diff options
author | MarcoFalke <falke.marco@gmail.com> | 2020-05-22 17:43:18 -0400 |
---|---|---|
committer | MarcoFalke <falke.marco@gmail.com> | 2020-05-22 17:43:20 -0400 |
commit | fa3288cda12537b2e99a3cc7e14dfff3e944884f (patch) | |
tree | fdcd20c66f4014286b52bf860e919b37124bf404 | |
parent | b5c423c48e094bd098e11c3d1f57acae7502a4da (diff) | |
parent | 6a239e72eb403d3ac5c7db203c5b3a24ef49a48b (diff) | |
download | bitcoin-fa3288cda12537b2e99a3cc7e14dfff3e944884f.tar.xz |
Merge #19052: tests: Don't limit fuzzing inputs to 1 MB for afl-fuzz (now: ∞ ∀ fuzzers)
6a239e72eb403d3ac5c7db203c5b3a24ef49a48b tests: Don't limit fuzzing inputs to 1 MB for afl-fuzz (now: ∞ ∀ fuzzers) (practicalswift)
Pull request description:
Don't limit fuzzing inputs to 1 MB for `afl-fuzz`.
This change provides a level playing field for all fuzzers which allows for fair benchmarking using projects such as the excellent [FuzzBench](https://github.com/google/fuzzbench) project.
Prior to this commit we limited `afl-fuzz` to ≤1 MB inputs but allowed unlimited length inputs for all other fuzzers.
ACKs for top commit:
MarcoFalke:
ACK 6a239e72eb403d3ac5c7db203c5b3a24ef49a48b The maximum data size should be a runtime option, not a compile time hardcoded value.
Tree-SHA512: dad176ae39aa09fe919e057008ab0670b9da72909bfeb8f0e8b9ae93b65514f2e25a1d51be89a32be9122fc412edf49234dfd9a44beb974b25fda387fd7bf174
-rw-r--r-- | src/test/fuzz/fuzz.cpp | 2 |
1 files changed, 0 insertions, 2 deletions
diff --git a/src/test/fuzz/fuzz.cpp b/src/test/fuzz/fuzz.cpp index 6e2188fe86..82e1d55c0b 100644 --- a/src/test/fuzz/fuzz.cpp +++ b/src/test/fuzz/fuzz.cpp @@ -19,8 +19,6 @@ static bool read_stdin(std::vector<uint8_t>& data) ssize_t length = 0; while ((length = read(STDIN_FILENO, buffer, 1024)) > 0) { data.insert(data.end(), buffer, buffer + length); - - if (data.size() > (1 << 20)) return false; } return length == 0; } |