of: find also nodes by mixture of alias and path ...

Let of_find_node_by_path_or_alias also find a node starting with an alias
followed by a path like "i2c0/tps@24"

Signed-off-by: Jan Weitzel <j.weitzel@phytec.de>
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

OF: parse OF clock providers after populate ...

This adds a call to of_clk_init right after platform devices have been
populated.

Signed-off-by: Sebastian Hesselbarth <sebastian.hesselbarth@gmail.com>
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

of: simplify phandle lookup ...

Instead of populating an extra list containing all phandles just iterate
over the whole tree. This is done as preparation for more dynamic devicetrees
where parts are loaded at runtime. Here we don't want to keep the list of
phandles in sync.

Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

Merge branch 'for-next/boardinfo' ...

Conflicts:
	arch/mips/boards/qemu-malta/init.c
	commands/bootm.c
	drivers/of/base.c

Merge branch 'for-next/of'

of: base: import of_find_node_by_type ...

Import of_find_node_by_type from Linux drivers/of/base.c -
commit id d8dfad3.
This function retrieves a node pointer based on the "device_type"
property of the node.

This is used by device tree update functions in PPC support.

Signed-off-by: Renaud Barbier <renaud.barbier@ge.com>
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

introduce barebox_set_model ...

Instead of calling of_get_model() in barebox_get_model() add a
barebox_set_model() and use it to set the boardinfo once it's
available from the devicetree.

Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

of: base.c: Use correct device-tree in of_add_initrd() ...

Otherwise only the internal DT is modified, not necessarily the one passed
to the kernel.

Signed-off-by: David Jander <david@protonic.nl>
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

of: introduce some new helpers ...

of_get_tree_max_phandle - get the maximum phandle of a tree. Needed for
                          creating new phandles without conflicts.
of_node_create_phandle - create a phandle for a node which doesn't have one.
of_find_node_by_alias - find a node by alias name
of_find_node_by_path_or_alias - find a node by full path or alias name
of_find_root_node - find the root node for a given device node

Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

of: default to internal tree in of_find_node_by_path_from ...

So that of_find_node_by_path_from can be used easier.

Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

of: parse phandles during unflatten ...

With this node->phandle becomes valid after unflattening a tree
and not during of_probe later.

Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

of: Fix endless loop in of_add_memory ...

of_add_memory is stuck in an endless loop when a memory bank with zero
size is passed. Fix this.

Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

Merge branch 'for-next/of' ...

Conflicts:
	arch/arm/boards/freescale-mx53-loco/board.c
	drivers/of/Makefile

Merge branch 'for-next/marvell' ...

Conflicts:
	arch/arm/boards/Makefile
	arch/arm/dts/Makefile

of: Use dts syntax when printing devicetrees ...

Our devicetree printing is close to correct dts syntax, so fix
some remaining differences:

- Use an equal sign instead of a colon to separate a poperty name and
  a value
- Add a semicolon at the end of properties
- Make sure we do not print a separator for empty properties

Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

of: Add convenience functions to en/disable devicenodes ...

These functions allow to manipulate the "status" property of
devicenodes effectively enabling/disabling devices.

Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

OF: base: initalize from node in of_find_node_with_property ...

This adds initialization of from device node with root_node if NULL
is passed. This corresponds to the behavior of all other node iterators.

Signed-off-by: Sebastian Hesselbarth <sebastian.hesselbarth@gmail.com>
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

OF: base: fix iterator in of_get_next_available_child ...

of_get_next_available_child does not iterate but always tries the
same node over and over again. This first prepares the entry and
then uses list_for_each_entry_continue, instead of for_each_child_of_node
before.

Signed-off-by: Sebastian Hesselbarth <sebastian.hesselbarth@gmail.com>
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

OF: base: fix inifinite looping in node iterators ...

of_find_node_by_name suffers from infinite looping, because it
does not check for root node of the tree iterated over. This
fixes this by checking for node->parent to determine whether
the last node has been reached.
Since of_tree_for_each_node does not iterate over the whole tree,
but only over the remaining nodes, rename it to
of_tree_for_each_node_from.

Reported-by: NISHIMOTO Hiroki <hiroki.nishimoto.if@gmail.com>
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

OF: base: rename of_free to of_delete_node ...

of_free is misleading about the actual purpose of the function. There is
already a of_create_node counterpart, so rename of_free to of_create_node
and update all users accordingly.

Signed-off-by: Sebastian Hesselbarth <sebastian.hesselbarth@gmail.com>
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

OF: base: convert of_add_memory to OF API ...

Convert of_add_memory parsing to make use of of_address_to_resource
instead of parsing memory ranges itself. This makes some functions
dead code which are also removed.

Signed-off-by: Sebastian Hesselbarth <sebastian.hesselbarth@gmail.com>
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

OF: remove device and resource pointer from struct device_node ...

struct device_node has its own resources and a pointer to associated
device_d. With recent platform related OF code, we can convert the
only user of it and remove those pointers from struct device_node.

Signed-off-by: Sebastian Hesselbarth <sebastian.hesselbarth@gmail.com>
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

OF: base: remove dead device related functions ...

With recent conversion to of_platfrom_populate_some functions are now
dead code and can be removed.

Signed-off-by: Sebastian Hesselbarth <sebastian.hesselbarth@gmail.com>
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

OF: base: use of_platform_populate for probing ...

With recent bus/device related functions in OF API, we can now
convert to use of_platform_populate.

Signed-off-by: Sebastian Hesselbarth <sebastian.hesselbarth@gmail.com>
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

OF: base: move OF_ROOT_NODE_ defines to local OF code ...

OF_ROOT_NODE_* defines should not be used outside of base.c. Move them
to drivers/of/base.c to ensure they will not be used elsewhere.

Signed-off-by: Sebastian Hesselbarth <sebastian.hesselbarth@gmail.com>
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

OF: convert of_translate_address to new API ...

This converts existing of_translate_address to recently added API. In
contrast to existing behavior, the new function honors ranges properties
properly. It now allows reg properties to be set as offset with respect
to the correspoding parent node.

Signed-off-by: Sebastian Hesselbarth <sebastian.hesselbarth@gmail.com>
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

OF: base: use of_delete_property where applicable ...

This patch makes OF API use of_delete_property where applicable instead
of freeing allocated data in different places.

Signed-off-by: Sebastian Hesselbarth <sebastian.hesselbarth@gmail.com>
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

OF: base: add sanity checks to of_new/delete_property ...

This adds some sanity checks to of_new_property and of_delete_property.
Also, value pointer is always allocated even with zero length to allow
empty properties to be distinguished from non-existing properties.
Finally, data passed to of_new_property is only copied if non-NULL.

Signed-off-by: Sebastian Hesselbarth <sebastian.hesselbarth@gmail.com>
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

OF: base: remove of_find_child_by_name ...

With of_get_child_by_name from Linux API, we can now convert and remove
of_find_child_by_name.

Signed-off-by: Sebastian Hesselbarth <sebastian.hesselbarth@gmail.com>

OF: base: remove of_tree_for_each_node from public API ...

This patch converts users of of_tree_for_each_node to recently added
for_eacg_compatible_node helper. Also of_tree_for_each_node is removed
from public OF API.

Signed-off-by: Sebastian Hesselbarth <sebastian.hesselbarth@gmail.com>