Updated changelog.

new iso_image_set_ignore_aclea() and iso_file_source_get_aa_string() flag bit3 to import all xattr namespaces

COPYRIGHT

@@ -1,7 +1,9 @@
 Vreixo Formoso <metalpain2002@yahoo.es>,
 Mario Danic <mario.danic@gmail.com>, 
+Vladimir Serbinenko <phcoder@gmail.com>
 Thomas Schmitt <scdbackup@gmx.net>
-Copyright (C) 2007-2011 Vreixo Formoso, Mario Danic, Thomas Schmitt
+Copyright (C) 2007-2016
+Vreixo Formoso, Mario Danic, Vladimir Serbinenko, Thomas Schmitt
 
 
     This program is free software; you can redistribute it and/or modify

ChangeLog

@@ -1,3 +1,125 @@
+libisofs-1.5.0.tar.gz Sat Sep 15 2018
+===============================================================================
+* New API call iso_image_get_ignore_aclea(),
+  new iso_image_set_ignore_aclea() and iso_file_source_get_aa_string()
+  flag bit3 to import all xattr namespaces
+* New API calls iso_image_was_blind_attrs(), iso_local_set_attrs_errno().
+* New flag bit7 with iso_local_set_attrs() to avoid unnecessary write attempts.
+* New return value 2 of IsoFileSource.get_aa_string() and iso_local_get_attrs().
+* Now putting user defined padding after appended partitions.
+* Bug fix: Add-on sessions with partition offset claimed too many blocks as
+           size. Regression of version 1.4.8.
+* Bug fix: Long Joliet names without dot were mangled with one character too
+           many. Long Joliet names with leading dot were mangled one char
+           too short.
+* Bug fix: Reading beyond array end for HFS+ production caused SIGSEGV with
+           FreeBSD 11 CLANG -O2. Thanks ASX of GhostBSD.
+
+libisofs-1.4.8.tar.gz Tue Sep 12 2017
+===============================================================================
+* Bug fix: iso_read_opts_set_no_rockridge() did not prevent reading of root
+           SUSP.
+* Bug fix: Non-SUSP data in System Use Area prevented image loading if
+           Rock Ridge was enabled. Thanks to Jonathan Dowland.
+* Bug fix: Protective MBR for GPT could emerge with boot flag set.
+* Bug fix: Appended partitions of size >= 4 GiB led to abort with error message
+           "FATAL : ISO overwrite". Thanks to Sven Haardiek.
+* Bug fix: Bit 15 of iso_write_opts_set_system_area did not work with generic
+           MBR.
+* Bug fix: Keeping and patching of loaded boot images failed.
+           Regression by version 1.4.4.
+* Bug fix: Program crashes by intentionally wrong ISO image input.
+           Found by American Fuzzy Lop and Jakub Wilk.
+           Debian bug reports: 872372, 872475, 872545, 872590, 872761.
+* New API calls el_torito_set_full_load(), el_torito_get_full_load().
+* New API call iso_write_opts_set_iso_mbr_part_type().
+
+
+libisofs-1.4.6.tar.gz Fri Sep 16 2016
+===============================================================================
+* Bug fix: SIGSEGV by NULL when a data file was larger than ISO level allows.
+* Bug fix: Interpretation of 17 digit timestamps was wrong.
+* New API calls iso_generate_gpt_guid() and iso_write_opts_set_gpt_guid().
+* Made several pseudo-random ids reproducible by overriding volume modification
+  time.
+
+libisofs-1.4.4.tar.gz Fri Jul 01 2016
+===============================================================================
+* Bug fix: HFS+ production could cause MBR partition of type 0xEE without GPT.
+* Bug fix: Protective MBR was not produced if no System Area data are given.
+* Bug fix: Protective MBR was not recognized if partition is appended.
+* Bug fix: The HFS+ filesystem was not marked in GPT of GRUB2 hybrid layout.
+* Bug fix: HFS+ directories could announce more children than they actually
+           have.
+* Bug fix: At image loading time GRUB2 MBR was not recognized if the partition
+           table is not the protective one as described by UEFI.
+* Bug fix: Oversized text in ISO_SYSAREA_REPORT_DOC_ALPHA.
+           Thanks to Etienne Bergeron.
+* New pseudo path for El Torito boot images:
+  --interval:appened_partition_N:all::
+* New bit15 with options of iso_write_opts_set_system_area() to enforce
+  MBR bootable/active flag.
+* New API calls iso_write_opts_set_appended_as_apm(),
+  iso_write_opts_set_part_like_isohybrid().
+* Introduced image size tolerance of 300 kB in order to recognize SUN Disk
+  Label that was generated by genisoimage -B "...".
+* Added "extern C" to libisofs.h
+* Removed option --silent from libtool runs.
+
+libisofs-1.4.2.tar.gz Sat Nov 28 2015
+===============================================================================
+* Bug fix: zisofs compression caused SIGSEGV (by reading) with files larger
+           than 524160 KiB.
+* Bug fix: iso_node_get_name() of root node returned NULL pointer rather than
+           an empty string
+* Bug fix: Names read from Joliet tree where stripped of trailing ";1"
+* Now sorting the data file content extents by ECMA-119 tree, rather than
+  by the red-black tree which shall consolidate files with identical
+  source object.
+* New API call iso_read_opts_set_ecma119_map().
+* New AAIP variable isofs.nt records name truncation parameters.
+* Rectified handling of oversized filenames by new API calls:
+  iso_image_set_truncate_mode, iso_image_get_truncate_mode,
+  iso_truncate_leaf_name, iso_image_set_node_name, iso_image_tree_clone,
+  iso_image_add_new_dir, iso_image_add_new_file, iso_image_add_new_special,
+  iso_image_add_new_symlink, iso_image_dir_get_node, iso_image_path_to_node
+* Result of a Coverity audit: 50+ code changes, but no easy-to-trigger bugs
+
+libisofs-1.4.0.tar.gz Sun May 17 2015
+===============================================================================
+* Bug fix: iso_image_report_system_area() caused SIGSEGV by NULL if no valid
+           ISO 9660 image was loeaded. Thanks to OmegaPhil.
+* Bug fix: A SIGSEGV could happen when loading a faulty ISO filesystem.
+           Debian bug 774152. Thanks to Jakub Wilk.
+* Bug fix: Rock Ridge Continuation Area could be produced crossing a block
+           boundary. This is heavily disliked by the Linux kernel and spoils
+           the representation of directories which contain many symbolic links.
+* Bug fix: If iso_write_opts_set_hardlinks() enabled automatic inode numbers,
+           then they did not get into effect with nodes were zisofs decoder
+           filters got attached during the image load process.
+* Bug fix: The header indicator of the last El Torito catalog section header
+           was set to 0x90 rather than 0x91 if more than one boot image is in
+           that section.
+* Bug fix: Only 128 bytes of an emerging GPT header block were zeroized.
+* Bug fix: iso_image_report_system_area() did not show GPT partitions of
+           size 0.
+* Bug fix: A zero sized GPT partition was marked after the last appended
+           GPT partition.
+* Bug fix: GPT production did not yield proper results with appended sessions
+           or with TOC emulation enabled.
+* Increased default weight of El Torito boot catalog to 1 billion.
+* Improved handling of cylinder alignment if the resulting image size is not
+  divisible by 2048. Old behavior was to not align. New is to pad up by a
+  few blocks of 512 bytes.
+* New API call iso_write_opts_set_appended_as_gpt()
+  and marking of appended partitions in GPT if GPT emerges for other reasons.
+* New system area type 6 = DEC Alpha SRM boot sector.
+  New API calls iso_image_set_alpha_boot(), iso_image_get_alpha_boot().
+  Thanks to Helge Deller.
+* New API object iso_interval_reader. Enabling flag bits for older API calls
+  iso_write_opts_set_prep_img(), iso_write_opts_set_efi_bootp(),
+  and iso_write_opts_set_partition_img().
+
 libisofs-1.3.8.tar.gz Sat Jun 28 2014
 ===============================================================================
 * Bug fix: Prevent allocation of empty hash tables. Thanks Richard Nolde.

Makefile.am

@@ -13,7 +13,7 @@ ACLOCAL_AMFLAGS = -I ./
 # Build libraries
 
 libisofs_libisofs_la_LDFLAGS = \
-	-version-info $(LT_CURRENT):$(LT_REVISION):$(LT_AGE)
+	-version-info $(LT_CURRENT):$(LT_REVISION):$(LT_AGE) $(LIBLDFLAGS)
 
 # Eventually enabling system adapters for ACL and EA.
 # ts A90409: Eventually enabling use of zlib.
@@ -89,7 +89,7 @@ libinclude_HEADERS = \
 	libisofs/libisofs.h
 
 install-exec-hook:
-	$(LIBBURNIA_LDCONFIG_CMD) "$(DESTDIR)$(libdir)" || echo 'NOTE: Explicite dynamic library configuration failed. If needed, configure manually for:' "$(DESTDIR)$(libdir)"
+	$(LIBBURNIA_LDCONFIG_CMD) "$(DESTDIR)$(libdir)" || echo 'NOTE: Explicit dynamic library configuration failed. If needed, configure manually for:' "$(DESTDIR)$(libdir)"
 
 ## ========================================================================= ##
 

README

@@ -4,7 +4,7 @@
 
 Released under GPL (see COPYING file for details).
 
-Copyright (C) 2008 - 2013 Vreixo Formoso,
+Copyright (C) 2008 - 2016 Vreixo Formoso,
                           Mario Danic,
                           Vladimir Serbinenko,
                           Thomas Schmitt
@@ -14,7 +14,7 @@ libisofs is part of the libburnia project (libburnia-project.org)
 
                         Download, Build and Installation
 
-libisofs code is mantained in a Bazaar repository at Launchpad 
+libisofs code is maintained in a Bazaar repository at Launchpad 
 (https://launchpad.net/libisofs/). You can download it with:
 
 $ bzr branch lp:libisofs/for-libisoburn
@@ -37,7 +37,7 @@ and execute
    ./configure --prefix=/usr
    make
 
-To make the libraries accessible for running resp. developing applications 
+To make the libraries accessible for running and developing applications 
    make install
 
 On GNU/Linux it will try to run program ldconfig with the library installation

acinclude.m4

@@ -16,15 +16,21 @@ AC_DEFUN([TARGET_SHIZZLE],
 
   AC_MSG_CHECKING([target operating system])
 
+  LIBBURNIA_SUPP_ACL=none
+  LIBBURNIA_SUPP_FATTR=none
   LIBBURNIA_LDCONFIG_CMD="echo 'No ldconfig run performed. If needed, configure manually for:'"
   case $target in
     *-*-linux*)
       ARCH=linux
       LIBBURN_ARCH_LIBS=
+      LIBBURNIA_SUPP_ACL=libacl
+      LIBBURNIA_SUPP_FATTR=xattr
       LIBBURNIA_LDCONFIG_CMD=ldconfig
       ;;
     *-*-freebsd*)
       ARCH=freebsd
+      LIBBURNIA_SUPP_ACL=libacl
+      LIBBURNIA_SUPP_FATTR=extattr
       LIBBURN_ARCH_LIBS=-lcam
 
       # This may later be overridden by configure --enable-libdir-pkgconfig
@@ -141,10 +147,12 @@ AC_DEFUN([LIBISOFS_ASSERT_VERS_LIBS],
     LDFLAGS="$LDFLAGS -Wl,--version-script=libisofs/libisofs.ver"
     AC_TRY_LINK([#include <stdio.h>], [printf("Hello\n");],
                 [vers_libs_test="yes"], [vers_libs_test="no"])
-    if test x$vers_libs_test = xno
+    if test x$vers_libs_test = xyes
     then
-        LDFLAGS="$libburnia_save_LDFLAGS"
+        LIBLDFLAGS="-Wl,--version-script=libisofs/libisofs.ver"
     fi
+    LDFLAGS="$libburnia_save_LDFLAGS"
+    AC_SUBST(LIBLDFLAGS)
 ])
 
 
@@ -198,3 +206,15 @@ dnl For debugging only
  
 ])
 
+dnl LIBBURNIA_TRY_TIMEZONE is by Thomas Schmitt, libburnia project
+dnl It tests whether the global variable exists and is suitable for
+dnl integer arithmetics.
+AC_DEFUN([LIBBURNIA_TRY_TIMEZONE],
+[
+    echo -n "checking for timezone variable ... "
+    AC_TRY_LINK([ #include <time.h> ], [long int i; i = 1 - timezone; ], 
+                [LIBBURNIA_TIMEZONE="timezone"], [LIBBURNIA_TIMEZONE="0"]
+    )
+    echo "$LIBBURNIA_TIMEZONE"
+])
+

configure.ac

@@ -1,4 +1,4 @@
-AC_INIT([libisofs], [1.3.8], [http://libburnia-project.org])
+AC_INIT([libisofs], [1.4.9], [http://libburnia-project.org])
 AC_PREREQ([2.50])
 dnl AC_CONFIG_HEADER([config.h])	
 
@@ -26,8 +26,8 @@ dnl
 dnl LT_CURRENT, LT_REVISION and LT_AGE get set directly now.
 dnl
 dnl SONAME of the emerging library is  LT_CURRENT - LT_AGE.
-dnl The linker will do no finer checks. Especially no age range check for
-dnl the cdrskin binary. If SONAME matches, then the couple starts.
+dnl The linker will do no finer checks. If SONAME matches, then the couple
+dnl starts.
 dnl 
 dnl Therefore a run time check is provided by libisofs function 
 dnl iso_lib_version(). It returns the major, minor and micro revision of the 
@@ -40,8 +40,8 @@ dnl
 dnl If LIBISOFS_*_VERSION changes, be sure to change AC_INIT above to match.
 dnl
 LIBISOFS_MAJOR_VERSION=1
-LIBISOFS_MINOR_VERSION=3
-LIBISOFS_MICRO_VERSION=8
+LIBISOFS_MINOR_VERSION=4
+LIBISOFS_MICRO_VERSION=9
 LIBISOFS_VERSION=$LIBISOFS_MAJOR_VERSION.$LIBISOFS_MINOR_VERSION.$LIBISOFS_MICRO_VERSION
 
 AC_SUBST(LIBISOFS_MAJOR_VERSION)
@@ -51,10 +51,10 @@ AC_SUBST(LIBISOFS_VERSION)
 
 dnl Libtool versioning
 LT_RELEASE=$LIBISOFS_MAJOR_VERSION.$LIBISOFS_MINOR_VERSION
-# 2014.06.28 development jump has not yet happened
-# SONAME = 80 - 74 = 6 . Library name = libisofs.6.74.0
-LT_CURRENT=80
-LT_AGE=74
+# 2017.09.12 development jump has not yet happened
+# SONAME = 90 - 84 = 6 . Library name = libisofs.6.84.0
+LT_CURRENT=90
+LT_AGE=84
 LT_REVISION=0
 LT_CURRENT_MINUS_AGE=`expr $LT_CURRENT - $LT_AGE`
 
@@ -92,7 +92,7 @@ LIBBURNIA_ASSERT_ICONV
 
 AC_PROG_LIBTOOL
 AC_SUBST(LIBTOOL_DEPS)
-LIBTOOL="$LIBTOOL --silent"
+# LIBTOOL="$LIBTOOL --silent"
 
 AC_PROG_INSTALL
 
@@ -114,6 +114,16 @@ AC_CHECK_DECL([timegm],
 	,
 	[#include <time.h>])
 
+dnl Whether timezone is an integer variable
+AH_TEMPLATE([Libburnia_timezonE], [Either timezone or 0])
+LIBBURNIA_TRY_TIMEZONE
+if test x$LIBBURNIA_TIMEZONE = xtimezone
+then
+   AC_DEFINE([Libburnia_timezonE], [timezone])
+else
+   AC_DEFINE([Libburnia_timezonE], [0])
+fi
+
 dnl Check if non standard eaccess() function is available
 AC_CHECK_DECL([eaccess], 
 	[AC_DEFINE(HAVE_EACCESS, 1, [Define this if eaccess function is available])],
@@ -162,40 +172,118 @@ dnl Add compiler-specific flags
 AC_ARG_ENABLE(libacl,
 [  --enable-libacl         Enable use of ACL functions by libisofs, default=yes],
                , enable_libacl=yes)
-if test "x$enable_libacl" = xyes; then
+LIBACL_DEF=
+has_acl_h_but_no_func=0
+if test x$LIBBURNIA_SUPP_ACL = xlibacl
+then
+    if test x$enable_libacl = xyes; then
 dnl Check whether there is libacl-devel and libacl-runtime.
 dnl If not, erase this macro which would enable use of acl_to_text and others
-    LIBACL_DEF="-DLibisofs_with_aaip_acL"
+        LIBACL_DEF="-DLibisofs_with_aaip_acL"
 dnl The empty yes case obviously causes -lacl to be linked
-    has_acl_h_but_no_func=0
-    AC_CHECK_HEADER(sys/acl.h, AC_CHECK_LIB(acl, acl_to_text, , has_acl_h_but_no_libacl=1 ), LIBACL_DEF= )
-    if test "$has_acl_h_but_no_libacl" = 1
+        AC_CHECK_HEADER(sys/acl.h, AC_CHECK_LIB(acl, acl_to_text, , has_acl_h_but_no_libacl=1 ), LIBACL_DEF= )
+        if test "$has_acl_h_but_no_libacl" = 1
+        then
+            AC_CHECK_LIB(c, acl_to_text, X= , LIBACL_DEF= )
+        fi
+    fi
+fi
+if test x$LIBACL_DEF = x-DLibisofs_with_aaip_acL
+then
+    if test x$has_acl_h_but_no_libacl = x1
     then
-       AC_CHECK_LIB(c, acl_to_text, X= , LIBACL_DEF= )
-     fi
+        echo "enabled  local processing of ACL"
+    else
+        echo "enabled  libacl, local processing of ACL"
+    fi
 else
-    LIBACL_DEF=
+    echo "disabled local processing of ACL"
 fi
 AC_SUBST(LIBACL_DEF)
 
 
-dnl ts A90123
+dnl ts A90123 - B80508
 AC_ARG_ENABLE(xattr,
-[  --enable-xattr          Enable use of xattr by libisofs, default=yes],
-               , enable_xattr=yes)
-if test "x$enable_xattr" = xyes; then
-dnl Check whether there is the header for Linux xattr. 
+              [  --enable-xattr          Enable use of extended file attributes by libisofs, default=yes],
+              , enable_xattr=yes)
+AC_ARG_ENABLE(xattr_h_pref_attr,
+              [  --enable-xattr-h-pref-attr   Prefer include file attr/xattr.h over sys/xattr.h, default=no],
+              , enable_xattr_h_pref_attr=no)
+
+XATTR_DEF=
+XATTR_ADDON_DEF=
+if test x"$LIBBURNIA_SUPP_FATTR" = xxattr
+then
+    if test "x$enable_xattr" = xyes; then
+dnl Check whether there is the header for Linux xattr.
 dnl If not, erase this macro which would enable use of listxattr and others
-    XATTR_DEF="-DLibisofs_with_aaip_xattR"
-    AC_CHECK_HEADER(attr/xattr.h, AC_CHECK_LIB(c, listxattr, X= , XATTR_DEF= ), XATTR_DEF= )
-    if test "x$XATTR_DEF" = x
-    then
-       XATTR_DEF="-DLibisofs_with_freebsd_extattR"
-       AC_CHECK_HEADER(sys/extattr.h, AC_CHECK_LIB(c, extattr_list_file, X=, XATTR_DEF= ), XATTR_DEF= )
+
+        XATTR_A_DEF=
+        XATTR_S_DEF=
+        if test x"$enable_xattr_h_pref_attr" = xyes
+        then
+            echo "prefering include file attr/xattr.h over sys/attr.h"
+            XATTR_A_DEF=1
+            AC_CHECK_HEADER(attr/xattr.h, AC_CHECK_LIB(c, listxattr, X= ,
+                                                 XATTR_A_DEF= ), XATTR_A_DEF= )
+            if test x"$XATTR_A_DEF" = x1
+            then
+                XATTR_DEF="-DLibisofs_with_aaip_xattR"
+            else
+                XATTR_S_DEF=1
+                AC_CHECK_HEADER(sys/xattr.h, AC_CHECK_LIB(c, listxattr, X= ,
+                                                 XATTR_S_DEF= ), XATTR_S_DEF= )
+                if test x"$XATTR_S_DEF" = x1
+                then
+                    XATTR_DEF="-DLibisofs_with_aaip_xattR"
+                    XATTR_ADDON_DEF="-DLibisofs_with_sys_xattR"
+                fi
+            fi
+        else
+            XATTR_S_DEF=1
+            AC_CHECK_HEADER(sys/xattr.h, AC_CHECK_LIB(c, listxattr, X= ,
+                                                 XATTR_S_DEF= ), XATTR_S_DEF= )
+            if test x"$XATTR_S_DEF" = x1
+            then
+                XATTR_DEF="-DLibisofs_with_aaip_xattR"
+                XATTR_ADDON_DEF="-DLibisofs_with_sys_xattR"
+            else
+                XATTR_A_DEF=1
+                AC_CHECK_HEADER(attr/xattr.h, AC_CHECK_LIB(c, listxattr, X= ,
+                                                 XATTR_A_DEF= ), XATTR_A_DEF= )
+                if test x"$XATTR_A_DEF" = x1
+                then
+                    XATTR_DEF="-DLibisofs_with_aaip_xattR"
+                fi
+            fi
+        fi
+        if test x"$XATTR_S_DEF" = x1
+        then
+            echo "decided  to include file sys/attr.h"
+        elif test x"$XATTR_A_DEF" = x1
+        then
+            echo "decided  to include file attr/xattr.h"
+        fi
+    fi
+elif test x"$LIBBURNIA_SUPP_FATTR" = xextattr
+then
+    if test "x$enable_xattr" = xyes; then
+        XATTR_DEF="-DLibisofs_with_freebsd_extattR"
+        AC_CHECK_HEADER(sys/extattr.h, AC_CHECK_LIB(c, extattr_list_file, X=,
+                                                    XATTR_DEF= ), XATTR_DEF= )
     fi
-else
-    XATTR_DEF=
 fi
+if test x$XATTR_DEF = x-DLibisofs_with_aaip_xattR
+then
+    echo "enabled  xattr, local processing of extended file attributes Linux style"
+elif test x$XATTR_DEF = x-DLibisofs_with_freebsd_extattR
+then
+    echo "enabled  extattr, local processing of extended file attributes FreeBSD style"
+else
+    echo "disabled local processing of extended file attributes"
+fi
+XATTR_DEF="$XATTR_DEF $XATTR_ADDON_DEF"
+
 AC_SUBST(XATTR_DEF)
 
 

demo/demo.c

@@ -1,6 +1,6 @@
 
 /*
- * Copyright (c) 2007 - 2009 Vreixo Formoso, Thomas Schmitt
+ * Copyright (c) 2007 - 2016 Vreixo Formoso, Thomas Schmitt
  * 
  * This file is part of the libisofs project; you can redistribute it and/or 
  * modify it under the terms of the GNU General Public License version 2 
@@ -24,7 +24,9 @@ static char helptext[][80] = {
 "               Output the contents of an iso image.",
 "  -iso_cat  image_file path_in_image",
 "               Extract a file from a given ISO image and put out its content",
-"               to stdout. The file is addressed by path_in_image.",
+"               to stdout. The file is addressed by path_in_image. The ISO",
+"               image does not get loaded but rather the lookups are done",
+"               directly in the image file.",
 "  -iso_modify  image_file absolute_directory_path output_file",
 "               Load an iso image, add a directory, and write complete image.",
 "  -iso_ms  image_lba nwa image_file directory_path output_file",
@@ -58,6 +60,22 @@ static char helptext[][80] = {
 #endif
 
 
+/* ----------------------------- utilities -------------------------- */
+
+
+void demo_report_iso_err(int err, char *occasion)
+{
+    char *severity;
+
+    fprintf(stderr, "%s : err = 0x%X", occasion, (unsigned int) err);
+    if (err < 0) {
+        iso_sev_to_text(iso_error_get_severity(err), &severity);
+        fprintf(stderr, " -> %s '%s'", severity, iso_error_to_msg(err));
+    }
+    fprintf(stderr, "\n");
+}
+
+
 /* ------------------------- from demo/tree.c ----------------------- */
 
 static void
@@ -357,7 +375,7 @@ int gesture_iso(int argc, char **argv)
 
     result = iso_write_opts_new(&opts, 0);
     if (result < 0) {
-        printf ("Cant create write opts, error %d\n", result);
+        printf ("Cannot create write opts, error %d\n", result);
         goto ex;
     }
     iso_write_opts_set_iso_level(opts, level);
@@ -367,7 +385,7 @@ int gesture_iso(int argc, char **argv)
 
     result = iso_image_create_burn_source(image, opts, &burn_src);
     if (result < 0) {
-        printf ("Cant create image, error %d\n", result);
+        printf ("Cannot create image, error %d\n", result);
         goto ex;
     }
 
@@ -446,7 +464,8 @@ iso_read_print_dir(IsoFileSource *dir, int level)
         sp[i+1] = ' ';
     }
 
-    sp[level * 2-1] = '-';
+    if (level > 0)
+        sp[level * 2 - 1] = '-';
     sp[level * 2] = '\0';
 
     ret = iso_file_source_open(dir);
@@ -474,37 +493,43 @@ iso_read_print_dir(IsoFileSource *dir, int level)
 
 int gesture_iso_read(int argc, char **argv)
 {
-    int result;
-    IsoImageFilesystem *fs;
-    IsoDataSource *src;
-    IsoFileSource *root;
-    IsoReadOpts *ropts;
+    int result, initialized = 0, return_val = 1;
+    IsoImageFilesystem *fs = NULL;
+    IsoDataSource *src = NULL;
+    IsoFileSource *root = NULL;
+    IsoReadOpts *ropts = NULL;
 
     if (argc != 2) {
         printf ("You need to specify a valid path\n");
-        return 1;
+        goto ex;
     }
 
-    iso_init();
+    result = iso_init();
+    if (result < 0) {
+        demo_report_iso_err(result, "Cannot init libisofs");
+        goto ex;
+    }
+    initialized = 1;
+
     iso_set_msgs_severities("NEVER", "ALL", "");
 
     result = iso_data_source_new_from_file(argv[1], &src);
     if (result < 0) {
-        printf ("Error creating data source\n");
-        return 1;
+        demo_report_iso_err(result, "Error creating data source");
+        goto ex;
     }
-
     result = iso_read_opts_new(&ropts, 0);
     if (result < 0) {
-        fprintf(stderr, "Error creating read options\n");
-        return 1;
+        demo_report_iso_err(result, "Error creating read options");
+        goto ex;
     }
     result = iso_image_filesystem_new(src, ropts, 1, &fs);
-    iso_read_opts_free(ropts);
     if (result < 0) {
-        printf ("Error creating filesystem\n");
-        return 1;
+        demo_report_iso_err(result, "Error creating filesystem");
+        goto ex;
     }
+    iso_read_opts_free(ropts);
+    ropts = NULL;
 
     printf("\nVOLUME INFORMATION\n");
     printf("==================\n\n");
@@ -523,18 +548,27 @@ int gesture_iso_read(int argc, char **argv)
 
     result = fs->get_root(fs, &root);
     if (result < 0) {
-        printf ("Can't get root %d\n", result);
-        return 1;
+        demo_report_iso_err(result, "Cannot get root object");
+        goto ex;
     }
     /* iso_read_print_file_src(root); */
     iso_read_print_dir(root, 0);
-    iso_file_source_unref(root);
 
-    fs->close(fs);
-    iso_filesystem_unref((IsoFilesystem*)fs);
-    iso_data_source_unref(src);
-    iso_finish();
-    return 0;
+    return_val = 0;
+ex:;
+    if (root != NULL)
+        iso_file_source_unref(root);
+    if (ropts != NULL)
+        iso_read_opts_free(ropts);
+    if (fs != NULL) {
+        fs->close(fs);
+        iso_filesystem_unref((IsoFilesystem*)fs);
+    }
+    if (src != NULL)
+        iso_data_source_unref(src);
+    if (initialized)
+        iso_finish();
+    return return_val;
 }
 
 
@@ -543,83 +577,101 @@ int gesture_iso_read(int argc, char **argv)
 
 int gesture_iso_cat(int argc, char **argv)
 {
-    int res, write_ret;
-    IsoFilesystem *fs;
-    IsoFileSource *file;
+    int res, write_ret, ret;
+    IsoFilesystem *fs = NULL;
+    IsoFileSource *file = NULL;
     struct stat info;
-    IsoDataSource *src;
-    IsoReadOpts *opts;
+    IsoDataSource *src = NULL;
+    IsoReadOpts *opts = NULL;
 
     if (argc != 3) {
-        fprintf(stderr, "Usage: isocat /path/to/image /path/to/file\n");
+        fprintf(stderr, "Usage: -iso_cat /path/to/image /path/to/file\n");
         return 1;
     }
 
     res = iso_init();
     if (res < 0) {
-        fprintf(stderr, "Can't init libisofs\n");
+        demo_report_iso_err(res, "Cannot init libisofs");
         return 1;
     }
-    
+
+    /* Important Note:
+       From here on memory objects get created which need to be freed in
+       the end. Therefore in case of problems no direct return, but rather
+       a hop to label "ex:", where cleanup happens.
+     */
+
     res = iso_data_source_new_from_file(argv[1], &src);
     if (res < 0) {
-        fprintf(stderr, "Error creating data source\n");
-        return 1;
+        demo_report_iso_err(res, "Error creating data source object");
+        ret = 1; goto ex;
     }
 
     res = iso_read_opts_new(&opts, 0);
     if (res < 0) {
-        fprintf(stderr, "Error creating read options\n");
-        return 1;
+        demo_report_iso_err(res, "Error creating read options object");
+        ret = 1; goto ex;
     }
     res = iso_image_filesystem_new(src, opts, 1, &fs);
     if (res < 0) {
-        fprintf(stderr, "Error creating filesystem\n");
-        return 1;
+        demo_report_iso_err(res, "Error creating filesystem object");
+        ret = 1; goto ex;
     }
     iso_read_opts_free(opts);
+    opts = NULL;
 
     res = fs->get_by_path(fs, argv[2], &file);
     if (res < 0) {
-        fprintf(stderr, "Can't get file, err = %d\n", res);
-        return 1;
+        demo_report_iso_err(res, "Cannot get file object with given path");
+        ret = 1; goto ex;
     }
 
     res = iso_file_source_lstat(file, &info);
     if (res < 0) {
-        fprintf(stderr, "Can't stat file, err = %d\n", res);
-        return 1;
+        demo_report_iso_err(res,
+                         "Cannot inquire type of file object with given path");
+        ret = 1; goto ex;
     }
 
     if (S_ISDIR(info.st_mode)) {
         fprintf(stderr, "Path refers to a directory!!\n");
-        return 1;
+        ret = 1; goto ex;
     } else {
         char buf[1024];
         res = iso_file_source_open(file);
         if (res < 0) {
-            fprintf(stderr, "Can't open file, err = %d\n", res);
-            return 1;
+            demo_report_iso_err(res,
+                                "Cannot open file object with given path");
+            ret = 1; goto ex;
         }
         while ((res = iso_file_source_read(file, buf, 1024)) > 0) {
             write_ret = fwrite(buf, 1, res, stdout);
             if (write_ret < res) {
                 printf ("Cannot write block to stdout. errno= %d\n", errno);
-                return 1;
+                iso_file_source_close(file);
+                ret = 1; goto ex;
             }
         }
-        if (res < 0) {
-            fprintf(stderr, "Error reading, err = %d\n", res);
-            return 1;
-        }
         iso_file_source_close(file);
+        if (res < 0) {
+            demo_report_iso_err(res, "Error while reading data content");
+            fprintf(stderr, "Error reading, err = 0x%X\n", (unsigned int) res);
+            ret = 1; goto ex;
+        }
     }
-    
-    iso_file_source_unref(file);
-    iso_filesystem_unref(fs);
-    iso_data_source_unref(src);
+
+    ret = 0;
+ex:;
+    if (file != NULL)
+        iso_file_source_unref(file);
+    if (fs != NULL)
+        iso_filesystem_unref(fs);
+    if (opts != NULL)
+        iso_read_opts_free(opts);
+    if (src != NULL)
+        iso_data_source_unref(src);
     iso_finish();
-    return 0;
+    return ret;
 }
 
 
@@ -633,14 +685,14 @@ void iso_modify_usage(char **argv)
 
 int gesture_iso_modify(int argc, char **argv)
 {
-    int result;
-    IsoImage *image;
-    IsoDataSource *src;
-    struct burn_source *burn_src;
+    int result, return_val = 1, initialized = 0;
+    IsoImage *image = NULL;
+    IsoDataSource *src = NULL;
+    struct burn_source *burn_src = NULL;
     unsigned char buf[2048];
     FILE *fp = NULL;
-    IsoWriteOpts *opts;
-    IsoReadOpts *ropts;
+    IsoWriteOpts *opts = NULL;
+    IsoReadOpts *ropts = NULL;
 	
     if (argc < 4) {
         iso_modify_usage(argv);
@@ -653,20 +705,25 @@ int gesture_iso_modify(int argc, char **argv)
         goto ex;
     }
 
-    iso_init();
+    result = iso_init();
+    if (result < 0) {
+        demo_report_iso_err(result, "Cannot init libisofs");
+        goto ex;
+    }
+    initialized = 1;
     iso_set_msgs_severities("NEVER", "ALL", "");
     
     /* create the data source to accesss previous image */
     result = iso_data_source_new_from_file(argv[1], &src);
     if (result < 0) {
-        printf ("Error creating data source\n");
+        demo_report_iso_err(result, "Error creating data source");
         goto ex;
     }
     
     /* create the image context */
     result = iso_image_new("volume_id", &image);
     if (result < 0) {
-        printf ("Error creating image\n");
+        demo_report_iso_err(result, "Error creating image");
         goto ex;
     }
     iso_tree_set_follow_symlinks(image, 0);
@@ -675,58 +732,75 @@ int gesture_iso_modify(int argc, char **argv)
     /* import previous image */
     result = iso_read_opts_new(&ropts, 0);
     if (result < 0) {
-        fprintf(stderr, "Error creating read options\n");
+        demo_report_iso_err(result, "Error creating read options");
         goto ex;
     }
     result = iso_image_import(image, src, ropts, NULL);
-    iso_read_opts_free(ropts);
-    iso_data_source_unref(src);
     if (result < 0) {
-        printf ("Error importing previous session %d\n", result);
+        demo_report_iso_err(result, "Error importing previous session");
         goto ex;
     }
+    /* (One could of course keep them alive until cleanup) */
+    iso_read_opts_free(ropts);
+    ropts = NULL;
+    iso_data_source_unref(src);
+    src = NULL;
     
     /* add new dir */
     result = iso_tree_add_dir_rec(image, iso_image_get_root(image), argv[2]);
     if (result < 0) {
-        printf ("Error adding directory %d\n", result);
+        demo_report_iso_err(result, "Error adding directory");
         goto ex;
     }
     
-    /* generate a new image with both previous and added contents */
+    /* Generate a new image with both previous and added contents.
+       Profile 1 means Rock Ridge and ISO level 3.
+    */
     result = iso_write_opts_new(&opts, 1);
     if (result < 0) {
-        printf("Cant create write opts, error %d\n", result);
+        demo_report_iso_err(result, "Cannot create write opts");
         goto ex;
     }
-    /* for isolinux: iso_write_opts_set_allow_full_ascii(opts, 1); */
-    
+    /* Prefer specs violation over relocation deep directories */
+    iso_write_opts_set_allow_deep_paths(opts, 1);
+
+    /* For MS-Windows readers : iso_write_opts_set_joliet(opts, 1); */
+ 
     result = iso_image_create_burn_source(image, opts, &burn_src);
     if (result < 0) {
-        printf ("Cant create image, error %d\n", result);
+        demo_report_iso_err(result, "Cannot create image object");
         goto ex;
     }
-    
     iso_write_opts_free(opts);
+    opts = NULL;
     
     while (burn_src->read_xt(burn_src, buf, 2048) == 2048) {
         result = fwrite(buf, 1, 2048, fp);
         if (result < 2048) {
-            printf ("Cannot write block. errno= %d\n", errno);
+            fprintf (stderr, "Cannot write block. errno= %d\n", errno);
             goto ex;
         }
     }
-    fclose(fp);
-    burn_src->free_data(burn_src);
-    free(burn_src);
-    
-    iso_image_unref(image);
-    iso_finish();
-    return 0;
+
+    return_val = 0;
 ex:
     if (fp != NULL)
         fclose(fp);
-    return 1;
+    if (opts != NULL)
+        iso_write_opts_free(opts);
+    if (burn_src != NULL) {
+        burn_src->free_data(burn_src);
+        free(burn_src);
+    }
+    if (image != NULL)
+        iso_image_unref(image);
+    if (ropts != NULL)
+        iso_read_opts_free(ropts);
+    if (src != NULL)
+        iso_data_source_unref(src);
+    if (initialized)
+        iso_finish();
+    return return_val;
 }
 
 
@@ -740,14 +814,14 @@ void iso_ms_usage(char **argv)
 
 int gesture_iso_ms(int argc, char **argv)
 {
-    int result;
-    IsoImage *image;
-    IsoDataSource *src;
-    struct burn_source *burn_src;
+    int result, return_val = 1, initialized = 0;
+    IsoImage *image = NULL;
+    IsoDataSource *src = NULL;
+    struct burn_source *burn_src = NULL;
     unsigned char buf[2048];
     FILE *fp = NULL;
-    IsoWriteOpts *opts;
-    IsoReadOpts *ropts;
+    IsoWriteOpts *opts = NULL;
+    IsoReadOpts *ropts = NULL;
     uint32_t ms_block;
 	
     if (argc < 6) {
@@ -767,20 +841,26 @@ int gesture_iso_ms(int argc, char **argv)
         goto ex;
     }
 
-    iso_init();
+    result = iso_init();
+    if (result < 0) {
+        demo_report_iso_err(result, "Cannot init libisofs");
+        goto ex;
+    }
+    initialized = 1;
+
     iso_set_msgs_severities("NEVER", "ALL", "");
     
     /* create the data source to accesss previous image */
     result = iso_data_source_new_from_file(argv[3], &src);
     if (result < 0) {
-        printf ("Error creating data source\n");
+        demo_report_iso_err(result, "Error creating data source");
         goto ex;
     }
     
     /* create the image context */
     result = iso_image_new("volume_id", &image);
     if (result < 0) {
-        printf ("Error creating image\n");
+        demo_report_iso_err(result, "Error creating image");
         goto ex;
     }
     iso_tree_set_follow_symlinks(image, 0);
@@ -795,23 +875,25 @@ int gesture_iso_ms(int argc, char **argv)
     iso_read_opts_set_start_block(ropts, atoi(argv[1]));
     result = iso_image_import(image, src, ropts, NULL);
     iso_read_opts_free(ropts);
+    ropts = NULL;
     iso_data_source_unref(src);
+    src = NULL;
     if (result < 0) {
-        printf ("Error importing previous session %d\n", result);
+        demo_report_iso_err(result, "Error importing previous session");
         goto ex;
     }
     
     /* add new dir */
     result = iso_tree_add_dir_rec(image, iso_image_get_root(image), argv[4]);
     if (result < 0) {
-        printf ("Error adding directory %d\n", result);
+        demo_report_iso_err(result, "Error adding directory");
         goto ex;
     }
     
     /* generate a multisession image with new contents */
     result = iso_write_opts_new(&opts, 1);
     if (result < 0) {
-        printf("Cant create write opts, error %d\n", result);
+        demo_report_iso_err(result, "Cannot create write opts");
         goto ex;
     }
     
@@ -822,10 +904,11 @@ int gesture_iso_ms(int argc, char **argv)
 
     result = iso_image_create_burn_source(image, opts, &burn_src);
     if (result < 0) {
-        printf ("Cant create image, error %d\n", result);
+        printf ("Cannot create image, error %d\n", result);
         goto ex;
     }
     iso_write_opts_free(opts);
+    opts = NULL;
     
     while (burn_src->read_xt(burn_src, buf, 2048) == 2048) {
         result = fwrite(buf, 1, 2048, fp);
@@ -834,17 +917,26 @@ int gesture_iso_ms(int argc, char **argv)
             goto ex;
         }
     }
-    fclose(fp);
-    burn_src->free_data(burn_src);
-    free(burn_src);
-    
-    iso_image_unref(image);
-    iso_finish();
-    return 0;
+
+    return_val = 0;
 ex:;
+    if (burn_src != NULL) {
+        burn_src->free_data(burn_src);
+        free(burn_src);
+    }
+    if (opts != NULL)
+        iso_write_opts_free(opts);
+    if (image)
+        iso_image_unref(image);
+    if (ropts != NULL)
+        iso_read_opts_free(ropts);
+    if (src != NULL)
+        iso_data_source_unref(src);
+    if (initialized)
+        iso_finish();
     if (fp != NULL)
         fclose(fp);
-    return 1;
+    return return_val;
 }
 
 

doc/boot_sectors.txt

@@ -4,8 +4,9 @@
             Collection of Boot Sector Formats for ISO 9660 Images
 
 
-            by Thomas Schmitt    - mailto:scdbackup@gmx.net
-            Libburnia project    - mailto:libburn-hackers@pykix.org
+            by Thomas Schmitt    - scdbackup@gmx.net
+            Libburnia project    - bug-xorriso@gnu.org
+                                   pkg-libburnia-devel@lists.alioth.debian.org
 
 This information is collected from various sources. Some is backed by
 specifications, some is just rumor which happens to work (maybe not even that).
@@ -54,6 +55,7 @@ Sources:
     El Torito, Bootable CD-ROM Format Specification, Version 1.0, 1995
     which refers to ECMA-119, the standard for ISO 9660 filesystems.
     libisofs/eltorito.[ch] by Vreixo Formoso.
+    http://www.uefi.org/sites/default/files/resources/2_4_Errata_B.pdf
 
 
 ECMA-119 prescribes that the first 32 kB of an ISO 9660 image are System Area
@@ -67,9 +69,9 @@ intervals for:
 - Directory trees, tables, boot catalog, embedded partitions and filesystems.
 - Data file content, including content of El Torito boot images.
 
-The Boot Record is an ECMA-119 Volume Descriptor which is eventually located
-at 2 kB block number 17 (decimal). Its content points to the location of the
-Boot Catalog.
+The Boot Record is an ECMA-119 Volume Descriptor which is located at 2 kB block
+number 17 (decimal), if present at all. Its content points to the location of
+the Boot Catalog.
 The format is described in part by ECMA-119 8.2 "Boot Record" and further
 specified by El Torito figure 7.
 
@@ -124,11 +126,11 @@ Defined by El Torito are:
     0 = "80x86" which is used for standard PCs with Intel x86 or compatible CPU
     1 = "PowerPC" (possibly for IBM machines with PowerPC CPU)
     2 = "Mac" (possibly for Apple computers with MC68000 or PowerPC CPU)
-Further in use by GRUB2 and ISOLINUX is:
- 0xef = EFI, a competitor resp. successor to PC-BIOS, possibly in use with
-             Intel ia64 Itanium and possibly with newer Apple machines.
+UEFI 2.4 specifies in 12.3.2.1 "ISO-9660 and El Torito":
+ 0xef = EFI, a competitor and successor to PC-BIOS, further in use with
+             Intel ia64 Itanium and newer Apple machines.
 
-Words resp. numbers are represented are little-endian. 
+Words and numbers are represented as little-endian. 
 
 Validation Entry:
 
@@ -176,7 +178,7 @@ Byte Range | Value      | Meaning
            |            |
    5 -   5 |          0 | Unused
            |            |
-   6 -   7 |  sec_count | Sector Count.
+   6 -   7 |  sec_count | Sector Count. Sector size 512:
            |            | "the number of virtual/emulated sectors the system
            |            |  will store at Load Segment during the initial boot
            |            |  procedure."
@@ -231,7 +233,7 @@ Byte Range | Value      | Meaning
            |            | 0 if not emulation == 4.
    5 -   5 |          0 | Unused
            |            |
-   6 -   7 |  sec_count | Sector Count.
+   6 -   7 |  sec_count | Sector Count. Sector size 512.
            |            | See above Initial/Default Entry
            |            | libisofs stores 1 for emulated boot_media and a
            |            | user defined value for boot_media == 0. Often: 4.
@@ -300,28 +302,44 @@ Byte Range | Value      | Meaning
 
 Sources:
     http://en.wikipedia.org/wiki/Master_boot_record
+    https://en.wikipedia.org/wiki/INT_13H
     Mailing list conversations with H. Peter Anvin and Vladimir Serbinenko.
+    Mail conversations with Natalia Portillo.
 
 
 The candidates for MBR booting will normally use El Torito rather than MBR
 if the ISO image is presented on CD, DVD, or BD media.
-The eventual MBR comes into effect if the image is on a media that is
-interpreted by the BIOS as some kind of hard disk. Usually real hard disks,
-floppy disks, USB sticks, memory cards.
+The MBR comes into effect if the image is on a media that is interpreted by
+the BIOS as some kind of hard disk. Usually real hard disks, floppy disks,
+USB sticks, memory cards.
 
 An important part of an MBR is the DOS style partition table. It describes up
 to four primary partitions. There are two formats used for block address:
 Cylinder/Head/Sector (C/H/S) and Logical Block Address (LBA). Both are based
 on units of 512 bytes. So MBR_LBA = ISO_LBA * 4.
 
+Contemporary x86 BIOS normally supports LBA addressing directly.
+If INT 0x13 AH 0x41 returns with CX bit0 set, then INT 0x13 AH 0x42 may be used
+for reading. (Sometimes even if the bit is not set to indicate the capability.)
+
 For C/H/S, the sector address is broken up into whole cylinders, remaining
 heads, and remaining sectors + 1. The nomenclature seems to stem from antique
 drum storage.
 There are two parameters, sectors_per_head and heads_per_cylinder which are not
-stored in the MBR. So it is more or less arbitray how to convert a LBA into
-a C/H/S address and vice versa. For maximum range of C/H/S addresses one
-may use sectors_per_head = 63 , heads_per_cylinder = 255.
+stored in the MBR. So at ISO production time it is more or less arbitray how
+to convert a LBA into a C/H/S address and vice versa.
+At boot time the x86 BIOS decides about the two parameters. The boot loader may
+inquire these values by INT 0x13 AH 0x08 and use them to convert LBA to C/H/S
+for the read operation INT 0x13 AH 0x02. So the C/H/S values in an ISO's
+partition table are quite fictional and of few impact on boot loaders.
 
+More important seems to align partition ends to a consistent cylinder size,
+because some partition editors deduce their idea of disk geometry from there
+and raise protest if they deem it inconsistent.
+For maximum range of C/H/S addresses one may use sectors_per_head = 63 ,
+heads_per_cylinder = 255. But that is not divisible by 4 and imposes alignment
+problems with ISO 9660 filesystems. So (32,64) for images up to 1 GiB
+or (63,252) for larger images are better.
 
 Words are composed little-endian style.
 
@@ -330,8 +348,9 @@ Byte Range | Value      | Meaning
    0 - 439 | = opaque = | Code Area filled with bytes for some boot system,
            |            | typically machine code.
            |            |
- 440 - 443 | disk_sgntr | Disc signature, an individual disk id of obscure
-           |            | usability.
+ 440 - 443 | disk_sgntr | Disc signature: An individual disk id. Some software
+           |            | might use it to recognize the same storage medium
+           |            | at different device addresses.
            |            | (The Code Area might extend up to this field.)
            |            |
  444 - 445 |          0 | "usually nulls"
@@ -339,8 +358,11 @@ Byte Range | Value      | Meaning
            |            |
  446 - 461 | ========== | Partition Table Entry for partition 1
            |            |
- 446 - 446 |     status | Governs bootability:
-           |            | 0x80 = bootable/active , 0x00 non-bootable/inactive
+ 446 - 446 |     status | For some generic MBRs this marks the one partition
+           |            | from which the MBR should load and run more code.
+           |            | 0x80 = bootflag/active , 0x00 = noboot/inactive
+           |            | Some BIOSes ignore MBRs with no bootflag in any of
+           |            | their partition table entries.
            |            |
  447 - 449 | ========== | C/H/S address of partition start
  447 - 447 | start_head | Heads part of start address.
@@ -385,7 +407,7 @@ on Linux a partition device file (e.g. /dev/sdb1) which cannot be used to mount
 the ISO filesystem.
 
 libisofs is able to produce a second set of trees and meta data which is
-suitable for being mounted at start block 16 (ISO) resp. 64 (MBR).
+suitable for being mounted at start block 16 (ISO) which is block 64 in MBR.
 See <libisofs/libisofs.h> for call iso_write_opts_set_part_offset()
 and http://libburnia-project.org/wiki/PartitionOffset for examples with
 program xorriso.
@@ -403,6 +425,8 @@ Sources:
     http://www.informit.com/articles/article.aspx?p=376123&seqNum=3
     syslinux-4.05/utils/isohybrid.c
     Mail conversations with Vladimir Serbinenko.
+    Mail conversations with Natalia Portillo of DiskImageChef,
+                            who quoted "Inside Macintosh" Volumes IV and V.
 
 
 APM has an adjustable block size. Because the ISO images shall always work
@@ -411,11 +435,19 @@ additional GPT, only block size 2048 is considered here.
 The role of APM in the boot process is to guide the firmware to a
 HFS+ filesystem.
 
-Block0 of an APM begins at byte 0 of the medium. Thus it collides with MBR and
-other boot sector formats. By lucky coincidence it is possible to compose
-a mock-up of a Block0 which is acceptable to firmware which expects APM,
-and is also harmless x86 machine code with no negative side effects.
-So it is possible to combine APM with an especially prepared MBR.
+Block0 (aka Driver Descriptor Map) of an APM begins at byte 0 of the medium.
+Thus it collides with MBR and other boot sector formats. By lucky coincidence
+it is possible to compose a mock-up of a Block0 which is acceptable to firmware
+which expects APM, and is also harmless x86 machine code with no negative
+side effects. So it is possible to combine APM with an especially prepared MBR.
+
+Block0 is optional. But in the context of bootable hybrid ISOs it is not only
+needed to announce block size 2048, but also it is much better suited for
+staging as harmless x86 machine code than is an APM partition entry.
+
+Usually there is no Device Partition Map block (signature "TS"), although it
+is demanded by the book "Inside Macintosh". It would sit where GPT has its
+header block. So DPM is not described here.
 
 The layout of a Block0 of an APM is:
 
@@ -426,13 +458,17 @@ Byte Range | Value      | Meaning (all numbers are stored big endian)
    4 -   7 | block_count| Number of blocks covered by APM
            |            | Often some x86-harmless dummy. E.g. 0x9090 = 37008
            |            | or 0xeb02ffff = 3,942,842,367                         
-   8 -   9 | dev_type   | obscure: "device type"
-  10 -  11 | dev_id     | obscure: "device id"
-  12 -  15 | drv_data   | obscure: "driver data"
-  16 -  17 | drv_count  | obscure: "driver descriptor count"
-  18 -  81 | drv_map    | obscure: "driver descriptor table"
-           |            |          with 8 entries of 16 bytes each
-  82 - 511 | reserved   |
+   8 -   9 | dev_type   | Device type: The id of the Mac driver which is in
+           |            | charge of the storage device.
+  10 -  11 | dev_id     | Device id: Address in an i/o bus system.
+  12 -  15 | drv_data   | Driver data: Not used.
+  16 -  17 | drv_count  | Driver count: Count of entries in drv_map.
+  18 - 505 | drv_map    | Driver descriptor table:
+           |            | Up to 61 entries of 8 bytes each.
+           |            | They contain the 32 bit 512-byte LBA of the driver's
+           |            | storage location, its 16 bit size in 512-byte blocks,
+           |            | and value 0x0001.
+ 506 - 511 | reserved   |
 ---------- | ---------- | ----------------------------------------------------
 
 The SYSLINUX program isohybrid.c overwrites the first 32 bytes of this
@@ -462,9 +498,12 @@ Byte Range | Value      | Meaning (all numbers are stored big endian)
   84 -  87 | lb_count   | Logical block count (same as block_count)
   88 -  91 | flags      | Status flags
            |            | bit0= entry is valid
-           |            | bit1= entry is allocated
+           |            | bit1= partition is allocated
+           |            | bit2= partition is in use
+           |            | bit3= partition contains valid boot information
            |            | bit4= partition is readable
            |            | bit5= partition is writable
+           |            | bit7= boot code is position independent
            |            | bit30= automatic mount (legacy Mac)
   92 -  95 | boot_block | Logical start block number of boot code = 0
   96 -  99 | boot_bytes | Number of bytes in boot code = 0
@@ -528,6 +567,7 @@ Sources:
   http://mjg59.fedorapeople.org/Fedora-LiveCD.iso
   http://en.wikipedia.org/wiki/GUID_Partition_Table
   http://en.wikipedia.org/wiki/GUID
+  http://www.uefi.org/sites/default/files/resources/2_4_Errata_B.pdf
 
 
 GPT is the partition map format of EFI, a successor of PC-BIOS.
@@ -535,8 +575,11 @@ Block size is always 512. GPT consists of a header block at block address 1 and
 a partition table near the start of the medium. This is called the primary GPT.
 There is a backup copy of header and table near the end of the medium.
 
-GPT is particularly designed to co-exist with MBR. If it is present, then the
-booting firmware may or may not prefer it over the MBR partition table.
+GPT is particularly designed to co-exist with MBR. Officially only with a
+Protective MBR which covers the whole medium (except the MBR itself) by
+a single partition of type 0xee. Inofficially often with filesystem partitions
+marked in both, GPT and MBR. In the latter case the booting firmware may
+or may not prefer GPT over the MBR partition table.
 GPT can co-exist with APM if APM block size is at least 1024. In this case,
 the primary partition table will begin after the last APM entry block.
 
@@ -574,8 +617,7 @@ bit1 becomes bit30, and so on. Further it gets exored with 0xffffffff.
 
 A GUID consists of a 32-bit integer, two 16-bit integers, and an array of
 8 bytes. The integers are to be stored big-endian.
-A globally registered class of GUID are the partition type GUIDs.
-This example uses two of them
+A globally registered class of GUID are the partition type GUIDs:
 Basic data partition:  a2 a0 d0 eb , e5 b9 , 33 44 , 87 c0 68 b6 b7 26 99 c7
 HFS+ partition      :  00 53 46 48 , 00 00 , aa 11 , aa 11 00 30 65 43 ec ac
 EFI System partition:  28 73 2a c1 , 1f f8 , d2 11 , ba 4b 00 a0 c9 3e c9 3b
@@ -645,6 +687,7 @@ Sources:
       There are traces in the web which relate this to specs by
                MIPS Computer Systems, Inc. , 1985
                Silicon Graphics Computer Systems, Inc. , 2000
+   Mail conversations with Natalia Portillo.
 
 
 The first 512 bytes of the media constitute the Volume Header.
@@ -655,7 +698,11 @@ Byte Range | Value      | Meaning
    0 -   3 | 0x0be5a941 | Magic number
    4 -   5 |          0 | Root partition number
    6 -   7 |          0 | Swap partition number
-   8 -  23 |          0 | Name of file to boot (unclear what this means)
+   8 -  23 |          0 | One of the boot_name items from the Volume Directory
+           |            | may be put here to choose for booting the entry with
+           |            | that name.
+           |            | (Obviously it may be empty if only one non-zero entry
+           |            |  exists in the Volume Directory.)
            |            |
   24 -  71 | ========== | Device Parameters
            |            |
@@ -959,6 +1006,7 @@ PReP boots via a MBR partition containing only raw ELF and having type 0x41.
 Sources:
     Mail conversations with Vladimir Serbinenko.
     http://stuff.mit.edu/afs/sipb/contrib/doc/specs/protocol/chrp/chrp1_7a.pdf
+    https://www.ibm.com/developerworks/community/wikis/home?lang=en#!/wiki/W51a7ffcf4dfd_4b40_9d82_446ebc23c550/page/PowerLinux%20Boot%20howto
 
 CHRP is marked by an MBR partition entry of type 0x96 spanning the whole
 ISO 9660 image.
@@ -966,12 +1014,16 @@ ISO 9660 image.
 The specs in chrp1_7a.pdf promise that CHRP also recognizes ISO 9660 file
 systems on unpartitioned disks. (See 11.1.1. Media Layout Format)
 
+The firmware looks up a file /ppc/bootinfo.txt which in SGML-ish tag
+<boot-script> contains firmware commands.
+E.g. to execute the binary /boot/grub/powerpc.elf as first stage of GRUB2:
+  <boot-script>boot &device;:\boot\grub\powerpc.elf</boot-script> 
+
 Vladimir Serbinenko stated:
 PReP boot may be preferable. At least it can co-exist with other partitions
 in the ISO image [without causing overlapping between partitions].
 
 
-
 ------------------------------------------------------------------------------
 
                       HP-PA via PALO header version 4
@@ -1100,7 +1152,44 @@ Byte Range | Value      | Meaning
 
 ------------------------------------------------------------------------------
 
->>> ??? DEC Alpha
+                          DEC Alpha SRM boot sector
+                           for Alpha architecture
+
+Sources:
+    http://www.tldp.org/HOWTO/text/SRM-HOWTO
+      SRM Firmware Howto - Rich Payne, and David Huggins-Daines
+    cdrkit-1.1.10/genisoimage/boot-alpha.c
+      by Steve McIntyre
+      who states "Heavily inspired by isomarkboot by David Mosberger in 1996"
+    mail conversations with Helge Deller
+    
+The SRM firmware expects a Secondary Bootstrap Loader program, which usually
+is a data file of the ISO filesystem. This loader is announced by size and
+block address in the first 512 bytes of the System Area.
+SRM accepts the boot sector and executes the loader if the checksum matches.
+
+All numbers are recorded as unsigned 64 bit little endian.
+
+Boot sector components:
+
+Byte Range | Value      | Meaning
+---------- | ---------- | ----------------------------------------------------
+   0 -   ? | boot_string| genisoimage writes
+           |            | "Linux/Alpha aboot for ISO filesystem." 
+           |            | with terminating zero byte.
+           |            |
+   ? - 479 |          0 | Unused / undefined.
+           |            |
+ 480 - 487 |     length | Size of boot loader file in units of 512 bytes.
+           |            |
+ 488 - 495 |    address | LBA of the boot loader file in units of 512 bytes.
+           |            |
+ 496 - 503 |       flag | "Always 0"
+           |            |
+ 504 - 511 |   checksum | Sum of 64 bit words 0 to 63 (bytes 0 to 503).
+           |            |
+---------- | ---------- | ----------------------------------------------------
+
 
 ------------------------------------------------------------------------------
 
@@ -1220,7 +1309,7 @@ intentionally non-essential:
 They may be overwritten by other bytes which must not produce errors or
 undesirable side effects when executed as x86 machine code.
 The following 32 bytes from block 0 of an Apple Partiton Map (APM) are such
-harmless code. They stem from Fedora-LiveCD.iso by Matthew Garret:
+harmless code. They stem from Fedora-LiveCD.iso by Matthew Garrett:
    45  52  08  00  00  00  90  90  00  00  00  00  00  00  00  00
    00  00  00  00  00  00  00  00  00  00  00  00  00  00  00  00
 They do not depend on any properties of the ISO image or the information
@@ -1438,6 +1527,9 @@ Start at block 0xa4 = 164 * 512 = 41 * 2048. The VFAT image file.
 Last block is 0x0513 = 1299 = 164 + 1135. This end is correct.
 (Potential isohybrid.c bug #4:
  Wrong character set and incidential bytes in GPT partition name.)
+(Potential isohybrid.c bug #5:
+ The EFI System Partition should have type GUID :
+ "C12A7328-F81F-11D2-BA4B-00A0C93EC93B")
 
 Next entry at byte 0x02100 = 8448:
 
@@ -1466,8 +1558,8 @@ The ISO image file gets padded up to full MiB with sufficient room for the GPT
 backup which is stored near the very end of the image file. There is need for
 at least 16.5 KiB, which effectively occupy 18 KiB.
 
-The backup partition array is stored 17 KiB before the end of MBR partition 1
-resp. the image file.
+The backup partition array is stored 17 KiB before the end of MBR partition 1,
+which is also the end of the image file.
 (Potential isohybrid.c bug #1:
  Wikipedia suggests "LBA -33" counted from end. This would be 16.5 KiB before
  end.)

doc/checksums.txt

@@ -189,7 +189,7 @@ With tag type 2:
 
 Keep the original MD5 context of the data blocks and clone one for obtaining
 the MD5 bytes.
-If the MD5s match, then compute the checksum block and all folowing ones into
+If the MD5s match, then compute the checksum block and all following ones into
 the kept MD5 context and go on with reading and computing for the tree checksum
 tag. This will be found at block address next_tag, verified and parsed by:
   iso_util_decode_md5_tag(block, &tag_type, &pos,
@@ -211,7 +211,7 @@ next_tag. Go on by looking for tag type 2 and follow above prescription.
 
                       Checking the Data Part of the Session
 
-In order to check the trustworthyness of a whole session, continue reading
+In order to check the trustworthiness of a whole session, continue reading
 and checksumming after the tree was verified. 
 
 Read and checksum the blocks. When reaching block address next_tag (from the

doc/doxygen.conf.in

@@ -394,12 +394,6 @@ MAX_INITIALIZER_LINES  = 30
 
 SHOW_USED_FILES        = YES
 
-# If the sources in your project are distributed over multiple directories 
-# then setting the SHOW_DIRECTORIES tag to YES will show the directory hierarchy 
-# in the documentation. The default is NO.
-
-SHOW_DIRECTORIES       = NO
-
 # The FILE_VERSION_FILTER tag can be used to specify a program or script that 
 # doxygen should invoke to get the current version for each file (typically from the 
 # version control system). Doxygen will invoke the program by executing (via 
@@ -527,7 +521,7 @@ EXCLUDE_SYMBOLS        =
 # directories that contain example code fragments that are included (see 
 # the \include command).
 
-EXAMPLE_PATH           = test
+EXAMPLE_PATH           = demo/demo.c
 
 # If the value of the EXAMPLE_PATH tag contains directories, you can use the 
 # EXAMPLE_PATTERNS tag to specify one or more wildcard pattern (like *.cpp 
@@ -698,12 +692,6 @@ HTML_FOOTER            =
 
 HTML_STYLESHEET        = 
 
-# If the HTML_ALIGN_MEMBERS tag is set to YES, the members of classes, 
-# files or namespaces will be aligned in HTML using tables. If set to 
-# NO a bullet list will be used.
-
-HTML_ALIGN_MEMBERS     = YES
-
 # If the GENERATE_HTMLHELP tag is set to YES, additional index files 
 # will be generated that can be used as input for tools like the 
 # Microsoft HTML help workshop to generate a compressed HTML help file (.chm) 
@@ -938,18 +926,6 @@ GENERATE_XML           = NO
 
 XML_OUTPUT             = xml
 
-# The XML_SCHEMA tag can be used to specify an XML schema, 
-# which can be used by a validating XML parser to check the 
-# syntax of the XML files.
-
-XML_SCHEMA             = 
-
-# The XML_DTD tag can be used to specify an XML DTD, 
-# which can be used by a validating XML parser to check the 
-# syntax of the XML files.
-
-XML_DTD                = 
-
 # If the XML_PROGRAMLISTING tag is set to YES Doxygen will 
 # dump the program listings (including syntax highlighting 
 # and cross-referencing information) to the XML output. Note that 

doc/susp_aaip_2_0.txt

@@ -217,7 +217,7 @@ S_IRWXG. If there is ACL_USER_N or ACL_GROUP_N there must also be ACL_MASK.
 
 A numeric qualifier is a binary number of variable length up to 4 bytes. The
 Most Significant Byte comes first. The number shall be the "POSIX File User ID"
-resp. "POSIX File Group ID" as also used in RRIP PX entries. The ids of owning
+or "POSIX File Group ID" as also used in RRIP PX entries. The ids of owning
 user and owning group shall be taken from the PX entry of the file object.
 
 Optional TRANSLATE entries may associate user or group names with numeric
@@ -235,7 +235,7 @@ An eventually needed qualifier is stored in one or more Qualifier Records.
   [b] "BP 2 - Qualifier Record Head" shall be present only if QUALIFIER is set
       to 1. It shall give the number of Qualifier Bytes and eventually
       indicate that the qualifier continues in a Qualifier Record which comes
-      imediately after this record.
+      immediately after this record.
           0 to 127   Q_LENGTH, the qualifier is complete by this record
         128 to 255   Q_LENGTH+128, the qualifier is continued by next record
       So a Qualifier Record can contain at most 127 Qualifier Bytes.

doc/susp_aaip_isofs_names.txt

@@ -4,11 +4,13 @@
                       Directory of Namespace "isofs."
 
           by Thomas Schmitt    - mailto:scdbackup@gmx.net
-          Libburnia project    - mailto:libburn-hackers@pykix.org
+          Libburnia project    - mailto:bug-xorriso@gnu.org
 
 
 The following names are defined for AAIP namespace "isofs." as mentioned in
-specification of AAIP :
+specification of AAIP. Unless explicitly stated otherwise, numbers with
+names like *_LEN are 8 bit unsigned integers, those with *_BYTES are 32 bit
+unsigned integers.
 
 -------------------------------------------------------------------------------
 
@@ -22,7 +24,7 @@ Purpose:
       END is also the block address of the start of the checksum recording
       area in the image.
       See also isofs.cx .
-      This attribute shall eventually be attached to the root directory entry
+      This attribute shall be attached to the root directory entry
       and be global for the whole image.
 
 Format of Value:
@@ -65,8 +67,8 @@ Purpose:
       Records the name of the character set that was used as output character
       set when writing the RRIP name tree of the ISO 9660 image. It shall be
       suitable as parameter for function iconv_open(3).
-      This attribute shall eventually be attached to the root directory entry
-      and be global for the whole image.
+      This attribute shall be attached to the root directory entry and be
+      global for the whole image.
 
 Format of Value:
       Shall hold the character set name without terminating 0-byte.
@@ -107,6 +109,7 @@ Name:
 Purpose:
       Records .st_dev and .st_ino of struct stat of the file source in the
       local filesystem. See man 2 stat.
+      Both values may be unsigned integers up to 255 bytes.
 
 Format of Value:
       DEV_LEN | DEV_BYTES | INO_LEN | INO_BYTES
@@ -128,7 +131,7 @@ Name:
 
 Purpose:
       Records the IsoHfsplusBlessings blessing of a IsoNode as defined
-      in libisofs.h. At image load time, this info shall be converted back
+      in libisofs.h. At image load time, this info may be converted back
       into a relation between IsoImage and IsoNode so that it is available for
       the HFS+ writer when a new ISO 9660 / HFS+ image gets produced.
 
@@ -152,7 +155,7 @@ Name:
 
 Purpose:
       Records the iso_hfsplus_xinfo_data information as defined in libisofs.h.
-      At image load time, this info shall be converted back into an xinfo
+      At image load time, this info may be converted back into an xinfo
       attachment for iso_hfsplus_xinfo_func so that it is available for
       the HFS+ writer when a new ISO 9660 / HFS+ image gets produced.
 
@@ -172,6 +175,26 @@ Registered:
 
 -------------------------------------------------------------------------------
 
+Name:
+      isofs.nt
+
+Purpose:
+      Records the name truncation mode and the truncation length for Rock Ridge
+      names. See iso_image_set_truncate_mode() in libisofs.h.
+      This attribute shall be attached to the root directory entry and be
+      global for the whole image.
+
+Format of Value:
+      MODE_LEN | MODE_BYTES | LENGTH_LEN | LENGTH_BYTES
+
+Example:
+      { 1, 1, 1, 255 }
+
+Registered:
+      24 Sep 2015 by Thomas Schmitt for libisofs.
+
+-------------------------------------------------------------------------------
+
 Name:
       isofs.st
 
@@ -183,7 +206,7 @@ Purpose:
       The RRIP timestamps have a blind second during which a change after
       node registration would not be recognizable for incremental backups
       which are based in "isofs.di" rather than on content comparison.
-      This attribute shall eventually be attached to the root directory entry
+      This attribute shall be attached to the root directory entry
       and be global for the whole image.
 
 Format of Value:
@@ -204,7 +227,7 @@ This text is under
 Copyright (c) 2009 - 2011 Thomas Schmitt <scdbackup@gmx.net>
 It shall only be modified in sync with libisofs and other software which
 makes use of AAIP. Please mail change requests to mailing list
-<libburn-hackers@pykix.org> or to the copyright holder in private.
+<bug-xorriso@gnu.org> or to the copyright holder in private.
 Only if you cannot reach the copyright holder for at least one month it is
 permissible to modify this text under the same license as the affected
 copy of libisofs.

doc/zisofs_format.txt

@@ -37,7 +37,7 @@ i.e. block sizes 32 kB, 64 kB, and 128 kB. Writers must not use other sizes.
 
                                Block Pointers
 
-There are ceil(input_size / block_size) input resp. output blocks.
+There are ceil(input_size / block_size) input and output blocks.
 Each input block is of fixed size whereas the output blocks have varying
 size (down to 0). For each output block there is an offset pointer giving
 its byte address in the overall file content. The next block pointer in the
@@ -68,7 +68,7 @@ when being read.
 ZF may only be applied to files with a single extent and less than 4 GiB of
 uncompressed size.
 
-The ZF entry follows the general layout of SUSP resp. RRIP.
+The ZF entry follows the general layout of SUSP and RRIP.
 Its fields are:
 
   [1] "BP 1 to BP 2 - Signature Word" shall be (5A)(46) ("ZF").
@@ -85,19 +85,18 @@ Its fields are:
   [5] "BP 7 - Header Size Div 4" shall specify as an 8-bit number the number of
       4-byte words in the header part of the file data recorded according
       to ISO 9660:7.1.1.
-      (This is a copy of header byte 12, resp. header BP 13).
+      (This is a copy of header byte 12 / BP 13).
 
   [6] "BP 8 - Log2 of Block Size" shall specify as an 8-bit number the binary
       logarithm of the compression block size recorded according to
       ISO 9660:7.1.1.
-      (This is a copy of header byte 13, resp. header BP 14.
+      (This is a copy of header byte 13 / BP 14.
        The value has to be 15, 16 or 17 i.e. 32 kiB, 64 kiB, or 128 kiB.)
 
   [7] "BP 9 to BP 16 - Uncompressed Size" shall tell the number of uncompressed
       bytes represented by the given extent. This field shall be recorded
       according to ISO 9660:7.3.3.
-      (This number is the same as in header bytes 8 to 11, resp header BP 9
-       to BP 12.)
+      (This number is the same as in header bytes 8 to 11 / BP 9 to BP 12.)
 
   | 'Z' | 'F' | LENGTH | 1 | 'p' | 'z' | HEADER SIZE DIV 4 | LOG2 BLOCK SIZE
   | UNCOMPRESSED SIZE |

libisofs/aaip-os-dummy.c

@@ -39,7 +39,7 @@
                           It is permissibile to set them to 1 already now.
              bit8 and higher: reserved, submit 0
    @return
-        Bitfield corresponding to flag. If bits are set, th
+        Bitfield corresponding to flag.
              bit0= ACL adapter is enabled
              bit1= xattr adapter is enabled
              bit2 - bit7= Reserved for future types.
@@ -99,10 +99,14 @@ int aaip_set_acl_text(char *path, char *text, int flag)
                        -7 support of ACL not enabled at compile time
 */
 int aaip_set_attr_list(char *path, size_t num_attrs, char **names,
-                       size_t *value_lengths, char **values, int flag)
+                       size_t *value_lengths, char **values, int *errnos,
+                       int flag)
 {
  size_t i;
 
+ for(i= 0; i < num_attrs; i++)
+   errnos[i]= 0;
+
  for(i= 0; i < num_attrs; i++) {
    if(names[i] == NULL || values[i] == NULL)
  continue;

libisofs/aaip-os-freebsd.c

@@ -5,9 +5,9 @@
  Arbitrary Attribute Interchange Protocol , system adapter for getting and
  setting of ACLs and xattr.
 
- To be included by aaip_0_2.c for FreeBSD and NetBSD
+ To be included by aaip_0_2.c for FreeBSD, NetBSD, and OpenBSD
 
- Copyright (c) 2009 - 2014 Thomas Schmitt, libburnia project, GPLv2+
+ Copyright (c) 2009 - 2016 Thomas Schmitt, libburnia project, GPLv2+
 
 */
 
@@ -49,7 +49,7 @@
                           It is permissibile to set them to 1 already now.
              bit8 and higher: reserved, submit 0
    @return
-        Bitfield corresponding to flag. If bits are set, th
+        Bitfield corresponding to flag.
              bit0= ACL adapter is enabled
              bit1= xattr adapter is enabled
              bit2 - bit7= Reserved for future types.
@@ -228,6 +228,10 @@ static int aaip_extattr_make_list(char *path, int attrnamespace,
      *list_size = 0;
      return(2);
    }
+   if(errno == EPERM && attrnamespace == EXTATTR_NAMESPACE_SYSTEM) {
+     *list_size = 0;
+     return(3);
+   }
    return(0);
  }
  if(*list_size == 0)
@@ -312,6 +316,54 @@ static int aaip_extattr_make_namelist(char *path, char *attrnamespace,
  return(1);
 }
 
+
+static int get_single_attr(char *path, char *name, size_t *value_length,
+                           char **value_bytes, int flag)
+{
+ char *namept;
+ int attrnamespace;
+ ssize_t value_ret;
+ 
+ *value_bytes= NULL;
+ *value_length= 0;
+ if(strncmp(name, "user.", 5) == 0) {
+   attrnamespace= EXTATTR_NAMESPACE_USER;
+   namept= name + 5;
+ } else {
+   if(!(flag & 8))
+     return(0);
+   attrnamespace= EXTATTR_NAMESPACE_SYSTEM;
+   namept= name + 7;
+ }
+ /* Predict length of value */
+ if(flag & 32) /* follow link */
+   value_ret= extattr_get_file(path, attrnamespace, namept, NULL, (size_t) 0);
+ else
+   value_ret= extattr_get_link(path, attrnamespace, namept, NULL, (size_t) 0);
+ if(value_ret == -1)
+   return(0);
+
+ *value_bytes= calloc(value_ret + 1, 1);
+ if(*value_bytes == NULL)
+   return(-1);
+
+ /* Obtain value */
+ if(flag & 32) /* follow link */
+   value_ret= extattr_get_file(path, attrnamespace, namept,
+                               *value_bytes, (size_t) value_ret);
+ else
+   value_ret= extattr_get_link(path, attrnamespace, namept,
+                               *value_bytes, (size_t) value_ret);
+ if(value_ret == -1) {
+   free(*value_bytes);
+   *value_bytes= NULL;
+   *value_length= 0;
+   return(0);
+ }
+ *value_length= value_ret;
+ return(1);
+}
+
 #endif /* Libisofs_with_freebsd_extattR */
 
 
@@ -332,7 +384,8 @@ static int aaip_extattr_make_namelist(char *path, char *attrnamespace,
                         bit4=  do not return trivial ACL that matches st_mode
                         bit5=  in case of symbolic link: inquire link target
                         bit15= free memory of names, value_lengths, values
-   @return              >0  ok
+   @return              1   ok
+                        2   ok, no permission to inspect non-user namespaces
                         <=0 error
                         -1= out of memory
                         -2= program error with prediction of result size
@@ -342,7 +395,7 @@ int aaip_get_attr_list(char *path, size_t *num_attrs, char ***names,
                        size_t **value_lengths, char ***values, int flag)
 {
  int ret;
- ssize_t i, num_names= 0, acl_names= 0;
+ ssize_t i, num_names= 0;
 
 #ifdef Libisofs_with_aaip_acL
  unsigned char *a_acl= NULL;
@@ -350,11 +403,12 @@ int aaip_get_attr_list(char *path, size_t *num_attrs, char ***names,
  size_t a_acl_len= 0;
 #endif
 #ifdef Libisofs_with_freebsd_extattR
- char *list= NULL, *user_list= NULL, *sys_list= NULL, *namept;
- ssize_t value_ret, retry= 0, list_size= 0, user_list_size= 0;
+ char *list= NULL, *user_list= NULL, *sys_list= NULL;
+ ssize_t value_ret, list_size= 0, user_list_size= 0;
  ssize_t sys_list_size= 0;
- int attrnamespace;
+ int acl_names= 0;
 #endif
+ int no_perm_for_system= 0;
 
  if(flag & (1 << 15)) { /* Free memory */
    {ret= 1; goto ex;}
@@ -390,6 +444,8 @@ int aaip_get_attr_list(char *path, size_t *num_attrs, char ***names,
                                  &sys_list, &sys_list_size, flag & 32);
      if(ret <= 0)
        {ret= -1; goto ex;}
+     if(ret == 3)
+       no_perm_for_system= 1;
    }
 
    /* Check for NUL in names, convert into a linuxish list of namespace.name */
@@ -408,7 +464,11 @@ int aaip_get_attr_list(char *path, size_t *num_attrs, char ***names,
 #ifdef Libisofs_with_aaip_acL
  if(flag & 1) {
    num_names++;
+
+#ifdef Libisofs_with_freebsd_extattR
    acl_names= 1;
+#endif
+
  }
 #endif
  
@@ -440,45 +500,10 @@ int aaip_get_attr_list(char *path, size_t *num_attrs, char ***names,
    }
  
    for(i= 0; (size_t) i < *num_attrs; i++) {
-     if(strncmp((*names)[i], "user.", 5) == 0) {
-       attrnamespace= EXTATTR_NAMESPACE_USER;
-       namept= (*names)[i] + 5;
-     } else {
-       if(!(flag & 8))
-   continue;
-       attrnamespace= EXTATTR_NAMESPACE_SYSTEM;
-       namept= (*names)[i] + 7;
-     }
-     /* Predict length of value */
-     if(flag & 32) /* follow link */
-       value_ret= extattr_get_file(path, attrnamespace, namept,
-                                   NULL, (size_t) 0);
-     else
-       value_ret= extattr_get_link(path, attrnamespace, namept,
-                                   NULL, (size_t) 0);
-     if(value_ret == -1)
- continue;
-
-     (*values)[i]= calloc(value_ret + 1, 1);
-     if((*values)[i] == NULL)
+     value_ret= get_single_attr(path, (*names)[i], *value_lengths + i,
+                                *values + i, flag & (8 | 32));
+     if(value_ret <= 0)
        {ret= -1; goto ex;}
-
-     /* Obtain value */
-     if(flag & 32) /* follow link */
-       value_ret= extattr_get_file(path, attrnamespace, namept,
-                                   (*values)[i], (size_t) value_ret);
-     else
-       value_ret= extattr_get_link(path, attrnamespace, namept,
-                                   (*values)[i], (size_t) value_ret);
-     if(value_ret == -1) { /* there could be a race condition */
-
-       if(retry++ > 5)
-         {ret= -1; goto ex;}
-       i--;
- continue;
-     }
-     (*value_lengths)[i]= value_ret;
-     retry= 0;
    }
  }
 
@@ -489,8 +514,11 @@ int aaip_get_attr_list(char *path, size_t *num_attrs, char ***names,
  if(flag & 1) { /* Obtain ACL */
   /* access-ACL */
    aaip_get_acl_text(path, &a_acl_text, flag & (16 | 32));
-   if(a_acl_text == NULL)
-     {ret= 1; goto ex;} /* empty ACL / only st_mode info was found in ACL */
+   if(a_acl_text == NULL) {
+     /* empty ACL / only st_mode info was found in ACL */
+     ret= 1 + no_perm_for_system;
+     goto ex;
+   }
    ret= aaip_encode_acl(a_acl_text, (mode_t) 0, &a_acl_len, &a_acl, flag & 2);
    if(ret <= 0)
      goto ex;
@@ -509,7 +537,7 @@ int aaip_get_attr_list(char *path, size_t *num_attrs, char ***names,
 
 #endif /* Libisofs_with_aaip_acL */
 
- ret= 1;
+ ret= 1 + no_perm_for_system;
 ex:;
 #ifdef Libisofs_with_aaip_acL
  if(a_acl != NULL)
@@ -687,7 +715,7 @@ ex:;
                          0 no suitable ACL manipulation adapter available
                         -1  failure of system ACL service (see errno)
                         -2 attempt to manipulate ACL of a symbolic link
-                           without bit5 resp. with no suitable link target
+                           without bit5 or with no suitable link target
 */
 int aaip_set_acl_text(char *path, char *text, int flag)
 {
@@ -768,6 +796,15 @@ static int aaip_extattr_delete_names(char *path, int attrnamespace,
 #endif /* Libisofs_with_freebsd_extattR */
 
 
+static void register_errno(int *errnos, int i, int in_errno)
+{
+ if(in_errno > 0)
+   errnos[i]= in_errno;
+ else
+   errnos[i]= -1;
+}
+
+
 /* Bring the given attributes and/or ACLs into effect with the given file.
    @param flag          Bitfield for control purposes
                         bit0= decode and set ACLs
@@ -779,6 +816,8 @@ static int aaip_extattr_delete_names(char *path, int attrnamespace,
                         bit5= in case of symbolic link: manipulate link target
                         bit6= tolerate inappropriate presence or absence of
                               directory default ACL
+                        bit7= void setting a name value pair if it already
+                              exists and has the desired value.
    @return              1 success
                        -1 error memory allocation
                        -2 error with decoding of ACL
@@ -791,17 +830,23 @@ static int aaip_extattr_delete_names(char *path, int attrnamespace,
     ISO_AAIP_ACL_MULT_OBJ multiple entries of user::, group::, other::
 */
 int aaip_set_attr_list(char *path, size_t num_attrs, char **names,
-                       size_t *value_lengths, char **values, int flag)
+                       size_t *value_lengths, char **values,
+                       int *errnos,  int flag)
 {
- int ret, has_default_acl= 0;
+ int ret, has_default_acl= 0, end_ret= 1;
  size_t i, consumed, acl_text_fill, acl_idx= 0;
  char *acl_text= NULL;
 #ifdef Libisofs_with_freebsd_extattR
- char *user_list= NULL, *sys_list= NULL, *namept;
- ssize_t user_list_size= 0, sys_list_size= 0;
+ char *user_list= NULL, *sys_list= NULL, *namept, *old_value;
+ ssize_t user_list_size= 0, sys_list_size= 0, value_ret;
  int attrnamespace;
+ size_t old_value_l;
+ int skip;
 #endif
 
+ for(i= 0; i < num_attrs; i++)
+   errnos[i]= 0;
+
 #ifdef Libisofs_with_freebsd_extattR
 
  if(flag & 2) { /* Delete all file attributes */
@@ -850,16 +895,35 @@ int aaip_set_attr_list(char *path, size_t num_attrs, char **names,
      attrnamespace= EXTATTR_NAMESPACE_SYSTEM;
      namept= names[i] + 7;
    } else {
-     {ret= -8; goto ex;}
+     register_errno(errnos, i, (int) EFAULT);
+     end_ret= -8;
+ continue;
+   }
+   skip= 0;
+   if(flag & 128) {
+     value_ret= get_single_attr(path, names[i], &old_value_l,
+                                &old_value, flag & (8 | 32));
+     if(value_ret > 0 && old_value_l == value_lengths[i]) {
+       if(memcmp(old_value, values[i], value_lengths[i]) == 0)
+         skip= 1;
+     }
+     if(old_value != NULL)
+       free(old_value);
+   }
+   if(!skip) {
+     if(flag & 32)
+       ret= extattr_set_file(path, attrnamespace, namept,
+                             values[i], value_lengths[i]);
+     else
+       ret= extattr_set_link(path, attrnamespace, namept,
+                             values[i], value_lengths[i]);
+     if(ret == -1) {
+       register_errno(errnos, i, errno);
+       if(end_ret != 1)
+         end_ret= -4;
+ continue;
+     }
    }
-   if(flag & 32)
-     ret= extattr_set_file(path, attrnamespace, namept,
-                           values[i], value_lengths[i]);
-   else
-     ret= extattr_set_link(path, attrnamespace, namept,
-                           values[i], value_lengths[i]);
-   if(ret == -1)
-     {ret= -4; goto ex;}
 
 #else
 
@@ -874,8 +938,12 @@ int aaip_set_attr_list(char *path, size_t num_attrs, char **names,
  }
 
  /* Decode ACLs */
+ /* It is important that this happens after restoring xattr which might be
+    representations of ACL, too. If isofs ACL are enabled, then they shall
+    override the xattr ones.
+ */
  if(acl_idx == 0)
-   {ret= 1; goto ex;}
+   {ret= end_ret; goto ex;}
  i= acl_idx - 1; 
 
  ret= aaip_decode_acl((unsigned char *) values[i], value_lengths[i],
@@ -897,6 +965,8 @@ int aaip_set_attr_list(char *path, size_t num_attrs, char **names,
 
 #ifdef Libisofs_with_aaip_acL
  ret= aaip_set_acl_text(path, acl_text, flag & (32 | 64));
+ if(ret == -1)
+    register_errno(errnos, i, errno);
  if(ret <= 0)
    {ret= -3; goto ex;}
 #else
@@ -906,7 +976,7 @@ int aaip_set_attr_list(char *path, size_t num_attrs, char **names,
  if(has_default_acl && !(flag & 64))
    {ret= -3; goto ex;}
 
- ret= 1;
+ ret= end_ret;
 ex:;
  if(acl_text != NULL)
    free(acl_text);

libisofs/aaip-os-linux.c

@@ -7,7 +7,7 @@
 
  To be included by aaip_0_2.c for Linux
 
- Copyright (c) 2009 - 2011 Thomas Schmitt, libburnia project, GPLv2+
+ Copyright (c) 2009 - 2018 Thomas Schmitt, libburnia project, GPLv2+
 
 */
 
@@ -30,8 +30,12 @@
 #endif
 
 #ifdef Libisofs_with_aaip_xattR
+#ifdef Libisofs_with_sys_xattR
+#include <sys/xattr.h>
+#else
 #include <attr/xattr.h>
 #endif
+#endif
 
 
 /* ------------------------------ Inquiry --------------------------------- */
@@ -89,7 +93,7 @@ int aaip_local_attr_support(int flag)
                             or filesystem does not support ACL
                          -1 failure of system ACL service (see errno)
                          -2 attempt to inquire ACL of a symbolic link without
-                            bit4 or bit5 resp. with no suitable link target
+                            bit4 or bit5 or with no suitable link target
 */
 int aaip_get_acl_text(char *path, char **text, int flag)
 {
@@ -159,6 +163,41 @@ int aaip_get_acl_text(char *path, char **text, int flag)
 }
 
 
+#ifdef Libisofs_with_aaip_xattR
+
+static int get_single_attr(char *path, char *name, size_t *value_length,
+                           char **value_bytes, int flag)
+{
+ ssize_t value_ret;
+
+ *value_bytes= NULL;
+ *value_length= 0;
+ if(flag & 32)
+   value_ret= getxattr(path, name, NULL, 0);
+ else
+   value_ret= lgetxattr(path, name, NULL, 0);
+ if(value_ret == -1)
+   return(0);
+ *value_bytes= calloc(value_ret + 1, 1);
+ if(*value_bytes == NULL)
+   return(-1);
+ if(flag & 32)
+   value_ret= getxattr(path, name, *value_bytes, value_ret);
+ else
+   value_ret= lgetxattr(path, name, *value_bytes, value_ret);
+ if(value_ret == -1) {
+   free(*value_bytes);
+   *value_bytes= NULL;
+   *value_length= 0;
+   return(0);
+ }
+ *value_length= value_ret;
+ return(1);
+}
+
+#endif /* Libisofs_with_aaip_xattR */
+
+
 /* Obtain the Extended Attributes and/or the ACLs of the given file in a form
    that is ready for aaip_encode().
    @param path          Path to the file
@@ -176,7 +215,9 @@ int aaip_get_acl_text(char *path, char **text, int flag)
                         bit4=  do not return trivial ACL that matches st_mode
                         bit5=  in case of symbolic link: inquire link target
                         bit15= free memory of names, value_lengths, values
-   @return              >0  ok
+   @return              1  ok
+                        (reserved for FreeBSD: 2 ok, no permission to inspect
+                                                 non-user namespaces.)
                         <=0 error
                         -1= out of memory
                         -2= program error with prediction of result size
@@ -186,16 +227,20 @@ int aaip_get_attr_list(char *path, size_t *num_attrs, char ***names,
                        size_t **value_lengths, char ***values, int flag)
 {
  int ret;
- ssize_t i, num_names= 0;
 
 #ifdef Libisofs_with_aaip_acL
  unsigned char *acl= NULL;
  char *a_acl_text= NULL, *d_acl_text= NULL;
  size_t acl_len= 0;
+#define Libisofs_aaip_get_attr_activE yes
 #endif
 #ifdef Libisofs_with_aaip_xattR
  char *list= NULL;
- ssize_t value_ret, retry= 0, list_size= 0;
+ ssize_t value_ret, list_size= 0;
+#define Libisofs_aaip_get_attr_activE yes
+#endif
+#ifdef Libisofs_aaip_get_attr_activE
+ ssize_t i, num_names= 0;
 #endif
 
  if(flag & (1 << 15)) { /* Free memory */
@@ -207,6 +252,14 @@ int aaip_get_attr_list(char *path, size_t *num_attrs, char ***names,
  *value_lengths= NULL;
  *values= NULL;
 
+#ifndef Libisofs_aaip_get_attr_activE
+
+ ret = 1;
+ex:;
+ return ret;
+
+#else /* Libisofs_aaip_get_attr_activE */
+
  /* Set up arrays */
 
 #ifdef Libisofs_with_aaip_xattR
@@ -276,27 +329,10 @@ int aaip_get_attr_list(char *path, size_t *num_attrs, char ***names,
      if(!(flag & 8))
        if(strncmp((*names)[i], "user.", 5))
    continue;
-     if(flag & 32)
-       value_ret= getxattr(path, (*names)[i], NULL, 0);
-     else
-       value_ret= lgetxattr(path, (*names)[i], NULL, 0);
-     if(value_ret == -1)
- continue;
-     (*values)[i]= calloc(value_ret + 1, 1);
-     if((*values)[i] == NULL)
+     value_ret= get_single_attr(path, (*names)[i], *value_lengths + i,
+                                *values + i, flag & 32);
+     if(value_ret <= 0)
        {ret= -1; goto ex;}
-     if(flag & 32)
-       value_ret= getxattr(path, (*names)[i], (*values)[i], value_ret);
-     else
-       value_ret= lgetxattr(path, (*names)[i], (*values)[i], value_ret);
-     if(value_ret == -1) { /* there could be a race condition */
-       if(retry++ > 5)
-         {ret= -1; goto ex;}
-       i--;
- continue;
-     }
-     (*value_lengths)[i]= value_ret;
-     retry= 0;
    }
  }
 
@@ -361,6 +397,9 @@ ex:;
    *num_attrs= 0;
  }
  return(ret);
+
+#endif /* Libisofs_aaip_get_attr_activE */
+
 }
 
 
@@ -377,7 +416,7 @@ ex:;
                          0 ACL support not enabled at compile time
                         -1 failure of system ACL service (see errno)
                         -2 attempt to manipulate ACL of a symbolic link
-                           without bit5 resp. with no suitable link target
+                           without bit5 or with no suitable link target
 */
 int aaip_set_acl_text(char *path, char *text, int flag)
 {
@@ -419,6 +458,15 @@ ex:
 }
 
 
+static void register_errno(int *errnos, int i)
+{
+ if(errno > 0)
+   errnos[i]= errno;
+ else
+   errnos[i]= -1;
+}
+
+
 /* Bring the given attributes and/or ACLs into effect with the given file.
    @param flag          Bitfield for control purposes
                         bit0= decode and set ACLs
@@ -428,8 +476,10 @@ ex:
                               I.e. those with a name which does not begin
                               by "user."
                         bit5= in case of symbolic link: manipulate link target
-                        bit6= tolerate inappropriate presence or absense of
+                        bit6= tolerate inappropriate presence or absence of
                               directory default ACL
+                        bit7= avoid setting a name value pair if it already
+                              exists and has the desired value.
    @return              1 success
                        -1 error memory allocation
                        -2 error with decoding of ACL
@@ -442,14 +492,25 @@ ex:
     ISO_AAIP_ACL_MULT_OBJ multiple entries of user::, group::, other::
 */
 int aaip_set_attr_list(char *path, size_t num_attrs, char **names,
-                       size_t *value_lengths, char **values, int flag)
+                       size_t *value_lengths, char **values,
+                       int *errnos, int flag)
 {
- int ret, has_default_acl= 0;
- size_t i, consumed, acl_text_fill, acl_idx= 0, h_consumed;
- char *acl_text= NULL, *list= NULL;
+ int ret, end_ret= 1;
+ size_t i, consumed, acl_text_fill, acl_idx= 0;
+ char *acl_text= NULL;
 #ifdef Libisofs_with_aaip_xattR
- ssize_t list_size= 0;
+ char *list= NULL, *old_value;
+ ssize_t list_size= 0, value_ret;
+ size_t old_value_l;
+ int skip;
 #endif
+#ifdef Libisofs_with_aaip_acL
+ size_t h_consumed;
+ int has_default_acl= 0;
+#endif
+
+ for(i= 0; i < num_attrs; i++)
+   errnos[i]= 0;
 
 #ifdef Libisofs_with_aaip_xattR
 
@@ -504,12 +565,28 @@ int aaip_set_attr_list(char *path, size_t num_attrs, char **names,
 
 #ifdef Libisofs_with_aaip_xattR
 
-   if(flag & 32)
-     ret= setxattr(path, names[i], values[i], value_lengths[i], 0);
-   else
-     ret= lsetxattr(path, names[i], values[i], value_lengths[i], 0);
-   if(ret == -1)
-     {ret= -4; goto ex;}
+   skip= 0;
+   if(flag & 128) {
+     value_ret= get_single_attr(path, names[i], &old_value_l,
+                                &old_value, flag & 32);
+     if(value_ret > 0 && old_value_l == value_lengths[i]) {
+       if(memcmp(old_value, values[i], value_lengths[i]) == 0)
+         skip= 1;
+     }
+     if(old_value != NULL)
+       free(old_value);
+   }
+   if(!skip) {
+     if(flag & 32)
+       ret= setxattr(path, names[i], values[i], value_lengths[i], 0);
+     else
+       ret= lsetxattr(path, names[i], values[i], value_lengths[i], 0);
+     if(ret == -1) {
+       register_errno(errnos, i);
+       end_ret= -4;
+ continue;
+     }
+   }
 
 #else
 
@@ -519,9 +596,13 @@ int aaip_set_attr_list(char *path, size_t num_attrs, char **names,
 
  }
 
-/* Decode ACLs */
+ /* Decode ACLs */
+ /* It is important that this happens after restoring xattr which might be
+    representations of ACL, too. If isofs ACL are enabled, then they shall
+    override the xattr ones.
+ */
  if(acl_idx == 0)
-   {ret= 1; goto ex;}
+   {ret= end_ret; goto ex;}
  i= acl_idx - 1;
                                                              /* "access" ACL */
  ret= aaip_decode_acl((unsigned char *) values[i], value_lengths[i],
@@ -539,15 +620,16 @@ int aaip_set_attr_list(char *path, size_t num_attrs, char **names,
    goto ex;
  if(ret <= 0)
    {ret= -2; goto ex;}
- has_default_acl= (ret == 2);
 
 #ifdef Libisofs_with_aaip_acL
+
+ has_default_acl= (ret == 2);
+
  ret= aaip_set_acl_text(path, acl_text, flag & 32);
+ if(ret == -1)
+   register_errno(errnos, i);
  if(ret <= 0)
    {ret= -3; goto ex;}
-#else
- {ret= -7; goto ex;}
-#endif
                                                             /* "default" ACL */
  if(has_default_acl) {
    free(acl_text);
@@ -570,6 +652,8 @@ int aaip_set_attr_list(char *path, size_t num_attrs, char **names,
    if(ret <= 0)
      {ret= -2; goto ex;}
    ret= aaip_set_acl_text(path, acl_text, 1 | (flag & 32));
+   if(ret == -1)
+     register_errno(errnos, i);
    if(ret <= 0)
      {ret= -3; goto ex;}
  } else {
@@ -581,12 +665,23 @@ int aaip_set_attr_list(char *path, size_t num_attrs, char **names,
 
    }
  }
- ret= 1;
+ ret= end_ret;
+
+#else
+
+ ret= -7;
+
+#endif /* !Libisofs_with_aaip_acL */
+
 ex:;
  if(acl_text != NULL)
    free(acl_text);
+
+#ifdef Libisofs_with_aaip_xattR
  if(list != NULL)
    free(list);
+#endif
+
  return(ret);
 }
 

libisofs/aaip_0_2.c

@@ -7,7 +7,7 @@
  See libisofs/aaip_0_2.h
      http://libburnia-project.org/wiki/AAIP
 
- Copyright (c) 2009 - 2014 Thomas Schmitt, libburnia project, GPLv2+
+ Copyright (c) 2009 - 2016 Thomas Schmitt, libburnia project, GPLv2+
 
 */
 
@@ -62,12 +62,12 @@
 #define Aaip_namespace_trusteD   0x05
 #define Aaip_namespace_securitY  0x06
 
-static char Aaip_namespace_textS[][16]= {"", "", "system.", "user.", "isofs.",
-                                         "trusted.", "security."};
-
 /* maximum expansion:  "security." */
 #define Aaip_max_name_expansioN  9
 
+static char Aaip_namespace_textS[][Aaip_max_name_expansioN + 1]=
+               {"", "", "system.", "user.", "isofs.", "trusted.", "security."};
+
 /* --------------------------------- Encoder ---------------------------- */
 
 
@@ -88,26 +88,30 @@ static int aaip_encode_pair(char *name, size_t attr_length, char *attr,
                         no longer needed 
    @param flag          Bitfield for control purposes
                         bit0= set CONTINUE bit of last AAIP field to 1
-   @return              >0 is the number of SUSP fields generated,
-                        0 means error 
+   @return              >= 0 is the number of SUSP fields generated,
+                        < 0 means error 
 */
-size_t aaip_encode(size_t num_attrs, char **names,
-                   size_t *value_lengths, char **values, 
-                   size_t *result_len, unsigned char **result, int flag)
+ssize_t aaip_encode(size_t num_attrs, char **names,
+                    size_t *value_lengths, char **values, 
+                    size_t *result_len, unsigned char **result, int flag)
 {
- size_t mem_size= 0, comp_size, ret;
+ size_t mem_size= 0, comp_size;
+ ssize_t ret;
  unsigned int number_of_fields, i, num_recs;
 
  /* Predict memory needs, number of SUSP fields and component records */
+ *result = NULL;
  *result_len= 0;
  for(i= 0; i < num_attrs; i++) {
    ret= aaip_encode_pair(names[i], value_lengths[i], values[i],
                          &num_recs, &comp_size, NULL, (size_t) 0, 1);
-   if(ret <= 0)
+   if(ret < 0)
      return(ret);
    mem_size+= comp_size;
  }
  number_of_fields= mem_size / 250 + !!(mem_size % 250);
+ if(number_of_fields == 0)
+   return(0);
  mem_size+= number_of_fields * 5;
 
 #ifdef Aaip_encode_debuG
@@ -118,14 +122,18 @@ size_t aaip_encode(size_t num_attrs, char **names,
 #endif
 
  if(*result == NULL)
-   return 0;
+   return ISO_OUT_OF_MEM;
 
  /* Encode pairs into result */
  for(i= 0; i < num_attrs; i++) {
    ret= aaip_encode_pair(names[i], value_lengths[i], values[i],
                          &num_recs, &comp_size, *result, *result_len, 0);
-   if(ret <= 0)
+   if(ret < 0) {
+     free(*result);
+     *result = NULL;
+     *result_len = 0;
      return(ret);
+   }
    (*result_len)+= comp_size;
  }
 
@@ -404,7 +412,8 @@ static ssize_t aaip_encode_acl_text(char *acl_text, mode_t st_mode,
          /* >>> Duplicate u:: entry. */;
          /* >>> ??? If it matches the previous one: ignore */
 
-         return((int) ISO_AAIP_ACL_MULT_OBJ);
+         ret = ISO_AAIP_ACL_MULT_OBJ;
+         goto ex;
        }
        has_u++;
      } else {
@@ -449,7 +458,8 @@ static ssize_t aaip_encode_acl_text(char *acl_text, mode_t st_mode,
          /* >>> Duplicate g:: entry. */;
          /* >>> ??? If it matches the previous one: ignore */
 
-         return((int) ISO_AAIP_ACL_MULT_OBJ);
+         ret = ISO_AAIP_ACL_MULT_OBJ;
+         goto ex;
        }
        has_g++;
      } else {
@@ -493,7 +503,8 @@ static ssize_t aaip_encode_acl_text(char *acl_text, mode_t st_mode,
        /* >>> Duplicate o:: entry. */;
        /* >>> ??? If it matches the previous one: ignore */
 
-       return((int) ISO_AAIP_ACL_MULT_OBJ);
+       ret = ISO_AAIP_ACL_MULT_OBJ;
+       goto ex;
      }
      has_o++;
    } else if(strncmp(rpt, "mask:", 5) == 0) {
@@ -637,7 +648,7 @@ ex:;
      mission bits.
 
 */
-/* Analyze occurence of ACL tag types in long text form. If not disabled by
+/* Analyze occurrence of ACL tag types in long text form. If not disabled by
    parameter flag remove the entries of type "user::" , "group::" , "other::" ,
    or "other:" from an ACL in long text form if they match the bits in st_mode
    as described by man 2 stat and man 5 acl.
@@ -1803,8 +1814,8 @@ int aaip_decode_attrs(struct aaip_state **handle,
    if(aaip->list_mem_used >= memory_limit)
      return(3);
    aaip->list_mem_used+= new_mem;
-   aaip->name_buf= calloc(sizeof(char *), Aaip_initial_name_leN);
-   aaip->value_buf= calloc(sizeof(char *), Aaip_initial_value_leN);
+   aaip->name_buf= calloc(1, Aaip_initial_name_leN);
+   aaip->value_buf= calloc(1, Aaip_initial_value_leN);
    if(aaip->name_buf == NULL || aaip->value_buf == NULL)
      return(-1);
    aaip->name_buf_size= Aaip_initial_name_leN;
@@ -2177,15 +2188,41 @@ ex:;
 /* ----------------------- Adapter for operating systems ----------------- */
 
 
+#ifdef Libisofs_use_os_dummY
+
+#include "aaip-os-dummy.c"
+
+#else
 #ifdef __FreeBSD__
 
 #include "aaip-os-freebsd.c"
 
+#else
+#ifdef __FreeBSD_kernel__
+
+#ifdef NIX
+#ifdef Libisofs_with_aaip_xattR
+/* ts B51213: xattr system library calls are only stubs */
+#include "aaip-os-linux.c"
+#else
+/* ts B51213: extattr system library calls are not even present */
+#include "aaip-os-freebsd.c"
+#endif /* ! Libisofs_with_aaip_xattR */
+#else /* NIX */
+/* ts B51213: so we still end up at the dummy */
+#include "aaip-os-dummy.c"
+#endif /* ! NIX */
+
 #else
 #ifdef __NetBSD__
 
 #include "aaip-os-freebsd.c"
 
+#else
+#ifdef __OpenBSD__
+
+#include "aaip-os-freebsd.c"
+
 #else
 #ifdef __linux
 
@@ -2203,6 +2240,9 @@ ex:;
 #include "aaip-os-dummy.c"
 
 #endif /* ! __linux */
+#endif /* ! __OpenBSD__ */
 #endif /* ! __NetBSD__ */
+#endif /* ! __FreeBSD_kernel__ */
 #endif /* ! __FreeBSD__ */
+#endif /* ! Libisofs_use_os_dummY */
 

libisofs/aaip_0_2.h

@@ -9,13 +9,16 @@
 
  test/aaip_0_2.h - Public declarations
 
- Copyright (c) 2009 Thomas Schmitt, libburnia project, GPLv2+
+ Copyright (c) 2009 - 2016 Thomas Schmitt, libburnia project, GPLv2+
 
 */
 
 #ifndef Aaip_h_is_includeD
 #define Aaip_h_is_includeD yes
 
+/* For ssize_t */
+#include <unistd.h>
+
 
 /* --------------------------------- Encoder ---------------------------- */
 
@@ -30,12 +33,12 @@
                         no longer needed 
    @param flag          Bitfield for control purposes
                         bit0= set CONTINUE bit of last AAIP field to 1
-   @return              >0 is the number of SUSP fields generated,
-                        0 means error 
+   @return              >= 0 is the number of SUSP fields generated,
+                        < 0 means error
 */
-size_t aaip_encode(size_t num_attrs, char **names,
-                   size_t *value_lengths, char **values, 
-                   size_t *result_len, unsigned char **result, int flag);
+ssize_t aaip_encode(size_t num_attrs, char **names,
+                    size_t *value_lengths, char **values, 
+                    size_t *result_len, unsigned char **result, int flag);
 
 
 /* ------ ACL representation ------ */
@@ -90,7 +93,7 @@ int aaip_encode_both_acl(char *a_acl_text, char *d_acl_text, mode_t st_mode,
                          size_t *result_len, unsigned char **result, int flag);
 
 
-/* Analyze occurence of ACL tag types in long text form. If not disabled by
+/* Analyze occurrence of ACL tag types in long text form. If not disabled by
    parameter flag remove the entries of type "user::" , "group::" , "other::" ,
    or "other:" from an ACL in long text form if they match the bits in st_mode
    as described by man 2 stat and man 5 acl.
@@ -504,7 +507,8 @@ int aaip_set_acl_text(char *path, char *text, int flag);
     ISO_AAIP_ACL_MULT_OBJ multiple entries of user::, group::, other::
 */
 int aaip_set_attr_list(char *path, size_t num_attrs, char **names,
-                       size_t *value_lengths, char **values, int flag);
+                       size_t *value_lengths, char **values,
+                       int *errnos, int flag);
 
 #endif /* ! Aaip_h_is_includeD */
 

libisofs/builder.c

@@ -1,6 +1,6 @@
 /*
  * Copyright (c) 2007 Vreixo Formoso
- * Copyright (c) 2009 - 2014 Thomas Schmitt
+ * Copyright (c) 2009 - 2015 Thomas Schmitt
  * 
  * This file is part of the libisofs project; you can redistribute it and/or 
  * modify it under the terms of the GNU General Public License version 2 
@@ -21,6 +21,7 @@
 #include "image.h"
 #include "aaip_0_2.h"
 #include "util.h"
+#include "messages.h"
 
 #include <stdlib.h>
 #include <string.h>
@@ -72,8 +73,15 @@ int default_create_file(IsoNodeBuilder *builder, IsoImage *image,
     iso_file_source_ref(src);
     
     name = iso_file_source_get_name(src);
-    if (strlen(name) > LIBISOFS_NODE_NAME_MAX)
-        name[LIBISOFS_NODE_NAME_MAX] = 0;
+    if ((int) strlen(name) > image->truncate_length) {
+        ret = iso_truncate_rr_name(image->truncate_mode,
+                                   image->truncate_length, name, 0);
+        if (ret < 0) {
+            iso_stream_unref(stream);
+            free(name);
+            return ret;
+        }
+    }
     ret = iso_node_new_file(name, stream, &node);
     if (ret < 0) {
         iso_stream_unref(stream);
@@ -98,11 +106,11 @@ static
 int default_create_node(IsoNodeBuilder *builder, IsoImage *image,
                         IsoFileSource *src, char *in_name, IsoNode **node)
 {
-    int ret;
+    int ret, name_is_attached = 0;
     struct stat info;
     IsoNode *new;
     IsoFilesystem *fs;
-    char *name;
+    char *name = NULL;
     unsigned char *aa_string = NULL;
     char *a_text = NULL, *d_text = NULL;
     char *dest = NULL;
@@ -131,8 +139,12 @@ int default_create_node(IsoNodeBuilder *builder, IsoImage *image,
         }
     }
 
-    if (strlen(name) > LIBISOFS_NODE_NAME_MAX)
-        name[LIBISOFS_NODE_NAME_MAX] = 0;
+    if ((int) strlen(name) > image->truncate_length) {
+        ret = iso_truncate_rr_name(image->truncate_mode,
+                                   image->truncate_length, name, 0);
+        if (ret < 0)
+            goto ex;
+    }
     fs = iso_file_source_get_filesystem(src);
     new = NULL;
 
@@ -203,12 +215,14 @@ int default_create_node(IsoNodeBuilder *builder, IsoImage *image,
             }
         }
         break;
-    }
-    
-    if (ret < 0) {
-        free(name);
+    default:
+        ret = ISO_BAD_FSRC_FILETYPE;
         goto ex;
     }
+    
+    if (ret < 0)
+        goto ex;
+    name_is_attached = 1;
 
     /* fill fields */
     iso_node_set_perms_internal(new, info.st_mode, 1);
@@ -222,9 +236,9 @@ int default_create_node(IsoNodeBuilder *builder, IsoImage *image,
     /* Eventually set S_IRWXG from ACL */
     if (image->builder_ignore_acl) {
         ret = iso_file_source_get_aa_string(src, &aa_string, 4);
-        if (aa_string != NULL)
+        if (ret >= 0 && aa_string != NULL)
             iso_aa_get_acl_text(aa_string, info.st_mode, &a_text, &d_text, 16);
-        if (a_text != NULL) {
+        if (ret >= 0 && a_text != NULL) {
             aaip_cleanout_st_mode(a_text, &(info.st_mode), 4 | 16);
             iso_node_set_perms_internal(new, info.st_mode, 1);
         }
@@ -238,8 +252,11 @@ int default_create_node(IsoNodeBuilder *builder, IsoImage *image,
     /* Obtain ownership of eventual AAIP string */
     ret = iso_file_source_get_aa_string(src, &aa_string,
             1 | (image->builder_ignore_acl << 1) |
-                (image->builder_ignore_ea << 2 ));
-    if (ret == 1 && aa_string != NULL) {
+                (image->builder_ignore_ea << 2) |
+                (image->builder_take_all_ea << 3));
+    if(ret == 2)
+      image->blind_on_local_get_attrs = 1;
+    if (ret > 0 && aa_string != NULL) {
         ret = iso_node_add_xinfo(new, aaip_xinfo_func, aa_string);
         if (ret < 0)
             goto ex;
@@ -251,6 +268,8 @@ int default_create_node(IsoNodeBuilder *builder, IsoImage *image,
 
     ret = ISO_SUCCESS;
 ex:;
+    if (name != NULL && !name_is_attached)
+        free(name);
     LIBISO_FREE_MEM(dest);
     return ret;
 }

libisofs/changelog.txt

@@ -0,0 +1,366 @@
+------------------------------------------------------------------------------
+       libburnia-project.org / dev.lovelyhq.com    libisofs
+------------------------------------------------------------------------------
+
+------------------------------------------------------------------------------
+                                  Changelog
+------------------------------------------------------------------------------
+
+13 Nov 2016  [8bf32d8] [1340]
+doc/susp_aaip_isofs_names.txt
+Prepared some of the documentation for move to dev.lovelyhq.com
+
+13 Nov 2016  [c51efce] [1341]
+libisofs/fs_image.c
+Bug fix: iso_read_opts_set_no_rockridge() did not prevent reading of root SUSP
+
+13 Nov 2016  [dc6cd94] [1342]
+libisofs/libisofs.h
+libisofs/eltorito.c
+libisofs/libisofs.ver
+New API calls el_torito_set_full_load(), el_torito_get_full_load()
+
+13 Nov 2016  [ed209e0] [1343]
+COPYRIGHT
+Mentioned Vladimir Serbinenko in libisofs copyright list
+
+13 Nov 2016  [2961bde] [1344]
+ChangeLog
++ libisofs/changelog.txt
+Updated change log
+
+23 Nov 2016  [01020ef] [1345]
+libisofs/eltorito.h
+Committed missing part of rev dc6cd94/1342
+
+23 Nov 2016  [8ec75ee] [1346]
+libisofs/fs_image.c
+libisofs/rockridge.h
+libisofs/rockridge_read.c
+Bug fix: Non-SUSP data in System Use Area prevented image loading if Rock Ridge was enabled. Thanks to Jonathan Dowland.
+
+25 Dec 2016  [76181d0] [1347]
+libisofs/ecma119.c
+Restricted volume size of PVD with non-zero partition offset to filesystem
+size even if the first PVD claims the whole output size
+
+25 Dec 2016  [afb10aa] [1348]
+libisofs/eltorito.c
+Claiming full output size in first PVD if partition offset is non-zero
+
+25 Dec 2016  [3043b5f] [1349]
+libisofs/fs_image.c
+Enabled recognition of partition offset of grub-mkrescue-sed.sh mode "gpt_appended"
+
+03 Jan 2017  [2152804] [1350]
+libisofs/system_area.c
+Bug fix: Protective MBR for GPT could emerge with boot flag set.
+
+24 Jan 2014  [5c1c5cd] [1351]
+libisofs/system_area.c
+Bug fix: Appended partitions of size >= 4 GiB led to abort with error message
+         "FATAL : ISO overwrite". Thanks to Sven Haardiek.
+
+24 Jan 2014  [094b3f7] [1352]
+libisofs/system_area.c
+Updated copyright year in system_area.c
+
+27 Feb 2014  [e66b9bf] [1353]
+libisofs/libisofs.h
+libisofs/ecma119.h
+libisofs/ecma119.c
+libisofs/system_area.c
+libisofs/libisofs.ver
+New API call iso_write_opts_set_iso_mbr_part_type()
+
+27 Feb 2014  [5600f3d] [1354]
+libisofs/system_area.c
+When deciding boot flag, consider MBR partition slot empty only if entirely 0
+
+19 Mar 2017  [86f6ffc] [1355]
+libisofs/system_area.c
+Let iso_mbr_part_type 0xee override ban on 0xee without GPT
+
+09 Apr 2017  [94e4bfb] [1356]
+libisofs/fs_image.c
+Adapted recognizing of partition offset to the changes of rev afb10aa / 1348
+
+10 Apr 2017  [0e7300b] [1357]
+doc/boot_sectors.txt
+Clarified meaning of MBR partition boot flag
+
+25 Apr 2017  [fb86970] [1358]
+libisofs/system_area.c
+Reacted on harmless compiler warning about uninitialized variable.
+
+03 Jun 2017  [d7737e3] []
+libisofs/fs_image.c
+Removed ban on reading El Torito platform ids other than 0 and 0xef
+
+29 Jun 2017  [6282bbc]
+libisofs/system_area.c
+Bug fix: Bit 15 of iso_write_opts_set_system_area did not work with generic MBR
+
+01 Jul 2017  [18ab601]
+libisofs/eltorito.c
+libisofs/system_area.c
+Let ISO size cover appended partitions if --protective-msdos-label or nonzero -partition_offset is given
+
+13 Aug 2017  [c6e4035]
+doc/boot_sectors.txt
+Added boot sector knowledge gained from Natalia Portillo
+
+13 Aug 2017  [e19a338]
+doc/boot_sectors.txt
+Re-added two empty lines which were lost by previous commit
+
+14 Aug 2014  [1e40ed3]
+libisofs/fs_image.c
+Bug fix: Keeping and patching of loaded boot images failed. Regression by version 1.4.4.
+
+14 Aug 2017  [280108d]
+ChangeLog
+libisofs/changelog.txt
+Updated change log
+
+17 Aug 2017  [860a91d]
+libisofs/libisofs.h
+libisofs/fs_image.c
+libisofs/messages.c
+Preventing NULL dereference if root directory bears a RRIP RE entry
+
+18 Aug 2017  [36c8800]
+libisofs/rockridge_read.c
+Preventing buffer underread with empty RRIP SL component.
+Debian bug 872475. Thanks Jakub Wilk and American Fuzzy Lop.
+
+18 Aug 2017  [16bde11]
+libisofs/fs_image.c
+Preventing memory leak caused by RRIP SL entry without PX entry that marks
+the file as symbolic link
+
+18 Aug 2017  [661b68c]
+libisofs/rockridge_read.c
+Preventing buffer overflow with AAIP AL entry of insufficient size.
+Debian bug 872545. Thanks Jakub Wilk and American Fuzzy Lop.
+
+19 Aug 2017  [91490d5]
+libisofs/libisofs.h
+libisofs/messages.c
+libisofs/rockridge_read.c
+Preventing use of zero sized SUSP CE entry which causes SIGSEGV.
+Debian bug 872590. Thanks Jakub Wilk and American Fuzzy Lop.
+
+19 Aug 2017  [31088d9]
+libisofs/rockridge_read.c
+Avoid to read blocks from start of CE area which do not belong to the given file
+
+19 Aug 2017  [2a64d89]
+libisofs/libisofs.h
+libisofs/rockridge.h
+libisofs/fs_image.c
+libisofs/rockridge_read.c
+libisofs/messages.c
+Refuse to read CE data blocks from after the end of ISO filesystem
+
+21 Aug 2017  [a7152f5]
+libisofs/fs_image.c
+Correcting previous commit for supporting multi-session
+
+21 Aug 2017  [78b0a7b]
+libisofs/fs_image.c
+Disallowed RRIP CL chaining in order to break any endless loops.
+Debian bug 872761. Thanks Jakub Wilk and American Fuzzy Lop.
+
+21 Aug 2017  [e599a57]
+libisofs/fs_image.c
+Closed a memory leak about RRIP CL following
+
+24 Aug 2017  [cace41e]
+libisofs/ecma119.h
+libisofs/ecma119.c
+libisofs/system_area.c
+Enabled partition intervals with source "imported_iso" with ISO growing
+
+26 Aug 2017  [028f927]
+libisofs/libisofs.h
+libisofs/ecma119.c
+libisofs/system_area.c
+libisofs/messages.c
+Throw error if imported_iso interval would be overwritten by multi-session
+
+09 Sep 2017  [4e5a54c]
+libisofs/fs_image.c
+Swapped at recognition time the precendence of MBR properties "isohybrid"
+and "protective-msdos-label"
+
+12 Sep 2017  [7234425]
+ChangeLog
+Updated change log
+
+12 Sep 2017  [dfc6de9]
+configure.ac
+libisofs/libisofs.h
+Version leap to 1.4.8.
+
+12 Sep 2017  [bdfd4c4]
+libisofs/changelog.txt
+Updated change log
+
+------------------------------------ release - libisofs-1.4.8 - 12 Sep 2017
+* Bug fix: iso_read_opts_set_no_rockridge() did not prevent reading of root SUSP
+* Bug fix: Non-SUSP data in System Use Area prevented image loading if
+           Rock Ridge was enabled. Thanks to Jonathan Dowland.
+* Bug fix: Protective MBR for GPT could emerge with boot flag set.
+* Bug fix: Appended partitions of size >= 4 GiB led to abort with error message
+           "FATAL : ISO overwrite". Thanks to Sven Haardiek.
+* Bug fix: Bit 15 of iso_write_opts_set_system_area did not work with
+           generic MBR.
+* Bug fix: Keeping and patching of loaded boot images failed.
+           Regression by version 1.4.4.
+* Bug fix: Program crashes by intentionally wrong ISO image input.
+           Found by American Fuzzy Lop and Jakub Wilk.
+           Debian bug reports: 872372, 872475, 872545, 872590, 872761.
+* New API calls el_torito_set_full_load(), el_torito_get_full_load().
+* New API call iso_write_opts_set_iso_mbr_part_type().
+
+
+12 Sep 2017  [48ee49a]
+configure.ac
+libisofs/libisofs.h
+Version leap to 1.4.9
+
+12 Sep 2017  [ce831f1]
+ChangeLog
+libisofs/changelog.txt
+Updated change log
+
+15 Sep 2017  [34e3586]
+libisofs/node.c
+Silenced harmless compiler warning -Wimplicit-fallthrough
+
+16 Sep 2017  [874dc16]
+libisofs/hfsplus.c
+Fixed a message typo found by lintian
+
+22 Sep 2017  [53b2d6d]
+libisofs/hfsplus_classes.c
+Bug fix: Reading beyond array end for HFS+ production caused SIGSEGV with
+         FreeBSD 11 CLANG -O2. Thanks ASX of GhostBSD.
+
+22 Sep 2017  [79baab3]
+libisofs/hfsplus_classes.c
+Fixed a harmless lapse with static array initialization
+
+07 Oct 2017  [7d45c88]
+libisofs/libisofs.h
+libisofs/image.h
+libisofs/image.c
+libisofs/builder.c
+libisofs/fs_local.c
+libisofs/libisofs.ver
+New API call iso_image_get_ignore_aclea(),
+new iso_image_set_ignore_aclea() and iso_file_source_get_aa_string() flag bit3
+to import all xattr namespaces
+
+23 Oct 2017  [4b031b5]
+libisofs/libisofs.h
+libisofs/image.h
+libisofs/image.c
+libisofs/fs_local.c
+libisofs/builder.c
+libisofs/aaip_0_2.h
+libisofs/aaip_0_2.c
+libisofs/aaip-os-linux.c
+libisofs/aaip-os-freebsd.c
+libisofs/aaip-os-dummy.c
+libisofs/libisofs.ver
+New flag bit7 with iso_local_set_attrs() to avoid unnecessary write attempts.
+New return value 2 of IsoFileSource.get_aa_string() and iso_local_get_attrs().
+New API calls iso_image_was_blind_attrs(), iso_local_set_attrs_errno().
+
+23 Oct 2017  [633b4d5]
+doc/boot_sectors.txt
+Updated project mail addresses 
+
+31 Oct 2017  [1da3b17]
+libisofs/aaip-os-linux.c
+Changed a comment in Linux OS adapter
+
+31 Oct 2017  [580b154]
+libisofs/node.c
+Adapted iso_node_merge_xattr to handling of all namespaces
+
+22 Nov 2017  [a936409]
+libisofs/system_area.c
+Fixed failure to compile with experimental Libisofs_appended_partitions_inlinE
+
+30 Mar 2018  [615dc7e]
+libisofs/ecma119.h
+libisofs/ecma119.c
+libisofs/system_area.c
+Bug fix: Add-on sessions with partition offset claimed too many blocks as size.
+         Regression of version 1.4.8.
+
+31 Mar 2018  [ad843f1]
+libisofs/joliet.c
+Bug fix: Long Joliet names without dot were mangled with one character too many
+
+31 Mar 2018  [3106121]
+libisofs/joliet.c
+Bug fix: Long Joliet names with leading dot were mangled one char too short
+
+01 May 2018  [c5a9cc5]
+libisofs/ecma119.h
+libisofs/ecma119.c
+libisofs/system_area.c
+Changed bug fix 615dc7e9978ea0ba1eed7b4b661fe4e9f6f2769e of Mar 30 13:51:21 2018 +0200
+
+18 May 2018  [848e039]
+configure.ac
+libisofs/aaip-os-linux.c
+Preferring Linux include file sys/xattr.h over attr/attr.h
+
+04 Jun 2018  [f39d4ee]
+libisofs/ecma119.h
+libisofs/system_area.h
+libisofs/ecma119.c
+libisofs/system_area.c
+Putting user defined padding after appended partitions
+
+10 Jun 2018  [69c8c54]
+libisofs/libisofs.h
+libisofs/messages.c
+libisofs/fs_image.c
+Improved message at image load time about hidden El Torito images for EFI
+
+15 Sep 2018  [d3c17d0]
+libisofs/libisofs.h
+Version leap to 1.5.0.
+
+15 Sep 2018  [e317a8d]
+ChangeLog
+libisofs/changelog.txt
+Updated change log
+
+16 Sep 2018  [066c6f6]
+libisofs/aaip-os-freebsd.c
+Fixed failure to build on NetBSD because of undeclared variable
+
+
+------------------------------------ release - libisofs-1.5.0 - 15 Sep 2018
+* New API call iso_image_get_ignore_aclea(),
+  new iso_image_set_ignore_aclea() and iso_file_source_get_aa_string()
+  flag bit3 to import all xattr namespaces
+* New API calls iso_image_was_blind_attrs(), iso_local_set_attrs_errno().
+* New flag bit7 with iso_local_set_attrs() to avoid unnecessary write attempts.
+* New return value 2 of IsoFileSource.get_aa_string() and iso_local_get_attrs().
+* Now putting user defined padding after appended partitions.
+* Bug fix: Add-on sessions with partition offset claimed too many blocks as
+           size. Regression of version 1.4.8.
+* Bug fix: Long Joliet names without dot were mangled with one character too
+           many. Long Joliet names with leading dot were mangled one char
+           too short.
+* Bug fix: Reading beyond array end for HFS+ production caused SIGSEGV with
+           FreeBSD 11 CLANG -O2. Thanks ASX of GhostBSD.
+

libisofs/data_source.c

@@ -21,6 +21,11 @@
 #include <fcntl.h>
 #include <unistd.h>
 
+/* O_BINARY is needed for Cygwin but undefined elsewhere */
+#ifndef O_BINARY
+#define O_BINARY 0
+#endif
+
 /**
  * Private data for File IsoDataSource
  */
@@ -65,7 +70,7 @@ int ds_open(IsoDataSource *src)
         return ISO_FILE_ALREADY_OPENED;
     }
 
-    fd = open(data->path, O_RDONLY);
+    fd = open(data->path, O_RDONLY | O_BINARY);
     if (fd == -1) {
         return ISO_FILE_ERROR;
     }
@@ -140,7 +145,7 @@ void ds_free_data(IsoDataSource *src)
 
 /**
  * Create a new IsoDataSource from a local file. This is suitable for
- * accessing regular .iso images, or to acces drives via its block device
+ * accessing regular .iso images, or to access drives via its block device
  * and standard POSIX I/O calls.
  * 
  * @param path

libisofs/ecma119.h

@@ -1,6 +1,6 @@
 /*
  * Copyright (c) 2007 Vreixo Formoso
- * Copyright (c) 2009 - 2013 Thomas Schmitt
+ * Copyright (c) 2009 - 2018 Thomas Schmitt
  *
  * This file is part of the libisofs project; you can redistribute it and/or
  * modify it under the terms of the GNU General Public License version 2 
@@ -59,7 +59,7 @@
 #define ISO_DISC_LABEL_SIZE 129
 
 
-/* The maximum lenght of an specs violating ECMA-119 file identifier.
+/* The maximum length of an specs violating ECMA-119 file identifier.
    The theoretical limit is  254 - 34 - 28 (len of SUSP CE entry) = 192
    Currently the practical limit is 254 - 34 - 96 (non-CE RR entries) - 28 (CE)
 */
@@ -223,7 +223,7 @@ struct iso_write_opts {
     /**
      * Write AAIP as extension according to SUSP 1.10 rather than SUSP 1.12.
      * I.e. without announcing it by an ER field and thus without the need
-     * to preceed the RRIP fields by an ES and to preceed the AA field by ES.
+     * to precede the RRIP fields by an ES and to precede the AA field by ES.
      * This saves bytes and might avoid problems with readers which dislike
      * ER fields other than the ones for RRIP.
      * On the other hand, SUSP 1.12 frowns on such unannounced extensions
@@ -468,17 +468,41 @@ struct iso_write_opts {
        to HFS+/FAT and IsoFileSrc areas and marked by an MBR partition entry.
     */
     char *prep_partition;
+    int prep_part_flag;
 
     /* Eventual disk file path of an EFI system partition image which shall
        be prepended to HFS+/FAT and IsoFileSrc areas and marked by a GPT entry.
     */
     char *efi_boot_partition;
+    int efi_boot_part_flag;
 
     /* Eventual disk file paths of prepared images which shall be appended
        after the ISO image and described by partiton table entries in a MBR
     */
     char *appended_partitions[ISO_MAX_PARTITIONS]; 
     uint8_t appended_part_types[ISO_MAX_PARTITIONS];
+    int appended_part_flags[ISO_MAX_PARTITIONS];
+
+    /* If 1: With appended partitions: create protective MBR and mark by GPT
+     */
+    int appended_as_gpt;
+
+    /* If 1: With appended partitions: mark by APM partition
+     */
+    int appended_as_apm;
+
+    /* If 1: Obey struct el_torito_boot_image.isolinux_options bit2-7 and bit8.
+             I.e. mention boot image as partition in GPT and/or APM.
+     */
+    int part_like_isohybrid;
+
+    /* The type to use for the mountable ISO partition if there is any and if
+       the type is not mandatorily determined for particular circumstances like
+       compliant GPT, CHRP, or PReP.
+       -1 = use the default value (e.g. 0xcd, 0x83, 0x17)
+       0x00 to 0xff = value to use if possible 
+    */
+    int iso_mbr_part_type;
 
     /* Eventual name of the non-ISO aspect of the image. E.g. SUN ASCII label.
      */
@@ -497,6 +521,13 @@ struct iso_write_opts {
      */
     int apm_block_size;
 
+    /* User defined GUID for GPT header and base of reproducible partition
+       GUIDs. (Not to be confused with volume "UUID", which is actually a
+       timestamp.)
+       See API call iso_write_opts_set_gpt_guid().
+     */
+    uint8_t gpt_disk_guid[16];
+    int gpt_disk_guid_mode;
 };
 
 typedef struct ecma119_image Ecma119Image;
@@ -547,10 +578,19 @@ struct ecma119_image
 
     time_t now; /**< Time at which writing began. */
 
-    /** Total size of the output. This only includes the current volume. */
+    /* Total size of the output. Counted in bytes.
+     * Includes ISO filesystem and appended data.
+     */
     off_t total_size;
+
+    /** Size actually governed by the ISO filesystem part of the output */
     uint32_t vol_space_size;
 
+    /* 1= write the total size into the PVD of the ISO,
+     * 0= write vol_space_size
+     */
+    int pvd_size_is_total_size;
+
     /* Bytes already written to image output */
     off_t bytes_written;
     /* just for progress notification */
@@ -633,6 +673,11 @@ struct ecma119_image
     int num_bootsrc;
     IsoFileSrc **bootsrc; /* location of the boot images in the new image */
 
+    int *boot_appended_idx; /* Appended partition which serve as boot images */
+
+    uint32_t *boot_intvl_start;   /* In blocks of 2048 bytes */
+    uint32_t *boot_intvl_size;    /* In blocks of  512 bytes */
+
     /*
      * System Area related information
      */
@@ -695,6 +740,8 @@ struct ecma119_image
     /* tree of files sources */
     IsoRBTree *files;
 
+    struct iso_filesrc_list_item *ecma119_hidden_list;
+
     unsigned int checksum_idx_counter;
     void *checksum_ctx;
     off_t checksum_counter;
@@ -752,10 +799,16 @@ struct ecma119_image
     */
     IsoFileSrc *sparc_core_src;
 
+    /* Trailing padding of ISO filesystem partition for cylinder alignment */
+    /* Only in effect with Libisofs_part_align_writeR */
+    uint32_t part_align_blocks;
+    uint32_t alignment_end_block;
+
     /* Counted in blocks of 2048 */
     uint32_t appended_part_prepad[ISO_MAX_PARTITIONS];
     uint32_t appended_part_start[ISO_MAX_PARTITIONS];
     uint32_t appended_part_size[ISO_MAX_PARTITIONS];
+    int have_appended_partitions;
 
     /* See IsoImage and libisofs.h */
     IsoNode *hfsplus_blessed[ISO_HFSPLUS_BLESS_MAX];
@@ -790,6 +843,12 @@ struct ecma119_image
     struct iso_mbr_partition_request *mbr_req[ISO_MBR_ENTRIES_MAX];
     int mbr_req_count;
 
+    /* Number of bytes which have to be added after the cylinder aligned end
+       of the overall ISO partition because clinder size is not a multiple
+       of 2048
+    */
+    int post_iso_part_pad;
+
     uint32_t prep_part_size;
 
     /* GPT description. To be composed during IsoImageWriter
@@ -802,6 +861,14 @@ struct ecma119_image
     /* bit0= GPT partitions may overlap */
     int gpt_req_flags;
 
+    /* Whether the eventual backup GPT is not part of the ISO filesystem */
+    int gpt_backup_outside;
+
+    /* The base UUID for the generated GPT UUIDs */
+    uint8_t gpt_uuid_base[16];
+    /* The counter which distinguishes the GPT UUIDs */
+    uint32_t gpt_uuid_counter;
+
     uint32_t efi_boot_part_size;
     IsoFileSrc *efi_boot_part_filesrc; /* Just a pointer. Do not free. */
 
@@ -822,6 +889,7 @@ struct ecma119_image
        write_data() methods of the writers.
     */
     uint8_t sys_area_as_written[16 * BLOCK_SIZE];
+    int sys_area_already_written;
 
     /* Size of the filesrc_writer area (data file content).
        This is available before any IsoImageWriter.compute_data_blocks()
@@ -969,5 +1037,22 @@ int iso_write_partition_file(Ecma119Image *target, char *path,
 
 void issue_ucs2_warning_summary(size_t failures);
 
+/* Tells whether ivr is a reader from imported_iso in a multi-session
+   add-on situation, and thus to be kept in place.
+*/
+int iso_interval_reader_keep(Ecma119Image *target,
+                             struct iso_interval_reader *ivr,
+                             int flag);
+
+/* @return: ISO_SUCCESS = ok, ISO_SUCCESS + 1 = keep , < 0 = error */
+int iso_interval_reader_start_size(Ecma119Image *t, char *path,
+                                   off_t *start_byte, off_t *byte_count,
+                                   int flag);
+
+/* Obtains start and end number of appended partition range and returns
+   the number of valid entries in the list of appended partitions.
+*/
+int iso_count_appended_partitions(Ecma119Image *target,
+                                  int *first_partition, int *last_partition);
 
 #endif /*LIBISO_ECMA119_H_*/

libisofs/ecma119_tree.c

@@ -1,6 +1,6 @@
 /*
  * Copyright (c) 2007 Vreixo Formoso
- * Copyright (c) 2009 - 2014 Thomas Schmitt
+ * Copyright (c) 2009 - 2016 Thomas Schmitt
  *
  * This file is part of the libisofs project; you can redistribute it and/or
  * modify it under the terms of the GNU General Public License version 2 
@@ -241,11 +241,11 @@ int create_file_src(Ecma119Image *img, IsoFile *iso, IsoFileSrc **src)
     size = iso_stream_get_size(iso->stream);
     if (size > (off_t)MAX_ISO_FILE_SECTION_SIZE && img->opts->iso_level != 3) {
         char *ipath = iso_tree_get_node_path(ISO_NODE(iso));
-        ret = iso_msg_submit(img->image->id, ISO_FILE_TOO_BIG, 0,
-                              "File \"%s\" can't be added to image because "
-                              "is greater than 4GB", ipath);
+        iso_msg_submit(img->image->id, ISO_FILE_TOO_BIG, 0,
+                              "File \"%s\" cannot be added to image because "
+                              "its size is 4 GiB or larger", ipath);
         free(ipath);
-        return ret;
+        return ISO_FILE_TOO_BIG;
     }
     ret = iso_file_src_create(img, iso, src);
     if (ret < 0) {
@@ -365,6 +365,35 @@ void ecma119_node_free(Ecma119Node *node)
     free(node);
 }
 
+
+static
+int add_to_hidden_list(Ecma119Image *image, IsoFileSrc *src)
+{
+    int ret;
+    struct iso_filesrc_list_item *item;
+
+    LIBISO_ALLOC_MEM(item, struct iso_filesrc_list_item, 1);
+    item->src = src;
+    item->next = image->ecma119_hidden_list;
+    image->ecma119_hidden_list = item;
+    ret = ISO_SUCCESS;
+ex:
+    return ret;
+}
+
+
+int iso_filesrc_list_destroy(struct iso_filesrc_list_item **start_item)
+{
+    struct iso_filesrc_list_item *item, *next;
+
+    for (item = *start_item; item != NULL; item = next) {
+        next = item->next;
+        LIBISO_FREE_MEM(item);
+    }
+    return ISO_SUCCESS;
+}
+ 
+
 /**
  * @param flag
  *      bit0= iso is in a hidden directory. Thus hide it.
@@ -381,11 +410,12 @@ int create_tree(Ecma119Image *image, IsoNode *iso, Ecma119Node **tree,
     int max_path;
     char *iso_name= NULL, *ipath = NULL;
     IsoFileSrc *src = NULL;
-    IsoWriteOpts *opts = image->opts;
+    IsoWriteOpts *opts;
 
     if (image == NULL || iso == NULL || tree == NULL) {
         return ISO_NULL_POINTER;
     }
+    opts = image->opts;
     *tree = NULL;
 
     hidden = flag & 1;
@@ -430,6 +460,9 @@ int create_tree(Ecma119Image *image, IsoNode *iso, Ecma119Node **tree,
     case LIBISO_FILE:
         if (hidden) {
             ret = create_file_src(image, (IsoFile *) iso, &src);
+            if (ret <= 0)
+                goto ex;
+            ret = add_to_hidden_list(image, src); 
         } else {
             ret = create_file(image, (IsoFile*)iso, &node);
         }
@@ -470,6 +503,9 @@ int create_tree(Ecma119Image *image, IsoNode *iso, Ecma119Node **tree,
         if (image->eltorito) {
             if (hidden) {
                 ret = el_torito_catalog_file_src_create(image, &src);
+                if (ret <= 0)
+                    goto ex;
+                ret = add_to_hidden_list(image, src); 
             } else {
                 ret = create_boot_cat(image, (IsoBoot*)iso, &node);
             }
@@ -561,7 +597,7 @@ int cmp_node_name(const void *f1, const void *f2)
 
 /**
  * Sorts a the children of each directory in the ECMA-119 tree represented
- * by \p root, acording to the order specified in ECMA-119, section 9.3.
+ * by \p root, according to the order specified in ECMA-119, section 9.3.
  */
 static
 void sort_tree(Ecma119Node *root)
@@ -618,6 +654,7 @@ int mangle_single_dir(Ecma119Image *img, Ecma119Node *dir, int max_file_len,
     for (i = 0; i < nchildren; ++i) {
         char *name, *ext;
         char full_name[40];
+        const int full_max_len = 40 - 1;
         int max; /* computed max len for name, without extension */
         int j = i;
         int digits = 1; /* characters to change per name */
@@ -656,7 +693,8 @@ int mangle_single_dir(Ecma119Image *img, Ecma119Node *dir, int max_file_len,
             int change = 0; /* number to be written */
 
             /* copy name to buffer */
-            strcpy(full_name, children[i]->iso_name);
+            strncpy(full_name, children[i]->iso_name, full_max_len);
+            full_name[full_max_len] = 0;
 
             /* compute name and extension */
             dot = strrchr(full_name, '.');
@@ -1232,7 +1270,7 @@ int ecma119_tree_create(Ecma119Image *img)
 
     if (img->opts->rockridge && !img->opts->allow_deep_paths) {
 
-        /* Relocate deep directories, acording to RRIP, 4.1.5 */
+        /* Relocate deep directories, according to RRIP, 4.1.5 */
         ret = reorder_tree(img, root, 1, 0);
         if (ret < 0) {
             return ret;

libisofs/ecma119_tree.h

@@ -79,6 +79,17 @@ struct ecma119_node
     } info;
 };
 
+
+/* For recording files which are hidden in ECMA-119 */
+struct iso_filesrc_list_item
+{
+    IsoFileSrc *src;
+    struct iso_filesrc_list_item *next;
+};
+
+int iso_filesrc_list_destroy(struct iso_filesrc_list_item **start_item);
+
+
 /**
  * 
  */

libisofs/eltorito.c

@@ -1,6 +1,6 @@
 /*
  * Copyright (c) 2007 Vreixo Formoso
- * Copyright (c) 2010 - 2014 Thomas Schmitt
+ * Copyright (c) 2010 - 2016 Thomas Schmitt
  *
  * This file is part of the libisofs project; you can redistribute it and/or
  * modify it under the terms of the GNU General Public License version 2 
@@ -19,6 +19,7 @@
 #include "image.h"
 #include "messages.h"
 #include "writer.h"
+#include "ecma119.h"
 
 #include <stdlib.h>
 #include <string.h>
@@ -112,9 +113,24 @@ void el_torito_set_load_size(ElToritoBootImage *bootimg, short sectors)
 /* API */
 int el_torito_get_load_size(ElToritoBootImage *bootimg)
 {
-   return (int) bootimg->load_size;
+    return (int) bootimg->load_size;
 }
 
+/* API */
+void el_torito_set_full_load(ElToritoBootImage *bootimg, int mode)
+{
+    if (bootimg->type != 0)
+        return;
+    bootimg->load_size_full= !!mode;
+}
+
+/* API */
+int el_torito_get_full_load(ElToritoBootImage *bootimg)
+{
+    return bootimg->load_size_full;
+}
+
+
 /**
  * Marks the specified boot image as not bootable
  */
@@ -206,6 +222,8 @@ void el_torito_patch_isolinux_image(ElToritoBootImage *bootimg)
 int el_torito_set_isolinux_options(ElToritoBootImage *bootimg, int options, int flag)
 {
     bootimg->isolinux_options = (options & 0x03ff);
+    bootimg->seems_boot_info_table = !!(options & 1);
+    bootimg->seems_grub2_boot_info = !!(options & (1 << 9));
     return ISO_SUCCESS;
 }
 
@@ -287,7 +305,7 @@ int iso_tree_add_boot_node(IsoDir *parent, const char *name, IsoBoot **boot)
     node->size = 0;
     node->content = NULL;
 
-    /* atributes from parent */
+    /* attributes from parent */
     node->node.mode = S_IFREG | (parent->node.mode & 0444);
     node->node.uid = parent->node.uid;
     node->node.gid = parent->node.gid;
@@ -310,6 +328,35 @@ int iso_tree_add_boot_node(IsoDir *parent, const char *name, IsoBoot **boot)
     return ++parent->nchildren;
 }
 
+/* Get start and size from "%d_start_%lus_size_%lud" */
+static
+void iso_parse_start_size(char *text, unsigned long *part_start,
+                          unsigned long *part_size)
+{
+    char *cpt;
+    unsigned long start, size;
+
+    cpt = strchr(text, '_');
+    if (cpt == NULL)
+        return;
+    if (strncmp(cpt, "_start_", 7) != 0)
+        return;
+    sscanf(cpt + 7, "%lu", &start);
+    cpt = strchr(cpt + 7, '_');
+    if (cpt == NULL)
+        return;
+    if (*(cpt - 1) != 's')
+        return;
+    if (strncmp(cpt, "_size_", 6) != 0)
+        return;
+    sscanf(cpt + 6, "%lu", &size);
+    for (cpt = cpt + 6; *cpt >= '0' && *cpt <= '9'; cpt++);
+    if (*cpt != 'd')
+        return;
+
+    *part_start = start;
+    *part_size = size;
+}
 
 static
 int create_image(IsoImage *image, const char *image_path,
@@ -321,36 +368,64 @@ int create_image(IsoImage *image, const char *image_path,
     struct el_torito_boot_image *boot;
     int boot_media_type = 0;
     int load_sectors = 0; /* number of sector to load */
+    int part_idx = -1;
+    unsigned long part_start = 0, part_size = 0;
     unsigned char partition_type = 0;
     off_t size;
-    IsoNode *imgfile;
-    IsoStream *stream;
+    IsoNode *imgfile = NULL;
+    IsoStream *stream = NULL;
 
     *bootnode = NULL;
-    ret = iso_tree_path_to_node(image, image_path, &imgfile);
-    if (ret < 0) {
-        return ret;
-    }
-    if (ret == 0) {
-        iso_msg_submit(image->id, ISO_NODE_DOESNT_EXIST, 0,
+
+    if (strncmp(image_path, "--interval:appended_partition_", 30) == 0) {
+        /* --interval:appended_partition_N... */ 
+        if (type != ELTORITO_NO_EMUL) {
+
+            /* >>> ??? lift this ban by making a temporary IsoStream from
+                       partition source, determine size,
+                       and read ELTORITO_HARD_DISC_EMUL MBR ?
+             */
+
+            iso_msg_submit(image->id, ISO_BOOT_IMAGE_NOT_VALID, 0,
+                           "Appended partition cannot serve as El Torito boot image with FD/HD emulation");
+            return ISO_BOOT_IMAGE_NOT_VALID;
+        }
+        sscanf(image_path + 30, "%d", &part_idx);
+        if (part_idx < 1 || part_idx > ISO_MAX_PARTITIONS) {
+            iso_msg_submit(image->id, ISO_BOOT_IMAGE_NOT_VALID, 0,
+          "Appended partition index for El Torito boot image is out of range");
+            return ISO_BOOT_IMAGE_NOT_VALID;
+        }
+        iso_parse_start_size((char *) (image_path + 30),
+                              &part_start, &part_size);
+        part_idx--;
+        size = 1;
+    } else {
+        ret = iso_tree_path_to_node(image, image_path, &imgfile);
+        if (ret < 0) {
+            return ret;
+        }
+        if (ret == 0) {
+            iso_msg_submit(image->id, ISO_NODE_DOESNT_EXIST, 0,
                        "El Torito boot image file missing in ISO image: '%s'",
                        image_path);
-        return ISO_NODE_DOESNT_EXIST;
+            return ISO_NODE_DOESNT_EXIST;
+        }
+
+        if (imgfile->type != LIBISO_FILE) {
+            return ISO_BOOT_IMAGE_NOT_VALID;
+        }
+        *bootnode = (IsoFile *) imgfile;
+
+        stream = ((IsoFile*)imgfile)->stream;
+
+        /* we need to read the image at least two times */
+        if (!iso_stream_is_repeatable(stream)) {
+            return ISO_BOOT_IMAGE_NOT_VALID;
+        }
+
+        size = iso_stream_get_size(stream);
     }
-
-    if (imgfile->type != LIBISO_FILE) {
-        return ISO_BOOT_IMAGE_NOT_VALID;
-    }
-    *bootnode = (IsoFile *) imgfile;
-
-    stream = ((IsoFile*)imgfile)->stream;
-
-    /* we need to read the image at least two times */
-    if (!iso_stream_is_repeatable(stream)) {
-        return ISO_BOOT_IMAGE_NOT_VALID;
-    }
-
-    size = iso_stream_get_size(stream);
     if (size <= 0) {
         iso_msg_submit(image->id, ISO_BOOT_IMAGE_NOT_VALID, 0,
                        "Boot image file is empty");
@@ -439,7 +514,11 @@ int create_image(IsoImage *image, const char *image_path,
         return ISO_OUT_OF_MEM;
     }
     boot->image = (IsoFile*)imgfile;
-    iso_node_ref(imgfile); /* get our ref */
+    boot->appended_idx = part_idx;
+    boot->appended_start = part_start;
+    boot->appended_size = part_size;
+    if (imgfile != NULL)
+        iso_node_ref(imgfile); /* get our ref */
     boot->bootable = 1;
     boot->seems_boot_info_table = 0;
     boot->seems_grub2_boot_info = 0;
@@ -449,12 +528,11 @@ int create_image(IsoImage *image, const char *image_path,
     boot->partition_type = partition_type;
     boot->load_seg = 0;
     boot->load_size = load_sectors;
+    boot->load_size_full = 0;
     boot->platform_id = 0; /* 80x86 */
     memset(boot->id_string, 0, sizeof(boot->id_string));
     memset(boot->selection_crit, 0, sizeof(boot->selection_crit));
-    if (bootimg) {
-        *bootimg = boot;
-    }
+    *bootimg = boot;
 
     return ISO_SUCCESS;
 }
@@ -537,9 +615,10 @@ int iso_image_set_boot_image(IsoImage *image, const char *image_path,
     for (i = 1; i < Libisofs_max_boot_imageS; i++)
         catalog->bootimages[i] = NULL;
     catalog->node = cat_node;
-    catalog->sort_weight = 1000;                            /* slightly high */
-    if (!boot_node->explicit_weight)
-        boot_node->sort_weight = 2;
+    catalog->sort_weight = 1000000000;                          /* very high */
+    if (boot_node != NULL)
+        if (!(boot_node->explicit_weight || boot_node->from_old_session))
+            boot_node->sort_weight = 2;
     iso_node_ref((IsoNode*)cat_node);
     image->bootcat = catalog;
 
@@ -555,7 +634,8 @@ boot_image_cleanup:;
         iso_node_unref((IsoNode*)cat_node);
     }
     if (boot_image) {
-        iso_node_unref((IsoNode*)boot_image->image);
+        if (boot_image->image != NULL)
+            iso_node_unref((IsoNode*)boot_image->image);
         free(boot_image);
     }
     return ret;
@@ -719,8 +799,9 @@ int iso_image_add_boot_image(IsoImage *image, const char *image_path,
     ret = create_image(image, image_path, type, &boot_img, &boot_node);
     if (ret < 0) 
         return ret;
-    if (!boot_node->explicit_weight)
-        boot_node->sort_weight = 2;
+    if (boot_node != NULL)
+        if (!(boot_node->explicit_weight || boot_node->from_old_session))
+            boot_node->sort_weight = 2;
     catalog->bootimages[catalog->num_bootimages] = boot_img;
     catalog->num_bootimages++;
     if (boot != NULL)
@@ -812,7 +893,8 @@ write_section_header(uint8_t *buf, Ecma119Image *t, int idx, int num_entries)
         (struct el_torito_section_header *) buf;
 
     /* 0x90 = more section headers follow , 0x91 = final section */
-    e->header_indicator[0] = 0x90 + (idx == t->catalog->num_bootimages - 1);
+    e->header_indicator[0] =
+                      0x90 + (idx == t->catalog->num_bootimages - num_entries);
     e->platform_id[0] = t->catalog->bootimages[idx]->platform_id;
     e->num_entries[0] = num_entries & 0xff;
     e->num_entries[1] = (num_entries >> 8) & 0xff;;
@@ -821,17 +903,44 @@ write_section_header(uint8_t *buf, Ecma119Image *t, int idx, int num_entries)
            sizeof(e->id_string));
 }
 
+static int
+write_section_load_size(struct el_torito_boot_image *img,
+                        struct el_torito_section_entry *se,
+                        uint16_t load_size, off_t full_byte_size, int flag)
+{
+    uint16_t size;
+    off_t blocks;
+
+    size= load_size;
+    if(img->type == 0 && img->load_size_full) {
+        blocks= ((full_byte_size + 2047) / 2048) * 4;
+        if (blocks > 65535) {
+            if (img->platform_id == 0xef)
+                size= 0;
+            else
+                size= 65535;
+        } else if(blocks <= 0) {
+            size= 1;
+        } else {
+            size= blocks;
+        }
+    }
+    iso_lsb(se->sec_count, size, 2);
+    return(1);
+}
+
 /**
  * Write one section entry.
  * Usable for the Default Entry
  * and for Section Entries with Selection criteria type == 0
  */
-static void
-write_section_entry(uint8_t *buf, Ecma119Image *t, int idx)
+static
+int write_section_entry(uint8_t *buf, Ecma119Image *t, int idx)
 {
     struct el_torito_boot_image *img;
     struct el_torito_section_entry *se =
         (struct el_torito_section_entry*)buf;
+    int app_idx, mode = 0;
 
     img = t->catalog->bootimages[idx];
 
@@ -839,16 +948,75 @@ write_section_entry(uint8_t *buf, Ecma119Image *t, int idx)
     se->boot_media_type[0] = img->type;
     iso_lsb(se->load_seg, img->load_seg, 2);
     se->system_type[0] = img->partition_type;
-    iso_lsb(se->sec_count, img->load_size, 2);
-    iso_lsb(se->block, t->bootsrc[idx]->sections[0].block, 4);
+
+    if (t->boot_appended_idx[idx] >= 0)
+        if (t->appended_part_size[t->boot_appended_idx[idx]] > 0)
+            mode = 2; /* appended partition */
+    if (mode == 0 && t->opts->appendable &&
+        (t->boot_intvl_start[idx] > 0 || t->boot_intvl_size[idx] > 0) &&
+         t->boot_intvl_start[idx] + (t->boot_intvl_size[idx] + 3) / 4 <=
+         t->opts->ms_block)
+        mode = 1; /* image interval */
+    if (mode == 0 && t->boot_appended_idx[idx] >= 0) {
+        iso_msg_submit(t->image->id, ISO_BOOT_IMAGE_NOT_VALID, 0,
+          "Appended partition which shall serve as boot image does not exist");
+        return ISO_BOOT_IMAGE_NOT_VALID;
+    }
+
+    if (mode == 1) {
+        if (t->boot_intvl_start[idx] + (t->boot_intvl_size[idx] + 3) / 4 >
+            t->total_size / 2048 + t->opts->ms_block - t->eff_partition_offset
+           ) {
+            iso_msg_submit(t->image->id, ISO_BOOT_IMAGE_NOT_VALID, 0,
+     "Block interval which shall serve as boot image is outside result range");
+            return ISO_BOOT_IMAGE_NOT_VALID;
+        }
+
+        /* >>> check for non-automatic load size */;
+
+        if (t->boot_intvl_size[idx] > 65535) {
+            if (img->platform_id == 0xef)
+                iso_lsb(se->sec_count, 0, 2);
+            else
+                iso_lsb(se->sec_count, 65535, 2);
+        } else {
+            if (t->boot_intvl_size[idx] == 0) {
+                iso_msg_submit(t->image->id, ISO_BOOT_IMAGE_NOT_VALID, 0,
+               "Block interval which shall serve as boot image has zero size");
+                return ISO_BOOT_IMAGE_NOT_VALID;
+            }
+            iso_lsb(se->sec_count, t->boot_intvl_size[idx], 2);
+        }
+        iso_lsb(se->block, t->boot_intvl_start[idx], 4);
+    } else if (mode == 2) {
+        app_idx = t->boot_appended_idx[idx];
+
+        /* >>> check for non-automatic load size */;
+
+        if (t->appended_part_size[app_idx] * 4 > 65535) {
+            if (img->platform_id == 0xef)
+                iso_lsb(se->sec_count, 0, 2);
+            else
+                iso_lsb(se->sec_count, 65535, 2);
+        } else {
+            iso_lsb(se->sec_count, t->appended_part_size[app_idx] * 4, 2);
+        }
+        iso_lsb(se->block, t->appended_part_start[app_idx], 4);
+    } else {
+        write_section_load_size(img, se, (uint16_t) img->load_size,
+                                (off_t) t->bootsrc[idx]->sections[0].size, 0);
+        iso_lsb(se->block, t->bootsrc[idx]->sections[0].block, 4);
+    }
+
     se->selec_criteria[0] = img->selection_crit[0];
     memcpy(se->vendor_sc, img->selection_crit + 1, 19);
+    return ISO_SUCCESS;
 }
 
 static
 int catalog_open(IsoStream *stream)
 {
-    int i, j, k, num_entries;
+    int i, j, k, num_entries, ret;
     struct catalog_stream *data;
     uint8_t *wpt;
     struct el_torito_boot_catalog *cat;
@@ -872,7 +1040,9 @@ int catalog_open(IsoStream *stream)
                            boots[0]->platform_id, boots[0]->id_string);
 
     /* write default entry = first boot image */
-    write_section_entry(data->buffer + 32, data->target, 0);
+    ret = write_section_entry(data->buffer + 32, data->target, 0);
+    if (ret < 0)
+        return ret;
 
     /* IMPORTANT: The maximum number of boot images must fit into BLOCK_SIZE */
     wpt = data->buffer + 64;
@@ -893,7 +1063,9 @@ int catalog_open(IsoStream *stream)
         write_section_header(wpt, data->target, i, num_entries);
         wpt += 32;
         for (j = 0; j < num_entries; j++) {
-            write_section_entry(wpt,  data->target, i);
+            ret = write_section_entry(wpt,  data->target, i);
+            if (ret < 0)
+                return ret;
             wpt += 32;
             i++;
         }
@@ -1131,6 +1303,9 @@ int patch_boot_info_table(uint8_t *buf, Ecma119Image *t,
         return iso_msg_submit(t->image->id, ISO_ISOLINUX_CANT_PATCH, 0,
             "Isolinux image too small. We won't patch it.");
     }
+    if (t->bootsrc[idx] == NULL)
+        return iso_msg_submit(t->image->id, ISO_ISOLINUX_CANT_PATCH, 0,
+            "Cannot apply ISOLINUX patching outside of ISO 9660 filesystem.");
     ret = make_boot_info_table(buf, t->opts->ms_block + (uint32_t) 16,
                                t->bootsrc[idx]->sections[0].block,
                                (uint32_t) imgsize);
@@ -1150,7 +1325,10 @@ int patch_grub2_boot_image(uint8_t *buf, Ecma119Image *t,
 
     if (imgsize < pos + 8)
         return iso_msg_submit(t->image->id, ISO_ISOLINUX_CANT_PATCH, 0,
-                     "Isolinux image too small for GRUB2. Will not patch it.");
+                     "Boot image too small for GRUB2. Will not patch it.");
+    if (t->bootsrc[idx] == NULL)
+        return iso_msg_submit(t->image->id, ISO_ISOLINUX_CANT_PATCH, 0,
+            "Cannot apply GRUB2 patching outside of ISO 9660 filesystem.");
     blk = ((uint64_t) t->bootsrc[idx]->sections[0].block) * 4 + offst;
     iso_lsb((buf + pos), blk & 0xffffffff, 4);
     iso_lsb((buf + pos + 4), blk >> 32, 4);
@@ -1173,12 +1351,16 @@ int iso_patch_eltoritos(Ecma119Image *t)
     for (idx = 0; idx < t->catalog->num_bootimages; idx++) {
         if (!(t->catalog->bootimages[idx]->isolinux_options & 0x201))
     continue;
+        if (t->bootsrc[idx] == NULL)
+            return iso_msg_submit(t->image->id, ISO_ISOLINUX_CANT_PATCH, 0,
+            "Cannot apply boot image patching outside of ISO 9660 filesystem");
+
         original = t->bootsrc[idx]->stream;
         size = (size_t) iso_stream_get_size(original);
-
-        /* >>> BOOT ts B00428 :
-               check whether size is not too large for buffering */;
-
+        if (size > Libisofs_elto_max_patchablE)
+            return ISO_PATCH_OVERSIZED_BOOT;
+        if (iso_stream_get_input_stream(original, 0) != NULL)
+            return ISO_PATCH_FILTERED_BOOT;
         buf = calloc(1, size);
         if (buf == NULL) {
             return ISO_OUT_OF_MEM;
@@ -1191,6 +1373,9 @@ int iso_patch_eltoritos(Ecma119Image *t)
         ret = iso_stream_read(original, buf, size);
         iso_stream_close(original);
         if (ret != (int) size) {
+            if (ret >= 0)
+               iso_msg_submit(t->image->id, ISO_FILE_READ_ERROR, 0,
+        "Cannot read all bytes from El Torito boot image for boot info table");
             return (ret < 0) ? ret : (int) ISO_FILE_READ_ERROR;
         }
 
@@ -1277,8 +1462,8 @@ int eltorito_writer_create(Ecma119Image *target)
 {
     int ret, idx, outsource_efi = 0;
     IsoImageWriter *writer;
-    IsoFile *bootimg;
-    IsoFileSrc *src;
+    IsoFile *bootimg = NULL;
+    IsoFileSrc *src = NULL;
 
     writer = calloc(1, sizeof(IsoImageWriter));
     if (writer == NULL) {
@@ -1311,6 +1496,18 @@ int eltorito_writer_create(Ecma119Image *target)
         if (strcmp(target->opts->efi_boot_partition, "--efi-boot-image") == 0)
             outsource_efi = 1;
     for (idx = 0; idx < target->catalog->num_bootimages; idx++) {
+        target->bootsrc[idx] = NULL;
+        if (target->catalog->bootimages[idx]->appended_idx >= 0) {
+            /* Use an appended partition as boot image rather than IsoFile */
+            target->boot_appended_idx[idx] =
+                                target->catalog->bootimages[idx]->appended_idx;
+            target->boot_intvl_start[idx] =
+                              target->catalog->bootimages[idx]->appended_start;
+            target->boot_intvl_size[idx] =
+                               target->catalog->bootimages[idx]->appended_size;
+    continue;
+        }
+
         bootimg = target->catalog->bootimages[idx]->image;
         ret = iso_file_src_create(target, bootimg, &src);
         if (ret < 0) {

libisofs/eltorito.h

@@ -53,6 +53,11 @@ struct el_torito_boot_catalog {
 struct el_torito_boot_image {
     IsoFile *image;
 
+    /* Overrides .image if >= 0 : array index of appended partition */
+    int appended_idx;
+    uint32_t appended_start; /* In blocks of 2048 bytes */
+    uint32_t appended_size;  /* In blocks of  512 bytes */
+
     unsigned int bootable:1; /**< If the entry is bootable. */
     /**
      * Whether the boot image seems to contain a boot_info_table
@@ -80,6 +85,7 @@ struct el_torito_boot_image {
     unsigned char partition_type; /**< type of partition for HD-emul images */
     uint16_t load_seg; /**< Load segment for the initial boot image. */
     uint16_t load_size; /**< Number of sectors to load. */
+    int load_size_full; /* 1= override load_size by image size */
 
     /* Byte 1 of Validation Entry or Section Header Entry:
        0= 80x86, 1= PowerPC, 2= Mac, 0xef= EFI */
@@ -168,4 +174,11 @@ int iso_patch_eltoritos(Ecma119Image *t);
 #define Libisofs_grub2_elto_patch_offsT  5
 
 
+/* Maximum size of a boot image which is marked by
+   el_torito_set_isolinux_options() for patching (boot info table,
+   GRUB2 boot info, maybe others).
+*/
+#define Libisofs_elto_max_patchablE      (32 * 1024 * 1024)
+
+
 #endif /* LIBISO_ELTORITO_H */

libisofs/filesrc.c

@@ -229,12 +229,23 @@ int shall_be_written(void *arg)
     return f->no_write ? 0 : 1;
 }
 
+static
+int shall_be_written_if_not_taken(void *arg)
+{
+    IsoFileSrc *f = (IsoFileSrc *)arg;
+    return f->no_write || f->taken ? 0 : 1;
+}
+
 int filesrc_writer_pre_compute(IsoImageWriter *writer)
 {
     size_t i, size, is_external;
     Ecma119Image *t;
     IsoFileSrc **filelist;
     int (*inc_item)(void *);
+    size_t omitted_count;
+    IsoFileSrc **iso_ecma119_to_filesrc_array(Ecma119Image *t,
+                                              int (*include_item)(void *),
+                                              size_t *size);
 
     if (writer == NULL) {
         return ISO_ASSERT_FAILURE;
@@ -257,7 +268,16 @@ int filesrc_writer_pre_compute(IsoImageWriter *writer)
     }
 
     /* store the filesrcs in a array */
-    filelist = (IsoFileSrc**)iso_rbtree_to_array(t->files, inc_item, &size);
+    filelist = (IsoFileSrc**) iso_ecma119_to_filesrc_array(t, inc_item, &size);
+    omitted_count = iso_rbtree_count_array(t->files, (size_t) 0,
+                                           shall_be_written_if_not_taken);
+    if (omitted_count > 0) {
+        iso_msg_submit(t->image->id, ISO_NOT_REPRODUCIBLE, 0,
+            "Cannot arrange content of data files in surely reproducible way");
+        LIBISO_FREE_MEM(filelist);
+        filelist = (IsoFileSrc**)iso_rbtree_to_array(
+                                                    t->files, inc_item, &size);
+    }
     if (filelist == NULL) {
         return ISO_OUT_OF_MEM;
     }

libisofs/filesrc.h

@@ -32,6 +32,10 @@ struct Iso_File_Src
     */
     unsigned int no_write :1;
 
+    /* Is 1 if the object was already put into the filelist array.
+    */
+    unsigned int taken :1;
+
     unsigned int checksum_index :31;
 
     /** File Sections of the file in the image */

libisofs/filters/external.c

@@ -655,14 +655,41 @@ IsoStreamIface extf_stream_class = {
 static
 int extf_cmp_ino(IsoStream *s1, IsoStream *s2)
 {
+    int i;
     ExternalFilterStreamData *data1, *data2;
+    IsoExternalFilterCommand *cmd1, *cmd2;
 
+    /* This function may rely on being called by iso_stream_cmp_ino()
+       only with s1, s2 which both point to it as their .cmp_ino() function.
+       It would be a programming error to let any other than extf_stream_class
+       point to extf_cmp_ino(). This fallback endangers transitivity of
+       iso_stream_cmp_ino().
+    */
     if (s1->class != &extf_stream_class || s2->class != &extf_stream_class)
         return iso_stream_cmp_ino(s1, s2, 1);
+
     data1 = (ExternalFilterStreamData*) s1->data;
     data2 = (ExternalFilterStreamData*) s2->data;
-    if (data1->cmd != data2->cmd)
-        return (data1->cmd < data2->cmd ? -1 : 1);
+    cmd1 = data1->cmd;
+    cmd2 = data2->cmd;
+    if (cmd1 != cmd2) {
+        if (strcmp(cmd1->name, cmd2->name) != 0)
+            return strcmp(cmd1->name, cmd2->name);
+        if (strcmp(cmd1->path, cmd2->path) != 0)
+            return strcmp(cmd1->path, cmd2->path);
+        if (cmd1->argc != cmd2->argc)
+            return cmd1->argc < cmd2->argc ? -1 : 1;
+        for (i = 0; i < cmd1->argc; i++) {
+            if (strcmp(cmd1->argv[i], cmd2->argv[i]) != 0)
+                return strcmp(cmd1->argv[i], cmd2->argv[i]);
+        }
+        if (cmd1->behavior != cmd2->behavior)
+            return cmd1->behavior < cmd2->behavior ? -1 : 1;
+        if (strcmp(cmd1->suffix, cmd2->suffix) != 0)
+            return strcmp(cmd1->suffix, cmd2->suffix);
+    }
+
+    /* Both streams apply the same treatment to their input streams */
     return iso_stream_cmp_ino(data1->orig, data2->orig, 0);
 }
 

libisofs/filters/gzip.c

@@ -573,6 +573,9 @@ int gzip_clone_stream(IsoStream *old_stream, IsoStream **new_stream, int flag)
 static
 int gzip_cmp_ino(IsoStream *s1, IsoStream *s2);
 
+static
+int gzip_uncompress_cmp_ino(IsoStream *s1, IsoStream *s2);
+
 
 IsoStreamIface gzip_stream_compress_class = {
     4,
@@ -603,17 +606,44 @@ IsoStreamIface gzip_stream_uncompress_class = {
     gzip_stream_free,
     gzip_update_size,
     gzip_get_input_stream,
-    gzip_cmp_ino,
+    gzip_uncompress_cmp_ino,
     gzip_clone_stream
 };
 
-    
+ 
 static
 int gzip_cmp_ino(IsoStream *s1, IsoStream *s2)
 {
+    /* This function may rely on being called by iso_stream_cmp_ino()
+       only with s1, s2 which both point to it as their .cmp_ino() function.
+       It would be a programming error to let any other than 
+       gzip_stream_compress_class point to gzip_cmp_ino().
+       This fallback endangers transitivity of iso_stream_cmp_ino().
+    */
     if (s1->class != s2->class || (s1->class != &gzip_stream_compress_class &&
                                    s2->class != &gzip_stream_compress_class))
         return iso_stream_cmp_ino(s1, s2, 1);
+
+    /* Both streams apply the same treatment to their input streams */
+    return iso_stream_cmp_ino(iso_stream_get_input_stream(s1, 0),
+                              iso_stream_get_input_stream(s2, 0), 0);
+}   
+
+ 
+static
+int gzip_uncompress_cmp_ino(IsoStream *s1, IsoStream *s2)
+{
+    /* This function may rely on being called by iso_stream_cmp_ino()
+       only with s1, s2 which both point to it as their .cmp_ino() function.
+       It would be a programming error to let any other than 
+       gzip_stream_uncompress_class point to gzip_uncompress_cmp_ino().
+    */
+    if (s1->class != s2->class ||
+        (s1->class != &gzip_stream_uncompress_class &&
+         s2->class != &gzip_stream_uncompress_class))
+        return iso_stream_cmp_ino(s1, s2, 1);
+
+    /* Both streams apply the same treatment to their input streams */
     return iso_stream_cmp_ino(iso_stream_get_input_stream(s1, 0),
                               iso_stream_get_input_stream(s2, 0), 0);
 }   

libisofs/filters/zisofs.c

@@ -130,9 +130,9 @@ int ziso_running_new(ZisofsFilterRuntime **running, int flag)
 
     o->block_size = ziso_block_size;
 #ifdef Libisofs_with_zliB
-    o->buffer_size= compressBound((uLong) ziso_block_size);
+    o->buffer_size = compressBound((uLong) ziso_block_size);
 #else
-    o->buffer_size= 2 * ziso_block_size;
+    o->buffer_size = 2 * ziso_block_size;
 #endif
     o->read_buffer = calloc(o->block_size, 1);
     o->block_buffer = calloc(o->buffer_size, 1);
@@ -381,7 +381,7 @@ int ziso_stream_compress(IsoStream *stream, void *buf, size_t desired)
                     if (todo * 4 > rng->buffer_size)
                         todo = rng->buffer_size / 4;
                     memcpy(rng->block_buffer,
-                           data->block_pointers + 4 * rng->block_pointer_rpos,
+                           data->block_pointers + rng->block_pointer_rpos,
                            todo * 4);
                     rng->buffer_rpos = 0;
                     rng->buffer_fill = todo * 4;
@@ -838,6 +838,9 @@ no_mem:
 static
 int ziso_cmp_ino(IsoStream *s1, IsoStream *s2);
 
+static
+int ziso_uncompress_cmp_ino(IsoStream *s1, IsoStream *s2);
+
 
 IsoStreamIface ziso_stream_compress_class = {
     4,
@@ -868,7 +871,7 @@ IsoStreamIface ziso_stream_uncompress_class = {
     ziso_stream_free,
     ziso_update_size,
     ziso_get_input_stream,
-    ziso_cmp_ino,
+    ziso_uncompress_cmp_ino,
     ziso_clone_stream
 };
 
@@ -876,9 +879,36 @@ IsoStreamIface ziso_stream_uncompress_class = {
 static
 int ziso_cmp_ino(IsoStream *s1, IsoStream *s2)
 {
+    /* This function may rely on being called by iso_stream_cmp_ino()
+       only with s1, s2 which both point to it as their .cmp_ino() function. 
+       It would be a programming error to let any other than
+       ziso_stream_compress_class point to ziso_cmp_ino().
+    */
     if (s1->class != s2->class || (s1->class != &ziso_stream_compress_class &&
                                    s2->class != &ziso_stream_uncompress_class))
         iso_stream_cmp_ino(s1, s2, 1);
+
+    /* Both streams apply the same treatment to their input streams */
+    return iso_stream_cmp_ino(iso_stream_get_input_stream(s1, 0),
+                              iso_stream_get_input_stream(s2, 0), 0);
+}
+
+
+static
+int ziso_uncompress_cmp_ino(IsoStream *s1, IsoStream *s2)
+{
+    /* This function may rely on being called by iso_stream_cmp_ino()
+       only with s1, s2 which both point to it as their .cmp_ino() function. 
+       It would be a programming error to let any other than
+       ziso_stream_uncompress_class point to ziso_uncompress_cmp_ino().
+       This fallback endangers transitivity of iso_stream_cmp_ino().
+    */
+    if (s1->class != s2->class ||
+        (s1->class != &ziso_stream_uncompress_class &&
+         s2->class != &ziso_stream_uncompress_class))
+        iso_stream_cmp_ino(s1, s2, 1);
+
+    /* Both streams apply the same treatment to their input streams */
     return iso_stream_cmp_ino(iso_stream_get_input_stream(s1, 0),
                               iso_stream_get_input_stream(s2, 0), 0);
 }

libisofs/fs_local.c

@@ -1,6 +1,6 @@
 /*
  * Copyright (c) 2007 Vreixo Formoso
- * Copyright (c) 2009 - 2011 Thomas Schmitt
+ * Copyright (c) 2009 - 2017 Thomas Schmitt
  * 
  * This file is part of the libisofs project; you can redistribute it and/or 
  * modify it under the terms of the GNU General Public License version 2 
@@ -30,6 +30,11 @@
 #include <libgen.h>
 #include <string.h>
 
+/* O_BINARY is needed for Cygwin but undefined elsewhere */
+#ifndef O_BINARY
+#define O_BINARY 0
+#endif
+
 static
 int iso_file_source_new_lfs(IsoFileSource *parent, const char *name, 
                             IsoFileSource **src);
@@ -222,7 +227,7 @@ int lfs_open(IsoFileSource *src)
         data->info.dir = opendir(path);
         data->openned = data->info.dir ? 2 : 0;
     } else {
-        data->info.fd = open(path, O_RDONLY);
+        data->info.fd = open(path, O_RDONLY | O_BINARY);
         data->openned = data->info.fd != -1 ? 1 : 0;
     }
     free(path);
@@ -282,6 +287,9 @@ static
 int lfs_read(IsoFileSource *src, void *buf, size_t count)
 {
     _LocalFsFileSource *data;
+    size_t to_read, done = 0;
+    int ret;
+    uint8_t *buf8;
 
     if (src == NULL || buf == NULL) {
         return ISO_NULL_POINTER;
@@ -293,28 +301,28 @@ int lfs_read(IsoFileSource *src, void *buf, size_t count)
     data = src->data;
     switch (data->openned) {
     case 1: /* not dir */
-        {
-            int ret;
-            ret = read(data->info.fd, buf, count);
+        buf8 = (uint8_t *) buf; /* for pointer arithmetic */
+        for (to_read = count; to_read > 0; to_read = count - done) {
+            if (to_read > 1024 * 1024)
+                to_read = 1024 * 1024;
+            ret = read(data->info.fd, buf8 + done, to_read);
             if (ret < 0) {
                 /* error on read */
                 switch (errno) {
                 case EINTR:
-                    ret = ISO_INTERRUPTED;
-                    break;
+                    return ISO_INTERRUPTED;
                 case EFAULT:
-                    ret = ISO_OUT_OF_MEM;
-                    break;
+                    return ISO_OUT_OF_MEM;
                 case EIO:
-                    ret = ISO_FILE_READ_ERROR;
-                    break;
-                default:
-                    ret = ISO_FILE_ERROR;
-                    break;
+                    return ISO_FILE_READ_ERROR;
                 }
+                return ISO_FILE_ERROR;
             }
-            return ret;
+            if (ret == 0) /* EOF */
+        break;
+            done += ret;
         }
+        return done;
     case 2: /* directory */
         return ISO_FILE_IS_DIR;
     default:
@@ -491,14 +499,15 @@ void lfs_free(IsoFileSource *src)
 static 
 int lfs_get_aa_string(IsoFileSource *src, unsigned char **aa_string, int flag)
 {
-    int ret;
-    size_t num_attrs = 0, *value_lengths = NULL, result_len, sret;
+    int ret, no_non_user_perm= 0;
+    size_t num_attrs = 0, *value_lengths = NULL, result_len;
+    ssize_t sret;
     char *path = NULL, **names = NULL, **values = NULL;
     unsigned char *result = NULL;
 
     *aa_string = NULL;
 
-    if ((flag & 3 ) == 3) {
+    if ((flag & 6 ) == 6) { /* Neither ACL nor xattr shall be read */
         ret = 1;
         goto ex;
     }
@@ -512,7 +521,7 @@ int lfs_get_aa_string(IsoFileSource *src, unsigned char **aa_string, int flag)
     }
     ret = aaip_get_attr_list(path, &num_attrs, &names,
                              &value_lengths, &values,
-                             (!(flag & 2)) | 2 | (flag & 4) | 16);
+                             (!(flag & 2)) | 2 | (flag & 4) | (flag & 8) | 16);
     if (ret <= 0) {
         if (ret == -2)
             ret = ISO_AAIP_NO_GET_LOCAL;
@@ -520,23 +529,26 @@ int lfs_get_aa_string(IsoFileSource *src, unsigned char **aa_string, int flag)
             ret = ISO_FILE_ERROR;
         goto ex;
     }
+    if(ret == 2)
+        no_non_user_perm= 1;
+      
     if (num_attrs == 0)
         result = NULL;
     else {
         sret = aaip_encode(num_attrs, names,
                            value_lengths, values, &result_len, &result, 0);
-        if (sret == 0) {
-            ret = ISO_OUT_OF_MEM;
+        if (sret < 0) {
+            ret = sret;
             goto ex;
-         }
+        }
     }
     *aa_string = result;
-    ret = 1;
+    ret = 1 + no_non_user_perm;
 ex:;
     if (path != NULL)
         free(path);
     if (names != NULL || value_lengths != NULL || values != NULL)
-        aaip_get_attr_list(path, &num_attrs, &names, &value_lengths, &values,
+        aaip_get_attr_list(NULL, &num_attrs, &names, &value_lengths, &values,
                            1 << 15); /* free memory */
     return ret;
 }
@@ -858,17 +870,19 @@ int iso_local_get_attrs(char *disk_path, size_t *num_attrs, char ***names,
                              (flag & (1 | 4 | 8 | 32 | (1 << 15))) | 2 | 16);
     if (ret <= 0)
         return ISO_AAIP_NO_GET_LOCAL;
-    return 1;
+    return 1 + (ret == 2);
 }
 
 
-int iso_local_set_attrs(char *disk_path, size_t num_attrs, char **names,
-                        size_t *value_lengths, char **values, int flag)
+int iso_local_set_attrs_errno(char *disk_path, size_t num_attrs, char **names,
+                              size_t *value_lengths, char **values,
+                              int *errnos, int flag)
 {
     int ret;
 
     ret = aaip_set_attr_list(disk_path, num_attrs, names, value_lengths,
-                             values, (flag & (8 | 32 | 64)) | !(flag & 1));
+                             values, errnos,
+                             (flag & (8 | 32 | 64 | 128)) | !(flag & 1));
     if (ret <= 0) {
         if (ret == -1)
             return ISO_OUT_OF_MEM;
@@ -886,6 +900,25 @@ int iso_local_set_attrs(char *disk_path, size_t num_attrs, char **names,
 }
 
 
+int iso_local_set_attrs(char *disk_path, size_t num_attrs, char **names,
+                        size_t *value_lengths, char **values, int flag)
+{
+    int ret;
+    int *errnos = NULL;
+
+    if(num_attrs > 0) {
+      errnos= calloc(num_attrs, sizeof(int));
+      if(errnos == NULL)
+        return ISO_OUT_OF_MEM;
+    }
+    ret= iso_local_set_attrs_errno(disk_path, num_attrs, names, value_lengths,
+                                   values, errnos, flag);
+    if(errnos != NULL)
+      free(errnos);
+    return ret;
+}
+
+
 int iso_local_get_perms_wo_acl(char *disk_path, mode_t *st_mode, int flag)
 {
     struct stat stbuf;

libisofs/fsource.h

@@ -1,6 +1,6 @@
 /*
  * Copyright (c) 2007 Vreixo Formoso
- * Copyright (c) 2009 - 2011 Thomas Schmitt
+ * Copyright (c) 2009 - 2016 Thomas Schmitt
  * 
  * This file is part of the libisofs project; you can redistribute it and/or 
  * modify it under the terms of the GNU General Public License version 2 
@@ -28,15 +28,19 @@
  * Create a new IsoFilesystem to deal with local filesystem.
  * 
  * @return
- *     1 sucess, < 0 error
+ *     1 success, < 0 error
  */
 int iso_local_filesystem_new(IsoFilesystem **fs);
 
 
 /* Rank two IsoFileSource of ifs_class by their eventual old image LBAs.
-   Other IsoFileSource classes will be ranked only roughly.
+ * @param cmp_ret  will return the reply value -1, 0, or 1.
+ * @return         1= *cmp_ret is a valid reply
+ *                 0= not both streams are of ifs_class,
+ *                    *cmp_ret is only a rough estimation.
 */
-int iso_ifs_sections_cmp(IsoFileSource *s1, IsoFileSource *s2, int flag);
+int iso_ifs_sections_cmp(IsoFileSource *s1, IsoFileSource *s2, int *cmp_ret,
+                         int flag);
 
 
 /* Create an independent copy of an ifs_class IsoFileSource.

libisofs/hfsplus.c

@@ -91,7 +91,7 @@ int filesrc_block_and_size(Ecma119Image *t, IsoFileSrc *src,
         *total_size += src->sections[i].size;
         if (pos != src->sections[i].block) {
             iso_msg_submit(t->image->id, ISO_SECT_SCATTERED, 0,
-                      "File sections do not form consequtive array of blocks");
+                      "File sections do not form consecutive array of blocks");
             return ISO_SECT_SCATTERED;
         }
         /* If .size is not aligned to blocks then there is a byte gap.
@@ -137,8 +137,10 @@ int iso_get_hfsplus_name(char *input_charset, int imgid, char *name,
     curlen = ucslen (ucs_name);
     *result = calloc ((curlen * HFSPLUS_MAX_DECOMPOSE_LEN + 1),
                      sizeof (uint16_t));
-    if (*result == NULL)
+    if (*result == NULL) {
+        free(ucs_name);
         return ISO_OUT_OF_MEM;
+    }
 
     for (iptr = ucs_name, optr = *result; *iptr; iptr++)
       {
@@ -208,8 +210,12 @@ int iso_get_hfsplus_name(char *input_charset, int imgid, char *name,
     while (done);
 
     *cmp_name = calloc ((ucslen (*result) + 1), sizeof (uint16_t));
-    if (*cmp_name == NULL)
-      return ISO_OUT_OF_MEM;
+    if (*cmp_name == NULL) {
+        free(ucs_name);
+        free(*result);
+        *result = NULL;
+        return ISO_OUT_OF_MEM;
+    }
 
     for (iptr = *result, optr = *cmp_name; *iptr; iptr++)
       {
@@ -399,7 +405,8 @@ int create_tree(Ecma119Image *t, IsoNode *iso, uint32_t parent_id)
 	    if (cret < 0)
 	      return cret;
 	    pos = pos->next;
-	    t->hfsp_leafs[cleaf].nchildren++;
+	    if (cret > 0)
+	      t->hfsp_leafs[cleaf].nchildren++;
 	  }
       }
     return ISO_SUCCESS;
@@ -1573,7 +1580,7 @@ int mangle_leafs(Ecma119Image *target, int flag)
 int hfsplus_writer_create(Ecma119Image *target)
 {
     int ret;
-    IsoImageWriter *writer;
+    IsoImageWriter *writer = NULL;
     int max_levels;
     int level = 0;
     IsoNode *pos;
@@ -1583,7 +1590,8 @@ int hfsplus_writer_create(Ecma119Image *target)
 
     writer = calloc(1, sizeof(IsoImageWriter));
     if (writer == NULL) {
-        return ISO_OUT_OF_MEM;
+        ret = ISO_OUT_OF_MEM;
+        goto ex;
     }
 
     make_hfsplus_decompose_pages();
@@ -1607,10 +1615,8 @@ int hfsplus_writer_create(Ecma119Image *target)
     target->hfsp_ndirs = 0;
     target->hfsp_cat_id = 16;
     ret = hfsplus_count_tree(target, (IsoNode*)target->image->root);
-    if (ret < 0) {
-        free((char *) writer);
-        return ret;
-    }
+    if (ret < 0)
+        goto ex;
 
     for (i = 0; i < ISO_HFSPLUS_BLESS_MAX; i++)
       target->hfsp_bless_id[i] = 0;
@@ -1620,12 +1626,13 @@ int hfsplus_writer_create(Ecma119Image *target)
 
     target->hfsp_leafs = calloc (target->hfsp_nleafs, sizeof (target->hfsp_leafs[0]));
     if (target->hfsp_leafs == NULL) {
-        return ISO_OUT_OF_MEM;
+        ret = ISO_OUT_OF_MEM;
+        goto ex;
     }
     ret = set_hfsplus_name (target, target->image->volume_id,
                             &target->hfsp_leafs[target->hfsp_curleaf]);
     if (ret < 0)
-        return ret;
+        goto ex;
     target->hfsp_leafs[target->hfsp_curleaf].node = (IsoNode *) target->image->root;
     target->hfsp_leafs[target->hfsp_curleaf].used_size = target->hfsp_leafs[target->hfsp_curleaf].strlen * 2 + 8 + 2 + sizeof (struct hfsplus_catfile_common);
 
@@ -1656,10 +1663,13 @@ int hfsplus_writer_create(Ecma119Image *target)
       {
 	int cret;
 	cret = create_tree(target, pos, 2);
-	if (cret < 0)
-	  return cret;
+	if (cret < 0) {
+	    ret = cret;
+	    goto ex;
+	}
 	pos = pos->next;
-	target->hfsp_leafs[0].nchildren++;
+	if (cret > 0)
+	    target->hfsp_leafs[0].nchildren++;
       }
 
     qsort(target->hfsp_leafs, target->hfsp_nleafs,
@@ -1667,13 +1677,14 @@ int hfsplus_writer_create(Ecma119Image *target)
 
     ret = mangle_leafs(target, 0);
     if (ret < 0)
-        return ret;
+        goto ex;
 
     for (max_levels = 0; target->hfsp_nleafs >> max_levels; max_levels++);
     max_levels += 2;
     target->hfsp_levels = calloc (max_levels, sizeof (target->hfsp_levels[0]));
     if (target->hfsp_levels == NULL) {
-        return ISO_OUT_OF_MEM;
+	ret = ISO_OUT_OF_MEM;
+	goto ex;
     }
 
     target->hfsp_nnodes = 1;
@@ -1683,9 +1694,10 @@ int hfsplus_writer_create(Ecma119Image *target)
       unsigned bytes_rem = cat_node_size - sizeof (struct hfsplus_btnode) - 2;
 
       target->hfsp_levels[level].nodes = calloc ((target->hfsp_nleafs + 1),  sizeof (target->hfsp_levels[level].nodes[0]));
-      if (!target->hfsp_levels[level].nodes)
-	return ISO_OUT_OF_MEM;
-    
+      if (!target->hfsp_levels[level].nodes) {
+	  ret = ISO_OUT_OF_MEM;
+	  goto ex;
+      }
       target->hfsp_levels[level].level_size = 0;  
       for (i = 0; i < target->hfsp_nleafs; i++)
 	{
@@ -1740,8 +1752,10 @@ int hfsplus_writer_create(Ecma119Image *target)
 	level++;
 
 	target->hfsp_levels[level].nodes = calloc (((last_size + 1) / 2),  sizeof (target->hfsp_levels[level].nodes[0]));
-	if (!target->hfsp_levels[level].nodes)
-	  return ISO_OUT_OF_MEM;
+	if (!target->hfsp_levels[level].nodes) {
+	    ret = ISO_OUT_OF_MEM;
+	    goto ex;
+        }
     
 	target->hfsp_levels[level].level_size = 0;  
 
@@ -1776,16 +1790,21 @@ int hfsplus_writer_create(Ecma119Image *target)
 
     if (target->hfsp_nnodes > (cat_node_size - 0x100) * 8)
       {
-	return iso_msg_submit(target->image->id, ISO_MANGLE_TOO_MUCH_FILES, 0,
-			      "HFS+ map nodes aren't implemented");
-
-	return ISO_MANGLE_TOO_MUCH_FILES;
+	iso_msg_submit(target->image->id, ISO_MANGLE_TOO_MUCH_FILES, 0,
+			"HFS+ map nodes aren't implemented");
+	ret = ISO_MANGLE_TOO_MUCH_FILES;
+	goto ex;
       }
 
     /* add this writer to image */
     target->writers[target->nwriters++] = writer;
+    writer = NULL;
 
-    return ISO_SUCCESS;
+    ret = ISO_SUCCESS;
+ex:;
+    if (writer != NULL)
+        free(writer);
+    return ret;
 }
 
 int hfsplus_tail_writer_create(Ecma119Image *target)
@@ -1832,7 +1851,7 @@ struct iso_hfsplus_xinfo_data *iso_hfsplus_xinfo_new(int flag)
 }
 
 /* The iso_node_xinfo_cloner function which gets associated to
- * iso_hfsplus_xinfo_func by iso_init() resp. iso_init_with_flag() via
+ * iso_hfsplus_xinfo_func by iso_init() or iso_init_with_flag() via
  * iso_node_xinfo_make_clonable()
  */
 int iso_hfsplus_xinfo_cloner(void *old_data, void **new_data, int flag)

libisofs/hfsplus_classes.c

@@ -422,6 +422,10 @@ static uint16_t class_page_data[] = {
   0x21, 0x230,
   0x22, 0x230,
   0x23, 0x230,
+  0x00,
+
+  /* End of list */
+  0x00
 };
 
 uint16_t *hfsplus_class_pages[256];
@@ -434,7 +438,7 @@ void make_hfsplus_class_pages()
     uint16_t *rpt, *page_pt;
     int page_count = 0;
 
-    memset(class_pages, 0, 19 * 256);
+    memset(class_pages, 0, 19 * 256 * sizeof(uint16_t));
     for (i = 0; i < 256; i++)
         hfsplus_class_pages[i] = NULL;  
 

libisofs/image.c

@@ -1,6 +1,6 @@
 /*
  * Copyright (c) 2007 Vreixo Formoso
- * Copyright (c) 2009 - 2014 Thomas Schmitt
+ * Copyright (c) 2009 - 2015 Thomas Schmitt
  *
  * This file is part of the libisofs project; you can redistribute it and/or
  * modify it under the terms of the GNU General Public License version 2 
@@ -51,6 +51,8 @@ int iso_imported_sa_new(struct iso_imported_sys_area **boots, int flag)
     b->hppa_kernel_64 = NULL;
     b->hppa_ramdisk = NULL;
 
+    b->alpha_boot_image = NULL;
+
     *boots = b;
     return 1;
 }
@@ -107,6 +109,7 @@ int iso_imported_sa_unref(struct iso_imported_sys_area **boots, int flag)
     LIBISO_FREE_MEM(b->hppa_kernel_32);
     LIBISO_FREE_MEM(b->hppa_kernel_64);
     LIBISO_FREE_MEM(b->hppa_ramdisk);
+    LIBISO_FREE_MEM(b->alpha_boot_image);
     LIBISO_FREE_MEM(b);
     *boots = NULL;
     return 1;
@@ -177,12 +180,17 @@ int iso_image_new(const char *name, IsoImage **image)
          img->mips_boot_file_paths[i] = NULL;
     img->sparc_core_node = NULL;
     img->hppa_cmdline= NULL;
-    img->hppa_bootloader= NULL;
-    img->hppa_kernel_32= NULL;
-    img->hppa_kernel_64= NULL;
-    img->hppa_ramdisk= NULL;
+    img->hppa_bootloader = NULL;
+    img->hppa_kernel_32 = NULL;
+    img->hppa_kernel_64 = NULL;
+    img->hppa_ramdisk = NULL;
+    img->alpha_boot_image = NULL;
+    img->import_src = NULL;
     img->builder_ignore_acl = 1;
     img->builder_ignore_ea = 1;
+    img->truncate_mode = 1;
+    img->truncate_length = LIBISOFS_NODE_NAME_MAX;
+    img->truncate_buffer[0] = 0;
     img->inode_counter = 0;
     img->used_inodes = NULL;
     img->used_inodes_start = 0;
@@ -195,6 +203,7 @@ int iso_image_new(const char *name, IsoImage **image)
         img->hfsplus_blessed[i] = NULL;
     img->collision_warnings = 0;
     img->imported_sa_info = NULL;
+    img->blind_on_local_get_attrs = 0;
 
     *image = img;
     return ISO_SUCCESS;
@@ -239,6 +248,10 @@ void iso_image_unref(IsoImage *image)
         if (image->sparc_core_node != NULL)
             iso_node_unref((IsoNode *) image->sparc_core_node);
         iso_image_set_hppa_palo(image, NULL, NULL, NULL, NULL, NULL, 1);
+        if (image->alpha_boot_image != NULL)
+            free(image->alpha_boot_image);
+        if (image->import_src != NULL)
+            iso_data_source_unref(image->import_src);
         free(image->volset_id);
         free(image->volume_id);
         free(image->publisher_id);
@@ -598,6 +611,15 @@ void iso_image_set_ignore_aclea(IsoImage *image, int what)
 {
     image->builder_ignore_acl = (what & 1);
     image->builder_ignore_ea = !!(what & 2);
+    image->builder_take_all_ea = !!(what & 8);
+}
+
+
+int iso_image_get_ignore_aclea(IsoImage *image)
+{
+    return image->builder_ignore_acl |
+           (image->builder_ignore_ea << 1) |
+           (image->builder_take_all_ea << 3);
 }
 
 
@@ -969,9 +991,9 @@ int iso_image_get_sparc_core(IsoImage *img, IsoFile **sparc_core, int flag)
                  Else only the non-NULL parameters of this call have an effect.
 */
 static int hppa_palo_set_path(IsoImage *img, char *path, char **target,
-                              int flag)
+                              char *what, int flag)
 {
-    int ret;
+    int ret, err;
     IsoNode *node;
     IsoFile *file;
 
@@ -986,16 +1008,19 @@ static int hppa_palo_set_path(IsoImage *img, char *path, char **target,
         return ret;
     if (ret == 0) {
         iso_msg_submit(img->id, ISO_BOOT_FILE_MISSING, 0,
-                       "Cannot find in ISO image: HP-PA file '%s'", path);
+                       "Cannot find in ISO image: %s file '%s'", what, path);
         return ISO_BOOT_FILE_MISSING;
     }
     if (iso_node_get_type(node) != LIBISO_FILE) {
-        iso_msg_submit(img->id, ISO_HPPA_PALO_NOTREG, 0,
-                       "HP-PA PALO file is not a data file: '%s'", path);
-        return ISO_HPPA_PALO_NOTREG;
+        err = ISO_HPPA_PALO_NOTREG;
+        if (strncmp(what, "DEC Alpha", 9) == 0)
+            err = ISO_ALPHA_BOOT_NOTREG;
+        iso_msg_submit(img->id, err, 0,
+                       "%s file is not a data file: '%s'", what, path);
+        return err;
     }
     file = (IsoFile *) node;
-    if (!file->explicit_weight)
+    if (!(file->explicit_weight || file->from_old_session))
         file->sort_weight = 2;
     return ISO_SUCCESS;
 }
@@ -1012,21 +1037,25 @@ int iso_image_set_hppa_palo(IsoImage *img, char *cmdline, char *bootloader,
                             int flag)
 {
     int ret;
+    static char *what = "HP-PA PALO";
 
     if (cmdline != NULL || (flag & 1))
         if (iso_clone_mgtd_mem(cmdline, &(img->hppa_cmdline), 0) < 0)
             return ISO_OUT_OF_MEM;
-    ret = hppa_palo_set_path(img, bootloader, &(img->hppa_bootloader),
+    ret = hppa_palo_set_path(img, bootloader, &(img->hppa_bootloader), what,
                              flag & 1);
     if (ret < 0)
         return ret;
-    ret = hppa_palo_set_path(img, kernel_32, &(img->hppa_kernel_32), flag & 1);
+    ret = hppa_palo_set_path(img, kernel_32, &(img->hppa_kernel_32), what,
+                             flag & 1);
     if (ret < 0)
         return ret;
-    ret = hppa_palo_set_path(img, kernel_64, &(img->hppa_kernel_64), flag & 1);
+    ret = hppa_palo_set_path(img, kernel_64, &(img->hppa_kernel_64), what,
+                             flag & 1);
     if (ret < 0)
         return ret;
-    ret = hppa_palo_set_path(img, ramdisk, &(img->hppa_ramdisk), flag & 1);
+    ret = hppa_palo_set_path(img, ramdisk, &(img->hppa_ramdisk), what,
+                             flag & 1);
     if (ret < 0)
         return ret;
     return ISO_SUCCESS;
@@ -1046,3 +1075,78 @@ int iso_image_get_hppa_palo(IsoImage *img, char **cmdline, char **bootloader,
 }
 
 
+/* API */
+int iso_image_set_alpha_boot(IsoImage *img, char *boot_loader_path, int flag)
+{
+    int ret;
+
+    ret = hppa_palo_set_path(img, boot_loader_path, &(img->alpha_boot_image),
+                             "DEC Alpha Bootloader", 1);
+    if (ret < 0)
+        return ret;
+    return ISO_SUCCESS;
+}
+
+
+/* API */
+int iso_image_get_alpha_boot(IsoImage *img, char **boot_loader_path)
+{
+    *boot_loader_path = img->alpha_boot_image;
+    return ISO_SUCCESS;
+}
+
+
+/* API */
+int iso_image_set_truncate_mode(IsoImage *img, int mode, int length)
+{
+    if (mode < 0 || mode > 1)
+        return ISO_WRONG_ARG_VALUE;
+    if (length < 64 || length > LIBISOFS_NODE_NAME_MAX)
+        return ISO_WRONG_ARG_VALUE;
+    img->truncate_mode = mode;
+    img->truncate_length = length;
+    return ISO_SUCCESS;
+}
+
+/* API */
+int iso_image_get_truncate_mode(IsoImage *img, int *mode, int *length)
+{
+    *mode = img->truncate_mode;
+    *length = img->truncate_length;
+    return ISO_SUCCESS;
+}
+
+/* Warning: Not thread-safe */
+int iso_image_truncate_name(IsoImage *image, const char *name, char **namept,
+                            int flag)
+{
+    int ret;
+
+    if (name == NULL) 
+        return ISO_NULL_POINTER;
+
+    if ((int) strlen(name) <= image->truncate_length) {
+        *namept = (char *) name;
+        return ISO_SUCCESS;
+    }
+    *namept = image->truncate_buffer;
+    if (name != image->truncate_buffer)
+        strncpy(image->truncate_buffer, name, 4095);
+    image->truncate_buffer[4095] = 0;
+    ret = iso_truncate_rr_name(image->truncate_mode, image->truncate_length,
+                               image->truncate_buffer, 0);
+    return ret;
+}
+
+
+/* API */
+int iso_image_was_blind_attrs(IsoImage *image, int flag)
+{
+    int ret;
+
+    ret = image->blind_on_local_get_attrs;
+    if (flag & 1)
+        image->blind_on_local_get_attrs = 0;
+    return ret;
+}
+

libisofs/image.h

@@ -1,6 +1,6 @@
 /*
  * Copyright (c) 2007 Vreixo Formoso
- * Copyright (c) 2009 - 2014 Thomas Schmitt
+ * Copyright (c) 2009 - 2017 Thomas Schmitt
  * 
  * This file is part of the libisofs project; you can redistribute it and/or 
  * modify it under the terms of the GNU General Public License version 2 
@@ -31,7 +31,7 @@
  * Global objects such as the message_queues must belogn to that
  * context. Thus we will have, for example, a msg queue per image,
  * so images are completelly independent and can be managed together.
- * (Usefull, for example, in Multiple-Document-Interface GUI apps.
+ * (Useful, for example, in Multiple-Document-Interface GUI apps.
  * [The stuff we have in init belongs really to image!]
  */
 
@@ -64,6 +64,8 @@ struct Iso_Image
     /* Eventually loaded system area data, or NULL */
     char *system_area_data;
     /* Prescribed/detected options, see iso_write_opts_set_system_area() */
+    /* >>> Needs to be coordinated with .imported_sa_info->system_area_options
+    */
     int system_area_options;
 
    /*
@@ -88,6 +90,9 @@ struct Iso_Image
     char *hppa_kernel_64;
     char *hppa_ramdisk;
 
+    /* Absolute DEC Alpha boot image path in the ISO image */
+    char *alpha_boot_image;
+
     /* image identifier, for message origin identifier */
     int id;
 
@@ -96,6 +101,11 @@ struct Iso_Image
      */
     IsoFilesystem *fs;
 
+    /**
+     * Block storage of imported ISO if demanded by IsoReadOpts.
+     */
+    IsoDataSource *import_src;
+
     /*
      * Default builder to use when adding files to the image tree.
      */
@@ -135,6 +145,12 @@ struct Iso_Image
      */
     unsigned int builder_ignore_ea : 1;
 
+    /**
+     * If not builder_ignore_ea : import all xattr namespaces from local
+     *                            filesystem, not only "user.
+     */
+    unsigned int builder_take_all_ea : 1;
+
     /**
      * Files to exclude. Wildcard support is included.
      */
@@ -150,6 +166,20 @@ struct Iso_Image
     /* TODO
     enum iso_replace_mode (*confirm_replace)(IsoFileSource *src, IsoNode *node);
     */
+
+    /**
+     * What to do in case of name longer than truncate_length:
+     *  0= throw FAILURE
+     *  1= truncate to truncate_length with MD5 of whole name at end
+     */
+    int truncate_mode;
+    int truncate_length;
+
+    /**
+     * This is a convenience buffer for name truncation during image
+     * manipulation where libisofs is not thread-safe anyway.
+     */
+    char truncate_buffer[4096];
     
     /**
      * When this is not NULL, it is a pointer to a function that will
@@ -217,9 +247,22 @@ struct Iso_Image
     /* Contains the assessment of boot aspects of the loaded image */
     struct iso_imported_sys_area *imported_sa_info;
 
+    /* Whether some local filesystem xattr namespace could not be explored
+     * during node building.
+     */
+    int blind_on_local_get_attrs;
+
 };
 
 
+/* Apply truncation mode to name, using image->truncate_buffer to perform
+   truncation if needed.
+
+   Warning: Not thread-safe !
+*/
+int iso_image_truncate_name(IsoImage *image, const char *name, char **namept,
+                            int flag);
+
     
 /* Collect the bitmap of used inode numbers in the range of
    _ImageFsData.used_inodes_start + ISO_USED_INODE_RANGE
@@ -385,6 +428,10 @@ struct iso_imported_sys_area {
     char *hppa_ramdisk;
     char *hppa_bootloader;
 
+    uint64_t alpha_boot_image_size;
+    uint64_t alpha_boot_image_adr;
+    char *alpha_boot_image;
+
     /* Some block addresses of active and first session:
          PVD, L Pathtable, Opt L, M Pathtable, Opt M, root directory
     */

libisofs/joliet.c

@@ -1,7 +1,7 @@
 /*
  * Copyright (c) 2007 Vreixo Formoso
  * Copyright (c) 2007 Mario Danic
- * Copyright (c) 2011-2014 Thomas Schmitt
+ * Copyright (c) 2011-2018 Thomas Schmitt
  *
  * This file is part of the libisofs project; you can redistribute it and/or
  * modify it under the terms of the GNU General Public License version 2 
@@ -401,6 +401,20 @@ int joliet_create_mangled_name(uint16_t *dest, uint16_t *src, int digits,
     return ISO_SUCCESS;
 }
 
+/*
+ * From Joliet specs:
+ * "ISO 9660 (Section 7.5.1) states that the sum of the following shall not
+ *  exceed 30:
+ *  - If there is a file name, the length of the file name.
+ *  - If there is a file name extension, the length of the file name extension.
+ *  On Joliet compliant media, however, the sum as calculated above shall not
+ *  exceed 128 [bytes], to allow for longer file identifiers."
+ *
+ * I.e. the dot does not count.
+ *
+ * (We have an option to lift the limit from 64*2 to 103*2, which is the
+ *  maximum to fit into an ISO 9660 directory record.)
+ */
 static
 int mangle_single_dir(Ecma119Image *t, JolietNode *dir)
 {
@@ -481,25 +495,26 @@ int mangle_single_dir(Ecma119Image *t, JolietNode *dir)
                 ext = dot + 1;
 
                 extlen = ucslen(ext);
-                max = maxchar + 1 - extlen - 1 - digits;
+                max = maxchar - extlen - digits;
                 if (max <= 0) {
-                    /* this can happen if extension is too long */
-                    if (extlen + max > 3) {
+                    /*
+                     * This can happen if the extension is too long.
+                     * Reduce its length, to give name at least one
+                     * original character, if it has any.
+                     */
+                    max = (dot > full_name);
+                    extlen = maxchar - max - digits;
+                    if (extlen < 3) {
                         /*
-                         * reduce extension len, to give name an extra char
-                         * note that max is negative or 0
-                         */
-                        extlen = extlen + max - 1;
-                        ext[extlen] = 0;
-                        max = maxchar + 2 - extlen - 1 - digits;
-                    } else {
-                        /*
-                         * error, we don't support extensions < 3
-                         * This can't happen with current limit of digits.
+                         * error, we do not reduce extensions to length < 3
+                         *
+                         * This cannot happen with current limit of digits
+                         * because maxchar is at least 64 and digits at most 7.
                          */
                         ret = ISO_ERROR;
                         goto mangle_cleanup;
                     }
+                    ext[extlen] = 0;
                 }
                 /* ok, reduce name by digits */
                 if (name + max < dot) {
@@ -508,10 +523,10 @@ int mangle_single_dir(Ecma119Image *t, JolietNode *dir)
             } else {
                 /* Directory, or file without extension */
                 if (children[i]->type == JOLIET_DIR) {
-                    max = maxchar + 1 - digits;
+                    max = maxchar - digits;
                     dot = NULL; /* dots have no meaning in dirs */
                 } else {
-                    max = maxchar + 1 - digits;
+                    max = maxchar - digits;
                 }
                 name = full_name;
                 if ((size_t) max < ucslen(name)) {

libisofs/libiso_msgs.c

@@ -1,7 +1,7 @@
 
 /* libiso_msgs   (generated from libdax_msgs : Fri Feb 22 19:42:52 CET 2008)
    Message handling facility of libisofs.
-   Copyright (C) 2006 - 2008 Thomas Schmitt <scdbackup@gmx.net>,
+   Copyright (C) 2006 - 2016 Thomas Schmitt <scdbackup@gmx.net>,
    provided under GPL version 2 or later
 */
 
@@ -33,14 +33,13 @@ static int libiso_msgs_item_new(struct libiso_msgs_item **item,
  int ret;
  struct libiso_msgs_item *o;
  struct timeval tv;
- struct timezone tz;
 
  (*item)= o= 
            (struct libiso_msgs_item *) malloc(sizeof(struct libiso_msgs_item));
  if(o==NULL)
    return(-1);
  o->timestamp= 0.0;
- ret= gettimeofday(&tv,&tz);
+ ret= gettimeofday(&tv, NULL);
  if(ret==0)
    o->timestamp= tv.tv_sec+0.000001*tv.tv_usec;
  o->process_id= getpid();

libisofs/libiso_msgs.h

@@ -1,7 +1,7 @@
 
 /* libiso_msgs   (generated from libdax_msgs : Fri Feb 22 19:42:52 CET 2008)
    Message handling facility of libisofs.
-   Copyright (C) 2006-2008 Thomas Schmitt <scdbackup@gmx.net>,
+   Copyright (C) 2006-2016 Thomas Schmitt <scdbackup@gmx.net>,
    provided under GPL version 2 or later
 */
 
@@ -221,7 +221,7 @@ struct libiso_msgs_item;
 */
 #define LIBISO_MSGS_SEV_ABORT                                        0x71000000
 
-/** A severity to exclude resp. discard any possible message.
+/** A severity to exclude or discard any possible message.
     Do not use this severity for submitting.
 */
 #define LIBISO_MSGS_SEV_NEVER                                        0x7fffffff
@@ -266,8 +266,9 @@ int libiso_msgs_new(struct libiso_msgs **m, int flag);
 
 /** Destroy a message handling facility and all its eventual messages.
     The submitted pointer gets set to NULL.
-    Actually only the last destroy call of all offical references to the object
-    will really dispose it. All others just decrement the reference counter.
+    Actually only the last destroy call of all official references to the
+    object will really dispose it. All others just decrement the reference
+    counter.
     Call this function only with official reference pointers obtained by
     libiso_msgs_new() or libiso_msgs_refer(), and only once per such pointer.
     @param flag Bitfield for control purposes (unused yet, submit 0)
@@ -510,7 +511,7 @@ Range "scdbackup"          :  0x00020000 to 0x0002ffff
  0x00020148 (SORRY,HIGH)   = Cannot write desired amount of data
  0x00020149 (SORRY,HIGH)   = Unsuitable filetype for pseudo-drive
  0x0002014a (SORRY,HIGH)   = Cannot read desired amount of data
- 0x0002014b (SORRY,HIGH)   = Drive is already registered resp. scanned
+ 0x0002014b (SORRY,HIGH)   = Drive is already registered and scanned
  0x0002014c (FATAL,HIGH)   = Emulated drive caught in SCSI function
  0x0002014d (SORRY,HIGH)   = Asynchromous SCSI error
  0x0002014f (SORRY,HIGH)   = Timeout with asynchromous SCSI command

libisofs/libisofs.ver

@@ -5,6 +5,7 @@ aaip_xinfo_func;
 el_torito_get_bootable;
 el_torito_get_boot_media_type;
 el_torito_get_boot_platform_id;
+el_torito_get_full_load;
 el_torito_get_id_string;
 el_torito_get_isolinux_options;
 el_torito_get_load_seg;
@@ -13,6 +14,7 @@ el_torito_get_selection_crit;
 el_torito_patch_isolinux_image;
 el_torito_seems_boot_info_table;
 el_torito_set_boot_platform_id;
+el_torito_set_full_load;
 el_torito_set_id_string;
 el_torito_set_isolinux_options;
 el_torito_set_load_seg;
@@ -68,6 +70,7 @@ iso_filesystem_ref;
 iso_filesystem_unref;
 iso_finish;
 iso_fs_global_id;
+iso_generate_gpt_guid;
 iso_get_local_charset;
 iso_get_messenger;
 iso_gzip_get_refcounts;
@@ -75,8 +78,13 @@ iso_hfsplus_xinfo_func;
 iso_hfsplus_xinfo_new;
 iso_image_add_boot_image;
 iso_image_add_mips_boot_file;
+iso_image_add_new_dir;
+iso_image_add_new_file;
+iso_image_add_new_special;
+iso_image_add_new_symlink;
 iso_image_attach_data;
 iso_image_create_burn_source;
+iso_image_dir_get_node;
 iso_image_filesystem_new;
 iso_image_fs_get_abstract_file_id;
 iso_image_fs_get_application_id;
@@ -90,6 +98,7 @@ iso_image_fs_get_volume_id;
 iso_image_generator_is_running;
 iso_image_get_abstract_file_id;
 iso_image_get_all_boot_imgs;
+iso_image_get_alpha_boot;
 iso_image_get_app_use;
 iso_image_get_application_id;
 iso_image_get_attached_data;
@@ -99,6 +108,7 @@ iso_image_get_boot_image;
 iso_image_get_copyright_file_id;
 iso_image_get_data_preparer_id;
 iso_image_get_hppa_palo;
+iso_image_get_ignore_aclea;
 iso_image_get_mips_boot_files;
 iso_image_get_msg_id;
 iso_image_get_publisher_id;
@@ -108,6 +118,7 @@ iso_image_get_session_md5;
 iso_image_get_sparc_core;
 iso_image_get_system_area;
 iso_image_get_system_id;
+iso_image_get_truncate_mode;
 iso_image_get_volset_id;
 iso_image_get_volume_id;
 iso_image_give_up_mips_boot;
@@ -115,11 +126,13 @@ iso_image_hfsplus_bless;
 iso_image_hfsplus_get_blessed;
 iso_image_import;
 iso_image_new;
+iso_image_path_to_node;
 iso_image_ref;
 iso_image_remove_boot_image;
 iso_image_report_el_torito;
 iso_image_report_system_area;
 iso_image_set_abstract_file_id;
+iso_image_set_alpha_boot;
 iso_image_set_app_use;
 iso_image_set_application_id;
 iso_image_set_biblio_file_id;
@@ -130,15 +143,22 @@ iso_image_set_copyright_file_id;
 iso_image_set_data_preparer_id;
 iso_image_set_hppa_palo;
 iso_image_set_ignore_aclea;
+iso_image_set_node_name;
 iso_image_set_publisher_id;
 iso_image_set_sparc_core;
 iso_image_set_system_id;
+iso_image_set_truncate_mode;
 iso_image_set_volset_id;
 iso_image_set_volume_id;
+iso_image_tree_clone;
 iso_image_unref;
 iso_image_update_sizes;
+iso_image_was_blind_attrs;
 iso_init;
 iso_init_with_flag;
+iso_interval_reader_destroy;
+iso_interval_reader_new;
+iso_interval_reader_read;
 iso_lib_is_compatible;
 iso_lib_version;
 iso_local_attr_support;
@@ -147,6 +167,7 @@ iso_local_get_attrs;
 iso_local_get_perms_wo_acl;
 iso_local_set_acl_text;
 iso_local_set_attrs;
+iso_local_set_attrs_errno;
 iso_md5_clone;
 iso_md5_compute;
 iso_md5_end;
@@ -214,11 +235,13 @@ iso_read_image_features_has_joliet;
 iso_read_image_features_has_rockridge;
 iso_read_opts_auto_input_charset;
 iso_read_opts_free;
+iso_read_opts_keep_import_src;
 iso_read_opts_load_system_area;
 iso_read_opts_new;
 iso_read_opts_set_default_gid;
 iso_read_opts_set_default_permissions;
 iso_read_opts_set_default_uid;
+iso_read_opts_set_ecma119_map;
 iso_read_opts_set_input_charset;
 iso_read_opts_set_new_inos;
 iso_read_opts_set_no_aaip;
@@ -274,6 +297,7 @@ iso_tree_set_ignore_hidden;
 iso_tree_set_ignore_special;
 iso_tree_set_replace_mode;
 iso_tree_set_report_callback;
+iso_truncate_leaf_name;
 iso_util_decode_md5_tag;
 iso_write_opts_attach_jte;
 iso_write_opts_detach_jte;
@@ -290,6 +314,8 @@ iso_write_opts_set_allow_longer_paths;
 iso_write_opts_set_allow_lowercase;
 iso_write_opts_set_always_gmt;
 iso_write_opts_set_appendable;
+iso_write_opts_set_appended_as_apm;
+iso_write_opts_set_appended_as_gpt;
 iso_write_opts_set_default_dir_mode;
 iso_write_opts_set_default_file_mode;
 iso_write_opts_set_default_gid;
@@ -300,12 +326,14 @@ iso_write_opts_set_disc_label;
 iso_write_opts_set_efi_bootp;
 iso_write_opts_set_fat;
 iso_write_opts_set_fifo_size;
+iso_write_opts_set_gpt_guid;
 iso_write_opts_set_hardlinks;
 iso_write_opts_set_hfsp_block_size;
 iso_write_opts_set_hfsp_serial_number;
 iso_write_opts_set_hfsplus;
 iso_write_opts_set_iso1999;
 iso_write_opts_set_iso_level;
+iso_write_opts_set_iso_mbr_part_type;
 iso_write_opts_set_joliet;
 iso_write_opts_set_joliet_long_names;
 iso_write_opts_set_joliet_longer_paths;
@@ -318,6 +346,7 @@ iso_write_opts_set_omit_version_numbers;
 iso_write_opts_set_output_charset;
 iso_write_opts_set_overwrite_buf;
 iso_write_opts_set_part_offset;
+iso_write_opts_set_part_like_isohybrid;
 iso_write_opts_set_partition_img;
 iso_write_opts_set_prep_img;
 iso_write_opts_set_pvd_times;

libisofs/make_isohybrid_mbr.c

@@ -1,3 +1,13 @@
+/*
+ * Copyright (c) 2008 - 2015 Thomas Schmitt
+ * with special credits to H. Peter Anvin for isohybrid
+ * and to Matthew Garrett for isohybrid with GPT and APM 
+ *
+ * This file is part of the libisofs project; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License version 2
+ * or later as published by the Free Software Foundation.
+ * See COPYING file for details.
+ */
 
 #ifdef HAVE_CONFIG_H
 #include "../config.h"
@@ -27,6 +37,8 @@
 #include "ecma119.h"
 #include "eltorito.h"
 #include "system_area.h"
+#include "image.h"
+#include "messages.h"
 
 
 /* This code stems from syslinux-3.72/utils/isohybrid, a perl script
@@ -50,7 +62,7 @@ license from above stem licenses, typically from LGPL.
 In case its generosity is needed, here is the 2-clause BSD license:
 
 make_isohybrid_mbr.c is copyright 2002-2008 H. Peter Anvin
-                              and 2008-2014 Thomas Schmitt
+                              and 2008-2015 Thomas Schmitt
 
 1. Redistributions of source code must retain the above copyright notice,
    this list of conditions and the following disclaimer.
@@ -146,7 +158,6 @@ int make_isohybrid_mbr(int bin_lba, int *img_blocks, char *mbr, int flag)
 
     /* For generating a weak random number */
     struct timeval tv;
-    struct timezone tz;
 
     if (bin_lba < 0 || bin_lba >= (1 << 29))
         return (0); /* 1 TB limit of signed 32 bit addressing of 512 byte blocks */
@@ -201,7 +212,7 @@ int make_isohybrid_mbr(int bin_lba, int *img_blocks, char *mbr, int flag)
      from. An environment variable ?
      125: Whatever, i use some 32-bit random value with no crypto strength.
      */
-    gettimeofday(&tv, &tz);
+    gettimeofday(&tv, NULL);
     id = 0xffffffff & (tv.tv_sec ^ (tv.tv_usec * 2000));
 
     /*
@@ -390,12 +401,15 @@ int lba512chs_to_buf(char **wpt, off_t lba, int head_count, int sector_count)
 
 /* Find out whether GPT and APM are desired
    flag bit0 = register APM and GPT requests in Ecma119Image
+        bit1 = do not asses and register APM
+        bit2 = do not register overall GPT partition
 */
 int assess_isohybrid_gpt_apm(Ecma119Image *t, int *gpt_count, int gpt_idx[128],
                              int *apm_count, int flag)
 {
     int i, ilx_opts, j, ret, num_img;
     uint32_t block_count;
+    uint64_t start_block;
     uint8_t gpt_name[72];
     static uint8_t zero_uuid[16] = {0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0};
     static uint8_t basic_data_uuid[16] = {
@@ -418,11 +432,13 @@ int assess_isohybrid_gpt_apm(Ecma119Image *t, int *gpt_count, int gpt_idx[128],
         num_img = 0;
     for (i = 0; i < num_img; i++) {
         ilx_opts = t->catalog->bootimages[i]->isolinux_options;
-        if (((ilx_opts >> 2) & 63) == 1 || ((ilx_opts >> 2) & 63) == 2) {
+        if ((((ilx_opts >> 2) & 63) == 1 || ((ilx_opts >> 2) & 63) == 2) &&
+           !(t->boot_appended_idx[i] >= 0 && t->opts->appended_as_gpt)) {
             if (*gpt_count < 128)
-                gpt_idx[*gpt_count]= i;
+                gpt_idx[*gpt_count] = i;
             (*gpt_count)++;
-            if ((flag & 1) && t->bootsrc[i] != NULL) {
+            if ((flag & 1) &&
+                (t->bootsrc[i] != NULL || t->boot_appended_idx[i] >= 0)) {
                 /* Register GPT entry */
                 memset(gpt_name, 0, 72);
                 sprintf((char *) gpt_name, "ISOHybrid%d", *gpt_count);
@@ -431,27 +447,44 @@ int assess_isohybrid_gpt_apm(Ecma119Image *t, int *gpt_count, int gpt_idx[128],
                     uuid = hfs_uuid;
                 else
                     uuid = basic_data_uuid;
-                block_count = 0;
-                for (j = 0; j < t->bootsrc[i]->nsections; j++)
-                     block_count += t->bootsrc[i]->sections[j].size / 2048;
+                if (t->boot_appended_idx[i] >= 0) {
+                    block_count = t->appended_part_size[
+                                                      t->boot_appended_idx[i]];
+                    start_block = ((uint64_t) t->appended_part_start[
+                                                 t->boot_appended_idx[i]]) * 4;
+                } else {
+                    block_count = 0;
+                    for (j = 0; j < t->bootsrc[i]->nsections; j++)
+                        block_count += t->bootsrc[i]->sections[j].size / 2048;
+                    start_block = ((uint64_t) t->bootsrc[i]->sections[0].block)
+                                  * 4;
+                }
                 ret = iso_quick_gpt_entry(
                             t->gpt_req, &(t->gpt_req_count),
-                            ((uint64_t) t->bootsrc[i]->sections[0].block) * 4,
-                            ((uint64_t) block_count) * 4,
+                            start_block, ((uint64_t) block_count) * 4,
                             uuid, zero_uuid, gpt_flags, (uint8_t *) gpt_name);
                 if (ret < 0)
                     return ret;
             }
         }
-        if (ilx_opts & 256) {
+        if ((ilx_opts & 256) && !(flag & 2)) {
             (*apm_count)++;
-            if ((flag & 1) && t->bootsrc[i] != NULL) {
+            if ((flag & 1) &&
+                (t->bootsrc[i] != NULL || t->boot_appended_idx[i] >= 0)) {
                 /* Register APM entry */
-                block_count = 0;
-                for (j = 0; j < t->bootsrc[i]->nsections; j++)
-                     block_count += t->bootsrc[i]->sections[j].size / 2048;
+                if (t->boot_appended_idx[i] >= 0) {
+                    block_count = t->appended_part_size[
+                                                      t->boot_appended_idx[i]];
+                    start_block = t->appended_part_start[
+                                                      t->boot_appended_idx[i]];
+                } else {
+                    block_count = 0;
+                    for (j = 0; j < t->bootsrc[i]->nsections; j++)
+                        block_count += t->bootsrc[i]->sections[j].size / 2048;
+                    start_block = t->bootsrc[i]->sections[0].block;
+                }
                 ret = iso_quick_apm_entry(t->apm_req, &(t->apm_req_count),
-                                          t->bootsrc[i]->sections[0].block,
+                                          (uint32_t) start_block,
                                           block_count, "EFI", "Apple_HFS");
                 if (ret < 0)
                     return ret;
@@ -461,7 +494,12 @@ int assess_isohybrid_gpt_apm(Ecma119Image *t, int *gpt_count, int gpt_idx[128],
             }
         }
     }
-    if ((flag & 1) && *gpt_count > 0) {
+    if (*gpt_count > 0 && !(flag & 4)) {
+        (*gpt_count)++;
+        if (*gpt_count < 128)
+            gpt_idx[*gpt_count] = -1;
+    }
+    if ((flag & 1) && *gpt_count > 0 && !(flag & 4)) {
         /* Register overall GPT partition */
         memset(gpt_name, 0, 72);
         sprintf((char *) gpt_name, "ISOHybrid");
@@ -526,19 +564,28 @@ static int insert_apm_head(uint8_t *buf, int apm_count)
 static int gpt_images_as_mbr_partitions(Ecma119Image *t, char *wpt,
                                         int gpt_idx[128], int *gpt_cursor)
 {
-    int ilx_opts;
+    int ilx_opts, skip = 0;
     off_t hd_blocks;
     static uint8_t dummy_chs[3] = {
         0xfe, 0xff, 0xff, 
     };
 
+    if (gpt_idx[*gpt_cursor] < 0)
+        skip = 1;
+    else if (t->bootsrc[gpt_idx[*gpt_cursor]] == NULL)
+        skip = 1;
+    if (skip) {
+        (*gpt_cursor)++;
+        return 2;
+    }
+
     wpt[0] = 0;
     memcpy(wpt + 1, dummy_chs, 3);
     ilx_opts = t->catalog->bootimages[gpt_idx[*gpt_cursor]]->isolinux_options;
     if (((ilx_opts >> 2) & 63) == 2)
         wpt[4] = 0x00; /* HFS gets marked as "Empty" */
     else
-        ((unsigned char *) wpt)[4] = 0xef; /* "EFI (FAT-12/16/" */
+        ((unsigned char *) wpt)[4] = 0xef; /* "EFI (FAT-12/16)" */
 
     memcpy(wpt + 5, dummy_chs, 3);
 
@@ -557,8 +604,29 @@ static int gpt_images_as_mbr_partitions(Ecma119Image *t, char *wpt,
 }
 
 
+/* For generating a weak random number */
+static uint32_t iso_make_mbr_id(Ecma119Image *t, int flag)
+{
+    uint32_t id;
+    struct timeval tv;
+
+    if(t->opts->vol_uuid[0]) {
+        id = iso_crc32_gpt((unsigned char *) t->opts->vol_uuid, 16, 0);
+    } else if(t->opts->vol_modification_time > 0) {
+        id = iso_crc32_gpt((unsigned char *) &(t->opts->vol_modification_time),
+                           sizeof(time_t), 0);
+    } else {
+        gettimeofday(&tv, NULL);
+        id = 0xffffffff & (tv.tv_sec ^ (tv.tv_usec * 2000));
+    }
+    return id;
+}
+
+
 /*
  * @param flag  bit0= make own random MBR Id from current time
+ *                    >>> or from overridden modification time
+ *              bit1= create protective MBR as of UEFI/GPT specs
  */
 int make_isolinux_mbr(uint32_t *img_blocks, Ecma119Image *t,
                       int part_offset, int part_number, int fs_type,
@@ -569,12 +637,22 @@ int make_isolinux_mbr(uint32_t *img_blocks, Ecma119Image *t,
     char *wpt;
     uint32_t boot_lba;
     int head_count, sector_count, ret;
-    int gpt_count = 0, gpt_idx[128], apm_count = 0, gpt_cursor;
-    /* For generating a weak random number */
-    struct timeval tv;
-    struct timezone tz;
+    int gpt_count = 0, gpt_idx[128], apm_count = 0, gpt_cursor, i;
 
-    hd_img_blocks = ((off_t) *img_blocks) * (off_t) 4;
+    if (t->bootsrc[0] == NULL)
+        return iso_msg_submit(t->image->id, ISO_BOOT_IMAGE_NOT_VALID, 0,
+      "Cannot refer by isohybrid MBR to data outside of ISO 9660 filesystem.");
+
+    for (i = 0; i < 128; i++)
+        gpt_idx[i] = -1;
+
+    if (flag & 2) {
+        part_number = 1;
+        part_offset = 1;
+    }
+
+    hd_img_blocks = ((off_t) *img_blocks) * (off_t) 4 -
+                    t->post_iso_part_pad / 512;
 
     boot_lba = t->bootsrc[0]->sections[0].block;
     head_count = t->partition_heads_per_cyl;
@@ -608,8 +686,7 @@ int make_isolinux_mbr(uint32_t *img_blocks, Ecma119Image *t,
        (here some 32-bit random value with no crypto strength)
     */
     if (flag & 1) {
-        gettimeofday(&tv, &tz);
-        id = 0xffffffff & (tv.tv_sec ^ (tv.tv_usec * 2000));
+        id = iso_make_mbr_id(t, 0);
         lsb_to_buf(&wpt, id, 32, 0);
     } else {
         wpt+= 4;
@@ -639,14 +716,17 @@ int make_isolinux_mbr(uint32_t *img_blocks, Ecma119Image *t,
             wpt+= 16;
     continue;
         }
-        /* write byte 0x80
+        /* write byte 0x80 if bootable
            write LBA_to_CHS(partition_offset)
            write byte filesystem_type
            write LBA_to_CHS(image_size-1)
            write dword partition_offset
            write dword image_size
         */
-        lsb_to_buf(&wpt, 0x80, 8, 0);
+        if (flag & 2)
+            lsb_to_buf(&wpt, 0x00, 8, 0);
+        else
+            lsb_to_buf(&wpt, 0x80, 8, 0);
         lba512chs_to_buf(&wpt, part_offset, head_count, sector_count);
         lsb_to_buf(&wpt, fs_type, 8, 0);
         lba512chs_to_buf(&wpt, hd_img_blocks - 1, head_count, sector_count);

libisofs/messages.c

@@ -1,6 +1,6 @@
 /*
  * Copyright (c) 2007 Vreixo Formoso
- * Copyright (c) 2009 - 2014 Thomas Schmitt
+ * Copyright (c) 2009 - 2016 Thomas Schmitt
  *
  * This file is part of the libisofs project; you can redistribute it and/or
  * modify it under the terms of the GNU General Public License version 2 
@@ -38,6 +38,7 @@
 
 #include "util.h"
 #include "node.h"
+#include "stream.h"
 
 
 /*
@@ -156,7 +157,7 @@ int iso_init_with_flag(int flag)
 #ifdef Libisofs_with_libjtE
 
  /* Ugly compile time check for header version compatibility.
-    If everthing matches, then it produces no C code. In case of mismatch,
+    If everything matches, then it produces no C code. In case of mismatch,
     intentionally faulty C code will be inserted.
  */ 
  /* The indendation is an advise of man gcc to help old compilers ignoring */
@@ -232,6 +233,7 @@ void iso_finish()
 {
     libiso_msgs_destroy(&libiso_msgr, 0);
     iso_node_xinfo_dispose_cloners(0);
+    iso_stream_destroy_cmpranks(0);
 }
 
 int iso_set_abort_severity(char *severity)
@@ -378,7 +380,7 @@ const char *iso_error_to_msg(int errcode)
     case ISO_DATA_SOURCE_MISHAP:
     case ISO_DATA_SOURCE_FAILURE:
     case ISO_DATA_SOURCE_FATAL:
-        return "Read error occured with IsoDataSource";
+        return "Read error occurred with IsoDataSource";
     case ISO_AAIP_IGNORED:
         return "AAIP info with ACL or xattr in ISO image will be ignored";
     case ISO_AAIP_BAD_ACL:
@@ -519,6 +521,40 @@ const char *iso_error_to_msg(int errcode)
         return "Problems encountered during inspection of System Area";
     case ISO_INQ_SYSAREA_PROP:
         return "Unrecognized inquiry for system area property";
+    case ISO_ALPHA_BOOT_NOTREG:
+        return "DEC Alpha Boot Loader file is not a data file";
+    case ISO_NO_KEPT_DATA_SRC:
+        return "No data source of imported ISO image available";
+    case ISO_MALFORMED_READ_INTVL:
+        return "Malformed description string for interval reader";
+    case ISO_INTVL_READ_PROBLEM:
+        return "Unreadable file, premature EOF, or failure to seek for interval reader";
+    case ISO_NOT_REPRODUCIBLE:
+        return "Cannot arrange content of data files in surely reproducible way";
+    case ISO_PATCH_FILTERED_BOOT:
+        return "May not write boot info into filtered stream of boot image";
+    case ISO_PATCH_OVERSIZED_BOOT:
+        return "Boot image to large to buffer for writing boot info";
+    case ISO_RR_NAME_TRUNCATED:
+        return "File name had to be truncated and MD5 marked";
+    case ISO_TRUNCATE_ISOFSNT:
+        return "File name truncation length changed by loaded image info";
+    case ISO_GENERAL_NOTE:
+        return "A general note message was issued";
+    case ISO_BAD_FSRC_FILETYPE:
+        return "Unrecognized file type of IsoFileSrc object";
+    case ISO_GPT_NO_VOL_UUID:
+        return "Cannot derive GPT GUID from undefined pseudo-UUID volume timestamp";
+    case ISO_BAD_GPT_GUID_MODE:
+        return "Unrecognized GPT disk GUID setup mode";
+    case ISO_NO_ROOT_DIR:
+        return "Unable to obtain root directory";
+    case ISO_SUSP_WRONG_CE_SIZE:
+        return "Zero sized, oversized, or mislocated SUSP CE area found";
+    case ISO_MULTI_OVER_IMPORTED:
+        return "Multi-session would overwrite imported_iso interval";
+    case ISO_ELTO_EFI_HIDDEN:
+        return "El-Torito EFI image is hidden";
     default:
         return "Unknown error";
     }
@@ -547,7 +583,8 @@ int iso_msg_submit(int imgid, int errcode, int causedby, const char *fmt, ...)
         vsnprintf(msg, MAX_MSG_LEN, fmt, ap);
         va_end(ap);
     } else {
-        strncpy(msg, iso_error_to_msg(errcode), MAX_MSG_LEN);
+        strncpy(msg, iso_error_to_msg(errcode), MAX_MSG_LEN - 1);
+        msg[MAX_MSG_LEN - 1] = 0;
     }
 
     libiso_msgs_submit(libiso_msgr, imgid, ISO_ERR_CODE(errcode),

libisofs/node.c

@@ -1,6 +1,6 @@
 /*
  * Copyright (c) 2007 Vreixo Formoso
- * Copyright (c) 2009 - 2014 Thomas Schmitt
+ * Copyright (c) 2009 - 2016 Thomas Schmitt
  *
  * This file is part of the libisofs project; you can redistribute it and/or
  * modify it under the terms of the GNU General Public License version 2 
@@ -327,32 +327,50 @@ enum IsoNodeType iso_node_get_type(IsoNode *node)
  * Set the name of a node.
  *
  * @param name  The name in UTF-8 encoding
+ * @param truncate_length (<64 = return on oversized name )
+ * @param flag  bit0= issue warning in case of truncation
  */
-int iso_node_set_name(IsoNode *node, const char *name)
+int iso_node_set_name_trunc(IsoNode *node, const char *in_name,
+                           int truncate_length, int flag)
 {
-    char *new;
+    char *new, *name, *trunc = NULL;
     int ret;
 
     if ((IsoNode*)node->parent == node) {
         /* you can't change name of the root node */
-        return ISO_WRONG_ARG_VALUE;
+        ret = ISO_WRONG_ARG_VALUE;
+        goto ex;
     }
 
+    name = (char *) in_name;
+    if (truncate_length >= 64) {
+        trunc = strdup(name);
+        if (trunc == 0) {
+            ret = ISO_OUT_OF_MEM;
+            goto ex;
+        }
+        ret = iso_truncate_rr_name(1, truncate_length, trunc, !(flag & 1));
+        if (ret < 0)
+            goto ex;
+        name = trunc;
+    }
     /* check if the name is valid */
     ret = iso_node_is_valid_name(name);
     if (ret < 0)
-        return ret;
+        goto ex;
 
     if (node->parent != NULL) {
         /* check if parent already has a node with same name */
         if (iso_dir_get_node(node->parent, name, NULL) == 1) {
-            return ISO_NODE_NAME_NOT_UNIQUE;
+            ret = ISO_NODE_NAME_NOT_UNIQUE;
+            goto ex;
         }
     }
 
     new = strdup(name);
     if (new == NULL) {
-        return ISO_OUT_OF_MEM;
+        ret = ISO_OUT_OF_MEM;
+        goto ex;
     }
     free(node->name);
     node->name = new;
@@ -364,10 +382,29 @@ int iso_node_set_name(IsoNode *node, const char *name)
         iso_node_take(node);
         res = iso_dir_add_node(parent, node, 0);
         if (res < 0) {
-            return res;
+            ret = res;
+            goto ex;
         }
     }
-    return ISO_SUCCESS;
+    ret = ISO_SUCCESS;
+ex:
+    if (trunc != NULL)
+        free(trunc);
+    return ret;
+}
+
+int iso_node_set_name(IsoNode *node, const char *name)
+{
+    return iso_node_set_name_trunc(node, name, 0, 0);
+}
+
+int iso_image_set_node_name(IsoImage *image, IsoNode *node, const char *name,
+                            int flag)
+{
+    if (image->truncate_mode == 0)
+        if ((int) strlen(name) > image->truncate_length)
+            return ISO_RR_NAME_TOO_LONG;
+    return iso_node_set_name_trunc(node, name, image->truncate_length, flag);
 }
 
 /**
@@ -377,6 +414,10 @@ int iso_node_set_name(IsoNode *node, const char *name)
  */
 const char *iso_node_get_name(const IsoNode *node)
 {
+    static char *root = {""};
+
+    if (node->name == NULL)
+        return root;
     return node->name;
 }
 
@@ -610,6 +651,43 @@ int iso_dir_get_node(IsoDir *dir, const char *name, IsoNode **node)
     return 1;
 }
 
+int iso_dir_get_node_trunc(IsoDir *dir, int truncate_length,
+                           const char *name, IsoNode **node)
+{
+    int ret;
+    char *trunc = NULL;
+
+    if ((int) strlen(name) <= truncate_length) {
+        ret = iso_dir_get_node(dir, name, node);
+        return ret;
+    }
+    trunc = strdup(name);
+    if (trunc == NULL)
+        return ISO_OUT_OF_MEM;
+    ret = iso_truncate_rr_name(1, truncate_length, trunc, 1);
+    if (ret < 0)
+        goto ex;
+    ret = iso_dir_get_node(dir, trunc, node);
+    if (ret == 0)
+        ret = 2;
+ex:;
+    LIBISO_FREE_MEM(trunc);
+    return ret;
+}
+
+/* API */
+int iso_image_dir_get_node(IsoImage *image, IsoDir *dir,
+                           const char *name, IsoNode **node, int flag)
+{
+    int ret;
+
+    if (image->truncate_mode == 0 || (flag & 1))
+        ret = iso_dir_get_node(dir, name, node);
+    else
+        ret = iso_dir_get_node_trunc(dir, image->truncate_length, name, node);
+    return ret;
+}
+
 /**
  * Get the number of children of a directory.
  *
@@ -1036,7 +1114,7 @@ int iso_symlink_set_dest(IsoSymlink *link, const char *dest)
  *      or regular files, this function has no effect.
  * @param w
  *      The weight as a integer number, the greater this value is, the
- *      closer from the begining of image the file will be written.
+ *      closer from the beginning of image the file will be written.
  */
 void iso_node_set_sort_weight(IsoNode *node, int w)
 {
@@ -1117,16 +1195,17 @@ int iso_file_get_old_image_lba(IsoFile *file, uint32_t *lba, int flag)
 {
     int ret;
     int section_count;
-    struct iso_file_section *sections;
+    struct iso_file_section *sections = NULL;
+
     if (file == NULL || lba == NULL) {
         return ISO_NULL_POINTER;
     }
-    ret = iso_file_get_old_image_sections(file, &section_count, &sections, flag);
-    if (ret <= 0) {
+    ret = iso_file_get_old_image_sections(file, &section_count, &sections, 0);
+    if (ret <= 0)
         return ret;
-    }
     if (section_count != 1) {
-        free(sections);
+        if (sections != NULL)
+            free(sections);
         return ISO_WRONG_ARG_VALUE;
     }
     *lba = sections[0].block;
@@ -1135,7 +1214,6 @@ int iso_file_get_old_image_lba(IsoFile *file, uint32_t *lba, int flag)
 }
 
 
-
 /*
  * Like iso_file_get_old_image_lba(), but take an IsoNode.
  *
@@ -1287,7 +1365,11 @@ int iso_dir_insert(IsoDir *dir, IsoNode *node, IsoNode **pos,
                 /* old file is newer */
                 return ISO_NODE_NAME_NOT_UNIQUE;
             }
-            /* fall down */
+            if ((node->mode & S_IFMT) != ((*pos)->mode & S_IFMT)) {
+                /* different file types */
+                return ISO_NODE_NAME_NOT_UNIQUE;
+            }
+            break;
         case ISO_REPLACE_IF_SAME_TYPE:
             if ((node->mode & S_IFMT) != ((*pos)->mode & S_IFMT)) {
                 /* different file types */
@@ -1531,11 +1613,17 @@ int attrs_cleanout_name(char *del_name, size_t *num_attrs, char **names,
     size_t i, w;
 
     for (w = i = 0; i < *num_attrs; i++) {
-        if ((strcmp(names[i], del_name) == 0) ^ (flag & 1))
-            continue;
+        if ((strcmp(names[i], del_name) == 0) ^ (flag & 1)) {
+            if (names[i] != NULL)
+                free(names[i]);
+            if (values[i] != NULL)
+                free(values[i]);
+             names[i] = values[i] = NULL;
+    continue;
+        }
         if (w == i) {
             w++;
-            continue;
+    continue;
         }
         names[w] = names[i];
         value_lengths[w] = value_lengths[i];
@@ -1734,12 +1822,13 @@ int attr_enlarge_list(char ***names, size_t **value_lengths, char ***values,
 
 /* Merge attribute list of node and given new attribute list into
    attribute list returned by  m_* parameters.
-   The m_* paramters have finally to be freed by a call with bit15 set.
+   The m_* parameters have finally to be freed by a call with bit15 set.
    @param flag          Bitfield for control purposes
                         bit0= delete all old names which begin by "user."     
                               (but not if bit2 is set)
                         bit2= delete the given names rather than overwrite
                               their content
+                        bit3= with bit0: delete all old non-"isofs." names
                         bit4= do not overwrite value of empty name
                         bit5= do not overwrite isofs attributes 
                         bit15= release memory and return 1
@@ -1765,9 +1854,11 @@ int iso_node_merge_xattr(IsoNode *node, size_t num_attrs, char **names,
         return ret;
 
     if ((flag & 1) && (!(flag & 4))) {
-        /* Delete unmatched user space pairs */
+        /* Delete unmatched settable pairs */
         for (j = 0; j < *m_num_attrs; j++) {
-            if (strncmp((*m_names)[j], "user.", 5) != 0)
+            if (strncmp((*m_names)[j], "isofs.", 6) == 0)
+                continue;
+            if (strncmp((*m_names)[j], "user.", 5) != 0 && !(flag & 8))
                 continue;
             for (i = 0; i < num_attrs; i++) {
                 if (names[i] == NULL || (*m_names)[j] == NULL)
@@ -1881,8 +1972,9 @@ int iso_node_set_attrs(IsoNode *node, size_t num_attrs, char **names,
                        size_t *value_lengths, char **values, int flag)
 {
     int ret, acl_saved = 0;
-    size_t sret, result_len, m_num = 0, *m_value_lengths = NULL, i;
-    unsigned char *result;
+    ssize_t sret;
+    size_t result_len, m_num = 0, *m_value_lengths = NULL, i;
+    unsigned char *result = NULL;
     char *a_acl = NULL, *d_acl = NULL, **m_names = NULL, **m_values = NULL;
 
     if (!(flag & 8))
@@ -1895,7 +1987,7 @@ int iso_node_set_attrs(IsoNode *node, size_t num_attrs, char **names,
                   node, num_attrs, names, value_lengths, values,
                   &m_num, &m_names, &m_value_lengths, &m_values,
                   (flag & 4) | (!(flag & 2)) | ((!(flag & 1)) << 4) |
-                  ((flag & 16) << 1));
+                  ((flag & 16) << 1) | (flag & 8));
         if (ret < 0)
             goto ex;
         num_attrs = m_num;
@@ -1921,28 +2013,34 @@ int iso_node_set_attrs(IsoNode *node, size_t num_attrs, char **names,
     }
     sret = aaip_encode(num_attrs, names, value_lengths, values,
                        &result_len, &result, 0);
-    if (sret == 0) {
-        ret = ISO_OUT_OF_MEM;
+    if (sret < 0) {
+        ret = sret;
         goto ex;
     }
 
     ret = iso_node_remove_xinfo(node, aaip_xinfo_func);
-    if (ret < 0)
-        goto ex;
-    ret = iso_node_add_xinfo(node, aaip_xinfo_func, result);
-    if (ret < 0)
-        goto ex;
-    if (ret == 0) {
-
-        /* >>> something is messed up with xinfo: an aa_string still exists */;
-
-        ret = ISO_ERROR;
+    if (ret < 0) {
+        if (result != NULL)
+            free(result);
         goto ex;
     }
-    if (acl_saved) {
-        ret = iso_node_set_acl_text(node, a_acl, d_acl, 0);
+    if (sret > 0) {
+        ret = iso_node_add_xinfo(node, aaip_xinfo_func, result);
         if (ret < 0)
             goto ex;
+        if (ret == 0) {
+
+            /* >>> something is messed up with xinfo:
+                   an aa_string still exists */;
+
+            ret = ISO_ERROR;
+            goto ex;
+        }
+        if (acl_saved) {
+            ret = iso_node_set_acl_text(node, a_acl, d_acl, 0);
+            if (ret < 0)
+                goto ex;
+        }
     }
     ret = 1;
 ex:;
@@ -2160,10 +2258,8 @@ int iso_node_set_acl_text(IsoNode *node, char *access_text, char *default_text,
                 ret = ISO_AAIP_BAD_ACL_TEXT;
                 goto ex;
             }
-            ret = 1;
-            if (a_text != NULL || d_text != NULL)
-                ret = aaip_encode_both_acl(a_text, d_text, st_mode,
-                                           &acl_len, &acl, 2 | 8);
+            ret = aaip_encode_both_acl(a_text, d_text, st_mode,
+                                       &acl_len, &acl, 2 | 8);
         } else {
             ret = 1;
             if (access_text != NULL || default_text != NULL)
@@ -2868,6 +2964,50 @@ ex:;
 }
 
 
+int iso_root_set_isofsnt(IsoNode *node, uint32_t truncate_mode,
+                         uint32_t truncate_length, int flag)
+{
+    char buffer[5 + 5], *wpt = buffer, *valuept = buffer;
+    int result_len, ret;
+    static char *names = "isofs.nt";
+    static size_t value_lengths[1];
+
+    iso_util_encode_len_bytes(truncate_mode, wpt, 0, &result_len, 0);
+    wpt += result_len;
+    iso_util_encode_len_bytes(truncate_length, wpt, 0, &result_len, 0);
+    wpt += result_len;
+    value_lengths[0] = wpt - buffer;
+    ret = iso_node_set_attrs(node, (size_t) 1,
+                             &names, value_lengths, &valuept, 2 | 8);
+    return ret;
+}
+
+
+int iso_root_get_isofsnt(IsoNode *node, uint32_t *truncate_mode,
+                         uint32_t *truncate_length, int flag)
+{
+    int ret, len;
+    size_t value_len;
+    char *value = NULL, *rpt;
+
+    ret = iso_node_lookup_attr(node, "isofs.nt", &value_len, &value, 0);
+    if (ret <= 0)
+        goto ex;
+
+    rpt = value;
+    iso_util_decode_len_bytes(truncate_mode, rpt, &len,
+                              value_len - (rpt - value), 0);
+    rpt += len + 1;
+    iso_util_decode_len_bytes(truncate_length, rpt, &len,
+                              value_len - (rpt - value), 0);
+    ret= ISO_SUCCESS;
+ex:;
+    if (value != NULL)
+        free(value);
+    return ret;
+}
+
+
 /* API */
 int iso_file_get_md5(IsoImage *image, IsoFile *file, char md5[16], int flag)
 {
@@ -2921,21 +3061,23 @@ int iso_file_make_md5(IsoFile *file, int flag)
 
     if (file->from_old_session)
         dig = 1;
-    md5= calloc(16, 1);
+    md5 = calloc(16, 1);
+    if (md5 == NULL) 
+        return ISO_OUT_OF_MEM;
     ret = iso_stream_make_md5(file->stream, md5, dig);
-    if (ret < 0)
-        goto ex;
+    if (ret < 0) {
+        free(md5);
+        return ret;
+    }
     iso_node_remove_xinfo((IsoNode *) file, checksum_md5_xinfo_func);
     ret = iso_node_add_xinfo((IsoNode *) file, checksum_md5_xinfo_func, md5);
     if (ret == 0)
         ret = ISO_ERROR; /* should not happen after iso_node_remove_xinfo() */
     if (ret < 0) {
         free(md5);
-        goto ex;
+        return ret;
     }
-    ret = 1;
-ex:;
-    return ret;
+    return 1;
 }
 
 

libisofs/node.h

@@ -1,6 +1,6 @@
 /*
  * Copyright (c) 2007 Vreixo Formoso
- * Copyright (c) 2009 - 2011 Thomas Schmitt
+ * Copyright (c) 2009 - 2016 Thomas Schmitt
  *
  * This file is part of the libisofs project; you can redistribute it and/or
  * modify it under the terms of the GNU General Public License version 2 
@@ -487,7 +487,7 @@ int iso_node_cmp_flag(IsoNode *n1, IsoNode *n2, int flag);
 
 
 /**
- * Set the checksum index (typically comming from IsoFileSrc.checksum_index)
+ * Set the checksum index (typically coming from IsoFileSrc.checksum_index)
  * of a regular file node. The index is encoded as xattr "isofs.cx" with
  * four bytes of value.
  */
@@ -512,6 +512,16 @@ int iso_root_get_isofsca(IsoNode *node, uint32_t *start_lba, uint32_t *end_lba,
                          int flag);
 
 
+/**
+ * Record and get truncation parameters as of iso_image_set_truncate_mode() by
+ * "isofs.nt".
+ */
+int iso_root_set_isofsnt(IsoNode *node, uint32_t truncate_mode,
+                         uint32_t truncate_length, int flag);
+int iso_root_get_isofsnt(IsoNode *node, uint32_t *truncate_mode,
+                         uint32_t *truncate_length, int flag);
+
+
 /**
  * Copy the xinfo list from one node to the another.
  */
@@ -543,4 +553,10 @@ int zisofs_zf_xinfo_func(void *data, int flag);
 int zisofs_zf_xinfo_cloner(void *old_data, void **new_data, int flag);
 
 
+/* Performing search for possibly truncated node name.
+ */
+int iso_dir_get_node_trunc(IsoDir *dir, int truncate_length,
+                           const char *name, IsoNode **node);
+
+
 #endif /*LIBISO_NODE_H_*/

libisofs/rockridge.c

@@ -1,7 +1,7 @@
 /*
  * Copyright (c) 2007 Vreixo Formoso
  * Copyright (c) 2007 Mario Danic
- * Copyright (c) 2009 - 2012 Thomas Schmitt
+ * Copyright (c) 2009 - 2015 Thomas Schmitt
  * 
  * This file is part of the libisofs project; you can redistribute it and/or 
  * modify it under the terms of the GNU General Public License version 2 
@@ -32,10 +32,16 @@
 #define ISO_ROCKRIDGE_IN_DIR_REC 124
 #endif
 
+#define ISO_CE_ENTRY_SIZE  28
+
 
 static
 int susp_add_ES(Ecma119Image *t, struct susp_info *susp, int to_ce, int seqno);
 
+static
+int susp_make_CE(Ecma119Image *t, uint8_t **CE,
+                 uint32_t block_offset, uint32_t byte_offset, uint32_t size);
+
 
 static
 int susp_append(Ecma119Image *t, struct susp_info *susp, uint8_t *data)
@@ -54,14 +60,76 @@ int susp_append(Ecma119Image *t, struct susp_info *susp, uint8_t *data)
 static
 int susp_append_ce(Ecma119Image *t, struct susp_info *susp, uint8_t *data)
 {
-    susp->n_ce_susp_fields++;
-    susp->ce_susp_fields = realloc(susp->ce_susp_fields, sizeof(void*)
-                                   * susp->n_ce_susp_fields);
-    if (susp->ce_susp_fields == NULL) {
-        return ISO_OUT_OF_MEM;
+    int to_alloc = 1, ret;
+    unsigned char *pad;
+    uint8_t *CE;
+    size_t next_alloc;
+
+    if (data[0] &&
+        (susp->ce_len + data[2] + ISO_CE_ENTRY_SIZE - 1) / BLOCK_SIZE !=
+                                                   susp->ce_len / BLOCK_SIZE) {
+        /* Linux fs/isofs wants byte_offset + ce_len <= BLOCK_SIZE.
+           So this Continuation Area needs to end by a CE which points
+           to the start of the next block.
+        */ 
+        to_alloc = 2;
+        if ((susp->ce_len + ISO_CE_ENTRY_SIZE) % BLOCK_SIZE)
+            to_alloc = 3; /* need a PAD pseudo entry */
+    }
+
+    if (susp->ce_susp_fields == NULL)
+        susp->alloc_ce_susp_fields = 0;
+    if (susp->n_ce_susp_fields + to_alloc > susp->alloc_ce_susp_fields) {
+        next_alloc = susp->alloc_ce_susp_fields;
+        while (susp->n_ce_susp_fields + to_alloc > next_alloc)
+            next_alloc += ISO_SUSP_CE_ALLOC_STEP;
+        susp->ce_susp_fields = realloc(susp->ce_susp_fields,
+                                       sizeof(uint8_t *) * next_alloc);
+        if (susp->ce_susp_fields == NULL)
+            return ISO_OUT_OF_MEM;
+        susp->alloc_ce_susp_fields = next_alloc;
+    }
+
+    if (to_alloc >= 2) {
+        /* Insert CE entry (actual CE size later by susp_update_CE_sizes) */
+        ret = susp_make_CE(t, &CE, (uint32_t) (susp->ce_block +
+                                               susp->ce_len / BLOCK_SIZE + 1),
+                           (uint32_t) 0, (uint32_t) 2048);
+        if (ret < 0)
+            return ret;
+        susp->ce_susp_fields[susp->n_ce_susp_fields] = CE;
+        susp->ce_len += ISO_CE_ENTRY_SIZE;
+        susp->n_ce_susp_fields++;
+    }
+    if (to_alloc >= 3) {
+
+
+#ifdef Libisofs_ce_calc_debuG
+
+        fprintf(stderr,
+                  "\nlibburn_DEBUG: Inserting %d bytes of CE padding\n\n",
+                  (int) (BLOCK_SIZE - (susp->ce_len % BLOCK_SIZE)));
+
+#endif /* Libisofs_ce_calc_debuG */
+
+        pad = malloc(1);
+        if (pad == NULL)
+            return ISO_OUT_OF_MEM;
+        pad[0] = 0;
+        susp->ce_susp_fields[susp->n_ce_susp_fields] = pad;
+        if (susp->ce_len % BLOCK_SIZE)
+            susp->ce_len += BLOCK_SIZE - (susp->ce_len % BLOCK_SIZE);
+        susp->n_ce_susp_fields++;
+    }
+    susp->ce_susp_fields[susp->n_ce_susp_fields] = data;
+    susp->n_ce_susp_fields++;
+
+    if (data[0] == 0) {
+        if (susp->ce_len % BLOCK_SIZE)
+            susp->ce_len += BLOCK_SIZE - (susp->ce_len % BLOCK_SIZE);
+    } else {
+        susp->ce_len += data[2];
     }
-    susp->ce_susp_fields[susp->n_ce_susp_fields - 1] = data;
-    susp->ce_len += data[2];
     return ISO_SUCCESS;
 }
 
@@ -315,9 +383,7 @@ int iso_get_rr_name(IsoWriteOpts *opts, char *input_charset,
             iso_msg_submit(imgid, ISO_FILENAME_WRONG_CHARSET, ret,
                    "Charset conversion error. Cannot convert %s from %s to %s",
                    str, input_charset, output_charset);
-
-        /* use the original name, it's the best we can do */
-        ret = iso_clone_mem(str, name, 0);
+        *name = NULL;
         return ISO_FILENAME_WRONG_CHARSET;
     }
 
@@ -511,12 +577,15 @@ int rrip_add_SL(Ecma119Image *t, struct susp_info *susp, uint8_t **comp,
              * debug purposes
              */
             if (ce == 0) {
+                free(SL);
                 return ISO_ASSERT_FAILURE;
             }
             ret = susp_append_ce(t, susp, SL);
             if (ret < 0) {
+                free(SL);
                 return ret;
             }
+            SL = NULL; /* now owned by susp */
             written = i;
             total_comp_len = comp[i][1] + 2;
         }
@@ -547,22 +616,56 @@ int rrip_add_SL(Ecma119Image *t, struct susp_info *susp, uint8_t **comp,
 }
 
 
+/* @param flag bit1= care about crossing block boundaries */
+static
+int susp_calc_add_to_ce(size_t *ce, size_t base_ce, int add, int flag)
+{
+    if (flag & 2) {
+        /* Account for inserted CE before size exceeds block size */
+        if ((*ce + base_ce + add + ISO_CE_ENTRY_SIZE - 1) / BLOCK_SIZE !=
+            (*ce + base_ce) / BLOCK_SIZE) {
+            /* Insert CE and padding */
+            *ce += ISO_CE_ENTRY_SIZE;
+            if ((*ce + base_ce) % BLOCK_SIZE)
+                *ce += BLOCK_SIZE - ((*ce + base_ce) % BLOCK_SIZE);
+        }
+    }
+    *ce += add;
+    return ISO_SUCCESS;
+}
+
+
 /*
-   @param flag bit0= only account sizes in sua_free resp. ce_len
-                     parameters susp and data may be NULL in this case
+   @param flag bit0= only account sizes in sua_free resp. ce_len.
+                     Parameter susp may be NULL in this case
+               bit1= account for crossing block boundaries
+                     (implied by bit0 == 0)
+   @param ce_len counts the freshly added CA size of the current node
+   @param ce_mem tells the CA size of previous nodes in the same directory
 */
 static
 int aaip_add_AL(Ecma119Image *t, struct susp_info *susp,
                 uint8_t **data, size_t num_data,
-                size_t *sua_free, size_t *ce_len, int flag)
+                size_t *sua_free, size_t *ce_len, size_t ce_mem, int flag)
 {
     int ret, done = 0, len, es_extra = 0;
     uint8_t *aapt, *cpt;
+    size_t count = 0;
 
+    if (!(flag & 1))
+        flag |= 2;
     if (!t->opts->aaip_susp_1_10)
         es_extra = 5;
     if (*sua_free < num_data + es_extra || *ce_len > 0) {
-        *ce_len += num_data + es_extra;
+        if (es_extra > 0)
+            susp_calc_add_to_ce(ce_len, ce_mem, es_extra, flag & 2);
+        done = 0;
+        for (aapt = *data; !done; aapt += aapt[2]) {
+            done = !(aapt[4] & 1);
+            len = aapt[2];
+            susp_calc_add_to_ce(ce_len, ce_mem, len, flag & 2);
+            count += len;
+        }
     } else {
         *sua_free -= num_data + es_extra;
     }
@@ -588,6 +691,7 @@ int aaip_add_AL(Ecma119Image *t, struct susp_info *susp,
     }
 
     /* Multiple fields have to be handed over as single field copies */
+    done = 0;
     for (aapt = *data; !done; aapt += aapt[2]) {
         done = !(aapt[4] & 1);
         len = aapt[2];
@@ -722,6 +826,35 @@ int aaip_add_ER(Ecma119Image *t, struct susp_info *susp, int flag)
 }
 
 
+/**
+ * Create the byte representation of a CE entry.
+ * (SUSP, 5.1).
+ */
+static
+int susp_make_CE(Ecma119Image *t, uint8_t **CE,
+                 uint32_t block_offset, uint32_t byte_offset, uint32_t size)
+{
+    uint8_t *data;
+
+    *CE = NULL;
+    data = calloc(1, 28);
+    if (data == NULL)
+        return ISO_OUT_OF_MEM;
+    *CE = data;
+
+    data[0] = 'C';
+    data[1] = 'E';
+    data[2] = 28;
+    data[3] = 1;
+
+    iso_bb(&data[4], block_offset - t->eff_partition_offset, 4);
+    iso_bb(&data[12], byte_offset, 4);
+    iso_bb(&data[20], size, 4);
+
+    return ISO_SUCCESS;
+}
+
+
 /**
  * Add a CE System Use Entry to the given tree node. A "CE" is used to add
  * a continuation area, where additional System Use Entry can be written.
@@ -730,20 +863,18 @@ int aaip_add_ER(Ecma119Image *t, struct susp_info *susp, int flag)
 static
 int susp_add_CE(Ecma119Image *t, size_t ce_len, struct susp_info *susp)
 {
-    uint8_t *CE = malloc(28);
-    if (CE == NULL) {
-        return ISO_OUT_OF_MEM;
-    }
-
-    CE[0] = 'C';
-    CE[1] = 'E';
-    CE[2] = 28;
-    CE[3] = 1;
-
-    iso_bb(&CE[4], susp->ce_block - t->eff_partition_offset, 4);
-    iso_bb(&CE[12], susp->ce_len, 4);
-    iso_bb(&CE[20], (uint32_t) ce_len, 4);
+    uint32_t block_offset, byte_offset;
+    uint8_t *CE;
+    int ret;
 
+    /* Linux fs/isofs wants byte_offset + ce_len <= BLOCK_SIZE
+     * Here the byte_offset is reduced to the minimum.
+     */
+    block_offset = susp->ce_block + susp->ce_len / BLOCK_SIZE;
+    byte_offset = susp->ce_len % BLOCK_SIZE;
+    ret = susp_make_CE(t, &CE, block_offset, byte_offset, (uint32_t) ce_len);
+    if (ret < 0)
+        return ret;
     return susp_append(t, susp, CE);
 }
 
@@ -799,6 +930,27 @@ int susp_add_ES(Ecma119Image *t, struct susp_info *susp, int to_ce, int seqno)
 }
 
 
+/** 
+ * A field beginning by 0 causes rrip_write_ce_fields() to pad up to the
+ * next block.
+ */
+static
+int pseudo_susp_add_PAD(Ecma119Image *t, struct susp_info *susp)
+{
+    unsigned char *pad;
+    int ret;
+
+    pad = malloc(1);
+    if (pad == NULL)
+        return ISO_OUT_OF_MEM;
+    pad[0] = 0;
+    ret = susp_append_ce(t, susp, pad);
+    if (ret < 0)
+        return ret;
+    return ISO_SUCCESS;
+}
+
+
 /**
  * see doc/zisofs_format.txt : "ZF System Use Entry Format"
  */
@@ -830,10 +982,12 @@ int zisofs_add_ZF(Ecma119Image *t, struct susp_info *susp, int to_ce,
 
 
 /* @param flag bit0= Do not add data but only count sua_free and ce_len
+               bit1= account for crossing block boundaries
+                     (implied by bit0 == 0)
 */
 static
 int add_zf_field(Ecma119Image *t, Ecma119Node *n, struct susp_info *info,
-                 size_t *sua_free, size_t *ce_len, int flag)
+                 size_t *sua_free, size_t *ce_len, size_t base_ce, int flag)
 {
     int ret, will_copy = 1, stream_type = 0, do_zf = 0;
     int header_size_div4 = 0, block_size_log2 = 0;
@@ -849,6 +1003,8 @@ int add_zf_field(Ecma119Image *t, Ecma119Node *n, struct susp_info *info,
                               int *header_size_div4, int *block_size_log2,
                               uint32_t *uncompressed_size, int flag);
 
+    if (!(flag & 1))
+        flag |= 2;
 
     if (iso_node_get_type(n->node) != LIBISO_FILE)
         return 2;
@@ -922,7 +1078,7 @@ int add_zf_field(Ecma119Image *t, Ecma119Node *n, struct susp_info *info,
 
     /* Account for field size */
     if (*sua_free < 16 || *ce_len > 0) {
-        *ce_len += 16;
+        susp_calc_add_to_ce(ce_len, base_ce, 16, flag & 2);
     } else {
         *sua_free -= 16;
     }
@@ -970,10 +1126,15 @@ int aaip_xinfo_cloner(void *old_data, void **new_data, int flag)
  * a CE entry of 28 bytes in SUA, this computation fails if not the 28 bytes
  * are taken into account at start. In this case the caller should retry with
  * bit0 set. 
+ * If the resulting *ce added to base_ce is in a different block than base_ce,
+ * then computation with bit0 fails and the caller should finally try bit1.
  * 
  * @param flag      bit0= assume CA usage (else return 0 on SUA overflow)
+ *                  bit1= let CA start at block start (else return 0 if
+ *                        *ce crosses a block boundary)
  * @return          1= ok, computation of *su_size and *ce is valid
  *                  0= not ok, CA usage is necessary but bit0 was not set
+ *                             or *ce crosses boundary and bit1 was not set
  *                     (*su_size and *ce stay unaltered in this case)
  *                 <0= error:
  *                 -1= not enough SUA space for 28 bytes of CE entry
@@ -981,22 +1142,50 @@ int aaip_xinfo_cloner(void *old_data, void **new_data, int flag)
  */
 static
 int susp_calc_nm_sl_al(Ecma119Image *t, Ecma119Node *n, size_t space,
-                       size_t *su_size, size_t *ce, int flag)
+                       size_t *su_size, size_t *ce, size_t base_ce, int flag)
 {
     char *name;
     size_t namelen, su_mem, ce_mem;
     void *xipt;
     size_t num_aapt = 0, sua_free = 0;
     int ret;
+    uint8_t *aapt;
+
+#ifdef Libisofs_ce_calc_debug_extrA
+
+    if (n->node->name != NULL)
+        fprintf(stderr, "libburn_DEBUG: susp_calc_nm_sl_al : %.f %s \n",
+                        (double) base_ce, n->node->name);
+
+#endif /* Libisofs_ce_calc_debug_extrA */
 
     su_mem = *su_size;
     ce_mem = *ce;
     if (*ce > 0 && !(flag & 1))
         goto unannounced_ca;
 
+    if (flag & 2) {
+        flag |= 1;
+        if (base_ce % BLOCK_SIZE) {
+
+#ifdef Libisofs_ce_calc_debuG
+
+            fprintf(stderr,
+                "\nlibburn_DEBUG: Accounting for %d bytes CE padding : %s\n\n",
+                (int) (BLOCK_SIZE - (base_ce % BLOCK_SIZE)), n->node->name);
+
+#endif /* Libisofs_ce_calc_debuG */
+
+            *ce += BLOCK_SIZE - (base_ce % BLOCK_SIZE);
+        }
+    }
+
+    namelen = 0;
     name = get_rr_fname(t, n->node->name);
-    namelen = strlen(name);
-    free(name);
+    if (name != NULL) {
+        namelen = strlen(name);
+        free(name);
+    }
 
     if (flag & 1) {
        /* Account for 28 bytes of CE field */
@@ -1020,7 +1209,7 @@ int susp_calc_nm_sl_al(Ecma119Image *t, Ecma119Node *n, size_t space,
                 of the name will always fit into the directory entry.)
         */;
 
-        *ce = 5 + namelen;
+        susp_calc_add_to_ce(ce, base_ce, 5 + namelen, flag & 2);
         *su_size = space;
     }
     if (n->type == ECMA119_SYMLINK) {
@@ -1090,21 +1279,26 @@ int susp_calc_nm_sl_al(Ecma119Image *t, Ecma119Node *n, size_t space,
                              * the component can be divided between this
                              * and another SL entry
                              */
-                            *ce += 255; /* this SL, full */
-                            sl_len = 5 + (clen - fit);
+                            /* Will fill up old SL and write it */
+                            susp_calc_add_to_ce(ce, base_ce, 255, flag & 2);
+                            sl_len = 5 + (clen - fit); /* Start new SL */
                         } else {
                             /*
                              * the component will need a 2rd SL entry in
                              * any case, so we prefer to don't write 
                              * anything in this SL
                              */
-                            *ce += sl_len + 255;
-                            sl_len = 5 + (clen - 250) + 2;
+                            /* Will write non-full old SL */
+                            susp_calc_add_to_ce(ce, base_ce, sl_len, flag & 2);
+                            /* Will write another full SL */
+                            susp_calc_add_to_ce(ce, base_ce, 255, flag & 2);
+                            sl_len = 5 + (clen - 250) + 2; /* Start new SL */
                         }
                     } else {
                         /* case 2, create a new SL entry */
-                        *ce += sl_len;
-                        sl_len = 5 + clen;
+                        /* Will write non-full old SL */
+                        susp_calc_add_to_ce(ce, base_ce, sl_len, flag & 2);
+                        sl_len = 5 + clen; /* Start new SL */
                     }
                 } else {
                     sl_len += clen;
@@ -1126,14 +1320,14 @@ int susp_calc_nm_sl_al(Ecma119Image *t, Ecma119Node *n, size_t space,
             /* the whole SL fits into the SUA */
             *su_size += sl_len;
         } else {
-            *ce += sl_len;
+            susp_calc_add_to_ce(ce, base_ce, sl_len, flag & 2);
         }
 
     }
 
     /* Find out whether ZF is to be added and account for its bytes */
     sua_free = space - *su_size;
-    add_zf_field(t, n, NULL, &sua_free, ce, 1);
+    add_zf_field(t, n, NULL, &sua_free, ce, base_ce, 1 | (flag & 2));
     *su_size = space - sua_free;
     if (*ce > 0 && !(flag & 1))
         goto unannounced_ca;
@@ -1150,12 +1344,46 @@ int susp_calc_nm_sl_al(Ecma119Image *t, Ecma119Node *n, size_t space,
     /* let the expert decide where to add num_aapt */
     if (num_aapt > 0) {
         sua_free = space - *su_size;
-        aaip_add_AL(t, NULL, NULL, num_aapt, &sua_free, ce, 1);
+        aapt = (uint8_t *) xipt;
+        aaip_add_AL(t, NULL, &aapt, num_aapt, &sua_free, ce, base_ce,
+                    1 | (flag & 2));
         *su_size = space - sua_free;
         if (*ce > 0 && !(flag & 1))
             goto unannounced_ca;
     }
 
+#ifdef Libisofs_ce_calc_debug_filetraP
+
+    if (n->node->name != NULL)
+        if (strcmp(n->node->name, "...insert.leaf.name.here...") == 0)
+            fprintf(stderr,
+                    "libburn_DEBUG: filename breakpoint susp_calc_nm_sl_al\n");
+
+#endif /* Libisofs_ce_calc_debug_filetraP */
+
+    if (*ce > 0 && !(flag & 2)) {
+        if (base_ce / BLOCK_SIZE !=
+            (base_ce + *ce + ISO_CE_ENTRY_SIZE - 1) / BLOCK_SIZE) {
+
+#ifdef Libisofs_ce_calc_debuG
+
+            fprintf(stderr,
+    "\nlibburn_DEBUG: Crossed block boundary: %.f (%lu) -> %.f (%lu) : %s\n\n",
+                    (double) base_ce, (unsigned long) (base_ce / BLOCK_SIZE),
+                    (double) (base_ce + *ce - 1),
+                    (unsigned long)
+                        ((base_ce + *ce + ISO_CE_ENTRY_SIZE - 1) / BLOCK_SIZE),
+                     n->node->name);
+
+#endif /* Libisofs_ce_calc_debuG */
+
+            /* Crossed a block boundary */
+            *su_size = su_mem;
+            *ce = ce_mem;
+            return 0;
+        }
+    }
+
     return 1;
 
 unannounced_ca:;
@@ -1167,10 +1395,12 @@ unannounced_ca:;
 
 /* @param flag bit0= Do not add data but only count sua_free and ce_len
                      param info may be NULL in this case
+               bit1= account for crossing block boundaries
+                     (implied by bit0 == 0)
 */
 static
 int add_aa_string(Ecma119Image *t, Ecma119Node *n, struct susp_info *info,
-                  size_t *sua_free, size_t *ce_len, int flag)
+                  size_t *sua_free, size_t *ce_len, size_t base_ce, int flag)
 {
     int ret;
     uint8_t *aapt;
@@ -1185,18 +1415,20 @@ int add_aa_string(Ecma119Image *t, Ecma119Node *n, struct susp_info *info,
         num_aapt = aaip_count_bytes((unsigned char *) xipt, 0);
         if (num_aapt > 0) {
             if (flag & 1) {
-                ret = aaip_add_AL(t, NULL,NULL, num_aapt, sua_free, ce_len, 1);
+                aapt = (unsigned char *) xipt;
+                ret = aaip_add_AL(t, NULL, &aapt, num_aapt, sua_free, ce_len,
+                                  base_ce, flag & (1 | 2));
             } else {
                 aapt = malloc(num_aapt);
                 if (aapt == NULL)
                     return ISO_OUT_OF_MEM;
                 memcpy(aapt, xipt, num_aapt);
                 ret = aaip_add_AL(t, info, &aapt, num_aapt, sua_free, ce_len,
-                                  0);
+                                  base_ce, 0);
+                /* aapt is NULL now and the memory is owned by t */
             }
             if (ret < 0) 
                 return ret;
-            /* aapt is NULL now and the memory is owned by t */
         }
     }
     return 1;
@@ -1214,11 +1446,13 @@ int add_aa_string(Ecma119Image *t, Ecma119Node *n, struct susp_info *info,
  *      Already occupied space in the directory record.
  * @param ce
  *      Will be filled with the space needed in a CE
+ * @param base_ce
+ *      Predicted fill of continuation area by previous nodes of same dir
  * @return
  *      The size needed for the RR entries in the System Use Area
  */
 size_t rrip_calc_len(Ecma119Image *t, Ecma119Node *n, int type, size_t used_up,
-                     size_t *ce)
+                     size_t *ce, size_t base_ce)
 {
     size_t su_size, space;
     int ret;
@@ -1281,9 +1515,11 @@ size_t rrip_calc_len(Ecma119Image *t, Ecma119Node *n, int type, size_t used_up,
     if (type == 0) {
 
         /* Try without CE */
-        ret = susp_calc_nm_sl_al(t, n, space, &su_size, ce, 0);
-        if (ret == 0) /* Retry with CE */
-            ret = susp_calc_nm_sl_al(t, n, space, &su_size, ce, 1);
+        ret = susp_calc_nm_sl_al(t, n, space, &su_size, ce, base_ce, 0);
+        if (ret == 0) /* Retry with CE but no block crossing */
+            ret = susp_calc_nm_sl_al(t, n, space, &su_size, ce, base_ce, 1);
+        if (ret == 0) /* Retry with aligned CE and block hopping */
+            ret = susp_calc_nm_sl_al(t, n, space, &su_size, ce, base_ce, 1 | 2);
         if (ret == -2)
            return ISO_OUT_OF_MEM;
 
@@ -1310,9 +1546,13 @@ size_t rrip_calc_len(Ecma119Image *t, Ecma119Node *n, int type, size_t used_up,
             if (t->opts->aaip && !t->opts->aaip_susp_1_10) {
                 *ce += 160; /* ER of AAIP */
             }
-            /* Compute length of AAIP string of root node */
+            /* Compute length of AAIP string of root node.
+               Will write all AIIP to CA, which already starts at
+               block boundary. So no need for three tries.
+             */
             aaip_sua_free= 0;
-            ret = add_aa_string(t, n, NULL, &aaip_sua_free, &aaip_len, 1);
+            ret = add_aa_string(t, n, NULL, &aaip_sua_free, &aaip_len, base_ce,
+                                1 | 2);
             if (ret < 0)
                return ret;
             *ce += aaip_len;
@@ -1376,7 +1616,7 @@ int rrip_get_susp_fields(Ecma119Image *t, Ecma119Node *n, int type,
     size_t rrip_er_len= 182;
     size_t su_size_pd, ce_len_pd; /* predicted sizes of SUA and CA */
     int ce_is_predicted = 0;
-    size_t aaip_sua_free= 0, aaip_len= 0;
+    size_t aaip_sua_free= 0, aaip_len= 0, ce_mem;
     int space;
 
     if (t == NULL || n == NULL || info == NULL) {
@@ -1393,6 +1633,18 @@ int rrip_get_susp_fields(Ecma119Image *t, Ecma119Node *n, int type,
         return ISO_ASSERT_FAILURE;
     }
 
+    /* Mark start index of node's continuation area for later update */
+    info->current_ce_start = info->n_ce_susp_fields;
+    ce_mem = info->ce_len;
+
+#ifdef Libisofs_ce_calc_debug_filetraP
+
+    if (n->node->name != NULL)
+        if (strcmp(n->node->name, "...put.leafname.here...") == 0)
+            fprintf(stderr, "libburn_DEBUG: filename breakpoint\n");
+
+#endif /* Libisofs_ce_calc_debug_filetraP */
+
     if (type == 2 && n->parent != NULL) {
         node = n->parent;
     } else {
@@ -1493,19 +1745,29 @@ int rrip_get_susp_fields(Ecma119Image *t, Ecma119Node *n, int type,
         uint8_t **comps= NULL; /* components of the SL field */
         size_t n_comp = 0; /* number of components */
 
+        namelen = 0;
         name = get_rr_fname(t, n->node->name);
+        if (name == NULL)
+            name = strdup("");
+        if (name == NULL) {
+            ret = ISO_OUT_OF_MEM;
+            goto add_susp_cleanup;
+        }
         namelen = strlen(name);
-
         sua_free = space - info->suf_len;
 
         /* Try whether NM, SL, AL will fit into SUA */
         su_size_pd = info->suf_len;
         ce_len_pd = ce_len;
         ret = susp_calc_nm_sl_al(t, n, (size_t) space,
-                                 &su_size_pd, &ce_len_pd, 0);
+                                 &su_size_pd, &ce_len_pd, info->ce_len, 0);
         if (ret == 0) { /* Have to use CA. 28 bytes of CE are necessary */
             ret = susp_calc_nm_sl_al(t, n, (size_t) space,
-                                     &su_size_pd, &ce_len_pd, 1);
+                                     &su_size_pd, &ce_len_pd, info->ce_len, 1);
+            if (ret == 0) /* Retry with aligned CE and block hopping */
+                ret = susp_calc_nm_sl_al(t, n, (size_t) space,
+                                         &su_size_pd, &ce_len_pd, info->ce_len,
+                                         1 | 2);
             sua_free -= 28;
             ce_is_predicted = 1;
         }
@@ -1535,6 +1797,13 @@ int rrip_get_susp_fields(Ecma119Image *t, Ecma119Node *n, int type,
             int cew = (nm_type == 1); /* are we writing to CE? */
 
             dest = get_rr_fname(t, ((IsoSymlink*)n->node)->dest);
+            if (dest == NULL)
+                dest = strdup("");
+            if (dest == NULL) {
+                ret = ISO_OUT_OF_MEM;
+                goto add_susp_cleanup;
+            }
+
             prev = dest;
             cur = strchr(prev, '/');
             while (1) {
@@ -1699,6 +1968,26 @@ int rrip_get_susp_fields(Ecma119Image *t, Ecma119Node *n, int type,
         }
 
         if (ce_is_predicted) {
+            if ((info->ce_len % BLOCK_SIZE) &&
+                (info->ce_len + ce_len_pd - 1 ) / BLOCK_SIZE !=
+                info->ce_len / BLOCK_SIZE) {
+                /* Linux fs/isofs wants byte_offset + ce_len <= BLOCK_SIZE
+                 * Insert padding to shift CE offset to next block start
+                 */
+
+#ifdef Libisofs_ce_calc_debuG
+
+                fprintf(stderr,
+                  "\nlibburn_DEBUG: Inserting %d bytes of CE padding : %s\n\n",
+                  (int) (BLOCK_SIZE - (info->ce_len % BLOCK_SIZE)),
+                   n->node->name);
+
+#endif /* Libisofs_ce_calc_debuG */
+
+                ret = pseudo_susp_add_PAD(t, info);
+                if (ret < 0)
+                    goto add_susp_cleanup;
+            }
             /* Add the CE entry */
             ret = susp_add_CE(t, ce_len_pd, info);
             if (ret < 0) {
@@ -1740,14 +2029,14 @@ int rrip_get_susp_fields(Ecma119Image *t, Ecma119Node *n, int type,
         }
 
         /* Eventually write zisofs ZF field */
-        ret = add_zf_field(t, n, info, &sua_free, &ce_len, 0);
+        ret = add_zf_field(t, n, info, &sua_free, &ce_len, ce_mem, 0);
         if (ret < 0)
             goto add_susp_cleanup;
 
         /* Eventually obtain AAIP field string from node
            and write it to directory entry or CE area.
         */
-        ret = add_aa_string(t, n, info, &sua_free, &ce_len, 0);
+        ret = add_aa_string(t, n, info, &sua_free, &ce_len, ce_mem, 0);
         if (ret < 0)
             goto add_susp_cleanup;
 
@@ -1808,7 +2097,8 @@ int rrip_get_susp_fields(Ecma119Image *t, Ecma119Node *n, int type,
 
             /* Compute length of AAIP string of root node */
             aaip_sua_free= 0;
-            ret = add_aa_string(t, n, NULL, &aaip_sua_free, &aaip_len, 1);
+            ret = add_aa_string(t, n, NULL, &aaip_sua_free, &aaip_len, ce_mem,
+                                1 | 2);
             if (ret < 0)
                 goto add_susp_cleanup;
 
@@ -1829,7 +2119,8 @@ int rrip_get_susp_fields(Ecma119Image *t, Ecma119Node *n, int type,
             }
             /* Write AAIP string of root node */
             aaip_sua_free= aaip_len= 0;
-            ret = add_aa_string(t, n, info, &aaip_sua_free, &aaip_len, 0);
+            ret = add_aa_string(t, n, info, &aaip_sua_free, &aaip_len, ce_mem,
+                                0);
             if (ret < 0)
                 goto add_susp_cleanup;
 
@@ -1848,12 +2139,63 @@ int rrip_get_susp_fields(Ecma119Image *t, Ecma119Node *n, int type,
     return ISO_SUCCESS;
 
     add_susp_cleanup: ;
-    free(name);
-    free(dest);
+    if (name != NULL)
+        free(name);
+    if (dest != NULL)
+        free(dest);
     susp_info_free(info);
     return ret;
 }
 
+
+/* Update the sizes of CE fields at end of info->susp_fields and in
+   single node range of info->ce_susp_fields.
+*/
+static
+int susp_update_CE_sizes(Ecma119Image *t, struct susp_info *info, int flag)
+{
+    size_t i, curr_pos;
+    uint8_t *curr_ce;
+    uint32_t size;
+
+    if (info->n_susp_fields == 0 ||
+        info->n_ce_susp_fields - info->current_ce_start == 0)
+        return ISO_SUCCESS;
+
+    for (i = 0; i < info->n_susp_fields; i++)
+        if (info->susp_fields[i][0] == 'C')
+            if(info->susp_fields[i][1] == 'E')
+    break;
+    if (i >= info->n_susp_fields) {
+        iso_msg_submit(t->image->id, ISO_ASSERT_FAILURE, 0,
+                       "System Use Area field contains no CE, but there are fields in Continuation Area");
+        return ISO_ASSERT_FAILURE;
+    }
+    curr_ce = info->susp_fields[i];
+    curr_pos = 0;
+    for (i = info->current_ce_start; i < info->n_ce_susp_fields; i++) {
+       if (info->ce_susp_fields[i][0] == 0) {
+           curr_pos = 0; /* pseudo SUSP PAD */
+    continue;
+       }
+       if (info->ce_susp_fields[i][0] == 'C' &&
+           info->ce_susp_fields[i][1] == 'E') {
+          size = (curr_pos + info->ce_susp_fields[i][2]) % BLOCK_SIZE;
+          if (size == 0)
+              size = BLOCK_SIZE;
+          iso_bb(curr_ce + 20, size, 4);
+          curr_ce = info->ce_susp_fields[i];
+       }
+       curr_pos = (curr_pos + info->ce_susp_fields[i][2]) % 2048;
+    }
+    if (curr_pos > 0) {
+        size = curr_pos % BLOCK_SIZE;
+        iso_bb(curr_ce + 20, size, 4);
+    }
+    return ISO_SUCCESS;
+}
+
+
 /**
  * Write the given SUSP fields into buf. Note that Continuation Area
  * fields are not written.
@@ -1866,11 +2208,16 @@ void rrip_write_susp_fields(Ecma119Image *t, struct susp_info *info,
 {
     size_t i;
     size_t pos = 0;
+    int ret;
 
     if (info->n_susp_fields == 0) {
         return;
     }
 
+    ret = susp_update_CE_sizes(t, info, 0);
+    if (ret < 0)
+        return;
+
     for (i = 0; i < info->n_susp_fields; i++) {
         memcpy(buf + pos, info->susp_fields[i], info->susp_fields[i][2]);
         pos += info->susp_fields[i][2];
@@ -1896,6 +2243,7 @@ int rrip_write_ce_fields(Ecma119Image *t, struct susp_info *info)
     size_t i;
     uint8_t *padding = NULL;
     int ret= ISO_SUCCESS;
+    uint64_t written = 0, pad_size;
 
     if (info->n_ce_susp_fields == 0) {
         goto ex;
@@ -1903,11 +2251,24 @@ int rrip_write_ce_fields(Ecma119Image *t, struct susp_info *info)
     LIBISO_ALLOC_MEM(padding, uint8_t, BLOCK_SIZE);
 
     for (i = 0; i < info->n_ce_susp_fields; i++) {
+        if (info->ce_susp_fields[i][0] == 0) {
+            /* Pseudo field: pad up to next block boundary */
+            pad_size = BLOCK_SIZE - (written % BLOCK_SIZE);
+            if (pad_size == BLOCK_SIZE)
+    continue;
+            memset(padding, 0, pad_size);
+            ret = iso_write(t, padding, pad_size);
+            if (ret < 0)
+                goto write_ce_field_cleanup;
+            written += pad_size;
+    continue;
+        }
         ret = iso_write(t, info->ce_susp_fields[i], 
                         info->ce_susp_fields[i][2]);
         if (ret < 0) {
             goto write_ce_field_cleanup;
         }
+        written += info->ce_susp_fields[i][2];
     }
 
     /* pad continuation area until block size */
@@ -1915,6 +2276,9 @@ int rrip_write_ce_fields(Ecma119Image *t, struct susp_info *info)
     if (i > 0 && i < BLOCK_SIZE) {
         memset(padding, 0, i);
         ret = iso_write(t, padding, i);
+        if (ret < 0)
+            goto write_ce_field_cleanup;
+        written += i;
     }
 
     write_ce_field_cleanup: ;
@@ -1925,6 +2289,7 @@ int rrip_write_ce_fields(Ecma119Image *t, struct susp_info *info)
     free(info->ce_susp_fields);
     info->ce_susp_fields = NULL;
     info->n_ce_susp_fields = 0;
+    info->alloc_ce_susp_fields = 0;
     info->ce_len = 0;
 ex:;
     LIBISO_FREE_MEM(padding);

libisofs/rockridge.h

@@ -1,7 +1,7 @@
 /*
  * Copyright (c) 2007 Vreixo Formoso
  * Copyright (c) 2007 Mario Danic
- * Copyright (c) 2009 Thomas Schmitt
+ * Copyright (c) 2009 - 2015 Thomas Schmitt
  * 
  * This file is part of the libisofs project; you can redistribute it and/or 
  * modify it under the terms of the GNU General Public License version 2 
@@ -63,10 +63,22 @@ struct susp_info
     uint32_t ce_block;
     uint32_t ce_len;
 
+    /* Storage for Continuation Area for a whole directory */
     size_t n_ce_susp_fields;
     uint8_t **ce_susp_fields;
+
+    /* The number of allocated members in ce_susp_fields */
+    size_t alloc_ce_susp_fields;
+
+    /* Marks the start index in ce_susp_fields of the current node */
+    size_t current_ce_start;
+
 };
 
+/* Step to increase allocated size of susp_info.ce_susp_fields */
+#define ISO_SUSP_CE_ALLOC_STEP 16
+
+
 /* SUSP 5.1 */
 struct susp_CE {
     uint8_t block[8];
@@ -186,11 +198,13 @@ struct susp_sys_user_entry
  *      Available space in the System Use Area for the directory record.
  * @param ce
  *      Will be filled with the space needed in a CE
+ * @param base_ce
+ *      Fill of continuation area by previous nodes of same dir
  * @return
  *      The size needed for the RR entries in the System Use Area
  */
 size_t rrip_calc_len(Ecma119Image *t, Ecma119Node *n, int type, size_t space,
-                     size_t *ce);
+                     size_t *ce, size_t base_ce);
 
 /**
  * Fill a struct susp_info with the RR/SUSP entries needed for a given
@@ -240,7 +254,7 @@ typedef struct susp_iterator SuspIterator;
 
 SuspIterator *
 susp_iter_new(IsoDataSource *src, struct ecma119_dir_record *record, 
-              uint8_t len_skp, int msgid);
+              uint32_t fs_blocks, uint8_t len_skp, int msgid);
 
 /**
  * Get the next SUSP System User Entry using given iterator.
@@ -252,7 +266,8 @@ susp_iter_new(IsoDataSource *src, struct ecma119_dir_record *record,
  * @return
  *      1 on success, 0 if no more entries, < 0 error
  */
-int susp_iter_next(SuspIterator *iter, struct susp_sys_user_entry **sue);
+int susp_iter_next(SuspIterator *iter, struct susp_sys_user_entry **sue,
+                   int flag);
 
 /**
  * Free a given susp iterator.
@@ -347,8 +362,7 @@ int read_zisofs_ZF(struct susp_sys_user_entry *zf, uint8_t algorithm[2],
                    uint32_t *uncompressed_size, int flag);
 
 /**
- * Convert a RR filename to the requested charset. On any conversion error,
- * the original name will be used.
+ * Convert a RR filename to the requested charset.
  * @param flag   bit0= do not issue error messages
  */
 int iso_get_rr_name(IsoWriteOpts *opts, char *input_charset,

libisofs/rockridge_read.c

@@ -35,6 +35,9 @@ struct susp_iterator
     IsoDataSource *src;
     int msgid;
 
+    /* Number of blocks in the ISO 9660 filesystem */
+    uint32_t fs_blocks;
+
     /* block and offset for next continuation area */
     uint32_t ce_block;
     uint32_t ce_off;
@@ -47,7 +50,7 @@ struct susp_iterator
 
 SuspIterator*
 susp_iter_new(IsoDataSource *src, struct ecma119_dir_record *record,
-              uint8_t len_skp, int msgid)
+              uint32_t fs_blocks, uint8_t len_skp, int msgid)
 {
     int pad = (record->len_fi[0] + 1) % 2;
     struct susp_iterator *iter = malloc(sizeof(struct susp_iterator));
@@ -60,6 +63,7 @@ susp_iter_new(IsoDataSource *src, struct ecma119_dir_record *record,
     iter->size = record->len_dr[0] - record->len_fi[0] - 33 - pad;
     iter->src = src;
     iter->msgid = msgid;
+    iter->fs_blocks = fs_blocks;
 
     iter->ce_len = 0;
     iter->buffer = NULL;
@@ -67,12 +71,30 @@ susp_iter_new(IsoDataSource *src, struct ecma119_dir_record *record,
     return iter;
 }
 
-int susp_iter_next(SuspIterator *iter, struct susp_sys_user_entry **sue)
+/* More than 1 MiB in a single file's CE area is suspicious */
+#define ISO_SUSP_MAX_CE_BYTES (1024 * 1024)
+
+
+/* @param flag bit0 = First call on root:
+                      Not yet clear whether this is SUSP at all
+*/
+int susp_iter_next(SuspIterator *iter, struct susp_sys_user_entry **sue,
+                   int flag)
 {
     struct susp_sys_user_entry *entry;
 
     entry = (struct susp_sys_user_entry*)(iter->base + iter->pos);
 
+    if (flag & 1) {
+        /* Yet unclear whether it is SUSP at all */
+        if (iter->size < 7)
+            return 0;
+        if (!SUSP_SIG(entry, 'S', 'P'))
+            return 0;
+        if (entry->len_sue[0] < 7)
+            return 0;
+        /* Looks like SUSP enough to pass the further processing here. */
+    }
     if ( (iter->pos + 4 > iter->size) || (SUSP_SIG(entry, 'S', 'T'))) {
 
         /* 
@@ -81,22 +103,31 @@ int susp_iter_next(SuspIterator *iter, struct susp_sys_user_entry **sue)
          * (IEEE 1281, SUSP. section 4) 
          */
         if (iter->ce_len) {
-            uint32_t block, nblocks;
+            uint32_t block, nblocks, skipped_blocks, skipped_bytes;
 
-            /* A CE has found, there is another continuation area */
-            nblocks = DIV_UP(iter->ce_off + iter->ce_len, BLOCK_SIZE);
+            /* A CE was found, there is another continuation area */
+            skipped_blocks = iter->ce_off / BLOCK_SIZE;
+            skipped_bytes = skipped_blocks * BLOCK_SIZE;
+            nblocks = DIV_UP(iter->ce_off - skipped_bytes + iter->ce_len,
+                             BLOCK_SIZE);
+            if (nblocks <= 0 || iter->ce_len > ISO_SUSP_MAX_CE_BYTES)
+                return ISO_SUSP_WRONG_CE_SIZE;
+            if (((uint64_t) iter->ce_block) + skipped_blocks + nblocks >
+                (uint64_t) iter->fs_blocks)
+                return ISO_SUSP_WRONG_CE_SIZE;
             iter->buffer = realloc(iter->buffer, nblocks * BLOCK_SIZE);
 
-            /* read all blocks needed to cache the full CE */
+            /* Read blocks needed to cache the given CE area range */
             for (block = 0; block < nblocks; ++block) {
                 int ret;
-                ret = iter->src->read_block(iter->src, iter->ce_block + block,
-                                            iter->buffer + block * BLOCK_SIZE);
+                ret = iter->src->read_block(iter->src,
+                                       iter->ce_block + skipped_blocks + block,
+                                       iter->buffer + block * BLOCK_SIZE);
                 if (ret < 0) {
                     return ret;
                 }
             }
-            iter->base = iter->buffer + iter->ce_off;
+            iter->base = iter->buffer + (iter->ce_off - skipped_bytes);
             iter->pos = 0;
             iter->size = iter->ce_len;
             iter->ce_len = 0;
@@ -134,10 +165,10 @@ int susp_iter_next(SuspIterator *iter, struct susp_sys_user_entry **sue)
         }
 
         /* we don't want to return CE entry to the user */
-        return susp_iter_next(iter, sue);
+        return susp_iter_next(iter, sue, 0);
     } else if (SUSP_SIG(entry, 'P', 'D')) {
         /* skip padding */
-        return susp_iter_next(iter, sue);
+        return susp_iter_next(iter, sue, 0);
     }
 
     *sue = entry;
@@ -211,10 +242,19 @@ int read_rr_TF(struct susp_sys_user_entry *tf, struct stat *st)
     
     /* 1. Creation time */
     if (tf->data.TF.flags[0] & (1 << 0)) {
-        
-        /* the creation is the recording time. we ignore this */
-        /* TODO maybe it would be good to manage it in ms discs, where
-         *      the recording time could be different than now!! */     
+        /* Linux accepts ctime by Creation time and by Attributes time.
+         * If both are given, then Attribute time will win.
+         */
+        if (tf->len_sue[0] < 5 + (nts+1) * s) {
+            /* RR TF entry too short. */
+            return ISO_WRONG_RR;
+        }
+        if (s == 7) {
+            time = iso_datetime_read_7(&tf->data.TF.t_stamps[nts*7]);
+        } else {
+            time = iso_datetime_read_17(&tf->data.TF.t_stamps[nts*17]);
+        }
+        st->st_ctime = time;
         ++nts;
     }
     
@@ -365,12 +405,18 @@ int read_rr_SL(struct susp_sys_user_entry *sl, char **dest, int *cont)
         if (*cont == 1) {
             /* new component */
             size_t size = strlen(*dest);
+            int has_slash;
+
             *dest = realloc(*dest, strlen(*dest) + len + 2);
             if (*dest == NULL) {
                 return ISO_OUT_OF_MEM;
             }
             /* it is a new compoenent, add the '/' */
-            if ((*dest)[size-1] != '/') {
+            has_slash = 0;
+            if (size > 0)
+                if ((*dest)[size - 1] == '/')
+                    has_slash = 1;
+            if (!has_slash) {
                 (*dest)[size] = '/';
                 (*dest)[size+1] = '\0';
             }
@@ -412,7 +458,7 @@ int read_rr_PN(struct susp_sys_user_entry *pn, struct stat *st)
 {
     int high_shift= 0;
 
-    if (pn == NULL || pn == NULL) {
+    if (pn == NULL || st == NULL) {
         return ISO_NULL_POINTER;
     }
     if (pn->sig[0] != 'P' || pn->sig[1] != 'N') {
@@ -453,24 +499,27 @@ int read_aaip_AA(struct susp_sys_user_entry *sue,
      if (*is_done) {
 
          /* To coexist with Apple ISO :
-            Gracefully react on eventually trailing Apple AA
+            Gracefully react on possibly trailing Apple AA
          */
          if (sue->version[0] != 1 || sue->len_sue[0] == 7)
              return ISO_SUCCESS;
 
          return ISO_WRONG_RR;
      }
-
-
-     /* Eventually create or grow storage */
      if (*aa_size == 0 || *aa_string == NULL) {
-
-         /* Gracefully react on eventually leading Apple AA
+         /* Gracefully react on possibly leading Apple AA
          */
-         if (sue->version[0] != 1 || sue->len_sue[0] < 9) {
+         if (sue->version[0] != 1 || sue->len_sue[0] < 9)
              return ISO_SUCCESS;
-         }
+     }
 
+     /* A valid AAIP AA entry has 5 header bytes and at least 1 component byte
+      */
+     if (sue->len_sue[0] < 6)
+         return ISO_WRONG_RR;
+
+     /* Possibly create or grow storage */
+     if (*aa_size == 0 || *aa_string == NULL) {
          *aa_size = *aa_len + sue->len_sue[0];
          *aa_string = calloc(*aa_size, 1);
          *aa_len = 0;
@@ -526,7 +575,12 @@ int read_aaip_AL(struct susp_sys_user_entry *sue,
      if (sue->version[0] != 1)
          return ISO_WRONG_RR;
 
-     /* Eventually create or grow storage */
+     /* A valid AL entry has 5 header bytes and at least 1 component byte
+      */
+     if (sue->len_sue[0] < 6)
+         return ISO_WRONG_RR;
+
+     /* Possibly create or grow storage */
      if (*aa_size == 0 || *aa_string == NULL) {
          *aa_size = *aa_len + sue->len_sue[0];
          *aa_string = calloc(*aa_size, 1);

libisofs/stream.c

@@ -1,6 +1,6 @@
 /*
  * Copyright (c) 2007 Vreixo Formoso
- * Copyright (c) 2009 - 2011 Thomas Schmitt
+ * Copyright (c) 2009 - 2015 Thomas Schmitt
  *
  * This file is part of the libisofs project; you can redistribute it and/or
  * modify it under the terms of the GNU General Public License version 2 
@@ -164,15 +164,6 @@ IsoStream *fsrc_get_input_stream(IsoStream *stream, int flag)
     return NULL;
 }
 
-static 
-int fsrc_cmp_ino(IsoStream *s1, IsoStream *s2)
-{
-    int ret;
-
-    ret = iso_stream_cmp_ino(s1, s2, 1);
-    return ret;
-}
-
 int fsrc_clone_stream(IsoStream *old_stream, IsoStream **new_stream,
                       int flag)
 {
@@ -227,7 +218,7 @@ IsoStreamIface fsrc_stream_class = {
     fsrc_free,
     fsrc_update_size,
     fsrc_get_input_stream,
-    fsrc_cmp_ino,
+    NULL,
     fsrc_clone_stream
 };
 
@@ -448,15 +439,6 @@ IsoStream* cut_out_get_input_stream(IsoStream *stream, int flag)
     return NULL;
 }
 
-static
-int cut_out_cmp_ino(IsoStream *s1, IsoStream *s2)
-{
-    int ret;
-
-    ret = iso_stream_cmp_ino(s1, s2, 1);
-    return ret;
-}
-
 static
 int cut_out_clone_stream(IsoStream *old_stream, IsoStream **new_stream,
                       int flag)
@@ -517,7 +499,7 @@ IsoStreamIface cut_out_stream_class = {
     cut_out_free,
     cut_out_update_size,
     cut_out_get_input_stream,
-    cut_out_cmp_ino,
+    NULL,
     cut_out_clone_stream
     
 };
@@ -698,15 +680,6 @@ IsoStream* mem_get_input_stream(IsoStream *stream, int flag)
     return NULL;
 }
 
-static
-int mem_cmp_ino(IsoStream *s1, IsoStream *s2)
-{
-    int ret;
-
-    ret = iso_stream_cmp_ino(s1, s2, 1);
-    return ret;
-}
-
 static
 int mem_clone_stream(IsoStream *old_stream, IsoStream **new_stream,
                       int flag)
@@ -763,7 +736,7 @@ IsoStreamIface mem_stream_class = {
     mem_free,
     mem_update_size,
     mem_get_input_stream,
-    mem_cmp_ino,
+    NULL,
     mem_clone_stream
 
 };
@@ -943,12 +916,22 @@ ex:;
     return path;
 }
 
-/* @return 1 = ok , 0 = not an ISO image stream , <0 = error */
+/*
+   @param flag bit0= in case of filter stream do not dig for base stream
+   @return 1 = ok , 0 = not an ISO image stream , <0 = error
+*/
 int iso_stream_set_image_ino(IsoStream *stream, ino_t ino, int flag)
 {
+    IsoStream *base_stream;
+
     if (stream == NULL) {
         return ISO_NULL_POINTER;
     }
+    if (!(flag & 1)) {
+        base_stream = iso_stream_get_input_stream(stream, 1);
+        if (base_stream != NULL)
+            stream = base_stream;
+    }
     if (stream->class == &fsrc_stream_class) {
         FSrcStreamData *fsrc_data = stream->data;
         fsrc_data->ino_id = ino;
@@ -957,6 +940,104 @@ int iso_stream_set_image_ino(IsoStream *stream, ino_t ino, int flag)
    return 0;
 }
 
+int iso_stream_cmp_ifs_sections(IsoStream *s1, IsoStream *s2, int *cmp_ret,
+                                int flag)
+{   
+    int ret;
+    FSrcStreamData *fssd1, *fssd2;
+    IsoFileSource *src1, *src2;
+
+    /* Must keep any suspect in the game to preserve transitivity of the
+       calling function by ranking applicable streams lower than
+       non-applicable. ones.
+    */
+    if (s1->class != &fsrc_stream_class && s2->class != &fsrc_stream_class)
+        return 0;
+
+    /* Compare eventual image data section LBA and sizes */
+    if (s1->class == &fsrc_stream_class) {
+        fssd1= (FSrcStreamData *) s1->data;
+        src1 = fssd1->src;
+    } else {
+        src1 = NULL;
+    }
+    if (s2->class == &fsrc_stream_class) {
+        fssd2= (FSrcStreamData *) s2->data;
+        src2 = fssd2->src;
+    } else {
+        src2 = NULL;
+    }
+    ret = iso_ifs_sections_cmp(src1, src2, cmp_ret, 1);
+    if (ret <= 0)
+        return 0;
+    return 1;
+}
+
+
+/* Maintain and exploit a list of stream compare functions seen by
+   iso_stream_cmp_ino(). This is needed to separate stream comparison
+   families in order to keep iso_stream_cmp_ino() transitive while
+   alternative stream->class->cmp_ino() decide inside the families.
+*/
+struct iso_streamcmprank {
+    int (*cmp_func)(IsoStream *s1, IsoStream *s2);
+    struct iso_streamcmprank *next;
+};
+
+static struct iso_streamcmprank *streamcmpranks = NULL;
+
+static
+int iso_get_streamcmprank(int (*cmp_func)(IsoStream *s1, IsoStream *s2),
+                          int flag)
+{
+    int idx;
+    struct iso_streamcmprank *cpr, *last_cpr = NULL;
+
+    idx = 0;
+    for (cpr = streamcmpranks; cpr != NULL; cpr = cpr->next) {
+        if (cpr->cmp_func == cmp_func)
+    break;
+        idx++;
+        last_cpr = cpr;
+    }
+    if (cpr != NULL)
+        return idx;
+    LIBISO_ALLOC_MEM_VOID(cpr, struct iso_streamcmprank, 1);
+    cpr->cmp_func = cmp_func;
+    cpr->next = NULL;
+    if (last_cpr != NULL)
+        last_cpr->next = cpr;
+    if (streamcmpranks == NULL)
+        streamcmpranks = cpr;
+    return idx;
+ex:;
+    return -1;
+}
+
+static
+int iso_cmp_streamcmpranks(int (*cf1)(IsoStream *s1, IsoStream *s2),
+                           int (*cf2)(IsoStream *s1, IsoStream *s2))
+{
+    int rank1, rank2;
+
+    rank1 = iso_get_streamcmprank(cf1, 0);
+    rank2 = iso_get_streamcmprank(cf2, 0);
+    return rank1 < rank2 ? -1 : 1;
+}
+
+int iso_stream_destroy_cmpranks(int flag)
+{
+    struct iso_streamcmprank *cpr, *next;
+
+    for (cpr = streamcmpranks; cpr != NULL; cpr = next) {
+        next = cpr->next;
+        LIBISO_FREE_MEM(cpr);
+    }
+    streamcmpranks = NULL;
+    return ISO_SUCCESS;
+}
+
+
 /* API */ 
 int iso_stream_cmp_ino(IsoStream *s1, IsoStream *s2, int flag)
 {
@@ -965,8 +1046,6 @@ int iso_stream_cmp_ino(IsoStream *s1, IsoStream *s2, int flag)
     dev_t dev_id1, dev_id2;
     ino_t ino_id1, ino_id2;
     off_t size1, size2;
-    FSrcStreamData *fssd1, *fssd2;
-
 
 /*
    #define Libisofs_stream_cmp_ino_debuG 1
@@ -983,10 +1062,72 @@ int iso_stream_cmp_ino(IsoStream *s1, IsoStream *s2, int flag)
     if (s2 == NULL)
         return 1;
 
-    if (s1->class->version >= 3 && !(flag & 1)) {
-       /* Filters may have smarter methods to compare themselves with others */
-       ret = s1->class->cmp_ino(s1, s2);
-       return ret;
+    /* This stays transitive by the fact that 
+       iso_stream_cmp_ifs_sections() is transitive,
+       returns > 0 if s1 or s2 are applicable,
+       ret is -1 if s1 is applicable but s2 is not,
+       ret is 1 if s1 is not applicable but s2 is.
+
+       Proof:
+       Be A the set of applicable streams, S and G transitive and
+       antisymmetric relations in respect to outcome {-1, 0, 1}.
+       The combined relation R shall be defined by
+         I.   R(a,b) = S(a,b) if a in A or b in A, else G(a,b)
+       Further S shall have the property
+         II.  S(a,b) = -1 if a in A and b not in A
+       Then R can be proven to be transitive:
+       By enumerating the 8 combinations of a,b,c being in A or not, we get
+       5 cases of pure S or pure G. Three cases are mixed:
+         a,b not in A, c in A : G(a,b) == -1, S(b,c) == -1 -> S(a,c) == -1 
+             Impossible because S(b,c) == -1 contradicts II.
+         a,c not in A, b in A : S(a,b) == -1, S(b,c) == -1 -> G(a,c) == -1
+             Impossible because S(a,b) == -1 contradicts II.
+         b,c not in A, a in A : S(a,b) == -1, G(b,c) == -1 -> S(a,c) == -1
+             Always true because S(a,c) == -1 by definition II.
+    */
+    if (iso_stream_cmp_ifs_sections(s1, s2, &ret, 0) > 0)
+        return ret; /* Both are unfiltered from loaded ISO filesystem */
+
+    if (!(flag & 1)) {
+       /* Filters may have smarter methods to compare themselves with others.
+          Transitivity is ensured by ranking mixed pairs by the rank of their
+          comparison functions, and by ranking streams with .cmp_ino lower
+          than streams without.
+          (One could merge (class->version < 3) and (cmp_ino == NULL).)
+
+          Here we define S for "and" rather than "or"
+            I.   R(a,b) = S(a,b) if a in A and b in A, else G(a,b)
+          and the function ranking in case of "exor" makes sure that
+            II.  G(a,b) = -1 if a in A and b not in A
+          Again we get three mixed cases:
+            a not in A, b,c in A : G(a,b) == -1, S(b,c) == -1 -> G(a,c) == -1 
+                Impossible because G(a,b) == -1 contradicts II.
+            b not in A, a,c in A : G(a,b) == -1, G(b,c) == -1 -> S(a,c) == -1
+                Impossible because G(b,c) == -1 contradicts II.
+            c not in A, a,b in A : S(a,b) == -1, G(b,c) == -1 -> G(a,c) == -1
+                Always true because G(a,c) == -1 by definition II.
+       */
+       if ((s1->class->version >= 3) ^ (s2->class->version >= 3)) {
+           /* One of both has no own com_ino function. Rank it as larger. */
+           return s1->class->version >= 3 ? -1 : 1;
+       } else if (s1->class->version >= 3) {
+           if (s1->class->cmp_ino == s2->class->cmp_ino) {
+               if (s1->class->cmp_ino == NULL) {
+                   /* Both are NULL. No decision by .cmp_ino(). */;
+               } else {
+                   /* Both are compared by the same function */
+                   ret = s1->class->cmp_ino(s1, s2);
+                   return ret;
+               }
+           } else {
+               /* Not the same cmp_ino() function. Decide by list rank of
+                  function while building the list on the fly.
+               */
+               ret = iso_cmp_streamcmpranks(s1->class->cmp_ino,
+                                            s2->class->cmp_ino);
+               return ret;
+           }
+       }
     }
 
     iso_stream_get_id(s1, &fs_id1, &dev_id1, &ino_id1);
@@ -1042,14 +1183,6 @@ int iso_stream_cmp_ino(IsoStream *s1, IsoStream *s2, int flag)
 
     if (s1->class != s2->class)
         return (s1->class < s2->class ? -1 : 1);
-    if (s1->class == &fsrc_stream_class) {
-        /* Compare eventual image data section LBA and sizes */
-        fssd1= (FSrcStreamData *) s1->data;
-        fssd2= (FSrcStreamData *) s2->data;
-        ret = iso_ifs_sections_cmp(fssd1->src, fssd2->src, 0);
-        if (ret != 0)
-            return ret;
-    }
     if (fs_id1 == 0 && dev_id1 == 0 && ino_id1 == 0) {
         return (s1 < s2 ? -1 : 1);
     }

libisofs/stream.h

@@ -1,6 +1,6 @@
 /*
  * Copyright (c) 2007 Vreixo Formoso
- * Copyright (c) 2009 - 2011 Thomas Schmitt
+ * Copyright (c) 2009 - 2016 Thomas Schmitt
  *
  * This file is part of the libisofs project; you can redistribute it and/or
  * modify it under the terms of the GNU General Public License version 2 
@@ -41,7 +41,7 @@ void iso_stream_get_file_name(IsoStream *stream, char *name);
  * take an extra ref.
  *
  * @return
- *      1 sucess, < 0 error
+ *      1 success, < 0 error
  *      Possible errors:
  *
  */
@@ -52,7 +52,7 @@ int iso_file_source_stream_new(IsoFileSource *src, IsoStream **stream);
  * The stream will add a ref. to the IsoFileSource.
  *
  * @return
- *      1 sucess, < 0 error
+ *      1 success, < 0 error
  */
 int iso_cut_out_stream_new(IsoFileSource *src, off_t offset, off_t size,
                            IsoStream **stream);
@@ -112,4 +112,13 @@ int iso_stream_clone_filter_common(IsoStream *old_stream,
                                    IsoStream **new_stream, 
                                    IsoStream **new_input, int flag);
 
+
+/**
+ * Dispose the internal list of stream class cmp_ino() functions. It is
+ * a static global of stream.c, created and used by iso_stream_cmp_ino().
+ * This function is supposed to be called by iso_finish() only.
+ */
+int iso_stream_destroy_cmpranks(int flag);
+
+
 #endif /*STREAM_H_*/

libisofs/system_area.h

@@ -1,6 +1,6 @@
 /*
  * Copyright (c) 2008 Vreixo Formoso
- * Copyright (c) 2012 Thomas Schmitt
+ * Copyright (c) 2012 - 2015 Thomas Schmitt
  *
  * This file is part of the libisofs project; you can redistribute it and/or
  * modify it under the terms of the GNU General Public License version 2 
@@ -173,12 +173,6 @@ int iso_quick_apm_entry(struct iso_apm_partition_request **req_array,
                         uint32_t start_block, uint32_t block_count,
                         char *name, char *type);
 
-
-/* CRC-32 as of GPT and Ethernet.
-*/
-unsigned int iso_crc32_gpt(unsigned char *data, int count, int flag);
-
-
 /* These two pseudo-random generators produce byte strings which will
    surely not duplicate in the first 256 calls. If more calls are necessary
    in the same process, then one must wait until the output of
@@ -188,10 +182,15 @@ unsigned int iso_crc32_gpt(unsigned char *data, int count, int flag);
    run on other machines with the same process number at the same time.
 */
 
-/* Produces a weakly random variation of a hardcoded real random uuid
+/* Produces a GPT disk or partition GUID.
+   Pseudo-random by iso_generate_gpt_guid() if t->gpt_uuid_counter is 0.
+   Else derived reproducibly by counter number from t->gpt_uuid_base.
 */
-void iso_random_uuid(Ecma119Image *t, uint8_t uuid[16]);
+void iso_gpt_uuid(Ecma119Image *t, uint8_t uuid[16]);
 
+/* Mark a given byte string as UUID version 4, RFC 4122.
+*/
+void iso_mark_guid_version_4(uint8_t *u);
 
 /* The parameter struct for production of a single GPT entry.
    See also the partial GPT description in doc/boot_sectors.txt.
@@ -287,10 +286,20 @@ struct iso_sun_disk_label_entry {
 */
 int partprepend_writer_create(Ecma119Image *target);
 
+/* Creates the Inline Partition Append Writer
+*/
+int partappend_writer_create(Ecma119Image *target);
+
 /* Creates the GPT backup tail writer.
 */
 int gpt_tail_writer_create(Ecma119Image *target);
 
+/* Not for execution but only to identify the writer by
+    ( writer->write_vol_desc == gpt_tail_writer_write_vol_desc )
+*/
+int gpt_tail_writer_write_vol_desc(IsoImageWriter *writer);
+
+
 /* Only for up to 36 characters ISO-8859-1 (or ASCII) input */
 void iso_ascii_utf_16le(uint8_t gap_name[72]);
 
@@ -307,4 +316,14 @@ void iso_ascii_utf_16le(uint8_t gap_name[72]);
 #define Libisofs_grub2_sparc_patch_adr_poS    0x228
 #define Libisofs_grub2_sparc_patch_size_poS   0x230    
 
+
+/* Put appended partitions into the writer range
+*/
+#define Libisofs_appended_partitions_inlinE yes
+#ifdef Libisofs_appended_partitions_inlinE
+/* For padding after appended partitions (and also after backup GPT)
+*/
+#define Libisofs_part_align_writeR yes
+#endif
+
 #endif /* SYSTEM_AREA_H_ */

libisofs/tree.c

@@ -1,6 +1,6 @@
 /*
  * Copyright (c) 2007 Vreixo Formoso
- * Copyright (c) 2011 - 2014 Thomas Schmitt
+ * Copyright (c) 2011 - 2015 Thomas Schmitt
  * 
  * This file is part of the libisofs project; you can redistribute it and/or 
  * modify it under the terms of the GNU General Public License version 2 
@@ -100,6 +100,19 @@ int iso_tree_add_new_dir(IsoDir *parent, const char *name, IsoDir **dir)
     return iso_dir_insert(parent, (IsoNode*)node, pos, ISO_REPLACE_NEVER);
 }
 
+int iso_image_add_new_dir(IsoImage *image, IsoDir *parent, const char *name,
+                          IsoDir **dir)
+{
+    int ret;
+    char *namept;
+
+    ret = iso_image_truncate_name(image, name, &namept, 0);
+    if (ret < 0)
+        return ret;
+    ret = iso_tree_add_new_dir(parent, namept, dir);
+    return ret;
+}
+
 /**
  * Add a new symlink to the directory tree. Permissions are set to 0777, 
  * owner and hidden atts are taken from parent. You can modify any of them 
@@ -175,6 +188,20 @@ int iso_tree_add_new_symlink(IsoDir *parent, const char *name,
     return iso_dir_insert(parent, (IsoNode*)node, pos, ISO_REPLACE_NEVER);
 }
 
+int iso_image_add_new_symlink(IsoImage *image, IsoDir *parent,
+                              const char *name, const char *dest,
+                              IsoSymlink **link)
+{
+    int ret;
+    char *namept;
+
+    ret = iso_image_truncate_name(image, name, &namept, 0);
+    if (ret < 0)
+        return ret;
+    ret = iso_tree_add_new_symlink(parent, namept, dest, link);
+    return ret;
+}
+
 /**
  * Add a new special file to the directory tree. As far as libisofs concerns,
  * an special file is a block device, a character device, a FIFO (named pipe)
@@ -264,6 +291,20 @@ int iso_tree_add_new_special(IsoDir *parent, const char *name, mode_t mode,
     return iso_dir_insert(parent, (IsoNode*)node, pos, ISO_REPLACE_NEVER);
 }
 
+int iso_image_add_new_special(IsoImage *image, IsoDir *parent,
+                              const char *name, mode_t mode,
+                              dev_t dev, IsoSpecial **special)
+{
+    int ret;
+    char *namept;
+
+    ret = iso_image_truncate_name(image, name, &namept, 0);
+    if (ret < 0)
+        return ret;
+    ret = iso_tree_add_new_special(parent, namept, mode, dev, special);
+    return ret;
+}
+
 /**
  * Add a new regular file to the iso tree. Permissions are set to 0444, 
  * owner and hidden atts are taken from parent. You can modify any of them 
@@ -339,6 +380,19 @@ int iso_tree_add_new_file(IsoDir *parent, const char *name, IsoStream *stream,
     return iso_dir_insert(parent, (IsoNode*)node, pos, ISO_REPLACE_NEVER);
 }
 
+int iso_image_add_new_file(IsoImage *image, IsoDir *parent, const char *name,
+                           IsoStream *stream, IsoFile **file)
+{
+    int ret;
+    char *namept;
+
+    ret = iso_image_truncate_name(image, name, &namept, 0);
+    if (ret < 0)
+        return ret;
+    ret = iso_tree_add_new_file(parent, namept, stream, file);
+    return ret;
+}
+
 /**
  * Set whether to follow or not symbolic links when added a file from a source
  * to IsoImage.
@@ -503,7 +557,7 @@ int iso_tree_add_node_builder(IsoImage *image, IsoDir *parent,
     int result;
     IsoNode *new;
     IsoNode **pos;
-    char *name = NULL;
+    char *name = NULL, *namept;
 
     if (parent == NULL || src == NULL || builder == NULL) {
         result = ISO_NULL_POINTER; goto ex;
@@ -514,14 +568,18 @@ int iso_tree_add_node_builder(IsoImage *image, IsoDir *parent,
 
     name = iso_file_source_get_name(src);
 
+    result = iso_image_truncate_name(image, name, &namept, 0);
+    if (result < 0)
+        return result;
+
     /* find place where to insert */
-    result = iso_dir_exists(parent, name, &pos);
+    result = iso_dir_exists(parent, namept, &pos);
     if (result) {
         /* a node with same name already exists */
         result = ISO_NODE_NAME_NOT_UNIQUE; goto ex;
     }
 
-    result = builder->create_node(builder, image, src, name, &new);
+    result = builder->create_node(builder, image, src, namept, &new);
     if (result < 0)
         goto ex;
 
@@ -568,6 +626,7 @@ int iso_tree_add_new_node(IsoImage *image, IsoDir *parent, const char *name,
     IsoFileSource *file;
     IsoNode *new;
     IsoNode **pos;
+    char *namept;
 
     if (image == NULL || parent == NULL || name == NULL || path == NULL) {
         return ISO_NULL_POINTER;
@@ -577,8 +636,12 @@ int iso_tree_add_new_node(IsoImage *image, IsoDir *parent, const char *name,
         *node = NULL;
     }
 
+    result = iso_image_truncate_name(image, name, &namept, 0);
+    if (result < 0)
+        return result;
+
     /* find place where to insert */
-    result = iso_dir_exists(parent, name, &pos);
+    result = iso_dir_exists(parent, namept, &pos);
     if (result) {
         /* a node with same name already exists */
         return ISO_NODE_NAME_NOT_UNIQUE;
@@ -591,7 +654,7 @@ int iso_tree_add_new_node(IsoImage *image, IsoDir *parent, const char *name,
     }
 
     result = image->builder->create_node(image->builder, image, file,
-                                         (char *) name, &new);
+                                         namept, &new);
     
     /* free the file */
     iso_file_source_unref(file);
@@ -620,6 +683,7 @@ int iso_tree_add_new_cut_out_node(IsoImage *image, IsoDir *parent,
     IsoFile *new;
     IsoNode **pos;
     IsoStream *stream;
+    char *namept;
 
     if (image == NULL || parent == NULL || name == NULL || path == NULL) {
         return ISO_NULL_POINTER;
@@ -629,8 +693,12 @@ int iso_tree_add_new_cut_out_node(IsoImage *image, IsoDir *parent,
         *node = NULL;
     }
 
+    result = iso_image_truncate_name(image, name, &namept, 0);
+    if (result < 0)
+        return result;
+
     /* find place where to insert */
-    result = iso_dir_exists(parent, name, &pos);
+    result = iso_dir_exists(parent, namept, &pos);
     if (result) {
         /* a node with same name already exists */
         return ISO_NODE_NAME_NOT_UNIQUE;
@@ -673,7 +741,7 @@ int iso_tree_add_new_cut_out_node(IsoImage *image, IsoDir *parent,
     iso_stream_unref(new->stream);
     new->stream = stream;
     
-    result = iso_node_set_name((IsoNode*)new, name);
+    result = iso_node_set_name((IsoNode*)new, namept);
     if (result < 0) {
         iso_node_unref((IsoNode*)new);
         return result;
@@ -1106,7 +1174,10 @@ int iso_tree_add_dir_rec(IsoImage *image, IsoDir *parent, const char *dir)
     return result;
 }
 
-int iso_tree_path_to_node(IsoImage *image, const char *path, IsoNode **node)
+/* @param flag bit0= truncate according to image truncate mode and length
+*/
+int iso_tree_path_to_node_flag(IsoImage *image, const char *path,
+                               IsoNode **node, int flag)
 {
     int result;
     IsoNode *n;
@@ -1128,6 +1199,8 @@ int iso_tree_path_to_node(IsoImage *image, const char *path, IsoNode **node)
     }
 
     ptr = strdup(path);
+    if (ptr == NULL)
+        return ISO_OUT_OF_MEM;
     result = 0;
 
     /* get the first component of the path */
@@ -1140,7 +1213,12 @@ int iso_tree_path_to_node(IsoImage *image, const char *path, IsoNode **node)
         }
         dir = (IsoDir *)n;
 
-        result = iso_dir_get_node(dir, component, &n);
+        if ((flag & 1) && image->truncate_mode == 1) {
+            result = iso_dir_get_node_trunc(dir, image->truncate_length,
+                                            component, &n);
+        } else {
+            result = iso_dir_get_node(dir, component, &n);
+        }
         if (result != 1) {
             n = NULL;
             break;
@@ -1156,6 +1234,16 @@ int iso_tree_path_to_node(IsoImage *image, const char *path, IsoNode **node)
     return result;
 }
 
+int iso_tree_path_to_node(IsoImage *image, const char *path, IsoNode **node)
+{
+    return iso_tree_path_to_node_flag(image, path, node, 0);
+}
+
+int iso_image_path_to_node(IsoImage *image, const char *path, IsoNode **node)
+{
+    return iso_tree_path_to_node_flag(image, path, node, 1);
+}
+
 char *iso_tree_get_node_path(IsoNode *node)
 {
     char *path = NULL, *parent_path = NULL;
@@ -1397,18 +1485,36 @@ int iso_tree_clone_special(IsoSpecial *node,
     return ISO_SUCCESS;
 }
 
-/* API */
-int iso_tree_clone(IsoNode *node,
-                   IsoDir *new_parent, char *new_name, IsoNode **new_node,
-                   int flag)
+
+/* @param flag bit0= Merge directories rather than ISO_NODE_NAME_NOT_UNIQUE.
+               bit1= issue warning in case of truncation
+*/
+int iso_tree_clone_trunc(IsoNode *node, IsoDir *new_parent, 
+                         char *new_name_in, IsoNode **new_node, 
+                         int truncate_length, int flag)
 {
     int ret = ISO_SUCCESS;
+    char *new_name, *trunc = NULL;
 
+    *new_node = NULL;
+    new_name = new_name_in;
+    if (truncate_length >= 64 && (int) strlen(new_name) > truncate_length) {
+        trunc = strdup(new_name);
+        if (trunc == 0) {
+            ret = ISO_OUT_OF_MEM;
+            goto ex;
+        }
+        ret = iso_truncate_rr_name(1, truncate_length, trunc, !(flag & 2));
+        if (ret < 0)
+            goto ex;
+        new_name = trunc;
+    }
     if (iso_dir_get_node(new_parent, new_name, new_node) == 1) {
         if (! (node->type == LIBISO_DIR && (*new_node)->type == LIBISO_DIR &&
                (flag & 1))) {
             *new_node = NULL;
-            return ISO_NODE_NAME_NOT_UNIQUE;
+            ret = ISO_NODE_NAME_NOT_UNIQUE;
+            goto ex;
         }
     } else
         flag &= ~1;
@@ -1429,10 +1535,42 @@ int iso_tree_clone(IsoNode *node,
         ret = ISO_SUCCESS; /* API says they are silently ignored */
     }
     if (ret < 0)
-        return ret;
-    if (flag & 1)
-        return 2; /* merged two directories, *new_node is not new */
+        goto ex;
+    if (flag & 1) {
+        ret = 2; /* merged two directories, *new_node is not new */
+        goto ex;
+    }
     ret = iso_tree_copy_node_attr(node, *new_node, 0);
+
+ex:;
+    if (trunc != NULL)
+        free(trunc);
+    return ret;
+}
+
+
+/* API */
+int iso_tree_clone(IsoNode *node,
+                   IsoDir *new_parent, char *new_name, IsoNode **new_node,
+                   int flag)
+{
+    return iso_tree_clone_trunc(node, new_parent, new_name, new_node, 0,
+                                flag & 1); 
+}
+
+
+/* API */
+int iso_image_tree_clone(IsoImage *image, IsoNode *node, IsoDir *new_parent,
+                         char *new_name, IsoNode **new_node, int flag)
+{
+    int length, ret;
+
+    if (image->truncate_mode == 0)
+        length = 0;
+    else
+        length = image->truncate_length;
+    ret = iso_tree_clone_trunc(node, new_parent, new_name, new_node, length,
+                               flag & 3);
     return ret;
 }
 

libisofs/util.c

@@ -1,7 +1,7 @@
 /*
  * Copyright (c) 2007 Vreixo Formoso
  * Copyright (c) 2007 Mario Danic
- * Copyright (c) 2009 - 2012 Thomas Schmitt
+ * Copyright (c) 2009 - 2015 Thomas Schmitt
  * 
  * This file is part of the libisofs project; you can redistribute it and/or 
  * modify it under the terms of the GNU General Public License version 2 
@@ -17,6 +17,7 @@
 #include "libisofs.h"
 #include "messages.h"
 #include "joliet.h"
+#include "node.h"
 #include "../version.h"
 
 #include <stdlib.h>
@@ -435,6 +436,7 @@ int str2ascii(const char *icharset, const char *input, char **output)
 
         ret_ = malloc(numchars + 1);
         if (ret_ == NULL) {
+            free(wsrc_);
             return ISO_OUT_OF_MEM;
         }
         outbytes = numchars;
@@ -444,7 +446,9 @@ int str2ascii(const char *icharset, const char *input, char **output)
         conv_ret = iso_iconv_open(&conv, "ASCII", "WCHAR_T", 0);
         if (conv_ret <= 0) {
             free(wsrc_);
+            wsrc_ = NULL;
             free(ret_);
+            ret = ret_ = NULL;
         }
     } else if (result != (int) ISO_CHARSET_CONV_ERROR)
         return result;
@@ -593,8 +597,10 @@ int str2ucs(const char *icharset, const char *input, uint16_t **output)
         loop_limit = inbytes + 3;
 
         ret_ = malloc((numchars+1) * sizeof(uint16_t));
-        if (ret_ == NULL)
+        if (ret_ == NULL) {
+            free(wsrc_);
             return ISO_OUT_OF_MEM;
+        }
         outbytes = numchars * sizeof(uint16_t);
         ret = ret_;
 
@@ -602,7 +608,9 @@ int str2ucs(const char *icharset, const char *input, uint16_t **output)
         conv_ret = iso_iconv_open(&conv, "UCS-2BE", "WCHAR_T", 0);
         if (conv_ret <= 0) {
             free(wsrc_);
+            wsrc_ = NULL;
             free(ret_);
+            ret = ret_ = NULL;
         }
     } else if (result != (int) ISO_CHARSET_CONV_ERROR)
         return result;
@@ -723,8 +731,10 @@ int str2utf16be(const char *icharset, const char *input, uint16_t **output)
             loop_limit = inbytes + 3;
 
             ret_ = malloc((2 * numchars+1) * sizeof(uint16_t));
-            if (ret_ == NULL)
+            if (ret_ == NULL) {
+                free(wsrc_);
                 return ISO_OUT_OF_MEM;
+            }
             outbytes = 2 * numchars * sizeof(uint16_t);
             ret = ret_;
 
@@ -1574,8 +1584,14 @@ void iso_datetime_7(unsigned char *buf, time_t t, int always_gmt)
 #else
     if (tm.tm_isdst < 0)
         tm.tm_isdst = 0;
-    tzoffset = ( - timezone / 60 / 15 ) + 4 * tm.tm_isdst;
+#ifndef Libburnia_timezonE
+#define Libburnia_timezonE timezone
 #endif
+ #if Libburnia_timezonE == 0
+    always_gmt = 1;
+ #endif
+    tzoffset = ( - Libburnia_timezonE / 60 / 15 ) + 4 * tm.tm_isdst;
+#endif /* ! HAVE_TM_GMTOFF */
 
     if (tzoffset > 52 || tzoffset < -48 || always_gmt) {
         /* absurd timezone offset, represent time in GMT */
@@ -1620,8 +1636,14 @@ void iso_datetime_17(unsigned char *buf, time_t t, int always_gmt)
 #else
     if (tm.tm_isdst < 0)
         tm.tm_isdst = 0;
-    tzoffset = ( - timezone / 60 / 15 ) + 4 * tm.tm_isdst;
+#ifndef Libburnia_timezonE
+#define Libburnia_timezonE timezone
 #endif
+ #if Libburnia_timezonE == 0
+    always_gmt = 1;
+ #endif
+    tzoffset = ( - Libburnia_timezonE / 60 / 15 ) + 4 * tm.tm_isdst;
+#endif /* ! HAVE_TM_GMTOFF */
 
     if (tzoffset > 52 || tzoffset < -48 || always_gmt) {
         /* absurd timezone offset, represent time in GMT */
@@ -1782,6 +1804,7 @@ time_t iso_datetime_read_7(const uint8_t *buf)
     tm.tm_hour = buf[3];
     tm.tm_min = buf[4];
     tm.tm_sec = buf[5];
+    tm.tm_isdst = 0;
 
     return timegm(&tm) - ((int8_t)buf[6]) * 60 * 15;
 }
@@ -1798,8 +1821,9 @@ time_t iso_datetime_read_17(const uint8_t *buf)
     sscanf((char*)&buf[12], "%2d", &tm.tm_sec);
     tm.tm_year -= 1900;
     tm.tm_mon -= 1;
+    tm.tm_isdst = 0;
 
-    return timegm(&tm) - ((int8_t)buf[6]) * 60 * 15;
+    return timegm(&tm) - ((int8_t)buf[16]) * 60 * 15;
 }
 
 /**
@@ -2214,6 +2238,7 @@ unexpected_type:;
         goto ex;
     } else if (range_start != ctx_start_lba) {
         ret = ISO_MD5_TAG_MISPLACED;
+        goto ex;
     }
     ret = iso_md5_clone(ctx, &cloned_ctx);
     if (ret < 0)
@@ -2282,7 +2307,7 @@ int iso_clone_mem(char *in, char **out, size_t size)
     if (*out == NULL)
         return ISO_OUT_OF_MEM;
     memcpy(*out, in, size);
-    return 1;
+    return ISO_SUCCESS;
 }
     
 
@@ -2293,3 +2318,133 @@ int iso_clone_mgtd_mem(char *in, char **out, size_t size)
     return iso_clone_mem(in, out, size);
 }
 
+
+/** Convert a text into a number of type double and multiply it by unit code
+    [kmgt] (2^10 to 2^40) or [s] (2048) or [d] (512).
+    (Also accepts capital letters.)
+    @param text Input like "42", "223062s", "3m" or "-1g"
+    @param flag Bitfield for control purposes:
+                bit0= return -1 rathern than 0 on failure
+                bit1= if scaled then compute the last byte of the last unit
+    @return The derived value
+*/
+off_t iso_scanf_io_size(char *text, int flag)
+{
+    int c;
+    off_t ret = 0, fac = 1;
+    char *rpt;
+
+    for (rpt = text; *rpt >= '0' && *rpt <= '9'; rpt++)
+        ret = ret * 10 + (*rpt - '0');
+    if (rpt == text)
+        return (off_t) (flag & 1 ? -1 : 0);
+    c = *rpt;
+    if (c=='k' || c=='K')
+        fac = 1024;
+    else if (c=='m' || c=='M')
+        fac = 1024 * 1024;
+    else if (c=='g' || c=='G')
+        fac = 1024 * 1024 * 1024;
+    else if (c=='t' || c=='T')
+        fac = ((off_t) 1024) * 1024 * 1024 * 1024;
+    else if (c=='s' || c=='S')
+        fac = 2048;
+    else if (c=='d' || c=='D')
+        fac = 512;
+    ret *= fac;
+    if (flag & 2)
+        ret += fac - 1;
+    return ret;
+}
+
+
+/* Find backward from idx the start byte of a possible UTF-8 character.
+     https://en.wikipedia.org/wiki/UTF-8#Description
+*/
+static
+int find_utf8_start(char *name, int idx, int flag)
+{
+    unsigned char *uname, uch;
+    int i;
+
+    uname= (unsigned char *) name;
+    if ((uname[idx] & 0xc0) != 0x80)
+        return idx;                                /* not an UTF-8 tail byte */
+    for (i = 0; i < 5 && idx - 1 - i >= 0; i++) {                                                                           /* up to deprecated 6-byte codes */
+        uch = uname[idx - 1 - i];
+        if ((uch & 0xe0) == 0xc0 || (uch & 0xf0) == 0xe0 ||
+            (uch & 0xf8) == 0xf0 || (uch & 0xfc) == 0xf8 ||
+            (uch & 0xfe) == 0xfc)
+            return (idx - 1 - i);                  /* UTF-8 start byte found */
+        if ((uch & 0xc0) != 0x80)
+          return idx;                 /* not an UTF-8 tail byte, so no UTF-8 */
+    }
+    return idx;                                      /* no UTF-8 start found */
+}
+
+/* @param flag bit0= do not issue warning message
+*/
+int iso_truncate_rr_name(int truncate_mode, int truncate_length,
+                         char *name, int flag)
+{
+    int neck, goal, ret, l, i;
+    static int hash_size = 32;
+    void *ctx = NULL;
+    char hashval[16];
+
+    l = strlen(name);
+    if (l <= truncate_length)
+        return ISO_SUCCESS;
+    if (truncate_mode == 0)
+        return ISO_RR_NAME_TOO_LONG;
+
+    /* Compute hash */
+    ret = iso_md5_start(&ctx);
+    if (ret < 0)
+        goto ex;
+    ret = iso_md5_compute(ctx, name, l > 4095 ? 4095 : l);
+    if (ret < 0)
+        goto ex;
+    ret = iso_md5_end(&ctx, hashval);
+    if (ret < 0)
+        goto ex;
+
+    if (!(flag & 1))
+        iso_msg_submit(-1, ISO_RR_NAME_TRUNCATED, 0,
+                     "File name had to be truncated and MD5 marked: %s", name);
+
+    /* Avoid to produce incomplete UTF-8 characters */
+    goal = truncate_length - hash_size - 1;
+    neck = find_utf8_start(name, goal, 0);
+    for (; neck < goal; neck++)
+        name[neck] = '_';
+
+    /* Write colon and hash text over end of truncated name */
+    name[goal] = ':';
+    goal++;
+    for (i = 0; goal < truncate_length - 1 && i < hash_size / 2; goal += 2) {
+        sprintf(name + goal, "%2.2x", *((unsigned char *) (hashval + i)));
+        i++;
+    }
+    name[truncate_length] = 0;
+
+    ret = ISO_SUCCESS;
+ex:;
+    if (ctx != NULL)
+        iso_md5_end(&ctx, hashval);
+    return ret;
+}
+
+/* API */
+int iso_truncate_leaf_name(int mode, int length, char *name, int flag)
+{
+    int ret;
+
+    if (mode < 0 || mode > 1)
+        return ISO_WRONG_ARG_VALUE;
+    if (length < 64 || length > LIBISOFS_NODE_NAME_MAX)
+        return ISO_WRONG_ARG_VALUE;
+    ret = iso_truncate_rr_name(mode, length, name, 1);
+    return ret;
+}
+

libisofs/util.h

@@ -21,6 +21,12 @@
 
 #include <time.h>
 
+#ifdef HAVE_STDLIB_H
+#include <stdlib.h>
+#endif
+
+#include <fcntl.h>
+
 #ifndef MAX
 #   define MAX(a, b) (((a) > (b)) ? (a) : (b))
 #endif
@@ -401,6 +407,12 @@ size_t iso_rbtree_get_size(IsoRBTree *tree);
 void **iso_rbtree_to_array(IsoRBTree *tree, int (*include_item)(void *), 
                            size_t *size);
 
+/** Predict the size of the array which gets returned by iso_rbtree_to_array().
+ */
+size_t iso_rbtree_count_array(IsoRBTree *tree, size_t initial_count,
+                              int (*include_item)(void *));
+
+
 /**
  * Create a new hash table.
  * 
@@ -552,6 +564,8 @@ int iso_util_bin_to_hex(char *target, uint8_t *bytes, int num_bytes, int flag);
 int iso_util_hex_to_bin(char *hex, char *bin, int bin_size, int *bin_count,
                         int flag);
 
+int iso_truncate_rr_name(int truncate_mode, int truncate_length,
+                         char *name, int flag);
 
 /* ------------------------------------------------------------------------- */
 
@@ -627,6 +641,16 @@ int iso_clone_mem(char *in, char **out, size_t size);
 */
 int iso_clone_mgtd_mem(char *in, char **out, size_t size);
 
+/** Convert a text into a number of type double and multiply it by unit code
+    [kmgt] (2^10 to 2^40) or [s] (2048) or [d] (512).
+    (Also accepts capital letters.)
+    @param text Input like "42", "223062s", "3m" or "-1g"
+    @param flag Bitfield for control purposes:
+                bit0= return -1 rathern than 0 on failure
+                bit1= if scaled then compute the last byte of the last unit
+    @return The derived value
+*/
+off_t iso_scanf_io_size(char *text, int flag);
 
 /* ------------------------------------------------------------------------- */
 

libisofs/util_rbtree.c

@@ -308,3 +308,38 @@ void ** iso_rbtree_to_array(IsoRBTree *tree, int (*include_item)(void *),
     return array;
 }
 
+
+static
+size_t rbtree_count_array_aux(struct iso_rbnode *root, size_t pos,
+                              int (*include_item)(void *))
+{
+    if (root == NULL) {
+        return pos;
+    }
+    pos = rbtree_count_array_aux(root->ch[0], pos, include_item);
+    if (include_item == NULL || include_item(root->data)) {
+
+/*
+{
+IsoFileSrc* src = (IsoFileSrc*) root->data;
+fprintf(stderr, "libisofs_DEBUG: rbtree_count_array_aux : not taken : '%s'\n",
+                iso_stream_get_source_path(src->stream, 0));
+}       
+*/
+
+        pos++;
+    }
+    pos = rbtree_count_array_aux(root->ch[1], pos, include_item);
+    return pos;
+}
+
+
+size_t iso_rbtree_count_array(IsoRBTree *tree, size_t initial_count,
+                          int (*include_item)(void *))
+{
+    size_t pos;
+
+    pos = rbtree_count_array_aux(tree->root, initial_count, include_item);
+    return pos;
+}
+

libisofs/writer.h

@@ -29,13 +29,13 @@ struct Iso_Image_Writer
 };
 
 /**
- * This is the function all Writers shoudl call to write data to image.
+ * This is the function all Writers should call to write data to image.
  * Currently, it is just a wrapper for write(2) Unix system call. 
  * 
  * It is implemented in ecma119.c
  * 
  * @return
- *      1 on sucess, < 0 error
+ *      1 on success, < 0 error
  */
 int iso_write(Ecma119Image *target, void *buf, size_t count);