linux-kernel - Re: [PATCH] tracing: Use linker magic instead of recasting ftrace_ops_list

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <CAG48ez2pOns4vF9M_4ubMJ+p9YFY29udMaH0wm8UuCwGQ4ZZAQ@mail.gmail.com>
Date:   Wed, 17 Jun 2020 23:30:07 +0200
From:   Jann Horn <jannh@...gle.com>
To:     Steven Rostedt <rostedt@...dmis.org>
Cc:     LKML <linux-kernel@...r.kernel.org>,
        Ingo Molnar <mingo@...nel.org>,
        Kees Cook <keescook@...omium.org>,
        Kernel Hardening <kernel-hardening@...ts.openwall.com>,
        Oscar Carter <oscar.carter@....com>,
        Andrew Morton <akpm@...ux-foundation.org>
Subject: Re: [PATCH] tracing: Use linker magic instead of recasting ftrace_ops_list_func()

On Wed, Jun 17, 2020 at 10:56 PM Steven Rostedt <rostedt@...dmis.org> wrote:
> In an effort to enable -Wcast-function-type in the top-level Makefile to
> support Control Flow Integrity builds, all function casts need to be
> removed.
>
> This means that ftrace_ops_list_func() can no longer be defined as
> ftrace_ops_no_ops(). The reason for ftrace_ops_no_ops() is to use that when
> an architecture calls ftrace_ops_list_func() with only two parameters
> (called from assembly). And to make sure there's no C side-effects, those
> archs call ftrace_ops_no_ops() which only has two parameters, as
> ftrace_ops_list_func() has four parameters.
>
> Instead of a typecast, use vmlinux.lds.h to define ftrace_ops_list_func() to
> arch_ftrace_ops_list_func() that will define the proper set of parameters.
>
> Link: https://lore.kernel.org/r/20200614070154.6039-1-oscar.carter@gmx.com
[...]
> diff --git a/include/asm-generic/vmlinux.lds.h b/include/asm-generic/vmlinux.lds.h
[...]
> +                       ftrace_ops_list_func = arch_ftrace_ops_list_func;
>  #else
>  # ifdef CONFIG_FUNCTION_TRACER
>  #  define MCOUNT_REC() ftrace_stub_graph = ftrace_stub;
> diff --git a/kernel/trace/ftrace.c b/kernel/trace/ftrace.c
[...]
> +/* Defined by vmlinux.lds.h see the commment above arch_ftrace_ops_list_func for details */
> +void ftrace_ops_list_func(unsigned long ip, unsigned long parent_ip,
> +                         struct ftrace_ops *op, struct pt_regs *regs);
[...]
> +void arch_ftrace_ops_list_func(unsigned long ip, unsigned long parent_ip)
>  {

Well, it's not like the function cast itself is the part that's
problematic for CFI; the problematic part is when you actually make a
C function call (in particular an indirect one) where the destination
is compiled with a prototype that is different from the prototype used
at the call site. Doing this linker hackery isn't really any better
than shutting up the compiler warning by piling on enough casts or
whatever. (There should be some combination of casts that'll shut up
this warning, right?)

IIUC the real issue here is that ftrace_func_t is defined as a fixed
type, but actually has different types depending on the architecture?
If so, it might be cleaner to define ftrace_func_t differently
depending on architecture, or something like that?

And if that's not feasible, I think it would be better to at least
replace this linker trickery with straightforward
shut-up-the-compiler-casts - it'd be much easier to understand what's
actually going on that way.