save progress

devops/docker/repro-builders/alpine/scripts/extract-functions.sh

@@ -0,0 +1,71 @@
+#!/bin/sh
+# Extract function fingerprints from ELF binaries
+# Outputs JSON array with function name, offset, size, and hashes
+#
+# Usage: extract-functions.sh <directory>
+#
+# Dependencies: objdump, readelf, sha256sum, jq
+
+set -eu
+
+DIR="${1:-.}"
+
+extract_functions_from_binary() {
+    local binary="$1"
+    
+    # Skip non-ELF files
+    file "$binary" | grep -q "ELF" || return 0
+    
+    # Get function symbols
+    objdump -t "$binary" 2>/dev/null | \
+    awk '/\.text.*[0-9a-f]+.*F/ {
+        # Fields: addr flags section size name
+        gsub(/\*.*\*/, "", $1)  # Clean address
+        if ($5 != "" && $4 != "00000000" && $4 != "0000000000000000") {
+            printf "%s %s %s\n", $1, $4, $NF
+        }
+    }' | while read -r offset size name; do
+        # Skip compiler-generated symbols
+        case "$name" in
+            __*|_GLOBAL_*|.plt*|.text*|frame_dummy|register_tm_clones|deregister_tm_clones)
+                continue
+                ;;
+        esac
+        
+        # Convert hex size to decimal
+        dec_size=$((16#$size))
+        
+        # Skip tiny functions (likely padding)
+        [ "$dec_size" -lt 16 ] && continue
+        
+        # Extract function bytes and compute hash
+        # Using objdump to get disassembly and hash the opcodes
+        local hash=$(objdump -d --start-address="0x$offset" --stop-address="0x$((16#$offset + dec_size))" "$binary" 2>/dev/null | \
+            grep "^[[:space:]]*[0-9a-f]*:" | \
+            awk '{for(i=2;i<=NF;i++){if($i~/^[0-9a-f]{2}$/){printf "%s", $i}}}' | \
+            sha256sum | cut -d' ' -f1)
+        
+        # Output JSON object
+        printf '{"name":"%s","offset":"0x%s","size":%d,"hash":"%s"}\n' \
+            "$name" "$offset" "$dec_size" "${hash:-unknown}"
+    done
+}
+
+# Find all ELF binaries in directory
+echo "["
+first=true
+find "$DIR" -type f -executable 2>/dev/null | while read -r binary; do
+    # Check if ELF
+    file "$binary" 2>/dev/null | grep -q "ELF" || continue
+    
+    extract_functions_from_binary "$binary" | while read -r json; do
+        [ -z "$json" ] && continue
+        if [ "$first" = "true" ]; then
+            first=false
+        else
+            echo ","
+        fi
+        echo "$json"
+    done
+done
+echo "]"