Fix remote stack overflow and uninit read in Content-Type/-Encoding parsing (#506)

treathead() parsed the Content-Type and Content-Encoding response-header
values with a width-less sscanf("%s") into a 1100-byte stack scratch,
bounds-checking the result only after the copy. The receive path in
back_wait hands treathead a line buffer of up to 2000 bytes (htsback.c
rcvd[2048], binput cap 2000), so a hostile server sending a header value
longer than 1099 non-whitespace bytes overflowed the stack. Bound both
scans to the buffer (%1099s), matching the existing %31s/%255s idiom in
the request-line parsers.

The Content-Encoding branch also ignored the sscanf return, so an empty
value (e.g. "Content-Encoding:") left tempo uninitialized and then ran
strlen() over it: an uninitialized read that can also over-read past the
buffer. The Content-Type branch already guarded this; mirror it.

Found via the P3-3 static-analysis baseline. New -#test=headerlong plus
empty-value cases in 01_engine-header.test build the hostile inputs in
the handler (the CLI caps argument length); they abort on the pre-fix
binary under ASan (overflow) and MSan (uninitialized read).

Signed-off-by: Xavier Roche <roche@httrack.com>
Co-authored-by: Claude Fable 5 <noreply@anthropic.com>

.github/workflows/codeql.yml

@@ -0,0 +1,57 @@
+# CodeQL static analysis (C). The security-extended suite covers the classes
+# this codebase actually fights: overflows, tainted-size allocs, format bugs.
+name: CodeQL
+
+on:
+  push:
+    branches: [master]
+  pull_request:
+  schedule:
+    # Weekly re-scan of master so new/updated queries land without a push.
+    - cron: "17 4 * * 1"
+
+permissions:
+  contents: read
+
+concurrency:
+  group: codeql-${{ github.ref }}
+  cancel-in-progress: true
+
+jobs:
+  analyze:
+    name: analyze (c-cpp)
+    runs-on: ubuntu-24.04
+    permissions:
+      contents: read
+      # Upload findings to the repo's code-scanning dashboard.
+      security-events: write
+    steps:
+      - uses: actions/checkout@v6
+        with:
+          submodules: recursive
+
+      - name: Install build dependencies
+        run: |
+          set -euo pipefail
+          sudo apt-get update
+          sudo apt-get install -y --no-install-recommends \
+            build-essential autoconf automake libtool autoconf-archive \
+            zlib1g-dev libssl-dev
+
+      - uses: github/codeql-action/init@v4
+        with:
+          languages: c-cpp
+          build-mode: manual
+          queries: security-extended
+
+      # Manual build: CodeQL traces the compiler, so build exactly what ships.
+      - name: Build
+        run: |
+          set -euo pipefail
+          autoreconf -fi
+          ./configure
+          make -j"$(nproc)"
+
+      - uses: github/codeql-action/analyze@v4
+        with:
+          category: "/language:c-cpp"

configure.ac

@@ -102,7 +102,8 @@ AX_CHECK_COMPILE_FLAG([-Wignored-qualifiers], [DEFAULT_CFLAGS="$DEFAULT_CFLAGS -
 AX_CHECK_COMPILE_FLAG([-Werror=attribute-warning], [DEFAULT_CFLAGS="$DEFAULT_CFLAGS -Werror=attribute-warning"])
 AX_CHECK_COMPILE_FLAG([-Werror=user-defined-warnings], [DEFAULT_CFLAGS="$DEFAULT_CFLAGS -Werror=user-defined-warnings"])
 AX_CHECK_COMPILE_FLAG([-fstrict-aliasing -Wstrict-aliasing], [DEFAULT_CFLAGS="$DEFAULT_CFLAGS -fstrict-aliasing -Wstrict-aliasing"])
-AX_CHECK_COMPILE_FLAG([-fstack-protector], [DEFAULT_CFLAGS="$DEFAULT_CFLAGS -fstack-protector"])
+AX_CHECK_COMPILE_FLAG([-fstack-protector-strong], [DEFAULT_CFLAGS="$DEFAULT_CFLAGS -fstack-protector-strong"],
+	[AX_CHECK_COMPILE_FLAG([-fstack-protector], [DEFAULT_CFLAGS="$DEFAULT_CFLAGS -fstack-protector"])])
 AX_CHECK_COMPILE_FLAG([-fstack-clash-protection], [DEFAULT_CFLAGS="$DEFAULT_CFLAGS -fstack-clash-protection"])
 AX_CHECK_COMPILE_FLAG([-fcf-protection], [DEFAULT_CFLAGS="$DEFAULT_CFLAGS -fcf-protection"])
 AX_CHECK_LINK_FLAG([-Wl,--discard-all], [DEFAULT_LDFLAGS="$DEFAULT_LDFLAGS -Wl,--discard-all"])
@@ -110,6 +111,13 @@ AX_CHECK_LINK_FLAG([-Wl,--no-undefined], [DEFAULT_LDFLAGS="$DEFAULT_LDFLAGS -Wl,
 AX_CHECK_LINK_FLAG([-Wl,-z,relro,-z,now], [DEFAULT_LDFLAGS="$DEFAULT_LDFLAGS -Wl,-z,relro,-z,now"])
 AX_CHECK_LINK_FLAG([-Wl,-z,noexecstack], [DEFAULT_LDFLAGS="$DEFAULT_LDFLAGS -Wl,-z,noexecstack"])
 
+# Fortify libc calls (=3, else =2) unless the toolchain predefines it; skip
+# sanitizer builds, whose interceptors want the unfortified calls.
+case "$CFLAGS" in
+*-fsanitize=*) ;;
+*) AX_ADD_FORTIFY_SOURCE ;;
+esac
+
 # Force libc back into DT_NEEDED for libraries that reach it only through
 # libhttrack (libhtsjava, the libtest callbacks), but only with a GNU-style
 # linker; Apple ld rejects these flags and links libSystem unconditionally.

m4/ax_add_fortify_source.m4

@@ -0,0 +1,119 @@
+# ===========================================================================
+#  https://www.gnu.org/software/autoconf-archive/ax_add_fortify_source.html
+# ===========================================================================
+#
+# SYNOPSIS
+#
+#   AX_ADD_FORTIFY_SOURCE
+#
+# DESCRIPTION
+#
+#   Check whether -D_FORTIFY_SOURCE=2 can be added to CPPFLAGS without macro
+#   redefinition warnings, other cpp warnings or linker. Some distributions
+#   (such as Ubuntu or Gentoo Linux) enable _FORTIFY_SOURCE globally in
+#   their compilers, leading to unnecessary warnings in the form of
+#
+#     <command-line>:0:0: error: "_FORTIFY_SOURCE" redefined [-Werror]
+#     <built-in>: note: this is the location of the previous definition
+#
+#   which is a problem if -Werror is enabled. This macro checks whether
+#   _FORTIFY_SOURCE is already defined, and if not, adds -D_FORTIFY_SOURCE=2
+#   to CPPFLAGS.
+#
+#   Newer mingw-w64 msys2 package comes with a bug in
+#   headers-git-7.0.0.5546.d200317d-1. It broke -D_FORTIFY_SOURCE support,
+#   and would need -lssp or -fstack-protector.  See
+#   https://github.com/msys2/MINGW-packages/issues/5803. Try to actually
+#   link it.
+#
+# LICENSE
+#
+#   Copyright (c) 2017 David Seifert <soap@gentoo.org>
+#   Copyright (c) 2019, 2023 Reini Urban <rurban@cpan.org>
+#
+#   Copying and distribution of this file, with or without modification, are
+#   permitted in any medium without royalty provided the copyright notice
+#   and this notice are preserved.  This file is offered as-is, without any
+#   warranty.
+
+#serial 10
+
+AC_DEFUN([AX_ADD_FORTIFY_SOURCE],[
+    ac_save_cflags=$CFLAGS
+    ac_cwerror_flag=yes
+    AX_CHECK_COMPILE_FLAG([-Werror],[CFLAGS="$CFLAGS -Werror"])
+    ax_add_fortify_3_failed=
+    AC_MSG_CHECKING([whether to add -D_FORTIFY_SOURCE=3 to CPPFLAGS])
+    AC_LINK_IFELSE([
+        AC_LANG_PROGRAM([],
+            [[
+                #ifndef _FORTIFY_SOURCE
+                    return 0;
+                #else
+                    _FORTIFY_SOURCE_already_defined;
+                #endif
+            ]]
+        )],
+        AC_LINK_IFELSE([
+            AC_LANG_SOURCE([[
+                #define _FORTIFY_SOURCE 3
+                #include <string.h>
+                int main(void) {
+                    char *s = " ";
+                    strcpy(s, "x");
+                    return strlen(s)-1;
+                }
+              ]]
+            )],
+            [
+              AC_MSG_RESULT([yes])
+              CFLAGS=$ac_save_cflags
+              CPPFLAGS="$CPPFLAGS -D_FORTIFY_SOURCE=3"
+            ], [
+              AC_MSG_RESULT([no])
+              ax_add_fortify_3_failed=1
+            ],
+        ),
+        [
+          AC_MSG_RESULT([no])
+          ax_add_fortify_3_failed=1
+        ])
+    if test -n "$ax_add_fortify_3_failed"
+    then
+    AC_MSG_CHECKING([whether to add -D_FORTIFY_SOURCE=2 to CPPFLAGS])
+    AC_LINK_IFELSE([
+        AC_LANG_PROGRAM([],
+            [[
+                #ifndef _FORTIFY_SOURCE
+                    return 0;
+                #else
+                    _FORTIFY_SOURCE_already_defined;
+                #endif
+            ]]
+        )],
+        AC_LINK_IFELSE([
+            AC_LANG_SOURCE([[
+                #define _FORTIFY_SOURCE 2
+                #include <string.h>
+                int main(void) {
+                    char *s = " ";
+                    strcpy(s, "x");
+                    return strlen(s)-1;
+                }
+              ]]
+            )],
+            [
+              AC_MSG_RESULT([yes])
+              CFLAGS=$ac_save_cflags
+              CPPFLAGS="$CPPFLAGS -D_FORTIFY_SOURCE=2"
+            ], [
+              AC_MSG_RESULT([no])
+              CFLAGS=$ac_save_cflags
+            ],
+        ),
+        [
+          AC_MSG_RESULT([no])
+          CFLAGS=$ac_save_cflags
+        ])
+    fi
+])

src/htslib.c

@@ -1562,7 +1562,7 @@ void treathead(t_cookie * cookie, const char *adr, const char *fil, htsblk * ret
       }
       // An empty/whitespace Content-Type value yields no token: keep the
       // sentinel default rather than reading an uninitialized tempo.
-      if (sscanf(rcvd + p, "%s", tempo) == 1) {
+      if (sscanf(rcvd + p, "%1099s", tempo) == 1) { // tempo[1100], server data
         if (strlen(tempo) < sizeof(retour->contenttype) - 2) // pas trop long!!
           strcpybuff(retour->contenttype, tempo);
         else
@@ -1657,11 +1657,11 @@ void treathead(t_cookie * cookie, const char *adr, const char *fil, htsblk * ret
         if (a)
           *a = '\0';
       }
-      sscanf(a, "%s", tempo);
-      if (strlen(tempo) < 64)   // pas trop long!!
+      // Bound to tempo[1100]; no token => leave empty, don't read uninit tempo.
+      if (sscanf(a, "%1099s", tempo) == 1 && strlen(tempo) < 64)
         strcpybuff(retour->contentencoding, tempo);
       else
-        retour->contentencoding[0] = '\0';      // erreur
+        retour->contentencoding[0] = '\0';
 #if HTS_USEZLIB
       /* Check known encodings */
       if (retour->contentencoding[0]) {

src/htsselftest.c

@@ -1163,6 +1163,26 @@ static int st_header(httrackp *opt, int argc, char **argv) {
     treathead(NULL, "www.example.com", "/", &r, line);
   }
   printf("contenttype=%s cdispo=%s\n", r.contenttype, r.cdispo);
+  printf("contentencoding=%s\n", r.contentencoding);
+  return 0;
+}
+
+/* An over-long header value must not overflow treathead's tempo[1100]. */
+static int st_headerlong(httrackp *opt, int argc, char **argv) {
+  htsblk r;
+  char BIGSTK line[HTS_URLMAXSIZE * 2];
+  const char *const name = argc >= 1 ? argv[0] : "Content-Type:";
+  const int pad = 1500; /* > tempo[1100] */
+  size_t n;
+
+  (void) opt;
+  memset(&r, 0, sizeof(r));
+  n = (size_t) snprintf(line, sizeof(line), "%s ", name);
+  memset(line + n, 'a', pad);
+  line[n + pad] = '\0';
+  treathead(NULL, "www.example.com", "/", &r, line);
+  printf("contenttype_len=%d contentencoding_len=%d\n",
+         (int) strlen(r.contenttype), (int) strlen(r.contentencoding));
   return 0;
 }
 
@@ -2185,6 +2205,9 @@ static const struct selftest_entry {
     {"identurl", "<url>", "split an absolute URL into (adr, fil)", st_identurl},
     {"header", "<raw-header-line> ...", "response header-line parsing",
      st_header},
+    {"headerlong", "[header-name:]",
+     "over-long header value must not overflow the parse scratch",
+     st_headerlong},
     {"savename", "<fil> <content-type> [key=value ...]",
      "local save-name for a URL", st_savename},
     {"sniff", "<content-type> <hex:..|text>", "MIME magic consistency",

tests/01_engine-header.test

@@ -27,3 +27,16 @@ hdr 'contenttype= cdispo=report.pdf' \
 # Path components in the filename are dropped on the wire (RFC 2616).
 hdr 'contenttype= cdispo=evil.pdf' \
     'Content-Disposition: attachment; filename="../../evil.pdf"'
+
+# An over-long header value must not overflow the parse scratch (ASan aborts
+# here on the pre-fix binary).
+test "$(httrack -O /dev/null -#test=headerlong 'Content-Type:')" \
+    == 'contenttype_len=32 contentencoding_len=0'
+test "$(httrack -O /dev/null -#test=headerlong 'Content-Encoding:')" \
+    == 'contenttype_len=0 contentencoding_len=0'
+
+# An empty Content-Encoding yields no token: leave it empty, don't read the
+# uninitialized scratch (MSan aborts here on the pre-fix binary).
+enc() { httrack -O /dev/null -#test=header "$1" | grep '^contentencoding='; }
+test "$(enc 'Content-Encoding:')" == 'contentencoding='
+test "$(enc 'Content-Encoding: gzip')" == 'contentencoding=gzip'