Re: [PATCH RFC kspp-next 0/3] compiler_types: add Endianness-dependent __counted_by_{le,be}

From: Alexander Lobakin <aleksander.lobakin@intel.com>
To: Kees Cook <keescook@chromium.org>,
	Przemek Kitszel <przemyslaw.kitszel@intel.com>
Cc: <linux-hardening@vger.kernel.org>,
	"Gustavo A. R. Silva" <gustavoars@kernel.org>,
	Marco Elver <elver@google.com>, Jakub Kicinski <kuba@kernel.org>,
	<netdev@vger.kernel.org>, <linux-kernel@vger.kernel.org>
Subject: Re: [PATCH RFC kspp-next 0/3] compiler_types: add Endianness-dependent __counted_by_{le,be}
Date: Tue, 19 Mar 2024 10:33:25 +0100	[thread overview]
Message-ID: <e7727b4e-c173-4edf-8d8e-a9dc00ba7653@intel.com> (raw)
In-Reply-To: <202403181038.EC9DF8CE3C@keescook>

From: Kees Cook <keescook@chromium.org>
Date: Mon, 18 Mar 2024 10:49:25 -0700

> On Mon, Mar 18, 2024 at 02:03:51PM +0100, Alexander Lobakin wrote:
>>  include/linux/compiler_types.h              | 11 ++++++++++
>>  drivers/net/ethernet/intel/idpf/virtchnl2.h | 24 ++++++++++-----------
>>  2 files changed, 23 insertions(+), 12 deletions(-)
> 
> Oh, I see the Subject says "kspp-next" -- normally I'd expect things
> touch net to go through netdev. I'm fine with this going through either
> tree. Perhaps better through netdev since that subsystem has the most
> users and may gain more using the new macros?

Yeah sure. I send it with "kspp-next", so that it would be clear it's a
security feature :>

Thanks for the ack. Re expressions -- Przemek suggested it would be nice
to have something like

	__le32 counter;
	struct a flex[] __counted_by(le32_to_cpu(counter));

but we don't know whether something like this is possible to implement
in the compiler.

> 
> -Kees

Thanks,
Olek