[PATCH 0/2] fix is_pure_ops_struct()

[PATCH 0/2] fix is_pure_ops_struct()

[Joonwon Kang]

Hello,

This series fixes unreachable code bug and removes dead code in
is_pure_ops_struct().

Thanks.

Joonwon Kang (2):
  randstruct: fix a bug in is_pure_ops_struct()
  randstruct: remove dead code in is_pure_ops_struct()

 scripts/gcc-plugins/randomize_layout_plugin.c | 11 ++++-------
 1 file changed, 4 insertions(+), 7 deletions(-)

-- 
2.17.1

[PATCH 1/2] randstruct: fix a bug in is_pure_ops_struct()

[Joonwon Kang]

Before this, there were false negatives in the case where a struct
contains other structs which contain only function pointers because
of unreachable code in is_pure_ops_struct().

Signed-off-by: Joonwon Kang <kjw1627@gmail.com>
---
 scripts/gcc-plugins/randomize_layout_plugin.c | 10 +++++-----
 1 file changed, 5 insertions(+), 5 deletions(-)

diff --git a/scripts/gcc-plugins/randomize_layout_plugin.c b/scripts/gcc-plugins/randomize_layout_plugin.c
index 6d5bbd31db7f..bd29e4e7a524 100644
--- a/scripts/gcc-plugins/randomize_layout_plugin.c
+++ b/scripts/gcc-plugins/randomize_layout_plugin.c
@@ -443,13 +443,13 @@ static int is_pure_ops_struct(const_tree node)
 		if (node == fieldtype)
 			continue;
 
-		if (!is_fptr(fieldtype))
-			return 0;
-
-		if (code != RECORD_TYPE && code != UNION_TYPE)
+		if (code == RECORD_TYPE || code == UNION_TYPE) {
+			if (!is_pure_ops_struct(fieldtype))
+				return 0;
 			continue;
+		}
 
-		if (!is_pure_ops_struct(fieldtype))
+		if (!is_fptr(fieldtype))
 			return 0;
 	}
 
-- 
2.17.1

[PATCH 2/2] randstruct: remove dead code in is_pure_ops_struct()

[Joonwon Kang]

Recursive declaration for struct which has member of the same struct
type, for example,

struct foo {
    struct foo f;
    ...
};

is not allowed. So, it is unnecessary to check if a struct has this
kind of member.

Signed-off-by: Joonwon Kang <kjw1627@gmail.com>
---
 scripts/gcc-plugins/randomize_layout_plugin.c | 3 ---
 1 file changed, 3 deletions(-)

diff --git a/scripts/gcc-plugins/randomize_layout_plugin.c b/scripts/gcc-plugins/randomize_layout_plugin.c
index bd29e4e7a524..e14efe23e645 100644
--- a/scripts/gcc-plugins/randomize_layout_plugin.c
+++ b/scripts/gcc-plugins/randomize_layout_plugin.c
@@ -440,9 +440,6 @@ static int is_pure_ops_struct(const_tree node)
 		const_tree fieldtype = get_field_type(field);
 		enum tree_code code = TREE_CODE(fieldtype);
 
-		if (node == fieldtype)
-			continue;
-
 		if (code == RECORD_TYPE || code == UNION_TYPE) {
 			if (!is_pure_ops_struct(fieldtype))
 				return 0;
-- 
2.17.1

Re: [PATCH 1/2] randstruct: fix a bug in is_pure_ops_struct()

[Kees Cook]

On Thu, Aug 01, 2019 at 03:01:10AM +0900, Joonwon Kang wrote:
> Before this, there were false negatives in the case where a struct
> contains other structs which contain only function pointers because
> of unreachable code in is_pure_ops_struct().
> 
> Signed-off-by: Joonwon Kang <kjw1627@gmail.com>

I've applied this (with some commit log tweaks) and it should be visible
in linux-next soon. I'll send this on to Linus before -rc3.

-- 
Kees Cook

Re: [PATCH 2/2] randstruct: remove dead code in is_pure_ops_struct()

[Kees Cook]

On Thu, Aug 01, 2019 at 03:01:49AM +0900, Joonwon Kang wrote:
> Recursive declaration for struct which has member of the same struct
> type, for example,
> 
> struct foo {
>     struct foo f;
>     ...
> };
> 
> is not allowed. So, it is unnecessary to check if a struct has this
> kind of member.

Is that the only case where this loop could happen? Seems also safe to
just leave it as-is...

-Kees

> 
> Signed-off-by: Joonwon Kang <kjw1627@gmail.com>
> ---
>  scripts/gcc-plugins/randomize_layout_plugin.c | 3 ---
>  1 file changed, 3 deletions(-)
> 
> diff --git a/scripts/gcc-plugins/randomize_layout_plugin.c b/scripts/gcc-plugins/randomize_layout_plugin.c
> index bd29e4e7a524..e14efe23e645 100644
> --- a/scripts/gcc-plugins/randomize_layout_plugin.c
> +++ b/scripts/gcc-plugins/randomize_layout_plugin.c
> @@ -440,9 +440,6 @@ static int is_pure_ops_struct(const_tree node)
>  		const_tree fieldtype = get_field_type(field);
>  		enum tree_code code = TREE_CODE(fieldtype);
>  
> -		if (node == fieldtype)
> -			continue;
> -
>  		if (code == RECORD_TYPE || code == UNION_TYPE) {
>  			if (!is_pure_ops_struct(fieldtype))
>  				return 0;
> -- 
> 2.17.1
> 

-- 
Kees Cook

Re: [PATCH 2/2] randstruct: remove dead code in is_pure_ops_struct()

[Joonwon Kang]

On Wed, Jul 31, 2019 at 12:59:30PM -0700, Kees Cook wrote:
> On Thu, Aug 01, 2019 at 03:01:49AM +0900, Joonwon Kang wrote:
> > Recursive declaration for struct which has member of the same struct
> > type, for example,
> > 
> > struct foo {
> >     struct foo f;
> >     ...
> > };
> > 
> > is not allowed. So, it is unnecessary to check if a struct has this
> > kind of member.
> 
> Is that the only case where this loop could happen? Seems also safe to
> just leave it as-is...
> 
> -Kees

I think it is pretty obvious that it is the only case. I compiled kernel
with allyesconfig and the condition never hit even once. However, it will
also be no problem to just leave it as-is as you mentioned.

> 
> > 
> > Signed-off-by: Joonwon Kang <kjw1627@gmail.com>
> > ---
> >  scripts/gcc-plugins/randomize_layout_plugin.c | 3 ---
> >  1 file changed, 3 deletions(-)
> > 
> > diff --git a/scripts/gcc-plugins/randomize_layout_plugin.c b/scripts/gcc-plugins/randomize_layout_plugin.c
> > index bd29e4e7a524..e14efe23e645 100644
> > --- a/scripts/gcc-plugins/randomize_layout_plugin.c
> > +++ b/scripts/gcc-plugins/randomize_layout_plugin.c
> > @@ -440,9 +440,6 @@ static int is_pure_ops_struct(const_tree node)
> >  		const_tree fieldtype = get_field_type(field);
> >  		enum tree_code code = TREE_CODE(fieldtype);
> >  
> > -		if (node == fieldtype)
> > -			continue;
> > -
> >  		if (code == RECORD_TYPE || code == UNION_TYPE) {
> >  			if (!is_pure_ops_struct(fieldtype))
> >  				return 0;
> > -- 
> > 2.17.1
> > 
> 
> -- 
> Kees Cook