David Zeuthen [Tue, 11 Nov 2008 13:18:45 +0000 (08:18 -0500)]
use right include path for devkit-gobject
David Zeuthen [Tue, 12 Aug 2008 16:33:52 +0000 (12:33 -0400)]
update location of git repos
Someone also needs to do s/devicekit/DeviceKit/ on the fd.o git servers...
David Zeuthen [Fri, 1 Aug 2008 19:46:42 +0000 (15:46 -0400)]
avoid closing the tray on optical drives
You're also going to need this patch to udev
http://marc.info/?l=linux-hotplug&m=
121749053301147&w=2
David Zeuthen [Mon, 21 Jul 2008 01:31:14 +0000 (21:31 -0400)]
update license so it's GPLv2
David Zeuthen [Mon, 21 Jul 2008 00:32:41 +0000 (20:32 -0400)]
fix 'make distcheck'
David Zeuthen [Mon, 21 Jul 2008 00:27:37 +0000 (20:27 -0400)]
fix marshaller changes
David Zeuthen [Fri, 18 Jul 2008 16:47:33 +0000 (12:47 -0400)]
include a private program to export the device-mapper information we need
Can't wait anymore for this to land in device-mapper upstream
https://bugzilla.redhat.com/show_bug.cgi?id=438604
We're using a separate name space for this so once this information
does land in device-mapper upstream we can use it from there.
David Zeuthen [Sun, 29 Jun 2008 01:05:27 +0000 (21:05 -0400)]
add some notes about how to support iSCSI target/initiator roles
David Zeuthen [Wed, 18 Jun 2008 15:00:01 +0000 (11:00 -0400)]
fix a bug introduced by commit
01f415b7e0049d71fbb221a43225d2ec7aa6404b
E.g. use the proper type boxed type for DBUS_G_TYPE_OBJECT_PATH for
the C closures when registering the signals
David Zeuthen [Wed, 18 Jun 2008 14:34:41 +0000 (10:34 -0400)]
register the correct marshaller for the devkit-disks client
David Zeuthen [Wed, 11 Jun 2008 15:37:37 +0000 (11:37 -0400)]
use object paths instead of strings for signals
Pointed out by Lennart Poettering via Jon McCann.
David Zeuthen [Wed, 11 Jun 2008 00:32:28 +0000 (20:32 -0400)]
don't S.M.A.R.T. data injected with simulate= to the database
David Zeuthen [Sun, 8 Jun 2008 01:15:32 +0000 (21:15 -0400)]
forgot to commit a few PolicyKit actions
David Zeuthen [Sun, 8 Jun 2008 01:05:03 +0000 (21:05 -0400)]
add a new method FilesystemListOpenFiles() returning (pid, uid, cmdline)
http://people.freedesktop.org/~david/gdu-fs-busy.png
David Zeuthen [Sat, 7 Jun 2008 03:39:48 +0000 (23:39 -0400)]
add DriveEject() and drive-is-media-ejectable, drive-requires-eject properties
David Zeuthen [Sat, 7 Jun 2008 01:33:31 +0000 (21:33 -0400)]
removed the .NotLabeled error, just return .Failed instead
David Zeuthen [Sat, 7 Jun 2008 01:17:04 +0000 (21:17 -0400)]
add support for optical discs and rework media type handling
In particular, use the data exported by the newly rewritten cdrom_id
program; reported a few bugs to Kay about that one.
David Zeuthen [Fri, 6 Jun 2008 22:12:22 +0000 (18:12 -0400)]
update devkit-disks to latest D-Bus API
David Zeuthen [Fri, 6 Jun 2008 01:24:47 +0000 (21:24 -0400)]
allow unmount/locking devices not set up by the daemon
Will require plenty of privilege (e.g. unmount-others and lock-others).
David Zeuthen [Fri, 6 Jun 2008 01:15:17 +0000 (21:15 -0400)]
reword some authentication messages and change some of the defaults
David Zeuthen [Fri, 6 Jun 2008 01:08:25 +0000 (21:08 -0400)]
rename some PolicyKit actions to be more consistent with the API
David Zeuthen [Thu, 5 Jun 2008 22:02:34 +0000 (18:02 -0400)]
use mlabel if label passed to mkfs.vfat is > 11 characters
David Zeuthen [Thu, 5 Jun 2008 21:43:06 +0000 (17:43 -0400)]
set an environment variable to skip checks
Kay Sievers reports these checks are broken. Either way, no need to
check the fstype since we already check the fstype from vol_id.
David Zeuthen [Thu, 5 Jun 2008 21:38:18 +0000 (17:38 -0400)]
use mlabel rather than dosfslabel
Because compared to dosfslabel
- mlabel actually works
- mlabel supports long labels (254 characters)
- mlabel is also available in e.g. SUSE; dosfslabel seem to be a RH hack
David Zeuthen [Thu, 5 Jun 2008 21:22:36 +0000 (17:22 -0400)]
revert "rename FilesystemSetLabel to SetLabel; some non-fs devices supports labels too"
This reverts commit
f7ad692c880c44dd9d52844f9963e08e24ec92cf.
Hmm.. thinking about it, it's probably better to have SwapSetLabel(),
LinuxMdSetLabel() and so forth.
David Zeuthen [Thu, 5 Jun 2008 20:42:04 +0000 (16:42 -0400)]
rename FilesystemSetLabel to SetLabel; some non-fs devices supports labels too
David Zeuthen [Thu, 5 Jun 2008 18:43:17 +0000 (14:43 -0400)]
sanitize/rework D-Bus errors returned
David Zeuthen [Thu, 5 Jun 2008 05:20:09 +0000 (01:20 -0400)]
avoid abstracting encrypted devices; instead be specific about it's LUKS
David Zeuthen [Thu, 5 Jun 2008 04:27:04 +0000 (00:27 -0400)]
add support for the xfs file system
David Zeuthen [Wed, 4 Jun 2008 03:40:42 +0000 (23:40 -0400)]
export daemon version and capabilities (encryption and fs support)
David Zeuthen [Tue, 3 Jun 2008 00:08:46 +0000 (20:08 -0400)]
add a new method FilesystemCheck
David Zeuthen [Mon, 2 Jun 2008 16:41:47 +0000 (12:41 -0400)]
include permissions and errors in the docs for the LinuxMdStart method
David Zeuthen [Mon, 2 Jun 2008 15:26:04 +0000 (11:26 -0400)]
rework what authorizations are needed
Lots of changes in this commit including new properties and ABI changes.
David Zeuthen [Sun, 1 Jun 2008 20:31:53 +0000 (16:31 -0400)]
add new property device-is-system-internal
David Zeuthen [Sun, 1 Jun 2008 20:07:33 +0000 (16:07 -0400)]
remove ability to use privileged mount options
People who wants this are advised to use /etc/fstab - as of commit
1aeeb7b6ec2236f20eef2281e30786266098e29e, FilesystemMount() and
FilesystemUnmount() will attempt to mount as the calling user if the
device is referenced in /etc/fstab.
David Zeuthen [Sun, 1 Jun 2008 19:42:07 +0000 (15:42 -0400)]
add options take_ownership_[uid|gid] to FilesystemCreate
This is to support features such as
http://people.freedesktop.org/~david/gdu-take-ownership.png
We might want to add take_ownership_secontext for setting the SELinux
context too. Need to check with the SELinux guys.
David Zeuthen [Sun, 1 Jun 2008 15:18:16 +0000 (11:18 -0400)]
add a --replace option to devkit-disks-daemon
This will replace an existing instance of devkit-disks-daemon.
David Zeuthen [Sat, 31 May 2008 18:38:19 +0000 (14:38 -0400)]
ensure that EncryptedLock doesn't return until cleartext device is removed
David Zeuthen [Sat, 31 May 2008 02:10:02 +0000 (22:10 -0400)]
remove debug spew
David Zeuthen [Sat, 31 May 2008 02:08:57 +0000 (22:08 -0400)]
fix unmount issue where directory isn't removed
David Zeuthen [Sat, 31 May 2008 01:48:45 +0000 (21:48 -0400)]
use a mount monitor based on /proc/mounts instead of gio's mount monitor
Also add a hack to work around the fact that device-mapper create it's
own device nodes.
David Zeuthen [Sat, 31 May 2008 00:06:35 +0000 (20:06 -0400)]
support UUID= and LABEL= in /etc/fstab
David Zeuthen [Fri, 30 May 2008 23:51:09 +0000 (19:51 -0400)]
fixup docs now that we support mounting/unmounting /etc/fstab devices
David Zeuthen [Fri, 30 May 2008 23:39:25 +0000 (19:39 -0400)]
for fstab mount points, call mount/umount as the calling user
With this change, the mount client (e.g. gnome-mount) don't have to
fool around with doing this himself.
David Zeuthen [Fri, 30 May 2008 21:40:56 +0000 (17:40 -0400)]
fix up some error handling
David Zeuthen [Fri, 30 May 2008 21:35:53 +0000 (17:35 -0400)]
add -Wformat and -Wformat-security warnings and fix up some code
I'm still baffled why gcc thinks this is something one needs to *turn* on.
David Zeuthen [Mon, 26 May 2008 15:06:23 +0000 (11:06 -0400)]
initialize error correctly
David Zeuthen [Mon, 26 May 2008 13:35:43 +0000 (09:35 -0400)]
properly detect mmc/sd cards and fix a bug in mount point naming
David Zeuthen [Wed, 7 May 2008 06:03:17 +0000 (02:03 -0400)]
update TODO list and bump version to 002
David Zeuthen [Sun, 4 May 2008 15:42:48 +0000 (11:42 -0400)]
store historical SMART data and add DriveSmartGetHistoricalData()
This add sqlite3 as a dependency.
David Zeuthen [Sat, 3 May 2008 22:21:39 +0000 (18:21 -0400)]
refresh SMART data on startup and every 30 minutes
David Zeuthen [Sat, 3 May 2008 21:43:24 +0000 (17:43 -0400)]
don't run DriveSmartRefreshData as a job
We also want to run this from the daemon every 30 minutes or so.
David Zeuthen [Sat, 3 May 2008 21:16:09 +0000 (17:16 -0400)]
rework how SMART is done; express it in properties and export attributes
Also add some test files from the smartmontools project and provide a
new simulate= option. This is very useful for testing clients
consuming this data. See test/smart/README for details.
David Zeuthen [Mon, 28 Apr 2008 22:59:40 +0000 (18:59 -0400)]
rip out the inhibit code from the daemon
In the future we want the daemon around so it can continously monitor
RAID and SMART. Need to figure out if we need to start it at boot
time.
David Zeuthen [Mon, 28 Apr 2008 18:52:16 +0000 (14:52 -0400)]
add some more docs
David Zeuthen [Mon, 28 Apr 2008 03:57:54 +0000 (23:57 -0400)]
rename some methods so the API is easier to read
David Zeuthen [Mon, 28 Apr 2008 00:22:20 +0000 (20:22 -0400)]
work more on the D-Bus docs
David Zeuthen [Sun, 27 Apr 2008 18:50:26 +0000 (14:50 -0400)]
add some polish to the D-Bus documentation
David Zeuthen [Sun, 27 Apr 2008 18:49:46 +0000 (14:49 -0400)]
finish port to new DeviceKit
David Zeuthen [Sat, 26 Apr 2008 01:15:31 +0000 (21:15 -0400)]
properly document the D-Bus interfaces
David Zeuthen [Fri, 25 Apr 2008 18:00:38 +0000 (14:00 -0400)]
port to new DeviceKit interface
This still needs a bit of work...
David Zeuthen [Thu, 10 Apr 2008 03:47:04 +0000 (23:47 -0400)]
Merge branch 'master' of ssh://david@people.freedesktop.org/~/DeviceKit-disks
David Zeuthen [Thu, 10 Apr 2008 03:46:31 +0000 (23:46 -0400)]
avoid gvfs for now
see http://bugzilla.gnome.org/show_bug.cgi?id=526454
David Zeuthen [Wed, 9 Apr 2008 05:32:07 +0000 (01:32 -0400)]
clear fs signatures if we detect a partition table
This is one reason we need to get partition table scanning into
vol_id. This is to fix a case where there's an old RAID signature for
a whole disk RAID component.
David Zeuthen [Wed, 9 Apr 2008 05:10:31 +0000 (01:10 -0400)]
fix warnings on x86_64
David Zeuthen [Sun, 6 Apr 2008 07:26:31 +0000 (03:26 -0400)]
add a 'last_self_test_result' return value to the RetrieveSmartData() method
David Zeuthen [Sat, 5 Apr 2008 04:09:52 +0000 (00:09 -0400)]
fix 'make distcheck'
David Zeuthen [Sat, 5 Apr 2008 03:07:23 +0000 (23:07 -0400)]
work some more on linux-md software RAID
Sorry for the big patch.
David Zeuthen [Mon, 31 Mar 2008 19:12:23 +0000 (15:12 -0400)]
add methods to assemble and stop linux md raid arrays
The AssembleLinuxMdArray() method is on the daemon; part of this patch
extends the job facility to be used when device == NULL.
David Zeuthen [Mon, 31 Mar 2008 01:32:08 +0000 (21:32 -0400)]
add support for recognizing md devices
David Zeuthen [Sun, 30 Mar 2008 23:49:28 +0000 (19:49 -0400)]
add support for recognizing Linux md components
Depends on a patch to udev for making vol_id export some more
information. Sent to Kay for review.
David Zeuthen [Sun, 30 Mar 2008 22:44:51 +0000 (18:44 -0400)]
improve md device handling
David Zeuthen [Sun, 30 Mar 2008 20:12:49 +0000 (16:12 -0400)]
also fail RetrieveSmartData if smartctl returns 0x01
David Zeuthen [Sun, 30 Mar 2008 18:38:29 +0000 (14:38 -0400)]
tell libparted to poke the kernel on partition table changes
David Zeuthen [Sun, 30 Mar 2008 18:38:13 +0000 (14:38 -0400)]
make the task name depend on the test we're running
David Zeuthen [Sun, 30 Mar 2008 06:43:38 +0000 (02:43 -0400)]
first stab at S.M.A.R.T. support
Yeah, yuck, screenscraping the output of smartctl(8) sucks. But there
is no library. Have to think about how to export more data such as the
attributes. Also needs to make this work with SCSI drives as the
smartctl(8) output depends on whether it's SCSI or ATA (!).
http://people.freedesktop.org/~david/gdu-smart-and-healthy.png
http://people.freedesktop.org/~david/gdu-smart-and-failing.png
David Zeuthen [Sat, 29 Mar 2008 18:38:35 +0000 (14:38 -0400)]
split drive-media into two properties (compatibility and current)
David Zeuthen [Sat, 29 Mar 2008 07:42:36 +0000 (03:42 -0400)]
add support for fslabel change, read_only and (partial) linux md raid
David Zeuthen [Thu, 27 Mar 2008 02:13:47 +0000 (22:13 -0400)]
add another card reader
David Zeuthen [Thu, 27 Mar 2008 01:30:35 +0000 (21:30 -0400)]
introduce a new property 'drive-media' to specify media compatibility
Also introduce a few hackish udev rules to set it to something
sane. As the comments in 95-devkit-disks.rules says, this needs to be
revisited.
David Zeuthen [Wed, 26 Mar 2008 19:20:40 +0000 (15:20 -0400)]
add new properties drive-connection-interface and drive-connection-speed
David Zeuthen [Tue, 25 Mar 2008 18:19:01 +0000 (14:19 -0400)]
only tear down crypto cleartext devices set up by ourselves
David Zeuthen [Tue, 25 Mar 2008 09:06:52 +0000 (05:06 -0400)]
fix a misleading comment
David Zeuthen [Tue, 25 Mar 2008 09:03:50 +0000 (05:03 -0400)]
tear down crypto devices when a device is unsafely removed
Streamline the whole way force removal works. It shouldn't be hard to
add force removal to RAID (e.g. md) and LVM arrays once we get around
to adding support for these.
We still need to check whether the encrypted device was set up by us
before tearing it down. That's for the next patch.
David Zeuthen [Tue, 25 Mar 2008 07:35:22 +0000 (03:35 -0400)]
forcibly unmount devices mounted by ourselves on unsafe media removal
David Zeuthen [Tue, 25 Mar 2008 04:16:52 +0000 (00:16 -0400)]
make the fd's used to communicate with childs non blocking
This fixes a grave bug where the daemon was unresponsive for many many
seconds. Especially during ext3 fs creation. Also read to end for
stderr and stdout when the child terminates. This fixes a long
standing bug where sometimes we didn't get all the data.
We now also pass stdout to the JobCompleted callback. Might come in
handy later.
David Zeuthen [Tue, 25 Mar 2008 02:40:57 +0000 (22:40 -0400)]
make sure holders/ is properly updated
... and remove some debugging spew as well.
David Zeuthen [Tue, 25 Mar 2008 01:54:38 +0000 (21:54 -0400)]
make libparted poke the kernel when adding/deleting partitions
Turns out that libparted is already using the right ioctl's mentioned
in commit
6a2a1f2b7cc7b52b277aa9788a3a0bb280227441. So now we can
correctly add/delete partitions even when other partitions on the same
disk are busy.
David Zeuthen [Tue, 25 Mar 2008 00:33:55 +0000 (20:33 -0400)]
to maintain integrity, use the new is-busy property to refuse service
For example, we now throw the .IsBusy exception on CreatePartitionTable()
if one of the existing partitions is busy.
Prior to this patch we'd happily create a new partition table even
though a dm-crypt target were holding one of the partitions. Which
broke in interesting ways when we issued BLKRRPART. This is because
the kernel is rather dumb and BLKRRPART won't work if any of the
partitions are busy.
The solution, I think, involves judicious use of the "add partition"
and "del partition" ioctls to spoonfeed the kernel this information.
Will try and do this next. As an upside, I think the annoying "remove
all partitions, then add them back again" artifact will go away with
this. So maybe it's all good. We'll see.
David Zeuthen [Mon, 24 Mar 2008 23:25:59 +0000 (19:25 -0400)]
introduce a new property device-is-busy
David Zeuthen [Mon, 24 Mar 2008 05:40:43 +0000 (01:40 -0400)]
add support for option encrypt= for CreateFilesystem
This makes it painfully simple to do UI's like this
http://people.freedesktop.org/~david/gdu-luks-easy.png
David Zeuthen [Sun, 23 Mar 2008 23:20:46 +0000 (19:20 -0400)]
ignore temporary private cryptsetup devices
For this to work you'll need HEAD of DeviceKit with this patch
http://gitweb.freedesktop.org/?p=users/david/DeviceKit.git;a=commitdiff;h=
3a11b7f9a5fe0efb1918055ec49516b648e45819
David Zeuthen [Sun, 23 Mar 2008 23:03:22 +0000 (19:03 -0400)]
add and implement new ChangeSecretForEncrypted method call
David Zeuthen [Sun, 23 Mar 2008 19:22:02 +0000 (15:22 -0400)]
update the job state before returning value of the the method called
David Zeuthen [Sun, 23 Mar 2008 18:10:11 +0000 (14:10 -0400)]
add support for UnlockEncrypted and LockEncrypted methods
On Linux this is using cryptsetup(8).
David Zeuthen [Sat, 22 Mar 2008 20:32:39 +0000 (16:32 -0400)]
add support for encrypted block devices
The namespace is abstract and I've implemented this for dm-crypt for
now (not sure what else there is to support on Linux).
This relies on some device-mapper patches, see
https://bugzilla.redhat.com/show_bug.cgi?id=438604
that needs to go upstream. Next patch will include methods for
locking, unlocking and managing keys.
David Zeuthen [Sat, 22 Mar 2008 06:40:22 +0000 (02:40 -0400)]
better way of avoiding detecting a VBR as an MBR
David Zeuthen [Sat, 22 Mar 2008 06:30:04 +0000 (02:30 -0400)]
Revert "rely on proper change events from the kernel and udev"
This reverts commit
071b5825390e537bfbc6650582a81c427017a9e6.
David Zeuthen [Sat, 22 Mar 2008 06:28:15 +0000 (02:28 -0400)]
rely on proper change events from the kernel and udev
Relies on patched udev rules as mentioned here
http://gitweb.freedesktop.org/?p=users/david/gnome-disk-utility.git;a=commitdiff;h=
d4098ad91e73414afdd8b49a8c57b42d31ff7998
David Zeuthen [Sat, 22 Mar 2008 06:17:47 +0000 (02:17 -0400)]
if creating a fs on the main device, signal kernel to reread partition table