When merging one device tree over the top of a previous tree, it is
possible to define a duplicate label that has the same name and points
to the same property or node. This is currently allowed by the duplicate
label checking code. However, alternative duplicate label checking
algorithms might not allow this. Add an explicit test to ensure this
capability is maintained.
Signed-off-by: Stephen Warren <swarren@nvidia.com>
Acked-by: David Gibson <david@gibson.dropbear.id.au>
When compiling the current code-base with gcc 4.6.1, the following warning
is raised, which is interpreted as an error:
cc1: warnings being treated as errors
tests/setprop_inplace.c: In function ‘main’:
tests/setprop_inplace.c:62: error: format ‘%016llx’ expects type ‘long long unsigned int’, but argument 2 has type ‘uint64_t’
tests/setprop_inplace.c:68: error: format ‘%016llx’ expects type ‘long long unsigned int’, but argument 2 has type ‘uint64_t’
Use printf format specifiers from <inttypes.h> to solve this.
Signed-off-by: Stephen Warren <swarren@nvidia.com>
Acked-by: David Gibson <david@gibson.dropbear.id.au>
The libfdt read/write functions are now usable enough that it's become a
moderately common pattern to use them to build and manipulate a device
tree from scratch. For example, we do so ourself in our rw_tree1 testcase,
and qemu is starting to use this model when building device trees for some
targets such as e500.
However, the read/write functions require some sort of valid tree to begin
with, so this necessitates either having a trivial canned dtb to begin with
or, more commonly, creating an empty tree using the serial-write functions
first.
This patch adds a helper function which uses the serial-write functions to
create a trivial, empty but complete and valid tree in a supplied buffer,
ready for manipulation with the read/write functions.
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
In device trees in the world, properties consisting of a single 64-bit
integer are not as common as those consisting of a single 32-bit, cell
sized integer, but they're common enough that they're worth including
convenience functions for.
This patch adds helper wrappers of fdt_setprop_inplace(), fdt_setprop() and
fdt_appendprop() for handling 64-bit integer quantities in properties. For
better consistency with the names of these new *_u64() functions we also
add *_u32() functions as alternative names for the existing *_cell()
functions handling 32-bit integers.
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
The testcases based on test_tree1_dts0.dts were added purely to test dtc's
backwards compatibility handling of the old dts-v0 format. Since that
support has been removed, the dts has been updated to use the current
dts-v1 syntax, which makes the testcases pass, but be completely useless.
This patch removes the now obsolete testcases.
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Several files were added, and should be in .gitignore. The *.test.dts
pattern should catch future source files which are generated by tests.
It also subsumes the old *.dtb.test.dts pattern.
Signed-off-by: Simon Glass <sjg@chromium.org>
Written by David Gibson <david@gibson.dropbear.id.au>. Additions by me:
* Ported to ToT dtc.
* Renamed cell to integer throughout.
* Implemented value range checks.
* Allow U/L/UL/LL/ULL suffix on literals.
* Enabled the commented test.
Signed-off-by: Stephen Warren <swarren@wwwdotorg.org>
On Ubuntu, /bin/sh is dash (at least by default), and dash's echo doesn't
accept the -e option. This means that fdtget-runtest.sh's EXPECT file will
contain "-e foo" rather than just "foo", which causes a test failure.
To work around this, run /bin/echo instead of (builtin) echo, which has
more chance of supporting the -e option.
Another possible fix is to change all the #! lines to /bin/bash rather
than /bin/sh, and change run_tests.sh to invoke sub-scripts using $SHELL
instead of just "sh". However, that would require bash specifically, which
may not be desirable.
Signed-off-by: Stephen Warren <swarren@wwwdotorg.org>
Acked-by: David Gibson <david@gibson.dropbear.id.au>
It is often inconvenient to place device tree files in the same directory
as their includes, or to specify the full path to include files.
An example of this is in U-Boot where we have a .dtsi file for each SOC
type, and this is included by the board .dts file. We need to either use
a mechanism like:
/include/ ARCH_CPU_DTS
with sed or cpp to perform the replacement with the correct path, or
we must specify the full path in the file:
/include/ "../../arch/arm/dts/tegra20.dtsi"
The first option is not desirable since it requires anyone compiling the
file to first pre-process it. The second is not desirable since it
introduces a path which is project-specific into a file which is supposed
to be a hardware description. For example Linux and U-Boot are unlikely to
put these include files in the same place.
It is much more convenient to specify the search patch on the command line
as is done with C pre-processors, for example.
Introduce a -i option to add to the list of search paths used to find
source and include files.
We cannot use -I as it is already in use. Other suggestions welcome.
Signed-off-by: Simon Glass <sjg@chromium.org>
Sometimes the requested node or property is not present in the device
tree. This option provides a way of reporting a default value in this
case, rather than halting with an error.
Signed-off-by: Simon Glass <sjg@chromium.org>
Acked-by: David Gibson <david@gibson.dropbear.id.au>
There is a rather unfortunate bug in fdtget in that if multiple argument
sets are provided, it just repeats displaying the first set ones for
each set.
Fix this bug and add a test for it.
Signed-off-by: Simon Glass <sjg@chromium.org>
Acked-by: David Gibson <david@gibson.dropbear.id.au>
Currently run_tests.sh generates several files of text test data. The
procedure it uses for this is somewhat torturous and has several problems:
* Since the test data is derived from a dts file, a cursory glance at the
test output suggests something is wrong with the processing of that dts.
This is misleading since in fact it's just being used as an arbirary
string.
* Since the base input has linefeeds removed, the head and sort commands
used later have no effect.
* Although an attempt is made to get rid of characters which the shell
will mangle, it's not thorough enough. Specifically it leaves in \ which
means that some string escapes found in the input data can get expanded
somewhere along the line in some shells.
This patch, therefore, replaces this generation of test data with a
pre-canned "Lorem ipsum" of approximately 2k. On my system, where /bin/sh
is dash, this fixes a test failure due to the aforementioned string
escapes being evaluated on one but not the other of the two comparison
paths (I haven't tracked down exactly where the expansion is happening).
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Currently the fdt{get,put}-runtest.sh scripts invoke diff to check if
fdt{get,put} did the right thing. This isn't great though: it's not
obvious from the diff output which is the expected and which is the
actual result; diff's line by line behaviour is useless here, since all
the results are a single line and finally, when there is a difference
it always prints information even when the tests are supposed to be
running in quiet mode.
This patch uses cmp instead, and explicitly prints the expected results,
when running in verbose mode (the invocation of fdtget itself will have
already displayed the actual results in this mode.
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
This patch cleans up how the fdtget and fdtput tests are invoked.
Specifically we no longer hide the full command lines with a wrapper
function - this makes it possible to distinguish fdtget from similar
fdtput tests and makes it easier to work out how to manually invoke an
individual failing test.
In addition, we remove the testing for errors from the
fdt{get,put}-runtest.sh script, instead using an internal wrapper
analagous to run_wrap_test which can test for any program invocation
that's expected to return an error.
For a couple of the fdtput tests this would result in printing out
ludicrously large command lines. Therefore we introduce a new
mechanism to cut those down to something reasonable.
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Several test scripts now have some code to check for a program returning
a signal, and reporting a suitable failure. This patch moves this
duplicated code into a helper function in tests.sh. At the same time we
remove a bashism found in the current copies (using the non portablr $[ ]
construct for arithmetic).
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
The patches introducing fdtget and fdtput inserted a peculiar bashism to
run_tests.sh using non-portable assignment within an (( )) expression.
This patch fixes it.
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Some of the test scripts create temporary files, which we remove at the
end. Except that we usually forgot to remove them on some exit paths. To
avoid this problem in future, this modifies the scripts to use the shell's
trap 0 functionality to automatically remove the temporaries on any exit.
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Several of the test scripts remove $TMPFILE, without ever having set
the TMPFILE variable. This patch fixes it.
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
We've add some test (generated) binaries that aren't currently listed in
.gitignore, in addition more scripts now generate various tmp.* files
during operation. This adds them all to .gitignore.
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
This simple utility allows writing of values into a device tree from the
command line. It aimes to be the opposite of fdtget.
What is it for:
- Updating fdt values when a binary blob already exists
(even though source may be available it might be easier to use this
utility rather than sed, etc.)
- Writing machine-specific fdt values within a build system
To use it, specify the fdt binary file on command line followed by the node
and property to set. Then, provide a list of values to put into that
property. Often there will be just one, but fdtput also supports arrays and
string lists.
fdtput does not try to guess the type of the property based on looking at
the arguments. Instead it always assumes that an integer is provided. To
indicate that you want to write a string, use -ts. You can also provide
hex values with -tx.
The command line arguments are joined together into a single value. For
strings, a nul terminator is placed between each string when it is packed
into the property. To avoid this, pass the string as a single argument.
Usage:
fdtput <options> <dt file> <<node> <property> [<value>...]
Options:
-t <type> Type of data
-v Verbose: display each value decoded from command line
-h Print this help
<type> s=string, i=int, u=unsigned, x=hex
Optional modifier prefix:
hh or b=byte, h=2 byte, l=4 byte (default)
To read from stdin and write to stdout, use - as the file. So you can do:
cat somefile.dtb | fdtput -ts - /node prop "My string value" > newfile.dtb
This commit also adds basic tests to verify the major features.
Signed-off-by: Simon Glass <sjg@chromium.org>
This simply utility makes it easy for scripts to read values from the device
tree. It is written in C and uses the same libfdt as the rest of the dtc
package.
What is it for:
- Reading fdt values from scripts
- Extracting fdt information within build systems
- Looking at particular values without having to dump the entire tree
To use it, specify the fdt binary file on command line followed by a list of
node, property pairs. The utility then looks up each node, finds the property
and displays the value.
Each value is printed on a new line.
fdtget tries to guess the type of each property based on its contents. This
is not always reliable, so you can use the -t option to force fdtget to decode
the value as a string, or byte, etc.
To read from stdin, use - as the file.
Usage:
fdtget <options> <dt file> [<node> <property>]...
Options:
-t <type> Type of data
-h Print this help
<type> s=string, i=int, u=unsigned, x=hex
Optional modifier prefix:
hh or b=byte, h=2 byte, l=4 byte (default)
Signed-off-by: Simon Glass <sjg@chromium.org>
This will allow callers to rebuild .dtb files when any of the /include/d
.dtsi files are modified, not just the top-level .dts file.
Signed-off-by: Stephen Warren <swarren@nvidia.com>
Acked-by: David Gibson <david@gibson.dropbear.id.au>
Commit a31e3ef83b introduced new libfdt
functions to append to existing properties. It also included a test case
for this, but neglected to update the Makefile and run_tests.sh script
to actually build and execute this testcase.
This patch corrects the oversight.
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Some properties may contain multiple values, these values may need
to be added to the property respectively. this patch provides this
functionality. The main purpose of fdt_append_prop() is to append
the values to a existing property, or create a new property if it
dose not exist.
Signed-off-by: Minghuan Lian <Minghuan.Lian@freescale.com>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Elements of size 8, 16, 32, and 64 bits are supported. The new
/bits/ syntax was selected so as to not pollute the reserved
keyword space with uint8/uint16/... type names.
With this patch the following property assignment:
property = /bits/ 16 <0x1234 0x5678 0x0 0xffff>;
is equivalent to:
property = <0x12345678 0x0000ffff>;
It is now also possible to directly specify a 64 bit literal in a
cell list, also known as an array using:
property = /bits/ 64 <0xdeadbeef00000000>;
It is an error to attempt to store a literal into an element that is
too small to hold the literal, and the compiler will generate an
error when it detects this. For instance:
property = /bits/ 8 <256>;
Will fail to compile. It is also an error to attempt to place a
reference in a non 32-bit element.
The documentation has been changed to reflect that the cell list
is now an array of elements that can be of sizes other than the
default 32-bit cell size.
The sized_cells test tests the creation and access of 8, 16, 32,
and 64-bit sized elements. It also tests that the creation of two
properties, one with 16 bit elements and one with 32 bit elements
result in the same property contents.
Signed-off-by: Anton Staaf <robotboy@chromium.org>
Acked-by: David Gibson <david@gibson.dropbear.id.au>
The load_blob() and save_blob() functions are very similar to the utilfdt
versions. This removes the duplicated code.
Signed-off-by: Simon Glass <sjg@chromium.org>
Acked-by: David Gibson <david@gibson.dropbear.id.au>
This adds higher-level libfdt operations for reading/writing an fdt
blob from/to a file, as well as a function to decode a data type string
as will be used by fdtget, fdtput.
This also adds a few tests for the simple type argument supported by
utilfdt_decode_type.
Signed-off-by: Simon Glass <sjg@chromium.org>
Acked-by: David Gibson <david@gibson.dropbear.id.au>
With this patch the following property assignment:
property = <0x12345678 'a' '\r' 100>;
is equivalent to:
property = <0x12345678 0x00000061 0x0000000D 0x00000064>
Signed-off-by: Anton Staaf <robotboy@chromium.org>
Acked-by: David Gibson <david@gibson.dropbear.id.au>
A number of the dtc testcases trigger the new "variable set but not
used" warning from gcc 4.6. That is they have variables which are
assigned, but then never read after that point.
In a couple of cases this is just because the variables aren't needed,
so this patch removes them. In subnode_offset.c, it's because one
pair of variables we clearly intended to test we don't actually test.
This patch also adds this missing check.
This patch makes the testsuite compile clean with gcc 4.6.
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
This patch adds a "dtdiff" script to do a useful form diff of two
device trees. This automatically converts the tree to dts form (if
it's not already) and uses a new "-s" option in dtc to "sort" the
tree. That is, it sorts the reserve entries, it sorts the properties
within each node by name, and it sorts nodes by name within their
parent.
This gives a pretty sensible diff between the trees, which will ignore
semantically null internal rearrangements (directly diffing the dts
files can give a lot of noise due to the order changes).
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
When nodes are modified by merging device trees, nodes to be updated/merged can
be specified by a label. Specifying nodes by full path (instead of label)
doesn't quite work. This patch fixes that.
Signed-off-by: John Bonesio <bones@secretlab.ca>
Acked-by: David Gibson <david@gibson.dropbear.id.au>
Current we check for various error codes with [ $x == "NN" ]. However
'==' is not actually a correct operator for the [ (test) command. It
should be either '=' for string comparison or '-eq' for integer
comparison. It appears that the bash builtin version of test
implements '==' though, so we were getting away with it, as long as
/bin/sh was bash - or the testsuite generated no errors.
This patch fixes the usage of test so that it should work on non-bash
shells.
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
The dtbs_equal_ordered test program is used to implement a number of
testcases. However, the test program itself has never been
particularly well tested. In addition there are testcases coming in
future for which it would be useful to have a corresponding
"dtbs_equal_unordered" which checks for equality of device trees, not
considering the internal ordering of elements. Finally, for some
tests we may want it would be useful to check trees for equality with
the PASS case being when they are *not* equal.
This patch addresses all of the above. A dtbs_equal_unordered is
added, and both it and the existing dtbs_equal_ordered program now
take a -n option to make the PASS case be where the trees are not
equal. A number of example trees with slight modifications from
test_tree1 are used to verify that both these programs correctly
identify when the tree is altered, and a dtb_reverse program is used
to verify that the unordered version does not depend on internal
ordering. These new testcases for the equality testing programs are
split out into a new test group in run_tests.sh.
dtbs_equal_unordered uses the new property iteration functions, and so
this also acts as further testing for those functions.
dtbs_equal_unordered will be useful for further testing the recently
added tree-merging code and its upcoming extensions.
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
This patch turns on a bunch of extra gcc warnings, most of which are
probably a good idea. Of the new warnings -Wnested-externs and
-Wstrict-prototypes need no code changes, we're already warning-clean.
The remaining one, -Wmissing-prototypes requires trivial changes in
some of the tests (making functions local).
This patch also rearranges the warnings flags into a separate make
variable for convenience, and turns on -Werror, to really encourage
people to keep the code warning-clean.
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
This patch allows the following construct:
/ {
property-a = "old";
property-b = "does not change";
};
/ {
property-a = "changed";
property-c = "new";
node-a {
};
};
Where the later device tree overrides the properties found in the
earlier tree. This is useful for laying down a template device tree
in an include file and modifying it for a specific board without having
to clone the entire tree.
Signed-off-by: Grant Likely <grant.likely@secretlab.ca>
The somewhat embarrasing bug in the first version of my previous patch
would have been detected by valgrind. Thus reminded, I've run the
testsuite under valgrind and fixed any errors I found. This turned
out to be just some uninitialized buffers in test programs. The
fragments of uninitialized data aren't particularly important, but we
might as well squash the valgrind warnings, so that future valgrind
errors will stand out.
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
At present, both the grammar and our internal data structures mean
that there can be only one label on a node or property. This is a
fairly arbitrary constraint, given that any number of value labels can
appear at the same point, and that in C you can have any number of
labels on the same statement.
This is pretty much a non-issue now, but it may become important with
some of the extensions that Grant and I have in mind. It's not that
hard to change, so this patch does so, allowing an arbitrary number of
labels on any given node or property. As usual a testcase is added
too.
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Acked-by: Grant Likely <grant.likely@secretlab.ca>
The only purpose of the dtc_references_dts0 testcase was to check
handling of references in the old dts v0 syntax. Since we no longer
support the old syntax, and the references_dts0.dts has been converted
to the new format, it's entirely redundant. This patch removes it.
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Currently, nothing will stop you from re-using the same label string
multiple times in a dts, e.g.:
/ {
samelabel: prop1 = "foo";
samelabel: prop2 = "bar";
};
or
/ {
samelabel: prop1 = "foo";
samelabel: subnode {
};
};
When using node references by label, this could lead to confusing
results (with no warning), and in -Oasm mode will result in output
which the assembler will complain about (since it too will have
duplicate labels).
This patch, therefore, adds code to checks.c to give errors if you
attempt to re-use the same label. It treats all labels (node,
property, and value) as residing in the same namespace, since the
assembler will treat them so for -Oasm mode.
Testcases for the new code are also added.
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Currently, when in -Idts -Odtb or -Ifs -Odtb modes, dtc always
defaults to using 0 as the value for the boot_cpuid_phys header field.
That's correct quite often, but there are some systems where there is
no CPU with hardware ID of 0, or where we don't want to use the CPU
with hardware ID 0 at all (e.g. for AMP-style partitioning). The only
way to override this default currently, is with the -b command line
option.
This patch improves dtc to instead base the default boot_cpuid_phys
value on the reg property of the first listed subnode of /cpus. This
means that dtc will get boot_cpuid_phys correct by default in a
greater proportion of cases (since the boot cpu is usually listed
first, and this way at least the boot_cpuid_phys default will match
some existing cpu node). If the node doesn't exist or has an invalid
'reg' property (missing or not 4 bytes in length), then
boot_cpuid_phys is set to 0.
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Currently, the Linux kernel, libfdt and dtc, when using flattened
device trees encode a node's phandle into a property named
"linux,phandle". The ePAPR specification, however - aiming as it is
to not be a Linux specific spec - requires that phandles be encoded in
a property named simply "phandle".
This patch adds support for this newer approach to dtc and libfdt.
Specifically:
- fdt_get_phandle() will now return the correct phandle if it
is supplied in either of these properties
- fdt_node_offset_by_phandle() will correctly find a node with
the given phandle encoded in either property.
- By default, when auto-generating phandles, dtc will encode
it into both properties for maximum compatibility. A new -H
option allows either only old-style or only new-style
properties to be generated.
- If phandle properties are explicitly supplied in the dts
file, dtc will not auto-generate ones in the alternate format.
- If both properties are supplied, dtc will check that they
have the same value.
- Some existing testcases are updated to use a mix of old and
new-style phandles, partially testing the changes.
- A new phandle_format test further tests the libfdt support,
and the -H option.
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
I've just tested building dtc as an x86_64 binary on a 32-bit i386
host by using:
make CC="gcc -m64"
This patch fixes a handful of minor bugs thus discovered:
* There is a printf() type mismatch on 64-bit in value-labels.c
* For the tests which use libdl, we were using the GNU make feature
where it will find libdl.so given a dependency in the form '-ldl'.
But this built-in make logic doesn't know we're compiling 64-bit so
finds the 32-bit version of the library. We avoid using this and
instead explicitly pass -ldl to CC, which being the 64-bit version
does know where to look.
* To process dtc's asm output into .so files, run_tests.sh was
directly invoking the (default instance of) the assembler and linker.
Instead invoke these via the CC driver, and allow that to be overriden
from the make environment.
* The x86_64 assembler doesn't 0 fill with the .balign directive
(presumably it is NOP filling). That doesn't produce strictly
incorrect trees, but it is confusing and confounds are testcases which
do byte-by-byte comparison of the trees produced by asm output with
direct dtb output (which does 0 pad where necessary, of course). This
patch uses the optional second argument to .balign to force gas to
zero-fill instead.
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
In one section, the run_tests script attempts to use the -e (interpret
escapes) option to echo. This option is not portable - for example
the echo built into dash, now the default /bin/sh on several
distributions does not support it and will just echo "-e" literally.
Since we don't actually use any of the escapes that -e enables, this
patch simply removes it.
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
When in -Odts mode, dtc will not produce correct output for
string-like properties which have more than one \0 character at the
end of the property's bytestring. In fact, it generates output which
is not syntactically correct. This patch fixes the bug, and adds a
testcase for future regressions here.
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
This patch adds a testcase using asm output mode to check that labels
within property values are correctly processed.
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Stephen Warren