Age | Commit message (Collapse) | Author | Files | Lines |
|
|
|
changes post Linux 2.6.17: (hotplug_event_begin_sysfs): Prepare for new
class devices showing up in /sys/devices instead of /sys/class.
Bump from 256 to 512 as the kernel emits longer sysfs paths post Linux
2.6.17. And udev now also uses 512.
|
|
|
|
|
|
|
|
|
|
enviroment by default (hald should never ever specify need to add a
path to the enviroment send to the runner, so we're not overriding
anything...).
|
|
|
|
Sleep three seconds on error to avoid attacks for cracking the password
|
|
my iBook clamshell. This means that the power button now emits
ButtonPressed conditions and the machine thus works with
gnome-power-manager.
|
|
by myself.
Teach libhal-storage about Blu-ray and HD DVD.
|
|
/dev/adb blocking). I've been running a little daemon on my system for
some years that also does this without any problems, so it should be
fine and makes the code somewhat nice imho. I've also removed some old
dead code, don't think there is much use in keeping it around right?
Patch from Sjoerd Simons <sjoerd@luon.net>.
|
|
HD DVD to storage.cdrom and volume.disc.type per the latest MMC-5
draft. See Table 87 - Profile List in
http://www.t10.org/drafts.htm#mmc5 . Probe-volume should do the right
thing, but probe-storage needs fixing (noted in FIXME comment).
|
|
Take an additional parameter "keep_auxgroups"
Add some comment that we *could* drop privileges provided we were in the
right group
Drop privileges
Pass right parameter to drop_privileges
Drop privileges
|
|
|
|
libhal_volume_crypto_get_backing_volume_udi
New function (libhal_volume_crypto_get_backing_volume_udi): New function
fix my commit entry for 2006-02-19 - missed changes to hald/hald_runner.c
|
|
|
|
org.freedesktop.Hal.NoSuchDevice if the device went away while we were
in a queue. Move freeing of the MethodInvocation structure into a
separate function and require callers to call this. Return FALSE if the
device wasn't there. (hald_exec_method_enqueue): Free the method
invocation (hald_exec_method_process_queue): Handle when device wasn't
there and process the next element instead
(local_server_message_handler): Comment out some debug information
|
|
|
|
worked fine, it didn't work so well for crypted volumes on removable
media. This patch should fix it.
New function (unmount_childs): If device is a crypto device, unmount all
cleartext devices and tear down the crypt link.
|
|
Add omnibook support.
don't prepend the acpi/ here as the new omnibook isn't under /proc/acpi.
See http://bugzilla.gnome.org/show_bug.cgi?id=331458 for details.
|
|
called by hal, not a user.
Change the path of hal-system-power-pmu as libexec is now in our path and
we are not installed in scripts anymore.
Add the sleep command for hal-system-power-pmu so that suspend works on ppc
hardware.
|
|
|
|
Pitt <martin@piware.de>.
|
|
would always return NULL.
|
|
|
|
checking symlinks. Patch from Ludwig Nussel <ludwig.nussel@suse.de>.
|
|
now maintain a queue of method calls per-device. One very real example
of this is gnome-mount when setting up crypto devices; after
Crypto.Setup() gnome-mount waits for the cleartext block device and
attempts to mount it; meanwhile g-v-m sees the cleartext block device
too and attempts to mount too. Since mounting is a relatively slow
operation both methods are executed at the same time and we end up with
two mounts. This patch fixes this problem.
|
|
Throw SetupPasswordError instead of just SetupError when cryptsetup is
failing - used to ask for password again if the one from e.g. the
keyring didn't work.
New function (hald_dbus_filter_handle_methods): Handle StringListRemove
|
|
|
|
Also allow comma "," as a valid character in a mount point
|
|
that the error is empty and initialised for each usage. If not we get
potentially error messages from dbus. :
|
|
|
|
Also export Teardown method
Remove, just use hardcoded paths
New file for tearing down the mapping
Use cryptsetup rather than luks-setup, add to CVS
Use cryptsetup rather than luks-setup, add to CVS
Add hal-luks-teardown
|
|
with '_' instead of '?' as the latter is not allowed as a mountpoint.
|
|
|
|
like "foobar der æøå サイトの". Set LANG to en_US.UTF-8 to get
bash to do this. Also allow - in GIVEN_MOUNTOPTIONS (Reported by
Jeffrey Stedfast <fejj@novell.com>) in order to get iocharset= to work.
Make this work for mountpoints with spaces in them
Also remove info.hal_mount properties.
(blockdev_generate_add_hotplug_event): For Reprobe, pass the model,
vendor, serial and firmware rev.
|
|
New file
Clean up mount point if we can see it was created by hal
|
|
info.hal_mount.created_mount_point
Export prototype for new function libhal_ctx_get_dbus_connection()
New function
Use a the Unmount() hal method to lazy unmount
Ugh ugh ugh.. clear buffer before receiving data from udev.. we didn't use
to do this and we got all sorts of weird side-effects since we'd for
instance pick up DEVNAME='/dev/sdb' and _then_ DEVNAME='/dev/sda1' for
/sys/block/sdb.. This was because this data was left in the buffer from
the old run. It now appears to work..
New function, pass end_token to hal_util_callout_device_remove()
(force_unmount): Use the Unmount method call to lazy unmount, thus
making sure we don't block (hotplug_event_begin_remove_blockdev): Pass
end_token to force_unmount and let that function invoke
hal_util_callout_device_remove()
Make this work for helpers using a direct connection to hald
(hald_dbus_filter_handle_methods): Pass local_interface to
hald_exec_method
Update to use 'pwd' just like run-hald.sh was updated
|
|
|
|
0). Use properties stored on hal to figure out mount location.
Don't advertise flush as a valid option as it is not upstream (SUSE guys:
you need to patch hal yourself until this is in an upstream kernel
version that a hal release can depend on)
|
|
mount Zip disks as e.g. 'mount /dev/hdd4 /media/foo' fails but 'mount
-t vfat /dev/hdd4 /media/foo' works.
|
|
the device. This also ensures we only unmount stuff that we mounted
ourselves ... where we = the hal mount/unmount methods. Also only allow
the user who mounted the volume to unmount it (also allow uid 0).
Allow mounting of drives we don't poll. Export mount point and what uid
mounted it as properties on the hal device object
Fix up detection of blank and audio discs for the purpose of adding Eject
methods. Export Mount, Unmount and Eject methods on drives we don't
poll. Give some leeway with what mount options we support.
|
|
|
|
add_blockdev_probing_helper_done() in the wrong position.
|
|
|
|
|
|
Don't set button.type or button.has_state as these are merged from the fdi
file.
Use absolute paths in HALD_RUNNER_PATH and PATH cuz otherwise helpers won't
get invoked...
Add 10-keyboard-policy.fdi
Forgot to add file. Also from Matthew Garrett <mjg59@srcf.ucam.org>. Tweak
file to merge button.type and button.has_state instead of setting
|
|
Two patches attached: 1) Changes the ACPI addon to report the type of
button pressed as well as the fact that a button has been pressed. 2)
Adds a keyboard addon. The included fdi file will bind it to anything
that has type input.keyboard. It adds the capability "button" to the
keyboard devices, and on any keypress checks the key against a table.
If the key is found there, a ButtonPressed event is generated with the
name of the key in the details field (one exception - KEY_SUSPEND is
given the name "hibernate").
There's a couple of extra features that could be added (export a list of
the available keys, for instance - not very helpful on PS/2, but useful
on USB), but could these be applied as is for now?
Move drop_privileges here so other code can use it too.
Add build rules for hald-addon-keybaord
Also send button.type as part of the event
New file. Also set button.type to "" and button.has_state to FALSE
|
|
BUS_HOST. These are pieces of built-in hardware, so won't be
hotpluggable. Patch from Matthew Garrett <mjg59@srcf.ucam.org>.
|