GoSe

Compile Benchmarks

1. Benchmark archives are provided in the benchmarks/ directory.

   bc	bsdtar	cjpeg	cmark	djpeg	gawk
   grep	img2sixel	jpegoptim	jpegtran	nano	nm
   pdftops	pdftotext	readelf	sed	speexdec	tar
   tcpprep	tcpreplay	tiffcp	tree	xmllint	xmlwf

2. Unpack and compile each benchmark with coverage instrumentation enabled. Store the build outputs in the target directories obj-gcov{1,2,3,...}.

Run GoSe

1. Install required packages

2. Build Instantiation Guidebook

• compile target program (with gcov support)
• prepare target program manual
• (implement your parsing logic in GoSe/utils/parse_manual.py)
• python3 main.py --generate-instantiation-guidebook --prog-path=/path/to/program --option-csv-dir=GoSe/input/instantiation_guidebook

3. Construct Fine-Grained Decision Probability

• python3 main.py --build-graph --strategy GoSe --prog-path=/path/to/program --prog-gcov-dir=/path/to/program_gcov_directory --option-csv-dir=GoSe/input/instantiation_guidebook --save-probability-dir=GoSe/output/probability --coverage-server-port=12321 --parallel-coverage-collection-num=32

4. Generate option sequences

• python3 main.py --test-seeds --strategy GoSe --prog-path=/path/to/program --prog-gcov-dir=/path/to/program_gcov_directory --option-csv-dir=GoSe/input/instantiation_guidebook --save-probability-dir=GoSe/output/probability --output-seeds-dir=GoSe/output/seeds --coverage-server-port=12321