dtb: Skip decoding strings if not nul terminated

If a property is defined to be a string, make sure it is nul terminated
before decoding it. This fixes a case where <1> int was decoded into
['', '', '', '']. This still will not fix the case where there's the last
byte is nul. There's only so much we can do for garbage in.

Signed-off-by: Rob Herring <robh@kernel.org>

dtschema/dtb.py

@@ -137,6 +137,9 @@ def prop_value(validator, nodename, p):
             return data
 
     if fmt.startswith('string'):
+        if not data.endswith(b'\0'):
+            # Wrong data type?, skip decoding to force errors
+            return data
         return data[:-1].decode(encoding='ascii').split('\0')
 
     if {'flag'} & prop_types: