Hi David,
On 19 February 2017 at 20:42, David Gibson <david@xxxxxxxxxxxxxxxxxxxxx> wrote:
> On Thu, Feb 16, 2017 at 09:48:28PM -0700, Simon Glass wrote:
>> Hi David,
>>
>> On 14 February 2017 at 22:44, David Gibson <david@xxxxxxxxxxxxxxxxxxxxx> wrote:
>> > On Tue, Feb 14, 2017 at 08:51:57PM -0700, Simon Glass wrote:
>> >> Add a set of tests to cover the functionality in pylibfdt.
>> >>
>> >> Signed-off-by: Simon Glass <sjg@xxxxxxxxxxxx>
>> >> ---
>> >>
>> >> Changes in v5:
>> >> - Adjust tests to match new swig bindings
>> >>
>> >> Changes in v4:
>> >> - Drop tests that are no-longer applicable
>> >> - Add a get for getprop()
>> >>
>> >> Changes in v3:
>> >> - Add some more tests
>> >>
>> >> Changes in v2:
>> >> - Update tests for new pylibfdt
>> >> - Add a few more tests to increase coverage
>> >>
>> >> tests/pylibfdt_tests.py | 267 ++++++++++++++++++++++++++++++++++++++++++++++++
>> >> tests/run_tests.sh | 19 +++-
>> >> 2 files changed, 285 insertions(+), 1 deletion(-)
>> >> create mode 100644 tests/pylibfdt_tests.py
>> >>
>>
>> [..]
>>
>> >> + def testPathOffset(self):
>> >> + """Check that we can find the offset of a node"""
>> >> + self.assertEquals(self.fdt.path_offset('/'), 0)
>> >> + self.assertEquals(self.fdt.path_offset('/subnode@1'), 124)
>> >
>> > This test is potentially fragile. Eventually it would be nice to be
>> > able to run the Python tests expecting test_tree1 on any of the copies
>> > of test_tree1 we generate. Those are required to be semantically
>> > identicaly (including node/property order) to test_tree1.dtb.
>> > However, some versions won't preserve exact offsets - for example
>> > there's a sequence of tests where we insert additional nops in the
>> > encoding to test handling of that. That's why tests/path_offset.c,
>> > for example, checks the behaviour of path_offset() against
>> > subnode_offset() and knowing what property and node names are supposed
>> > to be present, rather than against explicit known offsets.I'm
>>
>> Yes it is fragile, will check it's >0 which should be safe.
>
> Ok.
>
>> Re the tests, I feel we should try to avoid testing all the same
>> things as the C code, when we could just test the interface. But it
>> might be easier just to duplicate the tests you as say.
>
> I think so. The set of tree1 tests is a good model, because it's
> already a reasonably thorough test model of the basic libfdt
> interfaces.
>
>> >
>> >> + with self.assertRaisesRegexp(FdtException, get_err(libfdt.NOTFOUND)):
>> >> + self.fdt.path_offset('/wibble')
>> >> + self.assertEquals(self.fdt.path_offset('/wibble', QUIET_NOTFOUND),
>> >> + -libfdt.NOTFOUND)
>> >> +
>> >> + def testPropertyOffset(self):
>> >> + """Walk through all the properties in the root node"""
>> >> + self.assertEquals(self.fdt.first_property_offset(0), ROOT_PROPS[0])
>> >> + for pos in range(len(ROOT_PROPS) - 1):
>> >> + self.assertEquals(self.fdt.next_property_offset(ROOT_PROPS[pos]),
>> >> + ROOT_PROPS[pos + 1])
>> >> + self.assertEquals(self.fdt.next_property_offset(ROOT_PROPS[-1],
>> >> + QUIET_NOTFOUND),
>> >> + -libfdt.NOTFOUND)
>> >> +
>> >> + def testPropertyOffsetExceptions(self):
>> >> + """Check that exceptions are raised as expected"""
>> >> + with self.assertRaisesRegexp(FdtException, get_err(libfdt.NOTFOUND)):
>> >> + self.fdt.next_property_offset(108)
>> >
>> > Same issue here.
>>
>> OK, I can just drop this one.
>>
>> >
>> >> + with self.assertRaisesRegexp(FdtException, get_err(libfdt.BADOFFSET)):
>> >> + self.fdt.next_property_offset(107)
>> >> + with self.assertRaisesRegexp(FdtException, get_err(libfdt.BADOFFSET)):
>> >> + self.fdt.first_property_offset(107, QUIET_NOTFOUND)
>> >> + with self.assertRaisesRegexp(FdtException, get_err(libfdt.BADOFFSET)):
>> >> + self.fdt.next_property_offset(107, QUIET_NOTFOUND)
>> >> +
>> >> + node = self.fdt.path_offset('/subnode@1/ss1')
>> >> + self.assertEquals(self.fdt.first_property_offset(node, QUIET_NOTFOUND),
>> >> + -libfdt.NOTFOUND)
>> >> + with self.assertRaisesRegexp(FdtException, get_err(libfdt.NOTFOUND)):
>> >> + self.fdt.first_property_offset(node)
>> >> +
>> >> + def testGetName(self):
>> >> + """Check that we can get the name of a node"""
>> >> + self.assertEquals(self.fdt.get_name(0), '')
>> >> + node = self.fdt.path_offset('/subnode@1/subsubnode')
>> >> + self.assertEquals(self.fdt.get_name(node), 'subsubnode')
>> >> +
>> >> + with self.assertRaisesRegexp(FdtException, get_err(libfdt.BADOFFSET)):
>> >> + self.fdt.get_name(-2)
>> >> +
>> >> + def testGetPropertyByOffset(self):
>> >> + """Check that we can read the name and contents of a property"""
>> >> + root = self.fdt.path_offset('/')
>> >
>> > No point to this - offset of / is always 0. If you want to test that
>> > happens, make it a separate testcase.
>>
>> I already have it above so will drop this.
>>
>> >
>> >> + poffset = self.fdt.first_property_offset(root)
>> >> + prop = self.fdt.get_property_by_offset(poffset)
>> >> + self.assertEquals(prop.name, 'compatible')
>> >> + self.assertEquals(prop.value, 'test_tree1\0')
>> >> +
>> >> + with self.assertRaisesRegexp(FdtException, get_err(libfdt.BADOFFSET)):
>> >> + self.fdt.get_property_by_offset(-2)
>> >> + self.assertEquals(
>> >> + -libfdt.BADOFFSET,
>> >> + self.fdt.get_property_by_offset(-2, [libfdt.BADOFFSET]))
>> >> +
>> >> + def testGetProp(self):
>> >> + """Check that we can read the contents of a property by name"""
>> >> + root = self.fdt.path_offset('/')
>> >> + value = self.fdt.getprop(root, "compatible")
>> >> + self.assertEquals(value, 'test_tree1\0')
>> >> + self.assertEquals(-libfdt.NOTFOUND, self.fdt.getprop(root, 'missing',
>> >> + QUIET_NOTFOUND))
>> >
>> > For testing, it might be useful to add a special value you can set the
>> > quiet parameter to to make all errors quiet.
>>
>> Isn't that what I did?
>
> Sorry, I wasn't very clear. What I mean is that if you want to treat
> *all* errors as quiet, at the moment you have to do
> self.whatever(..., quiet=[NOTFOUND, EXISTS, NOSPACE,...])
>
> I was suggesting an extension to check_err() so you can instead say
> 'quiet=SOMETHING' where SOMETHING is a special value, and it will
> interpret that as making all errors quiet.
OK I see. Could SOMETHING be
QUIET_ALL = [NOTFOUND, EXISTS, all other errors]
or are you wanting to take the dynamic typing further and use an
integer or something?
Regards,
Simon
--
To unsubscribe from this list: send the line "unsubscribe devicetree-compiler" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
