merge script to summarize the System Calls and File Descriptors for e…

…ach benchmark (#51)

* sys_summary

* integrate sys summary into table generation pipline

---------

Co-authored-by: Zhuoxuan Zhang <hzhang6@oberlin.edu>

infrastructure/colossal_table.py

@@ -12,6 +12,12 @@
 data_path = root / 'infrastructure/target/dynamic_analysis.jsonl'
 input_size_path = root / 'infrastructure/data/size_inputs.jsonl'
 loc_data_path = root / 'infrastructure/target/lines_of_code.csv'
+csv_file_path = root / 'infrastructure/data/sys_summary.csv'
+
+def read_sys_results():
+    csv_data = pd.read_csv(csv_file_path)
+    csv_data.rename(columns={'Benchmark': 'benchmark', 'Sys Calls': 'sys_calls', 'File Descriptors': 'file_descriptors'}, inplace=True)
+    return csv_data
 
 benchmark_category_style = {
     'bio': ('XXX', 'XXX', 'XXX'),
@@ -145,6 +151,7 @@ def main():
     syntax_script_all_cmds, syntax_bench_all_cmds = stx.read_data(False)
     dyn_script,    dyn_bench = dyn.read_data()
     loc_data_script, loc_data_bench = read_loc_data()
+    sys_results = read_sys_results()
 
     syntax_script_all_cmds['unique_cmds'] = syntax_script_all_cmds['nodes'].apply(count_unique_cmds)
     syntax_bench_all_cmds['unique_cmds'] = syntax_bench_all_cmds['nodes'].apply(count_unique_cmds)
@@ -165,7 +172,8 @@ def main():
 
     big_bench = syntax_bench.merge(dyn_bench, on='benchmark')\
         .merge(loc_data_bench, on='benchmark')\
-        .merge(syntax_bench_all_cmds[['benchmark', 'unique_cmds']], on='benchmark')
+        .merge(syntax_bench_all_cmds[['benchmark', 'unique_cmds']], on='benchmark')\
+        .merge(sys_results, on='benchmark')
 
     big_script = syntax_script.merge(dyn_script, on='script')\
         .merge(loc_data_script, on='script')\

infrastructure/data/sys_summary.csv

@@ -0,0 +1,14 @@
+Benchmark,Sys Calls,File Descriptors
+aurpkg,207,0
+covid-mts,252,0
+file-enc,246,13
+log-analysis,225,13
+makeself,1237,13
+max-temp,202,0
+media-conv,225,13
+nlp,2054,15
+oneliners,677,13
+riker,254,13
+unix50,1556,0
+vps-audit,207,13
+vps-audit-negate,207,13

infrastructure/sys_summary.sh

@@ -0,0 +1,75 @@
+#!/bin/bash
+
+output_file="benchmark_results.csv"
+
+benchmarks=(
+  "aurpkg"
+  "bio"
+  "covid-mts"
+  "file-enc"
+  "log-analysis"
+  "makeself"
+  "max-temp"
+  "media-conv"
+  "nlp"
+  "oneliners"
+  "riker"
+  "sklearn"
+  "unix50"
+  "vps-audit"
+  "vps-audit-negate"
+  "web-index"
+)
+
+# Error handler
+error() {
+    echo "Error: $1" >&2
+    exit 1
+}
+
+echo "Benchmark,Sys Calls,File Descriptors" > "$output_file"
+
+for benchmark in "${benchmarks[@]}"; do
+    echo "Running benchmark: $benchmark"
+
+    strace_output="/tmp/${benchmark}_strace.txt"
+    lsof_output="/tmp/${benchmark}_lsof.txt"
+
+    if ! cd "./$benchmark"; then
+        echo "$benchmark [fail]: Directory not found" >> "$output_file"
+        continue
+    fi
+
+    strace -c -o "$strace_output" ./run.sh --small || {
+        echo "$benchmark [fail]: strace failed" >> "$output_file"
+        cd - > /dev/null
+        continue
+    }
+
+    ./run.sh --small &
+    pid=$!
+
+    sleep 1
+
+    if [[ -d /proc/$pid ]]; then
+        lsof -p "$pid" > "$lsof_output"
+    else
+        echo "Warning: Process $pid ended before lsof could capture file descriptors."
+        > "$lsof_output"
+    fi
+
+    wait "$pid"
+
+    total_syscalls=$(awk '/^100.00/ {print $4}' "$strace_output")
+    total_syscalls=${total_syscalls:-0}
+
+    fd_count=$(wc -l < "$lsof_output")
+    fd_count=${fd_count:-0}
+
+    echo "$benchmark,$total_syscalls,$fd_count" >> "../$output_file"
+
+    rm -f "$strace_output" "$lsof_output"
+    cd - > /dev/null
+done
+
+echo "Benchmark results saved to $output_file."