Intoduction to dizzy

Structure of .dizz files

A single packet is described by a so called .dizz file. Some example files can be found in the dizzfiles repository. These files are Python code, so they have to be written in proper Python syntax. They consist of three variables which need to be defined. The first variable is the name of the packet:

name = "test"

The second variable is called objects and describes the packet fields. It's a Python list with all the fields listed:

objects = [
    ...
]

Inside of that list you can use some pre-defined functions which generate the actual data during parsing. the functions are called field(), list(), rand() and link(). they take different arguments, as listed below:

• The field() function takes 4 arguments, which are: the name of the field [a string], the length of the field (in bits!) [an int] OR None for a field with variable length, the default value for that field [a string] and the fuzzing mode for that field [can be none for not fuzzing that field at all, std for fuzzing some values on the upper and lower value border, and full for fuzzing all possible values].

  objects = [
      field("len", 8, "\x00", none),
      ...
  ]

• The list() function takes 3 arguments: the name of the field [a string], the default value of the field [a string] and the path to a file, containing possible values for that field (one value per line, all values will be inserted while fuzzing).

  objects = [
      field("len", 8, "\x00", none),
      list("test4", "happens?", "lib/test.txt"),
      ...
  ]

• The rand() function takes 2 arguments: the name of the field [a string] and the length of the field (in bits!) [an int]. The value of that field will be a new random on, each time a packet is generated.

  objects = [
      field("len", 8, "\x00", none),
      list("test4", "happens?", "lib/test.txt"),
      rand("random", 12),
      ...
  ]

• The link() function takes 2 arguments: the name of the field [a string] and the name of an other (previous defined) field. The value of that field will always be the same as the source field, also the length will always be the same.

  objects = [
      field("len", 8, "\x00", none),
      list("test4", "happens?", "lib/test.txt"),
      rand("random", 12),
      link("same_random", "random"),
      ...
  ]

The third variable is called functions and it is also a Python list. It represents a set of functions that are called against the generated raw packet, before it is sent out. Currently the functions length(), lambda_legth(), csum(), lambda_csum() and lambda2_csum() are available.

• The length() function takes 3 argument: the name of the destination field, where the value should be updated with the calculated length [a string], the name of the first field, that should be in the length calculation (the starting point) [a string] and the name of the last field, that should be in the length calculation (the end point).

  functions = [
      length("len", "test4", "same_random"),
      ...
  ]

• The lambda_length() function takes 4 arguments: the name of the destination field, where the value should be updated with the calculated length [a string], the name of the first field, that should be in the length calculation (the starting point) [a string], the name of the last field, that should be in the length calculation (the end point) and a function, which will be called after the length is calculated, with the length as an argument [int].

  functions = [
      length("len", "align-mod", "value"),
      lambda_length("len2", "align-nomod", "align-mod", lambda x: x + 2),
      ...
  ]

• The csum() function takes 4 arguments: the name of the destination field, which value should be updated with the calculated checksum [a string], the name of the first field, that should be the input of the checksum calculation (the starting point) [a string], the name of the last field, that should be the input of the checksum calculation (the end point) [a string] and the name of the checksum [a string], were currently only 'inet' (rfc1071) is implemented.

  functions = [  
      length("len", "align-mod", "value"),
      lambda_length("len2", "align-nomod", "align-mod", lambda x: x + 2),
      csum("csum", "align-mod", "value", "inet"),
      ...
  ]

Note: There are some weird looking .dizz files, which are auto-generated from an old dizzy version. They are working and will be replaced by more readable ones in the future.

Structure of .act files

Once you want to get stateful, you need to write interaction in .act files. These file are Python code as well. they also got 3 variables, name for the name of the interaction [a string], objects which is a Python list of dizzes (you can use a pre-defined function here as well) and functions, which also is a Python list.

• The dizz() function takes 2 arguments: the name of the paket [a string] and the path of the .dizz file [a string]. These are the single packets of the interaction.

  objects = [
      dizz("test_dizz", "dizzes/test.dizz"),
      dizz("2nd step", "dizzes/example.dizz"),
      ...
  ]

There is a functions variable as well, which contains either copy() or adv_copy() functions:

• The copy() function takes 4 arguments: the step in which the function should be executed (1=on recv after the first packet [.dizz file], 2=on recv after the second, ...) [an int], the name of the destination field in the second dizz [a string] and two offsets, the start and the end point of the data that should be copied [ints]. These offsets are byte offsets inside of the received data (depending on the used session the received data starts at the ethernet dst [L2] or the tcp/udp/sctp payload [L4]).

  functions = [
      copy(1, "TEID", 4, 8),
      ...
  ]

• The adv_copy() function takes 2 arguments: the step in which the function should be executed [int] and a function reference. The function given will be called with the received data and the dizz of the next step (this should not be used without deep knowledge of the dizzy code ;)