linux-kernel - Re: [PATCH 1/3] lib/test_printf: Clean up test of hashed pointers

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, 2 Mar 2020 11:24:27 +0100
From:   Petr Mladek <pmladek@...e.com>
To:     Uwe Kleine-König <uwe@...ine-koenig.org>
Cc:     Andy Shevchenko <andriy.shevchenko@...ux.intel.com>,
        Sergey Senozhatsky <sergey.senozhatsky@...il.com>,
        Steven Rostedt <rostedt@...dmis.org>,
        Rasmus Villemoes <linux@...musvillemoes.dk>,
        Sergey Senozhatsky <sergey.senozhatsky.work@...il.com>,
        linux-kernel@...r.kernel.org, Ilya Dryomov <idryomov@...il.com>,
        Kees Cook <keescook@...omium.org>,
        "Tobin C . Harding" <me@...in.cc>
Subject: Re: [PATCH 1/3] lib/test_printf: Clean up test of hashed pointers

On Thu 2020-02-27 15:30:51, Uwe Kleine-König wrote:
> Hello Petr,
> 
> On 2/27/20 2:01 PM, Petr Mladek wrote:
> > The commit ad67b74d2469d9b82a ("printk: hash addresses printed with %p")
> > helps to prevent leaking kernel addresses.
> > 
> > The testing of this functionality is a bit problematic because the output
> > depends on a random key that is generated during boot. Though, it is
> > still possible to check some aspects:
> > 
> >   + output string length
> >   + hash differs from the original pointer value
> >   + top half bits are zeroed on 64-bit systems
> 
> Is "hash differs from the original pointer value" a valid check?
> Depending on the random value and the actual pointer I can imagine a
> valid match. Such a match is unlikely but not necessarily bogus, is it?

Yes, there is a small risk or false negative.

It might be possible to try if the problem persist with PTR+1 value or
so. But I am not sure if it is worth it.

The problem is only on 32-bit systems. The chance is really small.
I have added a comment above the check. It can be found via the added
error message.

Note that this check has been there even before in plain_hash().
But it was worse because it was without any comment or error message.

Best Regards,
Petr