kernel/drivers/of/selftest.c, branch linux-rolling-stable

of/unittest: Rename selftest.c to unittest.c

2014-11-04T16:43:07Z

This is unit testing code. It should use that name because it makes more sense than 'selftest'. Rename the files to match and rename the config variable. Signed-off-by: Grant Likely

of/unittest: Remove test devices after adding them

2014-11-04T16:43:04Z

The of_platform_populate() test cases don't remove the test devices after they are added. Fix this by adding tests for of_platform_depopulate(). At the same time rework the selftest() macro to return the test result value. This makes it easy to use the macro inside an if() condition. Signed-off-by: Grant Likely

of: Eliminate of_allnodes list

2014-11-04T13:29:38Z

The device tree structure is composed of two lists; the 'allnodes' list which is a singly linked list containing every node in the tree, and the child->parent structure where each parent node has a singly linked list of children. All of the data in the allnodes list can be easily reproduced with the parent-child lists, so of_allnodes is actually unnecessary. Remove it entirely which saves a bit of memory and simplifies the data structure quite a lot. Signed-off-by: Grant Likely Cc: Rob Herring Cc: Gaurav Minocha Cc: Pantelis Antoniou

of: Fix overflow bug in string property parsing functions

2014-11-04T10:19:48Z

The string property read helpers will run off the end of the buffer if it is handed a malformed string property. Rework the parsers to make sure that doesn't happen. At the same time add new test cases to make sure the functions behave themselves. The original implementations of of_property_read_string_index() and of_property_count_strings() both open-coded the same block of parsing code, each with it's own subtly different bugs. The fix here merges functions into a single helper and makes the original functions static inline wrappers around the helper. One non-bugfix aspect of this patch is the addition of a new wrapper, of_property_read_string_array(). The new wrapper is needed by the device_properties feature that Rafael is working on and planning to merge for v3.19. The implementation is identical both with and without the new static inline wrapper, so it just got left in to reduce the churn on the header file. Signed-off-by: Grant Likely Cc: Rafael J. Wysocki Cc: Mika Westerberg Cc: Rob Herring Cc: Arnd Bergmann Cc: Darren Hart Cc: # v3.3+: Drop selftest hunks that don't apply

of/selftest: Move hash table off stack to fix large frame size

2014-10-07T10:30:31Z

The new testcase that checks phandle consistency was using a hash table on the stack which made the frame size much large than it should be. Fix the problem by moving the hash table into the file scope. Signed-off-by: Grant Likely

of/selftest: Use the resolver to fixup phandles

2014-10-04T20:24:35Z

The selftest data ends up causing duplicate phandles in the live tree for the time that the testcase data is inserted into the live tree. This is obviously a bad situation because anything attempting to read the tree while the selftests are running make resolve phandles to one of the testcase data nodes. Fix the problem by using the of_resolve_phandles() function to eliminate duplicates. Signed-off-by: Grant Likely Cc: Pantelis Antoniou Cc: Gaurav Minocha

of/selftest: Add a test for duplicate phandles

2014-10-04T20:20:19Z

All phandles in the tree should be unique. Add a testcase to make sure that this is so. Note: this testcase fails on the current kernel because the selftest code itself ends up adding duplicate phandles. Before this testcase is merged the selftest code needs to be modified to resolve phandles before adding them. Signed-off-by: Grant Likely Cc: Pantelis Antoniou

of/selftest: Test structure of device tree

2014-10-04T20:20:13Z

Add a testcase to verify that the device tree is properly constructed and the lists are in a correct order. The new testcase gets run twice; once after adding the testcase data, and once after removing it again. It is run twice to make sure adding and removing the testcase data doesn't corrupt the data structure. Signed-off-by: Grant Likely Cc: Gaurav Minocha

of: Fix NULL dereference in selftest removal code

2014-10-01T16:01:43Z

The selftest code removes its testcase data from the live tree when exiting, but if the testcases data tree contains an empty child of the root, then it causes an oops due to a NULL dereference. The reason is that the code tries to directly dereference the child pointer without checking first if a child is actually there. The solution is to pass the parent node into detach_node_and_children() instead of trying to pass the child. This required removing the code that attempts to remove all of the sibling nodes in detach_node_and_children(), which was never sensible in the first place. At the same time add a check to make sure the bounds of the nodes list are not exceeded by the testdata tree. If they are then abort. Signed-off-by: Grant Likely Cc: Gaurav Minocha

Enabling OF selftest to run without machine's devicetree

2014-08-16T08:03:56Z

If there is no devicetree present, this patch adds the selftest data as a live devicetree. It also removes the same after the testcase execution is complete. Tested with and without machine's devicetree. Signed-off-by: Gaurav Minocha Signed-off-by: Grant Likely