All of lore.kernel.org
 help / color / mirror / Atom feed
* [PATCH] Print an error if typeof() lacks an argument
@ 2009-04-25 11:03 Martin Nagy
  2009-04-25 11:11 ` Martin Nagy
  2009-04-27  6:38 ` Christopher Li
  0 siblings, 2 replies; 4+ messages in thread
From: Martin Nagy @ 2009-04-25 11:03 UTC (permalink / raw)
  To: linux-sparse


We weren't checking if the initializer isn't NULL, which caused sparse
to segfault later on when performing lazy evaluation in classify_type().

Signed-off-by: Martin Nagy <nagy.martin@gmail.com>
---
 parse.c |   17 +++++++++++------
 1 files changed, 11 insertions(+), 6 deletions(-)

diff --git a/parse.c b/parse.c
index 9662122..18cfaef 100644
--- a/parse.c
+++ b/parse.c
@@ -924,12 +924,17 @@ static struct token *typeof_specifier(struct token *token, struct decl_state *ct
 		ctx->ctype.base_type = sym->ctype.base_type;
 		apply_ctype(token->pos, &sym->ctype, &ctx->ctype);
 	} else {
-		struct symbol *typeof_sym = alloc_symbol(token->pos, SYM_TYPEOF);
-		token = parse_expression(token->next, &typeof_sym->initializer);

^ permalink raw reply related	[flat|nested] 4+ messages in thread

* Re: [PATCH] Print an error if typeof() lacks an argument
  2009-04-25 11:03 [PATCH] Print an error if typeof() lacks an argument Martin Nagy
@ 2009-04-25 11:11 ` Martin Nagy
  2009-04-27  6:38 ` Christopher Li
  1 sibling, 0 replies; 4+ messages in thread
From: Martin Nagy @ 2009-04-25 11:11 UTC (permalink / raw)
  To: linux-sparse

[-- Attachment #1: Type: text/plain, Size: 425 bytes --]

Martin Nagy wrote:
> 
> We weren't checking if the initializer isn't NULL, which caused sparse
> to segfault later on when performing lazy evaluation in classify_type().
> 
> Signed-off-by: Martin Nagy <nagy.martin@gmail.com>

I accidentally sent this from my work email address, which is
different than the sign-off address, sorry. I'm not sure if that's
an issue, but just to make sure, I'm sending the patch again.

Martin

[-- Attachment #2: 0001-Print-an-error-if-typeof-lacks-an-argument.patch --]
[-- Type: text/x-patch, Size: 941 bytes --]

From 962e4b1ad3b3cb13c7427d07dfa44cd15af11693 Mon Sep 17 00:00:00 2001
From: Martin Nagy <nagy.martin@gmail.com>
Date: Sat, 25 Apr 2009 12:56:33 +0200
Subject: [PATCH] Print an error if typeof() lacks an argument

We weren't checking if the initializer isn't NULL, which caused sparse
to segfault later on when performing lazy evaluation in classify_type().

Signed-off-by: Martin Nagy <nagy.martin@gmail.com>
---
 parse.c |   17 +++++++++++------
 1 files changed, 11 insertions(+), 6 deletions(-)

diff --git a/parse.c b/parse.c
index 9662122..18cfaef 100644
--- a/parse.c
+++ b/parse.c
@@ -924,12 +924,17 @@ static struct token *typeof_specifier(struct token *token, struct decl_state *ct
 		ctx->ctype.base_type = sym->ctype.base_type;
 		apply_ctype(token->pos, &sym->ctype, &ctx->ctype);
 	} else {
-		struct symbol *typeof_sym = alloc_symbol(token->pos, SYM_TYPEOF);
-		token = parse_expression(token->next, &typeof_sym->initializer);

^ permalink raw reply related	[flat|nested] 4+ messages in thread

* Re: [PATCH] Print an error if typeof() lacks an argument
  2009-04-25 11:03 [PATCH] Print an error if typeof() lacks an argument Martin Nagy
  2009-04-25 11:11 ` Martin Nagy
@ 2009-04-27  6:38 ` Christopher Li
  2009-04-27  9:15   ` Martin Nagy
  1 sibling, 1 reply; 4+ messages in thread
From: Christopher Li @ 2009-04-27  6:38 UTC (permalink / raw)
  To: Martin Nagy; +Cc: linux-sparse

On Sat, Apr 25, 2009 at 4:03 AM, Martin Nagy <mnagy@redhat.com> wrote:
>
> We weren't checking if the initializer isn't NULL, which caused sparse
> to segfault later on when performing lazy evaluation in classify_type().
>
> Signed-off-by: Martin Nagy <nagy.martin@gmail.com>
> ---
>  parse.c |   17 +++++++++++------
>  1 files changed, 11 insertions(+), 6 deletions(-)
>
> diff --git a/parse.c b/parse.c
> index 9662122..18cfaef 100644
> --- a/parse.c
> +++ b/parse.c
> @@ -924,12 +924,17 @@ static struct token *typeof_specifier(struct token *token, struct decl_state *ct
>                ctx->ctype.base_type = sym->ctype.base_type;
>                apply_ctype(token->pos, &sym->ctype, &ctx->ctype);
>        } else {
> -               struct symbol *typeof_sym = alloc_symbol(token->pos, SYM_TYPEOF);
> -               token = parse_expression(token->next, &typeof_sym->initializer);
> -
> -               typeof_sym->endpos = token->pos;
> -               ctx->ctype.base_type = typeof_sym;
> -       }
> +               struct expression *expr;

I think you want expr = NULL here. Otherwise if(expr) will pick up crap.

I would just add two lines after "token =
parse_expression(token->next, &typeof_sym->initializer);"

if (!type->initializer)
        sparse_error(token->pos, "expected expression after the '(' token");

If there is compile error, the sparse should not continue the later
stage any way.

BTW, can you add a validation test case which will trigger the bug?

Thanks

Chris
--
To unsubscribe from this list: send the line "unsubscribe linux-sparse" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

^ permalink raw reply	[flat|nested] 4+ messages in thread

* Re: [PATCH] Print an error if typeof() lacks an argument
  2009-04-27  6:38 ` Christopher Li
@ 2009-04-27  9:15   ` Martin Nagy
  0 siblings, 0 replies; 4+ messages in thread
From: Martin Nagy @ 2009-04-27  9:15 UTC (permalink / raw)
  To: Christopher Li; +Cc: linux-sparse

[-- Attachment #1: Type: text/plain, Size: 1532 bytes --]

Christopher Li wrote:
> On Sat, Apr 25, 2009 at 4:03 AM, Martin Nagy <mnagy@redhat.com> wrote:
> > +               struct expression *expr;
> 
> I think you want expr = NULL here. Otherwise if(expr) will pick up crap.

Right. I somehow assumed that parse_expression() will set it to NULL in
case there is not any expression. I attached a new patch which fixes
this.

> I would just add two lines after "token =
> parse_expression(token->next, &typeof_sym->initializer);"
> 
> if (!type->initializer)
>         sparse_error(token->pos, "expected expression after the '(' token");
> 
> If there is compile error, the sparse should not continue the later
> stage any way.

That won't work. sparse_error() will not exit, and sparse will still
segfault later, so we have to return from the function. You could
instead do something like this:

if (!typeof_sym->initializer) {
	sparse_error(token->pos, "expected ...");
	return expect(token, ')', "after typeof");
}

Or use a goto to jump to the return statement. In any case, I didn't
want to repeat the code and cause a memory leak. And I didn't want to
use a goto. I guess that it's ultimately a matter of style. If you
think the memory leak would be acceptable I can rework the patch again.
The leak would only occur in this specific case, so I guess maybe it
would be acceptable, but I wasn't sure so I rather went with this
approach.

> BTW, can you add a validation test case which will trigger the bug?

Yup, it's in the new patch.

Martin

[-- Attachment #2: 0001-Print-an-error-if-typeof-lacks-an-argument.patch --]
[-- Type: text/x-patch, Size: 2018 bytes --]

From bbd2e88cdd9d36d47ce50204d18547e08f2e2bea Mon Sep 17 00:00:00 2001
From: Martin Nagy <nagy.martin@gmail.com>
Date: Mon, 27 Apr 2009 10:48:50 +0200
Subject: [PATCH] Print an error if typeof() lacks an argument

We weren't checking if the initializer isn't NULL, which caused sparse
to segfault later on when performing lazy evaluation in classify_type().

Signed-off-by: Martin Nagy <nagy.martin@gmail.com>
---
 parse.c                 |   17 +++++++++++------
 validation/bad-typeof.c |   15 +++++++++++++++
 2 files changed, 26 insertions(+), 6 deletions(-)
 create mode 100644 validation/bad-typeof.c

diff --git a/parse.c b/parse.c
index 9662122..604e528 100644
--- a/parse.c
+++ b/parse.c
@@ -924,12 +924,17 @@ static struct token *typeof_specifier(struct token *token, struct decl_state *ct
 		ctx->ctype.base_type = sym->ctype.base_type;
 		apply_ctype(token->pos, &sym->ctype, &ctx->ctype);
 	} else {
-		struct symbol *typeof_sym = alloc_symbol(token->pos, SYM_TYPEOF);
-		token = parse_expression(token->next, &typeof_sym->initializer);
-
-		typeof_sym->endpos = token->pos;
-		ctx->ctype.base_type = typeof_sym;
-	}		
+		struct expression *expr = NULL;
+		token = parse_expression(token->next, &expr);
+		if (expr) {
+			struct symbol *typeof_sym = alloc_symbol(token->pos, SYM_TYPEOF);
+			typeof_sym->endpos = token->pos;
+			typeof_sym->initializer = expr;
+			ctx->ctype.base_type = typeof_sym;
+		} else {
+			sparse_error(token->pos, "expected expression after the '(' token");
+		}
+	}
 	return expect(token, ')', "after typeof");
 }
 
diff --git a/validation/bad-typeof.c b/validation/bad-typeof.c
new file mode 100644
index 0000000..5c27de4
--- /dev/null
+++ b/validation/bad-typeof.c
@@ -0,0 +1,15 @@
+static int fun(void)
+{
+	typeof() a;
+	int b;
+
+	a = b;
+}
+/*
+ * check-name: Bad typeof syntax segfault
+ *
+ * check-error-start
+bad-typeof.c:3:16: error: expected expression after the '(' token
+bad-typeof.c:6:9: error: identifier 'a' has no type
+ * check-error-end
+ */
-- 
1.6.0.6


^ permalink raw reply related	[flat|nested] 4+ messages in thread

end of thread, other threads:[~2009-04-27  9:15 UTC | newest]

Thread overview: 4+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2009-04-25 11:03 [PATCH] Print an error if typeof() lacks an argument Martin Nagy
2009-04-25 11:11 ` Martin Nagy
2009-04-27  6:38 ` Christopher Li
2009-04-27  9:15   ` Martin Nagy

This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.