From 6898d26e8083900e5c61717de5dfb0edc409ebd9 Mon Sep 17 00:00:00 2001
From: Dmitry Stogov <dmitry@zend.com>
Date: Wed, 6 Dec 2023 16:30:44 +0300
Subject: [PATCH] Allow code generation for functions with unesolved symbols

---
 ir_aarch64.dasc | 17 +++++++++++++++--
 ir_emit.c       |  1 +
 ir_main.c       | 19 ++++++++++++++-----
 ir_x86.dasc     | 16 ++++++++++++++--
 4 files changed, 44 insertions(+), 9 deletions(-)

diff --git a/ir_aarch64.dasc b/ir_aarch64.dasc
index 86b9cde..873dc27 100644
--- a/ir_aarch64.dasc
+++ b/ir_aarch64.dasc
@@ -1172,8 +1172,16 @@ static void ir_emit_load(ir_ctx *ctx, ir_type type, ir_reg reg, ir_ref src)
 		if (IR_IS_TYPE_INT(type)) {
 			ir_insn *insn = &ctx->ir_base[src];
 
-			IR_ASSERT(insn->op != IR_STR && insn->op != IR_SYM && insn->op != IR_FUNC);
-			ir_emit_load_imm_int(ctx, type, reg, insn->val.i64);
+			if (insn->op == IR_SYM || insn->op == IR_FUNC) {
+				void *addr = (ctx->loader && ctx->loader->resolve_sym_name) ?
+					ctx->loader->resolve_sym_name(ctx->loader, ir_get_str(ctx, insn->val.name)) :
+					ir_resolve_sym_name(ir_get_str(ctx, insn->val.name));
+				IR_ASSERT(addr);
+				ir_emit_load_imm_int(ctx, type, reg, (intptr_t)addr);
+			} else {
+				IR_ASSERT(insn->op != IR_STR);
+				ir_emit_load_imm_int(ctx, type, reg, insn->val.i64);
+			}
 		} else {
 			ir_emit_load_imm_fp(ctx, type, reg, src);
 		}
@@ -4306,6 +4314,7 @@ static int32_t ir_emit_arguments(ir_ctx *ctx, ir_ref def, ir_insn *insn, ir_reg
 									void *addr = (ctx->loader && ctx->loader->resolve_sym_name) ?
 										ctx->loader->resolve_sym_name(ctx->loader, ir_get_str(ctx, val_insn->val.name)) :
 										ir_resolve_sym_name(ir_get_str(ctx, val_insn->val.name));
+									IR_ASSERT(addr);
 									ir_emit_load_imm_int(ctx, IR_ADDR, dst_reg, (intptr_t)addr);
 									continue;
 								}
@@ -4335,6 +4344,7 @@ static int32_t ir_emit_arguments(ir_ctx *ctx, ir_ref def, ir_insn *insn, ir_reg
 							void *addr = (ctx->loader && ctx->loader->resolve_sym_name) ?
 								ctx->loader->resolve_sym_name(ctx->loader, ir_get_str(ctx, val_insn->val.name)) :
 								ir_resolve_sym_name(ir_get_str(ctx, val_insn->val.name));
+							IR_ASSERT(addr);
 							ir_emit_load_imm_int(ctx, IR_ADDR, tmp_reg, (intptr_t)addr);
 							|	str Rx(tmp_reg), [sp, #stack_offset]
 						} else {
@@ -4387,6 +4397,7 @@ static void ir_emit_call_ex(ir_ctx *ctx, ir_ref def, ir_insn *insn, int32_t used
 			addr = (ctx->loader && ctx->loader->resolve_sym_name) ?
 				ctx->loader->resolve_sym_name(ctx->loader, ir_get_str(ctx, addr_insn->val.name)) :
 				ir_resolve_sym_name(ir_get_str(ctx, addr_insn->val.name));
+			IR_ASSERT(addr);
 		} else {
 			IR_ASSERT(addr_insn->op == IR_ADDR || addr_insn->op == IR_FUNC_ADDR);
 			addr = (void*)addr_insn->val.addr;
@@ -4472,6 +4483,7 @@ static void ir_emit_tailcall(ir_ctx *ctx, ir_ref def, ir_insn *insn)
 			addr = (ctx->loader && ctx->loader->resolve_sym_name) ?
 				ctx->loader->resolve_sym_name(ctx->loader, ir_get_str(ctx, addr_insn->val.name)) :
 				ir_resolve_sym_name(ir_get_str(ctx, addr_insn->val.name));
+			IR_ASSERT(addr);
 		} else {
 			IR_ASSERT(addr_insn->op == IR_ADDR || addr_insn->op == IR_FUNC_ADDR);
 			addr = (void*)addr_insn->val.addr;
@@ -4855,6 +4867,7 @@ static void ir_emit_exitcall(ir_ctx *ctx, ir_ref def, ir_insn *insn)
 			addr = (ctx->loader && ctx->loader->resolve_sym_name) ?
 				ctx->loader->resolve_sym_name(ctx->loader, ir_get_str(ctx, addr_insn->val.name)) :
 				ir_resolve_sym_name(ir_get_str(ctx, addr_insn->val.name));
+			IR_ASSERT(addr);
 		} else {
 			IR_ASSERT(addr_insn->op == IR_ADDR || addr_insn->op == IR_FUNC_ADDR);
 			addr = (void*)addr_insn->val.addr;
diff --git a/ir_emit.c b/ir_emit.c
index 40ddf10..520d875 100644
--- a/ir_emit.c
+++ b/ir_emit.c
@@ -337,6 +337,7 @@ static void *ir_jmp_addr(ir_ctx *ctx, ir_insn *insn, ir_insn *addr_insn)
 		addr = (ctx->loader && ctx->loader->resolve_sym_name) ?
 			ctx->loader->resolve_sym_name(ctx->loader, ir_get_str(ctx, addr_insn->val.name)) :
 			ir_resolve_sym_name(ir_get_str(ctx, addr_insn->val.name));
+		IR_ASSERT(addr);
 	} else {
 		IR_ASSERT(addr_insn->op == IR_ADDR || addr_insn->op == IR_FUNC_ADDR);
 		addr = (void*)addr_insn->val.addr;
diff --git a/ir_main.c b/ir_main.c
index 7186930..31b9a6f 100644
--- a/ir_main.c
+++ b/ir_main.c
@@ -231,6 +231,7 @@ int ir_compile_func(ir_ctx *ctx, int opt_level, uint32_t dump, FILE *dump_file,
 
 typedef struct _ir_sym {
 	void *addr;
+	void *thunk_addr;
 } ir_sym;
 
 typedef struct _ir_main_loader {
@@ -265,6 +266,9 @@ static bool ir_loader_add_sym(ir_loader *loader, const char *name, void *addr)
 		if (addr && !l->sym[old_val].addr) {
 			/* Update forward declaration */
 			l->sym[old_val].addr = addr;
+			if (l->sym[old_val].thunk_addr) {
+				// TODO: Fix thunk or relocation ???
+			}
 			return 1;
 		}
 		return 0;
@@ -274,6 +278,7 @@ static bool ir_loader_add_sym(ir_loader *loader, const char *name, void *addr)
 		l->sym = ir_mem_realloc(l->sym, sizeof(ir_sym) * l->sym_count);
 	}
 	l->sym[val].addr = addr;
+	l->sym[val].thunk_addr = NULL;
 	return 1;
 }
 
@@ -293,7 +298,15 @@ static void* ir_loader_resolve_sym_name(ir_loader *loader, const char *name)
 	void *addr;
 
 	if (val) {
-		return l->sym[val].addr;
+		if (l->sym[val].addr) {
+			return l->sym[val].addr;
+		}
+		if (!l->sym[val].thunk_addr) {
+			/* Undefined declaration */
+			// TODO: Add thunk or relocation ???
+			l->sym[val].thunk_addr = (void*)(intptr_t)sizeof(void*);
+		}
+		return l->sym[val].thunk_addr;
 	}
 	addr = ir_resolve_sym_name(name);
 	ir_loader_add_sym(loader, name, addr); /* cache */
@@ -312,11 +325,9 @@ static bool ir_loader_external_sym_dcl(ir_loader *loader, const char *name, uint
 		fprintf(l->dump_file, "extern %s %s;\n", (flags & IR_CONST) ? "const" : "var", name);
 	}
 	if (l->c_file) {
-		// TODO:
 		ir_emit_c_sym_decl(name, flags | IR_EXTERN, 0, l->c_file);
 	}
 	if (l->llvm_file) {
-		// TODO:
 		ir_emit_llvm_sym_decl(name, flags | IR_EXTERN, 0, l->llvm_file);
 	}
 	if (l->dump_asm || l->dump_size || l->run) {
@@ -430,11 +441,9 @@ static bool ir_loader_sym_dcl(ir_loader *loader, const char *name, uint32_t flag
 		fprintf(l->dump_file, "%s %s [%ld]%s\n", (flags & IR_CONST) ? "const" : "var", name, size, has_data ? " = {" : ";");
 	}
 	if (l->c_file) {
-		// TODO:
 		ir_emit_c_sym_decl(name, flags, has_data, l->c_file);
 	}
 	if (l->llvm_file) {
-		// TODO:
 		ir_emit_llvm_sym_decl(name, flags, has_data, l->llvm_file);
 	}
 	if (l->dump_asm || l->dump_size || l->run) {
diff --git a/ir_x86.dasc b/ir_x86.dasc
index c9639f2..e10ea47 100644
--- a/ir_x86.dasc
+++ b/ir_x86.dasc
@@ -2145,8 +2145,16 @@ static void ir_emit_load(ir_ctx *ctx, ir_type type, ir_reg reg, ir_ref src)
 		if (IR_IS_TYPE_INT(type)) {
 			ir_insn *insn = &ctx->ir_base[src];
 
-			IR_ASSERT(insn->op != IR_STR && insn->op != IR_SYM && insn->op != IR_FUNC);
-			ir_emit_load_imm_int(ctx, type, reg, insn->val.i64);
+			if (insn->op == IR_SYM || insn->op == IR_FUNC) {
+				void *addr = (ctx->loader && ctx->loader->resolve_sym_name) ?
+					ctx->loader->resolve_sym_name(ctx->loader, ir_get_str(ctx, insn->val.name)) :
+					ir_resolve_sym_name(ir_get_str(ctx, insn->val.name));
+				IR_ASSERT(addr);
+				ir_emit_load_imm_int(ctx, type, reg, (intptr_t)addr);
+			} else {
+				IR_ASSERT(insn->op != IR_STR);
+				ir_emit_load_imm_int(ctx, type, reg, insn->val.i64);
+			}
 		} else {
 			ir_emit_load_imm_fp(ctx, type, reg, src);
 		}
@@ -7154,6 +7162,7 @@ static int32_t ir_emit_arguments(ir_ctx *ctx, ir_ref def, ir_insn *insn, ir_reg
 									void *addr = (ctx->loader && ctx->loader->resolve_sym_name) ?
 										ctx->loader->resolve_sym_name(ctx->loader, ir_get_str(ctx, val_insn->val.name)) :
 										ir_resolve_sym_name(ir_get_str(ctx, val_insn->val.name));
+									IR_ASSERT(addr);
 									if (sizeof(void*) == 4 || IR_IS_SIGNED_32BIT(addr)) {
 										|	mov Ra(dst_reg), ((ptrdiff_t)addr)
 									} else {
@@ -7232,6 +7241,7 @@ static int32_t ir_emit_arguments(ir_ctx *ctx, ir_ref def, ir_insn *insn, ir_reg
 							void *addr = (ctx->loader && ctx->loader->resolve_sym_name) ?
 								ctx->loader->resolve_sym_name(ctx->loader, ir_get_str(ctx, val_insn->val.name)) :
 								ir_resolve_sym_name(ir_get_str(ctx, val_insn->val.name));
+							IR_ASSERT(addr);
 							if (sizeof(void*) == 4) {
 								|	mov aword [Ra(IR_REG_RSP)+stack_offset], ((ptrdiff_t)addr)
 |.if X64
@@ -7351,6 +7361,7 @@ static void ir_emit_call_ex(ir_ctx *ctx, ir_ref def, ir_insn *insn, int32_t used
 			addr = (ctx->loader && ctx->loader->resolve_sym_name) ?
 				ctx->loader->resolve_sym_name(ctx->loader, ir_get_str(ctx, addr_insn->val.name)) :
 				ir_resolve_sym_name(ir_get_str(ctx, addr_insn->val.name));
+			IR_ASSERT(addr);
 		} else {
 			IR_ASSERT(addr_insn->op == IR_ADDR || addr_insn->op == IR_FUNC_ADDR);
 			addr = (void*)addr_insn->val.addr;
@@ -7505,6 +7516,7 @@ static void ir_emit_tailcall(ir_ctx *ctx, ir_ref def, ir_insn *insn)
 			addr = (ctx->loader && ctx->loader->resolve_sym_name) ?
 				ctx->loader->resolve_sym_name(ctx->loader, ir_get_str(ctx, addr_insn->val.name)) :
 				ir_resolve_sym_name(ir_get_str(ctx, addr_insn->val.name));
+			IR_ASSERT(addr);
 		} else {
 			IR_ASSERT(addr_insn->op == IR_ADDR || addr_insn->op == IR_FUNC_ADDR);
 			addr = (void*)addr_insn->val.addr;