Age | Commit message (Collapse) | Author | Files | Lines |
|
|
|
The change in locking semantics is that a lock-holder can now access a
device even if other processes have locks on the device.
|
|
|
|
|
|
Now to write some docs how all this works...
|
|
This provides a mechanism for clients to acquire (and release) named
locks on device objects. More than one client can hold a named lock on
a device, this is by design. If a client disconnects from the bus, it
locks will be removed.
The rationale for this new API is to allow clients to lock out other
clients of service on given device objects; this is applicable for
- installers (inhibit mount requests)
- partitioning/formatting tools (inhibit mount requests)
- multi-seat power management (inhibit suspend/hibernate etc. requests
from idle (e.g. the user is idle) active (e.g. it is in the foreground
on some seat) sessions that wants to suspend due to user preferences.)
The next series of patches will make the various mechanisms use this
API by defining well-known locks.
|
|
Correct the python shebang line so that python can be found otuside of
/usr/bin.
|
|
This is useful when debugging the Mount/Unmount functions.
|
|
|
|
|
|
Remove libhal-policy, examples/volumed (it's defunt) and packaging
(both Fedora and SUSE spec files look vastly different nowadays).
|
|
and keyboard_backlight additions including example.
Remove ChangeLog and add rules to Makefile.am (from cairo) to generate
the ChangeLog file from git history.
Add support for light_sensor and keyboard_backlight on my Macbook
Pro. Add an examply python script to examples/ to show how it's done.
Move Unmount, Eject and related stuff to C. Start using the file
/media/.hal-mtab instead of .created-by-hal to specify if HAL created
the mount point. Also list tons of metadata such as uid, session id
(still unused). Fixes the bug where the unprivileged hald couldn't
look into /media/foobar/ for the .created-by-hal file. So now doing
'umount /dev/sda1' on a volume mounted by HAL, makes hald remove the
mount point as one would expect.
|
|
|
|
|
|
it wears out their drives (RH bug #157674)
Fixup this a little bit
Update the man page to reflect latest changes.
Fix a doublefree when acquiring lock (RH bug #158474)
|
|
consistent with new D-BUS naming, e.g. use info.locked.dbus_name
instead of info.lock.dbus_service (device_unlock): -do-
(service_deleted): Use NameOwnerChanged instead of NameLost as the
latter only goes to the service who's lost the name
(hald_dbus_filter_function): -do- (hald_dbus_init): Subscribe to
NameOwnerChanged rather than NameLost
Fixup to use get_unique_name() instead of get_base_service()
|
|
(agent_manager_commit_to_gdl): Removed (agent_manager_new_device):
Removed (filter_function): Removed all the AgentManager methods
(raise_udi_in_use): Removed
Add docs for info.locked. Remove section about HAL agents as they are now
gone. Fixed up renaming of scsi_device to scsi.
Remove HAL Agents
Update diagram since we now use a local socket from hal.hotplug and hal.dev
helpers.
New file; shows how to use locking
|
|
|
|
|
|
Add match for udev signals (osspec_hotplug): renamed to handle_hotplug
since old name didn't make any sense (handle_udev_node_created): new
function (handle_udev_node_created_found_device): new function
(osspec_filter_function): add check for udev signals
sleep a while so the kernel has time to populate sysfs (hmmm)
|
|
messages from udev (setup_udev_listener): setup filter for udev
messages (hal_monitor_enter): Listen to D-BUS messages
Use global sysfs path (device_hotplug_add): Use global sysfs path
(device_hotplug_remove): Use global sysfs path (drivers_collect): Use
global sysfs path (mainloop_integration): Save D-BUS connection object
for later use (main): Get sysfs mount path once and for all
Add dbus_connection and sysfs_mount_path as extern variables
new function (filter_function): add check for Device.QueryCapability
Minor formatting stuff
Crude example of volume manager now that we got udev integration in place.
It doesn't really mount anything but prints out when it should mount.
You'll need a very recent udev from BitKeeper with D-BUS enabled (got
the patch accepted today)
|