[PATCH] dissect: simplify lookup_member()

[PATCH] dissect: simplify lookup_member()

[Oleg Nesterov]

Change examine_sym_node() to recursively inspect the members of SYM_STRUCT.
This allows us to simplify lookup_member(): no need to do examine_sym_node(),
the member was already examined.

This is also good because now ->r_symdef(member) is called on declaration,
not when dissect detects the first usage of ->member.

Signed-off-by: Oleg Nesterov <oleg@tv-sign.ru>

--- DS/dissect.c~	2006-04-20 23:49:03.000000000 +0400
+++ DS/dissect.c	2006-04-21 00:04:18.000000000 +0400
@@ -210,13 +210,16 @@ static void examine_sym_node(struct symb
 		case SYM_STRUCT: case SYM_UNION: //case SYM_ENUM:
 			if (base->evaluated)
 				return;
+			if (!base->symbol_list)
+				return;
+			base->evaluated = 1;
+
 			if (!base->ident && name)
 				base->ident = mk_name(root, name);
-			if (!base->ident || !base->symbol_list)
-				return;
-			if (reporter->r_symdef)
+			if (base->ident && reporter->r_symdef)
 				reporter->r_symdef(base);
-			base->evaluated = 1;
+			DO_LIST(base->symbol_list, mem,
+				examine_sym_node(mem, base->ident ?: root));
 		default:
 			return;
 		}
@@ -263,14 +266,8 @@ found:
 
 static struct symbol *lookup_member(struct symbol *type, struct ident *name, int *addr)
 {
-	struct symbol *node = __lookup_member(type, name, addr);
-
-	if (node != NULL)
-		examine_sym_node(node, type->ident);
-	else
-		node = no_member(name);
-
-	return node;
+	return __lookup_member(type, name, addr)
+		?: no_member(name);
 }
 
 static struct expression *peek_preop(struct expression *expr, int op)

Re: [PATCH] dissect: simplify lookup_member()

[Josh Triplett]

[-- Attachment #1: Type: text/plain, Size: 446 bytes --]

Oleg Nesterov wrote:
> Change examine_sym_node() to recursively inspect the members of SYM_STRUCT.
> This allows us to simplify lookup_member(): no need to do examine_sym_node(),
> the member was already examined.
> 
> This is also good because now ->r_symdef(member) is called on declaration,
> not when dissect detects the first usage of ->member.
> 
> Signed-off-by: Oleg Nesterov <oleg@tv-sign.ru>

Applied.

- Josh Triplett



[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 252 bytes --]

Re: [PATCH] dissect: simplify lookup_member()

[Pavel Roskin]

On Tue, 2007-02-27 at 10:40 -0800, Josh Triplett wrote:
> Oleg Nesterov wrote:
> > Change examine_sym_node() to recursively inspect the members of SYM_STRUCT.
> > This allows us to simplify lookup_member(): no need to do examine_sym_node(),
> > the member was already examined.
> > 
> > This is also good because now ->r_symdef(member) is called on declaration,
> > not when dissect detects the first usage of ->member.
> > 
> > Signed-off-by: Oleg Nesterov <oleg@tv-sign.ru>
> 
> Applied.

Josh, I'll really appreciate if you push your changes to the publicly
available repository as soon as you post "applied".  This way, when I
have some time for sparse, I read the list, see your post, update sparse
and see how it's affecting my projects.

If the changes are only available the next day, I'll probably be busy
with other things.

-- 
Regards,
Pavel Roskin

Git mirror delay

[Josh Triplett]

[-- Attachment #1: Type: text/plain, Size: 1473 bytes --]

Pavel Roskin wrote:
> On Tue, 2007-02-27 at 10:40 -0800, Josh Triplett wrote:
>> Oleg Nesterov wrote:
>>> Change examine_sym_node() to recursively inspect the members of SYM_STRUCT.
>>> This allows us to simplify lookup_member(): no need to do examine_sym_node(),
>>> the member was already examined.
>>>
>>> This is also good because now ->r_symdef(member) is called on declaration,
>>> not when dissect detects the first usage of ->member.
>>>
>>> Signed-off-by: Oleg Nesterov <oleg@tv-sign.ru>
>> Applied.
> 
> Josh, I'll really appreciate if you push your changes to the publicly
> available repository as soon as you post "applied".  This way, when I
> have some time for sparse, I read the list, see your post, update sparse
> and see how it's affecting my projects.
> 
> If the changes are only available the next day, I'll probably be busy
> with other things.

To the best of my knowledge, I always push out changes before I send the
corresponding "applied" mail.  However, I often send the mail a few moments
after pushing out the change, and the kernel.org mirroring processes do not
run the instant I push changes out.  In particular, I just confirmed that a
fresh clone of git://git.kernel.org/pub/scm/linux/kernel/git/josh/sparse.git
did not include the changes I've pushed about 8 hours ago today.

The new git.kernel.org server may help alleviate some of the issues that lead
to this high mirroring delay.

- Josh Triplett


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 252 bytes --]