Age | Commit message (Collapse) | Author |
|
|
|
|
|
|
|
Symlinks can have priorities now, the priority is assigned to the device
and specified with OPTIONS="link_priority=100". Devices with higher
priorities overwrite the symlinks of devices with lower priorities.
If the device, that currently owns the link goes away, the symlink
will be removed, and recreated, pointing to the next device with the
highest actual priority.
This should solve the issue, that inserting an USB-stick may overwrite the
/dev/disk/by-id/-link of another disk, and removes the entire link after the
USB-stick is disconnected. If no priorities are specified, the new link will
overwrite the current one, and if the device goes away, it will restore
the old link. It should be possible to assign lower priorities to removable
devices, if needed.
In multipath setups, we see several devices, which all connect to the same
volume, and therefore all try to create the same metadata-links. The
different path-devices are combined into one device-mapper device, which also
contains the same metadata. It should be possible, to assign multipath-table
device-mapper devices a higher priority, so path-devices that appear and
disappear, will not overwrite or delete the device-mapper device links.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
udev_db_add_device() can be called when the corressponding database
entry already exists - it should overwrite the old entry in this case.
However, if the old entry was a symlink, fopen(filename, "w") will not
overwrite it properly - it will keep the symlink and create a file
named after the symlink target. Calling unlink(filename) before
trying to create the database file fixes the problem.
Signed-off-by: Sergey Vlasov <vsu@altlinux.ru>
|
|
|
|
|
|
|
|
|
|
devices
This scheme is more consistent and makes it obvious if a match happens
against the event device only, or the full chain of parent devices.
The old key names are now:
BUS -> SUBSYSTEMS
ID -> KERNELS
SYSFS -> ATTRS
DRIVER -> DRIVERS
Match keys for the event device:
KERNEL
SUBSYSTEM
ATTR
DRIVER (in a future release, for now the same as DRIVERS)
Match keys for all devices along the parent device chain:
KERNELS
SUBSYSTEMS
ATTRS
DRIVERS
ID, BUS, SYSFS are no longer mentioned in the man page but still work.
DRIVER must be converted to DRIVERS to match the new scheme. For now,
an error is logged, if DRIVER is used. In a future release, the DRIVER
key behaviour will change.
|
|
If we don't have any interesting values to store, we just use
a symlink (doesn't occupy a whole page on tmpfs like a file)
to store the name of the node belonging to a devpath. This will
allow udevinfo to resolve all names, even when we don't have a
full database entry.
|
|
We never used any of the libsysfs convenience features. Here we replace
it completely with 300 lines of code, which are much simpler and a bit
faster cause udev(d) does not open any syfs file for a simple event which
does not need any parent device information.
Signed-off-by: Kay Sievers <kay.sievers@suse.de>
|
|
Signed-off-by: Kay Sievers <kay.sievers@suse.de>
|
|
Thanks for Marco, who noticed it.
Signed-off-by: Kay Sievers <kay.sievers@suse.de>
|
|
All udev state is kept in /$udev_root/.udev/ now. No option to
configure that anymore, it will always be there.
Signed-off-by: Kay Sievers <kay.sievers@suse.de>
|
|
|
|
|
|
Signed-off-by: Kay Sievers <kay.sievers@suse.de>
|
|
Signed-off-by: Kay Sievers <kay.sievers@suse.de>
|
|
HAL soon wants to read the whole content of the udevdatabase while
starting up. This makes the whole udev structure available to the
udevinfo "dump".
Signed-off-by: Kay Sievers <kay.sievers@suse.de>
|
|
Signed-off-by: Kay Sievers <kay.sievers@suse.de>
|
|
Signed-off-by: Kay Sievers <kay.sievers@suse.de>
|
|
Signed-off-by: Kay Sievers <kay.sievers@suse.de>
|
|
strlcpy counts the sourec string lengt and is therefore not suitable
to copy a defined length of characters from one string to another.
Signed-off-by: Kay Sievers <kay.sievers@suse.de>
|
|
Signed-off-by: Kay Sievers <kay.sievers@suse.de>
|
|
Device nodes created with the default rule, without any symlink or option
are no longer saved to the udevdb. This saves us ~3 MB RAM for pretty much
useless files on tmpfs.
Note: HAL needs a fix to handle this correctly. It's already available on
the list.
Signed-off-by: Kay Sievers <kay.sievers@suse.de>
|
|
Any program can query with udevinfo for persistent device
attributes evaluated on device discovery now.
Signed-off-by: Kay Sievers <kay.sievers@suse.de>
|
|
udevinfo segfaults cause klibc's strlcpy writes behind the specified
size of the destination string. strlcat truncates the destination
string which is also not what you expect from a concatenation function.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Some broken ide drivers are generating high event traffic, with
add/remove events. With this attribute, it can be specified,
that the node is always available. It may be used in conjunction
with the new DRIVER= match to catch specific kernel device drivers.
|
|
|