linux-kernel - Re: [PATCH] samples/ftrace: mark my

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: <CABCJKufdFmoEEmx1-P_0d6vE6TOoDwJ43mq8eCQNQZ7rb0PxPQ@mail.gmail.com>
Date:   Mon, 16 Nov 2020 13:10:57 -0800
From:   Sami Tolvanen <samitolvanen@...gle.com>
To:     Steven Rostedt <rostedt@...dmis.org>
Cc:     Kees Cook <keescook@...omium.org>,
        Josh Poimboeuf <jpoimboe@...hat.com>,
        Colin Ian King <colin.king@...onical.com>,
        LKML <linux-kernel@...r.kernel.org>
Subject: Re: [PATCH] samples/ftrace: mark my_tramp[12]? global

On Mon, Nov 16, 2020 at 12:38 PM Steven Rostedt <rostedt@...dmis.org> wrote:
>
> On Mon, 16 Nov 2020 12:10:10 -0800
> Sami Tolvanen <samitolvanen@...gle.com> wrote:
>
> > On Mon, Nov 16, 2020 at 8:39 AM Steven Rostedt <rostedt@...dmis.org> wrote:
> > >
> > > On Fri, 13 Nov 2020 10:34:14 -0800
> > > Sami Tolvanen <samitolvanen@...gle.com> wrote:
> > >
> > > > my_tramp[12]? are declared as global functions in C, but they are not
> > > > marked global in the inline assembly definition. This mismatch confuses
> > > > Clang's Control-Flow Integrity checking. Fix the definitions by adding
> > > > .globl.
> > > >
> > >
> > > Actually, since that function is not really global, would it work if you
> > > removed the "extern" from the my_tramp declaration?
> >
> > Unfortunately not, removing the "extern" doesn't seem to change anything.
> >
> > > In other words, is there a way to tell C that a function is declared in an
> > > inline assembly block?
> >
> > I'm not sure if there's a way to tell C that a static function is
> > declared in inline assembly. At least I couldn't find a way that would
> > make the compiler happy.
>
> I'm trying to see the warning. What option makes clang trigger a warning on
> this?

Clang doesn't warn about this as we're building a module, it just
generates a reference to a non-existing global "my_tramp" symbol,
because the one defined in inline assembly has a local binding:

$ readelf --symbols --wide ftrace-direct.lto.o | grep my_tramp
    16: 0000000000000000    13 FUNC    LOCAL  DEFAULT    1 my_tramp
    33: 0000000000000000     0 NOTYPE  GLOBAL DEFAULT  UND my_tramp
    42: 0000000000000000     8 FUNC    GLOBAL HIDDEN     8 my_tramp.cfi_jt

This would prevent the module from loading, which modpost catches:

ERROR: modpost: "my_tramp" [samples/ftrace/ftrace-direct.ko] undefined!

> From user space, I'm just using the following file:

As this error happens only with Control-Flow Integrity, we need to
take the address of the "test" function to force the compiler to
generate a jump table entry for it. Here's a slightly tweaked
stand-alone reproducer:

https://godbolt.org/z/GnzjE4

Sami