bug 518641 - fix the "unify" script to properly handle Java .class files. r=bsmedberg

Java .class files and Mach-O fat binaries use the same 4-byte magic
number at the start of the file. 'unify' uses the magic number to detect
Mach-O binaries, so it chokes on Java .class files. This change makes us
use the same heuristic as file(1), which is to check the second 4 bytes
in the file. Java class files put a version number there, and Mach-O
fat binaries put the number of contained architectures there. Conveniently,
Mach defines only 18 architectures, and Java's lowest shipping version
number is 43, so there's no overlap in valid values.

--HG--
extra : rebase_source : 84f5382bca6fa4bde721686f029514a41dfaf6b4

build/Makefile.in

@@ -101,10 +101,47 @@ libs:: bloatcycle.html
 ifeq ($(OS_ARCH),Darwin)
 libs:: $(topsrcdir)/tools/rb/fix-macosx-stack.pl
 	$(INSTALL) $< $(DIST)/bin
+
+# Basic unit tests for some stuff in the unify script
+check::
+# build ppc/i386 binaries, and unify them
+	rm -f unify-test-ppc unify-test-i386 unify-test-universal
+	$(HOST_CC) -arch ppc $(srcdir)/unify-test.c -o unify-test-ppc
+	$(HOST_CC) -arch i386 $(srcdir)/unify-test.c -o unify-test-i386
+	@if ! $(srcdir)/macosx/universal/unify ./unify-test-ppc ./unify-test-i386 \
+          ./unify-test-universal; then \
+          echo "TEST-UNEXPECTED-FAIL | build/ | unify failed to produce a universal binary!"; \
+          false; \
+        fi
+	@if ! test -f ./unify-test-universal; then \
+          echo "TEST-UNEXPECTED-FAIL | build/ | unify failed to produce a universal binary!"; \
+          false; \
+        fi
+	@if ! file -b ./unify-test-universal | head -n1 | grep -q "^Mach-O universal binary"; then \
+          echo "TEST-UNEXPECTED-FAIL | build/ | unify failed to produce a universal binary!"; \
+          false; \
+        fi
+# try unifying two identical Java class files
+	rm -f unifytesta.class unifytestb.class unifytestc.class
+	cp $(srcdir)/unifytest.class ./unifytesta.class
+	cp $(srcdir)/unifytest.class ./unifytestb.class
+	@if ! $(srcdir)/macosx/universal/unify ./unifytesta.class ./unifytestb.class \
+          ./unifytestc.class; then \
+          echo "TEST-UNEXPECTED-FAIL | build/ | unify failed to unify a Java class file!"; \
+          false; \
+        fi
+	@if ! test -f ./unifytestc.class; then \
+          echo "TEST-UNEXPECTED-FAIL | build/ | unify failed to unify a Java class file!"; \
+          false; \
+        fi
+	@if ! diff -q ./unifytesta.class ./unifytestc.class; then \
+          echo "TEST-UNEXPECTED-FAIL | build/ | unify failed to unify a Java class file!"; \
+          false; \
+        fi
 endif
 
 ifeq ($(OS_ARCH),Linux)
 libs:: $(topsrcdir)/tools/rb/fix-linux-stack.pl
 	$(INSTALL) $< $(DIST)/bin
 endif
-endif
+endif # ENABLE_TESTS

build/macosx/universal/unify

@@ -214,6 +214,7 @@ sub readZipCRCs($);
   sub lstatMode($);
   sub lstatType($);
   sub magic($);
+  sub magic2($);
   sub path($);
   sub stat($);
   sub statSize($);
@@ -1053,6 +1054,7 @@ sub readZipCRCs($) {
       'lstatErrno'  => 0,
       'lstatInit'   => 0,
       'magic'       => undef,
+      'magic2'       => undef,
       'magicErrno'  => 0,
       'magicErrMsg' => undef,
       'magicInit'   => 0,
@@ -1069,7 +1071,7 @@ sub readZipCRCs($) {
   # Returns true if the file is a fat Mach-O file, false if it's not, and
   # undef if an error occurs.  See /usr/include/mach-o/fat.h.
   sub isFat($) {
-    my ($magic, $this);
+    my ($magic, $magic2, $this);
     ($this) = @_;
 
     # magic() caches, there's no separate cache because isFat() doesn't hit
@@ -1078,8 +1080,16 @@ sub readZipCRCs($) {
     if (!defined($magic = $this->magic())) {
       return undef;
     }
+    $magic2 = $this->magic2();
 
-    if ($magic == 0xcafebabe) {
+    # We have to sanity check the second four bytes, because Java class
+    # files use the same magic number as Mach-O fat binaries.
+    # This logic is adapted from file(1), which says that Mach-O uses
+    # these bytes to count the number of architectures within, while
+    # Java uses it for a version number. Conveniently, there are only
+    # 18 labelled Mach-O architectures, and Java's first released
+    # class format used the version 43.0.
+    if ($magic == 0xcafebabe && $magic2 < 20) {
       return 1;
     }
 
@@ -1105,7 +1115,7 @@ sub readZipCRCs($) {
     # Accept Mach-O fat files or Mach-O thin files of either endianness.
     if ($magic == 0xfeedface ||
         $magic == 0xcefaedfe ||
-        $magic == 0xcafebabe) {
+        $this->isFat()) {
       return 1;
     }
 
@@ -1292,7 +1302,7 @@ sub readZipCRCs($) {
     }
 
     $! = 0;
-    my ($bytes, $magic);
+    my ($bytes, $magic, $bytes2, $magic2);
     if (!defined($bytes = sysread($fh, $magic, 4))) {
       $$this{'magicErrno'} = $!;
       $$this{'magicErrMsg'} = 'read "'.$$this{'path'}.'": '.$!;
@@ -1301,6 +1311,9 @@ sub readZipCRCs($) {
       close($fh);
       return undef;
     }
+    else {
+      $bytes2 = sysread($fh, $magic2, 4);
+    }
 
     close($fh);
 
@@ -1308,13 +1321,35 @@ sub readZipCRCs($) {
       # The file is too short, didn't read a magic number.  This isn't really
       # an error.  Return an unlikely value.
       $$this{'magic'} = -1;
+      $$this{'magic2'} = -1;
       return -1;
     }
+    if ($bytes2 != 4) {
+      # File is too short to read a second 4 bytes.
+      $magic2 = -1;
+    }
 
     $$this{'magic'} = unpack('N', $magic);
+    $$this{'magic2'} = unpack('N', $magic2);
     return $$this{'magic'};
   }
 
+  # $FileAttrCache->magic2()
+  #
+  # Returns the second four bytes of the file as a 32-bit little endian number.
+  # See magic(), above for more info.
+  sub magic2($) {
+    my ($this);
+    ($this) = @_;
+
+    # we do the actual work (and cache it) in magic().
+    if (!$$this{'magicInit'}) {
+      my $magic = $$this->magic();
+    }
+
+    return $$this{'magic2'};
+  }
+
   # $FileAttrCache->path()
   #
   # Returns the file's pathname.

build/unify-test.c

@@ -0,0 +1,4 @@
+int main(int argc, char** argv)
+{
+  return 0;
+}

build/unifytest.java

@@ -0,0 +1,5 @@
+class unifytest {
+    public static void main(String[] args) {
+        return;
+    }
+}