1.0.0.rc16/configure | 3 3 + 0 - 0 !
1.0.0.rc16/configure.in | 5 4 + 1 - 0 !
1.0.0.rc16/lib/Makefile.in | 10 8 + 2 - 0 !
1.0.0.rc16/make.tmpl.in | 3 2 + 1 - 0 !
1.0.0.rc16/tools/Makefile.in | 8 6 + 2 - 0 !
5 files changed, 23 insertions(+), 6 deletions(-)

 fix library symbolic link creation and library linking.

1.0.0.rc16/lib/.export.sym | 2 0 + 2 - 0 !
1.0.0.rc16/lib/Makefile.in | 8 2 + 6 - 0 !
1.0.0.rc16/lib/activate/activate.c | 8 2 + 6 - 0 !
1.0.0.rc16/lib/metadata/reconfig.c | 8 2 + 6 - 0 !
1.0.0.rc16/make.tmpl.in | 2 1 + 1 - 0 !
1.0.0.rc16/tools/Makefile.in | 12 0 + 12 - 0 !
6 files changed, 7 insertions(+), 33 deletions(-)

---

1.0.0.rc16/man/dmraid.8 | 124 62 + 62 - 0 !
1 file changed, 62 insertions(+), 62 deletions(-)

 fix hyphen-used-as-minus-sign

1.0.0.rc16/lib/format/ataraid/pdc.h | 2 1 + 1 - 0 !
1 file changed, 1 insertion(+), 1 deletion(-)

 add extra offsets where promise metadata can be found.

1.0.0.rc16/lib/metadata/metadata.c | 4 2 + 2 - 0 !
1 file changed, 2 insertions(+), 2 deletions(-)

 fix exit-code for "no raid disks" and "no block devices found" errors (lp: #300825)

1.0.0.rc16/lib/device/scan.c | 2 1 + 1 - 0 !
1 file changed, 1 insertion(+), 1 deletion(-)

 support for checking virtio devices for dmraid metadata.

1.0.0.rc16/lib/format/ataraid/isw.c | 70 64 + 6 - 0 !
1.0.0.rc16/lib/format/ataraid/isw.h | 13 12 + 1 - 0 !
2 files changed, 76 insertions(+), 7 deletions(-)

---

1.0.0.rc16/lib/metadata/metadata.c | 2 1 + 1 - 0 !
1 file changed, 1 insertion(+), 1 deletion(-)

---

1.0.0.rc16/lib/format/ataraid/pdc.c | 4 3 + 1 - 0 !
1 file changed, 3 insertions(+), 1 deletion(-)

---

1.0.0.rc16/lib/format/ddf/ddf1.c | 10 10 + 0 - 0 !
1 file changed, 10 insertions(+)

---

1.0.0.rc16/lib/format/ataraid/isw.c | 4 3 + 1 - 0 !
1 file changed, 3 insertions(+), 1 deletion(-)

---

1.0.0.rc16/man/dmraid.8 | 10 5 + 5 - 0 !
1 file changed, 5 insertions(+), 5 deletions(-)

---

1.0.0.rc16/lib/format/ataraid/jm.c | 2 1 + 1 - 0 !
1 file changed, 1 insertion(+), 1 deletion(-)

 uninitialized variable prevented code from working
 as intended.

1.0.0.rc16/lib/device/scan.c | 4 2 + 2 - 0 !
1 file changed, 2 insertions(+), 2 deletions(-)

 testing with dm devices was failing on ubuntu because
 /dev/dm-X is the actual device node, but the code wanted it to be
 a symlink.  Fixed dm_test_device() to test that the file ( or node
 it points to ) is a block device, which seems a much more
 appropriate test.

1.0.0.rc16/lib/format/ataraid/pdc.c | 111 58 + 53 - 0 !
1 file changed, 58 insertions(+), 53 deletions(-)

 pdc supports up to 4 sets of metadata to describe

1.0.0.rc16/lib/format/partition/dos.c | 10 8 + 2 - 0 !
1 file changed, 8 insertions(+), 2 deletions(-)

 dmraid was changed at one point to insert a 'p' between
 the base device name and the partition number.  For some time
 debian and ubuntu reversed this change.  This patch modifies the
 behavior to add the 'p' iff the last character of the base name is
 a digit.  This makes dmraid comply with the behavior used by
 kpartx and "by linux since the dawn of time".

1.0.0.rc16/include/dmraid/metadata.h | 1 1 + 0 - 0 !
1.0.0.rc16/lib/activate/devmapper.c | 14 12 + 2 - 0 !
1.0.0.rc16/lib/format/ataraid/isw.c | 3 2 + 1 - 0 !
1.0.0.rc16/lib/format/format.c | 5 3 + 2 - 0 !
4 files changed, 18 insertions(+), 5 deletions(-)

 change the uuid for internal devices ( like
 each half of a raid10 ) so that the rest of the system
 can recognize that the internal device should not be
 probed for partitions or filesystems or mounted.  Also
 change partition's uuid to have the same "partNN-"
 prefix that kpartx gives them.

1.0.0.rc16/lib/activate/activate.c | 70 52 + 18 - 0 !
1.0.0.rc16/lib/events/libdmraid-events-isw.c | 30 16 + 14 - 0 !
1.0.0.rc16/lib/metadata/metadata.c | 10 5 + 5 - 0 !
3 files changed, 73 insertions(+), 37 deletions(-)

 convert dmraid to use dm-raid module for raid4 and raid5

The dmraid library used the dm-raid45 target for RAID4 and RAID5 sets.
This module however was never upstream and is suffering constant issues
when the Linux kernel changes.
Upstream introduced the dm-raid/raid module/target in 2.6.38. This can be
used to create device-mapper targets which use the MD RAID disciplines
(though not showing up in /proc/mdstat).

Most of the table constructor arguments can be mapped, only the member
devices may not have an offset.

From the old code it seems that RAID5 would always be build by set
elements which will not have an offset. So potentially only RAID4
could cause issues. That could be worked around if needed by creating
additional linear mappings.

NOTE: The event code is only compile tested. But we seem to have never
shipped the dmraid event tool which is needed to use this feature. And
we are about to remove support for ISW from dmraid (in favour of mdadm)
anyway.

Signed-off-by: Stefan Bader <stefan.bader@canonical.com>

1.0.0.rc16/lib/format/format.c | 12 12 + 0 - 0 !
1 file changed, 12 insertions(+)

 avoid meaningless error messages while probing for meta-data

This happens as device-mapper volumes are checked and kpartx creates
a 2 sector mapping if a device contains an extended partition.
Offsetting from the last sector backwards some probes overflow and
end up trying to seek to a very large offset (which fails and emits
a visible error message during device scanning).
This message is useless as such a small device would be pointless as
a RAID set anyway.
To fix this without larger changes, pragmatically ignore devices smaller
than 8 sectors (4K).

Signed-off-by: Stefan Bader <stefan.bader@canonical.com>