[PATCH] netfilter: ebtables: replace zero-length array members

[PATCH] netfilter: ebtables: replace zero-length array members

[GONG, Ruiqi]

From: "GONG, Ruiqi" <gongruiqi1@huawei.com>

As suggested by Kees[1], replace the old-style 0-element array members
of multiple structs in ebtables.h with modern C99 flexible array.

[1]: https://lore.kernel.org/all/5E8E0F9C-EE3F-4B0D-B827-DC47397E2A4A@kernel.org/

Link: https://github.com/KSPP/linux/issues/21
Signed-off-by: GONG, Ruiqi <gongruiqi1@huawei.com>
---
 include/uapi/linux/netfilter_bridge/ebtables.h | 10 +++++-----
 1 file changed, 5 insertions(+), 5 deletions(-)

diff --git a/include/uapi/linux/netfilter_bridge/ebtables.h b/include/uapi/linux/netfilter_bridge/ebtables.h
index b0caad82b693..673b00df162c 100644
--- a/include/uapi/linux/netfilter_bridge/ebtables.h
+++ b/include/uapi/linux/netfilter_bridge/ebtables.h
@@ -87,7 +87,7 @@ struct ebt_entries {
 	/* nr. of entries */
 	unsigned int nentries;
 	/* entry list */
-	char data[0] __attribute__ ((aligned (__alignof__(struct ebt_replace))));
+	char data[] __attribute__ ((aligned (__alignof__(struct ebt_replace))));
 };
 
 /* used for the bitmask of struct ebt_entry */
@@ -129,7 +129,7 @@ struct ebt_entry_match {
 	} u;
 	/* size of data */
 	unsigned int match_size;
-	unsigned char data[0] __attribute__ ((aligned (__alignof__(struct ebt_replace))));
+	unsigned char data[] __attribute__ ((aligned (__alignof__(struct ebt_replace))));
 };
 
 struct ebt_entry_watcher {
@@ -142,7 +142,7 @@ struct ebt_entry_watcher {
 	} u;
 	/* size of data */
 	unsigned int watcher_size;
-	unsigned char data[0] __attribute__ ((aligned (__alignof__(struct ebt_replace))));
+	unsigned char data[] __attribute__ ((aligned (__alignof__(struct ebt_replace))));
 };
 
 struct ebt_entry_target {
@@ -155,7 +155,7 @@ struct ebt_entry_target {
 	} u;
 	/* size of data */
 	unsigned int target_size;
-	unsigned char data[0] __attribute__ ((aligned (__alignof__(struct ebt_replace))));
+	unsigned char data[] __attribute__ ((aligned (__alignof__(struct ebt_replace))));
 };
 
 #define EBT_STANDARD_TARGET "standard"
@@ -190,7 +190,7 @@ struct ebt_entry {
 		/* sizeof ebt_entry + matches + watchers + target */
 		unsigned int next_offset;
 	);
-	unsigned char elems[0] __attribute__ ((aligned (__alignof__(struct ebt_replace))));
+	unsigned char elems[] __attribute__ ((aligned (__alignof__(struct ebt_replace))));
 };
 
 static __inline__ struct ebt_entry_target *
-- 
2.41.0

Re: [PATCH] netfilter: ebtables: replace zero-length array members

[Phil Sutter]

On Wed, Aug 09, 2023 at 03:51:36PM +0800, GONG, Ruiqi wrote:
> From: "GONG, Ruiqi" <gongruiqi1@huawei.com>
> 
> As suggested by Kees[1], replace the old-style 0-element array members
> of multiple structs in ebtables.h with modern C99 flexible array.
> 
> [1]: https://lore.kernel.org/all/5E8E0F9C-EE3F-4B0D-B827-DC47397E2A4A@kernel.org/
> 
> Link: https://github.com/KSPP/linux/issues/21
> Signed-off-by: GONG, Ruiqi <gongruiqi1@huawei.com>

I tried this once[1], but it was rejected pointing at a revert[2]. It
seems gcc was improved since then: The warning is gone and I don't find
-Wno-stringop-overflow flag in iptables sources.

Cheers, Phil

[1] https://lore.kernel.org/all/20200719100220.4666-1-phil@nwl.cc/
[2] https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=1e6e9d0f4859ec698d55381ea26f4136eff3afe1

Re: [PATCH] netfilter: ebtables: replace zero-length array members

[Kees Cook]

On Wed, Aug 09, 2023 at 03:51:36PM +0800, GONG, Ruiqi wrote:
> From: "GONG, Ruiqi" <gongruiqi1@huawei.com>
> 
> As suggested by Kees[1], replace the old-style 0-element array members
> of multiple structs in ebtables.h with modern C99 flexible array.
> 
> [1]: https://lore.kernel.org/all/5E8E0F9C-EE3F-4B0D-B827-DC47397E2A4A@kernel.org/
> 
> Link: https://github.com/KSPP/linux/issues/21
> Signed-off-by: GONG, Ruiqi <gongruiqi1@huawei.com>

In theory, this should be fine. It is possible there are userspace tools
that are doing (already) buggy things that will now turn into build
failures. If the userspace ebtable tools still build happily with these
UAPI changes, I imagine that would be a sufficient test.

Reviewed-by: Kees Cook <keescook@chromium.org>

-- 
Kees Cook