Skip to content
/ tonic Public

A Programmable Hardware Architecture for Network Transport Logic

License

Notifications You must be signed in to change notification settings

minmit/tonic

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

12 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Tonic

Tonic is a programmable hardware architecture for the transport logic, i.e., reliable data delivery and congestion control. All the design files in this repository are available for use under the 3-clause BSD License.

Code Structure

  • src/ - source files
    • tonic.v (Tonic's top-level)
    • dd_engine/ (Data Delivery Engine)
    • cr_engine/ (Credit Engine)
    • user_defined_logic/
    • Other components used by the engines (e.g., RAM, bitmap operations, etc.)
  • include/ system-wide constants
  • tb/ - testbench files
    • config/ (.yaml files for configuring simulations and test parameters)
    • sim/ (simulation files for Tonic, including a mock receiver in C)
  • tools/ (scripts for running simulations and tests)
  • build/ - temporary build files. The output of the build and simulation goes in this directory.
  • csim/ - Tonic's cycle-accurate simulator in C++.

Note: Some directories are empty as they are being ported into the new infrastructre.


Setting up the Environment

Before running any scripts, make sure to setup the environment variables by running

source settings.sh

Running Simulation and Other Tests

All the simulations and tests are executed through the tools/runtest script.

./tools/runtest --type [sim|unit|system] \
                --config path_of_config_file_from_config_root.yaml \
                [--debug]

Notes:

  • The system and unit options are not included yet and therefore cannot be used for now.
  • The sim option simulates the module that is designated in the config file using the initial parameters specified in the same file. Currently, we have simulation files for tonic, for which you can specify a protocol and initial parameters. See below for more information on the format of the configuration file.
  • Note that path_of_the_config_file_from_config_root.yaml is not the full path to the config file. It is the path to the file starting from tb/config.
  • The debug option is used to send a flag to simulation files to print debugging information. Please see tb/sim/dd_engine/sim_top.v for an example.

The Configuration File

Configuration files are in YAML format. Here is the basic structure of a configuration file.

module_name:
  params:
    module_specific_param1: val1
    module_specific_param2: val2
    ...
  specs:
    sim_cycles: val

    # sim_specific_params
  sim_specific_param1: val1
  sim_specific_param2: val2
  ...

Notes

  • module_name is, in most cases, the name of the top-level module for the test. Exception are top-level modules such as tonic and dd_engine that include one or more of the programmable modules (i.e., user_defined_incoming and user_defined_timeout). In those cases, module_name should be top_level_module/protocol_name. For instance, to test tonic with reno, the module_name should be tonic/reno.
  • All the relevant simulations and test files for module_name (except for the source files) should be in tb/test_type/module_name, where test_type is either of sim, unit, and system.
  • params is for specifying module-specific parameters. For tonic, these are init_wnd_size and init_rtx_timer_amnt. As another example, if one were to write a unit test for the fifo module used in Tonic, these paramters could include its depth and width.
  • specs are parameters used for customizing the simulation:
    • sim_cycles specifies how many cycles to run the test for and should be specified in all configuration files.
    • For end-to-end simulations for tonic, the following options are required (See tb/config/ for examples). active_flow_cnt, receiver, loss_prob (represented in X in thousands), and rtt are only used for end-to-end simulations for tonic.
    active_flow_cnt: val (total number of flows to simulate)
    receiver: True/False (whether the simulations requires a mock receiver)
    loss_prob: X (the resulting loss probability will be X/1000)
    rtt: val (in nanoseconds)
  cr_type: val (type of the credit engine, 0 -> cwnd, 1 -> rate, 2 -> tokens)
    ...
  • For top-level modules such as tonic and dd_engine that include one or more of the programmable modules, the following simulation variables should be specified as well:
  user_context_w: val (number of bits used for protocol-specific state)
  init_user_context: val (in hex, the initial value for protocol-specific state)
  send_dd_context: 0 or 1 (whether to send protocol-specific state to cr engine. 0 for all cwnd-based protocols)

Other Notes

Credit Engines

The current version of the code does not include the implementation of the rate-based and grant-based credit engine. Thus, as of now, only protocols that use congestion window as means of handling credit can be simulated.

Adding your Own Protocol

The current version of the code includes four example protocols in src/user_defined_logic/, all some variant of TCP: Reno, two different implementations of New Reno one based on RFC 6582, and an implementation of selective acknowledgment based on RFC 6675. To add your own protocol proto, create user_defined_logic/proto folder and implement user_defined_incomig.v and user_defined_timeout.v, following the input/output interface of the examples. Then create tb/sim/tonic/proto and follow the example in tb/sim/tonic to add protocol specific receiver behavior and simulation files.

About

A Programmable Hardware Architecture for Network Transport Logic

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published