11 years agoutil: fix test for text_version
NeilBrown [Mon, 11 May 2009 05:21:43 +0000 (15:21 +1000)] 
util: fix test for text_version

as text_version is a char array (not a pointer), testing the
address against NULL is the wrong thing to do.  Test the
content instead.

Signed-off-by: NeilBrown <>
11 years agoAssemble/Incr : minor tidy up of setting 'trustworthy'.
NeilBrown [Mon, 11 May 2009 05:19:30 +0000 (15:19 +1000)] 
Assemble/Incr : minor tidy up of setting 'trustworthy'.

Signed-off-by: NeilBrown <>
11 years agoconfig: allow Array line to contain array name without /dev/md/ prefix.
NeilBrown [Mon, 11 May 2009 05:18:35 +0000 (15:18 +1000)] 
config: allow Array line to contain array name without /dev/md/ prefix.

For consistency with --create and --assemble, allow the array name
given in mdadm.conf to exclude the "/dev/md/" prefix.  So e.g.

  ARRAY home uuid=whatever

is treated like

  ARRAY /dev/md/home uuid=whatever

Also exclude names which create_mddev will reject.

Signed-off-by: NeilBrown <>
11 years agoFix tests on ->container and ->member
NeilBrown [Mon, 11 May 2009 05:18:25 +0000 (15:18 +1000)] 
Fix tests on ->container and ->member

For container= and member= to be effective in an mdadm.conf line
they must both be present.  So when checking for their absence we
need container != NULL || member != NULL.

Signed-off-by: NeilBrown <>
11 years agoMake --brief even briefer.
NeilBrown [Mon, 11 May 2009 05:18:20 +0000 (15:18 +1000)] 
Make --brief even briefer.

Because ---examine --brief, or --detail --brief are
often used to create mdadm.conf, and because people don't want to
have to update their mdadm.conf unnecessarily, we don't want to
include information that might change.
And now that level changing is supported, that is almost everything
but UUID.

So move some more fields into the "Only print with --verbose" class.

Signed-off-by: NeilBrown <>
11 years agomdadm.8 general updates
NeilBrown [Mon, 11 May 2009 05:17:50 +0000 (15:17 +1000)] 
mdadm.8  general updates

Signed-off-by: NeilBrown <>
11 years agoconf/assemble: new config line "auto".
NeilBrown [Mon, 11 May 2009 05:17:33 +0000 (15:17 +1000)] 
conf/assemble:  new config line "auto".

The line 'auto' in mdadm.conf can be used to disable assembly
of specific metadata types, or of all arrays.

This does not affect assembly of arrays listed in mdadm.conf
or on command line.

 auto -all

will disable all auto-assembly.

 auto -ddf

will cause mdadm to ignore ddf arrays that are not explicitly
mentioned, and auto assemble anything else it finds.

Signed-off-by: NeilBrown <>
11 years agoconfig: support "ARRAY <ignore> ..." lines in mdadm.conf
NeilBrown [Mon, 11 May 2009 05:17:05 +0000 (15:17 +1000)] 
config: support "ARRAY <ignore> ..." lines in mdadm.conf

Sometimes we want to ensure particular arrays are never
assembled automatically.  This might include an array made of
devices that are shared between hosts.

To support this, allow ARRAY lines in mdadm.conf to use the word
"ignore" rather than a device name.  Arrays which match such lines
are never automatically assembled (though they can still be assembled
by explicitly giving identification information on the mdadm command

Signed-off-by: NeilBrown <>
11 years agoassemble: support arrays created with --homehost=any
NeilBrown [Mon, 11 May 2009 05:16:49 +0000 (15:16 +1000)] 
assemble:  support arrays created with --homehost=any

If an array is created with --homehost=any, then --assemble and
--incremental will treat it as being local to 'this' host, no matter
what the name of this host is.

This is useful for array that will be given unique names and be
moved between machines.

This needs to be documented.

Signed-off-by: NeilBrown <>
11 years agocreate_dev - allow array names like mdX and /dev/mdX to appear 'numeric'
NeilBrown [Mon, 11 May 2009 05:16:47 +0000 (15:16 +1000)] 
create_dev - allow array names like mdX and /dev/mdX to appear 'numeric'

When choosing the minor number to use with an array, we currently base
the number of the 'name' stored in the metadata if that name is
Extend that so that if it looks like a number md device name (/dev/md0
or just md0 or even /dev/md/0), then we use the number at the end to
suggest a minor number.

The means that if someone creates and array with "--name md0" or even
"--name /dev/md0" it will continue to do what they expect.

Signed-off-by: NeilBrown <>
11 years agoWait for POLLPRI on /proc or /sys files.
NeilBrown [Tue, 14 Apr 2009 04:59:24 +0000 (14:59 +1000)] 
Wait for POLLPRI on /proc or /sys files.

From 2.6.30, /proc/mounts and various /sys files will
probably always returns 'readable' to select, so we will need
to wait on POLLPRI to get the 'new data is available' signal.

When using select, this corresponds to an 'exception', so
adjust calls to select accordingly.
In one case we sometimes wait on a socket and sometime on
/proc/mounts, so we need to test which.

Signed-off-by: NeilBrown <>
11 years agoMerge branch 'master' of git:// into devel-3.0
NeilBrown [Tue, 14 Apr 2009 01:11:14 +0000 (11:11 +1000)] 
Merge branch 'master' of git:// into devel-3.0

Due to independent fixes for the "mdadm hangs if reshape finishes too quickly"

11 years agoudev rules fix for partitions.
NeilBrown [Tue, 14 Apr 2009 00:58:54 +0000 (10:58 +1000)] 
udev rules fix for partitions.

sysfs directories for partitions do not have md/* files, but
should not for that reason be ignored.

Thanks to Michal Soltys for original fix.

Signed-off-by: Michal Soltys <>
Signed-off-by: NeilBrown <>
11 years agomapfile: optionally store map file in /dev
NeilBrown [Tue, 14 Apr 2009 00:49:40 +0000 (10:49 +1000)] 
mapfile:  optionally store map file in /dev

During early boot, /var/run may not exist or be writable.
If that happens, sore the mapfile (which is very important for
incremental assembly) in /dev (which should exist for udev).

Thanks to Doug Ledford <> for identify this
problem and suggesting a solution.

Signed-off-by: NeilBrown <>
11 years agotests: basic ddf tests
NeilBrown [Tue, 14 Apr 2009 00:19:06 +0000 (10:19 +1000)] 
tests: basic ddf tests

Test script to create a ddf with three different member arrays,
and assemble it in a variety of ways.

Signed-off-by: NeilBrown <>
11 years agotests: add some simple data integrity tests for raid5/raid6
NeilBrown [Tue, 14 Apr 2009 00:19:05 +0000 (10:19 +1000)] 
tests:  add some simple data integrity tests for raid5/raid6

Make sure the data is preserved even when the array is degraded.

Signed-off-by: NeilBrown <>
11 years ago.gitignore update
NeilBrown [Tue, 14 Apr 2009 00:19:04 +0000 (10:19 +1000)] 
.gitignore update

Add a few more generated files to .gitignore

Signed-off-by: NeilBrown <>
11 years agoincremental_container: preserve 'in_sync' flag when adding to existing array.
NeilBrown [Tue, 14 Apr 2009 00:19:02 +0000 (10:19 +1000)] 
incremental_container: preserve 'in_sync' flag when adding to existing array.

When building container members with -IR, we need to ensure that
devices added to an active array preserve the 'in_sync' status so they
don't needlessly get rebuilt.

So allow sysfs_add_disk to do this (only works in kernels since
2.6.30) and pass the relevant flag down.

Signed-off-by: NeilBrown <>
11 years agomdmon: fix resync completion detection
Dan Williams [Sun, 12 Apr 2009 07:58:28 +0000 (00:58 -0700)] 
mdmon: fix resync completion detection

Starting with 2.6.30 the md/resync_start attribute will no longer return
a non-sensical number when resync is complete, instead it now returns

Signed-off-by: Dan Williams <>
11 years agoGrow: fix hang when reshape completes too fast
Dan Williams [Sun, 12 Apr 2009 07:58:28 +0000 (00:58 -0700)] 
Grow: fix hang when reshape completes too fast

For short reshapes the kernel may be done before mdadm can check that
progress has passed the critical section.

Signed-off-by: Dan Williams <>
11 years agoRebuildMap: handle missing disks
Dan Williams [Sun, 12 Apr 2009 07:58:28 +0000 (00:58 -0700)] 
RebuildMap: handle missing disks

When rebuilding the map file tolerate missing/offline disks, otherwise
we will segfault on the NULL return from sysfs_read.

Reported-by: Jacek Danecki <>
Signed-off-by: Dan Williams <>
11 years agoimsm: add the ddf field
Dan Williams [Sun, 12 Apr 2009 07:58:28 +0000 (00:58 -0700)] 
imsm: add the ddf field

This field is always one in arrays created by the Windows driver / OROM,
not sure why...

Signed-off-by: Dan Williams <>
11 years agoimsm: round down array size at Create
Dan Williams [Sun, 12 Apr 2009 07:58:28 +0000 (00:58 -0700)] 
imsm: round down array size at Create

Store the 1MB rounded down size of the array at create time.

Signed-off-by: Dan Williams <>
11 years agoimsm: set array size at Create/Assemble
Dan Williams [Sun, 12 Apr 2009 07:58:28 +0000 (00:58 -0700)] 
imsm: set array size at Create/Assemble

imsm arrays round down the effective array size to the closest 1
megabyte boundary so teach get_info_super_imsm and sysfs_set_array to
set 'md/array_size' if available (and make sure ddf uses the default

Signed-off-by: Dan Williams <>
11 years agoimsm: turn off curr_migr_unit updates
Dan Williams [Sun, 12 Apr 2009 07:58:28 +0000 (00:58 -0700)] 
imsm: turn off curr_migr_unit updates

New documentation shows that this field is not equivalent to
md/resync_start.  Disable updates until full support can be developed.

Writing '0' when a migration starts/re-starts remains correct.

Signed-off-by: Dan Williams <>
11 years agoimsm: defend against unsupported migrations (temporary)
Dan Williams [Sun, 12 Apr 2009 07:58:28 +0000 (00:58 -0700)] 
imsm: defend against unsupported migrations (temporary)

Until support for higher order migrations (online capacity expansion,
raid level migration, chunk size migration...) are implemented do not
allow arrays in these states to be assembled.

Signed-off-by: Dan Williams <>
11 years agoimsm: add 'verify', 'verify with fixup', and 'general' migration types
Dan Williams [Sun, 12 Apr 2009 07:58:27 +0000 (00:58 -0700)] 
imsm: add 'verify', 'verify with fixup', and 'general' migration types

imsm distinguishes parity initialization from parity checking in the
metadata. Older option roms marked the repair operation with the
'verify' type and a 'with fixup' flag in the raid device 'status' field.

Signed-off-by: Dan Williams <>
11 years agoimsm: fix imsm_map.num_domains
Dan Williams [Sun, 12 Apr 2009 07:58:27 +0000 (00:58 -0700)] 
imsm: fix imsm_map.num_domains

'num_domains' is the number of parity domains.  I.e. 2 in the raid10
case (2-mirrors), while raid0 through raid5 have 1 parity domain (even
though raid0 does not have parity).

Signed-off-by: Dan Williams <>
11 years agoimsm: ensure mpb buffer is zeroed
Dan Williams [Sun, 12 Apr 2009 07:58:27 +0000 (00:58 -0700)] 
imsm: ensure mpb buffer is zeroed

Don't leak unitialized data into the mpb.

Signed-off-by: Dan Williams <>
11 years agoimsm: support --examine --export
Dan Williams [Sun, 12 Apr 2009 04:53:25 +0000 (21:53 -0700)] 
imsm: support --examine --export

Signed-off-by: Dan Williams <>
11 years agoimsm: make uuid separator consistent with ddf
Dan Williams [Sun, 12 Apr 2009 04:53:25 +0000 (21:53 -0700)] 
imsm: make uuid separator consistent with ddf

'-' to ':'

Signed-off-by: Dan Williams <>
11 years agoimsm: extract right-most whitespace stripped serial number
Dan Williams [Wed, 8 Apr 2009 18:41:51 +0000 (11:41 -0700)] 
imsm: extract right-most whitespace stripped serial number

According to new documentation the metadata expects that all whitespace
(characters <= 0x20) are stripped from the incoming serial number.  If
the length remains longer than MAX_RAID_SERIAL_LEN then only the
right-most characters are preserved.

Signed-off-by: Dan Williams <>
11 years agomdadm udev rules change
Doug Ledford [Wed, 8 Apr 2009 07:32:55 +0000 (17:32 +1000)] 
mdadm udev rules change

I'm not attaching a patch for this because it's so simple.  Long story
short, watching both add and change events in udev rules is bad for md
devices.  Specifically, the kernel will generate a change event on
things like array stop, and on things like fdisk close.  In the case
of array stop, it can result in the array being assembled again
immediately.  In the case of fdisk close, the situation is worse.
Let's say you stop all the md devices on some block device in order to
repartition.  You run fdisk, change the partition table, then issue a
write of the table.  The write of the table triggers the change event
*before* the kernel updates the partition table in memory for the
block device, causing udev to rerun the incremental rules on the old
partition table and restart all the arrays you just stopped with the
old partition table layout, at which point the kernel is unable to
reread the partition table.  So, once you've enable incremental
assembly, it becomes apparent that what we really want is to only
start devices on add, not on add|change.


Doug Ledford <>

11 years agosuper-ddf: fix compile warnings on ppc64
Doug Ledford [Wed, 8 Apr 2009 07:25:52 +0000 (17:25 +1000)] 
super-ddf: fix compile warnings on ppc64

Simple patch to silence some compile warnings that only show up on
64bit arches.

Signed-off-by: Doug Ledford <>
Signed-off-by: NeilBrown <>
11 years agoIncremental - avoid NULL dereference.
NeilBrown [Tue, 7 Apr 2009 07:54:09 +0000 (17:54 +1000)] 
Incremental - avoid NULL dereference.

There structure returned by sysfs_read might not have any 'devs',
don't assume it does.

Signed-off-by: NeilBrown <>
11 years agotest: minor fixes
NeilBrown [Tue, 7 Apr 2009 07:50:34 +0000 (17:50 +1000)] 
test: minor fixes

 - allow core files to be dumped.
 - modprobe md_mod if needed to write to

Signed-off-by: NeilBrown <>
11 years agoIncremental: be more relaxed about member arrays not completely assembling.
NeilBrown [Tue, 7 Apr 2009 07:49:05 +0000 (17:49 +1000)] 
Incremental: be more relaxed about member arrays not completely assembling.

During incremental assembly, if the member array doesn't assemble
properly (yet), that isn't an error.

Signed-off-by: NeilBrown <>
11 years agoAssemble: call map_update even when not starting the array.
NeilBrown [Tue, 7 Apr 2009 07:45:42 +0000 (17:45 +1000)] 
Assemble: call map_update even when not starting the array.

For incremental assembly to work, we need to keep the 'map'
up-to-date even when not starting the array yet.

Signed-off-by: NeilBrown <>
11 years agoddf: fixed 'working_disks' reported by container_content.
NeilBrown [Tue, 7 Apr 2009 07:41:55 +0000 (17:41 +1000)] 
ddf: fixed 'working_disks' reported by container_content.

The 'work_disks' number should be the number that is expected, not the
number found so far.   This is needed for Incremental assembly to
start the array at the right time.

Signed-off-by: NeilBrown <>
11 years agoddf: improve print out of physical devices.
NeilBrown [Tue, 7 Apr 2009 07:40:24 +0000 (17:40 +1000)] 
ddf: improve print out of physical devices.

Just add some more space really.

Signed-off-by: NeilBrown <>
11 years agoddf: fix variable name overload
NeilBrown [Tue, 7 Apr 2009 07:38:25 +0000 (17:38 +1000)] 
ddf: fix variable name overload

The variable 'i' was being used as a loop variable, and also
for something else inside the loop.   So make the larger loop have a
more meaningful name.

Signed-off-by: NeilBrown <>
11 years agowait_for improvement.
NeilBrown [Tue, 7 Apr 2009 07:34:38 +0000 (17:34 +1000)] 
wait_for improvement.

wait not only for the name to appear, but for it to refer to the
correct device.
Sometimes old symlinks left lying around can be confusing.

Signed-off-by: NeilBrown <>
11 years agoconfig: treat "container=" and "member=" and providing the required identity information.
NeilBrown [Mon, 6 Apr 2009 06:09:52 +0000 (16:09 +1000)] 
config: treat "container=" and "member=" and providing the required identity information.

Signed-off-by: NeilBrown <>
11 years agoDetail: use meaningful names with --scan.
NeilBrown [Mon, 6 Apr 2009 05:54:51 +0000 (15:54 +1000)] 
Detail: use meaningful names with --scan.

When reporting "--detail --scan", use names like /dev/md/foo where
available rather than /dev/md/127

This is particularly needed for containers where the member arrays
will report "container=/dev/md/foo" and we want the container to have
the same name.

Signed-off-by: NeilBrown <>
11 years agoCreate: round to chunk size
NeilBrown [Mon, 6 Apr 2009 05:53:31 +0000 (15:53 +1000)] 
Create: round to chunk size

There are probably other places where rounding size to
chunksize is needed, or useful, but this is a good start.

Signed-off-by: NeilBrown <>
11 years agotest: support testing DDF
NeilBrown [Mon, 6 Apr 2009 05:52:13 +0000 (15:52 +1000)] 
test: support testing DDF

DDF requires larger devices as 32Meg is reserved.

So allocated larger loop devices for DDF tests.

Signed-off-by: NeilBrown <>
11 years agoutil.c: use correct range for minor numbers when finding free device.
NeilBrown [Mon, 6 Apr 2009 05:50:56 +0000 (15:50 +1000)] 
util.c: use correct range for minor numbers when finding free device.

Minor numbers are 20 bits, not 22.
So when looking for a free, high minor number, try (1<<20)-1,
not (1<<22)-1.

11 years agoddf: make sure Create fails if devices are too small.
NeilBrown [Thu, 2 Apr 2009 03:02:33 +0000 (14:02 +1100)] 
ddf: make sure Create fails if devices are too small.

Signed-off-by: NeilBrown <>
11 years agoddf: fix bug with error returns.
NeilBrown [Thu, 2 Apr 2009 02:37:34 +0000 (13:37 +1100)] 
ddf: fix bug with error returns.

In some cases we should only print an error message if
'devname' is defined.  In fact we were only returning
the error at all in that case!!

Signed-off-by: NeilBrown <>
11 years agoism-tests: fix the imsm-create-fail-rebuild test
NeilBrown [Wed, 1 Apr 2009 02:51:57 +0000 (13:51 +1100)] 
ism-tests: fix the imsm-create-fail-rebuild test

it was testing the return status of mdadm wrongly, and
I found I needed a 'udevadm settle'.

11 years agotest: return correct error status for mdadm
NeilBrown [Wed, 1 Apr 2009 02:50:19 +0000 (13:50 +1100)] 
test: return correct error status for mdadm

mdadm's return status is being hidden by the 'test' script so
individual tests cannot see it.

11 years agoMakefile: build 'mdmon' for 'test'
NeilBrown [Wed, 1 Apr 2009 02:49:34 +0000 (13:49 +1100)] 
Makefile:  build 'mdmon' for 'test'

  make test
might not prepare everything for running tests.

11 years agogrow: don't wait forever for critical section to pass.
NeilBrown [Wed, 1 Apr 2009 01:26:08 +0000 (12:26 +1100)] 
grow: don't wait forever for critical section to pass.

If an array reshape completed within 1 second, then --grow will not
notice that it has finished and will keep waiting for the critical
section to pass.

So be more cautious in the test.

Signed-off-by: NeilBrown <>
11 years agoRelease mdadm-3.0-devel3 mdadm-3.0-devel3
NeilBrown [Tue, 10 Mar 2009 05:59:57 +0000 (16:59 +1100)] 
Release mdadm-3.0-devel3

11 years agoMerge branch 'master' in devel-3.0
NeilBrown [Tue, 10 Mar 2009 05:47:02 +0000 (16:47 +1100)] 
Merge branch 'master' in devel-3.0

11 years agoRelease mdadm-2.6.9 mdadm-2.6.9
NeilBrown [Tue, 10 Mar 2009 05:38:37 +0000 (16:38 +1100)] 
Release mdadm-2.6.9

11 years agomdmon: allow incremental assembly of containers.
NeilBrown [Tue, 10 Mar 2009 05:28:22 +0000 (16:28 +1100)] 
mdmon: allow incremental assembly of containers.

If mdmon sees a device added to a container, it should assume it is
a new spare.  It could be a part of the array that just hadn't been
assembled yet.  So check first.

Signed-off-by: NeilBrown <>
11 years agoDDF: fix up container_content
NeilBrown [Tue, 10 Mar 2009 05:28:22 +0000 (16:28 +1100)] 
DDF: fix up container_content

If we haven't got hold of all the devices yet, we need to be
ready to skip over some while gathering content information.

Signed-off-by: NeilBrown <>
11 years agoAssemble/container: catch errors when starting a partial container.
NeilBrown [Tue, 10 Mar 2009 05:28:22 +0000 (16:28 +1100)] 
Assemble/container: catch errors when starting a partial container.

If we are assembling an array in a container and it isn't complete
enough to start yet, then
  - don't start mdmon
  - don't say the array is started
  - don't wait for the device to appear in /dev

Signed-off-by: NeilBrown <>
11 years agoIncremental/container: don't create when we should just be opening.
NeilBrown [Tue, 10 Mar 2009 05:28:22 +0000 (16:28 +1100)] 
Incremental/container:  don't create when we should just be opening.

When incr-assembling in a container, if the array already exists,
don't try to create it again, just open it.

Signed-off-by: NeilBrown <>
11 years agomdopen: be more careful when adding digit to names.
NeilBrown [Tue, 10 Mar 2009 05:28:22 +0000 (16:28 +1100)] 
mdopen:  be more careful when adding digit to names.

If we need to add digits to a name to make it unique, but don't have
to add '_', we need to avoid adding a digit immediately after a digit.
So if the last character of the name is a digit, add the '_' anyway.

Signed-off-by: NeilBrown <>
11 years agoIncremental: fix some handling of trustworthy.
NeilBrown [Tue, 10 Mar 2009 05:28:22 +0000 (16:28 +1100)] 
Incremental:  fix some handling of trustworthy.

1/ if homehost matches, then we need to set trustworthy to 'LOCAL'
2/ if we decide to set trustworthy to 'METADATA' because we have to
   use the metadata version name, do that *after* we have checked if
   we are going to assemble within a container, as inside the
   container there could be different sources of names to use.

Signed-off-by: NeilBrown <>
11 years agoddf: use better random number generation.
NeilBrown [Tue, 10 Mar 2009 05:28:18 +0000 (16:28 +1100)] 
ddf:  use better random number generation.

use /dev/urandom rather than calls to 'random()' if possible.

Signed-off-by: NeilBrown <>
11 years agoExamine: add examine_export for ddf and avoid crashes.
NeilBrown [Mon, 9 Mar 2009 03:10:52 +0000 (14:10 +1100)] 
Examine:  add examine_export for ddf and avoid crashes.

If the personality doesn't provide export_examine_super, don't crash.

Signed-off-by: NeilBrown <>
11 years agoddf: fix memory corruption bug.
NeilBrown [Mon, 9 Mar 2009 02:55:59 +0000 (13:55 +1100)] 
ddf: fix memory corruption bug.

When adding to a point, you don't need to multiple by the size
of the pointer - C does that for you!

11 years agoDDF: minor formatting improvement
NeilBrown [Mon, 9 Mar 2009 02:55:13 +0000 (13:55 +1100)] 
DDF: minor formatting improvement

avoid long lines in --examine output

11 years agoSupport new raid6 layouts needed for DDF
NeilBrown [Mon, 9 Mar 2009 00:16:53 +0000 (11:16 +1100)] 
Support new raid6 layouts needed for DDF

DDF raid6 layouts are subtly different from the standard 'md' layouts.
From 2.6.30 the kernel knows about these.
Teach mdadm about them, and also allow 'ddf' to set an appropriate default.

Signed-off-by: NeilBrown <>
11 years agosuper1/examine: remove confusing information.
NeilBrown [Sun, 8 Mar 2009 23:38:33 +0000 (10:38 +1100)] 
super1/examine:  remove confusing information.

The information about how slots and roles in the array lined up
turned out to be confusing.

So simplify it and one provide the interesting information.

Signed-off-by: NeilBrown <>
11 years agosuper1 - do metadata IO in sector_size units.
NeilBrown [Sun, 8 Mar 2009 23:17:42 +0000 (10:17 +1100)] 
super1 - do metadata IO in sector_size units.

If the sector size is > 512, we need to be more careful about
The largest known sector size is 4096 and (fortunately) both the
superblock and (in many cases) the bitmap are 4096-byte aligned.
So there should be no data-overlap problems.

The exception is when the bitmap is squeezed into the 3K after the
superblock.  This arrangement cannot currently be supported on
4K sector-size devices.

Signed-off-by: NeilBrown <>
11 years agosuper0: Do all metadata IO with 4096byte alignment
NeilBrown [Sun, 8 Mar 2009 23:16:17 +0000 (10:16 +1100)] 
super0:  Do all metadata IO with 4096byte alignment

.. because some devices (dasd) have 4096 byte sector size.

As the superblock is 4096 bytes and the bitmap is in a
60K region, this is safe from any possible corruption.

Signed-off-by: NeilBrown <>
11 years agoFix udev-rules for case where array isn't really there..
NeilBrown [Sun, 8 Mar 2009 23:14:48 +0000 (10:14 +1100)] 
Fix udev-rules for case where array isn't really there..


Signed-off-by: NeilBrown <>
11 years agosuper1: make sure max_dev grows enough when adding a device to an array.
NeilBrown [Sun, 8 Mar 2009 22:59:39 +0000 (09:59 +1100)] 
super1: make sure max_dev grows enough when adding a device to an array.

There was a few kernel releases where the kernel would shrink max_dev
to be just enough to hold the current number of devices.
More recent kernels never shrink it.
However to be as compatible as possible, if we notice that
max_dev is too small to successfully add a device, increase it.

Signed-off-by: NeilBrown <>
11 years agoimsm: display supported chunk sizes in --detail-platform
Dan Williams [Fri, 27 Feb 2009 22:35:20 +0000 (15:35 -0700)] 
imsm: display supported chunk sizes in --detail-platform

Signed-off-by: Dan Williams <>
11 years agoIncremental: honor --no-degraded to delay assembly
Dan Williams [Wed, 25 Feb 2009 01:45:57 +0000 (18:45 -0700)] 
Incremental: honor --no-degraded to delay assembly

Currently Incremental_container is being called after adding each disk.
In the imsm case where spares are not tracked in the raid_disks field we
can use --no-degraded to block premature assembly.

Signed-off-by: Dan Williams <>
11 years agoIncremental: fix 'name_to_use' in the container case
Dan Williams [Wed, 25 Feb 2009 01:45:57 +0000 (18:45 -0700)] 
Incremental: fix 'name_to_use' in the container case

Just like the Assemble case, default to the text_version of the
container if another name is not specified.

Signed-off-by: Dan Williams <>
11 years agomdmon: update cmdline when scanning
Dan Williams [Wed, 25 Feb 2009 01:45:57 +0000 (18:45 -0700)] 
mdmon: update cmdline when scanning

Allows ps -ax | grep mdmon to show:
mdmon md127
mdmon md126
...rather than:
mdmon /proc/mdstat
mdmon /proc/mdstat

Signed-off-by: Dan Williams <>
11 years agomdmon: man page
Dan Williams [Wed, 25 Feb 2009 01:45:57 +0000 (18:45 -0700)] 
mdmon: man page

Signed-off-by: Dan Williams <>
11 years agomdmon: fix missed 'clean' event
Dan Williams [Wed, 25 Feb 2009 01:45:57 +0000 (18:45 -0700)] 
mdmon: fix missed 'clean' event

mdmon may miss events because it re-reads state after read_and_act.  The
additional read is used to determine dirty status before allowing a
sigterm to proceed.  Since read_and_act is in the best position to
determine 'dirty' status and its return value is not used, modify it to
return true if the array is dirty.

Signed-off-by: Dan Williams <>
11 years agoimsm: auto layout
Dan Williams [Wed, 25 Feb 2009 01:45:57 +0000 (18:45 -0700)] 
imsm: auto layout

In support of auto-layout:

1/ collect and merge all extents to find the largest common-start free region
2/ verify that we meet the "all volumes must use the same set of disks"
2/ mark the disks to be added in add_to_super_imsm_volume

Signed-off-by: Dan Williams <>
11 years agoCreate: fixup 'insert_point', dependent on 'subdevs', for auto-layout
Dan Williams [Wed, 25 Feb 2009 01:45:57 +0000 (18:45 -0700)] 
Create: fixup 'insert_point', dependent on 'subdevs', for auto-layout

'subdevs' is read from the container in the auto-layout case so reset
subdevs dependent default values.  'insert_point' without this
change is always 2 blocking creation of arrays with > 2 raid disks.

Signed-off-by: Dan Williams <>
11 years agoCreate: wait_for container creation
Dan Williams [Wed, 25 Feb 2009 01:45:57 +0000 (18:45 -0700)] 
Create: wait_for container creation

Signed-off-by: Dan Williams <>
11 years agoManage: permit '--remove detached' for containers
Dan Williams [Wed, 25 Feb 2009 01:45:57 +0000 (18:45 -0700)] 
Manage: permit '--remove detached' for containers

Skip the unique holder check in the detached case... pretty sure no one is
holding on to it if open() returns ENXIO.

Signed-off-by: Dan Williams <>
11 years agomdmon: record added disks
Dan Williams [Wed, 25 Feb 2009 01:45:57 +0000 (18:45 -0700)] 
mdmon: record added disks

Prevent duplicate disks from being sent to the monitor thread.

Signed-off-by: Dan Williams <>
11 years agomdmon: fix removed disk handling
Dan Williams [Wed, 25 Feb 2009 01:45:57 +0000 (18:45 -0700)] 
mdmon: fix removed disk handling

Use SKIP_GONE_DEVS when reading the container, and correct some confused
logic in manage_new().

Signed-off-by: Dan Williams <>
11 years agosysfs: allow sysfs_read to detect and drop removed disks
Dan Williams [Wed, 25 Feb 2009 01:45:56 +0000 (18:45 -0700)] 
sysfs: allow sysfs_read to detect and drop removed disks

All operations that rely on loading from an existing container (like
--add) will fail after a disk has been removed.  Provide an option to
skip missing / offline disks rather than abort.  We attempt to do this
in the load_super_{imsm,ddf}_all cases when mdmon is running i.e. we
already have a consitent version of the metadata running in the system.
Otherwise, we fail as normal and let the administrator fix up the

Signed-off-by: Dan Williams <>
11 years agoimsm: retry load_imsm_mpb if we suspect mdmon has made modifications
Dan Williams [Wed, 25 Feb 2009 01:45:56 +0000 (18:45 -0700)] 
imsm: retry load_imsm_mpb if we suspect mdmon has made modifications

If the checksum verification fails and mdmon is running we retry the
load to get a consistent snapshot of the mpb.  Found by

Signed-off-by: Dan Williams <>
11 years agoimsm: verify single sector mpb checksums
Dan Williams [Wed, 25 Feb 2009 01:45:56 +0000 (18:45 -0700)] 
imsm: verify single sector mpb checksums

If the mpb is only one sector do not skip the checksum verification.

Signed-off-by: Dan Williams <>
11 years agoimsm: fix mark_failure / introduce mark_missing
Dan Williams [Wed, 25 Feb 2009 01:45:56 +0000 (18:45 -0700)] 
imsm: fix mark_failure / introduce mark_missing

Actually, rename mark_failure to mark_missing and then implement the
correct mark_failure which according to new documentation is to:

1/ Set the FAILED status bit
2/ Set IMSM_ORD_REBUILD to mark the disk out of sync
3/ Set map->failed_disk_num if this is the first failure detected
   failure (it is ~0 otherwise)

Previously the assumption was that IMSM_ORD_REBUILD only appeared in
map[1], so all routines that care about out-of-sync disks need to be

Signed-off-by: Dan Williams <>
11 years agoimsm: introduce get_imsm_disk_slot
Dan Williams [Wed, 25 Feb 2009 01:45:56 +0000 (18:45 -0700)] 
imsm: introduce get_imsm_disk_slot

Implement a common disk index to disk slot routine and replace open
coded versions.

Signed-off-by: Dan Williams <>
11 years agoimsm: fix activate spare to ignore foreign disks
Dan Williams [Tue, 24 Feb 2009 06:06:24 +0000 (23:06 -0700)] 
imsm: fix activate spare to ignore foreign disks

A foreign disk is one that all other drives believe is not-in-sync but
does not have the 'failed' status bit set.

This also reverts, because that commit is addressing the wrong problem.
Ideally mdmon would kick "non-fresh" drives like the kernel does at
native-md activation time, but that is too awkward to implement at the
moment because mdadm owns container manipulations.

Signed-off-by: Dan Williams <>
11 years agoimsm: fixup container spare uuids by default
Dan Williams [Mon, 23 Feb 2009 21:26:11 +0000 (14:26 -0700)] 
imsm: fixup container spare uuids by default

Spares in the imsm case are marked with the "match-all" uuid of
ffffffff-ffffffff-ffffffff-ffffffff.  When performing incremental
assembly we need to associate such devices with a populated container
uuid.  Also when performing --detail on a container with only spares
present we can make an attempt to return a real uuid.

Signed-off-by: Dan Williams <>
11 years agoimsm: fix missing initializations of the per-disk extents pointer
Dan Williams [Tue, 24 Feb 2009 06:06:24 +0000 (23:06 -0700)] 
imsm: fix missing initializations of the per-disk extents pointer

Fixes a glibc assertion when trying to free a pointer that was not

Signed-off-by: Dan Williams <>
11 years agotest: fix a call to udevsettle
Dan Williams [Tue, 24 Feb 2009 06:06:24 +0000 (23:06 -0700)] 
test: fix a call to udevsettle

udevsettle is deprecated, use udevadm settle

Signed-off-by: Dan Williams <>
11 years agoimsm: provide a simulated option-rom for regression tests
Dan Williams [Mon, 23 Feb 2009 21:26:10 +0000 (14:26 -0700)] 
imsm: provide a simulated option-rom for regression tests

IMSM_NO_PLATFORM turns off checks that should be tested, so provide a
IMSM_TEST_OROM variable to allow testing the orom constraints in the
mdadm regression suite.

Signed-off-by: Dan Williams <>
11 years agoMonitor: send --test message for arrays in /proc/mdstat that aren't in mdadm.conf
NeilBrown [Thu, 5 Feb 2009 06:06:03 +0000 (17:06 +1100)] 
Monitor: send --test message for arrays in /proc/mdstat that aren't in mdadm.conf

"mdadm --monitor --test --scan" currently only sends test messages for
arrays listed on the command line or in /etc/mdadm.conf.  With this
patch it also reports on any active arrays, which is more in line with
the description in the manpage.

Thanks to  Andrew Walrond <> for reporting this error.

Signed-off-by: NeilBrown <>
11 years agoimsm: block creation of devices with identical names
Dan Williams [Mon, 2 Feb 2009 22:01:13 +0000 (15:01 -0700)] 
imsm: block creation of devices with identical names

Signed-off-by: Dan Williams <>
11 years agoimsm: don't check raid1 chunk size
Dan Williams [Mon, 2 Feb 2009 17:54:58 +0000 (10:54 -0700)] 
imsm: don't check raid1 chunk size

mdadm -C /dev/md/r1d2n1s0-5 -amd -l1  --size 5242880 -n 2 /dev/sdb /dev/sdc  -R -f -v -c 64
mdadm: chunk size ignored for this level
mdadm: super0.90 cannot open /dev/sdb: Device or resource busy
mdadm: super1.x cannot open /dev/sdb: Device or resource busy
mdadm: platform does not support a chunk size of: 0
mdadm: device /dev/sdb not suitable for any style of array

Reported-by: Krzysztof Wojcik <>
Tested-by: Jacek Danecki <>
Signed-off-by: Dan Williams <>
11 years agoMerge branch 'master' into devel-3.0
NeilBrown [Mon, 2 Feb 2009 00:09:09 +0000 (11:09 +1100)] 
Merge branch 'master' into devel-3.0

11 years agoFix possible crash if bitmap metadata is bad.
NeilBrown [Mon, 2 Feb 2009 00:06:38 +0000 (11:06 +1100)] 
Fix possible crash if bitmap metadata is bad.

We really should never divide by 0.

Thanks to "Jon Nelson" <>
for finding the problem.

Signed-off-by: NeilBrown <>
11 years agoDocument 'max' option to --grow --size in --help output.
NeilBrown [Sun, 1 Feb 2009 23:58:08 +0000 (10:58 +1100)] 
Document 'max' option to --grow --size in --help output.

Suggestion from Christian Hudon <>

Signed-off-by: NeilBrown <>