You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
 
 
 
 
 
 
Sam Kleinman 8df7b6103f
proxy: collapse triforcated abci.Client (#8067)
3 years ago
..
mempool proxy: collapse triforcated abci.Client (#8067) 3 years ago
p2p/secretconnection Completed the existing `FinalizeBlock` PR and rebased to master (#7798) 3 years ago
rpc/jsonrpc/server testing: reduce usage of the MustDefaultLogger constructor (#7960) 3 years ago
Makefile fuzz: remove fuzz cases for deleted code (#7187) 3 years ago
README.md test/fuzz: move fuzz tests into this repo (#5918) 4 years ago
oss-fuzz-build.sh test/fuzz: fix wrong path for some p2p fuzzing packages (#6580) 3 years ago

README.md

fuzz

Fuzzing for various packages in Tendermint using go-fuzz library.

Inputs:

  • mempool CheckTx (using kvstore in-process ABCI app)
  • p2p Addrbook#AddAddress
  • p2p pex.Reactor#Receive
  • p2p SecretConnection#Read and SecretConnection#Write
  • rpc jsonrpc server

Directory structure

| test
|  |- corpus/
|  |- crashers/
|  |- init-corpus/
|  |- suppressions/
|  |- testdata/
|  |- <testname>.go

/corpus directory contains corpus data. The idea is to help the fuzzier to understand what bytes sequences are semantically valid (e.g. if we're testing PNG decoder, then we would put black-white PNG into corpus directory; with blockchain reactor - we would put blockchain messages into corpus).

/init-corpus (if present) contains a script for generating corpus data.

/testdata directory may contain an additional data (like addrbook.json).

Upon running the fuzzier, /crashers and /suppressions dirs will be created, along with .zip archive. /crashers will show any inputs, which have lead to panics (plus a trace). /suppressions will show any suppressed inputs.

Running

make fuzz-mempool
make fuzz-p2p-addrbook
make fuzz-p2p-pex
make fuzz-p2p-sc
make fuzz-rpc-server

Each command will create corpus data (if needed), generate a fuzz archive and call go-fuzz executable.

Then watch out for the respective outputs in the fuzzer output to announce new crashers which can be found in the directory crashers.

For example if we find

ls crashers/
61bde465f47c93254d64d643c3b2480e0a54666e
61bde465f47c93254d64d643c3b2480e0a54666e.output
61bde465f47c93254d64d643c3b2480e0a54666e.quoted
da39a3ee5e6b4b0d3255bfef95601890afd80709
da39a3ee5e6b4b0d3255bfef95601890afd80709.output
da39a3ee5e6b4b0d3255bfef95601890afd80709.quoted

the crashing bytes generated by the fuzzer will be in 61bde465f47c93254d64d643c3b2480e0a54666e the respective crash report in 61bde465f47c93254d64d643c3b2480e0a54666e.output

and the bug report can be created by retrieving the bytes in 61bde465f47c93254d64d643c3b2480e0a54666e and feeding those back into the Fuzz function.