It seems that we have endianness problem in internal lcms2

Change-Id: I1bec67344ddd5d1caf417355fafd344f48517f77

lcms2/lcms2-endian.patch

+--- misc/lcms2-2.3/include/lcms2.h	2011-12-15 16:45:47.000000000 +0100
++++ misc/build/lcms2-2.3/include/lcms2.h	2012-10-05 14:49:09.721146003 +0200
+@@ -30,8 +30,12 @@
+ 
+ // ********** Configuration toggles ****************************************
+ 
+-// Uncomment this one if you are using big endian machines
+-// #define CMS_USE_BIG_ENDIAN   1
++#ifndef _WIN32
++#include <endian.h>
++#if __BYTE_ORDER == __BIG_ENDIAN
++#define USE_BIG_ENDIAN   1
++#endif
++#endif /* !_WIN32 */
+ 
+ // Uncomment this one if your compiler/machine does NOT support the
+ // "long long" type.

lcms2/makefile.mk

@@ -39,7 +39,7 @@ TARGET=so_lcms2
 TARFILE_NAME=lcms2-2.3
 TARFILE_MD5=327348d67c979c88c2dec59a23a17d85
 
-PATCH_FILES = lcms2.patch
+PATCH_FILES = lcms2.patch lcms2-endian.patch
 
 .IF "$(SYSTEM_LCMS2)" == "YES"
 @all: