV + +SKIP_WITH_LVMLOCKD=1 +SKIP_WITH_LVMPOLLD=1 + +. lib/inittest + +which mkfs.ext2 || skip +which fsck || skip + +aux have_cache 1 5 0 || skip + +aux prepare_vg 2 + +lvcreate --type cache-pool -L1 $vg/cpool +lvcreate -H -L4 -n $lv1 $vg/cpool + +lvcreate -s -L2 -n $lv2 $vg/$lv1 +check lv_field $vg/$lv1 segtype cache + + +# Make some 'fs' data in snapshot +mkfs.ext2 "$DM_DEV_DIR/$vg/$lv2" +mkdir mnt +mount "$DM_DEV_DIR/$vg/$lv2" mnt +touch mnt/test +umount mnt + +sync +aux udev_wait + +# Merge snap to origin +lvconvert --merge $vg/$lv2 + +# Check cached origin has no valid fs. +fsck -n "$DM_DEV_DIR/$vg/$lv1" + +# Check deactivation +lvchange -an $vg + +# Check activation +lvchange -ay $vg + + +lvconvert --uncache $vg/$lv1 +check lv_field $vg/$lv1 segtype linear + +# Uncached origin is fine as well +fsck -n "$DM_DEV_DIR/$vg/$lv1" + + +vgremove -ff $vg

1 0

master - cache: support cached origin for snapshot
by Zdenek Kabelac 19 Dec '16

19 Dec '16

Gitweb: http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=0c56eb8f438a17a8… Commit: 0c56eb8f438a17a8c98470f9f6b2222235a7bf59 Parent: eb3f83357afd2b7462f5be9f409424f588a138c8 Author: Zdenek Kabelac <zkabelac(a)redhat.com> AuthorDate: Mon Dec 19 14:08:56 2016 +0100 Committer: Zdenek Kabelac <zkabelac(a)redhat.com> CommitterDate: Mon Dec 19 14:41:42 2016 +0100 cache: support cached origin for snapshot Enable 'lvcreate/lvconvert -s' for cached LV. and supported operations: Create a snapshot of cached LV Split/Join snapshot LV to cached origin LV. --- WHATS_NEW | 1 + lib/metadata/cache_manip.c | 2 +- lib/metadata/lv_manip.c | 2 +- tools/lvconvert.c | 16 +++++++++++++++- 4 files changed, 18 insertions(+), 3 deletions(-) diff --git a/WHATS_NEW b/WHATS_NEW index ab113ae..fa8f5c3 100644 --- a/WHATS_NEW +++ b/WHATS_NEW @@ -1,5 +1,6 @@ Version 2.02.169 - ===================================== + Enable usage of cached volumes as snapshot origin LV. Fix displayed lv name when splitting snapshot (2.02.146). Warn about command not making metadata backup just once per command. Enable usage of cached volume as thin volume's external origin. diff --git a/lib/metadata/cache_manip.c b/lib/metadata/cache_manip.c index 12b1054..404b1af 100644 --- a/lib/metadata/cache_manip.c +++ b/lib/metadata/cache_manip.c @@ -322,7 +322,7 @@ int validate_lv_cache_create_origin(const struct logical_volume *origin_lv) if (lv_is_cache_type(origin_lv) || lv_is_mirror_type(origin_lv) || lv_is_thin_volume(origin_lv) || lv_is_thin_pool_metadata(origin_lv) || - lv_is_origin(origin_lv) || lv_is_merging_origin(origin_lv) || + lv_is_merging_origin(origin_lv) || lv_is_cow(origin_lv) || lv_is_merging_cow(origin_lv) || lv_is_virtual(origin_lv)) { log_error("Cache is not supported with %s segment type of the original logical volume %s.", diff --git a/lib/metadata/lv_manip.c b/lib/metadata/lv_manip.c index 92a6d44..3862f11 100644 --- a/lib/metadata/lv_manip.c +++ b/lib/metadata/lv_manip.c @@ -7428,7 +7428,7 @@ static struct logical_volume *_lv_create_an_lv(struct volume_group *vg, return NULL; } - if (lv_is_cache_type(origin_lv)) { + if (lv_is_cache_type(origin_lv) && !lv_is_cache(origin_lv)) { log_error("Snapshots of cache type volume %s " "is not supported.", display_lvname(origin_lv)); return NULL; diff --git a/tools/lvconvert.c b/tools/lvconvert.c index 4103995..0f0f0ad 100644 --- a/tools/lvconvert.c +++ b/tools/lvconvert.c @@ -2336,7 +2336,7 @@ static int _lvconvert_snapshot(struct cmd_context *cmd, return_0; if (lv_is_locked(org) || - lv_is_cache_type(org) || + (lv_is_cache_type(org) && !lv_is_cache(org)) || lv_is_thin_type(org) || lv_is_pvmove(org) || lv_is_mirrored(org) || @@ -3771,6 +3771,16 @@ static int _convert_cache_volume_thin_pool(struct cmd_context *cmd, struct logic } /* + * Convert/Recombine cacheLV to be an origin for snapshot + * lvconvert --type snapshot cacheLV snapshotLV + */ +static int _convert_cache_volume_snapshot(struct cmd_context *cmd, struct logical_volume *lv, + struct lvconvert_params *lp) +{ + return _lvconvert_snapshot(cmd, lv, lp); +} + +/* * Split a cache volume from a cache pool LV. * lvconvert --splitcache LV */ @@ -4311,6 +4321,10 @@ static int _convert_cache_volume(struct cmd_context *cmd, struct logical_volume arg_is_set(cmd, thinpool_ARG)) return _convert_cache_volume_thin_pool(cmd, lv, lp); + if (!strcmp(lp->type_str, SEG_TYPE_NAME_SNAPSHOT) || + arg_is_set(cmd, snapshot_ARG)) + return _convert_cache_volume_snapshot(cmd, lv, lp); + /* The --thinpool alternative for --type thin-pool is not preferred, so not shown. */ log_error("Operation not permitted on cache LV %s.", display_lvname(lv));

1 0

master - lvconvert: fix shown lv name for snapshot split
by Zdenek Kabelac 19 Dec '16

19 Dec '16

Gitweb: http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=eb3f83357afd2b74… Commit: eb3f83357afd2b7462f5be9f409424f588a138c8 Parent: c90e9392e47a04f0d89c8ab0e0dca7fe9de00143 Author: Zdenek Kabelac <zkabelac(a)redhat.com> AuthorDate: Mon Dec 19 14:06:55 2016 +0100 Committer: Zdenek Kabelac <zkabelac(a)redhat.com> CommitterDate: Mon Dec 19 14:41:16 2016 +0100 lvconvert: fix shown lv name for snapshot split We can't keep 'display_lvname' for too long - it's using ringbuffer and keeps limited number of names. So it's safe only per few simple tests, but can't be used anymore after some function calls.. (Fixes 00e641ef37a977129acc503f3fa1b67f556ac5eb) --- WHATS_NEW | 1 + tools/lvconvert.c | 10 +++++----- 2 files changed, 6 insertions(+), 5 deletions(-) diff --git a/WHATS_NEW b/WHATS_NEW index fbaa369..ab113ae 100644 --- a/WHATS_NEW +++ b/WHATS_NEW @@ -1,5 +1,6 @@ Version 2.02.169 - ===================================== + Fix displayed lv name when splitting snapshot (2.02.146). Warn about command not making metadata backup just once per command. Enable usage of cached volume as thin volume's external origin. Support cache volume activation with -real layer. diff --git a/tools/lvconvert.c b/tools/lvconvert.c index 393d286..4103995 100644 --- a/tools/lvconvert.c +++ b/tools/lvconvert.c @@ -2167,7 +2167,7 @@ static int _lvconvert_splitsnapshot(struct cmd_context *cmd, struct logical_volu return_0; if (lv_is_pvmove(cow) || lv_is_mirror_type(cow) || lv_is_raid_type(cow) || lv_is_thin_type(cow)) { - log_error("LV %s type is unsupported with --splitsnapshot.", cow_name); + log_error("LV %s type is unsupported with --splitsnapshot.", display_lvname(cow)); return 0; } @@ -2179,8 +2179,8 @@ static int _lvconvert_splitsnapshot(struct cmd_context *cmd, struct logical_volu lv_is_visible(cow) && lv_is_active(cow)) { if (yes_no_prompt("Do you really want to split off active " - "logical volume %s? [y/n]: ", cow_name) == 'n') { - log_error("Logical volume %s not split.", cow_name); + "logical volume %s? [y/n]: ", display_lvname(cow)) == 'n') { + log_error("Logical volume %s not split.", display_lvname(cow)); return 0; } } @@ -2189,14 +2189,14 @@ static int _lvconvert_splitsnapshot(struct cmd_context *cmd, struct logical_volu if (!archive(vg)) return_0; - log_verbose("Splitting snapshot %s from its origin.", cow_name); + log_verbose("Splitting snapshot %s from its origin.", display_lvname(cow)); if (!vg_remove_snapshot(cow)) return_0; backup(vg); - log_print_unless_silent("Logical Volume %s split from its origin.", cow_name); + log_print_unless_silent("Logical Volume %s split from its origin.", display_lvname(cow)); return 1; }

1 0

master - libdm: add dm_stats_bind_from_fd()
by Bryn Reeves 19 Dec '16

19 Dec '16

Gitweb: http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=c90e9392e47a04f0… Commit: c90e9392e47a04f0d89c8ab0e0dca7fe9de00143 Parent: 009b711834d26610d23d14bd8b7540eb69c836a7 Author: Bryn M. Reeves <bmr(a)redhat.com> AuthorDate: Sun Dec 18 14:40:57 2016 +0000 Committer: Bryn M. Reeves <bmr(a)redhat.com> CommitterDate: Sun Dec 18 20:47:17 2016 +0000 libdm: add dm_stats_bind_from_fd() dmsetup already has a version of this function, and dmfilemapd will need it too: move it to libdevmapper to avoid copying it around. --- WHATS_NEW_DM | 1 + libdm/.exported_symbols.DM_1_02_138 | 1 + libdm/libdevmapper.h | 10 ++++++++++ libdm/libdm-stats.c | 19 +++++++++++++++++++ tools/dmsetup.c | 20 +------------------- 5 files changed, 32 insertions(+), 19 deletions(-) diff --git a/WHATS_NEW_DM b/WHATS_NEW_DM index d57ddd7..3f9eeac 100644 --- a/WHATS_NEW_DM +++ b/WHATS_NEW_DM @@ -1,5 +1,6 @@ Version 1.02.138 - ===================================== + Add dm_stats_bind_from_fd() to bind a stats handle from a file descriptor. Do not try call callback when reverting activation on error path. Fix file mapping for extents with physically adjacent extents. Validation vsnprintf result in runtime translate of dm_log (1.02.136). diff --git a/libdm/.exported_symbols.DM_1_02_138 b/libdm/.exported_symbols.DM_1_02_138 index 7535829..08f936f 100644 --- a/libdm/.exported_symbols.DM_1_02_138 +++ b/libdm/.exported_symbols.DM_1_02_138 @@ -1,3 +1,4 @@ dm_bit_get_last dm_bit_get_prev dm_bitset_parse_list +dm_stats_bind_from_fd diff --git a/libdm/libdevmapper.h b/libdm/libdevmapper.h index ed46795..363cf8e 100644 --- a/libdm/libdevmapper.h +++ b/libdm/libdevmapper.h @@ -518,6 +518,16 @@ int dm_stats_bind_name(struct dm_stats *dms, const char *name); int dm_stats_bind_uuid(struct dm_stats *dms, const char *uuid); /* + * Bind a dm_stats handle to the device backing the file referenced + * by the specified file descriptor. + * + * File descriptor fd must reference a regular file, open for reading, + * in a local file system, backed by a device-mapper device, that + * supports the FIEMAP ioctl, and that returns data describing the + * physical location of extents. + */ +int dm_stats_bind_from_fd(struct dm_stats *dms, int fd); +/* * Test whether the running kernel supports the precise_timestamps * feature. Presence of this feature also implies histogram support. * The library will check this call internally and fails any attempt diff --git a/libdm/libdm-stats.c b/libdm/libdm-stats.c index 3978244..6e79a09 100644 --- a/libdm/libdm-stats.c +++ b/libdm/libdm-stats.c @@ -16,6 +16,7 @@ */ #include "dmlib.h" +#include "kdev_t.h" #include "math.h" /* log10() */ @@ -452,6 +453,24 @@ int dm_stats_bind_uuid(struct dm_stats *dms, const char *uuid) return 1; } +int dm_stats_bind_from_fd(struct dm_stats *dms, int fd) +{ + int major, minor; + struct stat buf; + + if (fstat(fd, &buf)) { + log_error("fstat failed for fd %d.", fd); + return 0; + } + + major = (int) MAJOR(buf.st_dev); + minor = (int) MINOR(buf.st_dev); + + if (!dm_stats_bind_devno(dms, major, minor)) + return_0; + return 1; +} + static int _stats_check_precise_timestamps(const struct dm_stats *dms) { /* Already checked? */ diff --git a/tools/dmsetup.c b/tools/dmsetup.c index 3f2c619..fabb183 100644 --- a/tools/dmsetup.c +++ b/tools/dmsetup.c @@ -4640,24 +4640,6 @@ static int _bind_stats_device(struct dm_stats *dms, const char *name) return 1; } -static int _bind_stats_from_fd(struct dm_stats *dms, int fd) -{ - int major, minor; - struct stat buf; - - if (fstat(fd, &buf)) { - log_error("fstat failed for fd %d.", fd); - return 0; - } - - major = (int) MAJOR(buf.st_dev); - minor = (int) MINOR(buf.st_dev); - - if (!dm_stats_bind_devno(dms, major, minor)) - return_0; - return 1; -} - static int _stats_clear_one_region(struct dm_stats *dms, uint64_t region_id) { @@ -5068,7 +5050,7 @@ static int _stats_create_file(CMD_ARGS) goto bad; } - if (!_bind_stats_from_fd(dms, fd)) + if (!dm_stats_bind_from_fd(dms, fd)) goto_bad; if (!strlen(program_id))

1 0

master - libdm: clear region table in dm_stats_list()
by Bryn Reeves 19 Dec '16

19 Dec '16

Gitweb: http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=009b711834d26610… Commit: 009b711834d26610d23d14bd8b7540eb69c836a7 Parent: 8d6ac1c3ba2ee72e9832383609b6ff030d9883ce Author: Bryn M. Reeves <bmr(a)redhat.com> AuthorDate: Thu Dec 15 19:03:42 2016 +0000 Committer: Bryn M. Reeves <bmr(a)redhat.com> CommitterDate: Sun Dec 18 20:44:31 2016 +0000 libdm: clear region table in dm_stats_list() Call _stats_regions_destroy() from dm_stats_list() if dms->regions is non-NULL. This avoids leaking any pool allocations and ensures the handle is in a known state: if an error occurs during the list, dms->regions will be NULL and the handle will appear empty. --- libdm/libdm-stats.c | 3 +++ 1 files changed, 3 insertions(+), 0 deletions(-) diff --git a/libdm/libdm-stats.c b/libdm/libdm-stats.c index 8d202d2..3978244 100644 --- a/libdm/libdm-stats.c +++ b/libdm/libdm-stats.c @@ -1083,6 +1083,9 @@ int dm_stats_list(struct dm_stats *dms, const char *program_id) if (!_stats_set_name_cache(dms)) return_0; + if (dms->regions) + _stats_regions_destroy(dms); + r = dm_snprintf(msg, sizeof(msg), "@stats_list %s", program_id); if (r < 0) {

1 0

master - tests: using cached LV for external origin
by Zdenek Kabelac 19 Dec '16

19 Dec '16

Gitweb: http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=8d6ac1c3ba2ee72e… Commit: 8d6ac1c3ba2ee72e9832383609b6ff030d9883ce Parent: 8c17233af585b4e46a86acca8ddb126beb5faf95 Author: Zdenek Kabelac <zkabelac(a)redhat.com> AuthorDate: Sun Dec 18 17:43:05 2016 +0100 Committer: Zdenek Kabelac <zkabelac(a)redhat.com> CommitterDate: Sun Dec 18 19:38:51 2016 +0100 tests: using cached LV for external origin --- test/shell/lvconvert-thin-external-cache.sh | 107 +++++++++++++++++++++++++++ 1 files changed, 107 insertions(+), 0 deletions(-) diff --git a/test/shell/lvconvert-thin-external-cache.sh b/test/shell/lvconvert-thin-external-cache.sh new file mode 100644 index 0000000..5fc1ff8 --- /dev/null +++ b/test/shell/lvconvert-thin-external-cache.sh @@ -0,0 +1,107 @@ +#!/bin/sh + +# Copyright (C) 2016 Red Hat, Inc. All rights reserved. +# +# This copyrighted material is made available to anyone wishing to use, +# modify, copy, or redistribute it subject to the terms and conditions +# of the GNU General Public License v.2. +# +# You should have received a copy of the GNU General Public License +# along with this program; if not, write to the Free Software Foundation, +# Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + +# Test conversion cached LV to thin with cached external origin + +SKIP_WITH_LVMLOCKD=1 +SKIP_WITH_LVMPOLLD=1 + +export LVM_TEST_THIN_REPAIR_CMD=${LVM_TEST_THIN_REPAIR_CMD-/bin/false} + +. lib/inittest + +which mkfs.ext2 || skip +which fsck || skip + +# +# Main +# +aux have_thin 1 5 0 || skip +aux have_cache 1 7 0 || skip + +aux prepare_vg 2 64 + +# Test will use thin-pool +lvcreate -L10 -T $vg/tpool + +lvcreate -aey -L20 -n $lv1 $vg + + +mkfs.ext2 "$DM_DEV_DIR/$vg/$lv1" +mkdir mnt +mount "$DM_DEV_DIR/$vg/$lv1" mnt +touch mnt/test + +# Prepared cached LV - first in 'writeback' mode +lvcreate -H --cachemode writeback -L10 -n cpool $vg/$lv1 + +# Can't convert 'writeback' cache +not lvconvert --thin --thinpool $vg/tpool $vg/$lv1 + +# Switch to 'writethrough' - this should be supported +lvchange --cachemode writethrough $vg/$lv1 + +lvconvert --thin $vg/$lv1 --originname extorg --thinpool $vg/tpool + +# check cache exist as extorg-real +check grep_dmsetup table ${vg}-extorg-real "cache" + + +# Split cache from external origin (while in-use) +lvconvert --splitcache $vg/extorg + +# check linear exist as extorg-real +check grep_dmsetup table ${vg}-extorg-real "linear" +check lv_field $vg/extorg segtype linear + +# Cache external origin in-use again +lvconvert -y -H $vg/extorg --cachepool $vg/cpool + +get lv_field $vg/extorg attr | grep "^ori" + +umount mnt + +# Is filesystem still ok ? +fsck -n "$DM_DEV_DIR/$vg/$lv1" + +lvchange -an $vg +lvchange -ay $vg + +# Remove thin, external origin remains +lvremove -f $vg/$lv1 + +#lvchange -prw $vg/extorg +lvconvert --uncache $vg/extorg + +lvremove -f $vg + +# +# Check some more API variants +# + +lvcreate -L10 -n pool $vg + +lvcreate -aey -L2 -n $lv1 $vg +lvcreate -H -L2 $vg/$lv1 + +# Converts $vg/pool to thin-pool AND $vg/$lv1 to thin +lvconvert -y --type thin $vg/$lv1 --originname extorg --thinpool $vg/pool + +check lv_field $vg/$lv1 segtype thin +check lv_field $vg/pool segtype thin-pool +check lv_field $vg/extorg segtype cache + +lvconvert --uncache $vg/extorg + +check lv_field $vg/extorg segtype linear + +vgremove -ff $vg

1 0

master - debug: add debug message showing new lv
by Zdenek Kabelac 19 Dec '16

19 Dec '16

Gitweb: http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=8c17233af585b4e4… Commit: 8c17233af585b4e46a86acca8ddb126beb5faf95 Parent: 034931f68dec96265519af6f35ea3726ba7eec82 Author: Zdenek Kabelac <zkabelac(a)redhat.com> AuthorDate: Sun Dec 18 15:05:18 2016 +0100 Committer: Zdenek Kabelac <zkabelac(a)redhat.com> CommitterDate: Sun Dec 18 19:38:51 2016 +0100 debug: add debug message showing new lv Make trace easier to follow knowing which LV was added to dtree. --- lib/activate/dev_manager.c | 2 ++ 1 files changed, 2 insertions(+), 0 deletions(-) diff --git a/lib/activate/dev_manager.c b/lib/activate/dev_manager.c index 73229b3..685c055 100644 --- a/lib/activate/dev_manager.c +++ b/lib/activate/dev_manager.c @@ -2718,6 +2718,8 @@ static int _add_new_lv_to_dtree(struct dev_manager *dm, struct dm_tree *dtree, int save_pending_delete = dm->track_pending_delete; int merge_in_progress = 0; + log_debug_activation("Adding new LV %s%s%s to dtree", display_lvname(lv), + layer ? "-" : "", layer ? : ""); /* LV with pending delete is never put new into a table */ if (lv_is_pending_delete(lv) && !_cached_dm_info(dm->mem, dtree, lv, NULL)) return 1; /* Replace with error only when already exists */

1 0

master - activate: further _info API refinement
by Zdenek Kabelac 19 Dec '16

19 Dec '16

Gitweb: http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=034931f68dec9626… Commit: 034931f68dec96265519af6f35ea3726ba7eec82 Parent: 79121416dfd695900a02711e6f63146e532b825b Author: Zdenek Kabelac <zkabelac(a)redhat.com> AuthorDate: Sun Dec 18 14:57:16 2016 +0100 Committer: Zdenek Kabelac <zkabelac(a)redhat.com> CommitterDate: Sun Dec 18 19:38:51 2016 +0100 activate: further _info API refinement Another cleanup of internal _info() API simplifying code. Also make sure 'error' on _info() call is properly passed upward (return 0 is error path). --- lib/activate/dev_manager.c | 88 ++++++++++++++++++++----------------------- 1 files changed, 41 insertions(+), 47 deletions(-) diff --git a/lib/activate/dev_manager.c b/lib/activate/dev_manager.c index 520a0a1..73229b3 100644 --- a/lib/activate/dev_manager.c +++ b/lib/activate/dev_manager.c @@ -214,8 +214,8 @@ typedef enum { STATUS, /* DM_DEVICE_STATUS ioctl */ } info_type_t; -static int _info_run(info_type_t type, const char *dlid, - struct dm_info *dminfo, uint32_t *read_ahead, +static int _info_run(const char *dlid, struct dm_info *dminfo, + uint32_t *read_ahead, struct lv_seg_status *seg_status, int with_open_count, int with_read_ahead, uint32_t major, uint32_t minor) @@ -223,26 +223,21 @@ static int _info_run(info_type_t type, const char *dlid, int r = 0; struct dm_task *dmt; int dmtask; + int with_flush; /* TODO: arg for _info_run */ void *target = NULL; uint64_t target_start, target_length, start, length; char *target_name, *target_params; - int with_flush = 1; /* TODO: arg for _info_run */ - switch (type) { - case INFO: - dmtask = DM_DEVICE_INFO; - break; - case STATUS: - dmtask = DM_DEVICE_STATUS; - with_flush = 0; - break; - default: - log_error(INTERNAL_ERROR "_info_run: unhandled info type."); - return 0; + if (seg_status) { + dmtask = DM_DEVICE_STATUS; + with_flush = 0; + } else { + dmtask = DM_DEVICE_INFO; + with_flush = 1; /* doesn't really matter */ } - if (!(dmt = _setup_task_run(dmtask, dminfo, - NULL, dlid, 0, major, minor, with_open_count, with_flush, 0))) + if (!(dmt = _setup_task_run(dmtask, dminfo, NULL, dlid, 0, major, minor, + with_open_count, with_flush, 0))) return_0; if (with_read_ahead && dminfo->exists) { @@ -252,7 +247,7 @@ static int _info_run(info_type_t type, const char *dlid, *read_ahead = DM_READ_AHEAD_NONE; /* Query status only for active device */ - if ((type == STATUS) && dminfo->exists) { + if (seg_status && dminfo->exists) { start = length = seg_status->seg->lv->vg->extent_size; start *= seg_status->seg->le; length *= seg_status->seg->len; @@ -674,18 +669,24 @@ static int _original_uuid_format_check_required(struct cmd_context *cmd) return (_kernel_major == -1); } -static int _info(struct cmd_context *cmd, const char *dlid, int with_open_count, int with_read_ahead, +static int _info(struct cmd_context *cmd, + const char *name, const char *dlid, + int with_open_count, int with_read_ahead, struct dm_info *dminfo, uint32_t *read_ahead, struct lv_seg_status *seg_status) { - int r = 0; char old_style_dlid[sizeof(UUID_PREFIX) + 2 * ID_LEN]; const char *suffix, *suffix_position; unsigned i = 0; + log_debug_activation("Getting device info for %s [%s].", name, dlid); + /* Check for dlid */ - if ((r = _info_run(seg_status ? STATUS : INFO, dlid, dminfo, read_ahead, - seg_status, with_open_count, with_read_ahead, 0, 0)) && dminfo->exists) + if (!_info_run(dlid, dminfo, read_ahead, seg_status, + with_open_count, with_read_ahead, 0, 0)) + return_0; + + if (dminfo->exists) return 1; /* Check for original version of dlid before the suffixes got added in 2.02.106 */ @@ -696,33 +697,33 @@ static int _info(struct cmd_context *cmd, const char *dlid, int with_open_count, (void) strncpy(old_style_dlid, dlid, sizeof(old_style_dlid)); old_style_dlid[sizeof(old_style_dlid) - 1] = '\0'; - if ((r = _info_run(seg_status ? STATUS : INFO, old_style_dlid, dminfo, - read_ahead, seg_status, with_open_count, - with_read_ahead, 0, 0)) && dminfo->exists) + if (!_info_run(old_style_dlid, dminfo, read_ahead, seg_status, + with_open_count, with_read_ahead, 0, 0)) + return_0; + if (dminfo->exists) return 1; } } /* Must we still check for the pre-2.02.00 dm uuid format? */ if (!_original_uuid_format_check_required(cmd)) - return r; + return 1; /* Check for dlid before UUID_PREFIX was added */ - if ((r = _info_run(seg_status ? STATUS : INFO, dlid + sizeof(UUID_PREFIX) - 1, - dminfo, read_ahead, seg_status, with_open_count, - with_read_ahead, 0, 0)) && dminfo->exists) - return 1; + if (!_info_run(dlid + sizeof(UUID_PREFIX) - 1, dminfo, read_ahead, seg_status, + with_open_count, with_read_ahead, 0, 0)) + return_0; - return r; + return 1; } static int _info_by_dev(uint32_t major, uint32_t minor, struct dm_info *info) { - return _info_run(INFO, NULL, info, NULL, 0, 0, 0, major, minor); + return _info_run(NULL, info, NULL, 0, 0, 0, major, minor); } -int dev_manager_info(struct cmd_context *cmd, const struct logical_volume *lv, - const char *layer, +int dev_manager_info(struct cmd_context *cmd, + const struct logical_volume *lv, const char *layer, int with_open_count, int with_read_ahead, struct dm_info *dminfo, uint32_t *read_ahead, struct lv_seg_status *seg_status) @@ -736,9 +737,9 @@ int dev_manager_info(struct cmd_context *cmd, const struct logical_volume *lv, if (!(dlid = build_dm_uuid(cmd->mem, lv, layer))) goto_out; - log_debug_activation("Getting device info for %s [%s].", name, dlid); - r = _info(cmd, dlid, with_open_count, with_read_ahead, - dminfo, read_ahead, seg_status); + if (!(r = _info(cmd, name, dlid, with_open_count, with_read_ahead, + dminfo, read_ahead, seg_status))) + stack; out: dm_pool_free(cmd->mem, name); @@ -1713,12 +1714,8 @@ static int _add_dev_to_dtree(struct dev_manager *dm, struct dm_tree *dtree, if (!(dlid = build_dm_uuid(dm->mem, lv, layer))) return_0; - log_debug_activation("Getting device info for %s [%s].", name, dlid); - - if (!_info(dm->cmd, dlid, 1, 0, &info, NULL, NULL)) { - log_error("Failed to get info for %s [%s].", name, dlid); - return 0; - } + if (!_info(dm->cmd, name, dlid, 1, 0, &info, NULL, NULL)) + return_0; /* * For top level volumes verify that existing device match @@ -2241,11 +2238,8 @@ static char *_add_error_or_zero_device(struct dev_manager *dm, struct dm_tree *d seg->lv->name, errid))) return_NULL; - log_debug_activation("Getting device info for %s [%s].", name, dlid); - if (!_info(dm->cmd, dlid, 1, 0, &info, NULL, NULL)) { - log_error("Failed to get info for %s [%s].", name, dlid); - return 0; - } + if (!_info(dm->cmd, name, dlid, 1, 0, &info, NULL, NULL)) + return_NULL; if (!info.exists) { /* Create new node */

1 0

master - thin: add comment with future extension
by Zdenek Kabelac 19 Dec '16

19 Dec '16

Gitweb: http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=79121416dfd69590… Commit: 79121416dfd695900a02711e6f63146e532b825b Parent: 75f23880934a8c1cb192e9a898261320db15043a Author: Zdenek Kabelac <zkabelac(a)redhat.com> AuthorDate: Sun Dec 18 15:06:12 2016 +0100 Committer: Zdenek Kabelac <zkabelac(a)redhat.com> CommitterDate: Sun Dec 18 19:38:51 2016 +0100 thin: add comment with future extension It could be actually better to use even cache origin in read-only mode so there could no be some 'acidental' change being done on this volume. This however need further tools enhancment - where we would need to handle whole subtree on 'lvchange -pr/-prw'. --- lib/metadata/thin_manip.c | 4 ++++ 1 files changed, 4 insertions(+), 0 deletions(-) diff --git a/lib/metadata/thin_manip.c b/lib/metadata/thin_manip.c index e7a2c33..42ec552 100644 --- a/lib/metadata/thin_manip.c +++ b/lib/metadata/thin_manip.c @@ -99,6 +99,10 @@ int attach_thin_external_origin(struct lv_segment *seg, external_lv->name); external_lv->status &= ~LVM_WRITE; } + + // TODO: should we mark even origin read-only ? + //if (lv_is_cache(external_lv)) /* read-only corigin of cache LV */ + // seg_lv(first_seg(external_lv), 0)->status &= ~LVM_WRITE; } return 1;

1 0

master - backup: show warning once per command
by Zdenek Kabelac 19 Dec '16

19 Dec '16

Gitweb: http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=75f23880934a8c1c… Commit: 75f23880934a8c1cb192e9a898261320db15043a Parent: 5bb6266046b11e6e4b47596689e3f4a75ba692a3 Author: Zdenek Kabelac <zkabelac(a)redhat.com> AuthorDate: Sun Dec 18 16:36:33 2016 +0100 Committer: Zdenek Kabelac <zkabelac(a)redhat.com> CommitterDate: Sun Dec 18 19:38:30 2016 +0100 backup: show warning once per command When command calls backup() more then once (which is actually not wanted) this warning message is shown repeatedly: "WARNING: This metadata update is NOT backed up." Instead now print message just once and less confuse user. --- WHATS_NEW | 1 + lib/format_text/archiver.c | 4 +++- 2 files changed, 4 insertions(+), 1 deletions(-) diff --git a/WHATS_NEW b/WHATS_NEW index e7abf59..fbaa369 100644 --- a/WHATS_NEW +++ b/WHATS_NEW @@ -1,5 +1,6 @@ Version 2.02.169 - ===================================== + Warn about command not making metadata backup just once per command. Enable usage of cached volume as thin volume's external origin. Support cache volume activation with -real layer. Improve search of lock-holder for external origin and thin-pool. diff --git a/lib/format_text/archiver.c b/lib/format_text/archiver.c index 92799e4..d381155 100644 --- a/lib/format_text/archiver.c +++ b/lib/format_text/archiver.c @@ -35,6 +35,7 @@ struct archive_params { struct backup_params { int enabled; char *dir; + int suppress; }; int archive_init(struct cmd_context *cmd, const char *dir, @@ -235,7 +236,8 @@ static int _backup(struct volume_group *vg) int backup_locally(struct volume_group *vg) { if (!vg->cmd->backup_params->enabled || !vg->cmd->backup_params->dir) { - log_warn("WARNING: This metadata update is NOT backed up"); + log_warn_suppress(vg->cmd->backup_params->suppress++, + "WARNING: This metadata update is NOT backed up."); return 1; }

1 0

master - lvconvert: support cache to external origin conversion
by Zdenek Kabelac 19 Dec '16

19 Dec '16

Gitweb: http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=5bb6266046b11e6e… Commit: 5bb6266046b11e6e4b47596689e3f4a75ba692a3 Parent: 69434c2eca5bb7667d73916a28d03b8c687aea32 Author: Zdenek Kabelac <zkabelac(a)redhat.com> AuthorDate: Sat Dec 17 22:41:27 2016 +0100 Committer: Zdenek Kabelac <zkabelac(a)redhat.com> CommitterDate: Sun Dec 18 19:35:27 2016 +0100 lvconvert: support cache to external origin conversion Add this functionality to lvconvert: 'lvconvert --thin cachedLV --thinpool vg/poll' Converts cachedLV to external origin (which will be read-only). New thin volume is created in thinpool LV and it's using external origin as source for unprovisioned chunks. This conversion happens online (while volume is in use). Thin LV remains fully writable. Cached external origin no longer could be written so cache will be used ONLY for read operations. For this limitation we require cache mode to be writethrough (as writeback cannot write to read-only volumes). When thinLV is later removed cached external origin is again fully usable, just note, LV remain in 'read-only' mode. When read-write is needed, 'lvchange -prw' has to be used. Single external origin could be user by multiple thinLV in multiple differen thin pool. --- WHATS_NEW | 1 + lib/metadata/cache_manip.c | 1 - tools/lvconvert.c | 39 +++++++++++++++++++++++++++++++++++++-- 3 files changed, 38 insertions(+), 3 deletions(-) diff --git a/WHATS_NEW b/WHATS_NEW index 74638fd..e7abf59 100644 --- a/WHATS_NEW +++ b/WHATS_NEW @@ -1,5 +1,6 @@ Version 2.02.169 - ===================================== + Enable usage of cached volume as thin volume's external origin. Support cache volume activation with -real layer. Improve search of lock-holder for external origin and thin-pool. Support status checking of cache volume used in layer. diff --git a/lib/metadata/cache_manip.c b/lib/metadata/cache_manip.c index ed72fae..12b1054 100644 --- a/lib/metadata/cache_manip.c +++ b/lib/metadata/cache_manip.c @@ -324,7 +324,6 @@ int validate_lv_cache_create_origin(const struct logical_volume *origin_lv) lv_is_thin_volume(origin_lv) || lv_is_thin_pool_metadata(origin_lv) || lv_is_origin(origin_lv) || lv_is_merging_origin(origin_lv) || lv_is_cow(origin_lv) || lv_is_merging_cow(origin_lv) || - lv_is_external_origin(origin_lv) || lv_is_virtual(origin_lv)) { log_error("Cache is not supported with %s segment type of the original logical volume %s.", first_seg(origin_lv)->segtype->name, display_lvname(origin_lv)); diff --git a/tools/lvconvert.c b/tools/lvconvert.c index dc32e37..393d286 100644 --- a/tools/lvconvert.c +++ b/tools/lvconvert.c @@ -2825,7 +2825,6 @@ static int _lvconvert_thin(struct cmd_context *cmd, if (lv_is_locked(lv) || !lv_is_visible(lv) || - lv_is_cache_type(lv) || lv_is_cow(lv) || lv_is_pool(lv) || lv_is_pool_data(lv) || @@ -3725,6 +3724,12 @@ static int _convert_cache_volume_splitmirrors(struct cmd_context *cmd, struct lo * Convert a cache LV to a thin pool (using the cache LV for thin pool data). * lvconvert --type thin-pool LV * + * Convert a cache LV to a thin volume with cached external origin using given + * thinpool tpLV (when not yet Thinpool convert it to thin-pool first). + * Conversion is 2-step process in this case. + * Only writethrough cacheLV can be converted as external origin is read-only. + * lvconvert --thin cacheLV --thinpool tpLV + * * Alternate syntax: * This is equivalent to above, but not preferred because it's ambiguous and inconsistent. * lvconvert --thinpool LV @@ -3732,7 +3737,37 @@ static int _convert_cache_volume_splitmirrors(struct cmd_context *cmd, struct lo static int _convert_cache_volume_thin_pool(struct cmd_context *cmd, struct logical_volume *lv, struct lvconvert_params *lp) { - return _lvconvert_pool(cmd, lv, lp); + int is_clean; + const struct lv_segment *pool_seg; + + if (!_lvconvert_pool(cmd, lv, lp)) + return_0; + + if (lv_is_cache(lv) && !lv_is_pool_data(lv)) { + pool_seg = first_seg(first_seg(lv)->pool_lv); + if (pool_seg->cache_mode != CACHE_MODE_WRITETHROUGH) { + log_error("Cannot convert cache volume %s with %s cache mode to external origin.", + display_lvname(lv), + get_cache_mode_name(pool_seg)); + log_error("To proceed, run 'lvchange --cachemode writethrough %s'.", + display_lvname(lv)); + return 0; + } + + if (!lv_cache_wait_for_clean(lv, &is_clean)) + return_0; + + if (!is_clean) { + log_error("Cache %s is not clean, refusing to convert to external origin.", + display_lvname(lv)); + return 0; + } + + if (!_lvconvert_thin(cmd, lv, lp)) + return_0; + } + + return 1; } /*

1 0

master - cache: improve activation with -real
by Zdenek Kabelac 19 Dec '16

19 Dec '16

Gitweb: http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=69434c2eca5bb766… Commit: 69434c2eca5bb7667d73916a28d03b8c687aea32 Parent: 954c59779d35436a6d2f28482ec1ed6dc128844c Author: Zdenek Kabelac <zkabelac(a)redhat.com> AuthorDate: Sun Dec 18 15:05:31 2016 +0100 Committer: Zdenek Kabelac <zkabelac(a)redhat.com> CommitterDate: Sun Dec 18 19:30:50 2016 +0100 cache: improve activation with -real When cache volume may be converted from normal to -real layer LV we need to improve logic for call cache_check. With this patch, we register call for cache_check only when metadata LV is not yet present in active table slot (should match initial table load). This avoids unwanted checking when cache would become layer device online. --- WHATS_NEW | 1 + lib/activate/dev_manager.c | 4 ++++ 2 files changed, 5 insertions(+), 0 deletions(-) diff --git a/WHATS_NEW b/WHATS_NEW index 6ab2b19..74638fd 100644 --- a/WHATS_NEW +++ b/WHATS_NEW @@ -1,5 +1,6 @@ Version 2.02.169 - ===================================== + Support cache volume activation with -real layer. Improve search of lock-holder for external origin and thin-pool. Support status checking of cache volume used in layer. Avoid shifting by one number of blocks when clearing dirty cache volume. diff --git a/lib/activate/dev_manager.c b/lib/activate/dev_manager.c index 8245454..520a0a1 100644 --- a/lib/activate/dev_manager.c +++ b/lib/activate/dev_manager.c @@ -2901,6 +2901,10 @@ static int _add_new_lv_to_dtree(struct dev_manager *dm, struct dm_tree *dtree, return_0; if (lv_is_cache(lv) && + /* Register callback only for layer activation or non-layered cache LV */ + (layer || !lv_layer(lv)) && + /* Register callback when metadata LV is NOT already active */ + !_cached_dm_info(dm->mem, dtree, first_seg(first_seg(lv)->pool_lv)->metadata_lv, NULL) && !_pool_register_callback(dm, dnode, lv)) return_0;

1 0

master - libdm: drop callback on revert path
by Zdenek Kabelac 19 Dec '16

19 Dec '16

Gitweb: http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=954c59779d35436a… Commit: 954c59779d35436a6d2f28482ec1ed6dc128844c Parent: 29b0e42be3026a11c86fc05f82f6a14e71d26cc8 Author: Zdenek Kabelac <zkabelac(a)redhat.com> AuthorDate: Sat Dec 17 22:40:59 2016 +0100 Committer: Zdenek Kabelac <zkabelac(a)redhat.com> CommitterDate: Sun Dec 18 19:29:08 2016 +0100 libdm: drop callback on revert path The system is likely in some very inconsisten state. Do not try to make it even more problematic with trying to invoke tools like thin_check via callback. --- WHATS_NEW_DM | 1 + libdm/libdm-deptree.c | 4 ++++ 2 files changed, 5 insertions(+), 0 deletions(-) diff --git a/WHATS_NEW_DM b/WHATS_NEW_DM index ab32bdb..d57ddd7 100644 --- a/WHATS_NEW_DM +++ b/WHATS_NEW_DM @@ -1,5 +1,6 @@ Version 1.02.138 - ===================================== + Do not try call callback when reverting activation on error path. Fix file mapping for extents with physically adjacent extents. Validation vsnprintf result in runtime translate of dm_log (1.02.136). Separate filemap extent allocation from region table. diff --git a/libdm/libdm-deptree.c b/libdm/libdm-deptree.c index cf6a06e..c5226a7 100644 --- a/libdm/libdm-deptree.c +++ b/libdm/libdm-deptree.c @@ -2778,6 +2778,10 @@ static int _dm_tree_revert_activated(struct dm_tree_node *parent) dm_list_iterate_items_gen(child, &parent->activated, activated_list) { log_debug_activation("Reverting %s.", child->name); + if (child->callback) { + log_debug_activation("Dropping callback for %s.", child->name); + child->callback = NULL; + } if (!_deactivate_node(child->name, child->info.major, child->info.minor, &child->dtree->cookie, child->udev_flags, 0)) { log_error("Unable to deactivate %s (%" PRIu32

1 0

master - lv: fix lock holder for external origin
by Zdenek Kabelac 19 Dec '16

19 Dec '16

Gitweb: http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=29b0e42be3026a11… Commit: 29b0e42be3026a11c86fc05f82f6a14e71d26cc8 Parent: a24eae6e82f365c66f4faeabc975ead0c476916b Author: Zdenek Kabelac <zkabelac(a)redhat.com> AuthorDate: Sat Dec 17 22:40:14 2016 +0100 Committer: Zdenek Kabelac <zkabelac(a)redhat.com> CommitterDate: Sun Dec 18 19:25:25 2016 +0100 lv: fix lock holder for external origin External origin could be reloaded via more locks. It's actually even more complex then thin-pool, as it may be active on more nodes for linear LVs (and maybe even more types). External origin is always read-only thus unmodifiable device so there should not be a problem accesing it through multiple nodes. Also for thin-pool check first presence of active thin-pool. FIXME: It's not easy to detect on which nodes this device is active Thus manipulation with such device may require checking every node and it active state and refresh. But since such setup is quite complex to prepare and use, hopefully there are not user trying to 'explore' this usage yet. --- WHATS_NEW | 1 + lib/metadata/lv.c | 24 +++++++++++++----------- 2 files changed, 14 insertions(+), 11 deletions(-) diff --git a/WHATS_NEW b/WHATS_NEW index 6b9a310..6ab2b19 100644 --- a/WHATS_NEW +++ b/WHATS_NEW @@ -1,5 +1,6 @@ Version 2.02.169 - ===================================== + Improve search of lock-holder for external origin and thin-pool. Support status checking of cache volume used in layer. Avoid shifting by one number of blocks when clearing dirty cache volume. Extend metadata validation of external origin LV use count. diff --git a/lib/metadata/lv.c b/lib/metadata/lv.c index 1084149..8587902 100644 --- a/lib/metadata/lv.c +++ b/lib/metadata/lv.c @@ -1551,14 +1551,19 @@ const struct logical_volume *lv_lock_holder(const struct logical_volume *lv) if (lv_is_cow(lv)) return lv_lock_holder(origin_from_cow(lv)); - if (lv_is_thin_pool(lv)) { - /* Find any active LV from the pool */ - dm_list_iterate_items(sl, &lv->segs_using_this_lv) - if (lv_is_active(sl->seg->lv)) { - log_debug_activation("Thin volume %s is active.", - display_lvname(lv)); - return sl->seg->lv; - } + if (lv_is_thin_pool(lv) || + lv_is_external_origin(lv)) { + /* FIXME: Ensure cluster keeps thin-pool active exlusively. + * External origin can be activated on more nodes (depends on type). + */ + if (!lv_is_active(lv)) + /* Find any active LV from the pool or external origin */ + dm_list_iterate_items(sl, &lv->segs_using_this_lv) + if (lv_is_active(sl->seg->lv)) { + log_debug_activation("Thin volume %s is active.", + display_lvname(lv)); + return sl->seg->lv; + } return lv; } @@ -1573,9 +1578,6 @@ const struct logical_volume *lv_lock_holder(const struct logical_volume *lv) lv_is_thin_volume(sl->seg->lv) && first_seg(lv)->pool_lv == sl->seg->pool_lv) continue; /* Skip thin snaphost */ - if (lv_is_external_origin(lv) && - lv_is_thin_volume(sl->seg->lv)) - continue; /* Skip external origin */ if (lv_is_pending_delete(sl->seg->lv)) continue; /* Skip deleted LVs */ return lv_lock_holder(sl->seg->lv);

1 0

master - cache: prepare status checking for layer
by Zdenek Kabelac 19 Dec '16

19 Dec '16

Gitweb: http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=a24eae6e82f365c6… Commit: a24eae6e82f365c66f4faeabc975ead0c476916b Parent: bf157ed833f8732c4e264422bd8be46fdf75673a Author: Zdenek Kabelac <zkabelac(a)redhat.com> AuthorDate: Sat Dec 17 21:52:27 2016 +0100 Committer: Zdenek Kabelac <zkabelac(a)redhat.com> CommitterDate: Sun Dec 18 19:23:13 2016 +0100 cache: prepare status checking for layer To be ready to show status of cache volume, call the status with layer. Layer is automatically detected in this case when cache volume is used in 'layered' form (needs -real suffix). --- WHATS_NEW | 1 + lib/activate/activate.c | 4 ++-- lib/metadata/cache_manip.c | 2 +- 3 files changed, 4 insertions(+), 3 deletions(-) diff --git a/WHATS_NEW b/WHATS_NEW index 2e9c953..6b9a310 100644 --- a/WHATS_NEW +++ b/WHATS_NEW @@ -1,5 +1,6 @@ Version 2.02.169 - ===================================== + Support status checking of cache volume used in layer. Avoid shifting by one number of blocks when clearing dirty cache volume. Extend metadata validation of external origin LV use count. Fix dm table when the last user of active external origin is removed. diff --git a/lib/activate/activate.c b/lib/activate/activate.c index c554337..b7009e6 100644 --- a/lib/activate/activate.c +++ b/lib/activate/activate.c @@ -770,7 +770,7 @@ int lv_info_with_seg_status(struct cmd_context *cmd, /* INFO is not set as cache-pool cannot be active. * STATUS is collected from cache LV */ lv_seg = get_only_segment_using_this_lv(lv); - (void) _lv_info(cmd, lv_seg->lv, 0, NULL, lv_seg, &status->seg_status, 0, 0); + (void) _lv_info(cmd, lv_seg->lv, 1, NULL, lv_seg, &status->seg_status, 0, 0); return 1; } @@ -1171,7 +1171,7 @@ int lv_cache_status(const struct logical_volume *cache_lv, return 0; } - if (!lv_info(cache_lv->vg->cmd, cache_lv, 0, NULL, 0, 0)) { + if (!lv_info(cache_lv->vg->cmd, cache_lv, 1, NULL, 0, 0)) { log_error("Cannot check status for locally inactive cache volume %s.", display_lvname(cache_lv)); return 0; diff --git a/lib/metadata/cache_manip.c b/lib/metadata/cache_manip.c index 54d907c..ed72fae 100644 --- a/lib/metadata/cache_manip.c +++ b/lib/metadata/cache_manip.c @@ -475,7 +475,7 @@ int lv_cache_remove(struct logical_volume *cache_lv) } /* Localy active volume is needed for writeback */ - if (!lv_is_active_locally(cache_lv)) { + if (!lv_info(cache_lv->vg->cmd, cache_lv, 1, NULL, 0, 0)) { /* Give up any remote locks */ if (!deactivate_lv(cache_lv->vg->cmd, cache_lv)) { log_error("Cannot deactivate remotely active cache volume %s.",

1 0

master - cache: improve wait for cache clear
by Zdenek Kabelac 19 Dec '16

19 Dec '16

Gitweb: http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=bf157ed833f8732c… Commit: bf157ed833f8732c4e264422bd8be46fdf75673a Parent: 36f609e513262547aa4d92c2b55d98b1471ba2e2 Author: Zdenek Kabelac <zkabelac(a)redhat.com> AuthorDate: Sun Dec 18 15:05:57 2016 +0100 Committer: Zdenek Kabelac <zkabelac(a)redhat.com> CommitterDate: Sun Dec 18 19:22:11 2016 +0100 cache: improve wait for cache clear Avoid printing misleading message about single dirty block. Instead properly detect condition where the 'cleaner' policy needs to be installed without 'overloading' dirty variable. Also print warning if we would be clearing read-only volume. (it really shouldn't happen). --- WHATS_NEW | 1 + lib/metadata/cache_manip.c | 17 ++++++++++------- 2 files changed, 11 insertions(+), 7 deletions(-) diff --git a/WHATS_NEW b/WHATS_NEW index 5002308..2e9c953 100644 --- a/WHATS_NEW +++ b/WHATS_NEW @@ -1,5 +1,6 @@ Version 2.02.169 - ===================================== + Avoid shifting by one number of blocks when clearing dirty cache volume. Extend metadata validation of external origin LV use count. Fix dm table when the last user of active external origin is removed. Improve reported lvs status for active external origin volume. diff --git a/lib/metadata/cache_manip.c b/lib/metadata/cache_manip.c index d73142d..54d907c 100644 --- a/lib/metadata/cache_manip.c +++ b/lib/metadata/cache_manip.c @@ -383,7 +383,7 @@ int lv_cache_wait_for_clean(struct logical_volume *cache_lv, int *is_clean) const struct logical_volume *lock_lv = lv_lock_holder(cache_lv); struct lv_segment *cache_seg = first_seg(cache_lv); struct lv_status_cache *status; - int cleaner_policy; + int cleaner_policy, writeback; uint64_t dirty_blocks; *is_clean = 0; @@ -402,14 +402,11 @@ int lv_cache_wait_for_clean(struct logical_volume *cache_lv, int *is_clean) cleaner_policy = !strcmp(status->cache->policy_name, "cleaner"); dirty_blocks = status->cache->dirty_blocks; - - /* No clear policy and writeback mode means dirty */ - if (!cleaner_policy && - (status->cache->feature_flags & DM_CACHE_FEATURE_WRITEBACK)) - dirty_blocks++; + writeback = (status->cache->feature_flags & DM_CACHE_FEATURE_WRITEBACK); dm_pool_destroy(status->mem); - if (!dirty_blocks) + /* Only clear when policy is Clear or mode != writeback */ + if (!dirty_blocks && (cleaner_policy || !writeback)) break; log_print_unless_silent("Flushing " FMTu64 " blocks for cache %s.", @@ -420,6 +417,12 @@ int lv_cache_wait_for_clean(struct logical_volume *cache_lv, int *is_clean) continue; } + if (!(cache_lv->status & LVM_WRITE)) { + log_warn("WARNING: Dirty blocks found on read-only cache volume %s.", + display_lvname(cache_lv)); + /* TODO: can we actually clean something? */ + } + /* Switch to cleaner policy to flush the cache */ cache_seg->cleaner_policy = 1; /* Reaload kernel with "cleaner" policy */

1 0

master - validation: check external property is matching
by Zdenek Kabelac 19 Dec '16

19 Dec '16

Gitweb: http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=36f609e513262547… Commit: 36f609e513262547aa4d92c2b55d98b1471ba2e2 Parent: 7db46c4a39fd945f295a5bd4191bd48828072d24 Author: Zdenek Kabelac <zkabelac(a)redhat.com> AuthorDate: Sat Dec 17 21:58:35 2016 +0100 Committer: Zdenek Kabelac <zkabelac(a)redhat.com> CommitterDate: Sun Dec 18 19:17:59 2016 +0100 validation: check external property is matching Detect if number of external_count is matching referencing devices for external_origin LV. --- WHATS_NEW | 1 + lib/metadata/merge.c | 12 ++++++++++++ 2 files changed, 13 insertions(+), 0 deletions(-) diff --git a/WHATS_NEW b/WHATS_NEW index 6d36890..5002308 100644 --- a/WHATS_NEW +++ b/WHATS_NEW @@ -1,5 +1,6 @@ Version 2.02.169 - ===================================== + Extend metadata validation of external origin LV use count. Fix dm table when the last user of active external origin is removed. Improve reported lvs status for active external origin volume. Fix table load for splitted RAID LV and require explicit activation. diff --git a/lib/metadata/merge.c b/lib/metadata/merge.c index 6d52bc4..947513b 100644 --- a/lib/metadata/merge.c +++ b/lib/metadata/merge.c @@ -398,6 +398,18 @@ int check_lv_segments(struct logical_volume *lv, int complete_vg) inc_error_count; } } + + if (lv_is_external_origin(lv)) { + seg_found = 0; + dm_list_iterate_items(sl, &lv->segs_using_this_lv) + if (sl->seg->external_lv == lv) + seg_found++; + if (seg_found != lv->external_count) { + log_error("LV %s: external origin count does not match.", + lv->name); + inc_error_count; + } + } } dm_list_iterate_items(seg, &lv->segments) {

1 0

master - thin: reload external origin with last thin
by Zdenek Kabelac 19 Dec '16

19 Dec '16

Gitweb: http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=7db46c4a39fd945f… Commit: 7db46c4a39fd945f295a5bd4191bd48828072d24 Parent: c71fefad8d5e1d9a075553ab69791ddbf6bcb153 Author: Zdenek Kabelac <zkabelac(a)redhat.com> AuthorDate: Sat Dec 17 21:55:02 2016 +0100 Committer: Zdenek Kabelac <zkabelac(a)redhat.com> CommitterDate: Sun Dec 18 19:13:34 2016 +0100 thin: reload external origin with last thin External origin could be activated as stand-alone device. When the last thin LV is removed, external origin is no longer the external origin and it's layer property was dropped. Ensure dm table is correct by reloading external origin (when it's active). --- WHATS_NEW | 1 + lib/metadata/lv_manip.c | 10 ++++++++++ 2 files changed, 11 insertions(+), 0 deletions(-) diff --git a/WHATS_NEW b/WHATS_NEW index 93fc93e..6d36890 100644 --- a/WHATS_NEW +++ b/WHATS_NEW @@ -1,5 +1,6 @@ Version 2.02.169 - ===================================== + Fix dm table when the last user of active external origin is removed. Improve reported lvs status for active external origin volume. Fix table load for splitted RAID LV and require explicit activation. Always active splitted RAID LV exclusively locally. diff --git a/lib/metadata/lv_manip.c b/lib/metadata/lv_manip.c index 1889998..92a6d44 100644 --- a/lib/metadata/lv_manip.c +++ b/lib/metadata/lv_manip.c @@ -1264,6 +1264,7 @@ static int _lv_reduce(struct logical_volume *lv, uint32_t extents, int delete) uint32_t count = extents; uint32_t reduction; struct logical_volume *pool_lv; + struct logical_volume *external_lv = NULL; if (lv_is_merging_origin(lv)) { log_debug_metadata("Dropping snapshot merge of %s to removed origin %s.", @@ -1275,6 +1276,9 @@ static int _lv_reduce(struct logical_volume *lv, uint32_t extents, int delete) if (!count) break; + if (seg->external_lv) + external_lv = seg->external_lv; + if (seg->len <= count) { if (seg->merge_lv) { log_debug_metadata("Dropping snapshot merge of removed %s to origin %s.", @@ -1341,6 +1345,12 @@ static int _lv_reduce(struct logical_volume *lv, uint32_t extents, int delete) !lv->vg->fid->fmt->ops->lv_setup(lv->vg->fid, lv)) return_0; + /* Removal of last user enforces refresh */ + if (external_lv && !lv_is_external_origin(external_lv) && + lv_is_active(external_lv) && + !lv_update_and_reload(external_lv)) + return_0; + return 1; }

1 0

master - lvs: show status for layer
by Zdenek Kabelac 19 Dec '16

19 Dec '16

Gitweb: http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=c71fefad8d5e1d9a… Commit: c71fefad8d5e1d9a075553ab69791ddbf6bcb153 Parent: bdfc96cb089031c6d95dccb0ff66616efdb16783 Author: Zdenek Kabelac <zkabelac(a)redhat.com> AuthorDate: Sat Dec 17 21:54:51 2016 +0100 Committer: Zdenek Kabelac <zkabelac(a)redhat.com> CommitterDate: Sun Dec 18 19:12:12 2016 +0100 lvs: show status for layer When LV is external origin, show info for LV but status for -layer. So we expose more info to a user as otherwise active external origin is only linear mapping of -real layer. We do the same for i.e. old snaphost origin. --- WHATS_NEW | 1 + lib/activate/activate.c | 7 +++++++ 2 files changed, 8 insertions(+), 0 deletions(-) diff --git a/WHATS_NEW b/WHATS_NEW index eee6e98..93fc93e 100644 --- a/WHATS_NEW +++ b/WHATS_NEW @@ -1,5 +1,6 @@ Version 2.02.169 - ===================================== + Improve reported lvs status for active external origin volume. Fix table load for splitted RAID LV and require explicit activation. Always active splitted RAID LV exclusively locally. Do not use LV RAID status bit for segment status. diff --git a/lib/activate/activate.c b/lib/activate/activate.c index b4c8a2c..c554337 100644 --- a/lib/activate/activate.c +++ b/lib/activate/activate.c @@ -784,6 +784,13 @@ int lv_info_with_seg_status(struct cmd_context *cmd, status->info.exists = 0; /* So pool LV is not active */ } return 1; + } else if (lv_is_external_origin(lv)) { + if (!_lv_info(cmd, lv, 0, &status->info, NULL, NULL, + with_open_count, with_read_ahead)) + return_0; + + (void) _lv_info(cmd, lv, 1, NULL, lv_seg, &status->seg_status, 0, 0); + return 1; } else if (lv_is_origin(lv)) { /* Query segment status for 'layered' (-real) device most of the time, * only for merging snapshot, query its progress.

1 0

master - raid: fix activation of tracked image
by Zdenek Kabelac 19 Dec '16

19 Dec '16

Gitweb: http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=bdfc96cb089031c6… Commit: bdfc96cb089031c6d95dccb0ff66616efdb16783 Parent: a15f0d181ca2fb44c9c854dfba425d372a8756c6 Author: Zdenek Kabelac <zkabelac(a)redhat.com> AuthorDate: Wed Dec 14 21:47:38 2016 +0100 Committer: Zdenek Kabelac <zkabelac(a)redhat.com> CommitterDate: Sun Dec 18 19:10:38 2016 +0100 raid: fix activation of tracked image Activation of raid has brough up also splitted image with tracing (without taking lock for this). So when raid is now activate - such image is not put into table (with _rmeta). When user needs such device, just active it. --- WHATS_NEW | 3 ++- lib/activate/dev_manager.c | 3 +++ lib/metadata/metadata-exported.h | 1 + 3 files changed, 6 insertions(+), 1 deletions(-) diff --git a/WHATS_NEW b/WHATS_NEW index 4ac3377..eee6e98 100644 --- a/WHATS_NEW +++ b/WHATS_NEW @@ -1,8 +1,9 @@ Version 2.02.169 - ===================================== + Fix table load for splitted RAID LV and require explicit activation. Always active splitted RAID LV exclusively locally. Do not use LV RAID status bit for segment status. - Check segtype directly instead of checking RAID in segment status. + Check segtype directly instead of checking RAID in segment status. Reusing exiting code for raid image removal. Fix pvmove leaving -pvmove0 error device in clustered VG. Avoid adding extra '_' at end of raid extracted images or metadata. diff --git a/lib/activate/dev_manager.c b/lib/activate/dev_manager.c index 581e900..8245454 100644 --- a/lib/activate/dev_manager.c +++ b/lib/activate/dev_manager.c @@ -2681,12 +2681,15 @@ static int _add_segment_to_dtree(struct dev_manager *dm, /* Add any LVs used by this segment */ for (s = 0; s < seg->area_count; ++s) { if ((seg_type(seg, s) == AREA_LV) && + /* do not bring up tracked image */ + !lv_is_raid_image_with_tracking(seg_lv(seg, s)) && /* origin only for cache without pending delete */ (!dm->track_pending_delete || !seg_is_cache(seg)) && !_add_new_lv_to_dtree(dm, dtree, seg_lv(seg, s), laopts, NULL)) return_0; if (seg_is_raid_with_meta(seg) && seg->meta_areas && seg_metalv(seg, s) && + !lv_is_raid_image_with_tracking(seg_lv(seg, s)) && !_add_new_lv_to_dtree(dm, dtree, seg_metalv(seg, s), laopts, NULL)) return_0; diff --git a/lib/metadata/metadata-exported.h b/lib/metadata/metadata-exported.h index 4bf466b..aa40c6c 100644 --- a/lib/metadata/metadata-exported.h +++ b/lib/metadata/metadata-exported.h @@ -226,6 +226,7 @@ #define lv_is_raid(lv) (((lv)->status & RAID) ? 1 : 0) #define lv_is_raid_image(lv) (((lv)->status & RAID_IMAGE) ? 1 : 0) +#define lv_is_raid_image_with_tracking(lv) ((lv_is_raid_image(lv) && !((lv)->status & LVM_WRITE)) ? 1 : 0) #define lv_is_raid_metadata(lv) (((lv)->status & RAID_META) ? 1 : 0) #define lv_is_raid_type(lv) (((lv)->status & (RAID | RAID_IMAGE | RAID_META)) ? 1 : 0)

1 0

master - dmstats: don't declare _start_timestamp if HAVE_SYS_TIMERFD_H
by Bryn Reeves 18 Dec '16

18 Dec '16

Gitweb: http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=a15f0d181ca2fb44… Commit: a15f0d181ca2fb44c9c854dfba425d372a8756c6 Parent: 3e53adf7c0b83194a74a9fa75a7ebbf1c077de9a Author: Bryn M. Reeves <bmr(a)redhat.com> AuthorDate: Sun Dec 18 14:08:11 2016 +0000 Committer: Bryn M. Reeves <bmr(a)redhat.com> CommitterDate: Sun Dec 18 14:08:11 2016 +0000 dmstats: don't declare _start_timestamp if HAVE_SYS_TIMERFD_H The _start_timestamp is not used by the TIMERFD clock. --- tools/dmsetup.c | 8 +++++--- 1 files changed, 5 insertions(+), 3 deletions(-) diff --git a/tools/dmsetup.c b/tools/dmsetup.c index e5ac80a..3f2c619 100644 --- a/tools/dmsetup.c +++ b/tools/dmsetup.c @@ -282,13 +282,15 @@ const char *_stats_types[] = { /* report timekeeping */ static struct dm_timestamp *_cycle_timestamp = NULL; +#ifndef HAVE_SYS_TIMERFD_H static struct dm_timestamp *_start_timestamp = NULL; +#else /* HAVE_SYS_TIMERFD_H */ +static int _timer_fd = -1; /* timerfd file descriptor. */ +#endif /* !HAVE_SYS_TIMERFD_H */ + static uint64_t _interval = 0; /* configured interval in nsecs */ static uint64_t _new_interval = 0; /* flag top-of-interval */ static uint64_t _last_interval = 0; /* approx. measured interval in nsecs */ -#ifdef HAVE_SYS_TIMERFD_H -static int _timer_fd = -1; /* timerfd file descriptor. */ -#endif /* HAVE_SYS_TIMERFD_H */ /* Invalid fd value used to signal end-of-reporting. */ #define TIMER_STOPPED -2

1 0

master - dmstats: fix TIMERFD _timer_running() test
by Bryn Reeves 18 Dec '16

18 Dec '16

Gitweb: http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=3e53adf7c0b83194… Commit: 3e53adf7c0b83194a74a9fa75a7ebbf1c077de9a Parent: 5a4750d76c9930a6e4a181ab664f0cac4d1a400b Author: Bryn M. Reeves <bmr(a)redhat.com> AuthorDate: Sun Dec 18 14:07:25 2016 +0000 Committer: Bryn M. Reeves <bmr(a)redhat.com> CommitterDate: Sun Dec 18 14:07:25 2016 +0000 dmstats: fix TIMERFD _timer_running() test --- tools/dmsetup.c | 5 +---- 1 files changed, 1 insertions(+), 4 deletions(-) diff --git a/tools/dmsetup.c b/tools/dmsetup.c index 06d609b..e5ac80a 100644 --- a/tools/dmsetup.c +++ b/tools/dmsetup.c @@ -652,10 +652,7 @@ static int _do_timer_wait(void) static int _timer_running(void) { - /* - * Clock shutdown for exit - nothing to do. - */ - return ((_timer_fd == TIMER_STOPPED) && !_cycle_timestamp); + return ((_timer_fd != TIMER_STOPPED) || _cycle_timestamp); } #else /* !HAVE_SYS_TIMERFD_H */

1 0

master - dmstats: fix interval number reporting with --count=0
by Bryn Reeves 18 Dec '16

18 Dec '16

Gitweb: http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=5a4750d76c9930a6… Commit: 5a4750d76c9930a6e4a181ab664f0cac4d1a400b Parent: 5635cd3b0337ac854a82a1a15cef2609c2091e6c Author: Bryn M. Reeves <bmr(a)redhat.com> AuthorDate: Sun Dec 18 12:58:03 2016 +0000 Committer: Bryn M. Reeves <bmr(a)redhat.com> CommitterDate: Sun Dec 18 13:03:45 2016 +0000 dmstats: fix interval number reporting with --count=0 When --count=0 interval numbers are miscalculated: Interval #18446744069414584325 time delta: 999920887ns Interval #18446744069414584325 current err: -79113ns End interval #18446744069414584325 duration: 999920887ns This is because the command line argument is cast through the uint32_t type, and fixed to UINT32_MAX: _count = ((uint32_t)_int_args[COUNT_ARG]) ? : UINT32_MAX; We also need to handle --count=0 specially when calculating the interval number: since intervals count from #1, this must account for the implicit "minus one" when converting from zero to the UINT64_MAX value used (which is too large to store in _int_args). --- tools/dmsetup.c | 13 +++++++------ 1 files changed, 7 insertions(+), 6 deletions(-) diff --git a/tools/dmsetup.c b/tools/dmsetup.c index b1967db..06d609b 100644 --- a/tools/dmsetup.c +++ b/tools/dmsetup.c @@ -264,7 +264,7 @@ static struct dm_tree *_dtree; static struct dm_report *_report; static report_type_t _report_type; static dev_name_t _dev_name_type; -static uint32_t _count = 1; /* count of repeating reports */ +static uint64_t _count = 1; /* count of repeating reports */ static struct dm_timestamp *_initial_timestamp = NULL; static uint64_t _disp_factor = 512; /* display sizes in sectors */ static char _disp_units = 's'; @@ -557,7 +557,8 @@ static void _destroy_split_name(struct dm_split_name *split_name) */ static uint64_t _interval_num(void) { - return 1 + (uint64_t) _int_args[COUNT_ARG] - _count; + uint64_t count_arg = _int_args[COUNT_ARG]; + return ((uint64_t) _int_args[COUNT_ARG] - _count) + !!count_arg; } #ifdef HAVE_SYS_TIMERFD_H @@ -6873,10 +6874,10 @@ unknown: goto_out; } - if (_switches[COUNT_ARG]) - _count = ((uint32_t)_int_args[COUNT_ARG]) ? : UINT32_MAX; - else if (_switches[INTERVAL_ARG]) - _count = UINT32_MAX; + if (_switches[COUNT_ARG] && _int_args[COUNT_ARG]) + _count = (uint64_t)_int_args[COUNT_ARG]; + else if (_switches[COUNT_ARG] || _switches[INTERVAL_ARG]) + _count = UINT64_MAX; if (_switches[UNITS_ARG]) { _disp_factor = _factor_from_units(_string_args[UNITS_ARG],

1 0

master - dmstats: separate TIMERFD and useleep() exit conditions
by Bryn Reeves 18 Dec '16

18 Dec '16

Gitweb: http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=5635cd3b0337ac85… Commit: 5635cd3b0337ac854a82a1a15cef2609c2091e6c Parent: 886b4f755d7f33194f92dafbf2c22517db4de374 Author: Bryn M. Reeves <bmr(a)redhat.com> AuthorDate: Sun Dec 18 12:42:47 2016 +0000 Committer: Bryn M. Reeves <bmr(a)redhat.com> CommitterDate: Sun Dec 18 13:03:44 2016 +0000 dmstats: separate TIMERFD and useleep() exit conditions The time management code mixes tests of the _timer_fd value with code that should be timer agnostic: this causes problems for users of the usleep() timer, since it cannot properly detect the start of a new interval: Beginning first interval Interval #18446744069414584348 time delta: 1000000000ns Interval #18446744069414584348 current err: 0ns End interval #18446744069414584348 duration: 1000000000ns Adjusted sample interval duration: 1000000000ns [...] Beginning first interval Interval #18446744069414584349 time delta: 1000000000ns Interval #18446744069414584349 current err: 0ns End interval #18446744069414584349 duration: 1000000000ns Adjusted sample interval duration: 1000000000ns Separate these out, by defining a _timer_running() call that each timer implements, and only define _timer_fd if we are compiling with TIMERFD enabled. --- tools/dmsetup.c | 19 +++++++++++++++++-- 1 files changed, 17 insertions(+), 2 deletions(-) diff --git a/tools/dmsetup.c b/tools/dmsetup.c index 17c5742..b1967db 100644 --- a/tools/dmsetup.c +++ b/tools/dmsetup.c @@ -286,7 +286,9 @@ static struct dm_timestamp *_start_timestamp = NULL; static uint64_t _interval = 0; /* configured interval in nsecs */ static uint64_t _new_interval = 0; /* flag top-of-interval */ static uint64_t _last_interval = 0; /* approx. measured interval in nsecs */ +#ifdef HAVE_SYS_TIMERFD_H static int _timer_fd = -1; /* timerfd file descriptor. */ +#endif /* HAVE_SYS_TIMERFD_H */ /* Invalid fd value used to signal end-of-reporting. */ #define TIMER_STOPPED -2 @@ -647,6 +649,14 @@ static int _do_timer_wait(void) return _do_timerfd_wait(); } +static int _timer_running(void) +{ + /* + * Clock shutdown for exit - nothing to do. + */ + return ((_timer_fd == TIMER_STOPPED) && !_cycle_timestamp); +} + #else /* !HAVE_SYS_TIMERFD_H */ static int _start_usleep_timer(void) { @@ -718,6 +728,11 @@ static int _do_timer_wait(void) return _do_usleep_wait(); } +static int _timer_running(void) +{ + return (_start_timestamp != NULL); +} + #endif /* HAVE_SYS_TIMERFD_H */ static int _update_interval_times(void) @@ -729,7 +744,7 @@ static int _update_interval_times(void) /* * Clock shutdown for exit - nothing to do. */ - if ((_timer_fd == TIMER_STOPPED) && !_cycle_timestamp) + if (!_timer_running()) goto out; /* clock is running */ @@ -805,7 +820,7 @@ static int _update_interval_times(void) out: /* timer stopped or never started */ - if (!r || _timer_fd < 0) { + if (!r || !_timer_running()) { /* The _cycle_timestamp has not yet been allocated if we * fail to obtain this_timestamp on the first interval. */

1 0

master - dmstats: use better interval estimate for usleep() timer
by Bryn Reeves 18 Dec '16

18 Dec '16

Gitweb: http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=886b4f755d7f3319… Commit: 886b4f755d7f33194f92dafbf2c22517db4de374 Parent: 68ec42ebaf288911b1cb95882678ca778b795953 Author: Bryn M. Reeves <bmr(a)redhat.com> AuthorDate: Sun Dec 18 12:39:26 2016 +0000 Committer: Bryn M. Reeves <bmr(a)redhat.com> CommitterDate: Sun Dec 18 13:03:44 2016 +0000 dmstats: use better interval estimate for usleep() timer Although the usleep() interval timer is not used if the Linux TIMERFD interface is available it should still provide reasonably good timing. Instead of trying to estimate the error from the duration of the last sleep, peg it to the start time of the program, and use the value of ((start_time - now) % interval) to correct the current interval duration. This always pulls us back into sync at the end of each interval, rather than relying on trying to incrementally adjust the time duration at each interval start. This greatly reduces drift when the usleep() clock is used. --- tools/dmsetup.c | 20 +++++++++++--------- 1 files changed, 11 insertions(+), 9 deletions(-) diff --git a/tools/dmsetup.c b/tools/dmsetup.c index 1be19d5..17c5742 100644 --- a/tools/dmsetup.c +++ b/tools/dmsetup.c @@ -282,6 +282,7 @@ const char *_stats_types[] = { /* report timekeeping */ static struct dm_timestamp *_cycle_timestamp = NULL; +static struct dm_timestamp *_start_timestamp = NULL; static uint64_t _interval = 0; /* configured interval in nsecs */ static uint64_t _new_interval = 0; /* flag top-of-interval */ static uint64_t _last_interval = 0; /* approx. measured interval in nsecs */ @@ -650,12 +651,14 @@ static int _do_timer_wait(void) static int _start_usleep_timer(void) { log_debug("Using usleep for interval timekeeping."); + _start_timestamp = dm_timestamp_alloc(); + dm_timestamp_get(_start_timestamp); return 1; } static int _do_usleep_wait(void) { - static struct dm_timestamp *_last_sleep, *_now = NULL; + static struct dm_timestamp *_now = NULL; uint64_t this_interval; int64_t delta_t; @@ -664,9 +667,7 @@ static int _do_usleep_wait(void) * message ioctls by keeping track of the last wake time and * adjusting the sleep interval accordingly. */ - if (!_last_sleep && !_now) { - if (!(_last_sleep = dm_timestamp_alloc())) - return_0; + if (!_now) { if (!(_now = dm_timestamp_alloc())) return_0; dm_timestamp_get(_now); @@ -674,19 +675,19 @@ static int _do_usleep_wait(void) log_error("Using "FMTu64" as first interval.", this_interval); } else { dm_timestamp_get(_now); - delta_t = dm_timestamp_delta(_now, _last_sleep); - log_debug("Interval timer delta_t: "FMTi64, delta_t); + delta_t = dm_timestamp_delta(_now, _start_timestamp); + log_debug("Interval timer drift: "FMTi64, + (delta_t % _interval)); /* FIXME: usleep timer drift over large counts. */ /* adjust for time spent populating and reporting */ - this_interval = 2 * _interval - delta_t; + this_interval = _interval - (delta_t % _interval); log_debug("Using "FMTu64" as interval.", this_interval); } /* Signal that a new interval has begun. */ _new_interval = 1; - dm_timestamp_copy(_last_sleep, _now); if (usleep(this_interval / NSEC_PER_USEC)) { if (errno == EINTR) @@ -699,8 +700,9 @@ static int _do_usleep_wait(void) } if (_count == 2) { - dm_timestamp_destroy(_last_sleep); + dm_timestamp_destroy(_start_timestamp); dm_timestamp_destroy(_now); + _start_timestamp = _now = NULL; } return 1;

1 0

master - dmstats: improve tool help output and option coverage
by Bryn Reeves 18 Dec '16

18 Dec '16

Gitweb: http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=68ec42ebaf288911… Commit: 68ec42ebaf288911b1cb95882678ca778b795953 Parent: 4f9d901c7142a0f2a73eb0353db992407187a144 Author: Bryn M. Reeves <bmr(a)redhat.com> AuthorDate: Sun Dec 18 10:53:27 2016 +0000 Committer: Bryn M. Reeves <bmr(a)redhat.com> CommitterDate: Sun Dec 18 11:51:13 2016 +0000 dmstats: improve tool help output and option coverage --- tools/dmsetup.c | 104 +++++++++++++++++++++++++++++++++++++++++-------------- 1 files changed, 78 insertions(+), 26 deletions(-) diff --git a/tools/dmsetup.c b/tools/dmsetup.c index f068293..1be19d5 100644 --- a/tools/dmsetup.c +++ b/tools/dmsetup.c @@ -5608,48 +5608,100 @@ static int _stats_help(CMD_ARGS); * dmsetup stats <cmd> [options] [device_name] * dmstats <cmd> [options] [device_name] * - * clear [--regionid id] <device_name> - * create [--areas nr_areas] [--areasize size] - * [ [--start start] [--length len] | [--segments]] - * [--userdata data] [--programid id] [<device_name>] - * delete [--regionid] <device_name> - * delete_all [--programid id] - * group [--alias name] [--alldevices] [--regions <regions>] [<device_name>] - * list [--programid id] [<device_name>] - * print [--clear] [--programid id] [--regionid id] [<device_name>] - * report [--interval seconds] [--count count] [--units units] [--regionid id] - * [--programid id] [<device>] - * ungroup [--alldevices] [--groupid id] [<device_name>] + * clear [--allregions|--regionid id] [--alldevices|<device...>] + * create [--start <start> [--length <len>] + * [--areas <nr_areas>] [--areasize <size>] + * [--programid <id>] [--userdata <data> ] + * [--bounds histogram_boundaries] [--precise] + * [--alldevices|<device...>] + * create --filemap [--nogroup] + * [--programid <id>] [--userdata <data> ] + * [--bounds histogram_boundaries] [--precise] [<file_path>] + * delete [--allprograms|--programid id] + * [--allregions|--regionid id] + * [--alldevices|<device...>] + * group [--alias NAME] --regions <regions> + * [--allprograms|--programid id] [--alldevices|<device...>] + * list [--allprograms|--programid id] [--allregions|--regionid id] + * print [--clear] [--allprograms|--programid id] + * [--allregions|--regionid id] + * [--alldevices|<device...>] + * report [--interval <seconds>] [--count <cnt>] + * [--units <u>] [--programid <id>] [--regionid <id>] + * [-o <fields>] [-O|--sort <sort_fields>] + * [-S|--select <selection>] [--nameprefixes] + * [--noheadings] [--separator <separator>] + * [--allprograms|--programid id] [<device...>] + * ungroup --groupid <id> [--allprograms|--programid id] + * [--alldevices|<device...>] */ +#define INDENT "\n\t " +/* groups of commonly used options */ #define AREA_OPTS "[--areas <nr_areas>] [--areasize <size>] " -#define CREATE_OPTS "[--start <start> [--length <len>]]\n\t\t" AREA_OPTS +#define REGION_OPTS "[--start <start> [--length <len>]" INDENT AREA_OPTS #define ID_OPTS "[--programid <id>] [--userdata <data> ] " #define SELECT_OPTS "[--programid <id>] [--regionid <id>] " -#define PRINT_OPTS "[--clear] " SELECT_OPTS -#define REPORT_OPTS "[--interval <seconds>] [--count <cnt>]\n\t\t[--units <u>]" SELECT_OPTS -#define GROUP_OPTS "[--alias NAME] --regions <regions>" +#define HIST_OPTS "[--bounds histogram_boundaries] " +#define PRECISE_OPTS "[--precise] " +#define SEGMENTS_OPT "[--segments] " +#define EXTRA_OPTS HIST_OPTS PRECISE_OPTS +#define ALL_PROGS_OPT "[--allprograms|--programid id] " +#define ALL_REGIONS_OPT "[--allregions|--regionid id] " +#define ALL_DEVICES_OPT "[--alldevices|<device...>] " +#define ALL_PROGS_REGIONS_DEVICES ALL_PROGS_OPT INDENT ALL_REGIONS_OPT INDENT ALL_DEVICES_OPT +#define FIELD_OPTS "[-o <fields>] [-O|--sort <sort_fields>]" +#define DM_REPORT_OPTS FIELD_OPTS INDENT "[-S|--select <selection>] [--nameprefixes]" INDENT \ +"[--noheadings] [--separator <separator>]" + +/* command options */ +#define CREATE_OPTS REGION_OPTS INDENT ID_OPTS INDENT EXTRA_OPTS INDENT SEGMENTS_OPT +#define FILEMAP_OPTS "--filemap [--nogroup]" INDENT ID_OPTS INDENT EXTRA_OPTS +#define PRINT_OPTS "[--clear] " ALL_PROGS_REGIONS_DEVICES +#define REPORT_OPTS "[--interval <seconds>] [--count <cnt>]" INDENT \ +"[--units <u>] " SELECT_OPTS INDENT DM_REPORT_OPTS INDENT ALL_PROGS_OPT +#define GROUP_OPTS "[--alias NAME] --regions <regions>" INDENT ALL_PROGS_OPT ALL_DEVICES_OPT +#define UNGROUP_OPTS ALL_PROGS_OPT INDENT ALL_DEVICES_OPT +/* + * The 'create' command has two entries in the table, to allow for the + * the fact that 'create' and 'create --filemap' have largely disjoint + * sets of options. + */ static struct command _stats_subcommands[] = { {"help", "", 0, 0, 0, 0, _stats_help}, - {"clear", "--regionid <id> [<device>]", 0, -1, 1, 0, _stats_clear}, - {"create", CREATE_OPTS "\n\t\t" ID_OPTS "[<device>]", 0, -1, 1, 0, _stats_create}, - {"delete", "--regionid <id> <device>", 1, -1, 1, 0, _stats_delete}, + {"clear", ALL_REGIONS_OPT ALL_DEVICES_OPT, 0, -1, 1, 0, _stats_clear}, + {"create", CREATE_OPTS ALL_DEVICES_OPT, 0, -1, 1, 0, _stats_create}, + {"create", FILEMAP_OPTS "[<file_path>]", 0, -1, 1, 0, _stats_create}, + {"delete", ALL_PROGS_REGIONS_DEVICES, 1, -1, 1, 0, _stats_delete}, {"group", GROUP_OPTS, 1, -1, 1, 0, _stats_group}, - {"list", "[--programid <id>] [<device>]", 0, -1, 1, 0, _stats_report}, - {"print", PRINT_OPTS "[<device>]", 0, -1, 1, 0, _stats_print}, - {"report", REPORT_OPTS "[<device>]", 0, -1, 1, 0, _stats_report}, - {"ungroup", "--groupid <id> [device]", 1, -1, 1, 0, _stats_ungroup}, + {"list", ALL_PROGS_OPT ALL_REGIONS_OPT, 0, -1, 1, 0, _stats_report}, + {"print", PRINT_OPTS, 0, -1, 1, 0, _stats_print}, + {"report", REPORT_OPTS "[<device...>]", 0, -1, 1, 0, _stats_report}, + {"ungroup", "--groupid <id> " UNGROUP_OPTS, 1, -1, 1, 0, _stats_ungroup}, {"version", "", 0, -1, 1, 0, _version}, {NULL, NULL, 0, 0, 0, 0, NULL} }; #undef AREA_OPTS -#undef CREATE_OPTS +#undef REGION_OPTS #undef ID_OPTS +#undef SELECT_OPTS +#undef HIST_OPTS +#undef PRECISE_OPTS +#undef EXTRA_OPTS +#undef ALL_PROGS_OPT +#undef ALL_REGIONS_OPT +#undef ALL_DEVICES_OPT +#undef ALL_PROGS_REGIONS_DEVICES +#undef FIELD_OPTS +#undef DM_REPORT_OPTS +#undef CREATE_OPTS +#undef FILEMAP_OPTS #undef PRINT_OPTS #undef REPORT_OPTS -#undef SELECT_OPTS +#undef GROUP_OPTS +#undef UNGROUP_OPTS static int _dmsetup_help(CMD_ARGS); @@ -5723,7 +5775,7 @@ static void _stats_usage(FILE *out) for (i = 0; _stats_subcommands[i].name; i++) fprintf(out, "\t%s %s\n", _stats_subcommands[i].name, _stats_subcommands[i].help); - fprintf(out, "<device> may be device name or -u <uuid> or " + fprintf(out, "\n<device> may be device name or -u <uuid> or " "-j <major> -m <minor>\n"); fprintf(out, "<fields> are comma-separated. Use 'help -c' for list.\n"); fprintf(out, "\n");

1 0

master - man: fix 'dmstats create' formatting in dmstats.8.in
by Bryn Reeves 18 Dec '16

18 Dec '16

Gitweb: http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=4f9d901c7142a0f2… Commit: 4f9d901c7142a0f2a73eb0353db992407187a144 Parent: 14be8c4fad7cd50cbd0006b6824552733f40bafd Author: Bryn M. Reeves <bmr(a)redhat.com> AuthorDate: Sun Dec 18 10:23:12 2016 +0000 Committer: Bryn M. Reeves <bmr(a)redhat.com> CommitterDate: Sun Dec 18 10:23:12 2016 +0000 man: fix 'dmstats create' formatting in dmstats.8.in --- man/dmstats.8.in | 4 ++-- 1 files changed, 2 insertions(+), 2 deletions(-) diff --git a/man/dmstats.8.in b/man/dmstats.8.in index 5ee02a9..fbc7811 100644 --- a/man/dmstats.8.in +++ b/man/dmstats.8.in @@ -83,8 +83,8 @@ dmstats \(em device-mapper statistics management . ad l . BR create . RB [ device_name... -. RB | file_path... ] -. RB [ \-\-alldevices ] +. RB | file_path... +. RB | [ \-\-alldevices ]] . RB [ \-\-areas . IR nr_areas | \fB\-\-areasize . IR area_size ]

1 0

master - man: fix 'dmstats list' option formatting in dmstats.8.in
by Bryn Reeves 18 Dec '16

18 Dec '16

Gitweb: http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=14be8c4fad7cd50c… Commit: 14be8c4fad7cd50cbd0006b6824552733f40bafd Parent: 25dd3988c3a5bed78f8f5c76ca2b24a7c95866a2 Author: Bryn M. Reeves <bmr(a)redhat.com> AuthorDate: Sun Dec 18 10:12:56 2016 +0000 Committer: Bryn M. Reeves <bmr(a)redhat.com> CommitterDate: Sun Dec 18 10:12:56 2016 +0000 man: fix 'dmstats list' option formatting in dmstats.8.in --- man/dmstats.8.in | 2 +- 1 files changed, 1 insertions(+), 1 deletions(-) diff --git a/man/dmstats.8.in b/man/dmstats.8.in index 7f451fd..5ee02a9 100644 --- a/man/dmstats.8.in +++ b/man/dmstats.8.in @@ -155,7 +155,7 @@ dmstats \(em device-mapper statistics management . OPT_OBJECTS . RB \%[ \-\-nosuffix ] . RB [ \-\-notimesuffix ] -. RB \%[ \-v | \-\-verbose [ \-v | \-\-verbose ]] +. RB \%[ \-v | \-\-verbose] . ad b .. .CMD_LIST

1 0

master - man: fix 'dmstats <command>' formatting in dmstats.8.in
by Bryn Reeves 18 Dec '16

18 Dec '16

Gitweb: http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=25dd3988c3a5bed7… Commit: 25dd3988c3a5bed78f8f5c76ca2b24a7c95866a2 Parent: 35791689ba5ef95da45290fd12ce9cff55c86258 Author: Bryn M. Reeves <bmr(a)redhat.com> AuthorDate: Sun Dec 18 10:11:02 2016 +0000 Committer: Bryn M. Reeves <bmr(a)redhat.com> CommitterDate: Sun Dec 18 10:12:45 2016 +0000 man: fix 'dmstats <command>' formatting in dmstats.8.in --- man/dmstats.8.in | 8 +++++--- 1 files changed, 5 insertions(+), 3 deletions(-) diff --git a/man/dmstats.8.in b/man/dmstats.8.in index 2367b8b0..7f451fd 100644 --- a/man/dmstats.8.in +++ b/man/dmstats.8.in @@ -53,12 +53,14 @@ dmstats \(em device-mapper statistics management .de CMD_COMMAND . ad l . IR command -. RI [ device_name -. RB | \-\-uuid -. IR uuid | \fB\-\-major +. RI [ device_name | +. RB [ \-u | \-\-uuid +. IR uuid ] +. RB | [ \-\-major . IR major . BR \-\-minor . IR minor ] +. RB \%[ \-v | \-\-verbose] . ad b .. .CMD_COMMAND

1 0

master - libdm: use destination size as limit in dm_bit_copy()
by Bryn Reeves 14 Dec '16

14 Dec '16

Gitweb: http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=35791689ba5ef95d… Commit: 35791689ba5ef95da45290fd12ce9cff55c86258 Parent: 0f98d5c2e6d103a46bb5eca75ac496622933c475 Author: Bryn M. Reeves <bmr(a)redhat.com> AuthorDate: Mon Dec 12 20:28:29 2016 +0000 Committer: Bryn M. Reeves <bmr(a)redhat.com> CommitterDate: Wed Dec 14 11:28:11 2016 +0000 libdm: use destination size as limit in dm_bit_copy() The dm_bit_copy() macro uses the source (bs1) bitset size as the limit for memcpy: memcpy((bs1) + 1, (bs2) + 1, ((*(bs1) / DM_BITS_PER_INT) + 1)..) This is safe if the destination bitset is smaller than the source, or if the two bitsets are of the same size. With a destination that is larger (e.g. when resizing a bitmap to add more capacity), the memcpy will overrun the source bitset and set garbage bits in the destination. There are nine uses of the macro currently (8 in libdm/regex, and 1 in daemons/cmirrord): in each case the two bitsets are always of equal size so the behaviour is unchanged. Fix the macro to use bs2's size to simplify resizing bitsets and avoid the need for another copy macro. --- libdm/libdevmapper.h | 2 +- 1 files changed, 1 insertions(+), 1 deletions(-) diff --git a/libdm/libdevmapper.h b/libdm/libdevmapper.h index bcf784b..ed46795 100644 --- a/libdm/libdevmapper.h +++ b/libdm/libdevmapper.h @@ -2090,7 +2090,7 @@ int dm_bit_get_prev(dm_bitset_t bs, int last_bit); memset((bs) + 1, 0, ((*(bs) / DM_BITS_PER_INT) + 1) * sizeof(int)) #define dm_bit_copy(bs1, bs2) \ - memcpy((bs1) + 1, (bs2) + 1, ((*(bs1) / DM_BITS_PER_INT) + 1) * sizeof(int)) + memcpy((bs1) + 1, (bs2) + 1, ((*(bs2) / DM_BITS_PER_INT) + 1) * sizeof(int)) /* * Parse a string representation of a bitset into a dm_bitset_t. The

1 0

master - cleanup: use exiting function
by Zdenek Kabelac 14 Dec '16

14 Dec '16

Gitweb: http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=0f98d5c2e6d103a4… Commit: 0f98d5c2e6d103a46bb5eca75ac496622933c475 Parent: fecd043cca4b5f4d5a244d715c60ce35ff1b62d7 Author: Zdenek Kabelac <zkabelac(a)redhat.com> AuthorDate: Wed Dec 14 11:34:28 2016 +0100 Committer: Zdenek Kabelac <zkabelac(a)redhat.com> CommitterDate: Wed Dec 14 11:41:42 2016 +0100 cleanup: use exiting function Reuse existing code and some indent change. --- lib/metadata/lv_manip.c | 4 ++-- lib/metadata/raid_manip.c | 9 ++------- 2 files changed, 4 insertions(+), 9 deletions(-) diff --git a/lib/metadata/lv_manip.c b/lib/metadata/lv_manip.c index 211b190..1889998 100644 --- a/lib/metadata/lv_manip.c +++ b/lib/metadata/lv_manip.c @@ -6258,12 +6258,12 @@ static int _lv_update_and_reload(struct logical_volume *lv, int origin_only) int lv_update_and_reload(struct logical_volume *lv) { - return _lv_update_and_reload(lv, 0); + return _lv_update_and_reload(lv, 0); } int lv_update_and_reload_origin(struct logical_volume *lv) { - return _lv_update_and_reload(lv, 1); + return _lv_update_and_reload(lv, 1); } /* diff --git a/lib/metadata/raid_manip.c b/lib/metadata/raid_manip.c index b48b959..5ec86a3 100644 --- a/lib/metadata/raid_manip.c +++ b/lib/metadata/raid_manip.c @@ -1464,13 +1464,8 @@ int lv_raid_split(struct logical_volume *lv, const char *split_name, /* * Eliminate the residual LVs */ - dm_list_iterate_items(lvl, &removal_lvs) { - if (!deactivate_lv(cmd, lvl->lv)) - return_0; - - if (!lv_remove(lvl->lv)) - return_0; - } + if (!_deactivate_and_remove_lvs(lv->vg, &removal_lvs)) + return_0; if (!vg_write(lv->vg) || !vg_commit(lv->vg)) return_0;

1 0

master - raid: split preserves local exlusive activation
by Zdenek Kabelac 14 Dec '16

14 Dec '16

Gitweb: http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=fecd043cca4b5f4d… Commit: fecd043cca4b5f4d5a244d715c60ce35ff1b62d7 Parent: 77e09c3fb4d019ab3a0472059481ef990c64b415 Author: Zdenek Kabelac <zkabelac(a)redhat.com> AuthorDate: Wed Dec 14 10:19:25 2016 +0100 Committer: Zdenek Kabelac <zkabelac(a)redhat.com> CommitterDate: Wed Dec 14 11:40:01 2016 +0100 raid: split preserves local exlusive activation --- WHATS_NEW | 1 + lib/metadata/raid_manip.c | 19 ++----------------- 2 files changed, 3 insertions(+), 17 deletions(-) diff --git a/WHATS_NEW b/WHATS_NEW index 60e8fa5..4ac3377 100644 --- a/WHATS_NEW +++ b/WHATS_NEW @@ -1,5 +1,6 @@ Version 2.02.169 - ===================================== + Always active splitted RAID LV exclusively locally. Do not use LV RAID status bit for segment status. Check segtype directly instead of checking RAID in segment status. Reusing exiting code for raid image removal. diff --git a/lib/metadata/raid_manip.c b/lib/metadata/raid_manip.c index ed48239..b48b959 100644 --- a/lib/metadata/raid_manip.c +++ b/lib/metadata/raid_manip.c @@ -163,22 +163,6 @@ uint32_t lv_raid_image_count(const struct logical_volume *lv) return seg->area_count; } -static int _activate_sublv_preserving_excl(struct logical_volume *top_lv, - struct logical_volume *sub_lv) -{ - struct cmd_context *cmd = top_lv->vg->cmd; - - /* If top RAID was EX, use EX */ - if (lv_is_active_exclusive_locally(top_lv)) { - if (!activate_lv_excl_local(cmd, sub_lv)) - return_0; - } else { - if (!activate_lv(cmd, sub_lv)) - return_0; - } - return 1; -} - /* HM Helper: prohibit allocation on @pv if @lv already has segments allocated on it */ static int _avoid_pv_of_lv(struct logical_volume *lv, struct physical_volume *pv) { @@ -1553,7 +1537,8 @@ int lv_raid_split_and_track(struct logical_volume *lv, display_lvname(lv)); /* Activate the split (and tracking) LV */ - if (!_activate_sublv_preserving_excl(lv, seg_lv(seg, s))) + /* Preserving exclusive local activation also for tracked LV */ + if (!activate_lv_excl_local(lv->vg->cmd, seg_lv(seg, s))) return_0; log_print_unless_silent("Use 'lvconvert --merge %s' to merge back into %s.",

1 0

master - raid: activation with list
by Zdenek Kabelac 14 Dec '16

14 Dec '16

Gitweb: http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=77e09c3fb4d019ab… Commit: 77e09c3fb4d019ab3a0472059481ef990c64b415 Parent: 4a05f83278a1b1967b91d8f47fa1f7ce18341439 Author: Zdenek Kabelac <zkabelac(a)redhat.com> AuthorDate: Wed Dec 14 11:16:47 2016 +0100 Committer: Zdenek Kabelac <zkabelac(a)redhat.com> CommitterDate: Wed Dec 14 11:37:02 2016 +0100 raid: activation with list Commit 069039204002e5c8514050fe541bbd378c383a02 revealed a problem in raid metadata manipulation. We do two operations in one table reload: - raid leg/image extraction - rename remaining raid legs This should be made in separate steps. Otherwise we do an uncorrectable table change on error path (leaving tables for admin and dmsetup). As a hotfix - restore the previous logic and use a single new function _lv_update_and_reload_list which activates exclusively extracted LVs on the list before resuming suspended raid LV. This restore 'rename' functionality upon resume. Also still preserve the 'origin_only' logic - although we know it's not working properly for cluster and LV stacking. Further fixes are needed. --- lib/metadata/raid_manip.c | 67 +++++++++++++++++++++++++++++++++++++++----- 1 files changed, 59 insertions(+), 8 deletions(-) diff --git a/lib/metadata/raid_manip.c b/lib/metadata/raid_manip.c index 049102a..ed48239 100644 --- a/lib/metadata/raid_manip.c +++ b/lib/metadata/raid_manip.c @@ -233,11 +233,6 @@ static int _deactivate_and_remove_lvs(struct volume_group *vg, struct dm_list *r { struct lv_list *lvl; - /* Need to take lock/resume for proper deactivation */ - dm_list_iterate_items(lvl, removal_lvs) - if (!activate_lv_excl_local(vg->cmd, lvl->lv)) - return_0; - dm_list_iterate_items(lvl, removal_lvs) { if (!deactivate_lv(vg->cmd, lvl->lv)) return_0; @@ -363,6 +358,62 @@ static int _raid_remove_top_layer(struct logical_volume *lv, return 1; } +/* + * Assisted excl_local activation of lvl listed LVs before resume + * + * FIXME: code which needs to use this function is usually unsafe + * againt crashes as it's doing more then 1 operation per commit + * and as such is currently irreversible on error path. + * + * Function is not making backup as this is usually not the last + * metadata changing operation. + * + * Also we should take 'struct lv_list'... + */ +static int _lv_update_and_reload_list(struct logical_volume *lv, int origin_only, struct dm_list *lv_list) +{ + struct volume_group *vg = lv->vg; + const struct logical_volume *lock_lv = lv_lock_holder(lv); + struct lv_list *lvl; + int r; + + log_very_verbose("Updating logical volume %s on disk(s)%s.", + display_lvname(lock_lv), origin_only ? " (origin only)": ""); + + if (!vg_write(vg)) + return_0; + + if (!(r = (origin_only ? suspend_lv_origin(vg->cmd, lock_lv) : suspend_lv(vg->cmd, lock_lv)))) { + log_error("Failed to lock logical volume %s.", + display_lvname(lock_lv)); + vg_revert(vg); + } else if (!(r = vg_commit(vg))) + stack; /* !vg_commit() has implict vg_revert() */ + + if (r && lv_list) { + dm_list_iterate_items(lvl, lv_list) { + log_very_verbose("Activating logical volume %s before %s in kernel.", + display_lvname(lvl->lv), display_lvname(lock_lv)); + if (!activate_lv_excl_local(vg->cmd, lvl->lv)) { + log_error("Failed to activate %s before resuming %s.", + display_lvname(lvl->lv), display_lvname(lock_lv)); + r = 0; /* But lets try with the rest */ + } + } + } + + log_very_verbose("Updating logical volume %s in kernel.", + display_lvname(lock_lv)); + + if (!(origin_only ? resume_lv_origin(vg->cmd, lock_lv) : resume_lv(vg->cmd, lock_lv))) { + log_error("Problem reactivating logical volume %s.", + display_lvname(lock_lv)); + r = 0; + } + + return r; +} + /* Makes on-disk metadata changes * If LV is active: * clear first block of device @@ -1233,7 +1284,7 @@ static int _raid_remove_images(struct logical_volume *lv, if (!commit) return 1; - if (!lv_update_and_reload(lv)) + if (!_lv_update_and_reload_list(lv, 0, removal_lvs)) return_0; /* @@ -1927,7 +1978,7 @@ static int _raid0_add_or_remove_metadata_lvs(struct logical_volume *lv, return_0; if (update_and_reload) { - if (!lv_update_and_reload_origin(lv)) + if (!_lv_update_and_reload_list(lv, 1, removal_lvs)) return_0; /* If any residual LVs, eliminate them, write VG, commit it and take a backup */ @@ -1977,7 +2028,7 @@ static int _lv_update_reload_fns_reset_eliminate_lvs(struct logical_volume *lv, { int flags_were_cleared = 0, r = 0; - if (!lv_update_and_reload_origin(lv)) + if (!_lv_update_and_reload_list(lv, 1, removal_lvs)) return_0; /* Eliminate any residual LV and don't commit the metadata */

1 0

master - configure: just move new macro to right file
by Zdenek Kabelac 14 Dec '16

14 Dec '16

Gitweb: http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=4a05f83278a1b196… Commit: 4a05f83278a1b1967b91d8f47fa1f7ce18341439 Parent: f4401fe351f63679ae8f4bab448a9283fb631a4a Author: Zdenek Kabelac <zkabelac(a)redhat.com> AuthorDate: Tue Dec 13 22:48:56 2016 +0100 Committer: Zdenek Kabelac <zkabelac(a)redhat.com> CommitterDate: Tue Dec 13 22:49:59 2016 +0100 configure: just move new macro to right file aclocal is regenerated while acinclude is permanent. Move new macro to permanent file. --- acinclude.m4 | 171 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ aclocal.m4 | 170 --------------------------------------------------------- 2 files changed, 171 insertions(+), 170 deletions(-) diff --git a/acinclude.m4 b/acinclude.m4 index b6c9b18..cd02300 100644 --- a/acinclude.m4 +++ b/acinclude.m4 @@ -61,3 +61,174 @@ AC_DEFUN([AC_TRY_LDFLAGS], ifelse([$4], [], [:], [$4]) fi ]) + +# =========================================================================== +# http://www.gnu.org/software/autoconf-archive/ax_gcc_builtin.html +# =========================================================================== +# +# SYNOPSIS +# +# AX_GCC_BUILTIN(BUILTIN) +# +# DESCRIPTION +# +# This macro checks if the compiler supports one of GCC's built-in +# functions; many other compilers also provide those same built-ins. +# +# The BUILTIN parameter is the name of the built-in function. +# +# If BUILTIN is supported define HAVE_<BUILTIN>. Keep in mind that since +# builtins usually start with two underscores they will be copied over +# into the HAVE_<BUILTIN> definition (e.g. HAVE___BUILTIN_EXPECT for +# __builtin_expect()). +# +# The macro caches its result in the ax_cv_have_<BUILTIN> variable (e.g. +# ax_cv_have___builtin_expect). +# +# The macro currently supports the following built-in functions: +# +# __builtin_assume_aligned +# __builtin_bswap16 +# __builtin_bswap32 +# __builtin_bswap64 +# __builtin_choose_expr +# __builtin___clear_cache +# __builtin_clrsb +# __builtin_clrsbl +# __builtin_clrsbll +# __builtin_clz +# __builtin_clzl +# __builtin_clzll +# __builtin_complex +# __builtin_constant_p +# __builtin_ctz +# __builtin_ctzl +# __builtin_ctzll +# __builtin_expect +# __builtin_ffs +# __builtin_ffsl +# __builtin_ffsll +# __builtin_fpclassify +# __builtin_huge_val +# __builtin_huge_valf +# __builtin_huge_vall +# __builtin_inf +# __builtin_infd128 +# __builtin_infd32 +# __builtin_infd64 +# __builtin_inff +# __builtin_infl +# __builtin_isinf_sign +# __builtin_nan +# __builtin_nand128 +# __builtin_nand32 +# __builtin_nand64 +# __builtin_nanf +# __builtin_nanl +# __builtin_nans +# __builtin_nansf +# __builtin_nansl +# __builtin_object_size +# __builtin_parity +# __builtin_parityl +# __builtin_parityll +# __builtin_popcount +# __builtin_popcountl +# __builtin_popcountll +# __builtin_powi +# __builtin_powif +# __builtin_powil +# __builtin_prefetch +# __builtin_trap +# __builtin_types_compatible_p +# __builtin_unreachable +# +# Unsuppored built-ins will be tested with an empty parameter set and the +# result of the check might be wrong or meaningless so use with care. +# +# LICENSE +# +# Copyright (c) 2013 Gabriele Svelto <gabriele.svelto(a)gmail.com> +# +# Copying and distribution of this file, with or without modification, are +# permitted in any medium without royalty provided the copyright notice +# and this notice are preserved. This file is offered as-is, without any +# warranty. + +#serial 3 + +AC_DEFUN([AX_GCC_BUILTIN], [ + AS_VAR_PUSHDEF([ac_var], [ax_cv_have_$1]) + + AC_CACHE_CHECK([for $1], [ac_var], [ + AC_LINK_IFELSE([AC_LANG_PROGRAM([], [ + m4_case([$1], + [__builtin_assume_aligned], [$1("", 0)], + [__builtin_bswap16], [$1(0)], + [__builtin_bswap32], [$1(0)], + [__builtin_bswap64], [$1(0)], + [__builtin_choose_expr], [$1(0, 0, 0)], + [__builtin___clear_cache], [$1("", "")], + [__builtin_clrsb], [$1(0)], + [__builtin_clrsbl], [$1(0)], + [__builtin_clrsbll], [$1(0)], + [__builtin_clz], [$1(0)], + [__builtin_clzl], [$1(0)], + [__builtin_clzll], [$1(0)], + [__builtin_complex], [$1(0.0, 0.0)], + [__builtin_constant_p], [$1(0)], + [__builtin_ctz], [$1(0)], + [__builtin_ctzl], [$1(0)], + [__builtin_ctzll], [$1(0)], + [__builtin_expect], [$1(0, 0)], + [__builtin_ffs], [$1(0)], + [__builtin_ffsl], [$1(0)], + [__builtin_ffsll], [$1(0)], + [__builtin_fpclassify], [$1(0, 1, 2, 3, 4, 0.0)], + [__builtin_huge_val], [$1()], + [__builtin_huge_valf], [$1()], + [__builtin_huge_vall], [$1()], + [__builtin_inf], [$1()], + [__builtin_infd128], [$1()], + [__builtin_infd32], [$1()], + [__builtin_infd64], [$1()], + [__builtin_inff], [$1()], + [__builtin_infl], [$1()], + [__builtin_isinf_sign], [$1(0.0)], + [__builtin_nan], [$1("")], + [__builtin_nand128], [$1("")], + [__builtin_nand32], [$1("")], + [__builtin_nand64], [$1("")], + [__builtin_nanf], [$1("")], + [__builtin_nanl], [$1("")], + [__builtin_nans], [$1("")], + [__builtin_nansf], [$1("")], + [__builtin_nansl], [$1("")], + [__builtin_object_size], [$1("", 0)], + [__builtin_parity], [$1(0)], + [__builtin_parityl], [$1(0)], + [__builtin_parityll], [$1(0)], + [__builtin_popcount], [$1(0)], + [__builtin_popcountl], [$1(0)], + [__builtin_popcountll], [$1(0)], + [__builtin_powi], [$1(0, 0)], + [__builtin_powif], [$1(0, 0)], + [__builtin_powil], [$1(0, 0)], + [__builtin_prefetch], [$1("")], + [__builtin_trap], [$1()], + [__builtin_types_compatible_p], [$1(int, int)], + [__builtin_unreachable], [$1()], + [m4_warn([syntax], [Unsupported built-in $1, the test may fail]) + $1()] + ) + ])], + [AS_VAR_SET([ac_var], [yes])], + [AS_VAR_SET([ac_var], [no])]) + ]) + + AS_IF([test yes = AS_VAR_GET([ac_var])], + [AC_DEFINE_UNQUOTED(AS_TR_CPP(HAVE_$1), 1, + [Define to 1 if the system has the `$1' built-in function])], []) + + AS_VAR_POPDEF([ac_var]) +]) diff --git a/aclocal.m4 b/aclocal.m4 index 1f02c2b..4ab64a8 100644 --- a/aclocal.m4 +++ b/aclocal.m4 @@ -536,175 +536,5 @@ AC_DEFUN([AM_RUN_LOG], echo "$as_me:$LINENO: \$? = $ac_status" >&AS_MESSAGE_LOG_FD (exit $ac_status); }]) -# =========================================================================== -# http://www.gnu.org/software/autoconf-archive/ax_gcc_builtin.html -# =========================================================================== -# -# SYNOPSIS -# -# AX_GCC_BUILTIN(BUILTIN) -# -# DESCRIPTION -# -# This macro checks if the compiler supports one of GCC's built-in -# functions; many other compilers also provide those same built-ins. -# -# The BUILTIN parameter is the name of the built-in function. -# -# If BUILTIN is supported define HAVE_<BUILTIN>. Keep in mind that since -# builtins usually start with two underscores they will be copied over -# into the HAVE_<BUILTIN> definition (e.g. HAVE___BUILTIN_EXPECT for -# __builtin_expect()). -# -# The macro caches its result in the ax_cv_have_<BUILTIN> variable (e.g. -# ax_cv_have___builtin_expect). -# -# The macro currently supports the following built-in functions: -# -# __builtin_assume_aligned -# __builtin_bswap16 -# __builtin_bswap32 -# __builtin_bswap64 -# __builtin_choose_expr -# __builtin___clear_cache -# __builtin_clrsb -# __builtin_clrsbl -# __builtin_clrsbll -# __builtin_clz -# __builtin_clzl -# __builtin_clzll -# __builtin_complex -# __builtin_constant_p -# __builtin_ctz -# __builtin_ctzl -# __builtin_ctzll -# __builtin_expect -# __builtin_ffs -# __builtin_ffsl -# __builtin_ffsll -# __builtin_fpclassify -# __builtin_huge_val -# __builtin_huge_valf -# __builtin_huge_vall -# __builtin_inf -# __builtin_infd128 -# __builtin_infd32 -# __builtin_infd64 -# __builtin_inff -# __builtin_infl -# __builtin_isinf_sign -# __builtin_nan -# __builtin_nand128 -# __builtin_nand32 -# __builtin_nand64 -# __builtin_nanf -# __builtin_nanl -# __builtin_nans -# __builtin_nansf -# __builtin_nansl -# __builtin_object_size -# __builtin_parity -# __builtin_parityl -# __builtin_parityll -# __builtin_popcount -# __builtin_popcountl -# __builtin_popcountll -# __builtin_powi -# __builtin_powif -# __builtin_powil -# __builtin_prefetch -# __builtin_trap -# __builtin_types_compatible_p -# __builtin_unreachable -# -# Unsuppored built-ins will be tested with an empty parameter set and the -# result of the check might be wrong or meaningless so use with care. -# -# LICENSE -# -# Copyright (c) 2013 Gabriele Svelto <gabriele.svelto(a)gmail.com> -# -# Copying and distribution of this file, with or without modification, are -# permitted in any medium without royalty provided the copyright notice -# and this notice are preserved. This file is offered as-is, without any -# warranty. - -#serial 3 - -AC_DEFUN([AX_GCC_BUILTIN], [ - AS_VAR_PUSHDEF([ac_var], [ax_cv_have_$1]) - - AC_CACHE_CHECK([for $1], [ac_var], [ - AC_LINK_IFELSE([AC_LANG_PROGRAM([], [ - m4_case([$1], - [__builtin_assume_aligned], [$1("", 0)], - [__builtin_bswap16], [$1(0)], - [__builtin_bswap32], [$1(0)], - [__builtin_bswap64], [$1(0)], - [__builtin_choose_expr], [$1(0, 0, 0)], - [__builtin___clear_cache], [$1("", "")], - [__builtin_clrsb], [$1(0)], - [__builtin_clrsbl], [$1(0)], - [__builtin_clrsbll], [$1(0)], - [__builtin_clz], [$1(0)], - [__builtin_clzl], [$1(0)], - [__builtin_clzll], [$1(0)], - [__builtin_complex], [$1(0.0, 0.0)], - [__builtin_constant_p], [$1(0)], - [__builtin_ctz], [$1(0)], - [__builtin_ctzl], [$1(0)], - [__builtin_ctzll], [$1(0)], - [__builtin_expect], [$1(0, 0)], - [__builtin_ffs], [$1(0)], - [__builtin_ffsl], [$1(0)], - [__builtin_ffsll], [$1(0)], - [__builtin_fpclassify], [$1(0, 1, 2, 3, 4, 0.0)], - [__builtin_huge_val], [$1()], - [__builtin_huge_valf], [$1()], - [__builtin_huge_vall], [$1()], - [__builtin_inf], [$1()], - [__builtin_infd128], [$1()], - [__builtin_infd32], [$1()], - [__builtin_infd64], [$1()], - [__builtin_inff], [$1()], - [__builtin_infl], [$1()], - [__builtin_isinf_sign], [$1(0.0)], - [__builtin_nan], [$1("")], - [__builtin_nand128], [$1("")], - [__builtin_nand32], [$1("")], - [__builtin_nand64], [$1("")], - [__builtin_nanf], [$1("")], - [__builtin_nanl], [$1("")], - [__builtin_nans], [$1("")], - [__builtin_nansf], [$1("")], - [__builtin_nansl], [$1("")], - [__builtin_object_size], [$1("", 0)], - [__builtin_parity], [$1(0)], - [__builtin_parityl], [$1(0)], - [__builtin_parityll], [$1(0)], - [__builtin_popcount], [$1(0)], - [__builtin_popcountl], [$1(0)], - [__builtin_popcountll], [$1(0)], - [__builtin_powi], [$1(0, 0)], - [__builtin_powif], [$1(0, 0)], - [__builtin_powil], [$1(0, 0)], - [__builtin_prefetch], [$1("")], - [__builtin_trap], [$1()], - [__builtin_types_compatible_p], [$1(int, int)], - [__builtin_unreachable], [$1()], - [m4_warn([syntax], [Unsupported built-in $1, the test may fail]) - $1()] - ) - ])], - [AS_VAR_SET([ac_var], [yes])], - [AS_VAR_SET([ac_var], [no])]) - ]) - - AS_IF([test yes = AS_VAR_GET([ac_var])], - [AC_DEFINE_UNQUOTED(AS_TR_CPP(HAVE_$1), 1, - [Define to 1 if the system has the `$1' built-in function])], []) - - AS_VAR_POPDEF([ac_var]) -]) m4_include([acinclude.m4])

1 0

master - libdm: ensure first extent is always counted
by Bryn Reeves 14 Dec '16

14 Dec '16

Gitweb: http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=f4401fe351f63679… Commit: f4401fe351f63679ae8f4bab448a9283fb631a4a Parent: fce7449d7311b49839f339b714c48e953e9a837a Author: Bryn M. Reeves <bmr(a)redhat.com> AuthorDate: Tue Dec 13 21:36:11 2016 +0000 Committer: Bryn M. Reeves <bmr(a)redhat.com> CommitterDate: Tue Dec 13 21:41:31 2016 +0000 libdm: ensure first extent is always counted If FIEMAP returns a single extent after the first call, no extent boundary is detected and the first extent is not counted by the normal mechanism. In this case, increment nr_extents at the same time the extent is added to the region table, before returning. --- libdm/libdm-stats.c | 4 +++- 1 files changed, 3 insertions(+), 1 deletions(-) diff --git a/libdm/libdm-stats.c b/libdm/libdm-stats.c index 02a30dc..8d202d2 100644 --- a/libdm/libdm-stats.c +++ b/libdm/libdm-stats.c @@ -4286,8 +4286,10 @@ static uint64_t _stats_map_extents(struct dm_pool *mem, * If the file only has a single extent, no boundary is ever * detected to trigger addition of the first extent. */ - if (fm_ext[i - 1].fe_logical == 0) + if (fm_ext[i - 1].fe_logical == 0) { _stats_add_extent(mem, fm_pending, nr_extents); + nr_extents++; + } fiemap->fm_start = (fm_ext[i - 1].fe_logical + fm_ext[i - 1].fe_length);

1 0

master - cleanup: remove wrapping function
by Zdenek Kabelac 14 Dec '16

14 Dec '16

Gitweb: http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=fce7449d7311b498… Commit: fce7449d7311b49839f339b714c48e953e9a837a Parent: c7da16e5f162193db16f2a1845200fc99ed6ebc3 Author: Zdenek Kabelac <zkabelac(a)redhat.com> AuthorDate: Tue Dec 13 12:28:12 2016 +0100 Committer: Zdenek Kabelac <zkabelac(a)redhat.com> CommitterDate: Tue Dec 13 22:07:52 2016 +0100 cleanup: remove wrapping function backup is not 'tested' for success and also it should actually happen just when command is finished. We do not target to make backups with each inter-step metadata change. --- lib/metadata/raid_manip.c | 23 ++++++----------------- 1 files changed, 6 insertions(+), 17 deletions(-) diff --git a/lib/metadata/raid_manip.c b/lib/metadata/raid_manip.c index 56fe6a7..049102a 100644 --- a/lib/metadata/raid_manip.c +++ b/lib/metadata/raid_manip.c @@ -226,21 +226,6 @@ static void _clear_allocation_prohibited(struct dm_list *pvs) pvl->pv->status &= ~PV_ALLOCATION_PROHIBITED; } -/* FIXME Move this out */ -/* Write, commit and optionally backup metadata of vg */ -static int _vg_write_commit_backup(struct volume_group *vg) -{ - if (!vg_write(vg) || !vg_commit(vg)) { - log_error("Failed to commit VG %s metadata.", vg->name); - return 0; - } - - if (!backup(vg)) - log_warn("WARNING: Backup of VG %s metadata failed. Continuing.", vg->name); - - return 1; -} - /* * Deactivate and remove the LVs on removal_lvs list from vg. */ @@ -1898,8 +1883,12 @@ static int _eliminate_extracted_lvs_optional_write_vg(struct volume_group *vg, dm_list_init(removal_lvs); - if (vg_write_requested && !_vg_write_commit_backup(vg)) - return_0; + if (vg_write_requested) { + if (!vg_write(vg) || !vg_commit(vg)) + return_0; + + backup(vg); + } return 1; }

1 0

master - cleanup: log message updates
by Zdenek Kabelac 14 Dec '16

14 Dec '16

Gitweb: http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=c7da16e5f162193d… Commit: c7da16e5f162193db16f2a1845200fc99ed6ebc3 Parent: a8f5e1f2744a568779445aa5def423f7e1210ee7 Author: Zdenek Kabelac <zkabelac(a)redhat.com> AuthorDate: Tue Dec 13 00:10:01 2016 +0100 Committer: Zdenek Kabelac <zkabelac(a)redhat.com> CommitterDate: Tue Dec 13 22:07:52 2016 +0100 cleanup: log message updates --- lib/metadata/raid_manip.c | 42 ++++++++++++++++++++++++------------------ 1 files changed, 24 insertions(+), 18 deletions(-) diff --git a/lib/metadata/raid_manip.c b/lib/metadata/raid_manip.c index 81e2ac2..56fe6a7 100644 --- a/lib/metadata/raid_manip.c +++ b/lib/metadata/raid_manip.c @@ -1056,8 +1056,10 @@ static int _extract_image_components(struct lv_segment *seg, uint32_t idx, struct logical_volume *data_lv = seg_lv(seg, idx); struct logical_volume *meta_lv = seg_metalv(seg, idx); - log_very_verbose("Extracting image components %s and %s from %s", - data_lv->name, meta_lv->name, seg->lv->name); + log_very_verbose("Extracting image components %s and %s from %s.", + display_lvname(data_lv), + display_lvname(meta_lv), + display_lvname(seg->lv)); data_lv->status &= ~RAID_IMAGE; meta_lv->status &= ~RAID_META; @@ -1185,8 +1187,9 @@ static int _raid_extract_images(struct logical_volume *lv, } } if (!_extract_image_components(seg, s, &rmeta_lv, &rimage_lv)) { - log_error("Failed to extract %s from %s", - seg_lv(seg, s)->name, lv->name); + log_error("Failed to extract %s from %s.", + display_lvname(seg_lv(seg, s)), + display_lvname(lv)); return 0; } @@ -1362,7 +1365,7 @@ int lv_raid_split(struct logical_volume *lv, const char *split_name, if (!lv_is_on_pvs(tracking, splittable_pvs)) { log_error("Unable to split additional image from %s " "while tracking changes for %s.", - display_lvname(lv), tracking->name); + display_lvname(lv), display_lvname(tracking)); return 0; } @@ -1502,7 +1505,7 @@ int lv_raid_split_and_track(struct logical_volume *lv, } if (s < 0) { - log_error("Unable to find image to satisfy request"); + log_error("Unable to find image to satisfy request."); return 0; } @@ -2042,12 +2045,12 @@ static int _adjust_data_lvs(struct logical_volume *lv, enum mirror_raid_conv dir dlv = seg_lv(seg, s); if (!(sublv_name_suffix = first_substring(dlv->name, "_mimage_", "_rimage_", NULL))) { - log_error(INTERNAL_ERROR "name lags image part"); + log_error(INTERNAL_ERROR "Name %s lags image part.", dlv->name); return 0; } *(sublv_name_suffix + 1) = conv[direction].type_char; - log_debug_metadata("data LV renamed to %s.", display_lvname(dlv)); + log_debug_metadata("Data LV renamed to %s.", display_lvname(dlv)); dlv->status &= ~conv[direction].reset_flag; dlv->status |= conv[direction].set_flag; @@ -3084,7 +3087,7 @@ static int _raid456_to_raid0_or_striped_wrapper(TAKEOVER_FN_ARGS) if (!yes && yes_no_prompt("Are you sure you want to convert \"%s\" LV %s to \"%s\" " "type losing all resilience? [y/n]: ", lvseg_name(seg), display_lvname(lv), new_segtype->name) == 'n') { - log_error("Logical volume %s NOT converted to \"%s\"", + log_error("Logical volume %s NOT converted to \"%s\".", display_lvname(lv), new_segtype->name); return 0; } @@ -3108,7 +3111,8 @@ static int _raid456_to_raid0_or_striped_wrapper(TAKEOVER_FN_ARGS) if (segtype_is_any_raid0(new_segtype) && !(rename_sublvs = _rename_area_lvs(lv, "_"))) { - log_error("Failed to rename %s LV %s MetaLVs", lvseg_name(seg), display_lvname(lv)); + log_error("Failed to rename %s LV %s MetaLVs.", + lvseg_name(seg), display_lvname(lv)); return 0; } @@ -3138,7 +3142,8 @@ static int _raid456_to_raid0_or_striped_wrapper(TAKEOVER_FN_ARGS) if (rename_sublvs) { if (!_rename_area_lvs(lv, NULL)) { - log_error("Failed to rename %s LV %s MetaLVs", lvseg_name(seg), display_lvname(lv)); + log_error("Failed to rename %s LV %s MetaLVs.", + lvseg_name(seg), display_lvname(lv)); return 0; } if (!lv_update_and_reload(lv)) @@ -3181,19 +3186,19 @@ static int _striped_or_raid0_to_raid45610_wrapper(TAKEOVER_FN_ARGS) return _takeover_unsupported_yet(lv, new_stripes, new_segtype); if (new_data_copies > new_image_count) { - log_error("N number of data_copies \"--mirrors N-1\" may not be larger than number of stripes"); + log_error("N number of data_copies \"--mirrors N-1\" may not be larger than number of stripes."); return 0; } if (new_stripes && new_stripes != seg->area_count) { - log_error("Can't restripe LV %s during conversion", display_lvname(lv)); + log_error("Can't restripe LV %s during conversion.", display_lvname(lv)); return 0; } /* FIXME: restricted to raid4 for the time being... */ if (!segtype_is_raid4(new_segtype)) { /* Can't convert striped/raid0* to e.g. raid10_offset */ - log_error("Can't convert %s to %s", display_lvname(lv), new_segtype->name); + log_error("Can't convert %s to %s.", display_lvname(lv), new_segtype->name); return 0; } @@ -3211,7 +3216,7 @@ static int _striped_or_raid0_to_raid45610_wrapper(TAKEOVER_FN_ARGS) /* Add metadata LVs */ if (seg_is_raid0(seg)) { - log_debug_metadata("Adding metadata LVs to %s", display_lvname(lv)); + log_debug_metadata("Adding metadata LVs to %s.", display_lvname(lv)); if (!_raid0_add_or_remove_metadata_lvs(lv, 1 /* update_and_reload */, allocate_pvs, NULL)) return_0; /* raid0_meta -> raid4 needs clearing of MetaLVs in order to avoid raid disk role cahnge issues in the kernel */ @@ -3220,7 +3225,8 @@ static int _striped_or_raid0_to_raid45610_wrapper(TAKEOVER_FN_ARGS) return_0; /* Add the additional component LV pairs */ - log_debug_metadata("Adding %" PRIu32 " component LV pair(s) to %s", new_image_count - lv_raid_image_count(lv), + log_debug_metadata("Adding %" PRIu32 " component LV pair(s) to %s.", + new_image_count - lv_raid_image_count(lv), display_lvname(lv)); if (!_lv_raid_change_image_count(lv, new_image_count, allocate_pvs, NULL, 0, 1)) return_0; @@ -3228,7 +3234,7 @@ static int _striped_or_raid0_to_raid45610_wrapper(TAKEOVER_FN_ARGS) if (segtype_is_raid4(new_segtype) && (!_shift_parity_dev(seg) || !_rename_area_lvs(lv, "_"))) { - log_error("Can't convert %s to %s", display_lvname(lv), new_segtype->name); + log_error("Can't convert %s to %s.", display_lvname(lv), new_segtype->name); return 0; } @@ -3239,7 +3245,7 @@ static int _striped_or_raid0_to_raid45610_wrapper(TAKEOVER_FN_ARGS) _check_and_adjust_region_size(lv); - log_debug_metadata("Updating VG metadata and reloading %s LV %s", + log_debug_metadata("Updating VG metadata and reloading %s LV %s.", lvseg_name(seg), display_lvname(lv)); if (!_lv_update_reload_fns_reset_eliminate_lvs(lv, NULL)) return_0;

1 0

master - cleanup: more lv_is_ usage
by Zdenek Kabelac 14 Dec '16

14 Dec '16

Gitweb: http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=a8f5e1f2744a5687… Commit: a8f5e1f2744a568779445aa5def423f7e1210ee7 Parent: 47b96c3537968caf0cfb8ffa3874d64944715bec Author: Zdenek Kabelac <zkabelac(a)redhat.com> AuthorDate: Tue Dec 13 00:09:15 2016 +0100 Committer: Zdenek Kabelac <zkabelac(a)redhat.com> CommitterDate: Tue Dec 13 22:07:52 2016 +0100 cleanup: more lv_is_ usage --- lib/format1/import-export.c | 2 +- lib/format1/import-extents.c | 2 +- lib/format_pool/import_export.c | 2 +- lib/format_text/import_vsn1.c | 2 +- lib/locking/lvmlockd.c | 2 +- lib/metadata/lv.c | 2 +- lib/metadata/lv_manip.c | 7 +++---- lib/metadata/raid_manip.c | 2 +- lib/metadata/snapshot_manip.c | 2 +- tools/toollib.c | 2 +- tools/vgchange.c | 4 ++-- tools/vgconvert.c | 2 +- tools/vgsplit.c | 4 ++-- 13 files changed, 17 insertions(+), 18 deletions(-) diff --git a/lib/format1/import-export.c b/lib/format1/import-export.c index 9b7f5ef..146623e 100644 --- a/lib/format1/import-export.c +++ b/lib/format1/import-export.c @@ -510,7 +510,7 @@ int export_lvs(struct disk_list *dl, struct volume_group *vg, goto_out; dm_list_iterate_items(ll, &vg->lvs) { - if (ll->lv->status & SNAPSHOT) + if (lv_is_snapshot(ll->lv)) continue; if (!(lvdl = dm_pool_alloc(dl->mem, sizeof(*lvdl)))) diff --git a/lib/format1/import-extents.c b/lib/format1/import-extents.c index 5cad4f2..3ab3ac4 100644 --- a/lib/format1/import-extents.c +++ b/lib/format1/import-extents.c @@ -56,7 +56,7 @@ static struct dm_hash_table *_create_lv_maps(struct dm_pool *mem, } dm_list_iterate_items(ll, &vg->lvs) { - if (ll->lv->status & SNAPSHOT) + if (lv_is_snapshot(ll->lv)) continue; if (!(lvm = dm_pool_alloc(mem, sizeof(*lvm)))) diff --git a/lib/format_pool/import_export.c b/lib/format_pool/import_export.c index 1529df5..2f0f2eb 100644 --- a/lib/format_pool/import_export.c +++ b/lib/format_pool/import_export.c @@ -267,7 +267,7 @@ int import_pool_segments(struct dm_list *lvs, struct dm_pool *mem, dm_list_iterate_items(lvl, lvs) { lv = lvl->lv; - if (lv->status & SNAPSHOT) + if (lv_is_snapshot(lv)) continue; for (i = 0; i < subpools; i++) { diff --git a/lib/format_text/import_vsn1.c b/lib/format_text/import_vsn1.c index e11b16b..00caf21 100644 --- a/lib/format_text/import_vsn1.c +++ b/lib/format_text/import_vsn1.c @@ -512,7 +512,7 @@ static int _read_segments(struct logical_volume *lv, const struct dm_config_node count++; } /* FIXME Remove this restriction */ - if ((lv->status & SNAPSHOT) && count > 1) { + if (lv_is_snapshot(lv) && count > 1) { log_error("Only one segment permitted for snapshot"); return 0; } diff --git a/lib/locking/lvmlockd.c b/lib/locking/lvmlockd.c index 8acf69c..92a970a 100644 --- a/lib/locking/lvmlockd.c +++ b/lib/locking/lvmlockd.c @@ -2651,7 +2651,7 @@ int lockd_lv_uses_lock(struct logical_volume *lv) if (lv_is_cow(lv)) return 0; - if (lv->status & SNAPSHOT) + if (lv_is_snapshot(lv)) return 0; /* FIXME: lv_is_virtual_origin ? */ diff --git a/lib/metadata/lv.c b/lib/metadata/lv.c index 8f13f6b..1084149 100644 --- a/lib/metadata/lv.c +++ b/lib/metadata/lv.c @@ -971,7 +971,7 @@ int lv_mirror_image_in_sync(const struct logical_volume *lv) struct lv_segment *seg = first_seg(lv); struct lv_segment *mirror_seg; - if (!(lv->status & MIRROR_IMAGE) || !seg || + if (!lv_is_mirror_image(lv) || !seg || !(mirror_seg = find_mirror_seg(seg))) { log_error(INTERNAL_ERROR "Cannot find mirror segment."); return 0; diff --git a/lib/metadata/lv_manip.c b/lib/metadata/lv_manip.c index fdfee36..211b190 100644 --- a/lib/metadata/lv_manip.c +++ b/lib/metadata/lv_manip.c @@ -1139,7 +1139,8 @@ int set_lv_segment_area_lv(struct lv_segment *seg, uint32_t area_num, display_lvname(seg->lv), seg->le, area_num, display_lvname(lv), le); - if (status & RAID_META) { + lv->status |= status; + if (lv_is_raid_metadata(lv)) { seg->meta_areas[area_num].type = AREA_LV; seg_metalv(seg, area_num) = lv; if (le) { @@ -1152,7 +1153,6 @@ int set_lv_segment_area_lv(struct lv_segment *seg, uint32_t area_num, seg_lv(seg, area_num) = lv; seg_le(seg, area_num) = le; } - lv->status |= status; if (!add_seg_to_segs_using_this_lv(lv, seg)) return_0; @@ -4678,8 +4678,7 @@ static int _lvresize_check(struct logical_volume *lv, if (lv_is_raid_image(lv) || lv_is_raid_metadata(lv)) { log_error("Cannot resize a RAID %s directly", - (lv->status & RAID_IMAGE) ? "image" : - "metadata area"); + lv_is_raid_image(lv) ? "image" : "metadata area"); return 0; } diff --git a/lib/metadata/raid_manip.c b/lib/metadata/raid_manip.c index b8a47c0..81e2ac2 100644 --- a/lib/metadata/raid_manip.c +++ b/lib/metadata/raid_manip.c @@ -135,7 +135,7 @@ static int _lv_is_raid_with_tracking(const struct logical_volume *lv, *tracking = NULL; - if (!(lv->status & RAID)) + if (!lv_is_raid(lv)) return 0; for (s = 0; s < seg->area_count; s++) diff --git a/lib/metadata/snapshot_manip.c b/lib/metadata/snapshot_manip.c index 2f13ac9..b5fb60c 100644 --- a/lib/metadata/snapshot_manip.c +++ b/lib/metadata/snapshot_manip.c @@ -121,7 +121,7 @@ int lv_is_visible(const struct logical_volume *lv) if (lv_is_historical(lv)) return 1; - if (lv->status & SNAPSHOT) + if (lv_is_snapshot(lv)) return 0; if (lv_is_cow(lv)) { diff --git a/tools/toollib.c b/tools/toollib.c index 9f399b8..235e535 100644 --- a/tools/toollib.c +++ b/tools/toollib.c @@ -2402,7 +2402,7 @@ int process_each_lv_in_vg(struct cmd_context *cmd, struct volume_group *vg, goto_out; } - if (lvl->lv->status & SNAPSHOT) + if (lv_is_snapshot(lvl->lv)) continue; /* Skip availability change for non-virt snaps when processing all LVs */ diff --git a/tools/vgchange.c b/tools/vgchange.c index 455f688..ca0f450 100644 --- a/tools/vgchange.c +++ b/tools/vgchange.c @@ -107,11 +107,11 @@ static int _activate_lvs_in_vg(struct cmd_context *cmd, struct volume_group *vg, lv = origin_from_cow(lv); /* Only request activation of snapshot origin devices */ - if ((lv->status & SNAPSHOT) || lv_is_cow(lv)) + if (lv_is_snapshot(lv) || lv_is_cow(lv)) continue; /* Only request activation of mirror LV */ - if ((lv->status & MIRROR_IMAGE) || (lv->status & MIRROR_LOG)) + if (lv_is_mirror_image(lv) || lv_is_mirror_log(lv)) continue; /* Only request activation of the first replicator-dev LV */ diff --git a/tools/vgconvert.c b/tools/vgconvert.c index 4f28ba3..2c0ac08 100644 --- a/tools/vgconvert.c +++ b/tools/vgconvert.c @@ -97,7 +97,7 @@ static int vgconvert_single(struct cmd_context *cmd, const char *vg_name, if (cmd->fmt->features & FMT_RESTRICTED_LVIDS) { dm_list_iterate_items(lvl, &vg->lvs) { lv = lvl->lv; - if (lv->status & SNAPSHOT) + if (lv_is_snapshot(lv)) continue; if (lvnum_from_lvid(&lv->lvid) < MAX_RESTRICTED_LVS) continue; diff --git a/tools/vgsplit.c b/tools/vgsplit.c index 52f603f..e2ba6f7 100644 --- a/tools/vgsplit.c +++ b/tools/vgsplit.c @@ -112,7 +112,7 @@ static int _move_lvs(struct volume_group *vg_from, struct volume_group *vg_to) dm_list_iterate_safe(lvh, lvht, &vg_from->lvs) { lv = dm_list_item(lvh, struct lv_list)->lv; - if ((lv->status & SNAPSHOT)) + if (lv_is_snapshot(lv)) continue; if (lv_is_raid(lv)) @@ -193,7 +193,7 @@ static int _move_snapshots(struct volume_group *vg_from, dm_list_iterate_safe(lvh, lvht, &vg_from->lvs) { lv = dm_list_item(lvh, struct lv_list)->lv; - if (!(lv->status & SNAPSHOT)) + if (!lv_is_snapshot(lv)) continue; dm_list_iterate_items(seg, &lv->segments) {

1 0

master - cleanup: allocate NAME_LEN size for lv name
by Zdenek Kabelac 14 Dec '16

14 Dec '16

Gitweb: http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=47b96c3537968caf… Commit: 47b96c3537968caf0cfb8ffa3874d64944715bec Parent: d0fe3ec0c58ccd5dfcc4f5fd40ddabab0c9defe5 Author: Zdenek Kabelac <zkabelac(a)redhat.com> AuthorDate: Tue Dec 13 00:09:38 2016 +0100 Committer: Zdenek Kabelac <zkabelac(a)redhat.com> CommitterDate: Tue Dec 13 22:07:52 2016 +0100 cleanup: allocate NAME_LEN size for lv name --- lib/metadata/raid_manip.c | 18 ++++++++++-------- 1 files changed, 10 insertions(+), 8 deletions(-) diff --git a/lib/metadata/raid_manip.c b/lib/metadata/raid_manip.c index c70ce3e..b8a47c0 100644 --- a/lib/metadata/raid_manip.c +++ b/lib/metadata/raid_manip.c @@ -548,18 +548,15 @@ static char *_generate_raid_name(struct logical_volume *lv, const char *suffix, int count) { const char *format = (count >= 0) ? "%s_%s_%u" : "%s_%s"; - size_t len = strlen(lv->name) + strlen(suffix) + ((count >= 0) ? 5 : 2); - char *name; + char name[NAME_LEN], *lvname; int historical; - if (!(name = dm_pool_alloc(lv->vg->vgmem, len))) { - log_error("Failed to allocate new name."); + if (dm_snprintf(name, sizeof(name), format, lv->name, suffix, count) < 0) { + log_error("Failed to new raid name for %s.", + display_lvname(lv)); return NULL; } - if (dm_snprintf(name, len, format, lv->name, suffix, count) < 0) - return_NULL; - if (!validate_name(name)) { log_error("New logical volume name \"%s\" is not valid.", name); return NULL; @@ -571,7 +568,12 @@ static char *_generate_raid_name(struct logical_volume *lv, return NULL; } - return name; + if (!(lvname = dm_pool_strdup(lv->vg->vgmem, name))) { + log_error("Failed to allocate new name."); + return NULL; + } + + return lvname; } /* * Create an LV of specified type. Set visible after creation.

1 0

master - raid: avoid manipulation of segment status
by Zdenek Kabelac 14 Dec '16

14 Dec '16

Gitweb: http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=d0fe3ec0c58ccd5d… Commit: d0fe3ec0c58ccd5dfcc4f5fd40ddabab0c9defe5 Parent: d1e398c474032716679f704008026b5d17e9469a Author: Zdenek Kabelac <zkabelac(a)redhat.com> AuthorDate: Tue Dec 13 14:52:06 2016 +0100 Committer: Zdenek Kabelac <zkabelac(a)redhat.com> CommitterDate: Tue Dec 13 22:07:52 2016 +0100 raid: avoid manipulation of segment status RAID is LV property TODO: only 2 flags are seg->status: PVMOVE & MERGING At least the second one should be soon elimanted as again we merge LV not a segment. --- WHATS_NEW | 1 + lib/metadata/raid_manip.c | 6 ++---- lib/raid/raid.c | 1 - 3 files changed, 3 insertions(+), 5 deletions(-) diff --git a/WHATS_NEW b/WHATS_NEW index 0561074..60e8fa5 100644 --- a/WHATS_NEW +++ b/WHATS_NEW @@ -1,5 +1,6 @@ Version 2.02.169 - ===================================== + Do not use LV RAID status bit for segment status. Check segtype directly instead of checking RAID in segment status. Reusing exiting code for raid image removal. Fix pvmove leaving -pvmove0 error device in clustered VG. diff --git a/lib/metadata/raid_manip.c b/lib/metadata/raid_manip.c index f2c0b26..c70ce3e 100644 --- a/lib/metadata/raid_manip.c +++ b/lib/metadata/raid_manip.c @@ -1701,7 +1701,7 @@ static int _alloc_and_add_new_striped_segment(struct logical_volume *lv, /* Allocate a segment with seg->area_count areas */ if (!(new_seg = alloc_lv_segment(striped_segtype, lv, le, area_len * seg->area_count, - seg->status & ~RAID, + 0, seg->stripe_size, NULL, seg->area_count, area_len, seg->chunk_size, 0, 0, NULL))) return_0; @@ -2150,7 +2150,6 @@ static int _convert_mirror_to_raid1(struct logical_volume *lv, lv->status &= ~MIRROR; lv->status &= ~MIRRORED; lv->status |= RAID; - seg->status |= RAID; if (!lv_update_and_reload(lv)) return_0; @@ -2220,7 +2219,6 @@ static int _convert_raid1_to_mirror(struct logical_volume *lv, seg->segtype = new_segtype; seg->region_size = new_region_size; lv->status &= ~RAID; - seg->status &= ~RAID; lv->status |= (MIRROR | MIRRORED); if (!attach_mirror_log(first_seg(lv), log_lv)) @@ -2473,7 +2471,7 @@ static struct lv_segment *_convert_striped_to_raid0(struct logical_volume *lv, seg = first_seg(dm_list_item(dm_list_first(&data_lvs), struct lv_list)->lv); if (!(raid0_seg = alloc_lv_segment(segtype, lv, 0 /* le */, lv->le_count /* len */, - seg->status | SEG_RAID, + 0, stripe_size, NULL /* log_lv */, area_count, area_len, 0 /* chunk_size */, diff --git a/lib/raid/raid.c b/lib/raid/raid.c index 21fad92..1e562ff 100644 --- a/lib/raid/raid.c +++ b/lib/raid/raid.c @@ -167,7 +167,6 @@ static int _raid_text_import(struct lv_segment *seg, if (seg_is_any_raid0(seg)) seg->area_len /= seg->area_count; - seg->status |= RAID; return 1; }

1 0

master - segtype: check for seg type instead of status
by Zdenek Kabelac 14 Dec '16

14 Dec '16

Gitweb: http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=d1e398c474032716… Commit: d1e398c474032716679f704008026b5d17e9469a Parent: 069039204002e5c8514050fe541bbd378c383a02 Author: Zdenek Kabelac <zkabelac(a)redhat.com> AuthorDate: Tue Dec 13 14:50:48 2016 +0100 Committer: Zdenek Kabelac <zkabelac(a)redhat.com> CommitterDate: Tue Dec 13 22:07:52 2016 +0100 segtype: check for seg type instead of status RAID is LV property - which has single segment of raid type. --- WHATS_NEW | 1 + lib/format_text/export.c | 2 +- lib/metadata/merge.c | 8 ++++---- 3 files changed, 6 insertions(+), 5 deletions(-) diff --git a/WHATS_NEW b/WHATS_NEW index a75183e..0561074 100644 --- a/WHATS_NEW +++ b/WHATS_NEW @@ -1,5 +1,6 @@ Version 2.02.169 - ===================================== + Check segtype directly instead of checking RAID in segment status. Reusing exiting code for raid image removal. Fix pvmove leaving -pvmove0 error device in clustered VG. Avoid adding extra '_' at end of raid extracted images or metadata. diff --git a/lib/format_text/export.c b/lib/format_text/export.c index 1ecc362..199c185 100644 --- a/lib/format_text/export.c +++ b/lib/format_text/export.c @@ -623,7 +623,7 @@ int out_areas(struct formatter *f, const struct lv_segment *seg, break; case AREA_LV: /* FIXME This helper code should be target-independent! Check for metadata LV property. */ - if (!(seg->status & RAID)) { + if (!seg_is_raid(seg)) { outf(f, "\"%s\", %u%s", seg_lv(seg, s)->name, seg_le(seg, s), diff --git a/lib/metadata/merge.c b/lib/metadata/merge.c index bf0d2c0..6d52bc4 100644 --- a/lib/metadata/merge.c +++ b/lib/metadata/merge.c @@ -431,7 +431,7 @@ int check_lv_segments(struct logical_volume *lv, int complete_vg) } if (complete_vg && seg->log_lv && - !seg_is_mirrored(seg) && !(seg->status & RAID_IMAGE)) { + !seg_is_mirrored(seg) && lv_is_raid_image(lv)) { log_error("LV %s: segment %u log LV %s is not a " "mirror log or a RAID image", lv->name, seg_count, seg->log_lv->name); @@ -458,9 +458,9 @@ int check_lv_segments(struct logical_volume *lv, int complete_vg) } } - if (complete_vg && seg->status & MIRROR_IMAGE) { - if (!find_mirror_seg(seg) || - !seg_is_mirrored(find_mirror_seg(seg))) { + if (complete_vg && lv_is_mirror_image(lv)) { + if (!(seg2 = find_mirror_seg(seg)) || + !seg_is_mirrored(seg2)) { log_error("LV %s: segment %u mirror image " "is not mirrored", lv->name, seg_count);

1 0

master - raid: improve table reload sequence
by Zdenek Kabelac 14 Dec '16

14 Dec '16

Gitweb: http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=069039204002e5c8… Commit: 069039204002e5c8514050fe541bbd378c383a02 Parent: 7dff632c11e6ddce77dc9cde8f9011bb22970589 Author: Zdenek Kabelac <zkabelac(a)redhat.com> AuthorDate: Tue Dec 13 12:31:28 2016 +0100 Committer: Zdenek Kabelac <zkabelac(a)redhat.com> CommitterDate: Tue Dec 13 22:07:52 2016 +0100 raid: improve table reload sequence This is another place for 'common' use pattern or reload and activation of deleted devices. (Moving the exclusive activation to _deactivate_and_remove_lvs()). TODO: looks like halve of raid function is reloading just 'origin' - and the other full LV. --- WHATS_NEW | 1 + lib/metadata/raid_manip.c | 60 ++++++-------------------------------------- 2 files changed, 10 insertions(+), 51 deletions(-) diff --git a/WHATS_NEW b/WHATS_NEW index ea15903..a75183e 100644 --- a/WHATS_NEW +++ b/WHATS_NEW @@ -1,5 +1,6 @@ Version 2.02.169 - ===================================== + Reusing exiting code for raid image removal. Fix pvmove leaving -pvmove0 error device in clustered VG. Avoid adding extra '_' at end of raid extracted images or metadata. Optimize another _rmeta clearing code. diff --git a/lib/metadata/raid_manip.c b/lib/metadata/raid_manip.c index 98f30e2..f2c0b26 100644 --- a/lib/metadata/raid_manip.c +++ b/lib/metadata/raid_manip.c @@ -248,11 +248,10 @@ static int _deactivate_and_remove_lvs(struct volume_group *vg, struct dm_list *r { struct lv_list *lvl; - if (vg_is_clustered(vg)) - /* Need to take lock for proper deactivation */ - dm_list_iterate_items(lvl, removal_lvs) - if (!activate_lv_excl_local(vg->cmd, lvl->lv)) - return_0; + /* Need to take lock/resume for proper deactivation */ + dm_list_iterate_items(lvl, removal_lvs) + if (!activate_lv_excl_local(vg->cmd, lvl->lv)) + return_0; dm_list_iterate_items(lvl, removal_lvs) { if (!deactivate_lv(vg->cmd, lvl->lv)) @@ -1214,15 +1213,14 @@ static int _raid_remove_images(struct logical_volume *lv, struct dm_list *removal_lvs, int commit) { struct dm_list removed_lvs; - struct lv_list *lvl; - - dm_list_init(&removed_lvs); if (!archive(lv->vg)) return_0; - if (!removal_lvs) + if (!removal_lvs) { + dm_list_init(&removed_lvs); removal_lvs = &removed_lvs; + } if (!_raid_extract_images(lv, 0, new_count, allocate_pvs, 1, removal_lvs, removal_lvs)) { @@ -1245,48 +1243,8 @@ static int _raid_remove_images(struct logical_volume *lv, if (!commit) return 1; - if (!vg_write(lv->vg)) { - log_error("Failed to write changes for %s.", - display_lvname(lv)); - return 0; - } - - if (!suspend_lv(lv->vg->cmd, lv)) { - log_error("Failed to suspend %s before committing changes.", - display_lvname(lv)); - vg_revert(lv->vg); - return 0; - } - - if (!vg_commit(lv->vg)) { - log_error("Failed to commit changes for %s.", - display_lvname(lv)); - return 0; - } - - /* - * We activate the extracted sub-LVs first so they are renamed - * and won't conflict with the remaining (possibly shifted) - * sub-LVs. - */ - dm_list_iterate_items(lvl, removal_lvs) { - if (!activate_lv_excl_local(lv->vg->cmd, lvl->lv)) { - log_error("Failed to resume extracted LVs."); - return 0; - } - } - - if (!resume_lv(lv->vg->cmd, lv)) { - log_error("Failed to resume %s after committing changes.", - display_lvname(lv)); - return 0; - } - - if (!sync_local_dev_names(lv->vg->cmd)) { - log_error("Failed to sync local devices after committing changes for %s.", - display_lvname(lv)); - return 0; - } + if (!lv_update_and_reload(lv)) + return_0; /* * Eliminate the extracted LVs

1 0

master - libdm: add min_num_bits to dm_bitset_parse_list()
by Bryn Reeves 14 Dec '16

14 Dec '16

Gitweb: http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=7dff632c11e6ddce… Commit: 7dff632c11e6ddce77dc9cde8f9011bb22970589 Parent: e8d966bc31741f846dc6d0d4af9a565d19560b98 Author: Bryn M. Reeves <bmr(a)redhat.com> AuthorDate: Sun Dec 11 22:41:45 2016 +0000 Committer: Bryn M. Reeves <bmr(a)redhat.com> CommitterDate: Tue Dec 13 21:02:18 2016 +0000 libdm: add min_num_bits to dm_bitset_parse_list() It's useful to be able to specify a minimum number of bits for a new bitmap parsed from a list, for e.g. to allow for expansing a group without needing to copy/reallocate the bitmap. Add a backwards compatible symbol for programs linked against old versions of the library. --- libdm/.exported_symbols.DM_1_02_138 | 1 + libdm/datastruct/bitset.c | 22 +++++++++++++++++++++- libdm/libdevmapper.h | 3 ++- libdm/libdm-stats.c | 5 +++-- 4 files changed, 27 insertions(+), 4 deletions(-) diff --git a/libdm/.exported_symbols.DM_1_02_138 b/libdm/.exported_symbols.DM_1_02_138 index 63fc2b8..7535829 100644 --- a/libdm/.exported_symbols.DM_1_02_138 +++ b/libdm/.exported_symbols.DM_1_02_138 @@ -1,2 +1,3 @@ dm_bit_get_last dm_bit_get_prev +dm_bitset_parse_list diff --git a/libdm/datastruct/bitset.c b/libdm/datastruct/bitset.c index 90efec6..b0826e1 100644 --- a/libdm/datastruct/bitset.c +++ b/libdm/datastruct/bitset.c @@ -145,7 +145,8 @@ int dm_bit_get_last(dm_bitset_t bs) /* * Based on the Linux kernel __bitmap_parselist from lib/bitmap.c */ -dm_bitset_t dm_bitset_parse_list(const char *str, struct dm_pool *mem) +dm_bitset_t dm_bitset_parse_list(const char *str, struct dm_pool *mem, + size_t min_num_bits) { unsigned a, b; int c, old_c, totaldigits, ndigits, nmaskbits; @@ -221,6 +222,9 @@ scan: } while (len && c == ','); if (!mask) { + if (min_num_bits && (nmaskbits < min_num_bits)) + nmaskbits = min_num_bits; + if (!(mask = dm_bitset_create(mem, nmaskbits))) goto_bad; str = start; @@ -237,3 +241,19 @@ bad: } return NULL; } + +#if defined(__GNUC__) +/* + * Maintain backward compatibility with older versions that did not + * accept a 'min_num_bits' argument to dm_bitset_parse_list(). + */ +dm_bitset_t dm_bitset_parse_list_v1_02_129(const char *str, struct dm_pool *mem); +dm_bitset_t dm_bitset_parse_list_v1_02_129(const char *str, struct dm_pool *mem) +{ + return dm_bitset_parse_list(str, mem, 0); +} +DM_EXPORT_SYMBOL(dm_bitset_parse_list, 1_02_129); + +#else /* if defined(__GNUC__) */ + +#endif diff --git a/libdm/libdevmapper.h b/libdm/libdevmapper.h index 44aa55c..bcf784b 100644 --- a/libdm/libdevmapper.h +++ b/libdm/libdevmapper.h @@ -2100,7 +2100,8 @@ int dm_bit_get_prev(dm_bitset_t bs, int last_bit); * dm_malloc(). Otherwise the bitset will be allocated using the supplied * dm_pool. */ -dm_bitset_t dm_bitset_parse_list(const char *str, struct dm_pool *mem); +dm_bitset_t dm_bitset_parse_list(const char *str, struct dm_pool *mem, + size_t min_num_bits); /* Returns number of set bits */ static inline unsigned hweight32(uint32_t i) diff --git a/libdm/libdm-stats.c b/libdm/libdm-stats.c index 8210a93..02a30dc 100644 --- a/libdm/libdm-stats.c +++ b/libdm/libdm-stats.c @@ -675,6 +675,7 @@ static void _check_group_regions_present(struct dm_stats *dms, #define DMS_GROUP_TAG_LEN (sizeof(DMS_GROUP_TAG) - 1) #define DMS_GROUP_SEP ':' #define DMS_AUX_SEP "#" + static int _parse_aux_data_group(struct dm_stats *dms, struct dm_stats_region *region, struct dm_stats_group *group) @@ -718,7 +719,7 @@ static int _parse_aux_data_group(struct dm_stats *dms, end = c + strlen(c); *(end++) = '\0'; - if (!(regions = dm_bitset_parse_list(c, NULL))) { + if (!(regions = dm_bitset_parse_list(c, NULL, 0))) { log_error("Could not parse member list while " "reading group aux_data"); return 0; @@ -3992,7 +3993,7 @@ int dm_stats_create_group(struct dm_stats *dms, const char *members, return 0; }; - if (!(regions = dm_bitset_parse_list(members, NULL))) { + if (!(regions = dm_bitset_parse_list(members, NULL, 0))) { log_error("Could not parse list: '%s'", members); return 0; }

1 0

master - libdm: use dm_bit_get_last() in _stats_group_tag_fill()
by Bryn Reeves 14 Dec '16

14 Dec '16

Gitweb: http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=e8d966bc31741f84… Commit: e8d966bc31741f846dc6d0d4af9a565d19560b98 Parent: 5d1d65e735b6dcc5f02d0e536e3b63617f40ce83 Author: Bryn M. Reeves <bmr(a)redhat.com> AuthorDate: Tue Dec 13 10:32:29 2016 +0000 Committer: Bryn M. Reeves <bmr(a)redhat.com> CommitterDate: Tue Dec 13 21:02:18 2016 +0000 libdm: use dm_bit_get_last() in _stats_group_tag_fill() Instead of iterating over all bits, use dm_bit_get_last() to find the last set bit in the group bitmap. --- libdm/libdm-stats.c | 4 +--- 1 files changed, 1 insertions(+), 3 deletions(-) diff --git a/libdm/libdm-stats.c b/libdm/libdm-stats.c index 471bd7b..8210a93 100644 --- a/libdm/libdm-stats.c +++ b/libdm/libdm-stats.c @@ -1716,9 +1716,7 @@ static size_t _stats_group_tag_fill(const struct dm_stats *dms, int i, j, r, next, last = 0; size_t used = 0; - i = dm_bit_get_first(regions); - for (; i >= 0; i = dm_bit_get_next(regions, i)) - last = i; + last = dm_bit_get_last(regions); i = dm_bit_get_first(regions); for(; i >= 0; i = dm_bit_get_next(regions, i)) {

1 0

master - libdm: add dm_bit_get_last()/dm_bit_get_prev()
by Bryn Reeves 14 Dec '16

14 Dec '16

Gitweb: http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=5d1d65e735b6dcc5… Commit: 5d1d65e735b6dcc5f02d0e536e3b63617f40ce83 Parent: 107bc13db3b0117e32bacede53fbd382feaf4a17 Author: Bryn M. Reeves <bmr(a)redhat.com> AuthorDate: Sun Dec 11 12:44:45 2016 +0000 Committer: Bryn M. Reeves <bmr(a)redhat.com> CommitterDate: Tue Dec 13 21:01:58 2016 +0000 libdm: add dm_bit_get_last()/dm_bit_get_prev() It is sometimes convenient to iterate over the set bits in a dm bitset in reverse order (from the highest set bit toward zero), or to quickly find the last set bit. Add dm_bit_get_last() and dm_bit_get_prev(), mirroring the existing dm_bit_get_first() and dm_bit_get_next(). dm_bit_get_prev() uses __builtin_clz when available to efficiently test the bitset in reverse. --- libdm/.exported_symbols.DM_1_02_138 | 2 + libdm/datastruct/bitset.c | 36 +++++++++++++++++++++++++++++++++++ libdm/libdevmapper.h | 2 + 3 files changed, 40 insertions(+), 0 deletions(-) diff --git a/libdm/.exported_symbols.DM_1_02_138 b/libdm/.exported_symbols.DM_1_02_138 new file mode 100644 index 0000000..63fc2b8 --- /dev/null +++ b/libdm/.exported_symbols.DM_1_02_138 @@ -0,0 +1,2 @@ +dm_bit_get_last +dm_bit_get_prev diff --git a/libdm/datastruct/bitset.c b/libdm/datastruct/bitset.c index 3fd6a9c..90efec6 100644 --- a/libdm/datastruct/bitset.c +++ b/libdm/datastruct/bitset.c @@ -76,6 +76,13 @@ static int _test_word(uint32_t test, int bit) return (tb ? ffs(tb) + bit - 1 : -1); } +static int _test_word_rev(uint32_t test, int bit) +{ + uint32_t tb = test << (DM_BITS_PER_INT - 1 - bit); + + return (tb ? bit - clz(tb) : -1); +} + int dm_bit_get_next(dm_bitset_t bs, int last_bit) { int bit, word; @@ -101,11 +108,40 @@ int dm_bit_get_next(dm_bitset_t bs, int last_bit) return -1; } +int dm_bit_get_prev(dm_bitset_t bs, int last_bit) +{ + int bit, word; + uint32_t test; + + last_bit--; /* otherwise we'll return the same bit again */ + + /* + * bs[0] holds number of bits + */ + while (last_bit >= 0) { + word = last_bit >> INT_SHIFT; + test = bs[word + 1]; + bit = last_bit & (DM_BITS_PER_INT - 1); + + if ((bit = _test_word_rev(test, bit)) >= 0) + return (word * DM_BITS_PER_INT) + bit; + + last_bit = (last_bit & ~(DM_BITS_PER_INT - 1)) - 1; + } + + return -1; +} + int dm_bit_get_first(dm_bitset_t bs) { return dm_bit_get_next(bs, -1); } +int dm_bit_get_last(dm_bitset_t bs) +{ + return dm_bit_get_prev(bs, bs[0] + 1); +} + /* * Based on the Linux kernel __bitmap_parselist from lib/bitmap.c */ diff --git a/libdm/libdevmapper.h b/libdm/libdevmapper.h index ab69a56..44aa55c 100644 --- a/libdm/libdevmapper.h +++ b/libdm/libdevmapper.h @@ -2069,6 +2069,8 @@ void dm_bit_and(dm_bitset_t out, dm_bitset_t in1, dm_bitset_t in2); void dm_bit_union(dm_bitset_t out, dm_bitset_t in1, dm_bitset_t in2); int dm_bit_get_first(dm_bitset_t bs); int dm_bit_get_next(dm_bitset_t bs, int last_bit); +int dm_bit_get_last(dm_bitset_t bs); +int dm_bit_get_prev(dm_bitset_t bs, int last_bit); #define DM_BITS_PER_INT (sizeof(int) * CHAR_BIT)

1 0

master - util: add clz() and use __builtin_clz() if available
by Bryn Reeves 14 Dec '16

14 Dec '16

Gitweb: http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=107bc13db3b0117e… Commit: 107bc13db3b0117e32bacede53fbd382feaf4a17 Parent: 83a9cd5258fbd1acf293734cb4c3aba0f9ffcc9b Author: Bryn M. Reeves <bmr(a)redhat.com> AuthorDate: Mon Dec 12 12:03:50 2016 +0000 Committer: Bryn M. Reeves <bmr(a)redhat.com> CommitterDate: Tue Dec 13 20:41:29 2016 +0000 util: add clz() and use __builtin_clz() if available Add a macro for the clz (count leading zeros) operation. Use the GCC __builtin_clz() for clz() if it is available and fall back to a shift based implementation on systems that do not set HAVE___BUILTIN_CLZ. --- lib/misc/util.h | 41 +++++++++++++++++++++++++++++++++++++++++ 1 files changed, 41 insertions(+), 0 deletions(-) diff --git a/lib/misc/util.h b/lib/misc/util.h index 730b903..9231ea9 100644 --- a/lib/misc/util.h +++ b/lib/misc/util.h @@ -35,6 +35,47 @@ #define uninitialized_var(x) x = x #endif +/* + * GCC 3.4 adds a __builtin_clz, which uses the count leading zeros (clz) + * instruction on arches that have one. Provide a fallback using shifts + * and comparisons for older compilers. + */ +#ifdef HAVE___BUILTIN_CLZ +#define clz(x) __builtin_clz((x)) +#else /* ifdef HAVE___BUILTIN_CLZ */ +unsigned _dm_clz(unsigned x) +{ + int n; + + if ((int)x <= 0) return (~x >> 26) & 32; + + n = 1; + + if ((x >> 16) == 0) { + n = n + 16; + x = x << 16; + } + + if ((x >> 24) == 0) { + n = n + 8; + x = x << 8; + } + + if ((x >> 28) == 0) { + n = n + 4; + x = x << 4; + } + + if ((x >> 30) == 0) { + n = n + 2; + x = x << 2; + } + n = n - (x >> 31); + return n; +} +#define clz(x) _dm_clz((x)) +#endif /* ifdef HAVE___BUILTIN_CLZ */ + #define KERNEL_VERSION(major, minor, release) (((major) << 16) + ((minor) << 8) + (release)) /* Define some portable printing types */

1 0

master - configure: check for __builtin_clz()
by Bryn Reeves 14 Dec '16

14 Dec '16

Gitweb: http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=83a9cd5258fbd1ac… Commit: 83a9cd5258fbd1acf293734cb4c3aba0f9ffcc9b Parent: 930b0b4c9e4ca466d955f8729644ef26a939937c Author: Bryn M. Reeves <bmr(a)redhat.com> AuthorDate: Mon Dec 12 12:03:16 2016 +0000 Committer: Bryn M. Reeves <bmr(a)redhat.com> CommitterDate: Tue Dec 13 20:38:40 2016 +0000 configure: check for __builtin_clz() --- aclocal.m4 | 171 ++++++++++++++++++++++++++++++++++++++++++++++++ configure | 44 ++++++++++++ configure.in | 1 + include/configure.h.in | 3 + 4 files changed, 219 insertions(+), 0 deletions(-) diff --git a/aclocal.m4 b/aclocal.m4 index 35107ac..1f02c2b 100644 --- a/aclocal.m4 +++ b/aclocal.m4 @@ -536,4 +536,175 @@ AC_DEFUN([AM_RUN_LOG], echo "$as_me:$LINENO: \$? = $ac_status" >&AS_MESSAGE_LOG_FD (exit $ac_status); }]) +# =========================================================================== +# http://www.gnu.org/software/autoconf-archive/ax_gcc_builtin.html +# =========================================================================== +# +# SYNOPSIS +# +# AX_GCC_BUILTIN(BUILTIN) +# +# DESCRIPTION +# +# This macro checks if the compiler supports one of GCC's built-in +# functions; many other compilers also provide those same built-ins. +# +# The BUILTIN parameter is the name of the built-in function. +# +# If BUILTIN is supported define HAVE_<BUILTIN>. Keep in mind that since +# builtins usually start with two underscores they will be copied over +# into the HAVE_<BUILTIN> definition (e.g. HAVE___BUILTIN_EXPECT for +# __builtin_expect()). +# +# The macro caches its result in the ax_cv_have_<BUILTIN> variable (e.g. +# ax_cv_have___builtin_expect). +# +# The macro currently supports the following built-in functions: +# +# __builtin_assume_aligned +# __builtin_bswap16 +# __builtin_bswap32 +# __builtin_bswap64 +# __builtin_choose_expr +# __builtin___clear_cache +# __builtin_clrsb +# __builtin_clrsbl +# __builtin_clrsbll +# __builtin_clz +# __builtin_clzl +# __builtin_clzll +# __builtin_complex +# __builtin_constant_p +# __builtin_ctz +# __builtin_ctzl +# __builtin_ctzll +# __builtin_expect +# __builtin_ffs +# __builtin_ffsl +# __builtin_ffsll +# __builtin_fpclassify +# __builtin_huge_val +# __builtin_huge_valf +# __builtin_huge_vall +# __builtin_inf +# __builtin_infd128 +# __builtin_infd32 +# __builtin_infd64 +# __builtin_inff +# __builtin_infl +# __builtin_isinf_sign +# __builtin_nan +# __builtin_nand128 +# __builtin_nand32 +# __builtin_nand64 +# __builtin_nanf +# __builtin_nanl +# __builtin_nans +# __builtin_nansf +# __builtin_nansl +# __builtin_object_size +# __builtin_parity +# __builtin_parityl +# __builtin_parityll +# __builtin_popcount +# __builtin_popcountl +# __builtin_popcountll +# __builtin_powi +# __builtin_powif +# __builtin_powil +# __builtin_prefetch +# __builtin_trap +# __builtin_types_compatible_p +# __builtin_unreachable +# +# Unsuppored built-ins will be tested with an empty parameter set and the +# result of the check might be wrong or meaningless so use with care. +# +# LICENSE +# +# Copyright (c) 2013 Gabriele Svelto <gabriele.svelto(a)gmail.com> +# +# Copying and distribution of this file, with or without modification, are +# permitted in any medium without royalty provided the copyright notice +# and this notice are preserved. This file is offered as-is, without any +# warranty. + +#serial 3 + +AC_DEFUN([AX_GCC_BUILTIN], [ + AS_VAR_PUSHDEF([ac_var], [ax_cv_have_$1]) + + AC_CACHE_CHECK([for $1], [ac_var], [ + AC_LINK_IFELSE([AC_LANG_PROGRAM([], [ + m4_case([$1], + [__builtin_assume_aligned], [$1("", 0)], + [__builtin_bswap16], [$1(0)], + [__builtin_bswap32], [$1(0)], + [__builtin_bswap64], [$1(0)], + [__builtin_choose_expr], [$1(0, 0, 0)], + [__builtin___clear_cache], [$1("", "")], + [__builtin_clrsb], [$1(0)], + [__builtin_clrsbl], [$1(0)], + [__builtin_clrsbll], [$1(0)], + [__builtin_clz], [$1(0)], + [__builtin_clzl], [$1(0)], + [__builtin_clzll], [$1(0)], + [__builtin_complex], [$1(0.0, 0.0)], + [__builtin_constant_p], [$1(0)], + [__builtin_ctz], [$1(0)], + [__builtin_ctzl], [$1(0)], + [__builtin_ctzll], [$1(0)], + [__builtin_expect], [$1(0, 0)], + [__builtin_ffs], [$1(0)], + [__builtin_ffsl], [$1(0)], + [__builtin_ffsll], [$1(0)], + [__builtin_fpclassify], [$1(0, 1, 2, 3, 4, 0.0)], + [__builtin_huge_val], [$1()], + [__builtin_huge_valf], [$1()], + [__builtin_huge_vall], [$1()], + [__builtin_inf], [$1()], + [__builtin_infd128], [$1()], + [__builtin_infd32], [$1()], + [__builtin_infd64], [$1()], + [__builtin_inff], [$1()], + [__builtin_infl], [$1()], + [__builtin_isinf_sign], [$1(0.0)], + [__builtin_nan], [$1("")], + [__builtin_nand128], [$1("")], + [__builtin_nand32], [$1("")], + [__builtin_nand64], [$1("")], + [__builtin_nanf], [$1("")], + [__builtin_nanl], [$1("")], + [__builtin_nans], [$1("")], + [__builtin_nansf], [$1("")], + [__builtin_nansl], [$1("")], + [__builtin_object_size], [$1("", 0)], + [__builtin_parity], [$1(0)], + [__builtin_parityl], [$1(0)], + [__builtin_parityll], [$1(0)], + [__builtin_popcount], [$1(0)], + [__builtin_popcountl], [$1(0)], + [__builtin_popcountll], [$1(0)], + [__builtin_powi], [$1(0, 0)], + [__builtin_powif], [$1(0, 0)], + [__builtin_powil], [$1(0, 0)], + [__builtin_prefetch], [$1("")], + [__builtin_trap], [$1()], + [__builtin_types_compatible_p], [$1(int, int)], + [__builtin_unreachable], [$1()], + [m4_warn([syntax], [Unsupported built-in $1, the test may fail]) + $1()] + ) + ])], + [AS_VAR_SET([ac_var], [yes])], + [AS_VAR_SET([ac_var], [no])]) + ]) + + AS_IF([test yes = AS_VAR_GET([ac_var])], + [AC_DEFINE_UNQUOTED(AS_TR_CPP(HAVE_$1), 1, + [Define to 1 if the system has the `$1' built-in function])], []) + + AS_VAR_POPDEF([ac_var]) +]) + m4_include([acinclude.m4]) diff --git a/configure b/configure index c82a19b..1cb4b54 100755 --- a/configure +++ b/configure @@ -6321,6 +6321,50 @@ _ACEOF esac + + + { $as_echo "$as_me:${as_lineno-$LINENO}: checking for __builtin_clz" >&5 +$as_echo_n "checking for __builtin_clz... " >&6; } +if ${ax_cv_have___builtin_clz+:} false; then : + $as_echo_n "(cached) " >&6 +else + + cat confdefs.h - <<_ACEOF >conftest.$ac_ext +/* end confdefs.h. */ + +int +main () +{ + + __builtin_clz(0) + + ; + return 0; +} +_ACEOF +if ac_fn_c_try_link "$LINENO"; then : + ax_cv_have___builtin_clz=yes +else + ax_cv_have___builtin_clz=no +fi +rm -f core conftest.err conftest.$ac_objext \ + conftest$ac_exeext conftest.$ac_ext + +fi +{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ax_cv_have___builtin_clz" >&5 +$as_echo "$ax_cv_have___builtin_clz" >&6; } + + if test yes = $ax_cv_have___builtin_clz; then : + +cat >>confdefs.h <<_ACEOF +#define HAVE___BUILTIN_CLZ 1 +_ACEOF + +fi + + + + ################################################################################ for ac_func in ftruncate gethostname getpagesize gettimeofday localtime_r \ memchr memset mkdir mkfifo munmap nl_langinfo realpath rmdir setenv \ diff --git a/configure.in b/configure.in index cc77aab..8ec65c8 100644 --- a/configure.in +++ b/configure.in @@ -134,6 +134,7 @@ AC_TYPE_UINT8_T AC_TYPE_UINT16_T AC_TYPE_UINT32_T AC_TYPE_UINT64_T +AX_GCC_BUILTIN([__builtin_clz]) ################################################################################ dnl -- Check for functions diff --git a/include/configure.h.in b/include/configure.h.in index 3c43bd5..cf8dc94 100644 --- a/include/configure.h.in +++ b/include/configure.h.in @@ -341,6 +341,9 @@ /* Define to 1 if the system has the type `ptrdiff_t'. */ #undef HAVE_PTRDIFF_T +/* Define to 1 if the compiler has the `__builtin_clz` builtin. */ +#undef HAVE___BUILTIN_CLZ + /* Define to 1 if you have the <readline/history.h> header file. */ #undef HAVE_READLINE_HISTORY_H

1 0

master - libdm: fix start of file detection in _stats_map_extents()
by Bryn Reeves 14 Dec '16

14 Dec '16

Gitweb: http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=930b0b4c9e4ca466… Commit: 930b0b4c9e4ca466d955f8729644ef26a939937c Parent: eb6557221715b30a374be46e4a52f79cd7bbed44 Author: Bryn M. Reeves <bmr(a)redhat.com> AuthorDate: Tue Dec 13 18:14:13 2016 +0000 Committer: Bryn M. Reeves <bmr(a)redhat.com> CommitterDate: Tue Dec 13 20:25:47 2016 +0000 libdm: fix start of file detection in _stats_map_extents() --- libdm/libdm-stats.c | 2 +- 1 files changed, 1 insertions(+), 1 deletions(-) diff --git a/libdm/libdm-stats.c b/libdm/libdm-stats.c index d9f0427..471bd7b 100644 --- a/libdm/libdm-stats.c +++ b/libdm/libdm-stats.c @@ -4287,7 +4287,7 @@ static uint64_t _stats_map_extents(struct dm_pool *mem, * If the file only has a single extent, no boundary is ever * detected to trigger addition of the first extent. */ - if (fm_ext[i].fe_logical == 0) + if (fm_ext[i - 1].fe_logical == 0) _stats_add_extent(mem, fm_pending, nr_extents); fiemap->fm_start = (fm_ext[i - 1].fe_logical +

1 0

master - libdm: break up _stats_get_extents_for_file()
by Bryn Reeves 14 Dec '16

14 Dec '16

Gitweb: http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=eb6557221715b30a… Commit: eb6557221715b30a374be46e4a52f79cd7bbed44 Parent: ea9af3e290879403d99dd602153ff9d0ad0de8de Author: Bryn M. Reeves <bmr(a)redhat.com> AuthorDate: Tue Dec 13 12:03:00 2016 +0000 Committer: Bryn M. Reeves <bmr(a)redhat.com> CommitterDate: Tue Dec 13 20:25:45 2016 +0000 libdm: break up _stats_get_extents_for_file() Split out the loop that iterates over each batch of FIEMAP extent data from the function that sets up and calls the ioctl to reduce nesting and simplify local variable use: _stats_get_extents_for_file() -> _stats_map_extents() The _stats_map_extents() function is responsible for detecting eof and extent boundaries and adding whole, allocated extents to the file extent table for region creation. --- libdm/libdm-stats.c | 145 +++++++++++++++++++++++++++++++-------------------- 1 files changed, 88 insertions(+), 57 deletions(-) diff --git a/libdm/libdm-stats.c b/libdm/libdm-stats.c index 2d2a032..d9f0427 100644 --- a/libdm/libdm-stats.c +++ b/libdm/libdm-stats.c @@ -4226,6 +4226,82 @@ static int _stats_add_extent(struct dm_pool *mem, struct fiemap_extent *fm_ext, ((ext).fe_physical != (exp)) /* + * Copy fields from fiemap_extent 'from' to the fiemap_extent + * pointed to by 'to'. + */ +#define ext_copy(to, from) \ +do { \ + *(to) = *(from); \ +} while (0) + +static uint64_t _stats_map_extents(struct dm_pool *mem, + struct fiemap *fiemap, + struct fiemap_extent *fm_ext, + struct fiemap_extent *fm_last, + struct fiemap_extent *fm_pending, + uint64_t next_extent, + int *eof) +{ + uint64_t expected = 0, nr_extents = next_extent; + unsigned int i; + + /* + * Loop over the returned extents adding the fm_pending extent + * to the table of extents each time a discontinuity (or eof) + * is detected. + * + * We use a pointer to fm_pending in the caller since it is + * possible that logical extents comprising a single physical + * extent are returned by successive FIEMAP calls. + */ + for (i = 0; i < fiemap->fm_mapped_extents; i++) { + expected = fm_last->fe_physical + fm_last->fe_length; + + if (fm_ext[i].fe_flags & FIEMAP_EXTENT_LAST) + *eof = 1; + + /* cannot map extents that are not yet allocated. */ + if (fm_ext[i].fe_flags + & (FIEMAP_EXTENT_UNKNOWN | FIEMAP_EXTENT_DELALLOC)) + continue; + + if (ext_boundary(fm_ext[i], expected)) { + if (!_stats_add_extent(mem, fm_pending, nr_extents)) + goto_bad; + nr_extents++; + /* Begin a new pending extent. */ + ext_copy(fm_pending, fm_ext + i); + } else { + expected = 0; + /* Begin a new pending extent for extent 0. */ + if (fm_ext[i].fe_logical == 0) + ext_copy(fm_pending, fm_ext + i); + else + /* accumulate this logical extent's length */ + fm_pending->fe_length += fm_ext[i].fe_length; + } + *fm_last = fm_ext[i]; + } + + /* + * If the file only has a single extent, no boundary is ever + * detected to trigger addition of the first extent. + */ + if (fm_ext[i].fe_logical == 0) + _stats_add_extent(mem, fm_pending, nr_extents); + + fiemap->fm_start = (fm_ext[i - 1].fe_logical + + fm_ext[i - 1].fe_length); + + /* return the number of extents found in this call. */ + return nr_extents - next_extent; +bad: + /* signal mapping error to caller */ + *eof = -1; + return 0; +} + +/* * Read the extents of an open file descriptor into a table of struct _extent. * * Based on e2fsprogs/misc/filefrag.c::filefrag_fiemap(). @@ -4236,16 +4312,12 @@ static int _stats_add_extent(struct dm_pool *mem, struct fiemap_extent *fm_ext, static struct _extent *_stats_get_extents_for_file(struct dm_pool *mem, int fd, uint64_t *count) { - uint64_t *buf; + struct fiemap_extent fm_last = {0}, fm_pending = {0}, *fm_ext = NULL; struct fiemap *fiemap = NULL; - struct fiemap_extent *fm_ext = NULL; - struct fiemap_extent fm_last = {0}, fm_pending = {0}; + int eof = 0, nr_extents = 0; struct _extent *extents; - unsigned long long expected = 0; unsigned long flags = 0; - unsigned int i, num = 0; - int tot_extents = 0, n = 0; - int last = 0; + uint64_t *buf; int rc; buf = dm_zalloc(STATS_FIE_BUF_LEN); @@ -4284,67 +4356,26 @@ static struct _extent *_stats_get_extents_for_file(struct dm_pool *mem, int fd, goto bad; } - /* If 0 extents are returned, then more ioctls are not needed */ + /* If 0 extents are returned, more ioctls are not needed */ if (fiemap->fm_mapped_extents == 0) break; - for (i = 0; i < fiemap->fm_mapped_extents; i++) { - expected = fm_last.fe_physical + fm_last.fe_length; + nr_extents += _stats_map_extents(mem, fiemap, fm_ext, &fm_last, + &fm_pending, nr_extents, &eof); - if (fm_ext[i].fe_flags & FIEMAP_EXTENT_LAST) - last = 1; - - /* cannot map extents that are not yet allocated. */ - if (fm_ext[i].fe_flags - & (FIEMAP_EXTENT_UNKNOWN | FIEMAP_EXTENT_DELALLOC)) - continue; - - if (ext_boundary(fm_ext[i], expected)) { - tot_extents++; - if (!_stats_add_extent(mem, &fm_pending, - fm_pending.fe_flags)) - goto_bad; - /* Begin a new pending extent. */ - fm_pending.fe_flags = tot_extents - 1; - fm_pending.fe_physical = fm_ext[i].fe_physical; - fm_pending.fe_logical = fm_ext[i].fe_logical; - fm_pending.fe_length = fm_ext[i].fe_length; - } else { - expected = 0; - if (!tot_extents) - tot_extents = 1; - /* Begin a new pending extent for extent 0. */ - if (fm_ext[i].fe_logical == 0) { - fm_pending.fe_flags = tot_extents - 1; - fm_pending.fe_physical = fm_ext[i].fe_physical; - fm_pending.fe_logical = fm_ext[i].fe_logical; - fm_pending.fe_length = fm_ext[i].fe_length; - } else - fm_pending.fe_length += fm_ext[i].fe_length; - } - num += tot_extents; - fm_last = fm_ext[i]; - n++; - } - - /* - * If the file only has a single extent, no boundary is ever - * detected to trigger addition of the first extent. - */ - if (fm_ext[i].fe_logical == 0) - _stats_add_extent(mem, &fm_pending, fm_pending.fe_flags); + /* check for extent mapping error */ + if (eof < 0) + goto bad; - fiemap->fm_start = (fm_ext[i - 1].fe_logical + - fm_ext[i - 1].fe_length); - } while (last == 0); + } while (eof == 0); - if (!tot_extents) { + if (!nr_extents) { log_error("Cannot map file: no allocated extents."); goto bad; } /* return total number of extents */ - *count = tot_extents; + *count = nr_extents; extents = dm_pool_end_object(mem); /* free FIEMAP buffer. */

1 0

master - libdm: fix dm_stats_foreach_group() macro
by Bryn Reeves 14 Dec '16

14 Dec '16

Gitweb: http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=ea9af3e290879403… Commit: ea9af3e290879403d99dd602153ff9d0ad0de8de Parent: 8e339728283d3b83222a25c7bf12b79a944c6ef5 Author: Bryn M. Reeves <bmr(a)redhat.com> AuthorDate: Tue Dec 13 15:32:34 2016 +0000 Committer: Bryn M. Reeves <bmr(a)redhat.com> CommitterDate: Tue Dec 13 20:01:00 2016 +0000 libdm: fix dm_stats_foreach_group() macro --- libdm/libdevmapper.h | 6 +++--- 1 files changed, 3 insertions(+), 3 deletions(-) diff --git a/libdm/libdevmapper.h b/libdm/libdevmapper.h index fe2e410..ab69a56 100644 --- a/libdm/libdevmapper.h +++ b/libdm/libdevmapper.h @@ -1138,9 +1138,9 @@ for (dm_stats_walk_init((dms), DM_STATS_WALK_AREA), \ */ #define dm_stats_foreach_group(dms) \ for (dm_stats_walk_init((dms), DM_STATS_WALK_GROUP), \ - dm_stats_group_walk_start(dms); \ - !dm_stats_group_walk_end(dms); \ - dm_stats_group_walk_next(dms)) + dm_stats_walk_start(dms); \ + !dm_stats_walk_end(dms); \ + dm_stats_walk_next(dms)) /* * Start a walk iterating over the regions contained in dm_stats handle

1 0

master - libdm: check for non-existent region_id values in groups
by Bryn Reeves 14 Dec '16

14 Dec '16

Gitweb: http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=8e339728283d3b83… Commit: 8e339728283d3b83222a25c7bf12b79a944c6ef5 Parent: 99b6d82e2dac7ef097fa940c64456620930d7758 Author: Bryn M. Reeves <bmr(a)redhat.com> AuthorDate: Tue Dec 13 14:31:39 2016 +0000 Committer: Bryn M. Reeves <bmr(a)redhat.com> CommitterDate: Tue Dec 13 15:37:48 2016 +0000 libdm: check for non-existent region_id values in groups Check that all region_id values specified in a group bitmap are actually present: although this should not normally happen when using the dmstats tool, it is possible as a result of manual changes (or bugs) for a group descriptor to contain one or more group_id values that do not exist. Check for this situation when reading group descriptors, warn the user the user, and clear these bits in the bitmap when formatting it for output. --- libdm/libdm-stats.c | 27 +++++++++++++++++++++------ 1 files changed, 21 insertions(+), 6 deletions(-) diff --git a/libdm/libdm-stats.c b/libdm/libdm-stats.c index 5833ac6..2d2a032 100644 --- a/libdm/libdm-stats.c +++ b/libdm/libdm-stats.c @@ -645,6 +645,23 @@ static void _stats_update_groups(struct dm_stats *dms) } } +static void _check_group_regions_present(struct dm_stats *dms, + struct dm_stats_group *group) +{ + dm_bitset_t regions = group->regions; + int64_t i, group_id; + + group_id = i = dm_bit_get_first(regions); + + for (; i > 0; dm_bit_get_next(regions, i)) + if (!_stats_region_present(&dms->regions[i])) { + log_warn("Group descriptor " FMTi64 " contains " + "non-existent region_id " FMTi64 ".", + group_id, i); + dm_bit_clear(regions, i); + } +} + /* * Parse a DMS_GROUP group descriptor embedded in a region's aux_data. * @@ -1030,6 +1047,9 @@ static int _stats_parse_list(struct dm_stats *dms, const char *resp) dms->regions = dm_pool_end_object(mem); dms->groups = dm_pool_end_object(group_mem); + dm_stats_foreach_group(dms) + _check_group_regions_present(dms, &dms->groups[dms->cur_group]); + _stats_update_groups(dms); if (fclose(list_rows)) @@ -1742,17 +1762,12 @@ bad: static size_t _stats_group_tag_len(const struct dm_stats *dms, dm_bitset_t regions) { - int i, j, next, nr_regions = 0; + int64_t i, j, next, nr_regions = 0; size_t buflen = 0, id_len = 0; /* check region ids and find last set bit */ i = dm_bit_get_first(regions); for (; i >= 0; i = dm_bit_get_next(regions, i)) { - if (!dm_stats_region_present(dms, i)) { - log_error("Region identifier %d not found", i); - return 0; - } - /* length of region_id or range start in characters */ id_len = (i) ? 1 + (size_t) log10(i) : 1; buflen += id_len;

1 0

master - libdm: fix segfault with invalid group descriptor
by Bryn Reeves 14 Dec '16

14 Dec '16

Gitweb: http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=99b6d82e2dac7ef0… Commit: 99b6d82e2dac7ef097fa940c64456620930d7758 Parent: 138e4336fd8422623df7daa362b98a4899f5fbec Author: Bryn M. Reeves <bmr(a)redhat.com> AuthorDate: Tue Dec 13 13:56:10 2016 +0000 Committer: Bryn M. Reeves <bmr(a)redhat.com> CommitterDate: Tue Dec 13 14:37:41 2016 +0000 libdm: fix segfault with invalid group descriptor If a region has a a DMS_GROUP tag in aux_data where the first region_id in the bitmap is not the same as the containing region, dmstats will segfault: # '2' is never a valid group bitset list for region_id == 0 # dmsetup message vg_hex/root 0 "@stats_set_aux 0 DMS_GROUP=img:2#" # dmsetup message vg_hex/root 0 "@stats_list" 0: 45383680+16384 16384 dmstats DMS_GROUP=img:2# 1: 46071808+32768 32768 dmstats - 2: 47382528+16384 16384 dmstats - # dmstats list Segmentation fault (core dumped) The crash will occur in some arbitrary dm_stats_get_* property method - this happens while processing the 1st region_id in the bitset, because the region is marked as grouped, but there is no group bitmap present at dms->groups[2]->regions. Fix this by detecting a mismatch between the expected region_id and dm_bit_get_first() for the parsed bitset during _parse_aux_data_group(). --- libdm/libdm-stats.c | 8 +++++++- 1 files changed, 7 insertions(+), 1 deletions(-) diff --git a/libdm/libdm-stats.c b/libdm/libdm-stats.c index 8c0428e..5833ac6 100644 --- a/libdm/libdm-stats.c +++ b/libdm/libdm-stats.c @@ -708,8 +708,14 @@ static int _parse_aux_data_group(struct dm_stats *dms, } group->group_id = dm_bit_get_first(regions); - group->regions = regions; + if (group->group_id != region->region_id) { + log_error("Found invalid group descriptor in region " FMTu64 + " aux_data.", region->region_id); + group->group_id = DM_STATS_GROUP_NOT_PRESENT; + goto bad; + } + group->regions = regions; group->alias = NULL; if (strlen(alias)) { group->alias = dm_strdup(alias);

1 0

master - libdm: fix region overlap tests
by Bryn Reeves 13 Dec '16

13 Dec '16

Gitweb: http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=138e4336fd842262… Commit: 138e4336fd8422623df7daa362b98a4899f5fbec Parent: 93f420caf4e58f468919f40a9f3f9c20157affd0 Author: Bryn M. Reeves <bmr(a)redhat.com> AuthorDate: Mon Dec 12 22:49:31 2016 +0000 Committer: Bryn M. Reeves <bmr(a)redhat.com> CommitterDate: Tue Dec 13 09:09:29 2016 +0000 libdm: fix region overlap tests --- libdm/libdm-stats.c | 8 +++++++- 1 files changed, 7 insertions(+), 1 deletions(-) diff --git a/libdm/libdm-stats.c b/libdm/libdm-stats.c index 56a38f2..8c0428e 100644 --- a/libdm/libdm-stats.c +++ b/libdm/libdm-stats.c @@ -3880,6 +3880,12 @@ static int _stats_group_check_overlap(const struct dm_stats *dms, i++; } + /* A single region cannot overlap itself. */ + if (i == 1) { + dm_pool_free(dms->mem, map); + return 1; + } + /* sort by extent.start */ qsort(map, count, sizeof(*map), _extent_start_compare); @@ -3915,7 +3921,7 @@ merge: goto merge; dm_pool_free(dms->mem, map); - return overlap; + return (overlap == 0); } static void _stats_copy_histogram_bounds(struct dm_histogram *to,

1 0

master - libdm: fix _stats_get_extents_for_file()
by Bryn Reeves 13 Dec '16

13 Dec '16

Gitweb: http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=93f420caf4e58f46… Commit: 93f420caf4e58f468919f40a9f3f9c20157affd0 Parent: 87117c2b2546231c789f92c75590f053a8fb987c Author: Bryn M. Reeves <bmr(a)redhat.com> AuthorDate: Mon Dec 12 20:40:27 2016 +0000 Committer: Bryn M. Reeves <bmr(a)redhat.com> CommitterDate: Tue Dec 13 09:09:25 2016 +0000 libdm: fix _stats_get_extents_for_file() Handle files that contain multiple logical extents in a single physical extent properly: - In FIEMAP terms a logical extent is a contiguous range of sectors in the file's address space. - One or more physically adjacent logical extents comprise a physical extent: these are the disk areas that will be mapped to regions. - An extent boundary occurs when the start sector of extent n+1 is not equal to (n.start + n.length). This requires that we accumulate the length values of extents returned by FIEMAP until a discontinuity is found (since each struct fiemap_extent returned by FIEMAP only represents a single logical extent, which may be contiguous with other logical extents on-disk). This avoids creating large numbers of regions for physically adjacent (logical) extents and fixes the earlier behaviour which would only map the first logical extent of the physical extent, leaving gaps in the region table for these files. --- WHATS_NEW_DM | 3 +- libdm/libdm-stats.c | 56 +++++++++++++++++++++++++++++++++----------------- 2 files changed, 39 insertions(+), 20 deletions(-) diff --git a/WHATS_NEW_DM b/WHATS_NEW_DM index f402113..ab32bdb 100644 --- a/WHATS_NEW_DM +++ b/WHATS_NEW_DM @@ -1,8 +1,9 @@ Version 1.02.138 - ===================================== + Fix file mapping for extents with physically adjacent extents. Validation vsnprintf result in runtime translate of dm_log (1.02.136). Separate filemap extent allocation from region table. - Fix segmentation fault when filemap region creation fails + Fix segmentation fault when filemap region creation fails. Fix performance of region cleanup for failed filemap creation. Fix very slow region deletion with many regions. diff --git a/libdm/libdm-stats.c b/libdm/libdm-stats.c index 850aa49..56a38f2 100644 --- a/libdm/libdm-stats.c +++ b/libdm/libdm-stats.c @@ -4194,10 +4194,9 @@ static int _stats_add_extent(struct dm_pool *mem, struct fiemap_extent *fm_ext, } /* test for the boundary of an extent */ -#define ext_boundary(ext, exp, exp_dense) \ -(((ext).fe_logical != 0) && \ -((ext).fe_physical != (exp)) && \ -((ext).fe_physical != (exp_dense))) +#define ext_boundary(ext, exp) \ +((ext).fe_logical != 0) && \ +((ext).fe_physical != (exp)) /* * Read the extents of an open file descriptor into a table of struct _extent. @@ -4213,10 +4212,9 @@ static struct _extent *_stats_get_extents_for_file(struct dm_pool *mem, int fd, uint64_t *buf; struct fiemap *fiemap = NULL; struct fiemap_extent *fm_ext = NULL; - struct fiemap_extent fm_last = {0}; + struct fiemap_extent fm_last = {0}, fm_pending = {0}; struct _extent *extents; unsigned long long expected = 0; - unsigned long long expected_dense = 0; unsigned long flags = 0; unsigned int i, num = 0; int tot_extents = 0, n = 0; @@ -4264,31 +4262,51 @@ static struct _extent *_stats_get_extents_for_file(struct dm_pool *mem, int fd, break; for (i = 0; i < fiemap->fm_mapped_extents; i++) { - expected_dense = fm_last.fe_physical + - fm_last.fe_length; - expected = fm_last.fe_physical + - fm_ext[i].fe_logical - fm_last.fe_logical; - if (ext_boundary(fm_ext[i], expected, expected_dense)) { + expected = fm_last.fe_physical + fm_last.fe_length; + + if (fm_ext[i].fe_flags & FIEMAP_EXTENT_LAST) + last = 1; + + /* cannot map extents that are not yet allocated. */ + if (fm_ext[i].fe_flags + & (FIEMAP_EXTENT_UNKNOWN | FIEMAP_EXTENT_DELALLOC)) + continue; + + if (ext_boundary(fm_ext[i], expected)) { tot_extents++; - if (!_stats_add_extent(mem, fm_ext + i, - tot_extents - 1)) + if (!_stats_add_extent(mem, &fm_pending, + fm_pending.fe_flags)) goto_bad; + /* Begin a new pending extent. */ + fm_pending.fe_flags = tot_extents - 1; + fm_pending.fe_physical = fm_ext[i].fe_physical; + fm_pending.fe_logical = fm_ext[i].fe_logical; + fm_pending.fe_length = fm_ext[i].fe_length; } else { expected = 0; if (!tot_extents) tot_extents = 1; - if (fm_ext[i].fe_logical == 0) - if (!_stats_add_extent(mem, fm_ext + i, - tot_extents - 1)) - goto_bad; + /* Begin a new pending extent for extent 0. */ + if (fm_ext[i].fe_logical == 0) { + fm_pending.fe_flags = tot_extents - 1; + fm_pending.fe_physical = fm_ext[i].fe_physical; + fm_pending.fe_logical = fm_ext[i].fe_logical; + fm_pending.fe_length = fm_ext[i].fe_length; + } else + fm_pending.fe_length += fm_ext[i].fe_length; } num += tot_extents; - if (fm_ext[i].fe_flags & FIEMAP_EXTENT_LAST) - last = 1; fm_last = fm_ext[i]; n++; } + /* + * If the file only has a single extent, no boundary is ever + * detected to trigger addition of the first extent. + */ + if (fm_ext[i].fe_logical == 0) + _stats_add_extent(mem, &fm_pending, fm_pending.fe_flags); + fiemap->fm_start = (fm_ext[i - 1].fe_logical + fm_ext[i - 1].fe_length); } while (last == 0);

1 0

master - lvchange: allow a transiently failed RaidLV to be refreshed
by Heinz Mauelshagen 13 Dec '16

13 Dec '16

Gitweb: http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=87117c2b2546231c… Commit: 87117c2b2546231c789f92c75590f053a8fb987c Parent: 75ec7c8deeed95f005d78b4b962f70f783f7f0a0 Author: Heinz Mauelshagen <heinzm(a)redhat.com> AuthorDate: Mon Dec 12 22:06:17 2016 +0100 Committer: Heinz Mauelshagen <heinzm(a)redhat.com> CommitterDate: Mon Dec 12 22:08:47 2016 +0100 lvchange: allow a transiently failed RaidLV to be refreshed Enhance commit 0b8bf73a63d8 to refresh the top-level LV correctly in case of a clustered, remotely activated RaidLV. Related: rhbz1399844 --- lib/metadata/lv_manip.c | 26 ++++++++++++++++---------- 1 files changed, 16 insertions(+), 10 deletions(-) diff --git a/lib/metadata/lv_manip.c b/lib/metadata/lv_manip.c index 120217f..fdfee36 100644 --- a/lib/metadata/lv_manip.c +++ b/lib/metadata/lv_manip.c @@ -1417,17 +1417,23 @@ int lv_refresh_suspend_resume(const struct logical_volume *lv) * with transient failures of SubLVs. */ if (lv_is_raid(lv)) { - uint32_t s; - struct lv_segment *seg = first_seg(lv); - - for (s = 0; s < seg->area_count; s++) { - if (seg_type(seg, s) == AREA_LV && - !_lv_refresh_suspend_resume(seg_lv(seg, s))) - return 0; - if (seg->meta_areas && - seg_metatype(seg, s) == AREA_LV && - !_lv_refresh_suspend_resume(seg_metalv(seg, s))) + if (vg_is_clustered(lv->vg) && + lv_is_active_remotely(lv)) { + if (!_lv_refresh_suspend_resume(lv)) return 0; + } else { + uint32_t s; + struct lv_segment *seg = first_seg(lv); + + for (s = 0; s < seg->area_count; s++) { + if (seg_type(seg, s) == AREA_LV && + !_lv_refresh_suspend_resume(seg_lv(seg, s))) + return 0; + if (seg->meta_areas && + seg_metatype(seg, s) == AREA_LV && + !_lv_refresh_suspend_resume(seg_metalv(seg, s))) + return 0; + } } }

1 0

master - tests: update seg_size_pe
by Zdenek Kabelac 12 Dec '16

12 Dec '16

Gitweb: http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=75ec7c8deeed95f0… Commit: 75ec7c8deeed95f005d78b4b962f70f783f7f0a0 Parent: 3dd6e2c7e13a8b04829ab6a9a9df29a46b532bf0 Author: Zdenek Kabelac <zkabelac(a)redhat.com> AuthorDate: Mon Dec 12 11:49:26 2016 +0100 Committer: Zdenek Kabelac <zkabelac(a)redhat.com> CommitterDate: Mon Dec 12 11:51:50 2016 +0100 tests: update seg_size_pe Default prepare_vg uses 512K - so update test accordingly --- test/shell/error-usage.sh | 2 +- test/shell/zero-usage.sh | 2 +- 2 files changed, 2 insertions(+), 2 deletions(-) diff --git a/test/shell/error-usage.sh b/test/shell/error-usage.sh index c8765a6..062b71e 100644 --- a/test/shell/error-usage.sh +++ b/test/shell/error-usage.sh @@ -26,7 +26,7 @@ lvextend -L+1 $vg/$lv1 check lv_field $vg/$lv1 lv_modules "error" check lv_field $vg/$lv1 segtype "error" check lv_field $vg/$lv1 seg_count "1" -check lv_field $vg/$lv1 seg_size_pe "8" # 8 * 256 +check lv_field $vg/$lv1 seg_size_pe "4" # 4 * 512 => 2M # FIXME should we print info we are ignoring stripping? lvextend -L+1 -I64 -i2 $vg/$lv1 diff --git a/test/shell/zero-usage.sh b/test/shell/zero-usage.sh index 5d393f5..68afaaf 100644 --- a/test/shell/zero-usage.sh +++ b/test/shell/zero-usage.sh @@ -32,6 +32,6 @@ test "$sum1" = "$sum2" check lv_field $vg/$lv1 lv_modules "zero" check lv_field $vg/$lv1 segtype "zero" check lv_field $vg/$lv1 seg_count "1" -check lv_field $vg/$lv1 seg_size_pe "8" # 8 * 256 +check lv_field $vg/$lv1 seg_size_pe "4" # 4 * 512 vgremove -ff $vg

1 0

master - tests: use prepare_vg more often
by Zdenek Kabelac 12 Dec '16

12 Dec '16

Gitweb: http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=3dd6e2c7e13a8b04… Commit: 3dd6e2c7e13a8b04829ab6a9a9df29a46b532bf0 Parent: 8c9c6ab66060cfc3383297a6baed229f9f341d4a Author: Zdenek Kabelac <zkabelac(a)redhat.com> AuthorDate: Mon Dec 12 11:22:41 2016 +0100 Committer: Zdenek Kabelac <zkabelac(a)redhat.com> CommitterDate: Mon Dec 12 11:22:41 2016 +0100 tests: use prepare_vg more often --- test/shell/discards-thin.sh | 4 +--- test/shell/error-usage.sh | 4 +--- test/shell/lvconvert-mirror-basic.sh | 3 +-- test/shell/lvconvert-raid456.sh | 3 +-- test/shell/lvcreate-raid10.sh | 4 +--- test/shell/lvextend-thin-full.sh | 4 +--- test/shell/lvmetad-dump.sh | 4 +++- test/shell/lvresize-thin-external-origin.sh | 4 +--- test/shell/lvresize-thin-metadata.sh | 4 +--- test/shell/pvmove-all-segtypes.sh | 3 +-- test/shell/thin-foreign-dmeventd.sh | 4 +--- test/shell/thin-resize-match.sh | 4 +--- test/shell/zero-usage.sh | 4 +--- 13 files changed, 15 insertions(+), 34 deletions(-) diff --git a/test/shell/discards-thin.sh b/test/shell/discards-thin.sh index f85b22c..9a66dc8 100644 --- a/test/shell/discards-thin.sh +++ b/test/shell/discards-thin.sh @@ -25,12 +25,10 @@ export LVM_TEST_THIN_REPAIR_CMD=${LVM_TEST_THIN_REPAIR_CMD-/bin/false} # aux have_thin 1 1 0 || skip -aux prepare_pvs 2 64 +aux prepare_vg 2 64 aux extend_filter_LVMTEST -vgcreate $vg -s 64K $(cat DEVICES) - # Create named pool only lvcreate -l1 --discards ignore -T $vg/pool check lv_field $vg/pool discards "ignore" diff --git a/test/shell/error-usage.sh b/test/shell/error-usage.sh index 1f03071..c8765a6 100644 --- a/test/shell/error-usage.sh +++ b/test/shell/error-usage.sh @@ -16,9 +16,7 @@ SKIP_WITH_LVMPOLLD=1 . lib/inittest -aux prepare_pvs 1 - -vgcreate -s 256k $vg $(cat DEVICES) +aux prepare_vg 1 lvcreate --type error -L1 -n $lv1 $vg lvextend -L+1 $vg/$lv1 diff --git a/test/shell/lvconvert-mirror-basic.sh b/test/shell/lvconvert-mirror-basic.sh index 81fb0c6..c32b5c1 100644 --- a/test/shell/lvconvert-mirror-basic.sh +++ b/test/shell/lvconvert-mirror-basic.sh @@ -123,8 +123,7 @@ test_lvconvert() { fi } -aux prepare_pvs 5 5 -vgcreate -s 32k $vg $(cat DEVICES) +aux prepare_vg 5 5 MIRRORED="mirrored" # FIXME: Cluster is not supporting exlusive activation of mirrored log diff --git a/test/shell/lvconvert-raid456.sh b/test/shell/lvconvert-raid456.sh index 358ceec..8c87a03 100644 --- a/test/shell/lvconvert-raid456.sh +++ b/test/shell/lvconvert-raid456.sh @@ -28,8 +28,7 @@ get_image_pvs() { aux raid456_replace_works || skip aux have_raid 1 3 0 || skip -aux prepare_pvs 7 # 7 devices for 2 dev replacement of 5-dev RAID6 -vgcreate -s 256k $vg $(cat DEVICES) +aux prepare_vg 7 # 7 devices for 2 dev replacement of 5-dev RAID6 levels="5 6" aux have_raid4 && levels="4 5 6" diff --git a/test/shell/lvcreate-raid10.sh b/test/shell/lvcreate-raid10.sh index bebe221..68b4f46 100644 --- a/test/shell/lvcreate-raid10.sh +++ b/test/shell/lvcreate-raid10.sh @@ -23,9 +23,7 @@ lv_devices() { ######################################################## aux have_raid 1 3 0 || skip -aux prepare_pvs 6 20 # 6 devices for RAID10 (2-mirror,3-stripe) test -vgcreate -s 512k $vg $(cat DEVICES) - +aux prepare_vg 6 20 # 6 devices for RAID10 (2-mirror,3-stripe) test # # Create RAID10: # diff --git a/test/shell/lvextend-thin-full.sh b/test/shell/lvextend-thin-full.sh index 8fc89a8..3bafad4 100644 --- a/test/shell/lvextend-thin-full.sh +++ b/test/shell/lvextend-thin-full.sh @@ -27,9 +27,7 @@ test -n "$LVM_TEST_THIN_RESTORE_CMD" || LVM_TEST_THIN_RESTORE_CMD=$(which thin_r aux have_thin 1 10 0 || skip -aux prepare_pvs 3 256 - -vgcreate -s 1M $vg $(cat DEVICES) +aux prepare_vg 3 256 aux lvmconf 'activation/thin_pool_autoextend_percent = 30' \ 'activation/thin_pool_autoextend_threshold = 70' diff --git a/test/shell/lvmetad-dump.sh b/test/shell/lvmetad-dump.sh index 67ea01b..49cf9b7 100644 --- a/test/shell/lvmetad-dump.sh +++ b/test/shell/lvmetad-dump.sh @@ -15,11 +15,13 @@ SKIP_WITH_LVMPOLLD=1 . lib/inittest +# check we can talk to lvmetad +(echo | aux lvmetad_talk) || skip + aux prepare_pvs 2 vgcreate $vg1 "$dev1" "$dev2" lvcreate -n bar -l 1 $vg1 -(echo | aux lvmetad_talk) || skip aux lvmetad_dump | tee lvmetad.txt grep $vg1 lvmetad.txt diff --git a/test/shell/lvresize-thin-external-origin.sh b/test/shell/lvresize-thin-external-origin.sh index c1ebffe..686da39 100644 --- a/test/shell/lvresize-thin-external-origin.sh +++ b/test/shell/lvresize-thin-external-origin.sh @@ -23,9 +23,7 @@ aux have_thin 1 2 0 || skip # Pretend we miss the external_origin_extend feature aux lvmconf 'global/thin_disabled_features = [ "external_origin_extend" ]' -aux prepare_pvs 2 - -vgcreate -s 1M $vg $(cat DEVICES) +aux prepare_vg 2 lvcreate -L10 -n $lv1 $vg diff --git a/test/shell/lvresize-thin-metadata.sh b/test/shell/lvresize-thin-metadata.sh index 2ef4fce..a08fb40 100644 --- a/test/shell/lvresize-thin-metadata.sh +++ b/test/shell/lvresize-thin-metadata.sh @@ -18,9 +18,7 @@ export LVM_TEST_THIN_REPAIR_CMD=${LVM_TEST_THIN_REPAIR_CMD-/bin/false} aux have_thin 1 10 0 || skip -aux prepare_pvs 3 1256 - -vgcreate -s 1M $vg $(cat DEVICES) +aux prepare_vg 3 1256 for deactivate in true false; do # Create some thin volumes diff --git a/test/shell/pvmove-all-segtypes.sh b/test/shell/pvmove-all-segtypes.sh index e5f3b74..6a34ecf 100644 --- a/test/shell/pvmove-all-segtypes.sh +++ b/test/shell/pvmove-all-segtypes.sh @@ -17,8 +17,7 @@ SKIP_WITH_LVMLOCKD=1 which md5sum || skip -aux prepare_pvs 5 20 -vgcreate -s 256k $vg $(cat DEVICES) +aux prepare_vg 5 20 # Each of the following tests does: # 1) Create two LVs - one linear and one other segment type diff --git a/test/shell/thin-foreign-dmeventd.sh b/test/shell/thin-foreign-dmeventd.sh index 50ff44e..4ba3205 100644 --- a/test/shell/thin-foreign-dmeventd.sh +++ b/test/shell/thin-foreign-dmeventd.sh @@ -42,9 +42,7 @@ which mkfs.ext4 || skip export MKE2FS_CONFIG="$TESTOLDPWD/lib/mke2fs.conf" aux prepare_dmeventd -aux prepare_pvs 2 64 - -vgcreate $vg -s 64K $(cat DEVICES) +aux prepare_vg 2 64 # Create named pool only lvcreate --errorwhenfull y -L2 -T $vg/pool diff --git a/test/shell/thin-resize-match.sh b/test/shell/thin-resize-match.sh index 047225b..c78842f 100644 --- a/test/shell/thin-resize-match.sh +++ b/test/shell/thin-resize-match.sh @@ -23,9 +23,7 @@ which md5sum || skip aux have_thin 1 0 0 || skip -aux prepare_pvs 2 20 - -vgcreate -s 512K $vg $(< DEVICES) +aux prepare_vg 2 20 lvcreate -L1M -V2M -n $lv1 -T $vg/pool diff --git a/test/shell/zero-usage.sh b/test/shell/zero-usage.sh index 768df28..5d393f5 100644 --- a/test/shell/zero-usage.sh +++ b/test/shell/zero-usage.sh @@ -18,9 +18,7 @@ SKIP_WITH_LVMPOLLD=1 which md5sum || skip -aux prepare_pvs 1 - -vgcreate -s 256k $vg $(cat DEVICES) +aux prepare_vg 1 lvcreate --type zero -L1 -n $lv1 $vg lvextend -L+1 $vg/$lv1

1 0

master - tests: no left devices check for skipped test
by Zdenek Kabelac 12 Dec '16

12 Dec '16

Gitweb: http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=8c9c6ab66060cfc3… Commit: 8c9c6ab66060cfc3383297a6baed229f9f341d4a Parent: 48d33e5fb671af16f84d74afab9f86d47ca112fc Author: Zdenek Kabelac <zkabelac(a)redhat.com> AuthorDate: Mon Dec 12 11:22:10 2016 +0100 Committer: Zdenek Kabelac <zkabelac(a)redhat.com> CommitterDate: Mon Dec 12 11:22:10 2016 +0100 tests: no left devices check for skipped test --- test/lib/aux.sh | 5 ++++- 1 files changed, 4 insertions(+), 1 deletions(-) diff --git a/test/lib/aux.sh b/test/lib/aux.sh index 4749f81..a195883 100644 --- a/test/lib/aux.sh +++ b/test/lib/aux.sh @@ -520,7 +520,10 @@ teardown() { if test -f TESTNAME ; then - TEST_LEAKED_DEVICES=$(dmsetup table | grep "$PREFIX" | grep -v "${PREFIX}pv") || true + if test ! -f SKIP_THIS_TEST ; then + # Evaluate left devices only for non-skipped tests + TEST_LEAKED_DEVICES=$(dmsetup table | grep "$PREFIX" | grep -v "${PREFIX}pv") || true + fi kill_tagged_processes

1 0

master - cov: use unsigned for single bit values
by Zdenek Kabelac 12 Dec '16

12 Dec '16

Gitweb: http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=48d33e5fb671af16… Commit: 48d33e5fb671af16f84d74afab9f86d47ca112fc Parent: 56e4e2ce2ba05990eb4e8ced8a9a966b986f14ab Author: Zdenek Kabelac <zkabelac(a)redhat.com> AuthorDate: Mon Dec 12 11:21:42 2016 +0100 Committer: Zdenek Kabelac <zkabelac(a)redhat.com> CommitterDate: Mon Dec 12 11:21:42 2016 +0100 cov: use unsigned for single bit values Avoid using signed int. --- libdm/libdm-report.c | 4 ++-- 1 files changed, 2 insertions(+), 2 deletions(-) diff --git a/libdm/libdm-report.c b/libdm/libdm-report.c index ebd2439..93b8774 100644 --- a/libdm/libdm-report.c +++ b/libdm/libdm-report.c @@ -95,8 +95,8 @@ struct report_group_item { uint32_t finished_count; } store; struct report_group_item *parent; - int output_done:1; - int needs_closing:1; + unsigned output_done:1; + unsigned needs_closing:1; void *data; };

1 0

master - tests: track leaked devices in tests
by Zdenek Kabelac 12 Dec '16

12 Dec '16

Gitweb: http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=56e4e2ce2ba05990… Commit: 56e4e2ce2ba05990eb4e8ced8a9a966b986f14ab Parent: 273ccb11704075b1876f6931e9e8de19ff2b31a3 Author: Zdenek Kabelac <zkabelac(a)redhat.com> AuthorDate: Sun Dec 11 10:25:15 2016 +0100 Committer: Zdenek Kabelac <zkabelac(a)redhat.com> CommitterDate: Sun Dec 11 23:24:19 2016 +0100 tests: track leaked devices in tests When test calls teardown, no devices created by test are expected to be left in table. Trap such orphans and make the test fail. --- test/lib/aux.sh | 8 ++++++++ 1 files changed, 8 insertions(+), 0 deletions(-) diff --git a/test/lib/aux.sh b/test/lib/aux.sh index 7f3c84a..4749f81 100644 --- a/test/lib/aux.sh +++ b/test/lib/aux.sh @@ -520,6 +520,8 @@ teardown() { if test -f TESTNAME ; then + TEST_LEAKED_DEVICES=$(dmsetup table | grep "$PREFIX" | grep -v "${PREFIX}pv") || true + kill_tagged_processes if test -n "$LVM_TEST_LVMLOCKD_TEST" ; then @@ -567,6 +569,12 @@ teardown() { fi + test -z "$TEST_LEAKED_DEVICES" || { + echo "Unexpected devices left dm table:" + echo "$TEST_LEAKED_DEVICES" + return 1 + } + test -n "$TESTDIR" && { cd "$TESTOLDPWD" rm -rf "$TESTDIR" || echo BLA

1 0

master - tests: remove unwanted exit
by Zdenek Kabelac 12 Dec '16

12 Dec '16

Gitweb: http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=273ccb11704075b1… Commit: 273ccb11704075b1876f6931e9e8de19ff2b31a3 Parent: b902fd08f688c823937fec612a0ea72848289e56 Author: Zdenek Kabelac <zkabelac(a)redhat.com> AuthorDate: Sun Dec 11 10:21:26 2016 +0100 Committer: Zdenek Kabelac <zkabelac(a)redhat.com> CommitterDate: Sun Dec 11 23:24:19 2016 +0100 tests: remove unwanted exit Exit seemes slipped in from local testing in: eb6b2a11e3548f7598c726787494be2c3ea4dca1 --- test/shell/lvchange-thin.sh | 2 +- 1 files changed, 1 insertions(+), 1 deletions(-) diff --git a/test/shell/lvchange-thin.sh b/test/shell/lvchange-thin.sh index b5a6a86..3a8cc52 100644 --- a/test/shell/lvchange-thin.sh +++ b/test/shell/lvchange-thin.sh @@ -49,7 +49,7 @@ test -n "$no_discard" || check grep_dmsetup status $vg-pool " no_discard_passdow lvchange --discards passdown $vg/pool check grep_dmsetup table $vg-pool -v "passdown" test -n "$no_discard" || check grep_dmsetup status $vg-pool " discard_passdown" -exit + # zero_ARG (default is 'yes') check grep_dmsetup table $vg-pool -v "zeroing" lvchange --zero n $vg/pool

1 0

master - tests: remove some leaking device
by Zdenek Kabelac 12 Dec '16

12 Dec '16

Gitweb: http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=b902fd08f688c823… Commit: b902fd08f688c823937fec612a0ea72848289e56 Parent: bcf59a1b185c07a3f7e10f9de13e83501f9a852b Author: Zdenek Kabelac <zkabelac(a)redhat.com> AuthorDate: Sun Dec 11 10:24:06 2016 +0100 Committer: Zdenek Kabelac <zkabelac(a)redhat.com> CommitterDate: Sun Dec 11 23:24:19 2016 +0100 tests: remove some leaking device These tests leaks devices (known bugs). Remove them via dmsetup. TODO: fix actual commands --- test/shell/lvconvert-repair-cache.sh | 4 ++++ test/shell/lvcreate-repair.sh | 3 +++ 2 files changed, 7 insertions(+), 0 deletions(-) diff --git a/test/shell/lvconvert-repair-cache.sh b/test/shell/lvconvert-repair-cache.sh index 565ebf4..f709813 100644 --- a/test/shell/lvconvert-repair-cache.sh +++ b/test/shell/lvconvert-repair-cache.sh @@ -132,3 +132,7 @@ not "$FSCK" -n "$DM_DEV_DIR/$vg/$lv1" aux enable_dev "$dev1" vgremove -ff $vg + +# FIXME - device should not be here +should not dmsetup remove ${vg}-cpool_cmeta-missing_0_0 +should not dmsetup remove ${vg}-cpool_cdata-missing_0_0 diff --git a/test/shell/lvcreate-repair.sh b/test/shell/lvcreate-repair.sh index 53a6fc1..f3be28a 100644 --- a/test/shell/lvcreate-repair.sh +++ b/test/shell/lvcreate-repair.sh @@ -100,3 +100,6 @@ vgreduce --removemissing --force $vg # Failed to activate new LV. should lvcreate -l1 $vg "$dev1" +should not dmsetup remove ${vg}-lvol0 + +vgremove -ff $vg

1 0

master - tests: test should clean devices it has created
by Zdenek Kabelac 12 Dec '16

12 Dec '16

Gitweb: http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=bcf59a1b185c07a3… Commit: bcf59a1b185c07a3f7e10f9de13e83501f9a852b Parent: 198f335139f7a538272c8cb6657ad4f5d58b01db Author: Zdenek Kabelac <zkabelac(a)redhat.com> AuthorDate: Sun Dec 11 10:18:44 2016 +0100 Committer: Zdenek Kabelac <zkabelac(a)redhat.com> CommitterDate: Sun Dec 11 23:24:19 2016 +0100 tests: test should clean devices it has created To be able to detect lvm2 command is not leaking some 'unexpected' device - remove all devices before test exits by its own command so test teardown now can check what was 'left' unexpectedly. --- test/api/lvtest.sh | 2 ++ test/shell/error-usage.sh | 2 ++ test/shell/lvconvert-thin.sh | 3 +-- test/shell/lvconvert-twostep.sh | 3 ++- test/shell/lvcreate-small-snap.sh | 2 ++ test/shell/lvextend-percent-extents.sh | 2 ++ test/shell/lvresize-thin-external-origin.sh | 2 ++ test/shell/lvs-cache.sh | 2 ++ test/shell/mdata-strings.sh | 2 ++ test/shell/nomda-missing.sh | 2 ++ test/shell/nomda-restoremissing.sh | 2 ++ test/shell/profiles-thin.sh | 2 ++ test/shell/profiles.sh | 2 ++ test/shell/pv-range-overflow.sh | 1 + test/shell/select-report.sh | 2 ++ test/shell/vgcfgbackup-lvm1.sh | 2 +- test/shell/zero-usage.sh | 2 ++ 17 files changed, 31 insertions(+), 4 deletions(-) diff --git a/test/api/lvtest.sh b/test/api/lvtest.sh index f34f0f8..8cba14c 100644 --- a/test/api/lvtest.sh +++ b/test/api/lvtest.sh @@ -21,3 +21,5 @@ lvcreate -n test -l 5 $vg aux apitest lvtest $vg check lv_field $vg/test lv_name test + +vgremove -ff $vg diff --git a/test/shell/error-usage.sh b/test/shell/error-usage.sh index 531dad3..1f03071 100644 --- a/test/shell/error-usage.sh +++ b/test/shell/error-usage.sh @@ -32,3 +32,5 @@ check lv_field $vg/$lv1 seg_size_pe "8" # 8 * 256 # FIXME should we print info we are ignoring stripping? lvextend -L+1 -I64 -i2 $vg/$lv1 + +vgremove -ff $vg diff --git a/test/shell/lvconvert-thin.sh b/test/shell/lvconvert-thin.sh index 1999d86..4bd1acd 100644 --- a/test/shell/lvconvert-thin.sh +++ b/test/shell/lvconvert-thin.sh @@ -149,6 +149,5 @@ fail lvconvert --yes --type snapshot $vg/$lv1 $vg/$lv2 fail lvconvert --yes --type snapshot $vg/$lv2 $vg/$lv1 fail lvconvert --yes --type thin-pool $vg/$lv1 -lvremove -f $vg - vgremove -ff $vg +vgremove -ff $vg1 diff --git a/test/shell/lvconvert-twostep.sh b/test/shell/lvconvert-twostep.sh index f202ae2..9c54756 100644 --- a/test/shell/lvconvert-twostep.sh +++ b/test/shell/lvconvert-twostep.sh @@ -24,9 +24,10 @@ lvconvert --mirrorlog core $vg/mirror not lvconvert -m 1 --mirrorlog disk $vg/mirror "$dev3" 2>&1 | tee errs grep "two steps" errs -test -e LOCAL_CLVMD && exit 0 +if test ! -e LOCAL_CLVMD ; then # FIXME mirrored unsupported in cluster not lvconvert -m 1 --mirrorlog mirrored $vg/mirror "$dev3" "$dev4" 2>&1 | tee errs grep "two steps" errs +fi vgremove -ff $vg diff --git a/test/shell/lvcreate-small-snap.sh b/test/shell/lvcreate-small-snap.sh index 43424de..d7b5104 100644 --- a/test/shell/lvcreate-small-snap.sh +++ b/test/shell/lvcreate-small-snap.sh @@ -32,3 +32,5 @@ grep "smaller" lvcreate.out not lvcreate -s -l 4 -n snapB $vg/one 2>&1 | tee lvcreate.out not grep "suspend origin one" lvcreate.out grep "smaller" lvcreate.out + +vgremove -ff $vg diff --git a/test/shell/lvextend-percent-extents.sh b/test/shell/lvextend-percent-extents.sh index 67c36b2..56932a7 100644 --- a/test/shell/lvextend-percent-extents.sh +++ b/test/shell/lvextend-percent-extents.sh @@ -108,3 +108,5 @@ lvremove -f $vg/$lv lvcreate -i2 -I 64k -l10 -n $lv $vg lvreduce -f -l1 $vg/$lv check lv_field $vg/$lv lv_size "8.00m" + +vgremove -ff $vg diff --git a/test/shell/lvresize-thin-external-origin.sh b/test/shell/lvresize-thin-external-origin.sh index b864b81..c1ebffe 100644 --- a/test/shell/lvresize-thin-external-origin.sh +++ b/test/shell/lvresize-thin-external-origin.sh @@ -50,3 +50,5 @@ check lv_field $vg/$lv1 lv_size "5.00" --units m --nosuffix # Try to resize again back up to the size of external origin lvresize -L+5 -f $vg/$lv1 check lv_field $vg/$lv1 lv_size "10.00" --units m --nosuffix + +vgremove -ff $vg diff --git a/test/shell/lvs-cache.sh b/test/shell/lvs-cache.sh index 74077ae..f2463b8 100644 --- a/test/shell/lvs-cache.sh +++ b/test/shell/lvs-cache.sh @@ -88,3 +88,5 @@ lvremove -f $vg lvcreate -n foo -l 1 $vg lvs -a -S 'cache_policy=undefined' | grep foo + +vgremove -ff $vg diff --git a/test/shell/mdata-strings.sh b/test/shell/mdata-strings.sh index d3e831f..5d22953 100644 --- a/test/shell/mdata-strings.sh +++ b/test/shell/mdata-strings.sh @@ -45,3 +45,5 @@ vgcreate $vg "$created" # 'no parse errors and VG really exists' vgs $vg 2>err not grep "Parse error" err + +dmsetup remove "${PREFIX}${pv_ugly}" diff --git a/test/shell/nomda-missing.sh b/test/shell/nomda-missing.sh index f9c394e..9fe4bec 100644 --- a/test/shell/nomda-missing.sh +++ b/test/shell/nomda-missing.sh @@ -84,3 +84,5 @@ check active $vg linear2 check active $vg linear12 check active $vg mirror12 check active $vg mirror123 + +vgremove -ff $vg diff --git a/test/shell/nomda-restoremissing.sh b/test/shell/nomda-restoremissing.sh index e0d1da0..811637b 100644 --- a/test/shell/nomda-restoremissing.sh +++ b/test/shell/nomda-restoremissing.sh @@ -32,3 +32,5 @@ not lvcreate -aey --type mirror -m 1 -l 1 -n mirror $vg # write operations fail aux enable_dev "$dev1" lvcreate -aey --type mirror -m 1 -l 1 -n mirror $vg # no MDA => automatically restored vgck $vg + +vgremove -ff $vg diff --git a/test/shell/profiles-thin.sh b/test/shell/profiles-thin.sh index 6b8bc6a..d97bf59 100644 --- a/test/shell/profiles-thin.sh +++ b/test/shell/profiles-thin.sh @@ -74,3 +74,5 @@ check vg_field $vg profile "thin-performance" check lv_field $vg/pool_performance_inherited profile "" $SHOULD check lv_field $vg/pool_performance_inherited chunk_size 1.00m check lv_field $vg/pool_performance_inherited zero "" + +vgremove -ff $vg diff --git a/test/shell/profiles.sh b/test/shell/profiles.sh index 562b7b1..508f6db 100644 --- a/test/shell/profiles.sh +++ b/test/shell/profiles.sh @@ -131,3 +131,5 @@ lvm dumpconfig --type profilable-command --file etc/profile/generated.profile pvs --profile generated &> msg not grep "$MSG_NOT_PROFILABLE" msg not grep "$MSG_IGNORING_INVALID_CMD_PROFILE" msg + +vgremove -ff $vg1 diff --git a/test/shell/pv-range-overflow.sh b/test/shell/pv-range-overflow.sh index 37fd659..2d80f4e 100644 --- a/test/shell/pv-range-overflow.sh +++ b/test/shell/pv-range-overflow.sh @@ -32,3 +32,4 @@ grep "Logical volume bogus not found." err # 'run the offending pvmove command' not pvmove -v -n$lv "$dev1":4294967296 "$dev2" +vgremove -ff $vg diff --git a/test/shell/select-report.sh b/test/shell/select-report.sh index b39ce54..aad3857 100644 --- a/test/shell/select-report.sh +++ b/test/shell/select-report.sh @@ -218,3 +218,5 @@ sel lv '(lv_name=vol1 || lv_name=vol2) && lv_size < 8m' vol2 sel lv '(lv_name=vol1 && lv_size=8m) && vg_tags=vg_tag2' vol1 # negation of clause grouped by ( ) sel lv '!(lv_name=vol1 || lv_name=vol2)' abc xyz orig snap + +vgremove -ff $vg1 $vg2 $vg3 diff --git a/test/shell/vgcfgbackup-lvm1.sh b/test/shell/vgcfgbackup-lvm1.sh index 4a7f32a..f3c7ab4 100644 --- a/test/shell/vgcfgbackup-lvm1.sh +++ b/test/shell/vgcfgbackup-lvm1.sh @@ -30,7 +30,7 @@ vgcreate -M1 -c n $vg $(cat DEVICES) lvcreate -l1 -n $lv1 $vg "$dev1" pvremove -ff -y "$dev2" not lvcreate -l1 -n $lv1 $vg "$dev3" +lvchange -an $vg vgcfgbackup -f "backup.$$" $vg fi - diff --git a/test/shell/zero-usage.sh b/test/shell/zero-usage.sh index ee91bda..768df28 100644 --- a/test/shell/zero-usage.sh +++ b/test/shell/zero-usage.sh @@ -35,3 +35,5 @@ check lv_field $vg/$lv1 lv_modules "zero" check lv_field $vg/$lv1 segtype "zero" check lv_field $vg/$lv1 seg_count "1" check lv_field $vg/$lv1 seg_size_pe "8" # 8 * 256 + +vgremove -ff $vg

1 0

master - tests: slower device
by Zdenek Kabelac 12 Dec '16

12 Dec '16

Gitweb: http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=198f335139f7a538… Commit: 198f335139f7a538272c8cb6657ad4f5d58b01db Parent: 4a33e4c509fb72bf3540d20bd35e1f0944e1d37a Author: Zdenek Kabelac <zkabelac(a)redhat.com> AuthorDate: Sat Dec 10 15:07:14 2016 +0100 Committer: Zdenek Kabelac <zkabelac(a)redhat.com> CommitterDate: Sun Dec 11 23:24:19 2016 +0100 tests: slower device Some of test machines are too fast, slow raid syncing even more. --- test/shell/lvchange-rebuild-raid.sh | 2 +- 1 files changed, 1 insertions(+), 1 deletions(-) diff --git a/test/shell/lvchange-rebuild-raid.sh b/test/shell/lvchange-rebuild-raid.sh index 6b91941..df6a8cb 100644 --- a/test/shell/lvchange-rebuild-raid.sh +++ b/test/shell/lvchange-rebuild-raid.sh @@ -21,7 +21,7 @@ aux prepare_vg 8 # Delay legs so that rebuilding status characters can be read for d in $(< DEVICES) do - aux delay_dev "$d" 0 15 $(get first_extent_sector "$d") + aux delay_dev "$d" 0 20 $(get first_extent_sector "$d") done # rhbz 1064592

1 0

master - cov: declaration matching
by Zdenek Kabelac 12 Dec '16

12 Dec '16

Gitweb: http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=4a33e4c509fb72bf… Commit: 4a33e4c509fb72bf3540d20bd35e1f0944e1d37a Parent: b8c729efd094146437c3ed5b336b929608401535 Author: Zdenek Kabelac <zkabelac(a)redhat.com> AuthorDate: Fri Dec 9 23:10:00 2016 +0100 Committer: Zdenek Kabelac <zkabelac(a)redhat.com> CommitterDate: Sun Dec 11 23:24:19 2016 +0100 cov: declaration matching --- lib/format_text/format-text.h | 2 +- 1 files changed, 1 insertions(+), 1 deletions(-) diff --git a/lib/format_text/format-text.h b/lib/format_text/format-text.h index bfabdfd..d6e6b03 100644 --- a/lib/format_text/format-text.h +++ b/lib/format_text/format-text.h @@ -59,7 +59,7 @@ int add_da(struct dm_pool *mem, struct dm_list *das, void del_das(struct dm_list *das); int add_ba(struct dm_pool *mem, struct dm_list *eas, uint64_t start, uint64_t size); -void del_bas(struct dm_list *eas); +void del_bas(struct dm_list *bas); int add_mda(const struct format_type *fmt, struct dm_pool *mem, struct dm_list *mdas, struct device *dev, uint64_t start, uint64_t size, unsigned ignored); void del_mdas(struct dm_list *mdas);

1 0

master - cov: add internal error for impossible code path
by Zdenek Kabelac 12 Dec '16

12 Dec '16

Gitweb: http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=b8c729efd0941464… Commit: b8c729efd094146437c3ed5b336b929608401535 Parent: 794093722c657780171486dbfd8cd0ec392e4019 Author: Zdenek Kabelac <zkabelac(a)redhat.com> AuthorDate: Fri Dec 9 22:55:25 2016 +0100 Committer: Zdenek Kabelac <zkabelac(a)redhat.com> CommitterDate: Sun Dec 11 23:24:19 2016 +0100 cov: add internal error for impossible code path --- tools/vgchange.c | 5 +++++ 1 files changed, 5 insertions(+), 0 deletions(-) diff --git a/tools/vgchange.c b/tools/vgchange.c index 910f33a..455f688 100644 --- a/tools/vgchange.c +++ b/tools/vgchange.c @@ -533,6 +533,11 @@ static int _vgchange_locktype(struct cmd_context *cmd, struct logical_volume *lv; int lv_lock_count = 0; + if (!lock_type) { + log_error(INTERNAL_ERROR "No locktype_ARG."); + return 0; + } + /* * This is a special/forced exception to change the lock type to none. * It's needed for recovery cases and skips the normal steps of undoing

1 0

master - debug: missing stack traces
by Zdenek Kabelac 12 Dec '16

12 Dec '16

Gitweb: http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=794093722c657780… Commit: 794093722c657780171486dbfd8cd0ec392e4019 Parent: dc7d3dd5eda43186829ac4237ab049fe50cbc12a Author: Zdenek Kabelac <zkabelac(a)redhat.com> AuthorDate: Sat Dec 10 22:37:34 2016 +0100 Committer: Zdenek Kabelac <zkabelac(a)redhat.com> CommitterDate: Sun Dec 11 23:24:19 2016 +0100 debug: missing stack traces --- lib/metadata/raid_manip.c | 10 +++++----- 1 files changed, 5 insertions(+), 5 deletions(-) diff --git a/lib/metadata/raid_manip.c b/lib/metadata/raid_manip.c index 58e3bcf..98f30e2 100644 --- a/lib/metadata/raid_manip.c +++ b/lib/metadata/raid_manip.c @@ -3248,24 +3248,24 @@ static int _striped_or_raid0_to_raid45610_wrapper(TAKEOVER_FN_ARGS) log_debug_metadata("Converting LV %s from %s to %s", display_lvname(lv), SEG_TYPE_NAME_STRIPED, SEG_TYPE_NAME_RAID0); if (!(seg = _convert_striped_to_raid0(lv, 1 /* alloc_metadata_devs */, 0 /* update_and_reload */, allocate_pvs))) - return 0; + return_0; } /* Add metadata LVs */ if (seg_is_raid0(seg)) { log_debug_metadata("Adding metadata LVs to %s", display_lvname(lv)); if (!_raid0_add_or_remove_metadata_lvs(lv, 1 /* update_and_reload */, allocate_pvs, NULL)) - return 0; + return_0; /* raid0_meta -> raid4 needs clearing of MetaLVs in order to avoid raid disk role cahnge issues in the kernel */ } else if (segtype_is_raid4(new_segtype) && !_clear_meta_lvs(lv)) - return 0; + return_0; /* Add the additional component LV pairs */ log_debug_metadata("Adding %" PRIu32 " component LV pair(s) to %s", new_image_count - lv_raid_image_count(lv), display_lvname(lv)); if (!_lv_raid_change_image_count(lv, new_image_count, allocate_pvs, NULL, 0, 1)) - return 0; + return_0; if (segtype_is_raid4(new_segtype) && (!_shift_parity_dev(seg) || @@ -3289,7 +3289,7 @@ static int _striped_or_raid0_to_raid45610_wrapper(TAKEOVER_FN_ARGS) if (segtype_is_raid4(new_segtype)) { /* We had to rename SubLVs because of collision free sgifting, rename back... */ if (!_rename_area_lvs(lv, NULL)) - return 0; + return_0; if (!lv_update_and_reload(lv)) return_0; }

1 0

master - cleanup: use NAME_LEN stack buffer
by Zdenek Kabelac 12 Dec '16

12 Dec '16

Gitweb: http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=dc7d3dd5eda43186… Commit: dc7d3dd5eda43186829ac4237ab049fe50cbc12a Parent: 6cd68c55ddc280bcba72961944d24c6a66da1c18 Author: Zdenek Kabelac <zkabelac(a)redhat.com> AuthorDate: Sun Dec 11 12:19:38 2016 +0100 Committer: Zdenek Kabelac <zkabelac(a)redhat.com> CommitterDate: Sun Dec 11 23:24:19 2016 +0100 cleanup: use NAME_LEN stack buffer Using NAME_LEN we get at least easy max LV name size validation. Also code gets more simple. --- lib/metadata/mirror.c | 42 +++++++++++++++--------------------------- 1 files changed, 15 insertions(+), 27 deletions(-) diff --git a/lib/metadata/mirror.c b/lib/metadata/mirror.c index 73b1110..c668ea0 100644 --- a/lib/metadata/mirror.c +++ b/lib/metadata/mirror.c @@ -661,6 +661,7 @@ static int _split_mirror_images(struct logical_volume *lv, struct dm_list split_images; struct lv_list *lvl; struct cmd_context *cmd = lv->vg->cmd; + char layer_name[NAME_LEN], format[NAME_LEN]; if (!lv_is_mirrored(lv)) { log_error("Unable to split non-mirrored LV %s.", @@ -728,9 +729,6 @@ static int _split_mirror_images(struct logical_volume *lv, } if (!dm_list_empty(&split_images)) { - size_t len = strlen(new_lv->name) + 32; - char *layer_name, format[len]; - /* * A number of images have been split and * a new mirror layer must be formed @@ -747,23 +745,21 @@ static int _split_mirror_images(struct logical_volume *lv, dm_list_iterate_items(lvl, &split_images) { sub_lv = lvl->lv; - if (dm_snprintf(format, len, "%s_mimage_%%d", + if (dm_snprintf(format, sizeof(format), "%s_mimage_%%d", new_lv->name) < 0) { log_error("Failed to build new image name for %s.", display_lvname(new_lv)); return 0; } - if (!(layer_name = dm_pool_alloc(lv->vg->vgmem, len))) { - log_error("Unable to allocate memory."); - return 0; - } - if (!generate_lv_name(lv->vg, format, layer_name, len)|| - sscanf(layer_name, format, &i) != 1) { + if (!generate_lv_name(lv->vg, format, layer_name, sizeof(layer_name))) { log_error("Failed to generate new image names for %s.", display_lvname(new_lv)); return 0; } - sub_lv->name = layer_name; + if (!(sub_lv->name = dm_pool_strdup(lv->vg->vgmem, layer_name))) { + log_error("Unable to allocate memory."); + return 0; + } } if (!_merge_mirror_images(new_lv, &split_images)) { @@ -1403,15 +1399,11 @@ static int _create_mimage_lvs(struct alloc_handle *ah, int log) { uint32_t m, first_area; - char *img_name; - size_t len; - - len = strlen(lv->name) + 32; - img_name = alloca(len); + char img_name[NAME_LEN]; - if (dm_snprintf(img_name, len, "%s_mimage_%%d", lv->name) < 0) { - log_error("img_name allocation failed. " - "Remove new LV and retry."); + if (dm_snprintf(img_name, sizeof(img_name), "%s_mimage_%%d", lv->name) < 0) { + log_error("Failed to build new mirror image name for %s.", + display_lvname(lv)); return 0; } @@ -1784,14 +1776,10 @@ static struct logical_volume *_create_mirror_log(struct logical_volume *lv, const char *suffix) { struct logical_volume *log_lv; - char *log_name; - size_t len; - - len = strlen(lv_name) + 32; - log_name = alloca(len); /* alloca never fails */ + char log_name[NAME_LEN]; - if (dm_snprintf(log_name, len, "%s%s", lv_name, suffix) < 0) { - log_error("log_name allocation failed."); + if (dm_snprintf(log_name, sizeof(log_name), "%s%s", lv_name, suffix) < 0) { + log_error("Failed to build new mirror log name for %s.", lv_name); return NULL; }

1 0

master - cleanup: reuse existing function
by Zdenek Kabelac 12 Dec '16

12 Dec '16

Gitweb: http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=6cd68c55ddc280bc… Commit: 6cd68c55ddc280bcba72961944d24c6a66da1c18 Parent: 6f84d3c69cdc161508fcc29e82f1167334466084 Author: Zdenek Kabelac <zkabelac(a)redhat.com> AuthorDate: Sun Dec 11 14:31:47 2016 +0100 Committer: Zdenek Kabelac <zkabelac(a)redhat.com> CommitterDate: Sun Dec 11 23:24:19 2016 +0100 cleanup: reuse existing function Call lv_update_and_reload implementation. --- tools/pvmove_poll.c | 34 ++-------------------------------- 1 files changed, 2 insertions(+), 32 deletions(-) diff --git a/tools/pvmove_poll.c b/tools/pvmove_poll.c index 2cd746b..438d458 100644 --- a/tools/pvmove_poll.c +++ b/tools/pvmove_poll.c @@ -83,38 +83,8 @@ int pvmove_update_metadata(struct cmd_context *cmd, struct volume_group *vg, struct dm_list *lvs_changed __attribute__((unused)), unsigned flags __attribute__((unused))) { - log_verbose("Updating volume group metadata."); - if (!vg_write(vg)) { - log_error("ABORTING: Volume group metadata update failed."); - return 0; - } - - if (!suspend_lv(cmd, lv_mirr)) { - vg_revert(vg); - log_error("ABORTING: Temporary pvmove mirror reload failed."); - if (!revert_lv(cmd, lv_mirr)) - stack; - return 0; - } - - /* Commit on-disk metadata */ - if (!vg_commit(vg)) { - log_error("ABORTING: Volume group metadata update failed."); - if (!resume_lv(cmd, lv_mirr)) - log_error("Unable to reactivate logical volume \"%s\".", - lv_mirr->name); - if (!revert_lv(cmd, lv_mirr)) - stack; - return 0; - } - - if (!resume_lv(cmd, lv_mirr)) { - log_error("Unable to reactivate logical volume \"%s\".", - lv_mirr->name); - return 0; - } - - backup(vg); + if (!lv_update_and_reload(lv_mirr)) + return_0; return 1; }

1 0

master - libdm: validate vsnprintf
by Zdenek Kabelac 12 Dec '16

12 Dec '16

Gitweb: http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=6f84d3c69cdc1615… Commit: 6f84d3c69cdc161508fcc29e82f1167334466084 Parent: 3903f915f83634f91946550fda85d60b72ce3a88 Author: Zdenek Kabelac <zkabelac(a)redhat.com> AuthorDate: Fri Dec 9 22:54:08 2016 +0100 Committer: Zdenek Kabelac <zkabelac(a)redhat.com> CommitterDate: Sun Dec 11 23:24:17 2016 +0100 libdm: validate vsnprintf Avoid using buffer when no output has been generated. Missed in ee13f265f043b47a1b023321fb9e8470fb5703c1. --- WHATS_NEW_DM | 1 + libdm/libdm-common.c | 12 ++++++++---- 2 files changed, 9 insertions(+), 4 deletions(-) diff --git a/WHATS_NEW_DM b/WHATS_NEW_DM index 73707b6..f402113 100644 --- a/WHATS_NEW_DM +++ b/WHATS_NEW_DM @@ -1,5 +1,6 @@ Version 1.02.138 - ===================================== + Validation vsnprintf result in runtime translate of dm_log (1.02.136). Separate filemap extent allocation from region table. Fix segmentation fault when filemap region creation fails Fix performance of region cleanup for failed filemap creation. diff --git a/libdm/libdm-common.c b/libdm/libdm-common.c index 476f018..d8cf946 100644 --- a/libdm/libdm-common.c +++ b/libdm/libdm-common.c @@ -173,14 +173,16 @@ static void _log_to_default_log(int level, const char *file, int line, int dm_errno_or_class, const char *f, ...) { + int n; va_list ap; char buf[2 * PATH_MAX + 256]; /* big enough for most messages */ va_start(ap, f); - vsnprintf(buf, sizeof(buf), f, ap); + n = vsnprintf(buf, sizeof(buf), f, ap); va_end(ap); - dm_log(level, file, line, "%s", buf); + if (n > 0) /* Could be truncated */ + dm_log(level, file, line, "%s", buf); } /* @@ -195,14 +197,16 @@ __attribute__((format(printf, 4, 5))) static void _log_to_default_log_with_errno(int level, const char *file, int line, const char *f, ...) { + int n; va_list ap; char buf[2 * PATH_MAX + 256]; /* big enough for most messages */ va_start(ap, f); - vsnprintf(buf, sizeof(buf), f, ap); + n = vsnprintf(buf, sizeof(buf), f, ap); va_end(ap); - dm_log_with_errno(level, file, line, 0, "%s", buf); + if (n > 0) /* Could be truncated */ + dm_log_with_errno(level, file, line, 0, "%s", buf); } void dm_log_init(dm_log_fn fn)

1 0

master - pvmove: fix activation order
by Zdenek Kabelac 12 Dec '16

12 Dec '16

Gitweb: http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=3903f915f83634f9… Commit: 3903f915f83634f91946550fda85d60b72ce3a88 Parent: 67f9e6b175815d7a91a10ecc3d5098d30dbfed91 Author: Zdenek Kabelac <zkabelac(a)redhat.com> AuthorDate: Sun Dec 11 14:26:17 2016 +0100 Committer: Zdenek Kabelac <zkabelac(a)redhat.com> CommitterDate: Sun Dec 11 23:22:36 2016 +0100 pvmove: fix activation order For proper locking we need to gain lock first for mirror which needs to be deactivated later to be working in cluster. --- WHATS_NEW | 1 + tools/pvmove_poll.c | 10 +++++----- 2 files changed, 6 insertions(+), 5 deletions(-) diff --git a/WHATS_NEW b/WHATS_NEW index 63cdb89..ea15903 100644 --- a/WHATS_NEW +++ b/WHATS_NEW @@ -1,5 +1,6 @@ Version 2.02.169 - ===================================== + Fix pvmove leaving -pvmove0 error device in clustered VG. Avoid adding extra '_' at end of raid extracted images or metadata. Optimize another _rmeta clearing code. Fix deactivation of raid orphan devices for clustered VG. diff --git a/tools/pvmove_poll.c b/tools/pvmove_poll.c index ea4be24..2cd746b 100644 --- a/tools/pvmove_poll.c +++ b/tools/pvmove_poll.c @@ -157,17 +157,17 @@ int pvmove_finish(struct cmd_context *cmd, struct volume_group *vg, return 0; } + /* Unsuspend LVs */ + if (!resume_lvs(cmd, lvs_changed)) + stack; + /* Release mirror LV. (No pending I/O because it's been suspended.) */ - if (!resume_lv(cmd, lv_mirr)) { + if (!activate_lv_excl_local(cmd, lv_mirr)) { log_error("Unable to reactivate logical volume \"%s\"", lv_mirr->name); r = 0; } - /* Unsuspend LVs */ - if (!resume_lvs(cmd, lvs_changed)) - stack; - /* Deactivate mirror LV */ if (!deactivate_lv(cmd, lv_mirr)) { log_error("ABORTING: Unable to deactivate temporary logical "

1 0

master - raid: avoid _ at end of name of extracted metadata LV
by Zdenek Kabelac 12 Dec '16

12 Dec '16

Gitweb: http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=67f9e6b175815d7a… Commit: 67f9e6b175815d7a91a10ecc3d5098d30dbfed91 Parent: 55ca8043d4bb0ef4435a24728cad34f8b0ff6c8c Author: Zdenek Kabelac <zkabelac(a)redhat.com> AuthorDate: Sat Dec 10 14:51:00 2016 +0100 Committer: Zdenek Kabelac <zkabelac(a)redhat.com> CommitterDate: Sun Dec 11 23:20:51 2016 +0100 raid: avoid _ at end of name of extracted metadata LV Do not generate @PREFIX@vg/LV1_rmeta_1_extracted_. --- WHATS_NEW | 1 + lib/metadata/raid_manip.c | 2 +- 2 files changed, 2 insertions(+), 1 deletions(-) diff --git a/WHATS_NEW b/WHATS_NEW index 9e95792..63cdb89 100644 --- a/WHATS_NEW +++ b/WHATS_NEW @@ -1,5 +1,6 @@ Version 2.02.169 - ===================================== + Avoid adding extra '_' at end of raid extracted images or metadata. Optimize another _rmeta clearing code. Fix deactivation of raid orphan devices for clustered VG. Fix lvconvert raid1 to mirror table reload order. diff --git a/lib/metadata/raid_manip.c b/lib/metadata/raid_manip.c index 159bfa2..58e3bcf 100644 --- a/lib/metadata/raid_manip.c +++ b/lib/metadata/raid_manip.c @@ -1785,7 +1785,7 @@ static int _extract_image_component_error_seg(struct lv_segment *seg, if (!remove_seg_from_segs_using_this_lv(lv, seg)) return_0; - if (!(lv->name = _generate_raid_name(lv, "extracted_", -1))) + if (!(lv->name = _generate_raid_name(lv, "extracted", -1))) return_0; if (set_error_seg && !replace_lv_with_error_segment(lv))

1 0

master - raid: optimize clearing of lvs
by Zdenek Kabelac 12 Dec '16

12 Dec '16

Gitweb: http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=55ca8043d4bb0ef4… Commit: 55ca8043d4bb0ef4435a24728cad34f8b0ff6c8c Parent: 8831a541a831a2d25cb96ee09f6739526d29e5c1 Author: Zdenek Kabelac <zkabelac(a)redhat.com> AuthorDate: Sat Dec 10 20:53:17 2016 +0100 Committer: Zdenek Kabelac <zkabelac(a)redhat.com> CommitterDate: Sun Dec 11 23:19:41 2016 +0100 raid: optimize clearing of lvs Activate whole list of metadata lvs first before clearing them. (Similar to commit ada5733c5652d456ffa138b0d07e6897824813b0) TODO: make this clearing in a single common function. --- WHATS_NEW | 1 + lib/metadata/raid_manip.c | 92 ++++++++++++++++++++++---------------------- 2 files changed, 47 insertions(+), 46 deletions(-) diff --git a/WHATS_NEW b/WHATS_NEW index b6f74ec..9e95792 100644 --- a/WHATS_NEW +++ b/WHATS_NEW @@ -1,5 +1,6 @@ Version 2.02.169 - ===================================== + Optimize another _rmeta clearing code. Fix deactivation of raid orphan devices for clustered VG. Fix lvconvert raid1 to mirror table reload order. Add internal function for separate mirror log preparation. diff --git a/lib/metadata/raid_manip.c b/lib/metadata/raid_manip.c index 661c39b..159bfa2 100644 --- a/lib/metadata/raid_manip.c +++ b/lib/metadata/raid_manip.c @@ -379,10 +379,7 @@ static int _raid_remove_top_layer(struct logical_volume *lv, return 1; } -/* - * _clear_lv - * @lv - * +/* Makes on-disk metadata changes * If LV is active: * clear first block of device * otherwise: @@ -390,50 +387,15 @@ static int _raid_remove_top_layer(struct logical_volume *lv, * * Returns: 1 on success, 0 on failure */ -static int _clear_lv(struct logical_volume *lv) -{ - int was_active = lv_is_active_locally(lv); - - if (test_mode()) - return 1; - - lv->status |= LV_TEMPORARY; - if (!was_active && !activate_lv_local(lv->vg->cmd, lv)) { - log_error("Failed to activate localy %s for clearing.", - display_lvname(lv)); - return 0; - } - lv->status &= ~LV_TEMPORARY; - - log_verbose("Clearing metadata area of %s", - display_lvname(lv)); - /* - * Rather than wiping lv->size, we can simply - * wipe the first sector to remove the superblock of any previous - * RAID devices. It is much quicker. - */ - if (!wipe_lv(lv, (struct wipe_params) { .do_zero = 1, .zero_sectors = 1 })) { - log_error("Failed to zero %s.", - display_lvname(lv)); - return 0; - } - - if (!was_active && !deactivate_lv(lv->vg->cmd, lv)) { - log_error("Failed to deactivate %s.", - display_lvname(lv)); - return 0; - } - - return 1; -} - -/* Makes on-disk metadata changes */ static int _clear_lvs(struct dm_list *lv_list) { struct lv_list *lvl; struct volume_group *vg = NULL; + unsigned i = 0, sz = dm_list_size(lv_list); + char *was_active; + int r = 1; - if (dm_list_empty(lv_list)) { + if (!sz) { log_debug_metadata(INTERNAL_ERROR "Empty list of LVs given for clearing."); return 1; } @@ -447,6 +409,9 @@ static int _clear_lvs(struct dm_list *lv_list) vg = lvl->lv->vg; } + if (test_mode()) + return 1; + /* * FIXME: only vg_[write|commit] if LVs are not already written * as visible in the LVM metadata (which is never the case yet). @@ -454,11 +419,46 @@ static int _clear_lvs(struct dm_list *lv_list) if (!vg || !vg_write(vg) || !vg_commit(vg)) return_0; + was_active = alloca(sz); + dm_list_iterate_items(lvl, lv_list) - if (!_clear_lv(lvl->lv)) - return 0; + if (!(was_active[i++] = lv_is_active_locally(lvl->lv))) { + lvl->lv->status |= LV_TEMPORARY; + if (!activate_lv_excl_local(vg->cmd, lvl->lv)) { + log_error("Failed to activate localy %s for clearing.", + display_lvname(lvl->lv)); + r = 0; + goto out; + } + lvl->lv->status &= ~LV_TEMPORARY; + } - return 1; + dm_list_iterate_items(lvl, lv_list) { + log_verbose("Clearing metadata area %s.", display_lvname(lvl->lv)); + /* + * Rather than wiping lv->size, we can simply + * wipe the first sector to remove the superblock of any previous + * RAID devices. It is much quicker. + */ + if (!wipe_lv(lvl->lv, (struct wipe_params) { .do_zero = 1, .zero_sectors = 1 })) { + log_error("Failed to zero %s.", display_lvname(lvl->lv)); + r = 0; + goto out; + } + } +out: + /* TODO: deactivation is only needed with clustered locking + * in normal case we should keep device active + */ + sz = 0; + dm_list_iterate_items(lvl, lv_list) + if ((i > sz) && !was_active[sz++] && + !deactivate_lv(vg->cmd, lvl->lv)) { + log_error("Failed to deactivate %s.", display_lvname(lvl->lv)); + r = 0; /* continue deactivating */ + } + + return r; } /*

1 0

master - raid: fix delete on clustered vg
by Zdenek Kabelac 12 Dec '16

12 Dec '16

Gitweb: http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=8831a541a831a2d2… Commit: 8831a541a831a2d25cb96ee09f6739526d29e5c1 Parent: 0c8369099bc6105f39982b3e36dd9a53c347b07a Author: Zdenek Kabelac <zkabelac(a)redhat.com> AuthorDate: Sat Dec 10 20:00:32 2016 +0100 Committer: Zdenek Kabelac <zkabelac(a)redhat.com> CommitterDate: Sun Dec 11 23:18:22 2016 +0100 raid: fix delete on clustered vg For clustered VG ensure lock is grabbed first, so later deactivation works. TODO: fix tree to solve device removal automatically. --- WHATS_NEW | 1 + lib/metadata/raid_manip.c | 12 ++++++++++++ 2 files changed, 13 insertions(+), 0 deletions(-) diff --git a/WHATS_NEW b/WHATS_NEW index 74627ba..b6f74ec 100644 --- a/WHATS_NEW +++ b/WHATS_NEW @@ -1,5 +1,6 @@ Version 2.02.169 - ===================================== + Fix deactivation of raid orphan devices for clustered VG. Fix lvconvert raid1 to mirror table reload order. Add internal function for separate mirror log preparation. Fix segfault in lvmetad from missing NULL in daemon_reply_simple. diff --git a/lib/metadata/raid_manip.c b/lib/metadata/raid_manip.c index 5cbd819..661c39b 100644 --- a/lib/metadata/raid_manip.c +++ b/lib/metadata/raid_manip.c @@ -248,6 +248,12 @@ static int _deactivate_and_remove_lvs(struct volume_group *vg, struct dm_list *r { struct lv_list *lvl; + if (vg_is_clustered(vg)) + /* Need to take lock for proper deactivation */ + dm_list_iterate_items(lvl, removal_lvs) + if (!activate_lv_excl_local(vg->cmd, lvl->lv)) + return_0; + dm_list_iterate_items(lvl, removal_lvs) { if (!deactivate_lv(vg->cmd, lvl->lv)) return_0; @@ -2968,6 +2974,12 @@ static int _clear_meta_lvs(struct logical_volume *lv) !lv_update_and_reload(lv)) return_0; + /* Note: detached rmeta are NOT renamed */ + /* Grab locks first in case of clustered VG */ + if (vg_is_clustered(lv->vg)) + dm_list_iterate_items(lvl, &meta_lvs) + if (!activate_lv_excl_local(lv->vg->cmd, lvl->lv)) + return_0; /* * Now deactivate the MetaLVs before clearing, so * that _clear_lvs() will activate them visible.

1 0

master - raid: fix raid1 to mirror conversion
by Zdenek Kabelac 12 Dec '16

12 Dec '16

Gitweb: http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=0c8369099bc6105f… Commit: 0c8369099bc6105f39982b3e36dd9a53c347b07a Parent: 31564834db7610b37f52040fbba89f4c71de4c34 Author: Zdenek Kabelac <zkabelac(a)redhat.com> AuthorDate: Sat Dec 10 20:01:05 2016 +0100 Committer: Zdenek Kabelac <zkabelac(a)redhat.com> CommitterDate: Sun Dec 11 23:17:22 2016 +0100 raid: fix raid1 to mirror conversion Fix order of operation when converting raid1 into old mirror. Before any later metadata modification are initiated prepare mirror_log device with all clearing. Then directly convert raid1 into mirror with mirror_log. This convertion now properly see as precommitted metadata new 'mirror' and committed old 'raid' and is able to preload all LVs. --- WHATS_NEW | 1 + lib/metadata/raid_manip.c | 13 +++++++------ 2 files changed, 8 insertions(+), 6 deletions(-) diff --git a/WHATS_NEW b/WHATS_NEW index 274430f..74627ba 100644 --- a/WHATS_NEW +++ b/WHATS_NEW @@ -1,5 +1,6 @@ Version 2.02.169 - ===================================== + Fix lvconvert raid1 to mirror table reload order. Add internal function for separate mirror log preparation. Fix segfault in lvmetad from missing NULL in daemon_reply_simple. Simplify internal _info_run() and use _setup_task_run() for mknod. diff --git a/lib/metadata/raid_manip.c b/lib/metadata/raid_manip.c index 7c4ce61..5cbd819 100644 --- a/lib/metadata/raid_manip.c +++ b/lib/metadata/raid_manip.c @@ -2207,6 +2207,7 @@ static int _convert_raid1_to_mirror(struct logical_volume *lv, int update_and_reload, struct dm_list *removal_lvs) { + struct logical_volume *log_lv; struct lv_segment *seg = first_seg(lv); if (!seg_is_raid1(seg)) { @@ -2227,7 +2228,10 @@ static int _convert_raid1_to_mirror(struct logical_volume *lv, return 0; } - init_mirror_in_sync(new_image_count > seg->area_count ? 0 : 1); + if (!(log_lv = prepare_mirror_log(lv, (new_image_count <= seg->area_count) /* in sync */, + new_region_size, + allocate_pvs, lv->vg->alloc))) + return_0; /* TODO remove log_lv on error path */ /* Change image pair count to requested # of images */ if (new_image_count != seg->area_count) { @@ -2255,11 +2259,8 @@ static int _convert_raid1_to_mirror(struct logical_volume *lv, seg->status &= ~RAID; lv->status |= (MIRROR | MIRRORED); - /* Add mirror_log LV (should happen in wih image allocation */ - if (!add_mirror_log(lv->vg->cmd, lv, 1, seg->region_size, allocate_pvs, lv->vg->alloc)) { - log_error("Unable to add mirror log to %s.", display_lvname(lv)); - return 0; - } + if (!attach_mirror_log(first_seg(lv), log_lv)) + return_0; return update_and_reload ? _lv_update_reload_fns_reset_eliminate_lvs(lv, removal_lvs) : 1; }

1 0

master - mirror: add prepare_mirror_log
by Zdenek Kabelac 12 Dec '16

12 Dec '16

Gitweb: http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=31564834db7610b3… Commit: 31564834db7610b37f52040fbba89f4c71de4c34 Parent: b9f9ce52ecb0de3b33b1d0e2294a1b13c744e655 Author: Zdenek Kabelac <zkabelac(a)redhat.com> AuthorDate: Sat Dec 10 19:54:09 2016 +0100 Committer: Zdenek Kabelac <zkabelac(a)redhat.com> CommitterDate: Sun Dec 11 23:16:16 2016 +0100 mirror: add prepare_mirror_log Function prepares new mirror log LV in-sync optionaly. This is useful to have such device ready when converting raid1 to mirror. --- WHATS_NEW | 1 + lib/metadata/metadata-exported.h | 4 +++ lib/metadata/mirror.c | 43 ++++++++++++++++++++++++++++++++++++++ 3 files changed, 48 insertions(+), 0 deletions(-) diff --git a/WHATS_NEW b/WHATS_NEW index 37a0dc7..274430f 100644 --- a/WHATS_NEW +++ b/WHATS_NEW @@ -1,5 +1,6 @@ Version 2.02.169 - ===================================== + Add internal function for separate mirror log preparation. Fix segfault in lvmetad from missing NULL in daemon_reply_simple. Simplify internal _info_run() and use _setup_task_run() for mknod. Better API for internal function _setup_task_run. diff --git a/lib/metadata/metadata-exported.h b/lib/metadata/metadata-exported.h index cdd4984..4bf466b 100644 --- a/lib/metadata/metadata-exported.h +++ b/lib/metadata/metadata-exported.h @@ -1157,6 +1157,10 @@ struct logical_volume *detach_mirror_log(struct lv_segment *seg); int attach_mirror_log(struct lv_segment *seg, struct logical_volume *lv); int remove_mirror_log(struct cmd_context *cmd, struct logical_volume *lv, struct dm_list *removable_pvs, int force); +struct logical_volume *prepare_mirror_log(struct logical_volume *lv, + int in_sync, uint32_t region_size, + struct dm_list *allocatable_pvs, + alloc_policy_t alloc); int add_mirror_log(struct cmd_context *cmd, struct logical_volume *lv, uint32_t log_count, uint32_t region_size, struct dm_list *allocatable_pvs, alloc_policy_t alloc); diff --git a/lib/metadata/mirror.c b/lib/metadata/mirror.c index e84916e..73b1110 100644 --- a/lib/metadata/mirror.c +++ b/lib/metadata/mirror.c @@ -1912,6 +1912,49 @@ int attach_mirror_log(struct lv_segment *seg, struct logical_volume *log_lv) return add_seg_to_segs_using_this_lv(log_lv, seg); } +/* Prepare disk mirror log for raid1->mirror conversion */ +struct logical_volume *prepare_mirror_log(struct logical_volume *lv, + int in_sync, uint32_t region_size, + struct dm_list *allocatable_pvs, + alloc_policy_t alloc) +{ + struct cmd_context *cmd = lv->vg->cmd; + const struct segment_type *segtype; + struct dm_list *parallel_areas; + struct alloc_handle *ah; + struct logical_volume *log_lv; + + if (!(parallel_areas = build_parallel_areas_from_lv(lv, 0, 0))) + return_NULL; + + if (!(segtype = get_segtype_from_string(cmd, SEG_TYPE_NAME_MIRROR))) + return_NULL; + + /* Allocate destination extents */ + if (!(ah = allocate_extents(lv->vg, NULL, segtype, + 0, 0, 1, region_size, + lv->le_count, allocatable_pvs, + alloc, 0, parallel_areas))) { + log_error("Unable to allocate extents for mirror log."); + return NULL; + } + + if (!(log_lv = _create_mirror_log(lv, ah, alloc, lv->name, "_mlog"))) { + log_error("Failed to create mirror log."); + goto out; + } + + if (!_init_mirror_log(cmd, log_lv, in_sync, &lv->tags, 1)) { + log_error("Failed to initialise mirror log."); + log_lv = NULL; + goto out; + } +out: + alloc_destroy(ah); + + return log_lv; +} + int add_mirror_log(struct cmd_context *cmd, struct logical_volume *lv, uint32_t log_count, uint32_t region_size, struct dm_list *allocatable_pvs, alloc_policy_t alloc)

1 0

master - test: add results/ to .gitignore
by Bryn Reeves 11 Dec '16

11 Dec '16

Gitweb: http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=b9f9ce52ecb0de3b… Commit: b9f9ce52ecb0de3b33b1d0e2294a1b13c744e655 Parent: 14c20ae217ecb42e77f1775ae9185a08e74ea2d3 Author: Bryn M. Reeves <bmr(a)redhat.com> AuthorDate: Sat Dec 10 18:00:25 2016 +0000 Committer: Bryn M. Reeves <bmr(a)redhat.com> CommitterDate: Sat Dec 10 18:00:25 2016 +0000 test: add results/ to .gitignore --- test/.gitignore | 1 + 1 files changed, 1 insertions(+), 0 deletions(-) diff --git a/test/.gitignore b/test/.gitignore index 2450c66..5b377c7 100644 --- a/test/.gitignore +++ b/test/.gitignore @@ -1,3 +1,4 @@ .bin-dir-stamp bin init.sh +results/

1 0

master - lvmdbusd: add path.py to .gitignore
by Bryn Reeves 11 Dec '16

11 Dec '16

Gitweb: http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=14c20ae217ecb42e… Commit: 14c20ae217ecb42e77f1775ae9185a08e74ea2d3 Parent: f9c9772a80737e8f00f2c96dc2b033aaf85487c3 Author: Bryn M. Reeves <bmr(a)redhat.com> AuthorDate: Sat Dec 10 17:59:07 2016 +0000 Committer: Bryn M. Reeves <bmr(a)redhat.com> CommitterDate: Sat Dec 10 17:59:07 2016 +0000 lvmdbusd: add path.py to .gitignore --- daemons/lvmdbusd/.gitignore | 1 + 1 files changed, 1 insertions(+), 0 deletions(-) diff --git a/daemons/lvmdbusd/.gitignore b/daemons/lvmdbusd/.gitignore new file mode 100644 index 0000000..09375e7 --- /dev/null +++ b/daemons/lvmdbusd/.gitignore @@ -0,0 +1 @@ +path.py

1 0

master - scripts: add systemd unit files to .gitignore
by Bryn Reeves 11 Dec '16

11 Dec '16

Gitweb: http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=f9c9772a80737e8f… Commit: f9c9772a80737e8f00f2c96dc2b033aaf85487c3 Parent: 383683ffd390e1936474c3bced850ea887223478 Author: Bryn M. Reeves <bmr(a)redhat.com> AuthorDate: Sat Dec 10 17:58:10 2016 +0000 Committer: Bryn M. Reeves <bmr(a)redhat.com> CommitterDate: Sat Dec 10 17:58:10 2016 +0000 scripts: add systemd unit files to .gitignore com.redhat.lvmdbus1.service lvm2_lvmdbusd_systemd_red_hat.service .gitignore --- scripts/.gitignore | 2 ++ 1 files changed, 2 insertions(+), 0 deletions(-) diff --git a/scripts/.gitignore b/scripts/.gitignore index ed0d945..0368f23 100644 --- a/scripts/.gitignore +++ b/scripts/.gitignore @@ -3,6 +3,7 @@ blk_availability_systemd_red_hat.service blkdeactivate.sh clvmd_init_red_hat cmirrord_init_red_hat +com.redhat.lvmdbus1.service dm_event_systemd_red_hat.service dm_event_systemd_red_hat.socket lvm2_activation_generator_systemd_red_hat @@ -10,6 +11,7 @@ lvm2_cluster_activation_red_hat.sh lvm2_cluster_activation_systemd_red_hat.service lvm2_clvmd_systemd_red_hat.service lvm2_cmirrord_systemd_red_hat.service +lvm2_lvmdbusd_systemd_red_hat.service lvm2_lvmetad_init_red_hat lvm2_lvmetad_systemd_red_hat.service lvm2_lvmetad_systemd_red_hat.socket

1 0

master - scripts: add lvmdump.sh to .gitignore
by Bryn Reeves 11 Dec '16

11 Dec '16

Gitweb: http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=383683ffd390e193… Commit: 383683ffd390e1936474c3bced850ea887223478 Parent: b311122a5666489f5dbeae09952dfbcd533f5a36 Author: Bryn M. Reeves <bmr(a)redhat.com> AuthorDate: Sat Dec 10 17:56:56 2016 +0000 Committer: Bryn M. Reeves <bmr(a)redhat.com> CommitterDate: Sat Dec 10 17:56:56 2016 +0000 scripts: add lvmdump.sh to .gitignore --- scripts/.gitignore | 1 + 1 files changed, 1 insertions(+), 0 deletions(-) diff --git a/scripts/.gitignore b/scripts/.gitignore index 3a668ae..ed0d945 100644 --- a/scripts/.gitignore +++ b/scripts/.gitignore @@ -22,3 +22,4 @@ lvm2_monitoring_init_red_hat lvm2_monitoring_systemd_red_hat.service lvm2_pvscan_systemd_red_hat@.service lvm2_tmpfiles_red_hat.conf +lvmdump.sh

1 0

master - libdm: fix filemap cleanup loop condition
by Bryn Reeves 10 Dec '16

10 Dec '16

Gitweb: http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=b311122a5666489f… Commit: b311122a5666489f5dbeae09952dfbcd533f5a36 Parent: d8ba8ee9aea0b23a4ff03d19f7ddbe6511cdcbbd Author: Bryn M. Reeves <bmr(a)redhat.com> AuthorDate: Sat Dec 10 13:28:21 2016 +0000 Committer: Bryn M. Reeves <bmr(a)redhat.com> CommitterDate: Sat Dec 10 13:31:12 2016 +0000 libdm: fix filemap cleanup loop condition --- libdm/libdm-stats.c | 6 +++--- 1 files changed, 3 insertions(+), 3 deletions(-) diff --git a/libdm/libdm-stats.c b/libdm/libdm-stats.c index 78b66ab..850aa49 100644 --- a/libdm/libdm-stats.c +++ b/libdm/libdm-stats.c @@ -4323,7 +4323,7 @@ static uint64_t *_stats_create_file_regions(struct dm_stats *dms, int fd, struct dm_histogram *bounds, int precise, uint64_t *count) { - uint64_t *regions = NULL, i, max_region; + uint64_t *regions = NULL, i, fail_region; struct dm_pool *extent_mem = NULL; struct _extent *extents = NULL; char *hist_arg = NULL; @@ -4411,8 +4411,8 @@ out_remove: */ dm_stats_list(dms, NULL); - max_region = i; - for (i = max_region - 1; i < max_region; i++) + fail_region = i; + for (i = 0; i < fail_region; i++) if (!_stats_delete_region(dms, regions[i])) log_error("Could not delete region " FMTu64 ".", i);

1 0

master - libdm: use a private pool for filemap extent table
by Bryn Reeves 10 Dec '16

10 Dec '16

Gitweb: http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=d8ba8ee9aea0b23a… Commit: d8ba8ee9aea0b23a4ff03d19f7ddbe6511cdcbbd Parent: 1de3e106c96f6df7b32e072b6a1a3f924fa041bc Author: Bryn M. Reeves <bmr(a)redhat.com> AuthorDate: Sat Dec 10 13:07:03 2016 +0000 Committer: Bryn M. Reeves <bmr(a)redhat.com> CommitterDate: Sat Dec 10 13:31:12 2016 +0000 libdm: use a private pool for filemap extent table When mapping regions to a file descriptor, a temporary table of extent descriptors is built using the dm_pool object building interface. Previously this use borrowed the dms->mem region and counter table pool (since nothing can interleave with the allocation while the caller is still in dm_stats_create_regions_from_fd()). This turns out to be problematic for error recovery. When a region creation operation fails partway through file mapping, we need to roll back the set of already created regions and this requires a listed handle: the dm_stats_list() will then allocate from the same pool as the extents; we either have to throw away valid list data, or leak the extent table, to return the handle in a valid state. Avoid this problem by creating a new, temporary mem pool in _stats_create_file_regions() to hold the extent data, and discarding it on exit from the function. --- WHATS_NEW_DM | 1 + libdm/libdm-stats.c | 19 ++++++++++++++++--- 2 files changed, 17 insertions(+), 3 deletions(-) diff --git a/WHATS_NEW_DM b/WHATS_NEW_DM index 4113876..73707b6 100644 --- a/WHATS_NEW_DM +++ b/WHATS_NEW_DM @@ -1,5 +1,6 @@ Version 1.02.138 - ===================================== + Separate filemap extent allocation from region table. Fix segmentation fault when filemap region creation fails Fix performance of region cleanup for failed filemap creation. Fix very slow region deletion with many regions. diff --git a/libdm/libdm-stats.c b/libdm/libdm-stats.c index 34870e3..78b66ab 100644 --- a/libdm/libdm-stats.c +++ b/libdm/libdm-stats.c @@ -4324,6 +4324,7 @@ static uint64_t *_stats_create_file_regions(struct dm_stats *dms, int fd, int precise, uint64_t *count) { uint64_t *regions = NULL, i, max_region; + struct dm_pool *extent_mem = NULL; struct _extent *extents = NULL; char *hist_arg = NULL; struct statfs fsbuf; @@ -4357,9 +4358,19 @@ static uint64_t *_stats_create_file_regions(struct dm_stats *dms, int fd, return 0; } - if (!(extents = _stats_get_extents_for_file(dms->mem, fd, count))) + /* Use a temporary, private pool for the extent table. This avoids + * hijacking the dms->mem (region table) pool which would lead to + * interleaving temporary allocations with dm_stats_list() data, + * causing complications in the error path. + */ + if (!(extent_mem = dm_pool_create("extents", sizeof(*extents)))) return_0; + if (!(extents = _stats_get_extents_for_file(extent_mem, fd, count))) { + dm_pool_destroy(extent_mem); + return_0; + } + if (bounds) { /* _build_histogram_arg enables precise if vals < 1ms. */ if (!(hist_arg = _build_histogram_arg(bounds, &precise))) @@ -4386,7 +4397,8 @@ static uint64_t *_stats_create_file_regions(struct dm_stats *dms, int fd, if (bounds) dm_free(hist_arg); - dm_pool_free(dms->mem, extents); + dm_pool_free(extent_mem, extents); + dm_pool_destroy(extent_mem); return regions; out_remove: @@ -4407,7 +4419,8 @@ out_remove: *count = 0; out: - dm_pool_free(dms->mem, extents); + dm_pool_free(extent_mem, extents); + dm_pool_destroy(extent_mem); dm_free(hist_arg); dm_free(regions); return NULL;

1 0

master - doc: add filemap creation fixes to WHATS_NEW_DM
by Bryn Reeves 10 Dec '16

10 Dec '16

Gitweb: http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=1de3e106c96f6df7… Commit: 1de3e106c96f6df7b32e072b6a1a3f924fa041bc Parent: 2d1dbb9eddd9dbc3aedf4572926178e532899995 Author: Bryn M. Reeves <bmr(a)redhat.com> AuthorDate: Sat Dec 10 12:02:30 2016 +0000 Committer: Bryn M. Reeves <bmr(a)redhat.com> CommitterDate: Sat Dec 10 12:02:30 2016 +0000 doc: add filemap creation fixes to WHATS_NEW_DM --- WHATS_NEW_DM | 3 +++ 1 files changed, 3 insertions(+), 0 deletions(-) diff --git a/WHATS_NEW_DM b/WHATS_NEW_DM index c6a4902..4113876 100644 --- a/WHATS_NEW_DM +++ b/WHATS_NEW_DM @@ -1,5 +1,8 @@ Version 1.02.138 - ===================================== + Fix segmentation fault when filemap region creation fails + Fix performance of region cleanup for failed filemap creation. + Fix very slow region deletion with many regions. Version 1.02.137 - 30th November 2016 =====================================

1 0

master - libdm: fix performance of failed filemap cleanup
by Bryn Reeves 10 Dec '16

10 Dec '16

Gitweb: http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=2d1dbb9eddd9dbc3… Commit: 2d1dbb9eddd9dbc3aedf4572926178e532899995 Parent: 97c4490cc53157adde6da84933524f22149f449f Author: Bryn M. Reeves <bmr(a)redhat.com> AuthorDate: Fri Dec 9 23:59:51 2016 +0000 Committer: Bryn M. Reeves <bmr(a)redhat.com> CommitterDate: Sat Dec 10 11:59:16 2016 +0000 libdm: fix performance of failed filemap cleanup While cleaning up the table of already created regions during a failed dm_stats_create_regions_from_fd(), list the handle once, and call _stats_delete_region() directly. This avoids sending a @stats_list message for each region deleted, reducing runtime from 6s to 0.7s when cleaning up ~250 out of ~10000 regions: # time dmstats create --filemap b.img device-mapper: message ioctl on (253:0) failed: Cannot allocate memory Failed to create region 246 of 309 at 9388032. Could not create regions from file /root/b.img << pauses here >> Command failed real 0m6.267s user 0m3.770s sys 0m2.487s # time dmstats create --filemap b.img device-mapper: message ioctl on (253:0) failed: Cannot allocate memory Failed to create region 246 of 309 at 9388032. Could not create regions from file /root/b.img Command failed real 0m0.716s user 0m0.034s sys 0m0.581s Testing the error path requires region creation to start to fail part way through the operation (in order to have regions to clean up): the simplest way is to ensure the system is close to the kernel limit of 1/4 RAM or 1/2 vmalloc space consumed by dmstats data. --- libdm/libdevmapper.h | 21 +++++++++++++-------- libdm/libdm-stats.c | 20 +++++++++++++++----- 2 files changed, 28 insertions(+), 13 deletions(-) diff --git a/libdm/libdevmapper.h b/libdm/libdevmapper.h index 09e3fcc..fe2e410 100644 --- a/libdm/libdevmapper.h +++ b/libdm/libdevmapper.h @@ -1297,24 +1297,29 @@ int dm_stats_get_group_descriptor(const struct dm_stats *dms, * filesystem and optionally place them into a group. * * File descriptor fd must reference a regular file, open for reading, - * in a local file system that supports the FIEMAP ioctl and that + * in a local file system that supports the FIEMAP ioctl, and that * returns data describing the physical location of extents. * * The file descriptor can be closed by the caller following the call * to dm_stats_create_regions_from_fd(). * - * The function returns a pointer to an array of uint64_t containing - * the IDs of the newly created regions. The array is terminated by the - * value DM_STATS_REGIONS_ALL and should be freed using dm_free() when - * no longer required. - * * Unless nogroup is non-zero the regions will be placed into a group - * and the group alias is set to the value supplied. + * and the group alias set to the value supplied (if alias is NULL no + * group alias will be assigned). + * + * On success the function returns a pointer to an array of uint64_t + * containing the IDs of the newly created regions. The region_id + * array is terminated by the value DM_STATS_REGION_NOT_PRESENT and + * should be freed using dm_free() when no longer required. On error + * NULL is returned. + * + * Following a call to dm_stats_create_regions_from_fd() the handle + * is guaranteed to be in a listed state, and to contain any region + * and group identifiers created by the operation. * * The group_id for the new group is equal to the region_id value in * the first array element. * - * File mapped histograms will be supported in a future version. */ uint64_t *dm_stats_create_regions_from_fd(struct dm_stats *dms, int fd, int group, int precise, diff --git a/libdm/libdm-stats.c b/libdm/libdm-stats.c index 1cabd9a..34870e3 100644 --- a/libdm/libdm-stats.c +++ b/libdm/libdm-stats.c @@ -4323,8 +4323,8 @@ static uint64_t *_stats_create_file_regions(struct dm_stats *dms, int fd, struct dm_histogram *bounds, int precise, uint64_t *count) { + uint64_t *regions = NULL, i, max_region; struct _extent *extents = NULL; - uint64_t *regions = NULL, i; char *hist_arg = NULL; struct statfs fsbuf; struct stat buf; @@ -4390,11 +4390,22 @@ static uint64_t *_stats_create_file_regions(struct dm_stats *dms, int fd, return regions; out_remove: - /* clean up regions after create failure */ - for (--i; i != DM_STATS_REGION_NOT_PRESENT; i--) - if (!dm_stats_delete_region(dms, regions[i])) + /* New region creation may begin to fail part-way through creating + * a set of file mapped regions: in this case we need to roll back + * the regions that were already created and return the handle to + * a consistent state. A listed handle is required for this: use a + * single list operation and call _stats_delete_region() directly + * to avoid a @stats_list ioctl and list parsing for each region. + */ + dm_stats_list(dms, NULL); + + max_region = i; + for (i = max_region - 1; i < max_region; i++) + if (!_stats_delete_region(dms, regions[i])) log_error("Could not delete region " FMTu64 ".", i); + *count = 0; + out: dm_pool_free(dms->mem, extents); dm_free(hist_arg); @@ -4402,7 +4413,6 @@ out: return NULL; } - uint64_t *dm_stats_create_regions_from_fd(struct dm_stats *dms, int fd, int group, int precise, struct dm_histogram *bounds,

1 0

master - libdm: split off internal _stats_delete_region()
by Bryn Reeves 10 Dec '16

10 Dec '16

Gitweb: http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=97c4490cc53157ad… Commit: 97c4490cc53157adde6da84933524f22149f449f Parent: c459f23565e86eb9a8a116995cda7dcaed858b9f Author: Bryn M. Reeves <bmr(a)redhat.com> AuthorDate: Fri Dec 9 22:58:25 2016 +0000 Committer: Bryn M. Reeves <bmr(a)redhat.com> CommitterDate: Sat Dec 10 11:57:14 2016 +0000 libdm: split off internal _stats_delete_region() Split dm_stats_delete_region() so that internal callers can manage the handle state themselves. dm_stats_delete_region() now just handles checking the state of the handle, reporting validation errors, and calling dm_stats_list() if necessary, before calling _stats_delete_region(). The new _stats_delete_region() function performs the actual group member removal and region deletion, and requires a fully listed handle to operate. Callers that repeatedly delete regions can use a single listed handle for many operations on the same device, avoiding one message ioctl per region deleted: since @stats_list with many regions is expensive, this yields large runtime improvements. --- libdm/libdm-stats.c | 43 ++++++++++++++++++++++++++----------------- 1 files changed, 26 insertions(+), 17 deletions(-) diff --git a/libdm/libdm-stats.c b/libdm/libdm-stats.c index 438a1ed..1cabd9a 100644 --- a/libdm/libdm-stats.c +++ b/libdm/libdm-stats.c @@ -2018,10 +2018,34 @@ static int _stats_remove_region_id_from_group(struct dm_stats *dms, return _stats_set_aux(dms, group_id, dms->regions[group_id].aux_data); } -int dm_stats_delete_region(struct dm_stats *dms, uint64_t region_id) +static int _stats_delete_region(struct dm_stats *dms, uint64_t region_id) { char msg[STATS_MSG_BUF_LEN]; struct dm_task *dmt; + + if (_stats_region_is_grouped(dms, region_id)) + if (!_stats_remove_region_id_from_group(dms, region_id)) { + log_error("Could not remove region ID " FMTu64 " from " + "group ID " FMTu64, + region_id, dms->regions[region_id].group_id); + return 0; + } + + if (!dm_snprintf(msg, sizeof(msg), "@stats_delete " FMTu64, region_id)) { + log_error("Could not prepare @stats_delete message."); + return 0; + } + + dmt = _stats_send_message(dms, msg); + if (!dmt) + return_0; + dm_task_destroy(dmt); + + return 1; +} + +int dm_stats_delete_region(struct dm_stats *dms, uint64_t region_id) +{ int listed = 0; if (!_stats_bound(dms)) @@ -2065,23 +2089,8 @@ int dm_stats_delete_region(struct dm_stats *dms, uint64_t region_id) goto bad; } - if(_stats_region_is_grouped(dms, region_id)) - if (!_stats_remove_region_id_from_group(dms, region_id)) { - log_error("Could not remove region ID " FMTu64 " from " - "group ID " FMTu64, - region_id, dms->regions[region_id].group_id); - goto bad; - } - - if (!dm_snprintf(msg, sizeof(msg), "@stats_delete " FMTu64, region_id)) { - log_error("Could not prepare @stats_delete message."); + if (!_stats_delete_region(dms, region_id)) goto bad; - } - - dmt = _stats_send_message(dms, msg); - if (!dmt) - return_0; - dm_task_destroy(dmt); if (!listed) /* wipe region and mark as not present */

1 0

master - lvmetad: fix segfault in daemon_reply_simple
by David Teigland 10 Dec '16

10 Dec '16

Gitweb: http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=c459f23565e86eb9… Commit: c459f23565e86eb9a8a116995cda7dcaed858b9f Parent: 30ad254d84047ffe3eba98943855b53a893390e2 Author: David Teigland <teigland(a)redhat.com> AuthorDate: Fri Dec 9 15:17:53 2016 -0600 Committer: David Teigland <teigland(a)redhat.com> CommitterDate: Fri Dec 9 15:22:30 2016 -0600 lvmetad: fix segfault in daemon_reply_simple missing NULL termination --- WHATS_NEW | 1 + daemons/lvmetad/lvmetad-core.c | 9 ++++++--- 2 files changed, 7 insertions(+), 3 deletions(-) diff --git a/WHATS_NEW b/WHATS_NEW index 595b667..37a0dc7 100644 --- a/WHATS_NEW +++ b/WHATS_NEW @@ -1,5 +1,6 @@ Version 2.02.169 - ===================================== + Fix segfault in lvmetad from missing NULL in daemon_reply_simple. Simplify internal _info_run() and use _setup_task_run() for mknod. Better API for internal function _setup_task_run. Avoid using lv_has_target_type() call within lv_info_with_seg_status. diff --git a/daemons/lvmetad/lvmetad-core.c b/daemons/lvmetad/lvmetad-core.c index ee0c52f..1faae11 100644 --- a/daemons/lvmetad/lvmetad-core.c +++ b/daemons/lvmetad/lvmetad-core.c @@ -2745,7 +2745,8 @@ static response handler(daemon_state s, client_handle h, request r) "expected = %s", state->token, "received = %s", token, "update_pid = " FMTd64, (int64_t)state->update_pid, - "reason = %s", "another command has populated the cache"); + "reason = %s", "another command has populated the cache", + NULL); } DEBUGLOG(state, "token_update end len %d pid %d new token %s", @@ -2778,7 +2779,8 @@ static response handler(daemon_state s, client_handle h, request r) "expected = %s", state->token, "received = %s", token, "update_pid = " FMTd64, (int64_t)state->update_pid, - "reason = %s", "another command has populated the cache"); + "reason = %s", "another command has populated the cache", + NULL); } /* If a pid doing update was cancelled, ignore its update messages. */ @@ -2793,7 +2795,8 @@ static response handler(daemon_state s, client_handle h, request r) "expected = %s", state->token, "received = %s", token, "update_pid = " FMTd64, (int64_t)state->update_pid, - "reason = %s", "another command has populated the lvmetad cache"); + "reason = %s", "another command has populated the lvmetad cache", + NULL); } pthread_mutex_unlock(&state->token_lock);

1 0

master - libdm: use correct region_id when cleaning up a failed filemap
by Bryn Reeves 10 Dec '16

10 Dec '16

Gitweb: http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=30ad254d84047ffe… Commit: 30ad254d84047ffe3eba98943855b53a893390e2 Parent: 7fd2fa22ddb3ed97283305100567cc4b5122ddb3 Author: Bryn M. Reeves <bmr(a)redhat.com> AuthorDate: Fri Dec 9 15:50:41 2016 +0000 Committer: Bryn M. Reeves <bmr(a)redhat.com> CommitterDate: Fri Dec 9 16:04:13 2016 +0000 libdm: use correct region_id when cleaning up a failed filemap If we fail to create a region during dm_stats_create_regions_from_fd(), we must remove all regions that were created to do this to date. This needs to loop over the table of region_id values that were populated by _stats_create_file_regions() before the error. The code for this failure case in the out_remove branch incorrectly uses the table index as the region_id: for (--i; i != DM_STATS_REGION_NOT_PRESENT; i--) { if (!dm_stats_delete_region(dms, i)) log_error("Could not delete region " FMTu64 ".", i); } This causes the cleanup code to delete a completely unrelated set of regions (since the index here will always be nr_regions..0). Fix it to pass the actual region_id stored in regions[i] instead. --- libdm/libdm-stats.c | 5 ++--- 1 files changed, 2 insertions(+), 3 deletions(-) diff --git a/libdm/libdm-stats.c b/libdm/libdm-stats.c index 3900433..438a1ed 100644 --- a/libdm/libdm-stats.c +++ b/libdm/libdm-stats.c @@ -4382,10 +4382,9 @@ static uint64_t *_stats_create_file_regions(struct dm_stats *dms, int fd, out_remove: /* clean up regions after create failure */ - for (--i; i != DM_STATS_REGION_NOT_PRESENT; i--) { - if (!dm_stats_delete_region(dms, i)) + for (--i; i != DM_STATS_REGION_NOT_PRESENT; i--) + if (!dm_stats_delete_region(dms, regions[i])) log_error("Could not delete region " FMTu64 ".", i); - } out: dm_pool_free(dms->mem, extents);

1 0

master - libdm-stats: clear dms->groups in _stats_groups_destroy()
by Bryn Reeves 10 Dec '16

10 Dec '16

Gitweb: http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=7fd2fa22ddb3ed97… Commit: 7fd2fa22ddb3ed97283305100567cc4b5122ddb3 Parent: cb8c04760f73b0376a909440da9c20f78694ea4e Author: Bryn M. Reeves <bmr(a)redhat.com> AuthorDate: Fri Dec 9 15:49:06 2016 +0000 Committer: Bryn M. Reeves <bmr(a)redhat.com> CommitterDate: Fri Dec 9 16:04:13 2016 +0000 libdm-stats: clear dms->groups in _stats_groups_destroy() --- libdm/libdm-stats.c | 1 + 1 files changed, 1 insertions(+), 0 deletions(-) diff --git a/libdm/libdm-stats.c b/libdm/libdm-stats.c index 41adf42..3900433 100644 --- a/libdm/libdm-stats.c +++ b/libdm/libdm-stats.c @@ -378,6 +378,7 @@ static void _stats_groups_destroy(struct dm_stats *dms) for (i = dms->max_region; (i != DM_STATS_REGION_NOT_PRESENT); i--) _stats_group_destroy(&dms->groups[i]); dm_pool_free(dms->group_mem, dms->groups); + dms->groups = NULL; } static int _set_stats_device(struct dm_stats *dms, struct dm_task *dmt)

1 0

master - libdm-stats: clear dms->regions in _stats_regions_destroy()
by Bryn Reeves 10 Dec '16

10 Dec '16

Gitweb: http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=cb8c04760f73b037… Commit: cb8c04760f73b0376a909440da9c20f78694ea4e Parent: 900d203586faf1fef447002aa2e784cfc9575ad1 Author: Bryn M. Reeves <bmr(a)redhat.com> AuthorDate: Fri Dec 9 15:47:55 2016 +0000 Committer: Bryn M. Reeves <bmr(a)redhat.com> CommitterDate: Fri Dec 9 16:04:13 2016 +0000 libdm-stats: clear dms->regions in _stats_regions_destroy() --- libdm/libdm-stats.c | 1 + 1 files changed, 1 insertions(+), 0 deletions(-) diff --git a/libdm/libdm-stats.c b/libdm/libdm-stats.c index 411fd80..41adf42 100644 --- a/libdm/libdm-stats.c +++ b/libdm/libdm-stats.c @@ -347,6 +347,7 @@ static void _stats_regions_destroy(struct dm_stats *dms) } dm_pool_free(mem, dms->regions); + dms->regions = NULL; } static void _stats_group_destroy(struct dm_stats_group *group)

1 0

master - tests: fix missing exclusive activation
by Zdenek Kabelac 09 Dec '16

09 Dec '16

Gitweb: http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=900d203586faf1fe… Commit: 900d203586faf1fef447002aa2e784cfc9575ad1 Parent: c5aeb210155eff180ad5f1127ed451307bd7b854 Author: Zdenek Kabelac <zkabelac(a)redhat.com> AuthorDate: Thu Dec 8 16:27:24 2016 +0100 Committer: Zdenek Kabelac <zkabelac(a)redhat.com> CommitterDate: Fri Dec 9 15:15:02 2016 +0100 tests: fix missing exclusive activation For cluster conversion LV for caching needs to be activated exclusively. --- test/shell/lvs-cache.sh | 2 +- 1 files changed, 1 insertions(+), 1 deletions(-) diff --git a/test/shell/lvs-cache.sh b/test/shell/lvs-cache.sh index 2c518d0..74077ae 100644 --- a/test/shell/lvs-cache.sh +++ b/test/shell/lvs-cache.sh @@ -23,7 +23,7 @@ aux have_cache 1 3 0 || skip aux prepare_vg 5 8000 # Use 10M origin size -lvcreate -L10 -n $lv1 $vg +lvcreate -aey -L10 -n $lv1 $vg lvcreate -H -L5 $vg/$lv1 # replace 10M size with 5M size of cache device

1 0

master - cleanup: zero baton in struct initilizer
by Zdenek Kabelac 09 Dec '16

09 Dec '16

Gitweb: http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=c5aeb210155eff18… Commit: c5aeb210155eff180ad5f1127ed451307bd7b854 Parent: f9c6c115d3cfd203e99561d500293dbfc8662f49 Author: Zdenek Kabelac <zkabelac(a)redhat.com> AuthorDate: Fri Dec 9 15:08:04 2016 +0100 Committer: Zdenek Kabelac <zkabelac(a)redhat.com> CommitterDate: Fri Dec 9 15:15:02 2016 +0100 cleanup: zero baton in struct initilizer --- lib/metadata/metadata.c | 3 +-- 1 files changed, 1 insertions(+), 2 deletions(-) diff --git a/lib/metadata/metadata.c b/lib/metadata/metadata.c index ad6cb2a..c636fe2 100644 --- a/lib/metadata/metadata.c +++ b/lib/metadata/metadata.c @@ -2544,7 +2544,7 @@ static int _lv_mark_if_partial_collect(struct logical_volume *lv, void *data) static int _lv_mark_if_partial_single(struct logical_volume *lv, void *data) { unsigned s; - struct _lv_mark_if_partial_baton baton; + struct _lv_mark_if_partial_baton baton = { .partial = 0 }; struct lv_segment *lvseg; dm_list_iterate_items(lvseg, &lv->segments) { @@ -2556,7 +2556,6 @@ static int _lv_mark_if_partial_single(struct logical_volume *lv, void *data) } } - baton.partial = 0; if (!_lv_each_dependency(lv, _lv_mark_if_partial_collect, &baton)) return_0;

1 0

master - cleanup: easier code for raid plugin
by Zdenek Kabelac 09 Dec '16

09 Dec '16

Gitweb: http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=f9c6c115d3cfd203… Commit: f9c6c115d3cfd203e99561d500293dbfc8662f49 Parent: 15e4ab3e9375b561880fa83f0a5a3c438e192bcc Author: Zdenek Kabelac <zkabelac(a)redhat.com> AuthorDate: Fri Dec 9 13:58:19 2016 +0100 Committer: Zdenek Kabelac <zkabelac(a)redhat.com> CommitterDate: Fri Dec 9 15:15:02 2016 +0100 cleanup: easier code for raid plugin Set bits only when then were not yet assigned. --- daemons/dmeventd/plugins/raid/dmeventd_raid.c | 17 ++++++++++------- 1 files changed, 10 insertions(+), 7 deletions(-) diff --git a/daemons/dmeventd/plugins/raid/dmeventd_raid.c b/daemons/dmeventd/plugins/raid/dmeventd_raid.c index a207c99..911ae7a 100644 --- a/daemons/dmeventd/plugins/raid/dmeventd_raid.c +++ b/daemons/dmeventd/plugins/raid/dmeventd_raid.c @@ -38,6 +38,7 @@ static int _process_raid_event(struct dso_state *state, char *params, const char struct dm_status_raid *status; const char *d; int dead = 0, r = 1; + uint32_t dev; if (!dm_get_status_raid(state->mem, params, &status)) { log_error("Failed to process status line for %s.", device); @@ -46,24 +47,26 @@ static int _process_raid_event(struct dso_state *state, char *params, const char d = status->dev_health; while ((d = strchr(d, 'D'))) { - uint32_t dev = (uint32_t)(d - status->dev_health); + dev = (uint32_t)(d - status->dev_health); - if (!(state->raid_devs[dev / 64] & (UINT64_C(1) << (dev % 64)))) - log_error("Device #%u of %s array, %s, has failed.", - dev, status->raid_type, device); + if (!(state->raid_devs[dev / 64] & (UINT64_C(1) << (dev % 64)))) { + state->raid_devs[dev / 64] |= (UINT64_C(1) << (dev % 64)); + log_warn("WARNING: Device #%u of %s array, %s, has failed.", + dev, status->raid_type, device); + } - state->raid_devs[dev / 64] |= (UINT64_C(1) << (dev % 64)); d++; dead = 1; } if (dead) { if (status->insync_regions < status->total_regions) { - if (!state->warned) + if (!state->warned) { + state->warned = 1; log_warn("WARNING: waiting for resynchronization to finish " "before initiating repair on RAID device %s.", device); + } - state->warned = 1; goto out; /* Not yet done syncing with accessible devices */ }

1 0

master - cleanup: messages in raid
by Zdenek Kabelac 09 Dec '16

09 Dec '16

Gitweb: http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=15e4ab3e9375b561… Commit: 15e4ab3e9375b561880fa83f0a5a3c438e192bcc Parent: 6dd0bd0255d585feca61000432d2de9552ede4e0 Author: Zdenek Kabelac <zkabelac(a)redhat.com> AuthorDate: Fri Dec 9 14:27:45 2016 +0100 Committer: Zdenek Kabelac <zkabelac(a)redhat.com> CommitterDate: Fri Dec 9 15:15:02 2016 +0100 cleanup: messages in raid Use display_lvname and add 'dots'. Add some missing WARNING and log_debug_metadata. --- lib/metadata/raid_manip.c | 398 ++++++++++++++++++++++++--------------------- 1 files changed, 215 insertions(+), 183 deletions(-) diff --git a/lib/metadata/raid_manip.c b/lib/metadata/raid_manip.c index 7e591e9..7c4ce61 100644 --- a/lib/metadata/raid_manip.c +++ b/lib/metadata/raid_manip.c @@ -64,7 +64,8 @@ static void _ensure_min_region_size(const struct logical_volume *lv) region_size *= 2; if (seg->region_size != region_size) { - log_very_verbose("Setting region_size to %u for %s", seg->region_size, display_lvname(lv)); + log_very_verbose("Setting region_size to %u for %s.", + seg->region_size, display_lvname(lv)); seg->region_size = region_size; } } @@ -77,7 +78,7 @@ static void _ensure_min_region_size(const struct logical_volume *lv) static int _check_max_raid_devices(uint32_t image_count) { if (image_count > DEFAULT_RAID_MAX_IMAGES) { - log_error("Unable to handle raid arrays with more than %u devices", + log_error("Unable to handle raid arrays with more than %u devices.", DEFAULT_RAID_MAX_IMAGES); return 0; } @@ -88,7 +89,7 @@ static int _check_max_raid_devices(uint32_t image_count) static int _check_max_mirror_devices(uint32_t image_count) { if (image_count > DEFAULT_MIRROR_MAX_IMAGES) { - log_error("Unable to handle mirrors with more than %u devices", + log_error("Unable to handle mirrors with more than %u devices.", DEFAULT_MIRROR_MAX_IMAGES); return 0; } @@ -335,15 +336,15 @@ static int _raid_remove_top_layer(struct logical_volume *lv, if (!seg_is_mirrored(seg)) { log_error(INTERNAL_ERROR - "Unable to remove RAID layer from segment type %s", + "Unable to remove RAID layer from segment type %s.", lvseg_name(seg)); return 0; } if (seg->area_count != 1) { log_error(INTERNAL_ERROR - "Unable to remove RAID layer when there" - " is more than one sub-lv"); + "Unable to remove RAID layer when there is " + "more than one sub-lv."); return 0; } @@ -392,26 +393,28 @@ static int _clear_lv(struct logical_volume *lv) lv->status |= LV_TEMPORARY; if (!was_active && !activate_lv_local(lv->vg->cmd, lv)) { - log_error("Failed to activate localy %s for clearing", - lv->name); + log_error("Failed to activate localy %s for clearing.", + display_lvname(lv)); return 0; } lv->status &= ~LV_TEMPORARY; - log_verbose("Clearing metadata area of %s/%s", - lv->vg->name, lv->name); + log_verbose("Clearing metadata area of %s", + display_lvname(lv)); /* * Rather than wiping lv->size, we can simply * wipe the first sector to remove the superblock of any previous * RAID devices. It is much quicker. */ if (!wipe_lv(lv, (struct wipe_params) { .do_zero = 1, .zero_sectors = 1 })) { - log_error("Failed to zero %s", lv->name); + log_error("Failed to zero %s.", + display_lvname(lv)); return 0; } if (!was_active && !deactivate_lv(lv->vg->cmd, lv)) { - log_error("Failed to deactivate %s", lv->name); + log_error("Failed to deactivate %s.", + display_lvname(lv)); return 0; } @@ -425,14 +428,14 @@ static int _clear_lvs(struct dm_list *lv_list) struct volume_group *vg = NULL; if (dm_list_empty(lv_list)) { - log_debug_metadata(INTERNAL_ERROR "Empty list of LVs given for clearing"); + log_debug_metadata(INTERNAL_ERROR "Empty list of LVs given for clearing."); return 1; } dm_list_iterate_items(lvl, lv_list) { if (!lv_is_visible(lvl->lv)) { log_error(INTERNAL_ERROR - "LVs must be set visible before clearing"); + "LVs must be set visible before clearing."); return 0; } vg = lvl->lv->vg; @@ -484,17 +487,17 @@ static int _shift_and_rename_image_components(struct lv_segment *seg) * FIXME: Handling more would mean I'd have * to handle double digits */ - log_error("Unable handle arrays with more than 10 devices"); + log_error("Unable handle arrays with more than 10 devices."); return 0; } - log_very_verbose("Shifting images in %s", seg->lv->name); + log_very_verbose("Shifting images in %s.", display_lvname(seg->lv)); for (s = 0, missing = 0; s < seg->area_count; s++) { if (seg_type(seg, s) == AREA_UNASSIGNED) { if (seg_metatype(seg, s) != AREA_UNASSIGNED) { log_error(INTERNAL_ERROR "Metadata segment area" - " #%d should be AREA_UNASSIGNED", s); + " #%d should be AREA_UNASSIGNED.", s); return 0; } missing++; @@ -503,9 +506,9 @@ static int _shift_and_rename_image_components(struct lv_segment *seg) if (!missing) continue; - log_very_verbose("Shifting %s and %s by %u", - seg_metalv(seg, s)->name, - seg_lv(seg, s)->name, missing); + log_very_verbose("Shifting %s and %s by %u.", + display_lvname(seg_metalv(seg, s)), + display_lvname(seg_lv(seg, s)), missing); /* Alter rmeta name */ shift_name = dm_pool_strdup(cmd->mem, seg_metalv(seg, s)->name); @@ -595,7 +598,7 @@ static struct logical_volume *_alloc_image_component(struct logical_volume *lv, if (dm_snprintf(img_name, sizeof(img_name), "%s_%s_%%d", (alt_base_name) ? : lv->name, type_suffix) < 0) { - log_error("Component name for raid %s is too long.", lv->name); + log_error("Component name for raid %s is too long.", display_lvname(lv)); return 0; } @@ -610,7 +613,7 @@ static struct logical_volume *_alloc_image_component(struct logical_volume *lv, return_0; if (!lv_add_segment(ah, first_area, 1, tmp_lv, segtype, 0, status, 0)) { - log_error("Failed to add segment to LV, %s", img_name); + log_error("Failed to add segment to LV, %s.", img_name); return 0; } } @@ -738,15 +741,15 @@ static int _alloc_rmeta_for_lv(struct logical_volume *data_lv, allocate_pvs = &allocatable_pvs; if (!get_pv_list_for_lv(data_lv->vg->cmd->mem, data_lv, &allocatable_pvs)) { - log_error("Failed to build list of PVs for %s/%s", - data_lv->vg->name, data_lv->name); + log_error("Failed to build list of PVs for %s.", + display_lvname(data_lv)); return 0; } } if (!seg_is_linear(seg)) { log_error(INTERNAL_ERROR "Unable to allocate RAID metadata " - "area for non-linear LV, %s", data_lv->name); + "area for non-linear LV %s.", display_lvname(data_lv)); return 0; } @@ -789,13 +792,15 @@ static int _raid_add_images_without_commit(struct logical_volume *lv, } if (!_raid_in_sync(lv)) { - log_error("Can't add image to RAID LV that" - " is still initializing."); + log_error("Can't add image to RAID LV that is still initializing."); return 0; } - if (lv_is_active(lv_lock_holder(lv)) && (old_count == 1) && (lv_is_thin_pool_data(lv) || lv_is_thin_pool_metadata(lv))) { - log_error("Can't add image to active thin pool LV %s yet. Deactivate first.", display_lvname(lv)); + if (lv_is_active(lv_lock_holder(lv)) && + (old_count == 1) && + (lv_is_thin_pool_data(lv) || lv_is_thin_pool_metadata(lv))) { + log_error("Can't add image to active thin pool LV %s yet. Deactivate first.", + display_lvname(lv)); return 0; } @@ -814,7 +819,7 @@ static int _raid_add_images_without_commit(struct logical_volume *lv, status_mask = ~(LV_REBUILD); if (!(lvl = dm_pool_alloc(lv->vg->vgmem, sizeof(*lvl)))) { - log_error("Memory allocation failed"); + log_error("Memory allocation failed."); return 0; } @@ -823,8 +828,8 @@ static int _raid_add_images_without_commit(struct logical_volume *lv, dm_list_add(&meta_lvs, &lvl->list); } else if (!seg_is_raid(seg)) { - log_error("Unable to add RAID images to %s of segment type %s", - lv->name, lvseg_name(seg)); + log_error("Unable to add RAID images to %s of segment type %s.", + display_lvname(lv), lvseg_name(seg)); return 0; } @@ -873,7 +878,7 @@ static int _raid_add_images_without_commit(struct logical_volume *lv, /* MD's bitmap is limited to tracking 2^21 regions */ while (seg->region_size < (lv->size / (1 << 21))) { seg->region_size *= 2; - log_very_verbose("Setting RAID1 region_size to %uS", + log_very_verbose("Setting RAID1 region_size to %uS.", seg->region_size); } if (!(seg->segtype = get_segtype_from_string(lv->vg->cmd, SEG_TYPE_NAME_RAID1))) @@ -916,14 +921,16 @@ to be left for these sub-lvs. /* Set segment areas for metadata sub_lvs */ dm_list_iterate_items(lvl, &meta_lvs) { - log_debug_metadata("Adding %s to %s", - lvl->lv->name, lv->name); + log_debug_metadata("Adding %s to %s.", + display_lvname(lvl->lv), + display_lvname(lv)); lvl->lv->status &= status_mask; first_seg(lvl->lv)->status &= status_mask; if (!set_lv_segment_area_lv(seg, s, lvl->lv, 0, lvl->lv->status)) { - log_error("Failed to add %s to %s", - lvl->lv->name, lv->name); + log_error("Failed to add %s to %s.", + display_lvname(lvl->lv), + display_lvname(lv)); goto fail; } s++; @@ -933,14 +940,16 @@ to be left for these sub-lvs. /* Set segment areas for data sub_lvs */ dm_list_iterate_items(lvl, &data_lvs) { - log_debug_metadata("Adding %s to %s", - lvl->lv->name, lv->name); + log_debug_metadata("Adding %s to %s.", + display_lvname(lvl->lv), + display_lvname(lv)); lvl->lv->status &= status_mask; first_seg(lvl->lv)->status &= status_mask; if (!set_lv_segment_area_lv(seg, s, lvl->lv, 0, lvl->lv->status)) { - log_error("Failed to add %s to %s", - lvl->lv->name, lv->name); + log_error("Failed to add %s to %s.", + display_lvname(lvl->lv), + display_lvname(lv)); goto fail; } s++; @@ -1002,8 +1011,8 @@ static int _raid_add_images(struct logical_volume *lv, } if (rebuild_flag_cleared) { if (!vg_write(lv->vg) || !vg_commit(lv->vg)) { - log_error("Failed to clear REBUILD flag for %s/%s components", - lv->vg->name, lv->name); + log_error("Failed to clear REBUILD flag for %s components.", + display_lvname(lv)); return 0; } backup(lv->vg); @@ -1099,9 +1108,9 @@ static int _raid_extract_images(struct logical_volume *lv, struct segment_type *error_segtype; extract = seg->area_count - new_count; - log_verbose("Extracting %u %s from %s/%s", extract, + log_verbose("Extracting %u %s from %s.", extract, (extract > 1) ? "images" : "image", - lv->vg->name, lv->name); + display_lvname(lv)); if ((int) dm_list_size(target_pvs) < extract) { log_error("Unable to remove %d images: Only %d device%s given.", extract, dm_list_size(target_pvs), @@ -1146,7 +1155,7 @@ static int _raid_extract_images(struct logical_volume *lv, " the command."); return 0; } - log_debug("LVs with error segments to be removed: %s %s", + log_debug("LVs with error segments to be removed: %s %s.", display_lvname(seg_metalv(seg, s)), display_lvname(seg_lv(seg, s))); } else { @@ -1162,7 +1171,7 @@ static int _raid_extract_images(struct logical_volume *lv, if (!_raid_devs_sync_healthy(lv) && (!seg_is_mirrored(seg) || (s == 0 && !force))) { log_error("Unable to extract %sRAID image" - " while RAID array is not in-sync%s", + " while RAID array is not in-sync%s.", seg_is_mirrored(seg) ? "primary " : "", seg_is_mirrored(seg) ? " (use --force option to replace)" : ""); return 0; @@ -1175,7 +1184,7 @@ static int _raid_extract_images(struct logical_volume *lv, } if (shift && !_shift_and_rename_image_components(seg)) { - log_error("Failed to shift and rename image components"); + log_error("Failed to shift and rename image components."); return 0; } @@ -1187,7 +1196,7 @@ static int _raid_extract_images(struct logical_volume *lv, extract--; } if (extract) { - log_error("Unable to extract enough images to satisfy request"); + log_error("Unable to extract enough images to satisfy request."); return 0; } @@ -1211,16 +1220,16 @@ static int _raid_remove_images(struct logical_volume *lv, if (!_raid_extract_images(lv, 0, new_count, allocate_pvs, 1, removal_lvs, removal_lvs)) { - log_error("Failed to extract images from %s/%s", - lv->vg->name, lv->name); + log_error("Failed to extract images from %s.", + display_lvname(lv)); return 0; } /* Convert to linear? */ if (new_count == 1) { if (!_raid_remove_top_layer(lv, removal_lvs)) { - log_error("Failed to remove RAID layer" - " after linear conversion"); + log_error("Failed to remove RAID layer " + "after linear conversion."); return 0; } lv->status &= ~(LV_NOTSYNCED | LV_WRITEMOSTLY); @@ -1231,21 +1240,21 @@ static int _raid_remove_images(struct logical_volume *lv, return 1; if (!vg_write(lv->vg)) { - log_error("Failed to write changes to %s in %s", - lv->name, lv->vg->name); + log_error("Failed to write changes for %s.", + display_lvname(lv)); return 0; } if (!suspend_lv(lv->vg->cmd, lv)) { - log_error("Failed to suspend %s/%s before committing changes", - lv->vg->name, lv->name); + log_error("Failed to suspend %s before committing changes.", + display_lvname(lv)); vg_revert(lv->vg); return 0; } if (!vg_commit(lv->vg)) { - log_error("Failed to commit changes to %s in %s", - lv->name, lv->vg->name); + log_error("Failed to commit changes for %s.", + display_lvname(lv)); return 0; } @@ -1256,14 +1265,14 @@ static int _raid_remove_images(struct logical_volume *lv, */ dm_list_iterate_items(lvl, removal_lvs) { if (!activate_lv_excl_local(lv->vg->cmd, lvl->lv)) { - log_error("Failed to resume extracted LVs"); + log_error("Failed to resume extracted LVs."); return 0; } } if (!resume_lv(lv->vg->cmd, lv)) { - log_error("Failed to resume %s/%s after committing changes", - lv->vg->name, lv->name); + log_error("Failed to resume %s after committing changes.", + display_lvname(lv)); return 0; } @@ -1307,8 +1316,8 @@ static int _lv_raid_change_image_count(struct logical_volume *lv, uint32_t new_c uint32_t old_count = lv_raid_image_count(lv); if (old_count == new_count) { - log_warn("%s/%s already has image count of %d.", - lv->vg->name, lv->name, new_count); + log_warn("WARGNING: %s already has image count of %d.", + display_lvname(lv), new_count); return 1; } @@ -1317,8 +1326,8 @@ static int _lv_raid_change_image_count(struct logical_volume *lv, uint32_t new_c */ if (lv_is_active(lv_lock_holder(lv)) && vg_is_clustered(lv->vg) && !lv_is_active_exclusive_locally(lv_lock_holder(lv))) { - log_error("%s/%s must be active exclusive locally to" - " perform this operation.", lv->vg->name, lv->name); + log_error("%s must be active exclusive locally to " + "perform this operation.", display_lvname(lv)); return 0; } @@ -1349,33 +1358,33 @@ int lv_raid_split(struct logical_volume *lv, const char *split_name, dm_list_init(&data_list); if (is_lockd_type(lv->vg->lock_type)) { - log_error("Splitting raid image is not allowed with lock_type %s", + log_error("Splitting raid image is not allowed with lock_type %s.", lv->vg->lock_type); return 0; } if ((old_count - new_count) != 1) { - log_error("Unable to split more than one image from %s/%s", - lv->vg->name, lv->name); + log_error("Unable to split more than one image from %s.", + display_lvname(lv)); return 0; } if (!seg_is_mirrored(first_seg(lv)) || seg_is_raid10(first_seg(lv))) { - log_error("Unable to split logical volume of segment type, %s", + log_error("Unable to split logical volume of segment type, %s.", lvseg_name(first_seg(lv))); return 0; } if (lv_name_is_used_in_vg(lv->vg, split_name, &historical)) { - log_error("%sLogical Volume \"%s\" already exists in %s", + log_error("%sLogical Volume \"%s\" already exists in %s.", historical ? "historical " : "", split_name, lv->vg->name); return 0; } if (!_raid_in_sync(lv)) { - log_error("Unable to split %s/%s while it is not in-sync.", - lv->vg->name, lv->name); + log_error("Unable to split %s while it is not in-sync.", + display_lvname(lv)); return 0; } @@ -1386,8 +1395,8 @@ int lv_raid_split(struct logical_volume *lv, const char *split_name, if (_lv_is_raid_with_tracking(lv, &tracking)) { if (!lv_is_on_pvs(tracking, splittable_pvs)) { log_error("Unable to split additional image from %s " - "while tracking changes for %s", - lv->name, tracking->name); + "while tracking changes for %s.", + display_lvname(lv), tracking->name); return 0; } @@ -1401,14 +1410,14 @@ int lv_raid_split(struct logical_volume *lv, const char *split_name, if (!_raid_extract_images(lv, 0, new_count, splittable_pvs, 1, &removal_lvs, &data_list)) { - log_error("Failed to extract images from %s/%s", - lv->vg->name, lv->name); + log_error("Failed to extract images from %s.", + display_lvname(lv)); return 0; } /* Convert to linear? */ if ((new_count == 1) && !_raid_remove_top_layer(lv, &removal_lvs)) { - log_error("Failed to remove RAID layer after linear conversion"); + log_error("Failed to remove RAID layer after linear conversion."); return 0; } @@ -1419,21 +1428,21 @@ int lv_raid_split(struct logical_volume *lv, const char *split_name, lvl->lv->name = split_name; if (!vg_write(lv->vg)) { - log_error("Failed to write changes to %s in %s", - lv->name, lv->vg->name); + log_error("Failed to write changes for %s.", + display_lvname(lv)); return 0; } if (!suspend_lv(cmd, lv_lock_holder(lv))) { - log_error("Failed to suspend %s/%s before committing changes", - lv->vg->name, lv->name); + log_error("Failed to suspend %s before committing changes.", + display_lvname(lv_lock_holder(lv))); vg_revert(lv->vg); return 0; } if (!vg_commit(lv->vg)) { - log_error("Failed to commit changes to %s in %s", - lv->name, lv->vg->name); + log_error("Failed to commit changes for %s.", + display_lvname(lv)); return 0; } @@ -1451,8 +1460,8 @@ int lv_raid_split(struct logical_volume *lv, const char *split_name, return_0; if (!resume_lv(cmd, lv_lock_holder(lv))) { - log_error("Failed to resume %s/%s after committing changes", - lv->vg->name, lv->name); + log_error("Failed to resume %s after committing changes.", + display_lvname(lv)); return 0; } @@ -1502,19 +1511,19 @@ int lv_raid_split_and_track(struct logical_volume *lv, struct lv_segment *seg = first_seg(lv); if (!seg_is_mirrored(seg)) { - log_error("Unable to split images from non-mirrored RAID"); + log_error("Unable to split images from non-mirrored RAID."); return 0; } if (!_raid_in_sync(lv)) { - log_error("Unable to split image from %s/%s while not in-sync", - lv->vg->name, lv->name); + log_error("Unable to split image from %s while not in-sync.", + display_lvname(lv)); return 0; } /* Cannot track two split images at once */ if (lv_is_raid_with_tracking(lv)) { - log_error("Cannot track more than one split image at a time"); + log_error("Cannot track more than one split image at a time."); return 0; } @@ -1535,14 +1544,16 @@ int lv_raid_split_and_track(struct logical_volume *lv, return_0; log_print_unless_silent("%s split from %s for read-only purposes.", - seg_lv(seg, s)->name, lv->name); + display_lvname(seg_lv(seg, s)), + display_lvname(lv)); /* Activate the split (and tracking) LV */ if (!_activate_sublv_preserving_excl(lv, seg_lv(seg, s))) return_0; - log_print_unless_silent("Use 'lvconvert --merge %s/%s' to merge back into %s", - lv->vg->name, seg_lv(seg, s)->name, lv->name); + log_print_unless_silent("Use 'lvconvert --merge %s' to merge back into %s.", + display_lvname(seg_lv(seg, s)), + display_lvname(lv)); return 1; } @@ -1609,8 +1620,9 @@ int lv_raid_merge(struct logical_volume *image_lv) if (!lv_update_and_reload(lv)) return_0; - log_print_unless_silent("%s/%s successfully merged back into %s/%s", - vg->name, image_lv->name, vg->name, lv->name); + log_print_unless_silent("%s successfully merged back into %s.", + display_lvname(image_lv), + display_lvname(lv)); return 1; } @@ -1632,7 +1644,8 @@ static int _alloc_rmeta_devs_for_rimage_devs(struct logical_volume *lv, return_0; dm_list_iterate_items(lvl, new_data_lvs) { - log_debug_metadata("Allocating new metadata LV for %s", lvl->lv->name); + log_debug_metadata("Allocating new metadata LV for %s.", + display_lvname(lvl->lv)); /* * Try to collocate with DataLV first and @@ -1645,8 +1658,8 @@ static int _alloc_rmeta_devs_for_rimage_devs(struct logical_volume *lv, return_0; if (!_alloc_rmeta_for_lv(lvl->lv, &lvl_array[a].lv, allocate_pvs)) { - log_error("Failed to allocate metadata LV for %s in %s", - lvl->lv->name, lv->vg->name); + log_error("Failed to allocate metadata LV for %s.", + display_lvname(lvl->lv)); return 0; } } @@ -1685,7 +1698,8 @@ static int _add_image_component_list(struct lv_segment *seg, int delete_from_lis lvl->lv->status &= ~LV_REBUILD; if (!set_lv_segment_area_lv(seg, s++, lvl->lv, 0 /* le */, lvl->lv->status)) { - log_error("Failed to add sublv %s", lvl->lv->name); + log_error("Failed to add sublv %s.", + display_lvname(lvl->lv)); return 0; } } @@ -1756,7 +1770,8 @@ static int _extract_image_component_error_seg(struct lv_segment *seg, return 0; } - log_very_verbose("Extracting image component %s from %s", lv->name, lvseg_name(seg)); + log_very_verbose("Extracting image component %s from %s.", + display_lvname(lv), lvseg_name(seg)); lv->status &= ~(type | RAID); lv_set_visible(lv); @@ -1843,7 +1858,8 @@ static int _alloc_rmeta_devs_for_lv(struct logical_volume *lv, } if (!_alloc_rmeta_devs_for_rimage_devs(lv, &data_lvs, meta_lvs, allocate_pvs)) { - log_error("Failed to allocate metadata LVs for %s", lv->name); + log_error("Failed to allocate metadata LVs for %s.", + display_lvname(lv)); return 0; } @@ -1861,24 +1877,30 @@ static int _alloc_and_add_rmeta_devs_for_lv(struct logical_volume *lv, struct dm dm_list_init(&meta_lvs); - log_debug_metadata("Allocating metadata LVs for %s.", display_lvname(lv)); + log_debug_metadata("Allocating metadata LVs for %s.", + display_lvname(lv)); if (!_alloc_rmeta_devs_for_lv(lv, &meta_lvs, allocate_pvs, &seg_meta_areas)) { - log_error("Failed to allocate metadata LVs for %s.", display_lvname(lv)); + log_error("Failed to allocate metadata LVs for %s.", + display_lvname(lv)); return 0; } /* Metadata LVs must be cleared before being added to the array */ - log_debug_metadata("Clearing newly allocated metadata LVs for %s.", display_lvname(lv)); + log_debug_metadata("Clearing newly allocated metadata LVs for %s.", + display_lvname(lv)); if (!_clear_lvs(&meta_lvs)) { - log_error("Failed to initialize metadata LVs for %s.", display_lvname(lv)); + log_error("Failed to initialize metadata LVs for %s.", + display_lvname(lv)); return 0; } /* Set segment areas for metadata sub_lvs */ seg->meta_areas = seg_meta_areas; - log_debug_metadata("Adding newly allocated metadata LVs to %s.", display_lvname(lv)); + log_debug_metadata("Adding newly allocated metadata LVs to %s.", + display_lvname(lv)); if (!_add_image_component_list(seg, 1, 0, &meta_lvs, 0)) { - log_error("Failed to add newly allocated metadata LVs to %s.", display_lvname(lv)); + log_error("Failed to add newly allocated metadata LVs to %s.", + display_lvname(lv)); return 0; } @@ -2014,12 +2036,15 @@ static int _lv_update_reload_fns_reset_eliminate_lvs(struct logical_volume *lv, * and if successful, performs metadata backup. */ /* FIXME This needs to be done through hooks in the metadata */ - log_debug_metadata("Clearing any flags for %s passed to the kernel", display_lvname(lv)); + log_debug_metadata("Clearing any flags for %s passed to the kernel.", + display_lvname(lv)); _clear_rebuild_flags(lv, &flags_were_cleared); - log_debug_metadata("Updating metadata and reloading mappings for %s", display_lvname(lv)); + log_debug_metadata("Updating metadata and reloading mappings for %s.", + display_lvname(lv)); if ((r != 2 || flags_were_cleared) && !lv_update_and_reload(lv)) { - log_error("Update and reload of LV %s failed", display_lvname(lv)); + log_error("Update and reload of LV %s failed.", + display_lvname(lv)); return 0; } @@ -2056,7 +2081,7 @@ static int _adjust_data_lvs(struct logical_volume *lv, enum mirror_raid_conv dir } *(sublv_name_suffix + 1) = conv[direction].type_char; - log_debug_metadata("data LV renamed to %s", dlv->name); + log_debug_metadata("data LV renamed to %s.", display_lvname(dlv)); dlv->status &= ~conv[direction].reset_flag; dlv->status |= conv[direction].set_flag; @@ -2082,8 +2107,8 @@ static int _convert_mirror_to_raid1(struct logical_volume *lv, dm_list_init(&meta_lvs); if (!_raid_in_sync(lv)) { - log_error("Unable to convert %s/%s while it is not in-sync", - lv->vg->name, lv->name); + log_error("Unable to convert %s while it is not in-sync.", + display_lvname(lv)); return 0; } @@ -2097,26 +2122,28 @@ static int _convert_mirror_to_raid1(struct logical_volume *lv, return_0; for (s = 0; s < seg->area_count; s++) { - log_debug_metadata("Allocating new metadata LV for %s", - seg_lv(seg, s)->name); + log_debug_metadata("Allocating new metadata LV for %s.", + display_lvname(seg_lv(seg, s))); if (!_alloc_rmeta_for_lv(seg_lv(seg, s), &(lvl_array[s].lv), NULL)) { - log_error("Failed to allocate metadata LV for %s in %s", - seg_lv(seg, s)->name, lv->name); + log_error("Failed to allocate metadata LV for %s in %s.", + display_lvname(seg_lv(seg, s)), + display_lvname(lv)); return 0; } dm_list_add(&meta_lvs, &(lvl_array[s].list)); } - log_debug_metadata("Clearing newly allocated metadata LVs"); + log_debug_metadata("Clearing newly allocated metadata LVs."); if (!_clear_lvs(&meta_lvs)) { - log_error("Failed to initialize metadata LVs"); + log_error("Failed to initialize metadata LVs."); return 0; } if (seg->log_lv) { - log_debug_metadata("Removing mirror log, %s", seg->log_lv->name); + log_debug_metadata("Removing mirror log %s.", + display_lvname(seg->log_lv)); if (!remove_mirror_log(lv->vg->cmd, lv, NULL, 0)) { - log_error("Failed to remove mirror log"); + log_error("Failed to remove mirror log."); return 0; } } @@ -2125,7 +2152,9 @@ static int _convert_mirror_to_raid1(struct logical_volume *lv, s = 0; dm_list_iterate_items(lvl, &meta_lvs) { - log_debug_metadata("Adding %s to %s", lvl->lv->name, lv->name); + log_debug_metadata("Adding %s to %s.", + display_lvname(lvl->lv), + display_lvname(lv)); /* Images are known to be in-sync */ lvl->lv->status &= ~LV_REBUILD; @@ -2134,8 +2163,9 @@ static int _convert_mirror_to_raid1(struct logical_volume *lv, if (!set_lv_segment_area_lv(seg, s, lvl->lv, 0, lvl->lv->status)) { - log_error("Failed to add %s to %s", - lvl->lv->name, lv->name); + log_error("Failed to add %s to %s.", + display_lvname(lvl->lv), + display_lvname(lv)); return 0; } s++; @@ -2144,14 +2174,14 @@ static int _convert_mirror_to_raid1(struct logical_volume *lv, for (s = 0; s < seg->area_count; ++s) { if (!(new_name = _generate_raid_name(lv, "rimage", s))) return_0; - log_debug_metadata("Renaming %s to %s", seg_lv(seg, s)->name, new_name); + log_debug_metadata("Renaming %s to %s.", seg_lv(seg, s)->name, new_name); seg_lv(seg, s)->name = new_name; seg_lv(seg, s)->status &= ~MIRROR_IMAGE; seg_lv(seg, s)->status |= RAID_IMAGE; } init_mirror_in_sync(1); - log_debug_metadata("Setting new segtype for %s", lv->name); + log_debug_metadata("Setting new segtype for %s.", display_lvname(lv)); seg->segtype = new_segtype; lv->status &= ~MIRROR; lv->status &= ~MIRRORED; @@ -2180,19 +2210,19 @@ static int _convert_raid1_to_mirror(struct logical_volume *lv, struct lv_segment *seg = first_seg(lv); if (!seg_is_raid1(seg)) { - log_error(INTERNAL_ERROR "raid1 conversion supported only"); + log_error(INTERNAL_ERROR "raid1 conversion supported only."); return 0; } if ((new_image_count = new_image_count ?: seg->area_count) < 2) { - log_error("can't convert %s to fewer than 2 data_copies", display_lvname(lv)); + log_error("can't convert %s to fewer than 2 data_copies.", display_lvname(lv)); return 0; } if (!_check_max_mirror_devices(new_image_count)) { - log_error("Unable to convert %s LV %s with %u images to %s", + log_error("Unable to convert %s LV %s with %u images to %s.", SEG_TYPE_NAME_RAID1, display_lvname(lv), new_image_count, SEG_TYPE_NAME_MIRROR); - log_error("At least reduce to the maximum of %u images with \"lvconvert -m%u %s\"", + log_error("At least reduce to the maximum of %u images with \"lvconvert -m%u %s\".", DEFAULT_MIRROR_MAX_IMAGES, DEFAULT_MIRROR_MAX_IMAGES - 1, display_lvname(lv)); return 0; } @@ -2201,21 +2231,21 @@ static int _convert_raid1_to_mirror(struct logical_volume *lv, /* Change image pair count to requested # of images */ if (new_image_count != seg->area_count) { - log_debug_metadata("Changing image count to %u on %s", + log_debug_metadata("Changing image count to %u on %s.", new_image_count, display_lvname(lv)); if (!_lv_raid_change_image_count(lv, new_image_count, allocate_pvs, removal_lvs, 0, 0)) return 0; } /* Remove rmeta LVs */ - log_debug_metadata("Extracting and renaming metadata LVs"); + log_debug_metadata("Extracting and renaming metadata LVs."); if (!_extract_image_component_list(seg, RAID_META, 0, removal_lvs)) return 0; seg->meta_areas = NULL; /* Rename all data sub LVs from "*_rimage_*" to "*_mimage_*" and set their status */ - log_debug_metadata("Adjust data LVs of %s", display_lvname(lv)); + log_debug_metadata("Adjust data LVs of %s.", display_lvname(lv)); if (!_adjust_data_lvs(lv, RAID1_TO_MIRROR)) return 0; @@ -2227,7 +2257,7 @@ static int _convert_raid1_to_mirror(struct logical_volume *lv, /* Add mirror_log LV (should happen in wih image allocation */ if (!add_mirror_log(lv->vg->cmd, lv, 1, seg->region_size, allocate_pvs, lv->vg->alloc)) { - log_error("Unable to add mirror log to %s", display_lvname(lv)); + log_error("Unable to add mirror log to %s.", display_lvname(lv)); return 0; } @@ -2299,8 +2329,8 @@ static uint32_t _min_sublv_area_at_le(struct lv_segment *seg, uint32_t area_le) /* Find smallest segment of each of the data image LVs at offset area_le */ for (s = 0; s < seg->area_count; s++) { if (!(seg1 = find_seg_by_le(seg_lv(seg, s), area_le))) { - log_error("Failed to find segment for %s extent %" PRIu32, - seg_lv(seg, s)->name, area_le); + log_error("Failed to find segment for %s extent " FMTu32 ".", + display_lvname(seg_lv(seg, s)), area_le); return 0; } @@ -2352,8 +2382,8 @@ static int _raid0_to_striped_retrieve_segments_and_lvs(struct logical_volume *lv dm_list_iterate_items(seg_to, &new_segments) { for (s = 0; s < seg->area_count; s++) { if (!(data_seg = find_seg_by_le(seg_lv(seg, s), area_le))) { - log_error("Failed to find segment for %s extent %" PRIu32, - seg_lv(seg, s)->name, area_le); + log_error("Failed to find segment for %s extent " FMTu32 ".", + display_lvname(seg_lv(seg, s)), area_le); return 0; } @@ -2396,7 +2426,8 @@ static int _convert_raid0_to_striped(struct logical_volume *lv, /* Move the AREA_PV areas across to new top-level segments of type "striped" */ if (!_raid0_to_striped_retrieve_segments_and_lvs(lv, removal_lvs)) { - log_error("Failed to retrieve raid0 segments from %s.", lv->name); + log_error("Failed to retrieve raid0 segments from %s.", + display_lvname(lv)); return 0; } @@ -2441,13 +2472,13 @@ static struct lv_segment *_convert_striped_to_raid0(struct logical_volume *lv, /* Check for not (yet) supported varying area_count on multi-segment striped LVs */ if (!lv_has_constant_stripes(lv)) { - log_error("Cannot convert striped LV %s with varying stripe count to raid0", + log_error("Cannot convert striped LV %s with varying stripe count to raid0.", display_lvname(lv)); return NULL; } if (!is_power_of_2(seg->stripe_size)) { - log_error("Cannot convert striped LV %s with non-power of 2 stripe size %u", + log_error("Cannot convert striped LV %s with non-power of 2 stripe size %u.", display_lvname(lv), seg->stripe_size); return NULL; } @@ -2879,7 +2910,7 @@ static int _raid1_to_mirrored_wrapper(TAKEOVER_FN_ARGS) if (!yes && yes_no_prompt("Are you sure you want to convert %s back to the older \"%s\" type? [y/n]: ", display_lvname(lv), SEG_TYPE_NAME_MIRROR) == 'n') { - log_error("Logical volume %s NOT converted to \"%s\"", + log_error("Logical volume %s NOT converted to \"%s\".", display_lvname(lv), SEG_TYPE_NAME_MIRROR); return 0; } @@ -2923,7 +2954,7 @@ static int _clear_meta_lvs(struct logical_volume *lv) tmp_areas = seg->meta_areas; /* Extract all MetaLVs listing them on @meta_lvs */ - log_debug_metadata("Extracting all MetaLVs of %s to activate as raid0", + log_debug_metadata("Extracting all MetaLVs of %s to activate as raid0.", display_lvname(lv)); if (!_extract_image_component_sublist(seg, RAID_META, 0, seg->area_count, &meta_lvs, 0)) return_0; @@ -2946,9 +2977,9 @@ static int _clear_meta_lvs(struct logical_volume *lv) if (!deactivate_lv(lv->vg->cmd, lvl->lv)) return_0; - log_debug_metadata("Clearing allocated raid0_meta metadata LVs for conversion to raid4"); + log_debug_metadata("Clearing allocated raid0_meta metadata LVs for conversion to raid4."); if (!_clear_lvs(&meta_lvs)) { - log_error("Failed to initialize metadata LVs"); + log_error("Failed to initialize metadata LVs."); return 0; } @@ -2956,7 +2987,7 @@ static int _clear_meta_lvs(struct logical_volume *lv) seg->meta_areas = tmp_areas; seg->segtype = tmp_segtype; - log_debug_metadata("Adding metadata LVs back into %s", display_lvname(lv)); + log_debug_metadata("Adding metadata LVs back into %s.", display_lvname(lv)); s = 0; dm_list_iterate_items(lvl, &meta_lvs) { lv_set_hidden(lvl->lv); @@ -3070,7 +3101,7 @@ static int _raid456_to_raid0_or_striped_wrapper(TAKEOVER_FN_ARGS) dm_list_init(&removal_lvs); if (!seg_is_raid4(seg) && !seg_is_raid5_n(seg) && !seg_is_raid6_n_6(seg)) { - log_error("LV %s has to be of type raid4/raid5_n/raid6_n_6 to allow for this conversion", + log_error("LV %s has to be of type raid4/raid5_n/raid6_n_6 to allow for this conversion.", display_lvname(lv)); return 0; } @@ -3736,7 +3767,7 @@ int lv_raid_convert(struct logical_volume *lv, takeover_fn_t takeover_fn; if (!new_segtype) { - log_error(INTERNAL_ERROR "New segtype not specified"); + log_error(INTERNAL_ERROR "New segtype not specified."); return 0; } @@ -3790,7 +3821,7 @@ int lv_raid_convert(struct logical_volume *lv, /* LV must be in sync. */ if (!_raid_in_sync(lv)) { - log_error("Unable to convert %s while it is not in-sync", + log_error("Unable to convert %s while it is not in-sync.", display_lvname(lv)); return 0; } @@ -3832,8 +3863,8 @@ static int _remove_partial_multi_segment_image(struct logical_volume *lv, if (pv->status & MISSING_PV) extents_needed += rm_seg->len; } - log_debug("%u extents needed to repair %s", - extents_needed, rm_image->name); + log_debug_metadata("%u extents needed to repair %s.", + extents_needed, display_lvname(rm_image)); /* Second, do the other PVs have the space */ dm_list_iterate_items(rm_seg, &rm_image->segments) { @@ -3845,13 +3876,13 @@ static int _remove_partial_multi_segment_image(struct logical_volume *lv, if ((pv->pe_count - pv->pe_alloc_count) > extents_needed) { - log_debug("%s has enough space for %s", - pv_dev_name(pv), - rm_image->name); + log_debug_metadata("%s has enough space for %s.", + pv_dev_name(pv), + display_lvname(rm_image)); goto has_enough_space; } - log_debug("Not enough space on %s for %s", - pv_dev_name(pv), rm_image->name); + log_debug_metadata("Not enough space on %s for %s.", + pv_dev_name(pv), display_lvname(rm_image)); } } } @@ -3921,15 +3952,15 @@ static int _lv_raid_rebuild_or_replace(struct logical_volume *lv, lv->vg->cmd->partial_activation = 1; if (!lv_is_active_exclusive_locally(lv_lock_holder(lv))) { - log_error("%s/%s must be active %sto perform this operation.", - lv->vg->name, lv->name, + log_error("%s must be active %sto perform this operation.", + display_lvname(lv), vg_is_clustered(lv->vg) ? "exclusive locally " : ""); return 0; } if (!_raid_in_sync(lv)) { - log_error("Unable to replace devices in %s/%s while it is" - " not in-sync.", lv->vg->name, lv->name); + log_error("Unable to replace devices in %s while it is " + "not in-sync.", display_lvname(lv)); return 0; } @@ -3943,7 +3974,7 @@ static int _lv_raid_rebuild_or_replace(struct logical_volume *lv, if ((seg_type(raid_seg, s) == AREA_UNASSIGNED) || (seg_metatype(raid_seg, s) == AREA_UNASSIGNED)) { log_error("Unable to replace RAID images while the " - "array has unassigned areas"); + "array has unassigned areas."); return 0; } @@ -3960,19 +3991,18 @@ static int _lv_raid_rebuild_or_replace(struct logical_volume *lv, } if (!match_count) { - log_print_unless_silent("%s does not contain devices specified" - " to %s", display_lvname(lv), action_str); + log_print_unless_silent("%s does not contain devices specified to %s.", + display_lvname(lv), action_str); return 1; } else if (match_count == raid_seg->area_count) { - log_error("Unable to %s all PVs from %s/%s at once.", - action_str, lv->vg->name, lv->name); + log_error("Unable to %s all PVs from %s at once.", + action_str, display_lvname(lv)); return 0; } else if (raid_seg->segtype->parity_devs && (match_count > raid_seg->segtype->parity_devs)) { - log_error("Unable to %s more than %u PVs from (%s) %s/%s", + log_error("Unable to %s more than %u PVs from (%s) %s.", action_str, raid_seg->segtype->parity_devs, - lvseg_name(raid_seg), - lv->vg->name, lv->name); + lvseg_name(raid_seg), display_lvname(lv)); return 0; } else if (seg_is_raid10(raid_seg)) { uint32_t i, rebuilds_per_group = 0; @@ -4049,8 +4079,8 @@ try_again: lv->vg->cmd->partial_activation = 1; goto try_again; } - log_error("Failed to allocate replacement images for %s/%s", - lv->vg->name, lv->name); + log_error("Failed to allocate replacement images for %s.", + display_lvname(lv)); return 0; } @@ -4073,8 +4103,8 @@ try_again: partial_segment_removed ? &lv->vg->pvs : remove_pvs, 0, &old_lvs, &old_lvs)) { - log_error("Failed to remove the specified images from %s/%s", - lv->vg->name, lv->name); + log_error("Failed to remove the specified images from %s.", + display_lvname(lv)); return 0; } @@ -4106,8 +4136,9 @@ try_again: return_0; if (!set_lv_segment_area_lv(raid_seg, s, lvl->lv, 0, lvl->lv->status)) { - log_error("Failed to add %s to %s", - lvl->lv->name, lv->name); + log_error("Failed to add %s to %s.", + display_lvname(lvl->lv), + display_lvname(lv)); return 0; } lv_set_hidden(lvl->lv); @@ -4121,8 +4152,9 @@ try_again: return_0; if (!set_lv_segment_area_lv(raid_seg, s, lvl->lv, 0, lvl->lv->status)) { - log_error("Failed to add %s to %s", - lvl->lv->name, lv->name); + log_error("Failed to add %s to %s.", + display_lvname(lvl->lv), + display_lvname(lv)); return 0; } lv_set_hidden(lvl->lv); @@ -4200,16 +4232,16 @@ int lv_raid_remove_missing(struct logical_volume *lv) struct lv_segment *seg = first_seg(lv); if (!lv_is_partial(lv)) { - log_error(INTERNAL_ERROR "%s/%s is not a partial LV", - lv->vg->name, lv->name); + log_error(INTERNAL_ERROR "%s is not a partial LV.", + display_lvname(lv)); return 0; } if (!archive(lv->vg)) return_0; - log_debug("Attempting to remove missing devices from %s LV, %s", - lvseg_name(seg), lv->name); + log_debug("Attempting to remove missing devices from %s LV, %s.", + lvseg_name(seg), display_lvname(lv)); /* * FIXME: Make sure # of compromised components will not affect RAID @@ -4220,10 +4252,10 @@ int lv_raid_remove_missing(struct logical_volume *lv) (!seg->meta_areas || !seg_metalv(seg, s) || !lv_is_partial(seg_metalv(seg, s)))) continue; - log_debug("Replacing %s segments with error target", + log_debug("Replacing %s segments with error target.", display_lvname(seg_lv(seg, s))); if (seg->meta_areas && seg_metalv(seg, s)) - log_debug("Replacing %s segments with error target", + log_debug("Replacing %s segments with error target.", display_lvname(seg_metalv(seg, s))); if (!replace_lv_with_error_segment(seg_lv(seg, s))) { log_error("Failed to replace %s's extents with error target.", @@ -4325,7 +4357,7 @@ static int _lv_may_be_activated_in_degraded_mode(struct logical_volume *lv, void dm_list_iterate_items(seg, &lv->segments) for (s = 0; s < seg->area_count; s++) if (seg_type(seg, s) != AREA_LV) { - log_verbose("%s contains a segment incapable of degraded activation", + log_verbose("%s contains a segment incapable of degraded activation.", display_lvname(lv)); *not_capable = 1; }

1 0

master - libdm-stats: fix dm_stats_delete_region() performance
by Bryn Reeves 09 Dec '16

09 Dec '16

Gitweb: http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=6dd0bd0255d585fe… Commit: 6dd0bd0255d585feca61000432d2de9552ede4e0 Parent: 0ce9ae3cda9411f8f2a5491b07a6c4f06ab44da1 Author: Bryn M. Reeves <bmr(a)redhat.com> AuthorDate: Thu Dec 8 20:55:47 2016 +0000 Committer: Bryn M. Reeves <bmr(a)redhat.com> CommitterDate: Fri Dec 9 10:55:39 2016 +0000 libdm-stats: fix dm_stats_delete_region() performance Fix a silly bug in dm_stats_delete_region() that hugely inflates runtimes when deleting a large number of regions. For ~50,000 regions this change reduces the runtime from 98s to 6s on my test systems (a ~93% reduction). The bug exists because dm_stats_delete_region() applies a truth test to the return value of dm_stats_get_nr_areas(); this is never correct usage - it will walk the entire region table and calculate area counts for each region (which is roughly O(n^2) in the number of regions, as dm_stats_delete_region() is being called inside a region walk). Although the individual area calculation is not that costly, uselessly running anything 2,500,000,000 times over gets a bit slow. A much cheaper test (which is always true if the areas check is true) is to just test dm_stats_get_nr_regions() or dms->regions; if either is true it implies at least one area exists. Old: Performance counter stats for 'dmstats delete --allregions --alldevices': 98117.791458 task-clock (msec) # 1.000 CPUs utilized 127 context-switches # 0.001 K/sec 3 cpu-migrations # 0.000 K/sec 6,631 page-faults # 0.068 K/sec 307,711,724,562 cycles # 3.136 GHz 544,762,959,577 instructions # 1.77 insn per cycle 84,287,824,115 branches # 859.047 M/sec 2,538,875 branch-misses # 0.00% of all branches 98.119578733 seconds time elapsed New: Performance counter stats for 'dmstats delete --allregions --alldevices': 6427.251074 task-clock (msec) # 1.000 CPUs utilized 6 context-switches # 0.001 K/sec 0 cpu-migrations # 0.000 K/sec 6,634 page-faults # 0.001 M/sec 21,613,018,724 cycles # 3.363 GHz 3,794,755,445 instructions # 0.18 insn per cycle 852,974,026 branches # 132.712 M/sec 808,625 branch-misses # 0.09% of all branches 6.428953647 seconds time elapsed --- libdm/libdm-stats.c | 2 +- 1 files changed, 1 insertions(+), 1 deletions(-) diff --git a/libdm/libdm-stats.c b/libdm/libdm-stats.c index 2fc00da..411fd80 100644 --- a/libdm/libdm-stats.c +++ b/libdm/libdm-stats.c @@ -2051,7 +2051,7 @@ int dm_stats_delete_region(struct dm_stats *dms, uint64_t region_id) goto bad; } - if (!dm_stats_get_nr_areas(dms)) { + if (!dm_stats_get_nr_regions(dms)) { log_error("Could not delete region ID " FMTu64 ": " "no regions found", region_id); goto bad;

1 0

master - tests: more snaps
by Zdenek Kabelac 06 Dec '16

06 Dec '16

Gitweb: http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=0ce9ae3cda9411f8… Commit: 0ce9ae3cda9411f8f2a5491b07a6c4f06ab44da1 Parent: 3f6ade4b0d5901bb01bcbdfdb926823415a01703 Author: Zdenek Kabelac <zkabelac(a)redhat.com> AuthorDate: Sun Dec 4 22:27:57 2016 +0100 Committer: Zdenek Kabelac <zkabelac(a)redhat.com> CommitterDate: Mon Dec 5 17:12:42 2016 +0100 tests: more snaps Check lvs reports 'origin' not openned. --- test/shell/snapshot-merge-stack.sh | 7 +++++-- 1 files changed, 5 insertions(+), 2 deletions(-) diff --git a/test/shell/snapshot-merge-stack.sh b/test/shell/snapshot-merge-stack.sh index 3ee9eaa..bfafe71 100644 --- a/test/shell/snapshot-merge-stack.sh +++ b/test/shell/snapshot-merge-stack.sh @@ -37,14 +37,17 @@ snap_and_merge() { lvconvert -b --merge $vg/$lv2 lvs -a -o+lv_merging,lv_merge_failed $vg + get lv_field $vg/$lv1 lv_attr | grep "Owi-ao" + get lv_field $vg/$lv2 lv_attr | grep "Swi-a-s---" kill $SLEEP_PID aux delay_dev "$dev1" 0 200 $(get first_extent_sector "$dev1"): lvchange --poll n --refresh $vg/$lv1 dmsetup table - lvs -a -o+lv_merging,lv_merge_failed $vg + lvs -av -o+lv_merging,lv_merge_failed $vg + # Origin is closed and snapshot merge could run + get lv_field $vg/$lv1 lv_attr | grep "Owi-a-" sleep 1 - check lv_attr_bit state $vg/$lv1 "a" check lv_attr_bit state $vg/$lv2 "a" aux error_dev "$dev2" $(get first_extent_sector "$dev2"): aux enable_dev "$dev1"

1 0

master - tests: apostrof
by Zdenek Kabelac 06 Dec '16

06 Dec '16

Gitweb: http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=3f6ade4b0d5901bb… Commit: 3f6ade4b0d5901bb01bcbdfdb926823415a01703 Parent: 022d3af06853ba4e627e098ac04ef1a8f9806edc Author: Zdenek Kabelac <zkabelac(a)redhat.com> AuthorDate: Mon Nov 28 17:36:43 2016 +0100 Committer: Zdenek Kabelac <zkabelac(a)redhat.com> CommitterDate: Mon Dec 5 17:12:42 2016 +0100 tests: apostrof --- test/lib/aux.sh | 4 ++-- 1 files changed, 2 insertions(+), 2 deletions(-) diff --git a/test/lib/aux.sh b/test/lib/aux.sh index e4c3780..7f3c84a 100644 --- a/test/lib/aux.sh +++ b/test/lib/aux.sh @@ -322,9 +322,9 @@ prepare_lvmdbusd() { # Setup the python path so we can run export PYTHONPATH=$abs_top_builddir/daemons else - daemon="$(which lvmdbusd || :)" + daemon=$(which lvmdbusd || :) fi - [[ -n $daemon && -x $daemon ]] || skip "The daemon is missing" + [[ -n "$daemon" && -x "$daemon" ]] || skip "The daemon is missing" which python3 >/dev/null || skip "Missing python3" python3 -c "import pyudev, dbus, gi.repository" || skip "Missing python modules"

1 0

master - tests: check we recognize broken table entry
by Zdenek Kabelac 06 Dec '16

06 Dec '16

Gitweb: http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=022d3af06853ba4e… Commit: 022d3af06853ba4e627e098ac04ef1a8f9806edc Parent: 36b211b464a02cf054a838ccf9139379897cf973 Author: Zdenek Kabelac <zkabelac(a)redhat.com> AuthorDate: Thu Dec 1 10:45:37 2016 +0100 Committer: Zdenek Kabelac <zkabelac(a)redhat.com> CommitterDate: Mon Dec 5 17:12:42 2016 +0100 tests: check we recognize broken table entry One (initial) test to recognize we see a mismatching table entry for cache LV. --- test/shell/lvs-cache.sh | 19 +++++++++++++++++++ 1 files changed, 19 insertions(+), 0 deletions(-) diff --git a/test/shell/lvs-cache.sh b/test/shell/lvs-cache.sh index a06a023..2c518d0 100644 --- a/test/shell/lvs-cache.sh +++ b/test/shell/lvs-cache.sh @@ -22,6 +22,25 @@ SKIP_WITH_LVMPOLLD=1 aux have_cache 1 3 0 || skip aux prepare_vg 5 8000 +# Use 10M origin size +lvcreate -L10 -n $lv1 $vg +lvcreate -H -L5 $vg/$lv1 + +# replace 10M size with 5M size of cache device +NEWCLINE=$(dmsetup table $vg-$lv1 | sed 's/20480/10240/') +dmsetup reload $vg-$lv1 --table "$NEWCLINE" +dmsetup resume $vg-$lv1 + +# Check that mismatching cache target is shown by lvs +lvs -a $vg 2>&1 | grep "WARNING" +check lv_attr_bit state $vg/$lv1 "X" + +lvs -o+lv_active $vg + +lvremove -f $vg + + + lvcreate --type cache-pool -L10 $vg/cpool lvcreate --type cache -l 1 --cachepool $vg/cpool -n corigin $vg lvs -o lv_name,cache_policy

1 0

master - debug: add missing backtrace
by Zdenek Kabelac 06 Dec '16

06 Dec '16

Gitweb: http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=36b211b464a02cf0… Commit: 36b211b464a02cf054a838ccf9139379897cf973 Parent: 3331199cc9f4b3935bc531428f03fbf3bb017d4c Author: Zdenek Kabelac <zkabelac(a)redhat.com> AuthorDate: Sun Dec 4 22:29:18 2016 +0100 Committer: Zdenek Kabelac <zkabelac(a)redhat.com> CommitterDate: Mon Dec 5 17:12:42 2016 +0100 debug: add missing backtrace --- lib/metadata/lv.c | 3 ++- 1 files changed, 2 insertions(+), 1 deletions(-) diff --git a/lib/metadata/lv.c b/lib/metadata/lv.c index 4a9fed7..8f13f6b 100644 --- a/lib/metadata/lv.c +++ b/lib/metadata/lv.c @@ -255,7 +255,8 @@ char *lvseg_kernel_discards_dup(struct dm_pool *mem, const struct lv_segment *se if (!(status.info_ok = lv_info_with_seg_status(seg->lv->vg->cmd, seg, &status, 0, 0))) goto_bad; - ret = lvseg_kernel_discards_dup_with_info_and_seg_status(mem, &status); + if (!(ret = lvseg_kernel_discards_dup_with_info_and_seg_status(mem, &status))) + stack; bad: dm_pool_destroy(status.seg_status.mem);

1 0

master - cleanup: simplier code
by Zdenek Kabelac 06 Dec '16

06 Dec '16

Gitweb: http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=3331199cc9f4b393… Commit: 3331199cc9f4b3935bc531428f03fbf3bb017d4c Parent: 81ef4eb4f887381d67546884be5a717030fb505a Author: Zdenek Kabelac <zkabelac(a)redhat.com> AuthorDate: Thu Dec 1 14:59:02 2016 +0100 Committer: Zdenek Kabelac <zkabelac(a)redhat.com> CommitterDate: Mon Dec 5 17:12:42 2016 +0100 cleanup: simplier code --- lib/activate/dev_manager.c | 12 +++++------- 1 files changed, 5 insertions(+), 7 deletions(-) diff --git a/lib/activate/dev_manager.c b/lib/activate/dev_manager.c index 6a8e487..581e900 100644 --- a/lib/activate/dev_manager.c +++ b/lib/activate/dev_manager.c @@ -728,15 +728,13 @@ int dev_manager_info(struct cmd_context *cmd, const struct logical_volume *lv, struct lv_seg_status *seg_status) { char *dlid, *name; - int r; + int r = 0; if (!(name = dm_build_dm_name(cmd->mem, lv->vg->name, lv->name, layer))) return_0; - if (!(dlid = build_dm_uuid(cmd->mem, lv, layer))) { - r = 0; + if (!(dlid = build_dm_uuid(cmd->mem, lv, layer))) goto_out; - } log_debug_activation("Getting device info for %s [%s].", name, dlid); r = _info(cmd, dlid, with_open_count, with_read_ahead, @@ -1446,13 +1444,13 @@ int dev_manager_thin_pool_percent(struct dev_manager *dm, { char *name; const char *dlid; + const char *layer = lv_layer(lv); /* Build a name for the top layer */ - if (!(name = dm_build_dm_name(dm->mem, lv->vg->name, lv->name, - lv_layer(lv)))) + if (!(name = dm_build_dm_name(dm->mem, lv->vg->name, lv->name, layer))) return_0; - if (!(dlid = build_dm_uuid(dm->mem, lv, lv_layer(lv)))) + if (!(dlid = build_dm_uuid(dm->mem, lv, layer))) return_0; log_debug_activation("Getting device status percentage for %s.", name);

1 0

master - cleanup: indent and messsages updates
by Zdenek Kabelac 06 Dec '16

06 Dec '16

Gitweb: http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=81ef4eb4f887381d… Commit: 81ef4eb4f887381d67546884be5a717030fb505a Parent: 69c79ed2f486a9092cdf1586fcf320634e3cd44d Author: Zdenek Kabelac <zkabelac(a)redhat.com> AuthorDate: Thu Dec 1 14:53:35 2016 +0100 Committer: Zdenek Kabelac <zkabelac(a)redhat.com> CommitterDate: Mon Dec 5 17:12:42 2016 +0100 cleanup: indent and messsages updates --- lib/activate/activate.c | 4 ++ lib/activate/dev_manager.c | 77 +++++++++++++++++++++++++------------------- lib/metadata/lv.c | 1 + 3 files changed, 49 insertions(+), 33 deletions(-) diff --git a/lib/activate/activate.c b/lib/activate/activate.c index 3128f36..b4c8a2c 100644 --- a/lib/activate/activate.c +++ b/lib/activate/activate.c @@ -655,6 +655,10 @@ int target_present(struct cmd_context *cmd, const char *target_name, &maj, &min, &patchlevel); } +/* + * When '*info' is NULL, returns 1 only when LV is active. + * When '*info' != NULL, returns 1 when info structure is populated. + */ static int _lv_info(struct cmd_context *cmd, const struct logical_volume *lv, int use_layer, struct lvinfo *info, const struct lv_segment *seg, diff --git a/lib/activate/dev_manager.c b/lib/activate/dev_manager.c index af7f5d8..6a8e487 100644 --- a/lib/activate/dev_manager.c +++ b/lib/activate/dev_manager.c @@ -136,6 +136,7 @@ static struct dm_task *_setup_task_run(int task, struct dm_info *info, out: dm_task_destroy(dmt); + return NULL; } @@ -236,7 +237,7 @@ static int _info_run(info_type_t type, const char *dlid, with_flush = 0; break; default: - log_error(INTERNAL_ERROR "_info_run: unhandled info type"); + log_error(INTERNAL_ERROR "_info_run: unhandled info type."); return 0; } @@ -274,6 +275,7 @@ static int _info_run(info_type_t type, const char *dlid, out: dm_task_destroy(dmt); + return r; } @@ -326,14 +328,14 @@ static int _ignore_blocked_mirror_devices(struct device *dev, for (i = 0; i < sm->dev_count; ++i) if (sm->devs[i].health != DM_STATUS_MIRROR_ALIVE) { - log_debug_activation("%s: Mirror image %d marked as failed", + log_debug_activation("%s: Mirror image %d marked as failed.", dev_name(dev), i); check_for_blocking = 1; } if (!check_for_blocking && sm->log_count) { if (sm->logs[0].health != DM_STATUS_MIRROR_ALIVE) { - log_debug_activation("%s: Mirror log device marked as failed", + log_debug_activation("%s: Mirror log device marked as failed.", dev_name(dev)); check_for_blocking = 1; } else { @@ -382,7 +384,7 @@ static int _ignore_blocked_mirror_devices(struct device *dev, (p[15] == '\0' || p[15] == ' ')) || ((p = strstr(params, " handle_errors")) && (p[14] == '\0' || p[14] == ' '))) { - log_debug_activation("%s: I/O blocked to mirror device", + log_debug_activation("%s: I/O blocked to mirror device.", dev_name(dev)); goto out; } @@ -447,6 +449,7 @@ static int _ignore_suspended_snapshot_component(struct device *dev) out: dm_task_destroy(dmt); + return r; } @@ -735,7 +738,7 @@ int dev_manager_info(struct cmd_context *cmd, const struct logical_volume *lv, goto_out; } - log_debug_activation("Getting device info for %s [%s]", name, dlid); + log_debug_activation("Getting device info for %s [%s].", name, dlid); r = _info(cmd, dlid, with_open_count, with_read_ahead, dminfo, read_ahead, seg_status); out: @@ -874,7 +877,8 @@ bad: return (device_id == id); } -int add_linear_area_to_dtree(struct dm_tree_node *node, uint64_t size, uint32_t extent_size, int use_linear_target, const char *vgname, const char *lvname) +int add_linear_area_to_dtree(struct dm_tree_node *node, uint64_t size, uint32_t extent_size, + int use_linear_target, const char *vgname, const char *lvname) { uint32_t page_size; @@ -895,10 +899,12 @@ int add_linear_area_to_dtree(struct dm_tree_node *node, uint64_t size, uint32_t if (!dm_tree_node_add_striped_target(node, size, extent_size)) return_0; return 1; - } else - /* Some exotic cases are unsupported by striped. */ - log_warn("WARNING: Using linear target for %s/%s: Striped requires extent size (%" PRIu32 " sectors) >= page size (%" PRIu32 ").", - vgname, lvname, extent_size, page_size); + } + + /* Some exotic cases are unsupported by striped. */ + log_warn("WARNING: Using linear target for %s/%s: Striped requires extent size " + "(" FMTu32 " sectors) >= page size (" FMTu32 ").", + vgname, lvname, extent_size, page_size); } /* @@ -1148,6 +1154,7 @@ struct dev_manager *dev_manager_create(struct cmd_context *cmd, bad: dm_pool_destroy(mem); + return NULL; } @@ -1236,10 +1243,10 @@ int dev_manager_mirror_percent(struct dev_manager *dm, if (!(dlid = build_dm_uuid(dm->mem, lv, layer))) return_0; - log_debug_activation("Getting device %s status percentage for %s", + log_debug_activation("Getting device %s status percentage for %s.", target_type, name); - if (!_percent(dm, name, dlid, target_type, wait, lv, percent, - event_nr, 0)) + + if (!_percent(dm, name, dlid, target_type, wait, lv, percent, event_nr, 0)) return_0; return 1; @@ -1269,9 +1276,9 @@ int dev_manager_raid_status(struct dev_manager *dm, dm_get_next_target(dmt, NULL, &start, &length, &type, &params); - if (!type || strcmp(type, "raid")) { - log_error("Expected raid segment type but got %s instead", - type ? type : "NULL"); + if (!type || strcmp(type, TARGET_NAME_RAID)) { + log_error("Expected %s segment type but got %s instead.", + TARGET_NAME_RAID, type ? type : "NULL"); goto out; } @@ -1297,7 +1304,7 @@ int dev_manager_raid_message(struct dev_manager *dm, const char *layer = lv_layer(lv); if (!lv_is_raid(lv)) { - log_error(INTERNAL_ERROR "%s is not a RAID logical volume", + log_error(INTERNAL_ERROR "%s is not a RAID logical volume.", display_lvname(lv)); return 0; } @@ -1310,7 +1317,7 @@ int dev_manager_raid_message(struct dev_manager *dm, strcmp(msg, "check") && strcmp(msg, "repair") && strcmp(msg, "reshape")) { - log_error(INTERNAL_ERROR "Unknown RAID message: %s", msg); + log_error(INTERNAL_ERROR "Unknown RAID message: %s.", msg); return 0; } @@ -1361,8 +1368,8 @@ int dev_manager_cache_status(struct dev_manager *dm, dm_get_next_target(dmt, NULL, &start, &length, &type, &params); if (!type || strcmp(type, TARGET_NAME_CACHE)) { - log_error("Expected cache segment type but got %s instead", - type ? type : "NULL"); + log_error("Expected %s segment type but got %s instead.", + TARGET_NAME_CACHE, type ? type : "NULL"); goto out; } @@ -1448,7 +1455,8 @@ int dev_manager_thin_pool_percent(struct dev_manager *dm, if (!(dlid = build_dm_uuid(dm->mem, lv, lv_layer(lv)))) return_0; - log_debug_activation("Getting device status percentage for %s", name); + log_debug_activation("Getting device status percentage for %s.", name); + if (!(_percent(dm, name, dlid, TARGET_NAME_THIN_POOL, 0, (metadata) ? lv : NULL, percent, NULL, 1))) return_0; @@ -1472,6 +1480,7 @@ int dev_manager_thin_percent(struct dev_manager *dm, return_0; log_debug_activation("Getting device status percentage for %s", name); + if (!(_percent(dm, name, dlid, TARGET_NAME_THIN, 0, (mapped) ? NULL : lv, percent, NULL, 1))) return_0; @@ -1706,7 +1715,8 @@ static int _add_dev_to_dtree(struct dev_manager *dm, struct dm_tree *dtree, if (!(dlid = build_dm_uuid(dm->mem, lv, layer))) return_0; - log_debug_activation("Getting device info for %s [%s]", name, dlid); + log_debug_activation("Getting device info for %s [%s].", name, dlid); + if (!_info(dm->cmd, dlid, 1, 0, &info, NULL, NULL)) { log_error("Failed to get info for %s [%s].", name, dlid); return 0; @@ -1723,7 +1733,7 @@ static int _add_dev_to_dtree(struct dev_manager *dm, struct dm_tree *dtree, if (info.exists && ((int) info.minor != lv->minor)) { log_error("Volume %s (%" PRIu32 ":%" PRIu32")" " differs from already active device " - "(%" PRIu32 ":%" PRIu32")", + "(%" PRIu32 ":%" PRIu32").", display_lvname(lv), lv->major, lv->minor, info.major, info.minor); return 0; @@ -1731,7 +1741,7 @@ static int _add_dev_to_dtree(struct dev_manager *dm, struct dm_tree *dtree, if (!info.exists && _info_by_dev(lv->major, lv->minor, &info2) && info2.exists) { log_error("The requested major:minor pair " - "(%" PRIu32 ":%" PRIu32") is already used", + "(%" PRIu32 ":%" PRIu32") is already used.", lv->major, lv->minor); return 0; } @@ -1739,7 +1749,7 @@ static int _add_dev_to_dtree(struct dev_manager *dm, struct dm_tree *dtree, if (info.exists && !dm_tree_add_dev_with_udev_flags(dtree, info.major, info.minor, _get_udev_flags(dm, lv, layer, 0, 0))) { - log_error("Failed to add device (%" PRIu32 ":%" PRIu32") to dtree", + log_error("Failed to add device (%" PRIu32 ":%" PRIu32") to dtree.", info.major, info.minor); return 0; } @@ -1868,7 +1878,7 @@ static int _pool_callback(struct dm_tree_node *node, for (cv = cn->v; cv && args < 16; cv = cv->next) { if (cv->type != DM_CFG_STRING) { log_error("Invalid string in config file: " - "global/%s_check_options", + "global/%s_check_options.", data->global); return 0; } @@ -2491,7 +2501,7 @@ static int _add_target_to_dtree(struct dev_manager *dm, if (!seg->segtype->ops->add_target_line) { log_error(INTERNAL_ERROR "_emit_target cannot handle " - "segment type %s", seg->segtype->name); + "segment type %s.", seg->segtype->name); return 0; } @@ -2947,11 +2957,11 @@ static int _create_lv_symlinks(struct dev_manager *dm, struct dm_tree_node *root if (name && lvlayer->old_name && *lvlayer->old_name && strcmp(name, lvlayer->old_name)) { if (!dm_split_lvm_name(dm->mem, lvlayer->old_name, &old_vgname, &old_lvname, &old_layer)) { - log_error("_create_lv_symlinks: Couldn't split up old device name %s", lvlayer->old_name); + log_error("_create_lv_symlinks: Couldn't split up old device name %s.", lvlayer->old_name); return 0; } if (!dm_split_lvm_name(dm->mem, name, &new_vgname, &new_lvname, &new_layer)) { - log_error("_create_lv_symlinks: Couldn't split up new device name %s", name); + log_error("_create_lv_symlinks: Couldn't split up new device name %s.", name); return 0; } if (!fs_rename_lv(lvlayer->lv, name, old_vgname, old_lvname)) @@ -3079,7 +3089,7 @@ static int _tree_action(struct dev_manager *dm, const struct logical_volume *lv, return_0; if (!(root = dm_tree_find_node(dtree, 0, 0))) { - log_error("Lost dependency tree root node"); + log_error("Lost dependency tree root node."); goto out_no_root; } @@ -3226,14 +3236,14 @@ int dev_manager_device_uses_vg(struct device *dev, int r = 1; if (!(dtree = dm_tree_create())) { - log_error("partial dtree creation failed"); + log_error("Failed to create partial dtree."); return r; } dm_tree_set_optional_uuid_suffixes(dtree, &uuid_suffix_list[0]); if (!dm_tree_add_dev(dtree, (uint32_t) MAJOR(dev->dev), (uint32_t) MINOR(dev->dev))) { - log_error("Failed to add device %s (%" PRIu32 ":%" PRIu32") to dtree", + log_error("Failed to add device %s (%" PRIu32 ":%" PRIu32") to dtree.", dev_name(dev), (uint32_t) MAJOR(dev->dev), (uint32_t) MINOR(dev->dev)); goto out; } @@ -3242,7 +3252,7 @@ int dev_manager_device_uses_vg(struct device *dev, memcpy(dlid + sizeof(UUID_PREFIX) - 1, &vg->id.uuid[0], sizeof(vg->id)); if (!(root = dm_tree_find_node(dtree, 0, 0))) { - log_error("Lost dependency tree root node"); + log_error("Lost dependency tree root node."); goto out; } @@ -3253,5 +3263,6 @@ int dev_manager_device_uses_vg(struct device *dev, out: dm_tree_free(dtree); + return r; } diff --git a/lib/metadata/lv.c b/lib/metadata/lv.c index 081df71..4a9fed7 100644 --- a/lib/metadata/lv.c +++ b/lib/metadata/lv.c @@ -258,6 +258,7 @@ char *lvseg_kernel_discards_dup(struct dm_pool *mem, const struct lv_segment *se ret = lvseg_kernel_discards_dup_with_info_and_seg_status(mem, &status); bad: dm_pool_destroy(status.seg_status.mem); + return ret; }

1 0

master - cleanup: simplify code
by Zdenek Kabelac 06 Dec '16

06 Dec '16

Gitweb: http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=69c79ed2f486a909… Commit: 69c79ed2f486a9092cdf1586fcf320634e3cd44d Parent: 114f7e62854e7b62f613925881ad321c3d5f6ff3 Author: Zdenek Kabelac <zkabelac(a)redhat.com> AuthorDate: Mon Dec 5 14:30:13 2016 +0100 Committer: Zdenek Kabelac <zkabelac(a)redhat.com> CommitterDate: Mon Dec 5 17:12:42 2016 +0100 cleanup: simplify code Drop unneeded assignemnt in structs (var is set by called function). --- lib/metadata/lv.c | 10 +++------- 1 files changed, 3 insertions(+), 7 deletions(-) diff --git a/lib/metadata/lv.c b/lib/metadata/lv.c index 087b564..081df71 100644 --- a/lib/metadata/lv.c +++ b/lib/metadata/lv.c @@ -243,10 +243,7 @@ char *lvseg_kernel_discards_dup(struct dm_pool *mem, const struct lv_segment *se { char *ret = NULL; struct lv_with_info_and_seg_status status = { - .seg_status = { - .type = SEG_STATUS_NONE, - .seg = seg - }, + .seg_status.type = SEG_STATUS_NONE }; if (!lv_is_thin_pool(seg->lv)) @@ -416,8 +413,8 @@ dm_percent_t lvseg_percent_with_info_and_seg_status(const struct lv_with_info_an else { log_warn("WARNING: Thin volume %s maps %s while the size is only %s.", display_lvname(seg->lv), - display_size(lvdm->lv->vg->cmd, s->thin->mapped_sectors), - display_size(lvdm->lv->vg->cmd, csize)); + display_size(seg->lv->vg->cmd, s->thin->mapped_sectors), + display_size(seg->lv->vg->cmd, csize)); /* Don't show nonsense numbers like i.e. 1000% full */ p = DM_PERCENT_100; } @@ -1312,7 +1309,6 @@ char *lv_attr_dup(struct dm_pool *mem, const struct logical_volume *lv) char *ret = NULL; struct lv_with_info_and_seg_status status = { .seg_status.type = SEG_STATUS_NONE, - .lv = lv }; if (!(status.seg_status.mem = dm_pool_create("reporter_pool", 1024)))

1 0

master - dev_manager: use setup_task_run for mknod
by Zdenek Kabelac 06 Dec '16

06 Dec '16

Gitweb: http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=114f7e62854e7b62… Commit: 114f7e62854e7b62f613925881ad321c3d5f6ff3 Parent: 5163b8f697191caf13ae8192781f65b3f6335e74 Author: Zdenek Kabelac <zkabelac(a)redhat.com> AuthorDate: Fri Dec 2 13:46:47 2016 +0100 Committer: Zdenek Kabelac <zkabelac(a)redhat.com> CommitterDate: Mon Dec 5 17:12:39 2016 +0100 dev_manager: use setup_task_run for mknod Simplify info run for use only for INFO & STATUS. Drop handling MKNODES within _info_run() call and use more advanced _setup_task_run() directly. This allows to further simplify _info_run(). --- WHATS_NEW | 1 + lib/activate/dev_manager.c | 36 +++++++++++++++++------------------- 2 files changed, 18 insertions(+), 19 deletions(-) diff --git a/WHATS_NEW b/WHATS_NEW index eb64542..595b667 100644 --- a/WHATS_NEW +++ b/WHATS_NEW @@ -1,5 +1,6 @@ Version 2.02.169 - ===================================== + Simplify internal _info_run() and use _setup_task_run() for mknod. Better API for internal function _setup_task_run. Avoid using lv_has_target_type() call within lv_info_with_seg_status. Simplify internal lv_info_with_seg_status API. diff --git a/lib/activate/dev_manager.c b/lib/activate/dev_manager.c index a90b6c0..af7f5d8 100644 --- a/lib/activate/dev_manager.c +++ b/lib/activate/dev_manager.c @@ -211,10 +211,9 @@ static int _get_segment_status_from_target_params(const char *target_name, typedef enum { INFO, /* DM_DEVICE_INFO ioctl */ STATUS, /* DM_DEVICE_STATUS ioctl */ - MKNODES } info_type_t; -static int _info_run(info_type_t type, const char *name, const char *dlid, +static int _info_run(info_type_t type, const char *dlid, struct dm_info *dminfo, uint32_t *read_ahead, struct lv_seg_status *seg_status, int with_open_count, int with_read_ahead, @@ -236,17 +235,13 @@ static int _info_run(info_type_t type, const char *name, const char *dlid, dmtask = DM_DEVICE_STATUS; with_flush = 0; break; - case MKNODES: - dmtask = DM_DEVICE_MKNODES; - break; default: log_error(INTERNAL_ERROR "_info_run: unhandled info type"); return 0; } if (!(dmt = _setup_task_run(dmtask, dminfo, - (type == MKNODES) ? name : NULL, dlid, 0, - major, minor, with_open_count, with_flush, 0))) + NULL, dlid, 0, major, minor, with_open_count, with_flush, 0))) return_0; if (with_read_ahead && dminfo->exists) { @@ -686,7 +681,7 @@ static int _info(struct cmd_context *cmd, const char *dlid, int with_open_count, unsigned i = 0; /* Check for dlid */ - if ((r = _info_run(seg_status ? STATUS : INFO, NULL, dlid, dminfo, read_ahead, + if ((r = _info_run(seg_status ? STATUS : INFO, dlid, dminfo, read_ahead, seg_status, with_open_count, with_read_ahead, 0, 0)) && dminfo->exists) return 1; @@ -698,7 +693,7 @@ static int _info(struct cmd_context *cmd, const char *dlid, int with_open_count, (void) strncpy(old_style_dlid, dlid, sizeof(old_style_dlid)); old_style_dlid[sizeof(old_style_dlid) - 1] = '\0'; - if ((r = _info_run(seg_status ? STATUS : INFO, NULL, old_style_dlid, dminfo, + if ((r = _info_run(seg_status ? STATUS : INFO, old_style_dlid, dminfo, read_ahead, seg_status, with_open_count, with_read_ahead, 0, 0)) && dminfo->exists) return 1; @@ -710,7 +705,7 @@ static int _info(struct cmd_context *cmd, const char *dlid, int with_open_count, return r; /* Check for dlid before UUID_PREFIX was added */ - if ((r = _info_run(seg_status ? STATUS : INFO, NULL, dlid + sizeof(UUID_PREFIX) - 1, + if ((r = _info_run(seg_status ? STATUS : INFO, dlid + sizeof(UUID_PREFIX) - 1, dminfo, read_ahead, seg_status, with_open_count, with_read_ahead, 0, 0)) && dminfo->exists) return 1; @@ -720,7 +715,7 @@ static int _info(struct cmd_context *cmd, const char *dlid, int with_open_count, static int _info_by_dev(uint32_t major, uint32_t minor, struct dm_info *info) { - return _info_run(INFO, NULL, NULL, info, NULL, 0, 0, 0, major, minor); + return _info_run(INFO, NULL, info, NULL, 0, 0, 0, major, minor); } int dev_manager_info(struct cmd_context *cmd, const struct logical_volume *lv, @@ -1561,21 +1556,24 @@ static int _lv_has_mknode(const struct logical_volume *lv) int dev_manager_mknodes(const struct logical_volume *lv) { struct dm_info dminfo; + struct dm_task *dmt; char *name; int r = 0; if (!(name = dm_build_dm_name(lv->vg->cmd->mem, lv->vg->name, lv->name, NULL))) return_0; - if ((r = _info_run(MKNODES, name, NULL, &dminfo, NULL, NULL, 0, 0, 0, 0))) { - if (dminfo.exists) { - if (_lv_has_mknode(lv)) - r = _dev_manager_lv_mknodes(lv); - } else - r = _dev_manager_lv_rmnodes(lv); - } + if (!(dmt = _setup_task_run(DM_DEVICE_MKNODES, &dminfo, name, NULL, 0, 0, 0, 0, 0, 0))) + return_0; + + if (dminfo.exists) { + if (_lv_has_mknode(lv)) + r = _dev_manager_lv_mknodes(lv); + } else + r = _dev_manager_lv_rmnodes(lv); + + dm_task_destroy(dmt); - dm_pool_free(lv->vg->cmd->mem, name); return r; }

1 0

master - dev_manager: extend setup_task
by Zdenek Kabelac 06 Dec '16

06 Dec '16

Gitweb: http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=5163b8f697191caf… Commit: 5163b8f697191caf13ae8192781f65b3f6335e74 Parent: e2c7e0ad1196cc9323b63807ed199bcf4b10807c Author: Zdenek Kabelac <zkabelac(a)redhat.com> AuthorDate: Thu Dec 1 14:56:32 2016 +0100 Committer: Zdenek Kabelac <zkabelac(a)redhat.com> CommitterDate: Mon Dec 5 17:11:49 2016 +0100 dev_manager: extend setup_task Integrate also query for inactive table and handle dm_task_run() and dm_task_get_info() (thus switching to setup_task_run) Add one exception case for DM_DEVICE_TARGET_MSG. This allows further shortening and simplification of all other users of this function. --- WHATS_NEW | 1 + lib/activate/dev_manager.c | 185 ++++++++++++++++---------------------------- 2 files changed, 67 insertions(+), 119 deletions(-) diff --git a/WHATS_NEW b/WHATS_NEW index df4c49e..eb64542 100644 --- a/WHATS_NEW +++ b/WHATS_NEW @@ -1,5 +1,6 @@ Version 2.02.169 - ===================================== + Better API for internal function _setup_task_run. Avoid using lv_has_target_type() call within lv_info_with_seg_status. Simplify internal lv_info_with_seg_status API. Decide which status is needed in one place for lv_info_with_seg_status. diff --git a/lib/activate/dev_manager.c b/lib/activate/dev_manager.c index a72d5a9..a90b6c0 100644 --- a/lib/activate/dev_manager.c +++ b/lib/activate/dev_manager.c @@ -81,12 +81,16 @@ int read_only_lv(const struct logical_volume *lv, const struct lv_activate_opts /* * Low level device-layer operations. + * + * Unless task is DM_DEVICE_TARGET_MSG, also calls dm_task_run() */ -static struct dm_task *_setup_task(const char *name, const char *uuid, - uint32_t *event_nr, int task, - uint32_t major, uint32_t minor, - int with_open_count, - int with_flush) +static struct dm_task *_setup_task_run(int task, struct dm_info *info, + const char *name, const char *uuid, + uint32_t *event_nr, + uint32_t major, uint32_t minor, + int with_open_count, + int with_flush, + int query_inactive) { struct dm_task *dmt; @@ -108,14 +112,29 @@ static struct dm_task *_setup_task(const char *name, const char *uuid, if (activation_checks() && !dm_task_enable_checks(dmt)) goto_out; + if (query_inactive && !dm_task_query_inactive_table(dmt)) { + log_error("Failed to set query_inactive_table."); + goto out; + } + if (!with_open_count && !dm_task_no_open_count(dmt)) log_warn("WARNING: Failed to disable open_count."); if (!with_flush && !dm_task_no_flush(dmt)) log_warn("WARNING: Failed to set no_flush."); + if (task == DM_DEVICE_TARGET_MSG) + return dmt; /* TARGET_MSG needs more local tweaking before task_run() */ + + if (!dm_task_run(dmt)) + goto_out; + + if (info && !dm_task_get_info(dmt, info)) + goto_out; + return dmt; - out: + +out: dm_task_destroy(dmt); return NULL; } @@ -225,16 +244,11 @@ static int _info_run(info_type_t type, const char *name, const char *dlid, return 0; } - if (!(dmt = _setup_task((type == MKNODES) ? name : NULL, dlid, 0, dmtask, - major, minor, with_open_count, with_flush))) + if (!(dmt = _setup_task_run(dmtask, dminfo, + (type == MKNODES) ? name : NULL, dlid, 0, + major, minor, with_open_count, with_flush, 0))) return_0; - if (!dm_task_run(dmt)) - goto_out; - - if (!dm_task_get_info(dmt, dminfo)) - goto_out; - if (with_read_ahead && dminfo->exists) { if (!dm_task_get_read_ahead(dmt, read_ahead)) goto_out; @@ -357,11 +371,8 @@ static int _ignore_blocked_mirror_devices(struct device *dev, * We avoid another system call if we can, but if a device is * dead, we have no choice but to look up the table too. */ - if (!(dmt = _setup_task(NULL, NULL, NULL, DM_DEVICE_TABLE, - MAJOR(dev->dev), MINOR(dev->dev), 0, 1))) - goto_out; - - if (!dm_task_run(dmt)) + if (!(dmt = _setup_task_run(DM_DEVICE_TABLE, NULL, NULL, NULL, NULL, + MAJOR(dev->dev), MINOR(dev->dev), 0, 1, 0))) goto_out; do { @@ -397,22 +408,15 @@ static int _device_is_suspended(int major, int minor) { struct dm_task *dmt; struct dm_info info; - int r = 0; - if (!(dmt = _setup_task(NULL, NULL, NULL, DM_DEVICE_INFO, - major, minor, 0, 0))) + if (!(dmt = _setup_task_run(DM_DEVICE_INFO, &info, + NULL, NULL, NULL, + major, minor, 0, 0, 0))) return_0; - if (!dm_task_run(dmt) || - !dm_task_get_info(dmt, &info)) { - log_error("Failed to get info for device %d:%d", major, minor); - goto out; - } - - r = info.exists && info.suspended; -out: dm_task_destroy(dmt); - return r; + + return (info.exists && info.suspended); } static int _ignore_suspended_snapshot_component(struct device *dev) @@ -424,15 +428,11 @@ static int _ignore_suspended_snapshot_component(struct device *dev) int major1, minor1, major2, minor2; int r = 0; - if (!(dmt = _setup_task(NULL, NULL, NULL, DM_DEVICE_TABLE, - MAJOR(dev->dev), MINOR(dev->dev), 0, 1))) + if (!(dmt = _setup_task_run(DM_DEVICE_TABLE, NULL, + NULL, NULL, NULL, + MAJOR(dev->dev), MINOR(dev->dev), 0, 1, 0))) return_0; - if (!dm_task_run(dmt)) { - log_error("Failed to get state of snapshot or snapshot origin device."); - goto out; - } - do { next = dm_get_next_target(dmt, next, &start, &length, &target_type, &params); if (!target_type || !strcmp(target_type, TARGET_NAME_SNAPSHOT)) { @@ -471,14 +471,10 @@ static int _ignore_unusable_thins(struct device *dev) if (!(mem = dm_pool_create("unusable_thins", 128))) return_0; - if (!(dmt = _setup_task(NULL, NULL, NULL, DM_DEVICE_TABLE, - MAJOR(dev->dev), MINOR(dev->dev), 0, 1))) + if (!(dmt = _setup_task_run(DM_DEVICE_TABLE, NULL, NULL, NULL, NULL, + MAJOR(dev->dev), MINOR(dev->dev), 0, 1, 0))) goto_out; - if (!dm_task_run(dmt)) { - log_error("Failed to get state of mapped device."); - goto out; - } dm_get_next_target(dmt, next, &start, &length, &target_type, &params); if (!params || sscanf(params, "%d:%d", &major, &minor) != 2) { log_error("Failed to get thin-pool major:minor for thin device %d:%d.", @@ -487,15 +483,10 @@ static int _ignore_unusable_thins(struct device *dev) } dm_task_destroy(dmt); - if (!(dmt = _setup_task(NULL, NULL, NULL, DM_DEVICE_STATUS, - major, minor, 0, 0))) + if (!(dmt = _setup_task_run(DM_DEVICE_STATUS, NULL, NULL, NULL, NULL, + major, minor, 0, 0, 0))) goto_out; - if (!dm_task_run(dmt)) { - log_error("Failed to get state of mapped device."); - goto out; - } - dm_get_next_target(dmt, next, &start, &length, &target_type, &params); if (!dm_get_status_thin_pool(mem, params, &status)) goto_out; @@ -544,18 +535,10 @@ int device_is_usable(struct device *dev, struct dev_usable_check_params check) int only_error_target = 1; int r = 0; - if (!(dmt = _setup_task(NULL, NULL, NULL, DM_DEVICE_STATUS, - MAJOR(dev->dev), MINOR(dev->dev), 0, 0))) + if (!(dmt = _setup_task_run(DM_DEVICE_STATUS, &info, NULL, NULL, NULL, + MAJOR(dev->dev), MINOR(dev->dev), 0, 0, 0))) return_0; - if (!dm_task_run(dmt)) { - log_error("Failed to get state of mapped device"); - goto out; - } - - if (!dm_task_get_info(dmt, &info)) - goto_out; - if (!info.exists) goto out; @@ -810,29 +793,20 @@ int lv_has_target_type(struct dm_pool *mem, const struct logical_volume *lv, if (!(dlid = build_dm_uuid(mem, lv, layer))) return_0; - if (!(dmt = _setup_task(NULL, dlid, 0, DM_DEVICE_STATUS, 0, 0, 0, 0))) + if (!(dmt = _setup_task_run(DM_DEVICE_STATUS, &info, NULL, dlid, 0, 0, 0, 0, 0, 0))) goto_bad; - if (!dm_task_run(dmt)) - goto_out; - - if (!dm_task_get_info(dmt, &info) || !info.exists) + if (!info.exists) goto_out; /* If there is a preloaded table, use that in preference. */ if (info.inactive_table) { dm_task_destroy(dmt); - if (!(dmt = _setup_task(NULL, dlid, 0, DM_DEVICE_STATUS, 0, 0, 0, 0))) + if (!(dmt = _setup_task_run(DM_DEVICE_STATUS, &info, NULL, dlid, 0, 0, 0, 0, 0, 1))) goto_bad; - if (!dm_task_query_inactive_table(dmt)) - goto_out; - - if (!dm_task_run(dmt)) - goto_out; - - if (!dm_task_get_info(dmt, &info) || !info.exists || !info.inactive_table) + if (!info.exists || !info.inactive_table) goto_out; } @@ -868,29 +842,20 @@ static int _thin_lv_has_device_id(struct dm_pool *mem, const struct logical_volu if (!(dlid = build_dm_uuid(mem, lv, layer))) return_0; - if (!(dmt = _setup_task(NULL, dlid, 0, DM_DEVICE_TABLE, 0, 0, 0, 1))) + if (!(dmt = _setup_task_run(DM_DEVICE_TABLE, &info, NULL, dlid, 0, 0, 0, 0, 1, 0))) goto_bad; - if (!dm_task_run(dmt)) - goto_out; - - if (!dm_task_get_info(dmt, &info) || !info.exists) + if (!info.exists) goto_out; /* If there is a preloaded table, use that in preference. */ if (info.inactive_table) { dm_task_destroy(dmt); - if (!(dmt = _setup_task(NULL, dlid, 0, DM_DEVICE_TABLE, 0, 0, 0, 1))) + if (!(dmt = _setup_task_run(DM_DEVICE_TABLE, &info, NULL, dlid, 0, 0, 0, 0, 1, 1))) goto_bad; - if (!dm_task_query_inactive_table(dmt)) - goto_out; - - if (!dm_task_run(dmt)) - goto_out; - - if (!dm_task_get_info(dmt, &info) || !info.exists || !info.inactive_table) + if (!info.exists || !info.inactive_table) goto_out; } @@ -993,14 +958,11 @@ static int _percent_run(struct dev_manager *dm, const char *name, if (!(segtype = get_segtype_from_string(dm->cmd, target_type))) return_0; - if (!(dmt = _setup_task(name, dlid, event_nr, - wait ? DM_DEVICE_WAITEVENT : DM_DEVICE_STATUS, 0, 0, 0, 0))) + if (!(dmt = _setup_task_run(wait ? DM_DEVICE_WAITEVENT : DM_DEVICE_STATUS, &info, + name, dlid, event_nr, 0, 0, 0, 0, 0))) return_0; - if (!dm_task_run(dmt)) - goto_out; - - if (!dm_task_get_info(dmt, &info) || !info.exists) + if (!info.exists) goto_out; if (event_nr) @@ -1111,13 +1073,10 @@ int dev_manager_transient(struct dev_manager *dm, const struct logical_volume *l if (!(dlid = build_dm_uuid(dm->mem, lv, layer))) return_0; - if (!(dmt = _setup_task(0, dlid, NULL, DM_DEVICE_STATUS, 0, 0, 0, 0))) + if (!(dmt = _setup_task_run(DM_DEVICE_STATUS, &info, NULL, dlid, NULL, 0, 0, 0, 0, 0))) return_0; - if (!dm_task_run(dmt)) - goto_out; - - if (!dm_task_get_info(dmt, &info) || !info.exists) + if (!info.exists) goto_out; do { @@ -1307,13 +1266,10 @@ int dev_manager_raid_status(struct dev_manager *dm, if (!(dlid = build_dm_uuid(dm->mem, lv, layer))) return_0; - if (!(dmt = _setup_task(NULL, dlid, 0, DM_DEVICE_STATUS, 0, 0, 0, 0))) + if (!(dmt = _setup_task_run(DM_DEVICE_STATUS, &info, NULL, dlid, 0, 0, 0, 0, 0, 0))) return_0; - if (!dm_task_run(dmt)) - goto_out; - - if (!dm_task_get_info(dmt, &info) || !info.exists) + if (!info.exists) goto_out; dm_get_next_target(dmt, NULL, &start, &length, &type, &params); @@ -1366,7 +1322,7 @@ int dev_manager_raid_message(struct dev_manager *dm, if (!(dlid = build_dm_uuid(dm->mem, lv, layer))) return_0; - if (!(dmt = _setup_task(NULL, dlid, 0, DM_DEVICE_TARGET_MSG, 0, 0, 0, 1))) + if (!(dmt = _setup_task_run(DM_DEVICE_TARGET_MSG, NULL, NULL, dlid, 0, 0, 0, 0, 1, 0))) return_0; if (!dm_task_set_message(dmt, msg)) @@ -1401,13 +1357,10 @@ int dev_manager_cache_status(struct dev_manager *dm, if (!(*status = dm_pool_zalloc(dm->mem, sizeof(struct lv_status_cache)))) return_0; - if (!(dmt = _setup_task(NULL, dlid, 0, DM_DEVICE_STATUS, 0, 0, 0, 0))) + if (!(dmt = _setup_task_run(DM_DEVICE_STATUS, &info, NULL, dlid, 0, 0, 0, 0, 0, 0))) return_0; - if (!dm_task_run(dmt)) - goto_out; - - if (!dm_task_get_info(dmt, &info) || !info.exists) + if (!info.exists) goto_out; dm_get_next_target(dmt, NULL, &start, &length, &type, &params); @@ -1465,13 +1418,10 @@ int dev_manager_thin_pool_status(struct dev_manager *dm, if (!(dlid = build_dm_uuid(dm->mem, lv, lv_layer(lv)))) return_0; - if (!(dmt = _setup_task(NULL, dlid, 0, DM_DEVICE_STATUS, 0, 0, 0, flush))) + if (!(dmt = _setup_task_run(DM_DEVICE_STATUS, &info, NULL, dlid, 0, 0, 0, 0, flush, 0))) return_0; - if (!dm_task_run(dmt)) - goto_out; - - if (!dm_task_get_info(dmt, &info) || !info.exists) + if (!info.exists) goto_out; dm_get_next_target(dmt, NULL, &start, &length, &type, &params); @@ -1549,13 +1499,10 @@ int dev_manager_thin_device_id(struct dev_manager *dm, if (!(dlid = build_dm_uuid(dm->mem, lv, lv_layer(lv)))) return_0; - if (!(dmt = _setup_task(NULL, dlid, 0, DM_DEVICE_TABLE, 0, 0, 0, 1))) + if (!(dmt = _setup_task_run(DM_DEVICE_TABLE, &info, NULL, dlid, 0, 0, 0, 0, 1, 0))) return_0; - if (!dm_task_run(dmt)) - goto_out; - - if (!dm_task_get_info(dmt, &info) || !info.exists) + if (!info.exists) goto_out; if (dm_get_next_target(dmt, NULL, &start, &length,

1 0

master - activation: optimize away lv_has_target_type
by Zdenek Kabelac 06 Dec '16

06 Dec '16

Gitweb: http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=e2c7e0ad1196cc93… Commit: e2c7e0ad1196cc9323b63807ed199bcf4b10807c Parent: 6fd20be629b6dceb96702c5f875a2e612e6f465d Author: Zdenek Kabelac <zkabelac(a)redhat.com> AuthorDate: Mon Dec 5 15:23:18 2016 +0100 Committer: Zdenek Kabelac <zkabelac(a)redhat.com> CommitterDate: Mon Dec 5 17:10:14 2016 +0100 activation: optimize away lv_has_target_type It's actually not needed to call extra lv_has_target_type() to detect snapshot merge is in progress - decode this right during status capturing and save even few extra ioctl calls. --- WHATS_NEW | 1 + lib/activate/activate.c | 50 ++++++++++++++++++++++++++++------------------ 2 files changed, 31 insertions(+), 20 deletions(-) diff --git a/WHATS_NEW b/WHATS_NEW index a097477..df4c49e 100644 --- a/WHATS_NEW +++ b/WHATS_NEW @@ -1,5 +1,6 @@ Version 2.02.169 - ===================================== + Avoid using lv_has_target_type() call within lv_info_with_seg_status. Simplify internal lv_info_with_seg_status API. Decide which status is needed in one place for lv_info_with_seg_status. Fix matching of LV segment when checking for it info status. diff --git a/lib/activate/activate.c b/lib/activate/activate.c index afdf518..3128f36 100644 --- a/lib/activate/activate.c +++ b/lib/activate/activate.c @@ -757,7 +757,7 @@ int lv_info_with_seg_status(struct cmd_context *cmd, struct lv_with_info_and_seg_status *status, int with_open_count, int with_read_ahead) { - const struct logical_volume *lv = status->lv = lv_seg->lv; + const struct logical_volume *olv, *lv = status->lv = lv_seg->lv; if (!activation()) return 0; @@ -780,36 +780,46 @@ int lv_info_with_seg_status(struct cmd_context *cmd, status->info.exists = 0; /* So pool LV is not active */ } return 1; - } else if (lv_is_origin(lv) && - (!lv_is_merging_origin(lv) || - !lv_has_target_type(cmd->mem, lv, NULL, TARGET_NAME_SNAPSHOT_MERGE))) { + } else if (lv_is_origin(lv)) { /* Query segment status for 'layered' (-real) device most of the time, * only for merging snapshot, query its progress. * TODO: single LV may need couple status to be exposed at once.... * but this needs more logical background */ - /* Show INFO for actual origin */ - if (!_lv_info(cmd, lv, 0, &status->info, NULL, NULL, with_open_count, with_read_ahead)) + /* Show INFO for actual origin and grab status for merging origin */ + if (!_lv_info(cmd, lv, 0, &status->info, lv_seg, + lv_is_merging_origin(lv) ? &status->seg_status : NULL, + with_open_count, with_read_ahead)) return_0; - if (status->info.exists) + if (status->info.exists && + (status->seg_status.type != SEG_STATUS_SNAPSHOT)) /* Not merging */ /* Grab STATUS from layered -real */ (void) _lv_info(cmd, lv, 1, NULL, lv_seg, &status->seg_status, 0, 0); return 1; } else if (lv_is_cow(lv)) { - if (lv_is_merging_cow(lv) && - lv_has_target_type(cmd->mem, origin_from_cow(lv), NULL, TARGET_NAME_SNAPSHOT_MERGE)) { - /* - * When merge is in progress, query merging origin LV instead. - * COW volume is already mapped as error target in this case. - */ - status->lv = lv = origin_from_cow(lv); - lv_seg = first_seg(lv); - log_debug_activation("Snapshot merge is in progress, querying status of %s instead.", - display_lvname(lv)); - } else - /* Hadle fictional lvm2 snapshot and query snapshotX volume */ - lv_seg = find_snapshot(lv); + if (lv_is_merging_cow(lv)) { + olv = origin_from_cow(lv); + + if (!_lv_info(cmd, olv, 0, &status->info, first_seg(olv), &status->seg_status, + with_open_count, with_read_ahead)) + return_0; + + if (status->seg_status.type == SEG_STATUS_SNAPSHOT) { + log_debug_activation("Snapshot merge is in progress, querying status of %s instead.", + display_lvname(lv)); + /* + * When merge is in progress, query merging origin LV instead. + * COW volume is already mapped as error target in this case. + */ + status->lv = olv; + return 1; + } + + /* Merge not yet started, still a snapshot... */ + } + /* Hadle fictional lvm2 snapshot and query snapshotX volume */ + lv_seg = find_snapshot(lv); } return _lv_info(cmd, lv, 0, &status->info, lv_seg, &status->seg_status,

1 0

master - activation: lv_info_with_seg_status API change
by Zdenek Kabelac 06 Dec '16

06 Dec '16

Gitweb: http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=6fd20be629b6dceb… Commit: 6fd20be629b6dceb96702c5f875a2e612e6f465d Parent: ed93f0973a307634d1266aeab5c0d2bbceab086e Author: Zdenek Kabelac <zkabelac(a)redhat.com> AuthorDate: Mon Dec 5 14:31:25 2016 +0100 Committer: Zdenek Kabelac <zkabelac(a)redhat.com> CommitterDate: Mon Dec 5 17:09:47 2016 +0100 activation: lv_info_with_seg_status API change Drop LV from passed API arg - it's always segment being checked. Also use_layer is now in full control of lv_info_with_seg_status(). It decides which device needs to be checked to get 'the most info'. TODO: future version should be able to expose status from --- WHATS_NEW | 1 + lib/activate/activate.c | 47 +++++++++++++---------------------------------- lib/activate/activate.h | 16 +++++----------- lib/metadata/lv.c | 6 +++--- tools/reporter.c | 18 ++++++++---------- 5 files changed, 30 insertions(+), 58 deletions(-) diff --git a/WHATS_NEW b/WHATS_NEW index 2b0e822..a097477 100644 --- a/WHATS_NEW +++ b/WHATS_NEW @@ -1,5 +1,6 @@ Version 2.02.169 - ===================================== + Simplify internal lv_info_with_seg_status API. Decide which status is needed in one place for lv_info_with_seg_status. Fix matching of LV segment when checking for it info status. Report log_warn when status cannot be parsed. diff --git a/lib/activate/activate.c b/lib/activate/activate.c index 1b1f7e4..afdf518 100644 --- a/lib/activate/activate.c +++ b/lib/activate/activate.c @@ -741,32 +741,24 @@ int lv_info_by_lvid(struct cmd_context *cmd, const char *lvid_s, int use_layer, } /* - * Returns 1 if lv_seg_status structure populated, - * else 0 on failure or if device not active locally. - */ -int lv_status(struct cmd_context *cmd, const struct lv_segment *lv_seg, - int use_layer, struct lv_seg_status *lv_seg_status) -{ - if (!activation()) - return 0; - - return _lv_info(cmd, lv_seg->lv, use_layer, NULL, lv_seg, lv_seg_status, 0, 0); -} - -/* - * Returns 1 if lv_with_info_and_seg_status structure populated, + * Returns 1 if lv_with_info_and_seg_status info structure populated, * else 0 on failure or if device not active locally. * * When seg_status parsing had troubles it will set type to SEG_STATUS_UNKNOWN. * - * This is the same as calling lv_info and lv_status, - * but* it's done in one go with one ioctl if possible! ] + * Using usually one ioctl to obtain info and status. + * More complex segment do collect info from one device, + * but status from another device. + * + * TODO: further improve with more statuses (i.e. snapshot's origin/merge) */ -int lv_info_with_seg_status(struct cmd_context *cmd, const struct logical_volume *lv, - const struct lv_segment *lv_seg, int use_layer, +int lv_info_with_seg_status(struct cmd_context *cmd, + const struct lv_segment *lv_seg, struct lv_with_info_and_seg_status *status, int with_open_count, int with_read_ahead) { + const struct logical_volume *lv = status->lv = lv_seg->lv; + if (!activation()) return 0; @@ -774,23 +766,10 @@ int lv_info_with_seg_status(struct cmd_context *cmd, const struct logical_volume /* INFO is not set as cache-pool cannot be active. * STATUS is collected from cache LV */ lv_seg = get_only_segment_using_this_lv(lv); - (void) _lv_info(cmd, lv_seg->lv, use_layer, NULL, lv_seg, &status->seg_status, 0, 0); + (void) _lv_info(cmd, lv_seg->lv, 0, NULL, lv_seg, &status->seg_status, 0, 0); return 1; } - /* By default, take the first LV segment to report status for. */ - if (!lv_seg) - lv_seg = first_seg(lv); - - if (lv != lv_seg->lv) - /* - * If the info is requested for an LV and segment - * status for segment that belong to another LV, - * we need to acquire info and status separately! - */ - return _lv_info(cmd, lv, use_layer, &status->info, NULL, NULL, with_open_count, with_read_ahead) && - _lv_info(cmd, lv_seg->lv, use_layer, NULL, lv_seg, &status->seg_status, 0, 0); - if (lv_is_thin_pool(lv)) { /* Always collect status for '-tpool' */ if (_lv_info(cmd, lv, 1, &status->info, lv_seg, &status->seg_status, 0, 0) && @@ -824,7 +803,7 @@ int lv_info_with_seg_status(struct cmd_context *cmd, const struct logical_volume * When merge is in progress, query merging origin LV instead. * COW volume is already mapped as error target in this case. */ - lv = origin_from_cow(lv); + status->lv = lv = origin_from_cow(lv); lv_seg = first_seg(lv); log_debug_activation("Snapshot merge is in progress, querying status of %s instead.", display_lvname(lv)); @@ -833,7 +812,7 @@ int lv_info_with_seg_status(struct cmd_context *cmd, const struct logical_volume lv_seg = find_snapshot(lv); } - return _lv_info(cmd, lv, use_layer, &status->info, lv_seg, &status->seg_status, + return _lv_info(cmd, lv, 0, &status->info, lv_seg, &status->seg_status, with_open_count, with_read_ahead); } diff --git a/lib/activate/activate.h b/lib/activate/activate.h index 3922d78..db8d997 100644 --- a/lib/activate/activate.h +++ b/lib/activate/activate.h @@ -54,11 +54,12 @@ struct lv_seg_status { }; struct lv_with_info_and_seg_status { - const struct logical_volume *lv; /* input */ int info_ok; + const struct logical_volume *lv; /* output */ struct lvinfo info; /* output */ int seg_part_of_lv; /* output */ - struct lv_seg_status seg_status; /* input/output, see lv_seg_status */ + struct lv_seg_status seg_status; /* output, see lv_seg_status */ + /* TODO: add extra status for snapshot origin */ }; struct lv_activate_opts { @@ -133,21 +134,14 @@ int lv_info_by_lvid(struct cmd_context *cmd, const char *lvid_s, int use_layer, struct lvinfo *info, int with_open_count, int with_read_ahead); /* - * Returns 1 if lv_seg_status structure has been populated, - * else 0 on failure or if device not active locally. - */ -int lv_status(struct cmd_context *cmd, const struct lv_segment *lv_seg, - int use_layer, struct lv_seg_status *lv_seg_status); - -/* * Returns 1 if lv_info_and_seg_status structure has been populated, * else 0 on failure or if device not active locally. * * lv_info_with_seg_status is the same as calling lv_info and then lv_status, * but this fn tries to do that with one ioctl if possible. */ -int lv_info_with_seg_status(struct cmd_context *cmd, const struct logical_volume *lv, - const struct lv_segment *lv_seg, int use_layer, +int lv_info_with_seg_status(struct cmd_context *cmd, + const struct lv_segment *lv_seg, struct lv_with_info_and_seg_status *status, int with_open_count, int with_read_ahead); diff --git a/lib/metadata/lv.c b/lib/metadata/lv.c index eb5066a..087b564 100644 --- a/lib/metadata/lv.c +++ b/lib/metadata/lv.c @@ -255,7 +255,7 @@ char *lvseg_kernel_discards_dup(struct dm_pool *mem, const struct lv_segment *se if (!(status.seg_status.mem = dm_pool_create("reporter_pool", 1024))) return_NULL; - if (!(status.info_ok = lv_info_with_seg_status(seg->lv->vg->cmd, seg->lv, seg, 1, &status, 0, 0))) + if (!(status.info_ok = lv_info_with_seg_status(seg->lv->vg->cmd, seg, &status, 0, 0))) goto_bad; ret = lvseg_kernel_discards_dup_with_info_and_seg_status(mem, &status); @@ -407,7 +407,7 @@ dm_percent_t lvseg_percent_with_info_and_seg_status(const struct lv_with_info_an /* TODO: expose highest mapped sector */ p = DM_PERCENT_INVALID; else { - seg = first_seg(lvdm->lv); + seg = lvdm->seg_status.seg; /* Pool allocates whole chunk so round-up to nearest one */ csize = first_seg(seg->pool_lv)->chunk_size; csize = ((seg->lv->size + csize - 1) / csize) * csize; @@ -1318,7 +1318,7 @@ char *lv_attr_dup(struct dm_pool *mem, const struct logical_volume *lv) if (!(status.seg_status.mem = dm_pool_create("reporter_pool", 1024))) return_0; - if (!(status.info_ok = lv_info_with_seg_status(lv->vg->cmd, lv, first_seg(lv), 1, &status, 1, 1))) + if (!(status.info_ok = lv_info_with_seg_status(lv->vg->cmd, first_seg(lv), &status, 1, 1))) goto_bad; ret = lv_attr_dup_with_info_and_seg_status(mem, &status); diff --git a/tools/reporter.c b/tools/reporter.c index 0a9ca34..fc87c06 100644 --- a/tools/reporter.c +++ b/tools/reporter.c @@ -88,14 +88,13 @@ static int _vgs_single(struct cmd_context *cmd __attribute__((unused)), } static int _do_info_and_status(struct cmd_context *cmd, - const struct logical_volume *lv, const struct lv_segment *lv_seg, struct lv_with_info_and_seg_status *status, int do_info, int do_status) { - status->lv = lv; + status->lv = lv_seg->lv; - if (lv_is_historical(lv)) + if (lv_is_historical(lv_seg->lv)) return 1; if (do_status) { @@ -104,13 +103,12 @@ static int _do_info_and_status(struct cmd_context *cmd, if (do_info) /* both info and status */ - status->info_ok = lv_info_with_seg_status(cmd, lv, lv_seg, 0, status, 1, 1); + status->info_ok = lv_info_with_seg_status(cmd, lv_seg, status, 1, 1); else - /* status only */ - status->info_ok = lv_info_with_seg_status(cmd, lv, lv_seg, 0, status, 0, 0); + status->info_ok = lv_info_with_seg_status(cmd, lv_seg, status, 0, 0); } else if (do_info) /* info only */ - status->info_ok = lv_info(cmd, lv, 0, &status->info, 1, 1); + status->info_ok = lv_info(cmd, status->lv, 0, &status->info, 1, 1); return 1; } @@ -126,7 +124,7 @@ static int _do_lvs_with_info_and_status_single(struct cmd_context *cmd, }; int r = ECMD_FAILED; - if (!_do_info_and_status(cmd, lv, NULL, &status, do_info, do_status)) + if (!_do_info_and_status(cmd, first_seg(lv), &status, do_info, do_status)) goto_out; if (!report_object(sh ? : handle->custom_handle, sh != NULL, @@ -176,7 +174,7 @@ static int _do_segs_with_info_and_status_single(struct cmd_context *cmd, }; int r = ECMD_FAILED; - if (!_do_info_and_status(cmd, seg->lv, seg, &status, do_info, do_status)) + if (!_do_info_and_status(cmd, seg, &status, do_info, do_status)) goto_out; if (!report_object(sh ? : handle->custom_handle, sh != NULL, @@ -302,7 +300,7 @@ static int _do_pvsegs_sub_single(struct cmd_context *cmd, .lv = &_free_logical_volume }; - if (seg && !_do_info_and_status(cmd, seg->lv, seg, &status, do_info, do_status)) + if (seg && !_do_info_and_status(cmd, seg, &status, do_info, do_status)) goto_out; if (!report_object(sh ? : handle->custom_handle, sh != NULL,

1 0

master - activation: lv_info_with_seg_status unify status selection
by Zdenek Kabelac 06 Dec '16

06 Dec '16

Gitweb: http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=ed93f0973a307634… Commit: ed93f0973a307634d1266aeab5c0d2bbceab086e Parent: 0089201588d65bf131d3fadd1cf5be67a562c4b1 Author: Zdenek Kabelac <zkabelac(a)redhat.com> AuthorDate: Mon Dec 5 10:20:42 2016 +0100 Committer: Zdenek Kabelac <zkabelac(a)redhat.com> CommitterDate: Mon Dec 5 17:09:13 2016 +0100 activation: lv_info_with_seg_status unify status selection Start moving selection of status taken for a LV into a single place. The logic for showing info & status has been spread over multiple places and were doing too complex decision going agains each other. Unify selection of status of origin & cow scanned device. TODO: in future we want to grab status for LV and layered LV and have both statuses present for display - i.e. when 'old snapshot' of thinLV is takes and there is ongoing merge - at some moment we are not capable to show all needed info. --- WHATS_NEW | 1 + lib/activate/activate.c | 82 ++++++++++++++++++++++++++++++-------------- lib/activate/dev_manager.c | 56 +++++++++++++++-------------- tools/reporter.c | 41 +++------------------- 4 files changed, 91 insertions(+), 89 deletions(-) diff --git a/WHATS_NEW b/WHATS_NEW index 0f5ed02..2b0e822 100644 --- a/WHATS_NEW +++ b/WHATS_NEW @@ -1,5 +1,6 @@ Version 2.02.169 - ===================================== + Decide which status is needed in one place for lv_info_with_seg_status. Fix matching of LV segment when checking for it info status. Report log_warn when status cannot be parsed. Test segment type before accessing segment members when checking status. diff --git a/lib/activate/activate.c b/lib/activate/activate.c index efb8912..1b1f7e4 100644 --- a/lib/activate/activate.c +++ b/lib/activate/activate.c @@ -688,32 +688,6 @@ static int _lv_info(struct cmd_context *cmd, const struct logical_volume *lv, if (seg_status) { /* TODO: for now it's mess with seg_status */ seg_status->seg = seg; - if (lv_is_merging_cow(lv)) { - if (lv_has_target_type(cmd->mem, origin_from_cow(lv), NULL, TARGET_NAME_SNAPSHOT_MERGE)) { - /* - * When the snapshot-merge has not yet started, query COW LVs as is. - * When merge is in progress, query merging origin LV instead. - * COW volume is already mapped as error target in this case. - */ - lv = origin_from_cow(lv); - seg_status->seg = first_seg(lv); - log_debug_activation("Snapshot merge is in progress, querying status of %s instead.", - display_lvname(lv)); - } - } else if (!use_layer && lv_is_origin(lv) && !lv_is_external_origin(lv)) { - /* - * Query status for 'layered' (-real) device most of the time, - * only when snapshot merge started, query its progress. - * TODO: single LV may need couple status to be exposed at once.... - * but this needs more logical background - */ - if (!lv_is_merging_origin(lv) || - !lv_has_target_type(cmd->mem, origin_from_cow(lv), NULL, TARGET_NAME_SNAPSHOT_MERGE)) - use_layer = 1; - } else if (lv_is_cow(lv)) { - /* Hadle fictional lvm2 snapshot and query snapshotX volume */ - seg_status->seg = find_snapshot(lv); - } } if (!dev_manager_info(cmd, lv, @@ -783,6 +757,8 @@ int lv_status(struct cmd_context *cmd, const struct lv_segment *lv_seg, * Returns 1 if lv_with_info_and_seg_status structure populated, * else 0 on failure or if device not active locally. * + * When seg_status parsing had troubles it will set type to SEG_STATUS_UNKNOWN. + * * This is the same as calling lv_info and lv_status, * but* it's done in one go with one ioctl if possible! ] */ @@ -794,6 +770,18 @@ int lv_info_with_seg_status(struct cmd_context *cmd, const struct logical_volume if (!activation()) return 0; + if (lv_is_used_cache_pool(lv)) { + /* INFO is not set as cache-pool cannot be active. + * STATUS is collected from cache LV */ + lv_seg = get_only_segment_using_this_lv(lv); + (void) _lv_info(cmd, lv_seg->lv, use_layer, NULL, lv_seg, &status->seg_status, 0, 0); + return 1; + } + + /* By default, take the first LV segment to report status for. */ + if (!lv_seg) + lv_seg = first_seg(lv); + if (lv != lv_seg->lv) /* * If the info is requested for an LV and segment @@ -803,6 +791,48 @@ int lv_info_with_seg_status(struct cmd_context *cmd, const struct logical_volume return _lv_info(cmd, lv, use_layer, &status->info, NULL, NULL, with_open_count, with_read_ahead) && _lv_info(cmd, lv_seg->lv, use_layer, NULL, lv_seg, &status->seg_status, 0, 0); + if (lv_is_thin_pool(lv)) { + /* Always collect status for '-tpool' */ + if (_lv_info(cmd, lv, 1, &status->info, lv_seg, &status->seg_status, 0, 0) && + (status->seg_status.type == SEG_STATUS_THIN_POOL)) { + /* There is -tpool device, but query 'active' state of 'fake' thin-pool */ + if (!_lv_info(cmd, lv, 0, NULL, NULL, NULL, 0, 0) && + !status->seg_status.thin_pool->needs_check) + status->info.exists = 0; /* So pool LV is not active */ + } + return 1; + } else if (lv_is_origin(lv) && + (!lv_is_merging_origin(lv) || + !lv_has_target_type(cmd->mem, lv, NULL, TARGET_NAME_SNAPSHOT_MERGE))) { + /* Query segment status for 'layered' (-real) device most of the time, + * only for merging snapshot, query its progress. + * TODO: single LV may need couple status to be exposed at once.... + * but this needs more logical background + */ + /* Show INFO for actual origin */ + if (!_lv_info(cmd, lv, 0, &status->info, NULL, NULL, with_open_count, with_read_ahead)) + return_0; + + if (status->info.exists) + /* Grab STATUS from layered -real */ + (void) _lv_info(cmd, lv, 1, NULL, lv_seg, &status->seg_status, 0, 0); + return 1; + } else if (lv_is_cow(lv)) { + if (lv_is_merging_cow(lv) && + lv_has_target_type(cmd->mem, origin_from_cow(lv), NULL, TARGET_NAME_SNAPSHOT_MERGE)) { + /* + * When merge is in progress, query merging origin LV instead. + * COW volume is already mapped as error target in this case. + */ + lv = origin_from_cow(lv); + lv_seg = first_seg(lv); + log_debug_activation("Snapshot merge is in progress, querying status of %s instead.", + display_lvname(lv)); + } else + /* Hadle fictional lvm2 snapshot and query snapshotX volume */ + lv_seg = find_snapshot(lv); + } + return _lv_info(cmd, lv, use_layer, &status->info, lv_seg, &status->seg_status, with_open_count, with_read_ahead); } diff --git a/lib/activate/dev_manager.c b/lib/activate/dev_manager.c index 56c6d1d..a72d5a9 100644 --- a/lib/activate/dev_manager.c +++ b/lib/activate/dev_manager.c @@ -124,42 +124,41 @@ static int _get_segment_status_from_target_params(const char *target_name, const char *params, struct lv_seg_status *seg_status) { - struct segment_type *segtype; + const struct lv_segment *seg = seg_status->seg; + const struct segment_type *segtype = seg->segtype; seg_status->type = SEG_STATUS_UNKNOWN; /* Parsing failed */ + /* Switch to snapshot segtype status logic for merging origin */ + /* This is 'dynamic' decision, both states are valid */ + if (lv_is_merging_origin(seg->lv)) { + if (!strcmp(target_name, TARGET_NAME_SNAPSHOT_ORIGIN)) { + seg_status->type = SEG_STATUS_NONE; + return 1; /* Merge has not yet started */ + } + if (!strcmp(target_name, TARGET_NAME_SNAPSHOT_MERGE) && + !(segtype = get_segtype_from_string(seg->lv->vg->cmd, TARGET_NAME_SNAPSHOT))) + return_0; + /* Merging, parse 'snapshot' status of merge progress */ + } + if (!params) { log_warn("WARNING: Cannot find matching %s segment for %s.", - seg_status->seg->segtype->name, - display_lvname(seg_status->seg->lv)); + segtype->name, display_lvname(seg_status->seg->lv)); return 0; } - /* - * TODO: Add support for other segment types too! - * The segment to report status for must be properly - * selected for all the other types - mainly make sure - * linear/striped, old snapshots and raids have proper - * segment selected for status! - */ - if (!strcmp(target_name, TARGET_NAME_SNAPSHOT_MERGE) && - lv_is_merging_origin(seg_status->seg->lv)) { - /* Snapshot merge has started, check snapshot status */ - if (!(segtype = get_segtype_from_string(seg_status->seg->lv->vg->cmd, TARGET_NAME_SNAPSHOT))) - return_0; - } else { - if (!(segtype = get_segtype_from_string(seg_status->seg->lv->vg->cmd, target_name))) - return_0; - - /* Validate segtype from DM table and lvm2 metadata */ - if (segtype != seg_status->seg->segtype) { - log_warn("WARNING: segment type %s found does not match " - "expected segment type %s.", - segtype->name, seg_status->seg->segtype->name); - return 0; - } + /* Validate target_name segtype from DM table with lvm2 metadata segtype */ + if (strcmp(segtype->name, target_name) && + /* If kernel's type isn't an exact match is it compatible? */ + (!segtype->ops->target_status_compatible || + !segtype->ops->target_status_compatible(target_name))) { + log_warn(INTERNAL_ERROR "WARNING: Segment type %s found does not match expected type %s for %s.", + target_name, segtype->name, display_lvname(seg_status->seg->lv)); + return 0; } + /* TODO: move into segtype method */ if (segtype_is_cache(segtype)) { if (!dm_get_status_cache(seg_status->mem, params, &(seg_status->cache))) return_0; @@ -181,7 +180,10 @@ static int _get_segment_status_from_target_params(const char *target_name, return_0; seg_status->type = SEG_STATUS_SNAPSHOT; } else - /* Status not supported */ + /* + * TODO: Add support for other segment types too! + * Status not supported + */ seg_status->type = SEG_STATUS_NONE; return 1; diff --git a/tools/reporter.c b/tools/reporter.c index b7a2f39..0a9ca34 100644 --- a/tools/reporter.c +++ b/tools/reporter.c @@ -87,32 +87,12 @@ static int _vgs_single(struct cmd_context *cmd __attribute__((unused)), return ECMD_PROCESSED; } -static void _choose_lv_segment_for_status_report(const struct logical_volume *lv, const struct lv_segment **lv_seg) -{ - if (lv_is_used_cache_pool(lv)) { - /* For a used cache pool, choose cache volume segment */ - *lv_seg = get_only_segment_using_this_lv(lv); - return; - } - - /* - * By default, take the first LV segment to report status for. - * If there's any other specific segment that needs to be - * reported instead for the LV, choose it here and assign it - * to lvdm->seg_status->seg. This is the segment whose - * status line will be used for report exactly. - */ - *lv_seg = first_seg(lv); -} - static int _do_info_and_status(struct cmd_context *cmd, const struct logical_volume *lv, const struct lv_segment *lv_seg, struct lv_with_info_and_seg_status *status, int do_info, int do_status) { - unsigned use_layer = lv_is_thin_pool(lv) ? 1 : 0; - status->lv = lv; if (lv_is_historical(lv)) @@ -121,27 +101,16 @@ static int _do_info_and_status(struct cmd_context *cmd, if (do_status) { if (!(status->seg_status.mem = dm_pool_create("reporter_pool", 1024))) return_0; - if (!lv_seg || seg_is_used_cache_pool(lv_seg)) - _choose_lv_segment_for_status_report(lv, &lv_seg); - if (do_info) { + if (do_info) /* both info and status */ - status->info_ok = lv_info_with_seg_status(cmd, lv, lv_seg, use_layer, status, 1, 1); - /* for inactive thin-pools reset lv info struct */ - if (use_layer && status->info_ok && - !lv_info(cmd, lv, 0, NULL, 0, 0)) - memset(&status->info, 0, sizeof(status->info)); - /* for inactive cache reset lvinfo for its struct for cache-pool */ - if (lv_is_used_cache_pool(lv) && !status->info_ok) { - memset(&status->info, 0, sizeof(status->info)); - status->info_ok = 1; - } - } else + status->info_ok = lv_info_with_seg_status(cmd, lv, lv_seg, 0, status, 1, 1); + else /* status only */ - status->info_ok = lv_status(cmd, lv_seg, use_layer, &status->seg_status); + status->info_ok = lv_info_with_seg_status(cmd, lv, lv_seg, 0, status, 0, 0); } else if (do_info) /* info only */ - status->info_ok = lv_info(cmd, lv, use_layer, &status->info, 1, 1); + status->info_ok = lv_info(cmd, lv, 0, &status->info, 1, 1); return 1; }

1 0

master - cleanup: swap test order
by Zdenek Kabelac 06 Dec '16

06 Dec '16

Gitweb: http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=0089201588d65bf1… Commit: 0089201588d65bf131d3fadd1cf5be67a562c4b1 Parent: 5ba2d58d28ace811c33e7e66b777fcca3f42149e Author: Zdenek Kabelac <zkabelac(a)redhat.com> AuthorDate: Wed Nov 30 13:43:43 2016 +0100 Committer: Zdenek Kabelac <zkabelac(a)redhat.com> CommitterDate: Mon Dec 5 17:05:23 2016 +0100 cleanup: swap test order Check for lv != lvseg->lv. Make the logic of following patches look better and easier to read. --- lib/activate/activate.c | 22 +++++++++++----------- 1 files changed, 11 insertions(+), 11 deletions(-) diff --git a/lib/activate/activate.c b/lib/activate/activate.c index 8fc45b7..efb8912 100644 --- a/lib/activate/activate.c +++ b/lib/activate/activate.c @@ -794,17 +794,17 @@ int lv_info_with_seg_status(struct cmd_context *cmd, const struct logical_volume if (!activation()) return 0; - if (lv == lv_seg->lv) - return _lv_info(cmd, lv, use_layer, &status->info, lv_seg, &status->seg_status, - with_open_count, with_read_ahead); + if (lv != lv_seg->lv) + /* + * If the info is requested for an LV and segment + * status for segment that belong to another LV, + * we need to acquire info and status separately! + */ + return _lv_info(cmd, lv, use_layer, &status->info, NULL, NULL, with_open_count, with_read_ahead) && + _lv_info(cmd, lv_seg->lv, use_layer, NULL, lv_seg, &status->seg_status, 0, 0); - /* - * If the info is requested for an LV and segment - * status for segment that belong to another LV, - * we need to acquire info and status separately! - */ - return _lv_info(cmd, lv, use_layer, &status->info, NULL, NULL, with_open_count, with_read_ahead) && - _lv_info(cmd, lv_seg->lv, use_layer, NULL, lv_seg, &status->seg_status, 0, 0); + return _lv_info(cmd, lv, use_layer, &status->info, lv_seg, &status->seg_status, + with_open_count, with_read_ahead); } #define OPEN_COUNT_CHECK_RETRIES 25

1 0

master - activation: improve error handling for status reading
by Zdenek Kabelac 06 Dec '16

06 Dec '16

Gitweb: http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=5ba2d58d28ace811… Commit: 5ba2d58d28ace811c33e7e66b777fcca3f42149e Parent: 4a4b22e11413ac76164fbdd9b3c353151c8dde70 Author: Zdenek Kabelac <zkabelac(a)redhat.com> AuthorDate: Thu Dec 1 10:37:03 2016 +0100 Committer: Zdenek Kabelac <zkabelac(a)redhat.com> CommitterDate: Mon Dec 5 17:05:17 2016 +0100 activation: improve error handling for status reading When lvm2 wants to see a status, it needs to validate, segment for status reading is matching whan lvm2 expects in metadata. Also ensure status failure will not cause '0' from info reading when actual info was collected properly. Failure in 'status' reading is considered to be a 'log_warn()' event only. --- WHATS_NEW | 1 + lib/activate/dev_manager.c | 28 +++++++++++++++------------- 2 files changed, 16 insertions(+), 13 deletions(-) diff --git a/WHATS_NEW b/WHATS_NEW index 3aedebb..0f5ed02 100644 --- a/WHATS_NEW +++ b/WHATS_NEW @@ -1,5 +1,6 @@ Version 2.02.169 - ===================================== + Fix matching of LV segment when checking for it info status. Report log_warn when status cannot be parsed. Test segment type before accessing segment members when checking status. Implement compatible target function for stripe segment. diff --git a/lib/activate/dev_manager.c b/lib/activate/dev_manager.c index 8ace08c..56c6d1d 100644 --- a/lib/activate/dev_manager.c +++ b/lib/activate/dev_manager.c @@ -203,9 +203,8 @@ static int _info_run(info_type_t type, const char *name, const char *dlid, struct dm_task *dmt; int dmtask; void *target = NULL; - uint64_t target_start, target_length; - char *target_name, *target_params, *params_to_process = NULL; - uint32_t extent_size; + uint64_t target_start, target_length, start, length; + char *target_name, *target_params; int with_flush = 1; /* TODO: arg for _info_run */ switch (type) { @@ -240,21 +239,24 @@ static int _info_run(info_type_t type, const char *name, const char *dlid, } else if (read_ahead) *read_ahead = DM_READ_AHEAD_NONE; - if (type == STATUS) { - extent_size = seg_status->seg->lv->vg->extent_size; + /* Query status only for active device */ + if ((type == STATUS) && dminfo->exists) { + start = length = seg_status->seg->lv->vg->extent_size; + start *= seg_status->seg->le; + length *= seg_status->seg->len; + do { target = dm_get_next_target(dmt, target, &target_start, &target_length, &target_name, &target_params); - if (((uint64_t) seg_status->seg->le * extent_size == target_start) && - ((uint64_t) seg_status->seg->len * extent_size == target_length)) { - params_to_process = target_params; - break; - } + + if ((start == target_start) && (length == target_length)) + break; /* Keep target_params when matching segment is found */ + + target_params = NULL; /* Marking this target_params unusable */ } while (target); - if (params_to_process && - !_get_segment_status_from_target_params(target_name, params_to_process, seg_status)) - goto_out; + if (!_get_segment_status_from_target_params(target_name, target_params, seg_status)) + stack; } r = 1;

1 0

master - activation: status check switch to warn
by Zdenek Kabelac 06 Dec '16

06 Dec '16

Gitweb: http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=4a4b22e11413ac76… Commit: 4a4b22e11413ac76164fbdd9b3c353151c8dde70 Parent: 325c2c5687e42afd2c8611e4cefd73f4fb7fb6ed Author: Zdenek Kabelac <zkabelac(a)redhat.com> AuthorDate: Fri Dec 2 13:57:52 2016 +0100 Committer: Zdenek Kabelac <zkabelac(a)redhat.com> CommitterDate: Mon Dec 5 17:04:24 2016 +0100 activation: status check switch to warn When we can't parse status, switch to warning as this is not considered an errornous case. LVS is not supposed to return error status code when device is not what it's been expected to be - but it should be WARNING a user there is something unexpected. --- WHATS_NEW | 1 + lib/activate/dev_manager.c | 29 +++++++++++++++-------------- 2 files changed, 16 insertions(+), 14 deletions(-) diff --git a/WHATS_NEW b/WHATS_NEW index fbf4f61..3aedebb 100644 --- a/WHATS_NEW +++ b/WHATS_NEW @@ -1,5 +1,6 @@ Version 2.02.169 - ===================================== + Report log_warn when status cannot be parsed. Test segment type before accessing segment members when checking status. Implement compatible target function for stripe segment. Use status info to report merge failed and snapshot invalid lvs fields. diff --git a/lib/activate/dev_manager.c b/lib/activate/dev_manager.c index b8b1235..8ace08c 100644 --- a/lib/activate/dev_manager.c +++ b/lib/activate/dev_manager.c @@ -126,7 +126,14 @@ static int _get_segment_status_from_target_params(const char *target_name, { struct segment_type *segtype; - seg_status->type = SEG_STATUS_UNKNOWN; + seg_status->type = SEG_STATUS_UNKNOWN; /* Parsing failed */ + + if (!params) { + log_warn("WARNING: Cannot find matching %s segment for %s.", + seg_status->seg->segtype->name, + display_lvname(seg_status->seg->lv)); + return 0; + } /* * TODO: Add support for other segment types too! @@ -141,19 +148,14 @@ static int _get_segment_status_from_target_params(const char *target_name, if (!(segtype = get_segtype_from_string(seg_status->seg->lv->vg->cmd, TARGET_NAME_SNAPSHOT))) return_0; } else { - if (strcmp(target_name, TARGET_NAME_CACHE) && - strcmp(target_name, TARGET_NAME_SNAPSHOT) && - strcmp(target_name, TARGET_NAME_THIN_POOL) && - strcmp(target_name, TARGET_NAME_THIN)) - return 1; /* TODO: Do not know how to handle yet */ - if (!(segtype = get_segtype_from_string(seg_status->seg->lv->vg->cmd, target_name))) return_0; + /* Validate segtype from DM table and lvm2 metadata */ if (segtype != seg_status->seg->segtype) { - log_error(INTERNAL_ERROR "_get_segment_status_from_target_params: " - "segment type %s found does not match expected segment type %s", - segtype->name, seg_status->seg->segtype->name); + log_warn("WARNING: segment type %s found does not match " + "expected segment type %s.", + segtype->name, seg_status->seg->segtype->name); return 0; } } @@ -178,10 +180,9 @@ static int _get_segment_status_from_target_params(const char *target_name, if (!dm_get_status_snapshot(seg_status->mem, params, &seg_status->snapshot)) return_0; seg_status->type = SEG_STATUS_SNAPSHOT; - } else { - log_error(INTERNAL_ERROR "Unsupported segment type %s.", segtype->name); - return 0; - } + } else + /* Status not supported */ + seg_status->type = SEG_STATUS_NONE; return 1; }

1 0

master - lv: always check status type
by Zdenek Kabelac 06 Dec '16

06 Dec '16

Gitweb: http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=325c2c5687e42afd… Commit: 325c2c5687e42afd2c8611e4cefd73f4fb7fb6ed Parent: 6a450952add6bc389537aec23fc727ef6e156671 Author: Zdenek Kabelac <zkabelac(a)redhat.com> AuthorDate: Thu Dec 1 10:43:55 2016 +0100 Committer: Zdenek Kabelac <zkabelac(a)redhat.com> CommitterDate: Mon Dec 5 17:03:38 2016 +0100 lv: always check status type Always validate status type has an expected value, before accessing status struct members. --- WHATS_NEW | 1 + lib/metadata/lv.c | 33 ++++++++++++++++----------------- lib/report/report.c | 12 ++++++++---- 3 files changed, 25 insertions(+), 21 deletions(-) diff --git a/WHATS_NEW b/WHATS_NEW index 541e3ec..fbf4f61 100644 --- a/WHATS_NEW +++ b/WHATS_NEW @@ -1,5 +1,6 @@ Version 2.02.169 - ===================================== + Test segment type before accessing segment members when checking status. Implement compatible target function for stripe segment. Use status info to report merge failed and snapshot invalid lvs fields. diff --git a/lib/metadata/lv.c b/lib/metadata/lv.c index 14370bb..eb5066a 100644 --- a/lib/metadata/lv.c +++ b/lib/metadata/lv.c @@ -1184,7 +1184,8 @@ char *lv_attr_dup_with_info_and_seg_status(struct dm_pool *mem, const struct lv_ if (lv_is_historical(lv)) { repstr[4] = 'h'; repstr[5] = '-'; - } else if (!activation() || !lvdm->info_ok) { + } else if (!activation() || !lvdm->info_ok || + (lvdm->seg_status.type == SEG_STATUS_UNKNOWN)) { repstr[4] = 'X'; /* Unknown */ repstr[5] = 'X'; /* Unknown */ } else if (lvdm->info.exists) { @@ -1216,8 +1217,10 @@ char *lv_attr_dup_with_info_and_seg_status(struct dm_pool *mem, const struct lv_ /* 'c' when cache/thin-pool is active with needs_check flag * 'C' for suspend */ if ((lv_is_thin_pool(lv) && + (lvdm->seg_status.type == SEG_STATUS_THIN_POOL) && lvdm->seg_status.thin_pool->needs_check) || (lv_is_cache(lv) && + (lvdm->seg_status.type == SEG_STATUS_CACHE) && lvdm->seg_status.cache->needs_check)) repstr[4] = lvdm->info.suspended ? 'C' : 'c'; @@ -1259,6 +1262,10 @@ char *lv_attr_dup_with_info_and_seg_status(struct dm_pool *mem, const struct lv_ repstr[7] = '-'; repstr[8] = '-'; + /* TODO: also convert raid health + * lv_is_raid_type() is to wide + * NOTE: snapshot origin is 'mostly' showing it's layered status + */ if (lv_is_partial(lv)) repstr[8] = 'p'; else if (lv_is_raid_type(lv)) { @@ -1272,31 +1279,23 @@ char *lv_attr_dup_with_info_and_seg_status(struct dm_pool *mem, const struct lv_ repstr[8] = 'm'; /* RAID has 'm'ismatches */ } else if (lv->status & LV_WRITEMOSTLY) repstr[8] = 'w'; /* sub-LV has 'w'ritemostly */ - } else if (lv_is_cache(lv) && - (lvdm->seg_status.type != SEG_STATUS_NONE)) { - if (lvdm->seg_status.type == SEG_STATUS_UNKNOWN) - repstr[8] = 'X'; /* Unknown */ - else if (lvdm->seg_status.cache->fail) + } else if (lvdm->seg_status.type == SEG_STATUS_CACHE) { + if (lvdm->seg_status.cache->fail) repstr[8] = 'F'; else if (lvdm->seg_status.cache->read_only) repstr[8] = 'M'; - } else if (lv_is_thin_pool(lv) && - (lvdm->seg_status.type != SEG_STATUS_NONE)) { - if (lvdm->seg_status.type == SEG_STATUS_UNKNOWN) - repstr[8] = 'X'; /* Unknown */ - else if (lvdm->seg_status.thin_pool->fail) + } else if (lvdm->seg_status.type == SEG_STATUS_THIN_POOL) { + if (lvdm->seg_status.thin_pool->fail) repstr[8] = 'F'; else if (lvdm->seg_status.thin_pool->out_of_data_space) repstr[8] = 'D'; else if (lvdm->seg_status.thin_pool->read_only) repstr[8] = 'M'; - } else if (lv_is_thin_volume(lv) && - (lvdm->seg_status.type != SEG_STATUS_NONE)) { - if (lvdm->seg_status.type == SEG_STATUS_UNKNOWN) - repstr[8] = 'X'; /* Unknown */ - else if (lvdm->seg_status.thin->fail) + } else if (lvdm->seg_status.type == SEG_STATUS_THIN) { + if (lvdm->seg_status.thin->fail) repstr[8] = 'F'; - } + } else if (lvdm->seg_status.type == SEG_STATUS_UNKNOWN) + repstr[8] = 'X'; /* Unknown */ if (lv->status & LV_ACTIVATION_SKIP) repstr[9] = 'k'; diff --git a/lib/report/report.c b/lib/report/report.c index 358ae94..702befb 100644 --- a/lib/report/report.c +++ b/lib/report/report.c @@ -2936,12 +2936,16 @@ static int _metadatapercent_disp(struct dm_report *rh, const void *data, void *private) { const struct lv_with_info_and_seg_status *lvdm = (const struct lv_with_info_and_seg_status *) data; - dm_percent_t percent = DM_PERCENT_INVALID; + dm_percent_t percent; - if (lv_is_thin_pool(lvdm->lv) || - lv_is_cache(lvdm->lv) || - lv_is_used_cache_pool(lvdm->lv)) + switch (lvdm->seg_status.type) { + case SEG_STATUS_CACHE: + case SEG_STATUS_THIN_POOL: percent = lvseg_percent_with_info_and_seg_status(lvdm, PERCENT_GET_METADATA); + break; + default: + percent = DM_PERCENT_INVALID; + } return dm_report_field_percent(rh, field, &percent); }

1 0

master - striped: implement compatible target name
by Zdenek Kabelac 06 Dec '16

06 Dec '16

Gitweb: http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=6a450952add6bc38… Commit: 6a450952add6bc389537aec23fc727ef6e156671 Parent: 254f73e3efe27e55637e8d3c23abf607564650d7 Author: Zdenek Kabelac <zkabelac(a)redhat.com> AuthorDate: Thu Dec 1 16:09:53 2016 +0100 Committer: Zdenek Kabelac <zkabelac(a)redhat.com> CommitterDate: Mon Dec 5 17:02:01 2016 +0100 striped: implement compatible target name Linear is handled by striped target. --- WHATS_NEW | 1 + lib/striped/striped.c | 6 ++++++ 2 files changed, 7 insertions(+), 0 deletions(-) diff --git a/WHATS_NEW b/WHATS_NEW index 82222d7..541e3ec 100644 --- a/WHATS_NEW +++ b/WHATS_NEW @@ -1,5 +1,6 @@ Version 2.02.169 - ===================================== + Implement compatible target function for stripe segment. Use status info to report merge failed and snapshot invalid lvs fields. Version 2.02.168 - 30th November 2016 diff --git a/lib/striped/striped.c b/lib/striped/striped.c index 7d19786..e1de9c9 100644 --- a/lib/striped/striped.c +++ b/lib/striped/striped.c @@ -159,6 +159,11 @@ static int _striped_merge_segments(struct lv_segment *seg1, struct lv_segment *s } #ifdef DEVMAPPER_SUPPORT +static int _striped_target_status_compatible(const char *type) +{ + return (strcmp(type, TARGET_NAME_LINEAR) == 0); +} + static int _striped_add_target_line(struct dev_manager *dm, struct dm_pool *mem __attribute__((unused)), struct cmd_context *cmd __attribute__((unused)), @@ -218,6 +223,7 @@ static struct segtype_handler _striped_ops = { .text_export = _striped_text_export, .merge_segments = _striped_merge_segments, #ifdef DEVMAPPER_SUPPORT + .target_status_compatible = _striped_target_status_compatible, .add_target_line = _striped_add_target_line, .target_present = _striped_target_present, #endif

1 0

master - snapshot: reporting uses statusinfo
by Zdenek Kabelac 06 Dec '16

06 Dec '16

Gitweb: http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=254f73e3efe27e55… Commit: 254f73e3efe27e55637e8d3c23abf607564650d7 Parent: 07f9889b53eb3314e3ca816ffbeb72d087e9282a Author: Zdenek Kabelac <zkabelac(a)redhat.com> AuthorDate: Thu Dec 1 17:58:06 2016 +0100 Committer: Zdenek Kabelac <zkabelac(a)redhat.com> CommitterDate: Mon Dec 5 17:01:15 2016 +0100 snapshot: reporting uses statusinfo Convert lvs -o lv_merge_failed,lv_snapshot_invalid to use lv_info_and_status function. This makes it equal to attr value showing this info (as they were different since they were derived from different data set and different logic as well). Also saves couple extra ioctl that were needed to obtain this info. --- WHATS_NEW | 1 + lib/report/columns.h | 4 ++-- lib/report/report.c | 20 ++++++++------------ 3 files changed, 11 insertions(+), 14 deletions(-) diff --git a/WHATS_NEW b/WHATS_NEW index 3e551ec..82222d7 100644 --- a/WHATS_NEW +++ b/WHATS_NEW @@ -1,5 +1,6 @@ Version 2.02.169 - ===================================== + Use status info to report merge failed and snapshot invalid lvs fields. Version 2.02.168 - 30th November 2016 ===================================== diff --git a/lib/report/columns.h b/lib/report/columns.h index ffc92c0..6d9b851 100644 --- a/lib/report/columns.h +++ b/lib/report/columns.h @@ -60,8 +60,6 @@ FIELD(LVS, lv, BIN, "Converting", lvid, 0, lvconverting, lv_converting, "Set if FIELD(LVS, lv, STR, "AllocPol", lvid, 10, lvallocationpolicy, lv_allocation_policy, "LV allocation policy.", 0) FIELD(LVS, lv, BIN, "AllocLock", lvid, 10, lvallocationlocked, lv_allocation_locked, "Set if LV is locked against allocation changes.", 0) FIELD(LVS, lv, BIN, "FixMin", lvid, 10, lvfixedminor, lv_fixed_minor, "Set if LV has fixed minor number assigned.", 0) -FIELD(LVS, lv, BIN, "MergeFailed", lvid, 15, lvmergefailed, lv_merge_failed, "Set if snapshot merge failed.", 0) -FIELD(LVS, lv, BIN, "SnapInvalid", lvid, 15, lvsnapshotinvalid, lv_snapshot_invalid, "Set if snapshot LV is invalid.", 0) FIELD(LVS, lv, BIN, "SkipAct", lvid, 15, lvskipactivation, lv_skip_activation, "Set if LV is skipped on activation.", 0) FIELD(LVS, lv, STR, "WhenFull", lvid, 15, lvwhenfull, lv_when_full, "For thin pools, behavior when full.", 0) FIELD(LVS, lv, STR, "Active", lvid, 0, lvactive, lv_active, "Active state of the LV.", 0) @@ -145,6 +143,8 @@ FIELD(LVSSTATUS, lv, STR, "KCachePolicy", lvid, 18, kernel_cache_policy, kernel_ FIELD(LVSSTATUS, lv, STR, "Health", lvid, 15, lvhealthstatus, lv_health_status, "LV health status.", 0) FIELD(LVSSTATUS, lv, STR, "KDiscards", lvid, 0, kdiscards, kernel_discards, "For thin pools, how discards are handled in kernel.", 0) FIELD(LVSSTATUS, lv, BIN, "CheckNeeded", lvid, 15, lvcheckneeded, lv_check_needed, "For thin pools and cache volumes, whether metadata check is needed.", 0) +FIELD(LVSSTATUS, lv, BIN, "MergeFailed", lvid, 15, lvmergefailed, lv_merge_failed, "Set if snapshot merge failed.", 0) +FIELD(LVSSTATUS, lv, BIN, "SnapInvalid", lvid, 15, lvsnapshotinvalid, lv_snapshot_invalid, "Set if snapshot LV is invalid.", 0) /* * End of LVSSTATUS type fields */ diff --git a/lib/report/report.c b/lib/report/report.c index 1ca4dfb..358ae94 100644 --- a/lib/report/report.c +++ b/lib/report/report.c @@ -3363,30 +3363,26 @@ static int _lvmergefailed_disp(struct dm_report *rh, struct dm_pool *mem, struct dm_report_field *field, const void *data, void *private) { - const struct logical_volume *lv = (const struct logical_volume *) data; - dm_percent_t snap_percent; - int merge_failed; + const struct lv_with_info_and_seg_status *lvdm = (const struct lv_with_info_and_seg_status *) data; - if (!lv_is_cow(lv) || !lv_snapshot_percent(lv, &snap_percent)) + if (lvdm->seg_status.type != SEG_STATUS_SNAPSHOT) return _binary_undef_disp(rh, mem, field, private); - merge_failed = snap_percent == LVM_PERCENT_MERGE_FAILED; - return _binary_disp(rh, mem, field, merge_failed, GET_FIRST_RESERVED_NAME(lv_merge_failed_y), private); + return _binary_disp(rh, mem, field, lvdm->seg_status.snapshot->merge_failed, + GET_FIRST_RESERVED_NAME(lv_merge_failed_y), private); } static int _lvsnapshotinvalid_disp(struct dm_report *rh, struct dm_pool *mem, struct dm_report_field *field, const void *data, void *private) { - const struct logical_volume *lv = (const struct logical_volume *) data; - dm_percent_t snap_percent; - int snap_invalid; + const struct lv_with_info_and_seg_status *lvdm = (const struct lv_with_info_and_seg_status *) data; - if (!lv_is_cow(lv)) + if (lvdm->seg_status.type != SEG_STATUS_SNAPSHOT) return _binary_undef_disp(rh, mem, field, private); - snap_invalid = !lv_snapshot_percent(lv, &snap_percent) || snap_percent == DM_PERCENT_INVALID; - return _binary_disp(rh, mem, field, snap_invalid, GET_FIRST_RESERVED_NAME(lv_snapshot_invalid_y), private); + return _binary_disp(rh, mem, field, lvdm->seg_status.snapshot->invalid, + GET_FIRST_RESERVED_NAME(lv_snapshot_invalid_y), private); } static int _lvsuspended_disp(struct dm_report *rh, struct dm_pool *mem,

1 0

master - report: order fields by type for field defintions in columns.h
by Peter Rajnoha 01 Dec '16

01 Dec '16

Gitweb: http://git.fedorahosted.org/git/?p=lvm2.git;a=commitdiff;h=07f9889b53eb3314… Commit: 07f9889b53eb3314e3ca816ffbeb72d087e9282a Parent: bb5eb324e3d503bf6dd8923531f63d3e1933ff0d Author: Peter Rajnoha <prajnoha(a)redhat.com> AuthorDate: Thu Dec 1 14:39:21 2016 +0100 Committer: Peter Rajnoha <prajnoha(a)redhat.com> CommitterDate: Thu Dec 1 14:55:29 2016 +0100 report: order fields by type for field defintions in columns.h When displaying <reporting_command> -o help, we'd like to have fields grouped nicely, not starting having groups interleaved as it was before. The code that displays the help output for fields takes the order as written in columns.h file - this caused output like: $ lvs -o help Logical Volume Fields --------------------- ...field list... Logical Volume Device Info and Status Combined Fields ----------------------------------------------------- ...field list... Logical Volume Fields --------------------- ...field list... Logical Volume Device Status Fields ----------------------------------- ...field list... Logical Volume Fields --------------------- ...field list... Instead, let's have it without groups interleaved which may be a bit confusing, so: Logical Volume Fields --------------------- ...field list... Logical Volume Device Status Fields ----------------------------------- ...field list... Logical Volume Device Info and Status Combined Fields ----------------------------------------------------- ...field list... ..and so on. --- lib/report/columns.h | 77 ++++++++++++++++++++++++++++++++++++++++++++++---- 1 files changed, 71 insertions(+), 6 deletions(-) diff --git a/lib/report/columns.h b/lib/report/columns.h index 212f9ed..ffc92c0 100644 --- a/lib/report/columns.h +++ b/lib/report/columns.h @@ -21,9 +21,19 @@ * determines the order the entries appear in this file. * * When adding new entries take care to use the existing style. + * + * Do not interleave fields from different report types - for example, + * if you have a field of type "LVS" add it in between "LVS type fields" + * and "End of LVS type fields" comment. If you interleaved fields of + * different types here in this file, they would end up interleaved in + * the <reporting_command> -o help output too which may be confusing + * for users. + * * Displayed fields names normally have a type prefix and use underscores. + * * Field-specific internal functions names normally match the displayed * field names but without underscores. + * * Help text ends with a full stop. */ @@ -32,13 +42,15 @@ */ /* *INDENT-OFF* */ +/* + * LVS type fields + */ FIELD(LVS, lv, STR, "LV UUID", lvid, 38, lvuuid, lv_uuid, "Unique identifier.", 0) FIELD(LVS, lv, STR, "LV", lvid, 4, lvname, lv_name, "Name. LVs created for internal use are enclosed in brackets.", 0) FIELD(LVS, lv, STR, "LV", lvid, 4, lvfullname, lv_full_name, "Full name of LV including its VG, namely VG/LV.", 0) FIELD(LVS, lv, STR, "Path", lvid, 0, lvpath, lv_path, "Full pathname for LV. Blank for internal LVs.", 0) FIELD(LVS, lv, STR, "DMPath", lvid, 0, lvdmpath, lv_dm_path, "Internal device-mapper pathname for LV (in /dev/mapper directory).", 0) FIELD(LVS, lv, STR, "Parent", lvid, 0, lvparent, lv_parent, "For LVs that are components of another LV, the parent LV.", 0) -FIELD(LVSINFOSTATUS, lv, STR, "Attr", lvid, 0, lvstatus, lv_attr, "Various attributes - see man page.", 0) FIELD(LVS, lv, STR_LIST, "Layout", lvid, 10, lvlayout, lv_layout, "LV layout.", 0) FIELD(LVS, lv, STR_LIST, "Role", lvid, 10, lvrole, lv_role, "LV role.", 0) FIELD(LVS, lv, BIN, "InitImgSync", lvid, 10, lvinitialimagesync, lv_initial_image_sync, "Set if mirror/RAID images underwent initial resynchronization.", 0) @@ -69,11 +81,6 @@ FIELD(LVS, lv, STR_LIST, "Ancestors", lvid, 0, lvancestors, lv_ancestors, "LV an FIELD(LVS, lv, STR_LIST, "FAncestors", lvid, 0, lvfullancestors, lv_full_ancestors, "LV ancestors including stored history of the ancestry chain.", 0) FIELD(LVS, lv, STR_LIST, "Descendants", lvid, 0, lvdescendants, lv_descendants, "LV descendants ignoring any stored history of the ancestry chain.", 0) FIELD(LVS, lv, STR_LIST, "FDescendants", lvid, 0, lvfulldescendants, lv_full_descendants, "LV descendants including stored history of the ancestry chain.", 0) -FIELD(LVSSTATUS, lv, PCT, "Data%", lvid, 6, datapercent, data_percent, "For snapshot, cache and thin pools and volumes, the percentage full if LV is active.", 0) -FIELD(LVSSTATUS, lv, PCT, "Snap%", lvid, 6, snpercent, snap_percent, "For snapshots, the percentage full if LV is active.", 0) -FIELD(LVSSTATUS, lv, PCT, "Meta%", lvid, 6, metadatapercent, metadata_percent, "For cache and thin pools, the percentage of metadata full if LV is active.", 0) -FIELD(LVSSTATUS, lv, PCT, "Cpy%Sync", lvid, 0, copypercent, copy_percent, "For Cache, RAID, mirrors and pvmove, current percentage in-sync.", 0) -FIELD(LVSSTATUS, lv, PCT, "Cpy%Sync", lvid, 0, copypercent, sync_percent, "For Cache, RAID, mirrors and pvmove, current percentage in-sync.", 0) FIELD(LVS, lv, NUM, "Mismatches", lvid, 0, raidmismatchcount, raid_mismatch_count, "For RAID, number of mismatches found or repaired.", 0) FIELD(LVS, lv, STR, "SyncAction", lvid, 0, raidsyncaction, raid_sync_action, "For RAID, the current synchronization action being performed.", 0) FIELD(LVS, lv, NUM, "WBehind", lvid, 0, raidwritebehind, raid_write_behind, "For RAID1, the number of outstanding writes allowed to writemostly devices.", 0) @@ -99,7 +106,13 @@ FIELD(LVS, lv, TIM, "RTime", lvid, 26, lvtimeremoved, lv_time_removed, "Removal FIELD(LVS, lv, STR, "Host", lvid, 10, lvhost, lv_host, "Creation host of the LV, if known.", 0) FIELD(LVS, lv, STR_LIST, "Modules", lvid, 0, modules, lv_modules, "Kernel device-mapper modules required for this LV.", 0) FIELD(LVS, lv, BIN, "Historical", lvid, 0, lvhistorical, lv_historical, "Set if the LV is historical.", 0) +/* + * End of LVS type fields + */ +/* + * LVSINFO type fields + */ FIELD(LVSINFO, lv, SNUM, "KMaj", lvid, 0, lvkmaj, lv_kernel_major, "Currently assigned major number or -1 if LV is not active.", 0) FIELD(LVSINFO, lv, SNUM, "KMin", lvid, 0, lvkmin, lv_kernel_minor, "Currently assigned minor number or -1 if LV is not active.", 0) FIELD(LVSINFO, lv, SIZ, "KRahead", lvid, 0, lvkreadahead, lv_kernel_read_ahead, "Currently-in-use read ahead setting in current units.", 0) @@ -108,7 +121,18 @@ FIELD(LVSINFO, lv, BIN, "Suspended", lvid, 10, lvsuspended, lv_suspended, "Set i FIELD(LVSINFO, lv, BIN, "LiveTable", lvid, 20, lvlivetable, lv_live_table, "Set if LV has live table present.", 0) FIELD(LVSINFO, lv, BIN, "InactiveTable", lvid, 20, lvinactivetable, lv_inactive_table, "Set if LV has inactive table present.", 0) FIELD(LVSINFO, lv, BIN, "DevOpen", lvid, 10, lvdeviceopen, lv_device_open, "Set if LV device is open.", 0) +/* + * End of LVSINFO type fields + */ +/* + * LVSSTATUS type fields + */ +FIELD(LVSSTATUS, lv, PCT, "Data%", lvid, 6, datapercent, data_percent, "For snapshot, cache and thin pools and volumes, the percentage full if LV is active.", 0) +FIELD(LVSSTATUS, lv, PCT, "Snap%", lvid, 6, snpercent, snap_percent, "For snapshots, the percentage full if LV is active.", 0) +FIELD(LVSSTATUS, lv, PCT, "Meta%", lvid, 6, metadatapercent, metadata_percent, "For cache and thin pools, the percentage of metadata full if LV is active.", 0) +FIELD(LVSSTATUS, lv, PCT, "Cpy%Sync", lvid, 0, copypercent, copy_percent, "For Cache, RAID, mirrors and pvmove, current percentage in-sync.", 0) +FIELD(LVSSTATUS, lv, PCT, "Cpy%Sync", lvid, 0, copypercent, sync_percent, "For Cache, RAID, mirrors and pvmove, current percentage in-sync.", 0) FIELD(LVSSTATUS, lv, NUM, "CacheTotalBlocks", lvid, 0, cache_total_blocks, cache_total_blocks, "Total cache blocks.", 0) FIELD(LVSSTATUS, lv, NUM, "CacheUsedBlocks", lvid, 16, cache_used_blocks, cache_used_blocks, "Used cache blocks.", 0) FIELD(LVSSTATUS, lv, NUM, "CacheDirtyBlocks", lvid, 0, cache_dirty_blocks, cache_dirty_blocks, "Dirty cache blocks.", 0) @@ -121,7 +145,21 @@ FIELD(LVSSTATUS, lv, STR, "KCachePolicy", lvid, 18, kernel_cache_policy, kernel_ FIELD(LVSSTATUS, lv, STR, "Health", lvid, 15, lvhealthstatus, lv_health_status, "LV health status.", 0) FIELD(LVSSTATUS, lv, STR, "KDiscards", lvid, 0, kdiscards, kernel_discards, "For thin pools, how discards are handled in kernel.", 0) FIELD(LVSSTATUS, lv, BIN, "CheckNeeded", lvid, 15, lvcheckneeded, lv_check_needed, "For thin pools and cache volumes, whether metadata check is needed.", 0) +/* + * End of LVSSTATUS type fields + */ +/* + * LVSINFOSTATUS type fields + */ +FIELD(LVSINFOSTATUS, lv, STR, "Attr", lvid, 0, lvstatus, lv_attr, "Various attributes - see man page.", 0) +/* + * End of LVSINFOSTATUS type fields + */ + +/* + * LABEL type fields + */ FIELD(LABEL, label, STR, "Fmt", type, 0, pvfmt, pv_fmt, "Type of metadata.", 0) FIELD(LABEL, label, STR, "PV UUID", type, 38, pvuuid, pv_uuid, "Unique identifier.", 0) FIELD(LABEL, label, SIZ, "DevSize", dev, 0, devsize, dev_size, "Size of underlying device in current units.", 0) @@ -131,7 +169,13 @@ FIELD(LABEL, label, STR, "Min", dev, 0, devminor, pv_minor, "Device minor number FIELD(LABEL, label, SIZ, "PMdaFree", type, 9, pvmdafree, pv_mda_free, "Free metadata area space on this device in current units.", 0) FIELD(LABEL, label, SIZ, "PMdaSize", type, 9, pvmdasize, pv_mda_size, "Size of smallest metadata area on this device in current units.", 0) FIELD(LABEL, label, NUM, "PExtVsn", type, 0, pvextvsn, pv_ext_vsn, "PV header extension version.", 0) +/* + * End of LABEL type fields + */ +/* + * PVS type fields + */ FIELD(PVS, pv, NUM, "1st PE", pe_start, 7, size64, pe_start, "Offset to the start of data on the underlying device.", 0) FIELD(PVS, pv, SIZ, "PSize", id, 0, pvsize, pv_size, "Size of PV in current units.", 0) FIELD(PVS, pv, SIZ, "PFree", id, 0, pvfree, pv_free, "Total amount of unallocated space in current units.", 0) @@ -149,7 +193,13 @@ FIELD(PVS, pv, SIZ, "BA Start", ba_start, 0, size64, pv_ba_start, "Offset to the FIELD(PVS, pv, SIZ, "BA Size", ba_size, 0, size64, pv_ba_size, "Size of PV Bootloader Area in current units.", 0) FIELD(PVS, pv, BIN, "PInUse", id, 0, pvinuse, pv_in_use, "Set if PV is used.", 0) FIELD(PVS, pv, BIN, "Duplicate", id, 0, pvduplicate, pv_duplicate, "Set if PV is an unchosen duplicate.", 0) +/* + * End of PVS type fields + */ +/* + * VGS type fields + */ FIELD(VGS, vg, STR, "Fmt", cmd, 0, vgfmt, vg_fmt, "Type of metadata.", 0) FIELD(VGS, vg, STR, "VG UUID", id, 38, uuid, vg_uuid, "Unique identifier.", 0) FIELD(VGS, vg, STR, "VG", name, 0, string, vg_name, "Name.", 0) @@ -183,7 +233,13 @@ FIELD(VGS, vg, NUM, "#VMdaUse", cmd, 0, vgmdasused, vg_mda_used_count, "Number o FIELD(VGS, vg, SIZ, "VMdaFree", cmd, 9, vgmdafree, vg_mda_free, "Free metadata area space for this VG in current units.", 0) FIELD(VGS, vg, SIZ, "VMdaSize", cmd, 9, vgmdasize, vg_mda_size, "Size of smallest metadata area for this VG in current units.", 0) FIELD(VGS, vg, NUM, "#VMdaCps", cmd, 0, vgmdacopies, vg_mda_copies, "Target number of in use metadata areas in the VG.", 1) +/* + * End of VGS type fields + */ +/* + * SEGS type fields + */ FIELD(SEGS, seg, STR, "Type", list, 0, segtype, segtype, "Type of LV segment.", 0) FIELD(SEGS, seg, NUM, "#Str", area_count, 0, uint32, stripes, "Number of stripes or mirror legs.", 0) FIELD(SEGS, seg, SIZ, "Stripe", stripe_size, 0, size32, stripe_size, "For stripes, amount of data placed on one device before switching to the next.", 0) @@ -208,7 +264,16 @@ FIELD(SEGS, seg, STR_LIST, "Metadata Devs", list, 0, metadatadevices, metadata_d FIELD(SEGS, seg, STR, "Monitor", list, 0, segmonitor, seg_monitor, "Dmeventd monitoring status of the segment.", 0) FIELD(SEGS, seg, STR, "CachePolicy", list, 0, cache_policy, cache_policy, "The cache policy (cached segments only).", 0) FIELD(SEGS, seg, STR_LIST, "CacheSettings", list, 0, cache_settings, cache_settings, "Cache settings/parameters (cached segments only).", 0) +/* + * End of SEGS type fields + */ +/* + * PVSEGS type fields + */ FIELD(PVSEGS, pvseg, NUM, "Start", pe, 0, uint32, pvseg_start, "Physical Extent number of start of segment.", 0) FIELD(PVSEGS, pvseg, NUM, "SSize", len, 0, uint32, pvseg_size, "Number of extents in segment.", 0) +/* + * End of PVSEGS type fields + */ /* *INDENT-ON* */

1 0

2026

2025

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

lvm2-commits December 2016