Re: [PATCH v2 2/5] modules: Refactor + kdoc elf_validity_cached_copy

From: Greg KH <greg@kroah.com>
To: Matthew Maurer <mmaurer@google.com>
Cc: Masahiro Yamada <masahiroy@kernel.org>,
	Nick Desaulniers <ndesaulniers@google.com>,
	Miguel Ojeda <ojeda@kernel.org>, Gary Guo <gary@garyguo.net>,
	Luis Chamberlain <mcgrof@kernel.org>,
	Nathan Chancellor <nathan@kernel.org>,
	Nicolas Schier <nicolas@fjasle.eu>,
	linuxppc-dev@lists.ozlabs.org, linux-kernel@vger.kernel.org,
	linux-modules@vger.kernel.org, linux-kbuild@vger.kernel.org,
	rust-for-linux@vger.kernel.org, Laura Abbott <laura@labbott.name>
Subject: Re: [PATCH v2 2/5] modules: Refactor + kdoc elf_validity_cached_copy
Date: Sat, 18 Nov 2023 06:36:49 -0500	[thread overview]
Message-ID: <2023111818-agent-verdict-99a5@gregkh> (raw)
In-Reply-To: <20231118025748.2778044-3-mmaurer@google.com>

On Sat, Nov 18, 2023 at 02:54:43AM +0000, Matthew Maurer wrote:
> Functionality is almost identical, just structured for better
> documentation and readability. Changes:
> 
> * Section names are checked for *all* non-SHT_NULL sections, not just
>   SHF_ALLOC sections. We have code that accesses section names of
>   non-SHF_ALLOC sections (see find_any_sec for example)
> * The section name check occurs *before* strcmping on section names.
>   Previously, it was possible to use an out-of-bounds offset to strcmp
>   against ".modinfo" or ".gnu.linkonce.this_module"
> * strtab is checked for NUL lead+termination and nonzero size
> * The symbol table is swept to ensure offsets are inbounds of strtab
> 
> While some of these oversights would normally be worrying, all of the
> potentially unverified accesses occur after signature check, and only in
> response to a user who can load a kernel module.
> 
> Signed-off-by: Matthew Maurer <mmaurer@google.com>
> ---
>  kernel/module/internal.h |   7 +-
>  kernel/module/main.c     | 585 +++++++++++++++++++++++++++++----------
>  2 files changed, 444 insertions(+), 148 deletions(-)

Again, this needs to be broken into much smaller pieces before we can
even review it.  Would you want to review this?

thanks,

greg "think of the reviewers" k-h