X-Git-Url: https://ruderich.org/simon/gitweb/?a=blobdiff_plain;f=bin%2Fblhc;h=e602106b5fd8f6d0140634cf517f21e67471a6f8;hb=7e0037c0788371d6911d2dae5e919bd1b963aedb;hp=bf5f8ff58dc1edddd4208c703ee8c702d808266a;hpb=fa2e7db77f6c7c9ae52841d5b4cfc8d9a1f221a9;p=blhc%2Fblhc.git

diff --git a/bin/blhc b/bin/blhc
index bf5f8ff..e602106 100755
--- a/bin/blhc
+++ b/bin/blhc
@@ -23,6 +23,7 @@ use warnings;
 
 use Getopt::Long ();
 use Term::ANSIColor ();
+use Text::ParseWords ();
 
 our $VERSION = '0.01';
 
@@ -36,6 +37,81 @@ my $cc_regex = qr/(?:[a-z0-9_]+-(?:linux|kfreebsd)-gnu(?:eabi|eabihf)?-)?
 # Regex to catch (GCC) compiler warnings.
 my $warning_regex = qr/^(.+?):([0-9]+):[0-9]+: warning: (.+?) \[(.+?)\]$/;
 
+# Regex for source files which require preprocessing.
+my $source_preprocess_compile_regex = qr/
+    # C
+      c
+    # Objective-C
+    | m
+    # C++
+    | cc | cp | cxx | cpp | CPP | c\+\+ | C
+    # Objective-C++
+    | mm | M
+    # Fortran
+    | F | FOR | fpp | FPP | FTN | F90 | F95 | F03 | F08
+    /x;
+my $source_preprocess_no_compile_regex = qr/
+    # Assembly
+      s
+    /x;
+my $source_preprocess_regex = qr/
+      $source_preprocess_compile_regex
+    | $source_preprocess_no_compile_regex
+    /x;
+# Regex for source files which don't require preprocessing.
+my $source_no_preprocess_compile_regex = qr/
+    # C
+      i
+    # C++
+    | ii
+    # Objective-C
+    | mi
+    # Objective-C++
+    | mii
+    # Fortran
+    | f | for | ftn | f90 | f95 | f03 | f08
+    /x;
+my $source_no_preprocess_no_compile_regex = qr/
+    # Assembly
+      S | sx
+    /x;
+my $source_no_preprocess_regex = qr/
+      $source_no_preprocess_compile_regex
+    | $source_no_preprocess_no_compile_regex
+    /x;
+# Regex for header files which require preprocessing.
+my $header_preprocess_regex = qr/
+    # C, C++, Objective-C, Objective-C++
+      h
+    # C++
+    | hh | H | hp | hxx | hpp | HPP | h\+\+ | tcc
+    /x;
+# Regexps to match files with the given characteristics.
+my $file_no_preprocess_regex = qr/
+    $cc_regex.+?
+    \.(?: $source_no_preprocess_regex)\b
+    /x;
+my $file_preprocess_regex = qr/
+    $cc_regex.+?
+    \.(?: $header_preprocess_regex
+        | $source_preprocess_regex)\b
+    /x;
+my $file_compile_link_regex = qr/
+    $cc_regex.+?
+    \.(?: $source_preprocess_regex
+        | $source_no_preprocess_regex)\b
+    /x;
+my $file_compile_regex = qr/
+    $cc_regex.+?
+    \.(?: $source_preprocess_compile_regex
+        | $source_no_preprocess_compile_regex)\b
+    /x;
+my $file_no_compile_regex = qr/
+    $cc_regex.+
+    \.(?: $source_preprocess_no_compile_regex
+        | $source_no_preprocess_no_compile_regex)\b
+    /x;
+
 # Expected (hardening) flags. All flags are used as regexps.
 my @cflags = (
     '-g',
@@ -294,6 +370,10 @@ while (my $line = <>) {
     # false positives.
     $start = 1 if $line =~ /^dpkg-buildpackage:/;
     next if not $start;
+    # And stop at the end of the build log. Package details (reported by the
+    # buildd logs) are not important for us. This also prevents false
+    # positives.
+    last if $line =~ /^Build finished at \d{8}-\d{4}$/;
 
     # Detect architecture automatically unless overridden.
     if (not $option_arch
@@ -318,8 +398,10 @@ while (my $line = <>) {
     my $non_verbose = is_non_verbose_build($line);
 
     # One line may contain multiple commands (";"). Treat each one as single
-    # line.
-    my @line = split /(?<!\\);/, $line;
+    # line. parse_line() is slow, only use it when necessary.
+    my @line = (not $line =~ /;/)
+             ? ($line)
+             : Text::ParseWords::parse_line(';', 1, $line);
     foreach $line (@line) {
         # Add newline, drop all other whitespace at the end of a line.
         $line =~ s/\s+$//;
@@ -443,49 +525,68 @@ for (my $i = 0; $i < scalar @input; $i++) {
     # Even if it's a verbose build, we might have to skip this line.
     next if $skip;
 
+    # Skip unnecessary tests when only preprocessing.
+    my $flag_preprocess = 0;
+
+    my $preprocess = 0;
+    my $compile    = 0;
+    my $link       = 0;
 
-    # Is this a compiler or linker command?
-    my $compiler = 1;
-    my $linker   = 0;
-
-    # Linker commands.
-    if ($line =~ m{\s-o                        # -o
-                   [\s\\]*\s+                  # possible line continuation
-                   (?:[/.A-Za-z0-9~_-]+/)?     # path to file
-                        [A-Za-z0-9~_-]+        # binary name (no dots!)
-                   (?:[0-9.]*\.so[0-9.]*[a-z]? # library (including version)
-                    |\.la
-                    |\.cgi)?                   # CGI binary
-                   (?:\s|\\|$)                 # end of file name
-                  }x
-            or $line =~ /^libtool: link: /
-            or $line =~ m{\s*/bin/bash .+?libtool\s+(.+?\s+)?--mode=(re)?link}) {
-        $compiler = 0;
-        $linker   = 1;
+    # Preprocess, compile, assemble.
+    if ($line =~ /$cc_regex.*?\s(-E|-S|-c)\b/) {
+        $preprocess      = 1;
+        $flag_preprocess = 1 if $1 eq '-E';
+        $compile         = 1 if $1 eq '-S' or $1 eq '-c';
+    # Otherwise assume we are linking.
+    } else {
+        $link = 1;
+    }
+
+    # These file types don't require preprocessing.
+    if ($line =~ /$file_no_preprocess_regex/) {
+        $preprocess = 0;
+    }
+    # These file types require preprocessing.
+    if ($line =~ /$file_preprocess_regex/) {
+        $preprocess = 1;
     }
 
     # If there are source files then it's compiling/linking in one step and we
-    # must check both.
-    if ($line =~ /\.(?:c|cc|cpp)\b/) {
-        $compiler = 1;
+    # must check both. We only check for source files here, because header
+    # files cause too many false positives.
+    if (not $flag_preprocess and $line =~ /$file_compile_link_regex/) {
+        # Assembly files don't need CFLAGS.
+        if (not $line =~ /$file_compile_regex/
+                and $line =~ /$file_no_compile_regex/) {
+            $compile = 0;
+        # But the rest does.
+        } else {
+            $compile = 1;
+        }
     }
 
     # Check hardening flags.
     my @missing;
-    if ($compiler and not all_flags_used($line, \@missing, @cflags)
+    if ($compile and not all_flags_used($line, \@missing, @cflags)
             # Libraries linked with -fPIC don't have to (and can't) be linked
             # with -fPIE as well. It's no error if only PIE flags are missing.
-            and not pic_pie_conflict($line, $harden_pie, \@missing, @cflags_pie)) {
+            and not pic_pie_conflict($line, $harden_pie, \@missing, @cflags_pie)
+            # Assume dpkg-buildflags returns the correct flags.
+            and not $line =~ /`dpkg-buildflags --get (?:CFLAGS|CXXFLAGS)`/) {
         error_flags('CFLAGS missing', \@missing, \%flag_renames, $line);
         $exit |= 1 << 3;
     }
-    if ($compiler and not all_flags_used($line, \@missing, @cppflags)) {
+    if ($preprocess and not all_flags_used($line, \@missing, @cppflags)
+            # Assume dpkg-buildflags returns the correct flags.
+            and not $line =~ /`dpkg-buildflags --get CPPFLAGS`/) {
         error_flags('CPPFLAGS missing', \@missing, \%flag_renames, $line);
         $exit |= 1 << 3;
     }
-    if ($linker and not all_flags_used($line, \@missing, @ldflags)
+    if ($link and not all_flags_used($line, \@missing, @ldflags)
             # Same here, -fPIC conflicts with -fPIE.
-            and not pic_pie_conflict($line, $harden_pie, \@missing, @ldflags_pie)) {
+            and not pic_pie_conflict($line, $harden_pie, \@missing, @ldflags_pie)
+            # Assume dpkg-buildflags returns the correct flags.
+            and not $line =~ /`dpkg-buildflags --get LDFLAGS`/) {
         error_flags('LDFLAGS missing', \@missing, \%flag_renames, $line);
         $exit |= 1 << 3;
     }