os

commit df3145130f30f5a6499f50e5a1111cb58b6ab40c
parent 5bf8ba671da36de6e837384bf20b59ad0511eca6
Author: erai <erai@omiltem.net>
Date:   Fri, 31 Jan 2025 21:00:32 +0000

use rodata for strings

Diffstat:
M
as.om
 | 
120
+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++--------
M
cc0.c
 | 
144
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++---------------------
M
ir.om
 | 
34
+---------------------------------

3 files changed, 216 insertions(+), 82 deletions(-)
diff --git a/as.om b/as.om
@@ -189,6 +189,15 @@ struct assembler {
 	lineno: int;
 	prevfilename: *byte;
 	prevlineno: int;
+	blobs: *ablob;
+}
+
+struct ablob {
+	left: *ablob;
+	right: *ablob;
+	label: *label;
+	s: *byte;
+	slen: int;
 }
 
 struct symbol {
@@ -423,6 +432,79 @@ func fixup_label(c: *assembler, l: *label) {
 	}
 }
 
+func as_blob(c: *assembler, s: *byte, slen: int): *label {
+	var b: *ablob;
+	var link: **ablob;
+	var dir: int;
+
+	link = &c.blobs;
+	loop {
+		b = *link;
+		if !b {
+			break;
+		}
+
+		if b.slen < slen {
+			dir = memcmp(b.s, s, b.slen);
+			if dir == 0 {
+				dir = -1;
+			}
+		} else if b.slen > slen {
+			dir = memcmp(b.s, s, slen);
+			if dir == 0 {
+				dir = 1;
+			}
+		} else {
+			dir = memcmp(b.s, s, slen);
+		}
+
+		if dir == 0 {
+			return b.label;
+		} else if dir > 0 {
+			link = &b.left;
+		} else {  // dir < 0
+			link = &b.right;
+		}
+	}
+
+	b = alloc(c.a, sizeof(*b)) as *ablob;
+
+	b.label = mklabel(c);
+	b.s = alloc(c.a, slen);
+	b.slen = slen;
+
+	memcpy(b.s, s, slen);
+
+	*link = b;
+
+	return b.label;
+}
+
+func emit_blobs(c: *assembler, b: *ablob) {
+	var i: int;
+
+	if !b {
+		return;
+	}
+
+	emit_blobs(c, b.left);
+
+	fixup_label(c, b.label);
+
+	i = 0;
+	loop {
+		if i == b.slen {
+			break;
+		}
+
+		as_emit(c, b.s[i] as int);
+
+		i = i + 1;
+	}
+
+	emit_blobs(c, b.right);
+}
+
 func add_symbol(c: *assembler, name: *byte, l: *label) {
 	var s: *symbol;
 
@@ -471,6 +553,14 @@ func emit_sections(c: *assembler): int {
 	s.start = 0;
 	s.end = 0;
 
+	emit_align(c, 4096, OP_NOP);
+
+	add_section(c, ".rodata", SHT_PROGBITS);
+
+	emit_blobs(c, c.blobs);
+
+	emit_align(c, 4096, 0);
+
 	add_section(c, ".strtab", SHT_STRTAB);
 
 	y = c.symbols;
@@ -598,14 +688,21 @@ func emit_sections(c: *assembler): int {
 		as_emit(c, n >> 24);
 
 		// Flags
-		as_emit(c, 6);
-		as_emit(c, 0);
-		as_emit(c, 0);
-		as_emit(c, 0);
-		as_emit(c, 0);
-		as_emit(c, 0);
-		as_emit(c, 0);
-		as_emit(c, 0);
+		if strcmp(s.name, ".text") == 0 {
+			n = 6;
+		} else if strcmp(s.name, ".rodata") == 0 {
+			n = 2;
+		} else {
+			n = 0;
+		}
+		as_emit(c, n);
+		as_emit(c, n >> 8);
+		as_emit(c, n >> 16);
+		as_emit(c, n >> 24);
+		as_emit(c, n >> 32);
+		as_emit(c, n >> 40);
+		as_emit(c, n >> 48);
+		as_emit(c, n >> 56);
 
 		// Addr
 		if s.start == s.end {
@@ -709,7 +806,6 @@ func writeout(c: *assembler, start: *label, kstart: *label) {
 	}
 
 	load_addr = 0x100000;
-	text_size = c.at;
 
 	if (!start || !start.fixed) {
 		if !kstart {
@@ -719,9 +815,6 @@ func writeout(c: *assembler, start: *label, kstart: *label) {
 		entry = load_addr + start.at;
 	}
 
-	text_size = text_size;
-	text_end = load_addr + text_size;
-
 	mb_magic = 0x1badb002;
 	mb_flags = 0x00010003;
 	mb_checksum = -(mb_magic + mb_flags);
@@ -736,6 +829,9 @@ func writeout(c: *assembler, start: *label, kstart: *label) {
 
 	shoff = emit_sections(c as *assembler);
 
+	text_size = c.at;
+	text_end = load_addr + text_size;
+
 	// magic
 	putchar(c, 0x7f);
 	putchar(c, 'E');
diff --git a/cc0.c b/cc0.c
@@ -4,6 +4,7 @@
 #ifndef my_syscall
 #define my_syscall syscall
 #endif
+struct my_ablob;
 struct my_alloc;
 struct my_assembler;
 struct my_chunk;
@@ -31,6 +32,13 @@ struct my_section;
 struct my_sigaction;
 struct my_symbol;
 struct my_type;
+struct my_ablob {
+	struct my_ablob* my_left;
+	struct my_ablob* my_right;
+	struct my_label* my_label;
+	unsigned char* my_s;
+	unsigned long my_slen;
+};
 struct my_alloc {
 	struct my_page* my_page;
 };
@@ -50,6 +58,7 @@ struct my_assembler {
 	unsigned long my_lineno;
 	unsigned char* my_prevfilename;
 	unsigned long my_prevlineno;
+	struct my_ablob* my_blobs;
 };
 struct my_chunk {
 	struct my_chunk* my_next;
@@ -664,6 +673,7 @@ void( my_addfixup)(struct my_assembler* my_c,struct my_label* my_l);
 unsigned char*( my_alloc)(struct my_alloc* my_c,unsigned long my_size);
 unsigned long( my_any)(struct my_peg* my_c);
 void( my_args_to_ir)(struct my_irfunc* my_ic,struct my_node* my_n);
+struct my_label*( my_as_blob)(struct my_assembler* my_c,unsigned char* my_s,unsigned long my_slen);
 void( my_as_emit)(struct my_assembler* my_a,unsigned long my_b);
 void( my_as_jmp)(struct my_assembler* my_a,unsigned long my_op,struct my_label* my_l);
 void( my_as_modm)(struct my_assembler* my_a,unsigned long my_op,unsigned long my_b,unsigned long my_i,unsigned long my_s,unsigned long my_d);
@@ -712,6 +722,7 @@ void( my_defunion)(struct my_compiler* my_c,struct my_node* my_n);
 void( my_die)(unsigned char* my_msg);
 void( my_emit)(struct my_assembler* my_c,unsigned long my_x);
 void( my_emit_align)(struct my_assembler* my_c,unsigned long my_n,unsigned long my_b);
+void( my_emit_blobs)(struct my_assembler* my_c,struct my_ablob* my_b);
 void( my_emit_builtin)(struct my_compiler* my_c);
 void( my_emit_hook)(struct my_assembler* my_c);
 void( my_emit_isr)(struct my_compiler* my_c);
@@ -771,6 +782,7 @@ void( my_main)(unsigned long my_argc,unsigned char** my_argv,unsigned char** my_
 void( my_mark_expr_used)(struct my_compiler* my_c,struct my_decl* my_d,struct my_node* my_n);
 void( my_mark_func_used)(struct my_compiler* my_c,struct my_decl* my_d);
 void( my_mark_stmt_used)(struct my_compiler* my_c,struct my_decl* my_d,struct my_node* my_n);
+unsigned long( my_memcmp)(unsigned char* my_a,unsigned char* my_b,unsigned long my_n);
 void( my_memcpy)(unsigned char* my_dest,unsigned char* my_src,unsigned long my_size);
 void( my_memset)(unsigned char* my_dest,unsigned long my_c,unsigned long my_size);
 struct my_irblock*( my_mkirblock)(struct my_irfunc* my_ic);
@@ -1432,6 +1444,45 @@ void( my_args_to_ir)(struct my_irfunc* my_ic,struct my_node* my_n){
 	(my_n)=((my_n)->my_b);
 	}
 }
+struct my_label*( my_as_blob)(struct my_assembler* my_c,unsigned char* my_s,unsigned long my_slen){
+	struct my_ablob* my_b = 0;
+	struct my_ablob** my_link = 0;
+	unsigned long my_dir = 0;
+	(my_link)=(&((my_c)->my_blobs));
+	while (1) {
+	(my_b)=(*(my_link));
+	if ((unsigned long)(!(my_b))) {
+	break;
+	}
+	if ((unsigned long)(((long)((my_b)->my_slen))<((long)(my_slen)))) {
+	(my_dir)=((my_memcmp)(((my_b)->my_s),(my_s),((my_b)->my_slen)));
+	if ((unsigned long)(((long)(my_dir))==((long)(0UL)))) {
+	(my_dir)=((unsigned long)(-(unsigned long)(1UL)));
+	}
+	} else if ((unsigned long)(((long)((my_b)->my_slen))>((long)(my_slen)))) {
+	(my_dir)=((my_memcmp)(((my_b)->my_s),(my_s),(my_slen)));
+	if ((unsigned long)(((long)(my_dir))==((long)(0UL)))) {
+	(my_dir)=(1UL);
+	}
+	} else {
+	(my_dir)=((my_memcmp)(((my_b)->my_s),(my_s),(my_slen)));
+	}
+	if ((unsigned long)(((long)(my_dir))==((long)(0UL)))) {
+	return (my_b)->my_label;
+	} else if ((unsigned long)(((long)(my_dir))>((long)(0UL)))) {
+	(my_link)=(&((my_b)->my_left));
+	} else {
+	(my_link)=(&((my_b)->my_right));
+	}
+	}
+	(my_b)=((struct my_ablob*)(my_alloc)(((my_c)->my_a),(40UL)));
+	((my_b)->my_label)=((my_mklabel)((my_c)));
+	((my_b)->my_s)=((my_alloc)(((my_c)->my_a),(my_slen)));
+	((my_b)->my_slen)=(my_slen);
+	(my_memcpy)(((my_b)->my_s),(my_s),(my_slen));
+	(*(my_link))=(my_b);
+	return (my_b)->my_label;
+}
 void( my_as_emit)(struct my_assembler* my_a,unsigned long my_b){
 	(my_emit)((my_a),(my_b));
 }
@@ -2713,6 +2764,23 @@ void( my_emit_align)(struct my_assembler* my_c,unsigned long my_n,unsigned long 
 	(my_pad)=((unsigned long)(((unsigned long)(my_pad))+((unsigned long)(1UL))));
 	}
 }
+void( my_emit_blobs)(struct my_assembler* my_c,struct my_ablob* my_b){
+	unsigned long my_i = 0;
+	if ((unsigned long)(!(my_b))) {
+	return;
+	}
+	(my_emit_blobs)((my_c),((my_b)->my_left));
+	(my_fixup_label)((my_c),((my_b)->my_label));
+	(my_i)=(0UL);
+	while (1) {
+	if ((unsigned long)(((long)(my_i))==((long)((my_b)->my_slen)))) {
+	break;
+	}
+	(my_as_emit)((my_c),((unsigned long)((my_b)->my_s)[my_i]));
+	(my_i)=((unsigned long)(((unsigned long)(my_i))+((unsigned long)(1UL))));
+	}
+	(my_emit_blobs)((my_c),((my_b)->my_right));
+}
 void( my_emit_builtin)(struct my_compiler* my_c){
 	struct my_decl* my_d = 0;
 	(my_d)=((my_find)((my_c),((unsigned char *)"syscall"),((void *)0),(1UL)));
@@ -3378,6 +3446,10 @@ unsigned long( my_emit_sections)(struct my_assembler* my_c){
 	(my_s)=((my_find_section)((my_c),((unsigned char *)"")));
 	((my_s)->my_start)=(0UL);
 	((my_s)->my_end)=(0UL);
+	(my_emit_align)((my_c),(4096UL),(my_OP_NOP));
+	(my_add_section)((my_c),((unsigned char *)".rodata"),(my_SHT_PROGBITS));
+	(my_emit_blobs)((my_c),((my_c)->my_blobs));
+	(my_emit_align)((my_c),(4096UL),(0UL));
 	(my_add_section)((my_c),((unsigned char *)".strtab"),(my_SHT_STRTAB));
 	(my_y)=((my_c)->my_symbols);
 	while (1) {
@@ -3471,14 +3543,21 @@ unsigned long( my_emit_sections)(struct my_assembler* my_c){
 	(my_as_emit)((my_c),((unsigned long)(((unsigned long)(my_n))>>((unsigned long)(8UL)))));
 	(my_as_emit)((my_c),((unsigned long)(((unsigned long)(my_n))>>((unsigned long)(16UL)))));
 	(my_as_emit)((my_c),((unsigned long)(((unsigned long)(my_n))>>((unsigned long)(24UL)))));
-	(my_as_emit)((my_c),(6UL));
-	(my_as_emit)((my_c),(0UL));
-	(my_as_emit)((my_c),(0UL));
-	(my_as_emit)((my_c),(0UL));
-	(my_as_emit)((my_c),(0UL));
-	(my_as_emit)((my_c),(0UL));
-	(my_as_emit)((my_c),(0UL));
-	(my_as_emit)((my_c),(0UL));
+	if ((unsigned long)(((long)((my_strcmp)(((my_s)->my_name),((unsigned char *)".text"))))==((long)(0UL)))) {
+	(my_n)=(6UL);
+	} else if ((unsigned long)(((long)((my_strcmp)(((my_s)->my_name),((unsigned char *)".rodata"))))==((long)(0UL)))) {
+	(my_n)=(2UL);
+	} else {
+	(my_n)=(0UL);
+	}
+	(my_as_emit)((my_c),(my_n));
+	(my_as_emit)((my_c),((unsigned long)(((unsigned long)(my_n))>>((unsigned long)(8UL)))));
+	(my_as_emit)((my_c),((unsigned long)(((unsigned long)(my_n))>>((unsigned long)(16UL)))));
+	(my_as_emit)((my_c),((unsigned long)(((unsigned long)(my_n))>>((unsigned long)(24UL)))));
+	(my_as_emit)((my_c),((unsigned long)(((unsigned long)(my_n))>>((unsigned long)(32UL)))));
+	(my_as_emit)((my_c),((unsigned long)(((unsigned long)(my_n))>>((unsigned long)(40UL)))));
+	(my_as_emit)((my_c),((unsigned long)(((unsigned long)(my_n))>>((unsigned long)(48UL)))));
+	(my_as_emit)((my_c),((unsigned long)(((unsigned long)(my_n))>>((unsigned long)(56UL)))));
 	if ((unsigned long)(((long)((my_s)->my_start))==((long)((my_s)->my_end)))) {
 	(my_n)=(0UL);
 	} else {
@@ -5074,6 +5153,22 @@ void( my_mark_stmt_used)(struct my_compiler* my_c,struct my_decl* my_d,struct my
 	return;
 	}
 }
+unsigned long( my_memcmp)(unsigned char* my_a,unsigned char* my_b,unsigned long my_n){
+	unsigned long my_i = 0;
+	(my_i)=(0UL);
+	while (1) {
+	if ((unsigned long)(((long)(my_i))==((long)(my_n)))) {
+	return 0UL;
+	}
+	if ((unsigned long)(((long)((my_a)[my_i]))>((long)((my_b)[my_i])))) {
+	return 1UL;
+	}
+	if ((unsigned long)(((long)((my_a)[my_i]))<((long)((my_b)[my_i])))) {
+	return (unsigned long)(-(unsigned long)(1UL));
+	}
+	(my_i)=((unsigned long)(((unsigned long)(my_i))+((unsigned long)(1UL))));
+	}
+}
 void( my_memcpy)(unsigned char* my_dest,unsigned char* my_src,unsigned long my_size){
 	unsigned long my_i = 0;
 	if ((unsigned long)(((long)(my_size))<((long)(0UL)))) {
@@ -5854,31 +5949,7 @@ void( my_output_irstmt)(struct my_irfunc* my_ic,struct my_irblock* my_b,struct m
 }
 void( my_output_irstr)(struct my_irfunc* my_ic,struct my_irblock* my_b,struct my_irop* my_o){
 	struct my_label* my_s = 0;
-	struct my_label* my_out = 0;
-	unsigned long my_i = 0;
-	(my_s)=((my_mklabel)(((my_ic)->my_s)));
-	(my_out)=((my_mklabel)(((my_ic)->my_s)));
-	(my_as_jmp)(((my_ic)->my_s),(my_OP_JMP),(my_out));
-	(my_fixup_label)(((my_ic)->my_s),(my_s));
-	(my_i)=(0UL);
-	while (1) {
-	if ((unsigned long)(((long)(my_i))==((long)((my_o)->my_slen)))) {
-	break;
-	}
-	(my_as_emit)(((my_ic)->my_s),((unsigned long)((my_o)->my_s)[my_i]));
-	(my_i)=((unsigned long)(((unsigned long)(my_i))+((unsigned long)(1UL))));
-	}
-	(my_as_emit)(((my_ic)->my_s),(0UL));
-	(my_as_op)(((my_ic)->my_s),(my_OP_NOP));
-	(my_as_op)(((my_ic)->my_s),(my_OP_NOP));
-	(my_as_op)(((my_ic)->my_s),(my_OP_NOP));
-	(my_as_op)(((my_ic)->my_s),(my_OP_NOP));
-	(my_as_op)(((my_ic)->my_s),(my_OP_NOP));
-	(my_as_op)(((my_ic)->my_s),(my_OP_NOP));
-	(my_as_op)(((my_ic)->my_s),(my_OP_NOP));
-	(my_as_op)(((my_ic)->my_s),(my_OP_NOP));
-	(my_fixup_label)(((my_ic)->my_s),(my_out));
-	(my_reserve)(((my_ic)->my_s),(16UL));
+	(my_s)=((my_as_blob)(((my_ic)->my_s),((my_o)->my_s),((unsigned long)(((unsigned long)((my_o)->my_slen))+((unsigned long)(1UL))))));
 	(my_as_modrm)(((my_ic)->my_s),(my_OP_LEA),(my_R_RAX),(my_R_RIP),(0UL),(0UL),(128UL));
 	(my_addfixup)(((my_ic)->my_s),(my_s));
 }
@@ -9974,7 +10045,7 @@ void( my_setup_alloc)(struct my_alloc* my_c){
 }
 struct my_assembler*( my_setup_assembler)(struct my_alloc* my_a){
 	struct my_assembler* my_c = 0;
-	(my_c)=((struct my_assembler*)(my_alloc)((my_a),(120UL)));
+	(my_c)=((struct my_assembler*)(my_alloc)((my_a),(128UL)));
 	((my_c)->my_a)=(my_a);
 	((my_c)->my_out)=((void *)0);
 	((my_c)->my_at)=(160UL);
@@ -11043,7 +11114,6 @@ void( my_writeout)(struct my_assembler* my_c,struct my_label* my_start,struct my
 	(my_die)(((unsigned char *)"output not opened"));
 	}
 	(my_load_addr)=(1048576UL);
-	(my_text_size)=((my_c)->my_at);
 	if ((unsigned long)(((unsigned long)(!(my_start)))||((unsigned long)(!((my_start)->my_fixed))))) {
 	if ((unsigned long)(!(my_kstart))) {
 	(my_die)(((unsigned char *)"_start is not defined"));
@@ -11051,8 +11121,6 @@ void( my_writeout)(struct my_assembler* my_c,struct my_label* my_start,struct my
 	} else {
 	(my_entry)=((unsigned long)(((unsigned long)(my_load_addr))+((unsigned long)((my_start)->my_at))));
 	}
-	(my_text_size)=(my_text_size);
-	(my_text_end)=((unsigned long)(((unsigned long)(my_load_addr))+((unsigned long)(my_text_size))));
 	(my_mb_magic)=(464367618UL);
 	(my_mb_flags)=(65539UL);
 	(my_mb_checksum)=((unsigned long)(-(unsigned long)((unsigned long)(((unsigned long)(my_mb_magic))+((unsigned long)(my_mb_flags))))));
@@ -11064,6 +11132,8 @@ void( my_writeout)(struct my_assembler* my_c,struct my_label* my_start,struct my
 	(my_kentry)=(0UL);
 	}
 	(my_shoff)=((my_emit_sections)(((struct my_assembler*)my_c)));
+	(my_text_size)=((my_c)->my_at);
+	(my_text_end)=((unsigned long)(((unsigned long)(my_load_addr))+((unsigned long)(my_text_size))));
 	(my_putchar)((my_c),(127UL));
 	(my_putchar)((my_c),(69));
 	(my_putchar)((my_c),(76));
diff --git a/ir.om b/ir.om
@@ -1889,40 +1889,8 @@ func output_irretval(ic: *irfunc, b: *irblock, o: *irop) {
 
 func output_irstr(ic: *irfunc, b: *irblock, o: *irop) {
 	var s: *label;
-	var out: *label;
-	var i: int;
-
-	s = mklabel(ic.s);
-	out = mklabel(ic.s);
-
-	as_jmp(ic.s, OP_JMP, out);
-
-	fixup_label(ic.s, s);
-
-	i = 0;
-	loop {
-		if i == o.slen {
-			break;
-		}
-
-		as_emit(ic.s, o.s[i] as int);
-
-		i = i + 1;
-	}
-	as_emit(ic.s, 0);
-
-	as_op(ic.s, OP_NOP);
-	as_op(ic.s, OP_NOP);
-	as_op(ic.s, OP_NOP);
-	as_op(ic.s, OP_NOP);
-	as_op(ic.s, OP_NOP);
-	as_op(ic.s, OP_NOP);
-	as_op(ic.s, OP_NOP);
-	as_op(ic.s, OP_NOP);
 
-	fixup_label(ic.s, out);
-
-	reserve(ic.s, 16);
+	s = as_blob(ic.s, o.s, o.slen + 1);
 	as_modrm(ic.s, OP_LEA, R_RAX, R_RIP, 0, 0, 128);
 	addfixup(ic.s, s);
 }