FDT property data should not be const -- sometimes we need to update it,
for example when fixing up phandles in an overlay. On the other hand
it's occasionally desirable to put a string constant in there without
having to strdup() it all the time... let's just live with the tiny
implicit assumption that the data we'd want to modify (phandle
references, mostly) will never be added from string constants, and put a
cast in dt_add_string_prop().
Change-Id: Ifac103fcff0520cc427ab9a2aa141c65e12507ac
Signed-off-by: Julius Werner <jwerner@chromium.org>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/32868
Reviewed-by: Hung-Te Lin <hungte@chromium.org>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
This patch updates the device tree dumping functions (not compiled by
default but available for debugging) to output properties and nodes in a
format similar to .dts files that is very close to what dtc outputs when
you decompile a .dtb with it. This makes it easier to match device tree
dumps from coreboot with device tree dumps generated by other device
tree tooling.
This patch was adapted from depthcharge's http://crosreview.com/1536386
Change-Id: Ib40e50d906aff05473a70c4fc9b124d63232558c
Signed-off-by: Julius Werner <jwerner@chromium.org>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/32867
Reviewed-by: Hung-Te Lin <hungte@chromium.org>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Currently DT paths are *not* expected to start with '/'. This is not
what the spec says (see Devicetree Specification v0.2, 2.2.3 Path Names)
and also not what is done by Linux.
Change dt_find_node_by_path() to expect paths to start with '/' and add
a leading '/' to all DT path strings. Besides the compatibility with the
spec this change is also needed to support aliases in the future.
This patch was adapted from depthcharge's http://crosreview.com/1252770
Change-Id: Ibdf59ccbb4ead38c6193b630642fd1f1e847dd89
Signed-off-by: Julius Werner <jwerner@chromium.org>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/32865
Reviewed-by: Hung-Te Lin <hungte@chromium.org>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Besides looking up a node with an absolute path dt_find_node_by_path()
currently also supports finding a sub-node of a non-root node. All
callers of the function pass the root node though, so it seems there
is no real need for this functionality. Also it is planned to support
DT path names with aliases, which would become messy in combination with
the lookup from a sub-node.
Change the interface of dt_find_node_by_path() to receive the DT tree
object instead of a parent node and adapt all callers accordingly.
This patch was adapted from depthcharge's http://crosreview.com/1252769
Change-Id: Iff56be4da2461ae73a7301dcaa315758d2a8c999
Signed-off-by: Julius Werner <jwerner@chromium.org>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/32864
Reviewed-by: Hung-Te Lin <hungte@chromium.org>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
This patch caches phandles when unflattening the device tree, so we
don't have to look up the phandle property again every time we're trying
to find the phandle of a node. This is especially important when
supporting phandle lookups, which are also added. In addition we keep
track of the highest phandle in the whole tree, which will be important
for applying overlays later.
With this, dt_get_phandle(node) becomes obsolete because the phandle is
already available as a member variable in the node.
This patch was adapted from depthcharge's http://crosreview.com/1536385
Change-Id: I9cbd67d1d13e57c25d068b3db18bb75c709d7ebe
Signed-off-by: Julius Werner <jwerner@chromium.org>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/32863
Reviewed-by: Hung-Te Lin <hungte@chromium.org>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
The FIT code is already using xzalloc() everywhere, and that's the only
real consumer of device tree code right now. Chances are if you're
trying to unflatten an FDT and it doesn't fit into the heap you're
pretty much screwed anyway, so all the OOM handling feels a bit
unnecessary (and some functions will just silently fail because they
don't have a return value, which is bad). Let's just switch this all to
die on failed allocations.
Change-Id: I738f24d550a776653b2becd3d4f7d4d2cb3cc048
Signed-off-by: Julius Werner <jwerner@chromium.org>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/32861
Reviewed-by: Hung-Te Lin <hungte@chromium.org>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
* Adapt to coreboot coding style.
* Use coreboot's endian conversion functions.
* Fix header and header guards.
* Get rid of unused functions.
* Add Kconfig to build it on ramstage.
* Replace size32 with ALIGN_UP and DIV_ROUND_UP.
* Add NULL pointer checks
* Convert constants to defines
Required for Cavium's BDK and uImage FIT support.
Change-Id: I6e6cd9f78fb402bd54d684097326d26eb78d552a
Signed-off-by: Patrick Rudolph <patrick.rudolph@9elements.com>
Reviewed-on: https://review.coreboot.org/25523
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Julius Werner <jwerner@chromium.org>
