linux-kernel - Re: [PATCH] objtool,x86: Verify poke_int3

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]

Date:   Mon, 27 Jul 2020 14:50:19 +0200
From:   peterz@...radead.org
To:     Ingo Molnar <mingo@...nel.org>
Cc:     Josh Poimboeuf <jpoimboe@...hat.com>,
        Thomas Gleixner <tglx@...utronix.de>, x86@...nel.org,
        linux-kernel@...r.kernel.org, mbenes@...e.cz
Subject: Re: [PATCH] objtool,x86: Verify poke_int3_handler() is self contained

On Mon, Jul 27, 2020 at 01:21:44PM +0200, Ingo Molnar wrote:
> 
> * peterz@...radead.org <peterz@...radead.org> wrote:
> 
> > 
> > Abuse the SMAP rules to ensure poke_int3_handler() doesn't call out to
> > anything.
> > 
> > Signed-off-by: Peter Zijlstra (Intel) <peterz@...radead.org>
> > ---
> >  tools/objtool/check.c |    8 ++++++++
> >  1 file changed, 8 insertions(+)
> > 
> > --- a/tools/objtool/check.c
> > +++ b/tools/objtool/check.c
> > @@ -551,6 +551,14 @@ static const char *uaccess_safe_builtin[
> >  	"__memcpy_mcsafe",
> >  	"mcsafe_handle_tail",
> >  	"ftrace_likely_update", /* CONFIG_TRACE_BRANCH_PROFILING */
> > +	/*
> > +	 * Abuse alert!
> > +	 *
> > +	 * poke_int3_handler() is not in fact related to uaccess at all, we
> > +	 * abuse the uaccess rules to ensure poke_int3_handler() is self
> > +	 * contained and doesn't CALL out to other code.
> > +	 */
> > +	"poke_int3_handler",
> 
> So ->uaccess_safe makes sure that we don't CALL into non-uaccess-safe 
> functions, but it still allows CALLs into *other* uaccess-safe 
> functions, right?
> 
> So unless I missed something in the logic, the comment should say 
> something like "doesn't CALL out to other non-uaccess safe code" or 
> so? Which is, arguably, like 99% of all functions - but still, a whole 
> bunch are allowed, such as low level instrumentation and other utility 
> functions.

Right, so poke_int3_handler() is also noinstr and by that isn't allowed
to call out into !noinstr code. The intersection should be small.

But yeah, perhaps this is a bad idea and I should add another annotation
for this,.. dunno.