Add config script to generate Makefile based on flags.

Add install target and installation readme file.
A few comment changes.

INSTALL

@@ -0,0 +1,46 @@
+Copyright (c) 2012 Moinak Ghosh
+
+Basic Installation
+==================
+The simplest process to build and install this utility is:
+
+./config
+make
+make install
+
+Strictly speaking the 'make install' step is not required to
+run the utility as it is a single stand alone program for now.
+
+The config script controls various compile time parameters and
+generates the actual Makefile. You can get a short description
+of all the options by running:
+
+./config --help
+
+A more descriptive account is given below:
+
+./config [<options>]
+
+<options> can be one of the following:
+
+--prefix=<val>		The installation prefix. 
+			This is set to /usr by default. However alternate
+			prefix needs to eb used during packaging.
+
+--enable-debug		Enable debug mode compilation.
+			This reduces the compiler optimization level to basic
+			and taks out all the loop optimization flags. This is
+			primary to aid debugging.
+
+--disable-allocator	Disable use of internal memory allocator mechanism.
+			The internal allocator can be totally disabled by setting
+			this build time flag. It is also possible to dynamically
+			disable the allocator by setting the following env variable:
+			ALLOCATOR_BYPASS=1
+
+--enable-debug-stats	Enable printing of some verbose debug info.
+			This at present shows some info related to Dedupe
+			efficiency.
+
+--help			Display the help message.
+

Makefile.in

@@ -65,39 +65,45 @@ LZPOBJS = $(LZPSRCS:.c=.o)
 BAKFILES = *~ lzma/*~ lzfx/*~ lz4/*~ rabin/*~ bsdiff/*~ lzp/*~
 
 RM = rm -f
-CPPFLAGS = -I. -I./lzma -I./lzfx -I./lz4 -I./rabin -I./bsdiff -DNODEFAULT_PROPS \
+COMMON_CPPFLAGS = -I. -I./lzma -I./lzfx -I./lz4 -I./rabin -I./bsdiff -DNODEFAULT_PROPS \
 	-DFILE_OFFSET_BITS=64 -D_REENTRANT -D__USE_SSE_INTRIN__ -D_LZMA_PROB32 \
 	-I./lzp
-VEC_FLAGS = -ftree-vectorize
+COMMON_VEC_FLAGS = -ftree-vectorize
-LOOP_OPTFLAGS = $(VEC_FLAGS) -floop-interchange -floop-block
+COMMON_LOOP_OPTFLAGS = $(VEC_FLAGS) -floop-interchange -floop-block
 LDLIBS = -ldl -lbz2 $(ZLIB_DIR) -lz -lm
 OBJS = $(MAINOBJS) $(LZMAOBJS) $(PPMDOBJS) $(LZFXOBJS) $(LZ4OBJS) $(CRCOBJS) \
 $(RABINOBJS) $(BSDIFFOBJS) $(LZPOBJS)
 
-ifdef DEBUG
+DEBUG_LINK = g++ -m64 -pthread -msse3
-LINK = g++ -m64 -pthread -msse3
+DEBUG_COMPILE = gcc -m64 -g -msse3 -c
-COMPILE = gcc -m64 -g -msse3 -c
+DEBUG_COMPILE_cpp = g++ -m64 -g -msse3 -c
-COMPILE_cpp = g++ -m64 -g -msse3 -c
+DEBUG_VEC_FLAGS = 
-VEC_FLAGS = 
+DEBUG_LOOP_OPTFLAGS = 
-LOOP_OPTFLAGS = 
+DEBUG_GEN_OPT = -O -fno-omit-frame-pointer
-GEN_OPT = -O -fno-omit-frame-pointer
+DEBUG_RABIN_OPT = -O -fno-omit-frame-pointer
-RABIN_OPT = -O -fno-omit-frame-pointer
+DEBUG_CPPFLAGS = $(COMMON_CPPFLAGS)
-else
-GEN_OPT = -O3
-RABIN_OPT = -O2
-LINK = g++ -m64 -pthread -msse3
-COMPILE = gcc -m64 -msse3 -c
-COMPILE_cpp = g++ -m64 -msse3 -c
-CPPFLAGS += -DNDEBUG
-endif
 
-ifdef DEBUG_NO_SLAB
+RELEASE_LINK = g++ -m64 -pthread -msse3
-CPPFLAGS += -DDEBUG_NO_SLAB
+RELEASE_COMPILE = gcc -m64 -msse3 -c
-endif
+RELEASE_COMPILE_cpp = g++ -m64 -msse3 -c
+RELEASE_VEC_FLAGS = $(COMMON_VEC_FLAGS)
+RELEASE_LOOP_OPTFLAGS = $(COMMON_LOOP_OPTFLAGS)
+RELEASE_CPPFLAGS = $(COMMON_CPPFLAGS) -DNDEBUG
+RELEASE_GEN_OPT = -O3
+RELEASE_RABIN_OPT = -O2
 
-ifdef DEBUG_STATS
+NO_SLAB_CPPFLAGS = -DDEBUG_NO_SLAB
-CPPFLAGS += -DDEBUG_STATS
+DEBUG_STATS_CPPFLAGS = -DDEBUG_STATS
-endif
+
+LINK = @LINK@
+COMPILE = @COMPILE@
+COMPILE_cpp = @COMPILE_cpp@
+VEC_FLAGS = @VEC_FLAGS@
+LOOP_OPTFLAGS = @LOOP_OPTFLAGS@
+CPPFLAGS = @CPPFLAGS@ @NO_SLAB_CPPFLAGS@ @DEBUG_STATS_CPPFLAGS@
+GEN_OPT = @GEN_OPT@
+RABIN_OPT = @RABIN_OPT@
+PREFIX=@PREFIX@
 
 all: $(PROG)
 
@@ -134,3 +140,17 @@ $(PROG): $(OBJS)
 clean:
 	$(RM) $(PROG) $(OBJS) $(BAKFILES)
 
+distclean: clean
+	$(RM) Makefile
+
+install: $(PROG)
+	@mkdir -p $(PREFIX)/bin
+	@chmod 0755 $(PREFIX)/bin
+	@cp $(PROG) $(PREFIX)/bin
+	@chmod 0555 $(PREFIX)/bin/$(PROG)
+	@mkdir -p $(PREFIX)/share/doc/$(PROG)
+	@chmod 0755 $(PREFIX)/share $(PREFIX)/share/doc $(PREFIX)/share/doc/$(PROG)
+	@cp README.md $(PREFIX)/share/doc/$(PROG)/README
+	@chmod 0444 $(PREFIX)/share/doc/$(PROG)/README
+
+

config

@@ -0,0 +1,82 @@
+#!/bin/sh
+
+usage() {
+	prog=$1
+	cat << _EOF
+${prog} [<options>]
+
+<options> can be one of the following:
+
+--prefix=<val>		The installation prefix (default: /usr).
+--enable-debug		Enable debug mode compilation (default: disabled).
+--disable-allocator	Disable use of internal memory allocator mechanism (default: enabled).
+--enable-debug-stats	Enable printing of some verbose debug info (default: disabled).
+--help			Display this help message.
+
+_EOF
+}
+
+arg1=$1
+debug=0
+allocator=1
+debug_stats=0
+prefix=/usr
+while [ "${arg1}" != "" ]
+do
+	case "$arg1" in
+	--enable-debug) debug=1;;
+	--disable-allocator) allocator=0;;
+	--enable-debug-stats) debug_stats=1;;
+	--prefix=*)
+		pval=`echo ${arg1} | cut -f2 -d"="`
+		prefix=$pval
+	;;
+	--help) usage $0;;
+	*)
+		echo "Unrecognized option: ${arg1}"
+		exit 1
+	;;
+	esac
+	shift
+	arg1=$1
+done
+
+if [ $debug -eq 1 ]
+then
+	typ="DEBUG"
+else
+	typ="RELEASE"
+fi
+
+linkvar="LINK"
+compilevar="COMPILE"
+compilecppvar="COMPILE_cpp"
+vecflagsvar="VEC_FLAGS"
+loopoptflagsvar="LOOP_OPTFLAGS"
+cppflagsvar="CPPFLAGS"
+genoptvar="GEN_OPT"
+rabinoptvar="RABIN_OPT"
+noslabcppflagsvar="NO_SLAB_CPPFLAGS"
+debugstatscppflagsvar="DEBUG_STATS_CPPFLAGS"
+prefixvar="PREFIX"
+
+noslabcppflagsval=
+debugstatscppflagsval=
+
+[ $allocator -eq 0 ] && noslabcppflagsval='\$\(NO_SLAB_CPPFLAGS\)'
+[ $debug_stats -eq 1 ] && debugstatscppflagsval='\$\(DEBUG_STATS_CPPFLAGS\)'
+
+cat Makefile.in | sed "
+s#@${linkvar}@#\\\$\\(${typ}_${linkvar}\\)#g
+s#@${compilevar}@#\\\$\\(${typ}_${compilevar}\\)#g
+s#@${compilecppvar}@#\\\$\\(${typ}_${compilecppvar}\\)#g
+s#@${vecflagsvar}@#\\\$\\(${typ}_${vecflagsvar}\\)#g
+s#@${loopoptflagsvar}@#\\\$\\(${typ}_${loopoptflagsvar}\\)#g
+s#@${cppflagsvar}@#\\\$\\(${typ}_${cppflagsvar}\\)#g
+s#@${genoptvar}@#\\\$\\(${typ}_${genoptvar}\\)#g
+s#@${rabinoptvar}@#\\\$\\(${typ}_${rabinoptvar}\\)#g
+s#@${noslabcppflagsvar}@#${noslabcppflagsval}#g
+s#@${debugstatscppflagsvar}@#${debugstatscppflagsval}#g
+s#@${prefixvar}@#${prefix}#g
+" > Makefile
+

lzma/LzmaEnc.h

@@ -1,8 +1,3 @@
-/*
- * Copyright 2009 Sun Microsystems, Inc.  All rights reserved.
- * Use is subject to license terms.
- */
-
 /*  LzmaEnc.h -- LZMA Encoder
 2008-10-04 : Igor Pavlov : Public domain */
 

lzp/lzp.c

@@ -438,6 +438,11 @@ ssize_t lzp_decompress(const unsigned char * input, unsigned char * output, ssiz
     return (result == LZP_NO_ERROR) ? dataSize : result;
 }
 
+/*
+ * Counter-intuitively we use a larger hash (with better LZP compression) for lower global
+ * compression levels. So that LZP preprocessing plays along nicely with the primary
+ * compression algorithm being used and actually provides a benefit.
+ */
 int lzp_hash_size(int level) {
     if (level > 7) {
         return (LZP_DEFAULT_LZPHASHSIZE + 2);