Add missing checks for Waddress-space
PATCH 1/1 Add missing checks for Waddress-space
Some of these are missing in evaluate.c. Maybe we should change the option to not consider address space at all, if unset. It would be much easier to make attribute_address_space() ignore it.
--- evaluate.c | 4 ++-- 1 files changed, 2 insertions(+), 2 deletions(-)
evaluate.c +2 -2
@@ -1149,7 +1149,7 @@ static struct symbol *evaluate_conditional_expression(struct expression *expr) goto Err; }/* OK, it's pointer on pointer */ -if (ltype->ctype.as != rtype->ctype.as) { +if (Waddress_space && ltype->ctype.as != rtype->ctype.as) { typediff ="different address spaces" ; goto Err; }
@@ -1339,7 +1339,7 @@ static int compatible_assignment_types(struct expression *expr, struct symbol *t* we do not remove qualifiers from pointed to [C] * or mix address spaces [sparse]. */ -if (t->ctype.as != s->ctype.as) { +if (Waddress_space && t->ctype.as != s->ctype.as) { typediff ="different address spaces" ; goto Err; }
Signature · 1 line
1.6.0.6
Christopher Li
maintainerQuoted 4 lines · Martin Nagy
Some of these are missing in
evaluate.c. Maybe we should change the option to not consider address space at all, if unset. It would be much easier to makeattribute_address_space()ignore it.
I think disable it at attribute_address_space() might be better. Otherwise is_same_type() will still return false for address space difference. There are too many places to track down the address space comparison.
Chris
Quoted 9 lines · Christopher Li
Quoted 4 lines · Martin Nagy
Some of these are missing in
evaluate.c. Maybe we should change the option to not consider address space at all, if unset. It would be much easier to makeattribute_address_space()ignore it.I think disable it at
attribute_address_space()might be better. Otherwiseis_same_type()will still return false for address space difference. There are too many places to track down the address space comparison.
OK, new patch attached.
Martin
From f05ae87e33092ff9fab9f518887c9dfcb9e32a70 Mon Sep 17 00:00:00 2001
From: Martin Nagy <[email protected]>
Date: Sat, 25 Apr 2009 06:44:49 +0200
Subject: [PATCH] Ignore address space if !Waddress_space
Remove all previous checks for Waddress_space and add one centralized to the address_space attribute handler. If user passes the -Wno-address-space option, we behave as if every pointer had no address space.
--- evaluate.c | 8 ++++---- parse.c | 2 +- 2 files changed, 5 insertions(+), 5 deletions(-)
evaluate.c +4 -4
@@ -689,7 +689,7 @@ const char *type_difference(struct ctype *c1, struct ctype *c2,/* XXX: we ought to compare sizes */ break ;case SYM_PTR : -if (Waddress_space && as1 != as2) +if (as1 != as2)return "different address spaces" ;/* MOD_SPECIFIER is due to idiocy in parse.c */ if ((mod1 ^ mod2) & ~MOD_IGNORE & ~MOD_SPECIFIER )
@@ -706,7 +706,7 @@ const char *type_difference(struct ctype *c1, struct ctype *c2,struct symbol *arg1, *arg2;int i; -if (Waddress_space && as1 != as2) +if (as1 != as2)return "different address spaces" ;if ((mod1 ^ mod2) & ~MOD_IGNORE & ~MOD_SIGNEDNESS )return "different modifiers" ;
@@ -745,7 +745,7 @@ const char *type_difference(struct ctype *c1, struct ctype *c2,break ; }case SYM_BASETYPE : -if (Waddress_space && as1 != as2) +if (as1 != as2)return "different address spaces" ;if (base1 != base2)return "different base types" ;
@@ -762,7 +762,7 @@ const char *type_difference(struct ctype *c1, struct ctype *c2, t1 = base1; t2 = base2; } -if (Waddress_space && as1 != as2) +if (as1 != as2)return "different address spaces" ;if ((mod1 ^ mod2) & ~MOD_IGNORE & ~MOD_SIGNEDNESS )return "different modifiers" ;
parse.c +1 -1
@@ -987,7 +987,7 @@ static struct token *attribute_address_space(struct token *token, struct symbol token =conditional_expression (token, &expr);if (expr) { as =const_expression_value (expr); -if (as) +if (Waddress_space && as) ctx->ctype.as = as; } token =expect (token,')' ,"after address_space attribute" );
Signature · 1 line
1.6.0.6
Christopher Li
maintainerQuoted 1 line · Martin Nagy
OK, new patch attached.
Applied and thanks.
Chris