Adding MS-DOS .com program support (msdos86). Still incomplete.

build.lua

@@ -11,6 +11,7 @@ vars.plats = {
 	"linux68k",
 	"linuxppc",
 	"linuxmips",
+	"msdos86",
 	"osx386",
 	"osxppc",
 	"pc86",

plat/msdos86/boot.s

@@ -0,0 +1,145 @@
+#
+! $Source$
+! $State$
+! $Revision$
+
+! Declare segments (the order is important).
+
+.sect .text
+.sect .rom
+.sect .data
+.sect .bss
+
+.sect .text
+
+begtext:
+	! Make sure we are running under MS-DOS 2 or above.
+	!
+	! While at it, also remember the actual DOS version, so that we know
+	! whether DOS gives us the program's name in the environment
+	! segment.  (DOS 3+ does; DOS 2.x does not.)
+	movb ah, 0x30
+	int 0x21
+	cmpb al, 2
+	xchg bx, ax
+	jc bad_sys
+
+	! Clear BSS.
+	mov di, begbss
+	mov cx, endbss+1
+	sub cx, di
+	shr cx, 1
+	xor ax, ax
+	cld
+	rep stosw
+
+	! Get the size of the environment variables plus (if present) the
+	! program name.  Also count the number of environment variables.
+	mov es, (0x002C)
+	xor di, di
+	! ax = 0 from above
+	cwd				! dx = count of env. vars.
+	mov cx, -1
+	scasb				! handle special case of empty env.
+	jz is_empty_env
+size_env:
+	inc dx
+	repnz scasb
+	scasb
+	jnz size_env
+is_empty_env:
+	cmpb bl, 2
+	jz no_argv0
+	scasw
+	repnz scasb
+no_argv0:
+
+	! Copy out the environment variables and (possibly) program name
+	! onto the stack.
+	mov si, di
+	dec si
+	and si, -2
+	std
+copy_env:
+	test si, si
+	eseg lodsw
+	push ax
+	jnz copy_env
+	mov cx, sp
+
+	! Reset DF and es properly.
+	cld
+	push ss
+	pop es
+
+	! Reserve space for argc and the argv and envp pointers on the
+	! stack.  These will be passed to __m_a_i_n later.
+	sub sp, 6
+	mov ax, sp
+
+	! Build up argc, argv[], and envp[].
+	push ax				! output buffer for argc, argv, envp
+	push bx				! MS-DOS version
+	push cx				! env. string data
+	push dx				! count of env. vars.
+	mov ax, 0x0080
+	push ax				! raw command line
+	call __sys_initmain
+	add sp, 10
+
+	! Bail out if something went wrong.
+	test ax, ax
+	jnz no_room
+
+	! argc, argv, and envp are now at the stack top.  Now go.
+	call __m_a_i_n
+	add sp, 6
+	push ax
+	call _exit
+
+bad_sys:
+	mov dx, bad_sys_msg
+dos_msg:
+	movb ah, 9
+	int 0x21
+	ret
+
+no_room:
+	mov dx, no_room_msg
+	call dos_msg
+	movb al, -1
+	jmp al_exit
+
+	! Exit.
+.define __exit
+.extern __exit
+.define EXIT
+.extern EXIT
+__exit:
+EXIT:
+	pop bx
+	pop ax
+al_exit:
+	movb ah, 0x4c
+	int 0x21
+
+! Define symbols at the beginning of our various segments, so that we can find
+! them. (Except .text, which has already been done.)
+
+.define begtext, begdata, begbss
+.sect .data;       begdata:
+.sect .rom;        begrom:
+.sect .bss;        begbss:
+
+.sect .rom
+
+! Some text messages.
+bad_sys_msg: .ascii 'Bad DOS$'
+no_room_msg: .ascii 'No room$'
+
+! Some magic data. All EM systems need these.
+
+.define .trppc, .ignmask, _errno
+.comm .trppc, 4
+.comm .ignmask, 4
+.comm _errno, 4

plat/msdos86/build-pkg.lua

@@ -0,0 +1,25 @@
+include("plat/build.lua")
+
+ackfile {
+	name = "boot",
+	srcs = { "./boot.s" },
+	vars = { plat = "msdos86" }
+}
+
+build_plat_libs {
+	name = "libs",
+	arch = "i86",
+	plat = "msdos86",
+}
+
+installable {
+	name = "pkg",
+	map = {
+		"+tools",
+		"+libs",
+        "./include+pkg",
+		["$(PLATIND)/msdos86/boot.o"] = "+boot",
+        ["$(PLATIND)/msdos86/libsys.a"] = "./libsys+lib",
+	}
+}
+

plat/msdos86/build-tools.lua

@@ -0,0 +1,21 @@
+include("plat/build.lua")
+
+build_as {
+	name = "as",
+	arch = "i86",
+}
+
+build_ncg {
+	name = "ncg",
+	arch = "i86",
+}
+
+return installable {
+	name = "tools",
+	map = {
+		["$(PLATDEP)/msdos86/as"] = "+as",
+		["$(PLATDEP)/msdos86/ncg"] = "+ncg",
+		["$(PLATIND)/descr/msdos86"] = "./descr",
+		"util/opt+pkg",
+	}
+}

plat/msdos86/descr

@@ -0,0 +1,86 @@
+# $Source$
+# $State$
+# $Revision$
+
+var w=2
+var wa=2
+var p=2
+var pa=2
+var s=2
+var sa=2
+var l=4
+var la=2
+var f=4
+var fa=2
+var d=8
+var da=2
+var x=8
+var xa=2
+var ARCH=i86
+var PLATFORM=msdos86
+var PLATFORMDIR={EM}/share/ack/{PLATFORM}
+var CPP_F=-D__MSDOS__ -D__DOS__ -D_DOS
+var ALIGN=-a0:2 -a1:2 -a2:2 -a3:2
+var MACHOPT_F=-m8
+var EGO_PLAT_FLAGS=-M{EM}/share/ack/ego/{ARCH}.descr
+
+# Override the setting in fe so that files compiled for this platform can see
+# the platform-specific headers.
+
+var C_INCLUDES=-I{PLATFORMDIR}/include -I{EM}/share/ack/include/ansi
+
+# These will be overridden under the small (separate I/D) memory model.
+
+var SEPID=-b0:0x100
+var LOD=aslod
+
+name be
+	from .m.g
+	to .s
+	program {EM}/lib/ack/{PLATFORM}/ncg
+	args <
+	stdout
+	need .e
+end
+name as
+	from .s.so
+	to .o
+	program {EM}/lib/ack/{PLATFORM}/as
+	args - -o > <
+	prep cond
+end
+name led
+	from .o.a
+	to .out
+	program {EM}/lib/ack/em_led
+	mapflag -l* LNAME={PLATFORMDIR}/lib*
+	mapflag -i SEPID=-b1:0
+	mapflag -fp FLOATS={EM}/{ILIB}fp
+	args {ALIGN} {SEPID} \
+		({RTS}:.b=-u _i_main) \
+	    (.e:{HEAD}={PLATFORMDIR}/boot.o) \
+		({RTS}:.ocm.bas.b={PLATFORMDIR}/c-ansi.o) \
+		({RTS}:.c={PLATFORMDIR}/c-ansi.o) \
+		({RTS}:.mod={PLATFORMDIR}/modula2.o) \
+		({RTS}:.p={PLATFORMDIR}/pascal.o) \
+		-o > < \
+		(.p:{TAIL}={PLATFORMDIR}/libpascal.a) \
+		(.b:{TAIL}={PLATFORMDIR}/libb.a) \
+		(.bas:{TAIL}={PLATFORMDIR}/libbasic.a) \
+		(.mod:{TAIL}={PLATFORMDIR}/libmodula2.a) \
+		(.ocm:{TAIL}={PLATFORMDIR}/liboccam.a) \
+		(.ocm.bas.mod.b.c.p:{TAIL}={PLATFORMDIR}/libc.a) \
+		{FLOATS?} \
+		(.e:{TAIL}={PLATFORMDIR}/libem.a \
+		           {PLATFORMDIR}/libsys.a \
+		           {PLATFORMDIR}/libend.a)
+	linker
+end
+name cv
+	from .out
+	to .exe
+	mapflag -i LOD=amzlod
+	program {EM}/bin/{LOD}
+	args < >
+	outfile msdos86.exe
+end

plat/msdos86/libsys/sys_initmain.c

@@ -0,0 +1,112 @@
+/* $Source$
+ * $State$
+ * $Revision$
+ */
+
+#include <stdlib.h>
+#include <string.h>
+#include <unistd.h>
+
+#define	OUT_OF_MEMORY (void*)(-1)	/* sbrk returns this on failure */
+
+struct for_main {
+	int argc;
+	char **argv;
+	char **envp;
+};
+
+/*
+ * Marshal the command line and environment variable data provided by
+ * MS-DOS, into the standard format expected by a C main(...) routine.  The
+ * environment variable data has been copied to the near data segment.
+ *
+ * Return zero if everything went well, non-zero otherwise.
+ */
+int _sys_initmain(char *arg_data, size_t n_env_vars, char *env_data,
+    unsigned msdos_ver, struct for_main *out)
+{
+	char **argv, **envp;
+	char c, *p, **pp;
+	unsigned char n;
+
+	/*
+	 * Allocate space for argv[] for the maximum possible number of
+	 * arguments.  We can shrink this later.
+	 */
+	if ((argv = sbrk(65 * sizeof(char *))) == OUT_OF_MEMORY)
+		return -1;
+	out->argv = argv;
+
+	/* Sanity-check the original command line. */
+	p = arg_data;
+	n = (unsigned char)*p;
+	if (n > 0x7E || p[1 + n] != '\r')
+		return -1;
+
+	/* Initialize argv[0] to an empty string first. */
+	*p = 0;
+	argv[0] = p;
+
+	/*
+	 * Split the command line string into separate arguments.
+	 * TODO: handle double-quoting, backslashes, etc.
+	 */
+	pp = argv + 1;
+	do
+	{
+		do
+			c = *++p;
+		while (c == ' ' || c == '\t');
+
+		if (c != '\r')
+		{
+			*pp++ = p;
+			do
+				c = *++p;
+			while (c != ' ' && c != '\t' && c != '\r');
+			*p = 0;
+		}
+	} while (c != '\r');
+
+	/* Wrap up argv[].  Compute argc.  Free up unneeded space. */
+	out->argc = pp - argv;
+	*pp++ = NULL;
+	if (brk(pp) != 0)
+		return -1;
+
+	/* Allocate space for envp[]. */
+	if ((envp = sbrk((n_env_vars + 1) * sizeof(char *))) == OUT_OF_MEMORY)
+		return -1;
+	out->envp = envp;
+
+	/*
+	 * Populate envp[].  The environment data is simply a series of ASCIIZ
+	 * strings, one after another, terminated by an empty string.
+	 */
+	pp = envp;
+	p = env_data;
+	while (*p)
+	{
+		*pp++ = p;
+		while (*++p);
+		++p;
+	}
+	*pp = NULL;
+
+	/*
+	 * If there is a program name (MS-DOS version is 3 or above), point
+	 * argv[0] to it.
+	 *
+	 * p points to the zero byte just before the count of strings (a
+	 * shortword) following the environment variables, so advance p by 3
+	 * bytes to get at the program name.
+	 */
+	if ((msdos_ver & 0x00ff) >= 3)
+	{
+		p += 3;
+		argv[0] = p;
+	}
+
+	/* We are done. */
+	return 0;
+}