Accept more asm expressions

In particular subtracting a defined symbol from current section
makes the value PC relative, and .org accepts symbolic expressions
as well, if the symbol is from the current section.

i386-asm.c

@@ -356,8 +356,7 @@ static void parse_operand(TCCState *s1, Operand *op)
         next();
         asm_expr(s1, &e);
         op->type = OP_IM32;
-        op->e.v = e.v;
+        op->e = e;
-        op->e.sym = e.sym;
         if (!op->e.sym) {
             if (op->e.v == (uint8_t)op->e.v)
                 op->type |= OP_IM8;
@@ -378,8 +377,7 @@ static void parse_operand(TCCState *s1, Operand *op)
         op->shift = 0;
         if (tok != '(') {
             asm_expr(s1, &e);
-            op->e.v = e.v;
+            op->e = e;
-            op->e.sym = e.sym;
         } else {
             next();
             if (tok == '%') {
@@ -395,6 +393,7 @@ static void parse_operand(TCCState *s1, Operand *op)
                 op->e.v = e.v;
                 op->e.sym = e.sym;
             }
+	    op->e.pcrel = 0;
         }
         if (tok == '(') {
 	    int type = 0;
@@ -425,7 +424,12 @@ static void parse_operand(TCCState *s1, Operand *op)
 /* XXX: unify with C code output ? */
 ST_FUNC void gen_expr32(ExprValue *pe)
 {
-    gen_addr32(pe->sym ? VT_SYM : 0, pe->sym, pe->v);
+    if (pe->pcrel)
+        /* If PC-relative, always set VT_SYM, even without symbol,
+	   so as to force a relocation to be emitted.  */
+	gen_addrpc32(VT_SYM, pe->sym, pe->v);
+    else
+	gen_addr32(pe->sym ? VT_SYM : 0, pe->sym, pe->v);
 }
 
 #ifdef TCC_TARGET_X86_64

tcc.h

@@ -555,6 +555,7 @@ typedef struct CachedInclude {
 typedef struct ExprValue {
     uint64_t v;
     Sym *sym;
+    int pcrel;
 } ExprValue;
 
 #define MAX_ASM_OPERANDS 30

tccasm.c

@@ -67,11 +67,13 @@ static void asm_expr_unary(TCCState *s1, ExprValue *pe)
                     sym->type.t = VT_STATIC | VT_VOID;
                 }
             }
-            pe->v = 0;
+	    pe->v = 0;
-            pe->sym = sym;
+	    pe->sym = sym;
+	    pe->pcrel = 0;
         } else if (*p == '\0') {
             pe->v = n;
             pe->sym = NULL;
+	    pe->pcrel = 0;
         } else {
             tcc_error("invalid number syntax");
         }
@@ -97,6 +99,7 @@ static void asm_expr_unary(TCCState *s1, ExprValue *pe)
     case TOK_LCHAR:
 	pe->v = tokc.i;
 	pe->sym = NULL;
+	pe->pcrel = 0;
 	next();
 	break;
     case '(':
@@ -107,6 +110,7 @@ static void asm_expr_unary(TCCState *s1, ExprValue *pe)
     case '.':
         pe->v = 0;
         pe->sym = &sym_dot;
+	pe->pcrel = 0;
         sym_dot.type.t = VT_VOID | VT_STATIC;
         sym_dot.r = cur_text_section->sh_num;
         sym_dot.jnext = ind;
@@ -125,9 +129,11 @@ static void asm_expr_unary(TCCState *s1, ExprValue *pe)
                 /* if absolute symbol, no need to put a symbol value */
                 pe->v = sym->jnext;
                 pe->sym = NULL;
+		pe->pcrel = 0;
             } else {
                 pe->v = 0;
                 pe->sym = sym;
+		pe->pcrel = 0;
             }
             next();
         } else {
@@ -230,20 +236,21 @@ static inline void asm_expr_sum(TCCState *s1, ExprValue *pe)
             pe->v -= e2.v;
             /* NOTE: we are less powerful than gas in that case
                because we store only one symbol in the expression */
-            if (!pe->sym && !e2.sym) {
+	    if (!e2.sym) {
-                /* OK */
+		/* OK */
-            } else if (pe->sym && !e2.sym) {
+	    } else if (pe->sym == e2.sym) { 
-                /* OK */
+		/* OK */
-            } else if (pe->sym && e2.sym) {
+		pe->sym = NULL; /* same symbols can be subtracted to NULL */
-                if (pe->sym == e2.sym) { 
+	    } else if (pe->sym && pe->sym->r == e2.sym->r && pe->sym->r != 0) {
-                    /* OK */
+		/* we also accept defined symbols in the same section */
-                } else if (pe->sym->r == e2.sym->r && pe->sym->r != 0) {
+		pe->v += pe->sym->jnext - e2.sym->jnext;
-                    /* we also accept defined symbols in the same section */
+		pe->sym = NULL;
-                    pe->v += pe->sym->jnext - e2.sym->jnext;
+	    } else if (e2.sym->r == cur_text_section->sh_num) {
-                } else {
+		/* When subtracting a defined symbol in current section
-                    goto cannot_relocate;
+		   this actually makes the value PC-relative.  */
-                }
+		pe->v -= e2.sym->jnext - ind - 4;
-                pe->sym = NULL; /* same symbols can be subtracted to NULL */
+		pe->pcrel = 1;
+		e2.sym = NULL;
             } else {
             cannot_relocate:
                 tcc_error("invalid operation with label");
@@ -531,9 +538,15 @@ static void asm_parse_directive(TCCState *s1)
     case TOK_ASMDIR_org:
         {
             unsigned long n;
+	    ExprValue e;
             next();
-            /* XXX: handle section symbols too */
+	    asm_expr(s1, &e);
-            n = asm_int_expr(s1);
+	    n = e.v;
+	    if (e.sym) {
+		if (e.sym->r != cur_text_section->sh_num)
+		  expect("constant or same-section symbol");
+		n += e.sym->jnext;
+	    }
             if (n < ind)
                 tcc_error("attempt to .org backwards");
             v = 0;
@@ -703,6 +716,7 @@ static void asm_parse_directive(TCCState *s1)
     case TOK_ASMDIR_section:
         {
             char sname[256];
+	    int old_nb_section = s1->nb_sections;
 
 	    tok1 = tok;
             /* XXX: support more options */
@@ -733,6 +747,11 @@ static void asm_parse_directive(TCCState *s1)
 	        use_section(s1, sname);
 	    else
 	        push_section(s1, sname);
+	    /* If we just allocated a new section reset its alignment to
+	       1.  new_section normally acts for GCC compatibility and
+	       sets alignment to PTR_SIZE.  The assembler behaves different. */
+	    if (old_nb_section != s1->nb_sections)
+	        cur_text_section->sh_addralign = 1;
         }
         break;
     case TOK_ASMDIR_previous:

tests/asmtest.S

@@ -727,6 +727,19 @@ nop
 .popsection
 .popsection
 
+1: ud2
+.pushsection __bug_table,"a"
+.align 8
+2: .long 1b - 2b
+   .long 0x600000 - 2b
+   .long 1b + 42
+   .long 43 + 1b
+   .long 2b + 144
+   .long 145 + 2b
+   .word 164, 0
+   .org 2b+32
+.popsection
+
     movd %esi, %mm1
     movd %edi, %xmm2
     movd (%ebx), %mm3