The following sections might be updated with supplementary metadata relevant to reviewers and maintainers.

Conflicts

Reviewers, this pull request conflicts with the following ones:

• fuzz: Hide script_assets_test_minimizer #20663 (fuzz: Hide script_assets_test_minimizer by MarcoFalke)

If you consider this pull request important, please also help to review the conflicting pull requests. Ideally, start with the one that should be merged first.

Seeing 5% speedup here in the fuzz framework part (non-fuzz-target part)

Nice! Can you share the benchmarks and/or details on how to reproduce? :)

Concept ACK , can be rebased and maybe opened for review now that #20464 is merged.

To reproduce:

diff --git a/src/test/fuzz/string.cpp b/src/test/fuzz/string.cpp
index 282a2cd8ca..7a1c7b00d6 100644
--- a/src/test/fuzz/string.cpp
+++ b/src/test/fuzz/string.cpp
@@ -33,6 +33,16 @@
 #include <string>
 #include <vector>
 
+FUZZ_TARGET(max_throughput)
+{
+    FuzzedDataProvider fuzzed_data_provider(buffer.data(), buffer.size());
+    assert(
+        // "Teach" libFuzzer to explore long sequences (needs -use_value_profile=1 -len_control=1 -max_len=2000000)
+        fuzzed_data_provider.remaining_bytes() < 500'000 ||
+        // Make it crash when done
+        fuzzed_data_provider.ConsumeBool());
+}
+
 FUZZ_TARGET(string)
 {
     FuzzedDataProvider fuzzed_data_provider(buffer.data(), buffer.size());

Then run a bunch and plot all (lim, exec/s).

FUZZ=max_throughput ./src/test/fuzz/fuzz -use_value_profile=1 -len_control=1 -max_len=2000000

My result (N=10, Firefox was running in the background, so take with a span of salt):

cr ACK faf7d74: patch looks correct :)

That chart is a work of art 😄

Code review ACK faf7d74