B-C-1.04_05

git-svn-id: http://perl-compiler.googlecode.com/svn/trunk@6 ed534f1a-1453-0410-ab30-dfc593a8b23c

B/Bytecode.pm

@@ -738,13 +738,15 @@ sub compile {
 	    *newasm = *endasm = sub { };
 	    *asm = sub { print "    @_\n" };
 	    *nice = sub ($) { print "\n@_\n" };
+	} elsif (/^-v/) {
+	    *nice = sub ($) { print STDERR "@_\n" };
 	} elsif (/^-H/) {
 	    require ByteLoader;
 	    $head = "#! $^X
 use ByteLoader $ByteLoader::VERSION;
 ";
 	    #FIXME!
-	    undef $head;
+	    #undef $head;
 	} elsif (/^-k/) {
 	    keep_syn;
 	} elsif (/^-o(.*)$/) {
@@ -841,16 +843,15 @@ later by the ByteLoader module and executed as a regular Perl script.
 
 =over 4
 
+=item B<-H>
+
+prepend a C<use ByteLoader VERSION;> line to the produced bytecode.
+
 =item B<-b>
 
 Save all the BEGIN blocks. Normally only BEGIN blocks that C<require>
 other files (ex. C<use Foo;>) are saved.
 
-=item B<-H>
-
-prepend a C<use ByteLoader VERSION;> line to the produced bytecode.
-Does not work currently.
-
 =item B<-k>
 
 keep the syntax tree - it is stripped by default.
@@ -864,6 +865,11 @@ put the bytecode in <outfile> instead of dumping it to STDOUT.
 scan the script for C<# line ..> directives and for <goto LABEL>
 expressions. When gotos are found keep the syntax tree.
 
+=item B<-S>
+
+Output assembler source rather than piping it
+through the assembler and outputting bytecode.
+
 =back
 
 =head1 KNOWN BUGS

B/C.pm

@@ -8,7 +8,7 @@
 
 package B::C;
 
-our $VERSION = '1.04_04';
+our $VERSION = '1.04_05';
 
 package B::C::Section;
 

ByteLoader/bytecode.h

@@ -201,7 +201,6 @@ typedef HEK *hekindex;
         if(av_len((AV*) PL_regex_pad[0]) > -1) { \
             repointer = av_pop((AV*)PL_regex_pad[0]); \
             cPMOPx(o)->op_pmoffset = SvIV(repointer); \
-            SvREPADTMP_off(repointer); \
             sv_setiv(repointer,PTR2IV(rx)); \
         } else { \
             repointer = newSViv(PTR2IV(rx)); \
@@ -291,6 +290,10 @@ typedef HEK *hekindex;
 #define set_clonex(fp)
 #endif
 
+#ifndef PL_preprocess
+#define PL_preprocess 0
+#endif
+
 #define BSET_data(dummy,arg)						\
     STMT_START {							\
 	GV *gv;								\
@@ -305,7 +308,7 @@ typedef HEK *hekindex;
 	set_clonex(PL_RSFP);						\
 	/* Mark this internal pseudo-handle as clean */			\
 	IoFLAGS(GvIOp(gv)) |= IOf_UNTAINT;				\
-	if (PL_preprocess)						\
+	if ((PERL_VERSION < 11) && PL_preprocess)			\
 	    IoTYPE(GvIOp(gv)) = IoTYPE_PIPE;				\
 	else if ((PerlIO*)PL_RSFP == PerlIO_stdin())			\
 	    IoTYPE(GvIOp(gv)) = IoTYPE_STD;				\
@@ -479,4 +482,9 @@ typedef HEK *hekindex;
 	    if (sz != PTRSIZE) {				\
 		HEADER_FAIL("different PTRSIZE");		\
 	    }							\
+	    BGET_strconst(str); /* 12345678 */	      		\
+	    if (strNE(str, "12345678")) {			\
+		HEADER_FAIL2("cannot yet convert different byteorders (want %s, you have %s)",	\
+			"12345678", str);			\
+	    }							\
 	} STMT_END

ByteLoader/byterun.c

@@ -55,7 +55,7 @@ byterun(pTHX_ struct byteloader_state *bstate)
     register int insn;
     U32 isjit = 0;
     U32 ix;
-    SV *specialsv_list[6];
+    SV *specialsv_list[7];
 
     BYTECODE_HEADER_CHECK;	/* croak if incorrect platform, set isjit if PLJC magic header */
     if (isjit) {
@@ -65,6 +65,10 @@ byterun(pTHX_ struct byteloader_state *bstate)
         bstate->bs_obj_list_fill = 31;
         bstate->bs_obj_list[0] = NULL; /* first is always Null */
         bstate->bs_ix = 1;
+	DEBUG_l( Perl_deb(aTHX_ "(bstate.bs_fdata.idx %d)\n", bstate->bs_fdata->idx));
+	DEBUG_l( Perl_deb(aTHX_ "(bstate.bs_fdata.next_out %d)\n", bstate->bs_fdata->next_out));
+	DEBUG_l( Perl_deb(aTHX_ "(bstate.bs_fdata.datasv %p:\"%s\")\n", bstate->bs_fdata->datasv,
+				 SvPV_nolen(bstate->bs_fdata->datasv)));
 
         specialsv_list[0] = Nullsv;
         specialsv_list[1] = &PL_sv_undef;
@@ -75,7 +79,8 @@ byterun(pTHX_ struct byteloader_state *bstate)
         specialsv_list[6] = (SV*)pWARN_STD;
 
         while ((insn = BGET_FGETC()) != EOF) {
-	    switch (insn) {
+	  DEBUG_l( Perl_deb(aTHX_ "(insn %d)\n", insn));
+	  switch (insn) {
 	  case INSN_COMMENT:		/* 35 */
 	    {
 		comment_t arg;

ByteLoader/jitrun.c

@@ -94,7 +94,7 @@ jitrun(pTHX_ struct byteloader_state *bstate)
 
     jit_func bc_func = (jit_func) (jit_set_ip(codeBuffer).iptr); /* Function ptr */
 #ifdef DEBUGGING
-    disassemble(stderr, codeBuffer, jit_get_ip().ptr);
+    //disassemble(stderr, codeBuffer, jit_get_ip().ptr);
 #endif
     jit_flush_code(codeBuffer, jit_get_ip().ptr);
 

ByteLoader/jitrun.h

@@ -59,8 +59,9 @@
 */
 
 /* DATA handle reader */
-int jit_getc(struct byteloader_fdata *);
-int jit_read(struct byteloader_fdata *, char *, size_t, size_t);
+/* int bl_getc(struct byteloader_fdata *);
+   int bl_read(struct byteloader_fdata *, char *, size_t, size_t);
+*/
 
 #include <lightning.h>
 

Changes

@@ -1,6 +1,17 @@
 Started with B-C-1.04_01. The perl compiler was previously in CORE.
 TODO: Try to get info about earlier versions.
 
+1.04_05 2008-02-18 rurban
+	* added t/test.pl and t/test*.sh to MANIFEST. 
+	* fixed ByteLoader reading from the <DATA> filter.
+	* fixed -H .plc header parsing
+	* updated Bytecode options in NOTES and pod
+	* added -O=Bytecode,-v option
+	* fixed PL_preprocess and SvREPADTMP_off for 5.11
+
+1.04_04 2008-02-08 rurban
+	* added jitcompiler macros
+
 1.04_03 2008-02-05 rurban
 	* added Jit and Asm layout, fixed Makefile deps
 

MANIFEST

@@ -58,6 +58,9 @@ t/c.t
 t/cc.t
 t/o.t
 t/stash.t
+t/test.pl
+t/testplc.sh
+t/testc.sh
 TESTS
 Todo
 META.yml                                 Module meta-data (added by MakeMaker)

META.yml

@@ -1,6 +1,6 @@
 --- #YAML:1.0
 name:                B-C
-version:             1.04_04
+version:             1.04_05
 abstract:            ~
 license:             ~
 author:              ~

Makefile.PL

@@ -1,15 +1,15 @@
 use ExtUtils::MakeMaker;
 use Config;
 use File::Spec;
-use 5.009005;
+#use 5.009005;
 
 my $core = grep { $_ eq 'PERL_CORE=1' } @ARGV;
 
 WriteMakefile(
     NAME	    => "B::C",
     VERSION_FROM    => "B/C.pm",
 	         # 'scripts/assemble','scripts/disassemble',
-    PL_FILES => {'script/perlcc.PL'  => 'bin/perlcc' },
+    PL_FILES => {'script/perlcc.PL'  => '$(INST_BIN)/perlcc' },
     PREREQ_PM 	    => {
 	'B::Concise' => '0.74',
 	'B' => '1.17',
@@ -28,6 +28,10 @@ sub headerpath {
 
 package MY;
 
+sub x_libscan {
+  return 0 unless $_[1] =~ /\.pm$/;
+}
+
 sub post_constants {
     "\nLIBS = $Config::Config{libs}\n"
 }

NOTES

@@ -2,6 +2,8 @@ C backend invocation
 	If there are any non-option arguments, they are taken to be
 	names of objects to be saved (probably doesn't work properly yet).
 	Without extra arguments, it saves the main program.
+        -q 		Be quiet. STDOUT goes to $O::BEGIN_output
+        -qq		Be very quiet. Also suppress "Syntax OK"
 	-ofilename	Output to filename instead of STDOUT
 	-v		Verbose (currently gives a few compilation statistics)
 	--		Force end of options
@@ -35,7 +37,10 @@ Examples
 CC backend invocation
 	If there are any non-option arguments, they are taken to be names of
 	subs to be saved. Without extra arguments, it saves the main program.
+        -q 		Be quiet. STDOUT goes to $O::BEGIN_output
+        -qq		Be very quiet. Also suppress "Syntax OK"
 	-ofilename	Output to filename instead of STDOUT
+NYI	-v              Be verbose.
 	--		Force end of options
 	-uPackname	Force apparently unused subs from package Packname to
 			be compiled. This allows programs to use eval "foo()"
@@ -101,7 +106,10 @@ Bytecode backend invocation
 	If there are any non-option arguments, they are taken to be
 	names of objects to be saved (probably doesn't work properly yet).
 	Without extra arguments, it saves the main program.
+        -q 		Be quiet. STDOUT goes to $O::BEGIN_output
+        -qq		Be very quiet. Also suppress "Syntax OK"
 	-ofilename	Output to filename instead of STDOUT.
+NYI	-v              Be verbose.
 	--		Force end of options.
 	-f		Force optimisations on or off one at a time.
 			Each can be preceded by no- to turn the option off.
@@ -121,24 +129,34 @@ Bytecode backend invocation
 			run-time but leaving them out will make it impossible
 			to recompile or disassemble the resulting program.
 			It will also stop "goto label" statements from working.
-	-On		Optimisation level (n = 0, 1, 2, ...). -O means -O1.
+NYI	-On		Optimisation level (n = 0, 1, 2, ...). -O means -O1.
 			-O1 sets -fcompress-nullops -fomit-sequence numbers.
 			-O6 adds -fstrip-syntax-tree.
-	-D		Debug options (concat or separate flags like perl -D)
+NYI	-D		Debug options (concat or separate flags like perl -D)
 		o	OPs, prints each OP as it's processed.
 		b	print debugging information about bytecompiler progress
 		a	tells the assembler to include source assembler lines
 			in its output as bytecode comments.
 		C	prints each CV taken from the final symbol tree walk.
 	-S		Output assembler source rather than piping it
 			through the assembler and outputting bytecode.
-	-m		Compile as a module rather than a standalone program.
+	-H		add #! perl shebang header
+	-s		scan and keep keep syntax tree if goto op found.
+			scan the script for C<# line ..> directives and for <goto LABEL>
+			expressions. When gotos are found keep the syntax tree.
+	-b		Save all the BEGIN blocks. Normally only BEGIN blocks that require
+			other files (ex. use Foo;) are saved.
+	-k		keep syntax tree to disassemble the plc.
+			it is stripped by default.
+	-TI		testing, dump the @INC av
+	-TF     file	testing, sets COP::file	
+   	-m		Compile as a module rather than a standalone program.
 			Currently this just means that the bytecodes for
 			initialising main_start, main_root and curpad are
 			omitted.
 
 Example
-	perl -MO=Bytecode,-O6,-o,foo.plc foo.pl
+	perl -MO=Bytecode,-O6,-H,-ofoo.plc foo.pl
 
 	perl -MO=Bytecode,-S foo.pl > foo.S
 	assemble foo.S > foo.plc

Todo

@@ -16,7 +16,6 @@ CvOUTSIDE for ordinary subs
 DATA filehandle for standalone Bytecode program (easy)
 DATA filehandle for multiple bytecode-compiled modules (harder)
 DATA filehandle for C-compiled program (yet harder)
-parse -H Bytecode header
 
 * Features
 
@@ -31,6 +30,7 @@ module dependencies
 new Asm and JIT backend, PLJC ByteLoader magic
 
 * Optimisations
+
 collapse LISTOPs to UNOPs or BASEOPs
 compile-time qw(), constant subs
 global analysis of variables, type hints etc.

bytecode.pl

@@ -110,7 +110,9 @@ package B::Asmdata;
     register int insn;
     U32 isjit = 0;
     U32 ix;
-    SV *specialsv_list[6];
+EOT
+printf BYTERUN_C "    SV *specialsv_list[%d];\n", scalar @specialsv_name;
+print BYTERUN_C <<'EOT';
 
     BYTECODE_HEADER_CHECK;	/* croak if incorrect platform, set isjit if PLJC magic header */
     if (isjit) {
@@ -120,6 +122,10 @@ package B::Asmdata;
         bstate->bs_obj_list_fill = 31;
         bstate->bs_obj_list[0] = NULL; /* first is always Null */
         bstate->bs_ix = 1;
+	DEBUG_l( Perl_deb(aTHX_ "(bstate.bs_fdata.idx %d)\n", bstate->bs_fdata->idx));
+	DEBUG_l( Perl_deb(aTHX_ "(bstate.bs_fdata.next_out %d)\n", bstate->bs_fdata->next_out));
+	DEBUG_l( Perl_deb(aTHX_ "(bstate.bs_fdata.datasv %p:\"%s\")\n", bstate->bs_fdata->datasv,
+				 SvPV_nolen(bstate->bs_fdata->datasv)));
 
 EOT
 
@@ -130,7 +136,8 @@ package B::Asmdata;
 print BYTERUN_C <<'EOT';
 
         while ((insn = BGET_FGETC()) != EOF) {
-	    switch (insn) {
+	  DEBUG_l( Perl_deb(aTHX_ "(insn %d)\n", insn));
+	  switch (insn) {
 EOT
 
 

jitcompiler.pl

@@ -4,24 +4,31 @@
 
 =head1 JIT or ASM?
 
-Two possible native code attempts:
+Possible native code attempts:
 
-1. Use B<GNU lightning> to compile to jit at eval-time, and
-   jitrun the whole codebuffer, the appended insns instead
-   of looping through the opcode tree.
-   Possibly dump the jitted codebuffer as PLJC .plc files, to save
-   next startup parse time.
+1. Use some jit library, such  as B<GNU lightning>,  to compile to jit
+   at  eval-time, and jitrun the  whole codebuffer, the appended insns
+   instead  of  looping through  the  opcode tree.  Possibly  dump the
+   jitted codebuffer  as PLJC .plc files, to  save next  startup parse
+   time.
 
    This must be done with jit macro trickery on the C level.
 
-2. Use a self-written B<B::Asm> library to dump architectur specific
+   LLVM has some nice SSA optimizations.
+   See L<http://llvm.org/ProjectsWithLLVM/2002-Spring-CS497CZ-Jello.pdf>
+
+2. Use perl6 jit.
+   http://www.parrotcode.org/docs/jit.html#EXAMPLE
+   This already has optimizations for immediate and strict integer ops.
+
+3. Use a self-written B<B::Asm> library to dump architectur specific
    asm insns directly.
    Either dump the jitted codebuffer as PLJC .plc files, or construct
    the necessary OS headers directly. For writing the PE/COFF format
    see e.g. my L<C::DynaLib> pl2exe or Audreys L<Win32::Exe>, for writing
    Elf see e.g. L<GNU libjit>.
 
-   This could be done on a higher Perl level, and avoids the lightning
+   This could be done on a higher Perl level, and avoids the C-level
    macro abstraction.
 
 =head1 OPTIMIZATIONS
@@ -166,7 +173,7 @@ BEGIN
 
     jit_func bc_func = (jit_func) (jit_set_ip(codeBuffer).iptr); /* Function ptr */
 #ifdef DEBUGGING
-    disassemble(stderr, codeBuffer, jit_get_ip().ptr);
+    //disassemble(stderr, codeBuffer, jit_get_ip().ptr);
 #endif
     jit_flush_code(codeBuffer, jit_get_ip().ptr);
 
@@ -235,8 +242,9 @@ BEGIN
 */
 
 /* DATA handle reader */
-int jit_getc(struct byteloader_fdata *);
-int jit_read(struct byteloader_fdata *, char *, size_t, size_t);
+/* int bl_getc(struct byteloader_fdata *);
+   int bl_read(struct byteloader_fdata *, char *, size_t, size_t);
+*/
 
 #include <lightning.h>
 

script/perlcc

@@ -1,5 +1,5 @@
-#!/usr/local/bin/perl5.10.0-nt
-    eval 'exec /usr/local/bin/perl5.10.0-nt -S $0 ${1+"$@"}'
+#!/usr/bin/perl
+    eval 'exec /usr/bin/perl -S $0 ${1+"$@"}'
     if $running_under_some_shell;
 --$running_under_some_shell;