From 83dcc8f1441a2a71179ee6ab7a45afdfa1574ff8 Mon Sep 17 00:00:00 2001
From: Mingye Wang <arthur200126@gmail.com>
Date: Sun, 5 Mar 2023 14:29:42 +0800
Subject: [PATCH] FS_comparison: do sorted comparison to get 10x speed up

* New linear time O(n+m) comparison to replace O(nm) thing
  Gives 10x speed up on /bin/bash
* Modify chk_fortify to sastify the new precondition
---
 checksec                     | 77 ++++++++++++++++++------------------
 src/functions/chk_fortify.sh |  8 ++--
 src/functions/fs_funcs.sh    | 67 ++++++++++++++++---------------
 3 files changed, 75 insertions(+), 77 deletions(-)

diff --git a/checksec b/checksec
index b97e3e8..55a22fe 100755
--- a/checksec
+++ b/checksec
@@ -392,8 +392,8 @@ chk_fortify_file() {
 
   FS_chk_func_libc=()
   FS_functions=()
-  while IFS='' read -r line; do FS_chk_func_libc+=("$line"); done < <(${readelf} -s "${FS_libc}" 2> /dev/null | grep _chk@@ | awk '{ print $8 }' | cut -c 3- | sed -e 's/_chk@.*//')
-  while IFS='' read -r line; do FS_functions+=("$line"); done < <(${readelf} -s "${CHK_FORTIFY_FILE}" 2> /dev/null | awk '{ print $8 }' | sed 's/_*//' | sed -e 's/@.*//')
+  while IFS='' read -r line; do FS_chk_func_libc+=("$line"); done < <(${readelf} -s "${FS_libc}" 2> /dev/null | grep _chk@@ | awk '{ print $8 }' | cut -c 3- | sed -e 's/_chk@.*//' | sort -u)
+  while IFS='' read -r line; do FS_functions+=("$line"); done < <(${readelf} -s "${CHK_FORTIFY_FILE}" 2> /dev/null | awk '{ print $8 }' | sed 's/_*//' | sed -e 's/@.*//' | sort -u)
   echo_message "" "" "<fortify-test name='${CHK_FORTIFY_FILE}' " "{ \"fortify-test\": { \"name\":\"${CHK_FORTIFY_FILE}\" "
   FS_libc_check
   FS_binary_check
@@ -433,8 +433,8 @@ chk_fortify_proc() {
 
     FS_chk_func_libc=()
     FS_functions=()
-    while IFS='' read -r line; do FS_chk_func_libc+=("$line"); done < <(${readelf} -s "${FS_libc}" 2> /dev/null | grep _chk@@ | awk '{ print $8 }' | cut -c 3- | sed -e 's/_chk@.*//')
-    while IFS='' read -r line; do FS_functions+=("$line"); done < <(${readelf} -s "${CHK_FORTIFY_PROC}/exe" 2> /dev/null | awk '{ print $8 }' | sed 's/_*//' | sed -e 's/@.*//')
+    while IFS='' read -r line; do FS_chk_func_libc+=("$line"); done < <(${readelf} -s "${FS_libc}" 2> /dev/null | grep _chk@@ | awk '{ print $8 }' | cut -c 3- | sed -e 's/_chk@.*//' | sort -u)
+    while IFS='' read -r line; do FS_functions+=("$line"); done < <(${readelf} -s "${CHK_FORTIFY_PROC}/exe" 2> /dev/null | awk '{ print $8 }' | sed 's/_*//' | sed -e 's/@.*//' | sort -u)
     echo_message "" "" "<fortify-test name='${name}' pid='${N}' " "{ \"fortify-test\": { \"name\":\"${name}\", \"pid\":\"${N}\" "
     FS_libc_check
     FS_binary_check
@@ -882,42 +882,41 @@ FS_comparison() {
   echo_message " Fortifiable library functions | Checked function names\n" "" "" ""
   echo_message " -------------------------------------------------------\n" "" "" ""
 
-  for ((FS_elem_libc = 0; FS_elem_libc < ${#FS_chk_func_libc[@]}; FS_elem_libc++)); do
-    for ((FS_elem_functions = 0; FS_elem_functions < ${#FS_functions[@]}; FS_elem_functions++)); do
-      FS_tmp_func=${FS_functions[$FS_elem_functions]}
-      FS_tmp_libc=${FS_chk_func_libc[$FS_elem_libc]}
-
-      if [[ ${FS_tmp_func} =~ ^${FS_tmp_libc}$ ]]; then
-        if [[ ${format} == "cli" ]]; then
-          printf " \033[31m%-30s\033[m | __%s%s\n" "${FS_tmp_func}" "${FS_tmp_libc}" "${FS_end}"
-        else
-          if [[ $FS_elem_functions == 0 ]]; then
-            echo_message "" "${FS_tmp_func},${FS_tmp_libc},yes\n" "    <function name='${FS_tmp_func}' libc='${FS_tmp_libc}' fortifiable='yes' />\n" ", \"function\": { \"name\":\"${FS_tmp_func}\", \"libc\":\"${FS_tmp_libc}\", \"fortifiable\":\"yes\" }"
-          elif [[ $FS_elem_functions == $((${#FS_functions[@]} - 1)) ]]; then
-            echo_message "" "${FS_tmp_func},${FS_tmp_libc},yes\n" "    <function name='${FS_tmp_func}' libc='${FS_tmp_libc}' fortifiable='yes' />\n" ", \"function\": { \"name\":\"${FS_tmp_func}\", \"libc\":\"${FS_tmp_libc}\", \"fortifiable\":\"yes\" }"
-          else
-            echo_message "" "${FS_tmp_func},${FS_tmp_libc},yes\n" "    <function name='${FS_tmp_func}' libc='${FS_tmp_libc}' fortifiable='yes' />\n" ", \"function\": { \"name\":\"${FS_tmp_func}\", \"libc\":\"${FS_tmp_libc}\", \"fortifiable\":\"yes\" }"
-          fi
-        fi
-        ((FS_cnt_total++))
-        ((FS_cnt_unchecked++))
-      elif [[ ${FS_tmp_func} =~ ^${FS_tmp_libc}(_chk)$ ]]; then
-        if [[ ${format} == "cli" ]]; then
-          printf " \033[32m%-30s\033[m | __%s%s\n" "${FS_tmp_func}" "${FS_tmp_libc}" "${FS_end}"
-        else
-          if [[ $FS_elem_functions == 0 ]]; then
-            echo_message "" "${FS_tmp_func},${FS_tmp_libc},no\n" "    <function name='${FS_tmp_func}' libc='${FS_tmp_libc}' fortifiable='no' />\n" ", \"function\": { \"name\":\"${FS_tmp_func}\", \"libc\":\"${FS_tmp_libc}\", \"fortifiable\":\"no\" }"
-          elif [[ $FS_elem_functions == $((${#FS_functions[@]} - 1)) ]]; then
-            echo_message "" "${FS_tmp_func},${FS_tmp_libc},no\n" "    <function name='${FS_tmp_func}' libc='${FS_tmp_libc}' fortifiable='no' />\n" ", \"function\": { \"name\":\"${FS_tmp_func}\", \"libc\":\"${FS_tmp_libc}\", \"fortifiable\":\"no\" }"
-          else
-            echo_message "" "${FS_tmp_func},${FS_tmp_libc},no\n" "    <function name='${FS_tmp_func}' libc='${FS_tmp_libc}' fortifiable='no' />\n" ", \"function\": { \"name\":\"${FS_tmp_func}\", \"libc\":\"${FS_tmp_libc}\", \"fortifiable\":\"no\" }"
-          fi
-        fi
-        ((FS_cnt_total++))
-        ((FS_cnt_checked++))
+  # PRECONDITION. FS_chk_func_libc[] and FS_functions[] are both sorted to current locale and unique (sort -u).
+  for ((FS_elem_libc = 0, FS_elem_functions = 0;  \
+  FS_elem_libc < ${#FS_chk_func_libc[@]} && \
+  FS_elem_functions < ${#FS_functions[@]}; )); do
+    FS_tmp_func=${FS_functions[$FS_elem_functions]}
+    FS_tmp_libc=${FS_chk_func_libc[$FS_elem_libc]}
+
+    if [[ ${FS_tmp_func} < ${FS_tmp_libc} ]]; then
+      ((FS_elem_functions++))
+    elif [[ ${FS_tmp_func} == "${FS_tmp_libc}" ]]; then
+      if [[ ${format} == "cli" ]]; then
+        printf " \033[31m%-30s\033[m | __%s%s\n" "${FS_tmp_func}" "${FS_tmp_libc}" "${FS_end}"
+      else
+        echo_message "" "${FS_tmp_func},${FS_tmp_libc},yes\n" "    <function name='${FS_tmp_func}' libc='${FS_tmp_libc}' fortifiable='yes' />\n" ", \"function\": { \"name\":\"${FS_tmp_func}\", \"libc\":\"${FS_tmp_libc}\", \"fortifiable\":\"yes\" }"
       fi
-
-    done
+      ((FS_cnt_total++))
+      ((FS_cnt_unchecked++))
+      ((FS_elem_functions++))
+      # HERE LIES DRAGONS. If you advance the libc pointer now, you will miss the _chk!
+    elif [[ ${FS_tmp_func} < ${FS_tmp_libc}_chk ]]; then
+      ((FS_elem_functions++))
+    elif [[ ${FS_tmp_func} == "${FS_tmp_libc}_chk" ]]; then
+      if [[ ${format} == "cli" ]]; then
+        printf " \033[32m%-30s\033[m | __%s%s\n" "${FS_tmp_func}" "${FS_tmp_libc}" "${FS_end}"
+      else
+        echo_message "" "${FS_tmp_func},${FS_tmp_libc},no\n" "    <function name='${FS_tmp_func}' libc='${FS_tmp_libc}' fortifiable='no' />\n" ", \"function\": { \"name\":\"${FS_tmp_func}\", \"libc\":\"${FS_tmp_libc}\", \"fortifiable\":\"no\" }"
+      fi
+      ((FS_cnt_total++))
+      ((FS_cnt_checked++))
+      ((FS_elem_functions++))
+      ((FS_elem_libc++))
+    else
+      #  [[ ${FS_tmp_func} > ${FS_tmp_libc}_chk ]]
+      ((FS_elem_libc++))
+    fi
   done
 }
 
diff --git a/src/functions/chk_fortify.sh b/src/functions/chk_fortify.sh
index 91ce2c6..8a48f90 100644
--- a/src/functions/chk_fortify.sh
+++ b/src/functions/chk_fortify.sh
@@ -35,8 +35,8 @@ chk_fortify_file() {
 
   FS_chk_func_libc=()
   FS_functions=()
-  while IFS='' read -r line; do FS_chk_func_libc+=("$line"); done < <(${readelf} -s "${FS_libc}" 2> /dev/null | grep _chk@@ | awk '{ print $8 }' | cut -c 3- | sed -e 's/_chk@.*//')
-  while IFS='' read -r line; do FS_functions+=("$line"); done < <(${readelf} -s "${CHK_FORTIFY_FILE}" 2> /dev/null | awk '{ print $8 }' | sed 's/_*//' | sed -e 's/@.*//')
+  while IFS='' read -r line; do FS_chk_func_libc+=("$line"); done < <(${readelf} -s "${FS_libc}" 2> /dev/null | grep _chk@@ | awk '{ print $8 }' | cut -c 3- | sed -e 's/_chk@.*//' | sort -u)
+  while IFS='' read -r line; do FS_functions+=("$line"); done < <(${readelf} -s "${CHK_FORTIFY_FILE}" 2> /dev/null | awk '{ print $8 }' | sed 's/_*//' | sed -e 's/@.*//' | sort -u)
   echo_message "" "" "<fortify-test name='${CHK_FORTIFY_FILE}' " "{ \"fortify-test\": { \"name\":\"${CHK_FORTIFY_FILE}\" "
   FS_libc_check
   FS_binary_check
@@ -76,8 +76,8 @@ chk_fortify_proc() {
 
     FS_chk_func_libc=()
     FS_functions=()
-    while IFS='' read -r line; do FS_chk_func_libc+=("$line"); done < <(${readelf} -s "${FS_libc}" 2> /dev/null | grep _chk@@ | awk '{ print $8 }' | cut -c 3- | sed -e 's/_chk@.*//')
-    while IFS='' read -r line; do FS_functions+=("$line"); done < <(${readelf} -s "${CHK_FORTIFY_PROC}/exe" 2> /dev/null | awk '{ print $8 }' | sed 's/_*//' | sed -e 's/@.*//')
+    while IFS='' read -r line; do FS_chk_func_libc+=("$line"); done < <(${readelf} -s "${FS_libc}" 2> /dev/null | grep _chk@@ | awk '{ print $8 }' | cut -c 3- | sed -e 's/_chk@.*//' | sort -u)
+    while IFS='' read -r line; do FS_functions+=("$line"); done < <(${readelf} -s "${CHK_FORTIFY_PROC}/exe" 2> /dev/null | awk '{ print $8 }' | sed 's/_*//' | sed -e 's/@.*//' | sort -u)
     echo_message "" "" "<fortify-test name='${name}' pid='${N}' " "{ \"fortify-test\": { \"name\":\"${name}\", \"pid\":\"${N}\" "
     FS_libc_check
     FS_binary_check
diff --git a/src/functions/fs_funcs.sh b/src/functions/fs_funcs.sh
index 96cf660..3212a59 100644
--- a/src/functions/fs_funcs.sh
+++ b/src/functions/fs_funcs.sh
@@ -34,42 +34,41 @@ FS_comparison() {
   echo_message " Fortifiable library functions | Checked function names\n" "" "" ""
   echo_message " -------------------------------------------------------\n" "" "" ""
 
-  for ((FS_elem_libc = 0; FS_elem_libc < ${#FS_chk_func_libc[@]}; FS_elem_libc++)); do
-    for ((FS_elem_functions = 0; FS_elem_functions < ${#FS_functions[@]}; FS_elem_functions++)); do
-      FS_tmp_func=${FS_functions[$FS_elem_functions]}
-      FS_tmp_libc=${FS_chk_func_libc[$FS_elem_libc]}
+  # PRECONDITION. FS_chk_func_libc[] and FS_functions[] are both sorted to current locale and unique (sort -u).
+  for ((FS_elem_libc = 0, FS_elem_functions = 0;  \
+  FS_elem_libc < ${#FS_chk_func_libc[@]} && \
+  FS_elem_functions < ${#FS_functions[@]}; )); do
+    FS_tmp_func=${FS_functions[$FS_elem_functions]}
+    FS_tmp_libc=${FS_chk_func_libc[$FS_elem_libc]}
 
-      if [[ ${FS_tmp_func} =~ ^${FS_tmp_libc}$ ]]; then
-        if [[ ${format} == "cli" ]]; then
-          printf " \033[31m%-30s\033[m | __%s%s\n" "${FS_tmp_func}" "${FS_tmp_libc}" "${FS_end}"
-        else
-          if [[ $FS_elem_functions == 0 ]]; then
-            echo_message "" "${FS_tmp_func},${FS_tmp_libc},yes\n" "    <function name='${FS_tmp_func}' libc='${FS_tmp_libc}' fortifiable='yes' />\n" ", \"function\": { \"name\":\"${FS_tmp_func}\", \"libc\":\"${FS_tmp_libc}\", \"fortifiable\":\"yes\" }"
-          elif [[ $FS_elem_functions == $((${#FS_functions[@]} - 1)) ]]; then
-            echo_message "" "${FS_tmp_func},${FS_tmp_libc},yes\n" "    <function name='${FS_tmp_func}' libc='${FS_tmp_libc}' fortifiable='yes' />\n" ", \"function\": { \"name\":\"${FS_tmp_func}\", \"libc\":\"${FS_tmp_libc}\", \"fortifiable\":\"yes\" }"
-          else
-            echo_message "" "${FS_tmp_func},${FS_tmp_libc},yes\n" "    <function name='${FS_tmp_func}' libc='${FS_tmp_libc}' fortifiable='yes' />\n" ", \"function\": { \"name\":\"${FS_tmp_func}\", \"libc\":\"${FS_tmp_libc}\", \"fortifiable\":\"yes\" }"
-          fi
-        fi
-        ((FS_cnt_total++))
-        ((FS_cnt_unchecked++))
-      elif [[ ${FS_tmp_func} =~ ^${FS_tmp_libc}(_chk)$ ]]; then
-        if [[ ${format} == "cli" ]]; then
-          printf " \033[32m%-30s\033[m | __%s%s\n" "${FS_tmp_func}" "${FS_tmp_libc}" "${FS_end}"
-        else
-          if [[ $FS_elem_functions == 0 ]]; then
-            echo_message "" "${FS_tmp_func},${FS_tmp_libc},no\n" "    <function name='${FS_tmp_func}' libc='${FS_tmp_libc}' fortifiable='no' />\n" ", \"function\": { \"name\":\"${FS_tmp_func}\", \"libc\":\"${FS_tmp_libc}\", \"fortifiable\":\"no\" }"
-          elif [[ $FS_elem_functions == $((${#FS_functions[@]} - 1)) ]]; then
-            echo_message "" "${FS_tmp_func},${FS_tmp_libc},no\n" "    <function name='${FS_tmp_func}' libc='${FS_tmp_libc}' fortifiable='no' />\n" ", \"function\": { \"name\":\"${FS_tmp_func}\", \"libc\":\"${FS_tmp_libc}\", \"fortifiable\":\"no\" }"
-          else
-            echo_message "" "${FS_tmp_func},${FS_tmp_libc},no\n" "    <function name='${FS_tmp_func}' libc='${FS_tmp_libc}' fortifiable='no' />\n" ", \"function\": { \"name\":\"${FS_tmp_func}\", \"libc\":\"${FS_tmp_libc}\", \"fortifiable\":\"no\" }"
-          fi
-        fi
-        ((FS_cnt_total++))
-        ((FS_cnt_checked++))
+    if [[ ${FS_tmp_func} < ${FS_tmp_libc} ]]; then
+      ((FS_elem_functions++))
+    elif [[ ${FS_tmp_func} == "${FS_tmp_libc}" ]]; then
+      if [[ ${format} == "cli" ]]; then
+        printf " \033[31m%-30s\033[m | __%s%s\n" "${FS_tmp_func}" "${FS_tmp_libc}" "${FS_end}"
+      else
+        echo_message "" "${FS_tmp_func},${FS_tmp_libc},yes\n" "    <function name='${FS_tmp_func}' libc='${FS_tmp_libc}' fortifiable='yes' />\n" ", \"function\": { \"name\":\"${FS_tmp_func}\", \"libc\":\"${FS_tmp_libc}\", \"fortifiable\":\"yes\" }"
       fi
-
-    done
+      ((FS_cnt_total++))
+      ((FS_cnt_unchecked++))
+      ((FS_elem_functions++))
+      # HERE LIES DRAGONS. If you advance the libc pointer now, you will miss the _chk!
+    elif [[ ${FS_tmp_func} < ${FS_tmp_libc}_chk ]]; then
+      ((FS_elem_functions++))
+    elif [[ ${FS_tmp_func} == "${FS_tmp_libc}_chk" ]]; then
+      if [[ ${format} == "cli" ]]; then
+        printf " \033[32m%-30s\033[m | __%s%s\n" "${FS_tmp_func}" "${FS_tmp_libc}" "${FS_end}"
+      else
+        echo_message "" "${FS_tmp_func},${FS_tmp_libc},no\n" "    <function name='${FS_tmp_func}' libc='${FS_tmp_libc}' fortifiable='no' />\n" ", \"function\": { \"name\":\"${FS_tmp_func}\", \"libc\":\"${FS_tmp_libc}\", \"fortifiable\":\"no\" }"
+      fi
+      ((FS_cnt_total++))
+      ((FS_cnt_checked++))
+      ((FS_elem_functions++))
+      ((FS_elem_libc++))
+    else
+      #  [[ ${FS_tmp_func} > ${FS_tmp_libc}_chk ]]
+      ((FS_elem_libc++))
+    fi
   done
 }