Delete nonreduced fuzz inputs

[maflcko]

As per the usual process to avoid wasted CI resources and timeouts when CI runs on large and presumed irrelevant inputs.

Normally, deletion of non-reduced fuzz inputs should happen after feature-freeze on the master Bitcoin Core branch, but before branch-off, so that the latest release branch retains mostly valid fuzz inputs.

Previous: #204

To "reproduce"

Install a fresh VM, as explained in the bash script's doc, and run it:

apt update && apt install curl -y
curl -L -O https://raw.githubusercontent.com/bitcoin-core/bitcoin-maintainer-tools/main/delete_nonreduced_fuzz_inputs.sh
bash delete_nonreduced_fuzz_inputs.sh

To "test"

• Keep an eye on coverage stats, to ensure it doesn't drop
• Re-run the script, to ensure it is "reproducible" to some extent
• Anything else you think is important to test or review

CI

CI should pass, except for a lint failure, which should light up on any changes like this pull request, which delete fuzz inputs.

[maflcko]

Storage device usage (du -sh ./fuzz_corpora/)

5.8G -> 2.7G

Determinism

• ~125k fuzz input files deleted

git diff --stat origin/main..2b825cd506de7403c69ecf3e8a6f495b8dc1dcad | tail -1
 125400 files changed

• Cross-diff with a second run of the script: 4k fuzz input files

git -c diff.renameLimit=2007 diff --stat HEAD..2b825cd506de7403c69ecf3e8a6f495b8dc1dcad | tail -1 
 3737 files changed

Coverage

• main: https://drahtbot.space/host_reports/DrahtBot/reports/coverage_fuzz/monotree/dfb7d58108daf372/c62dc520e0c7caee/fuzz.coverage/index.html
• pull: https://drahtbot.space/host_reports/DrahtBot/reports/coverage_fuzz/monotree/dfb7d58108daf372/2b825cd506de7403/fuzz.coverage/index.html

[maflcko]

The coverage run completed, but it looks like the coverage decreases in some lines. Not sure if this happened in the past.

For example, Assert(expect_work) is no longer hit, when comparing https://drahtbot.space/host_reports/DrahtBot/reports/coverage_fuzz/monotree/dfb7d58108daf372/c62dc520e0c7caee/fuzz.coverage/src/test/fuzz/txdownloadman.cpp.gcov.html and https://drahtbot.space/host_reports/DrahtBot/reports/coverage_fuzz/monotree/dfb7d58108daf372/2b825cd506de7403/fuzz.coverage/src/test/fuzz/txdownloadman.cpp.gcov.html

Not sure why this would happen, other than due to bitcoin/bitcoin#29018

[maflcko]

Though, the cross-diff is at a historic low, so the result is one of the most deterministic ones (so far)

[maflcko]

This is ready for review, now that CI passed, except for the expected 'lint' failure.

[marcofleon]

Reviewing this now. About how long is the script expected to take?

[maflcko]

Looking at the commit timestamps, it took around 4 hours, but it could be more or less, depending on your compute, memory and storage.

[marcofleon]

Original: https://marcofleon.github.io/coverage/original_fuzz_corpora/
Reduced: https://marcofleon.github.io/coverage/reduced_fuzz_corpora/

Both coverage reports generated on the most commit in master. Looks generally good to me, there was a little drop in function coverage (e.g. chain.cpp, hash.cpp, rpc/server.cpp) but must just be some instability.

Corpora went from 5.7G to 2.7G.