linux-kernel - Re: [PATCH] vsprintf: Do not break early boot with probing addresses

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: <20190515075339.7biocrjfpxj77l3b@pathway.suse.cz>
Date:   Wed, 15 May 2019 09:53:39 +0200
From:   Petr Mladek <pmladek@...e.com>
To:     Geert Uytterhoeven <geert@...ux-m68k.org>
Cc:     Steven Rostedt <rostedt@...dmis.org>,
        David Laight <David.Laight@...lab.com>,
        Sergey Senozhatsky <sergey.senozhatsky.work@...il.com>,
        Andy Shevchenko <andriy.shevchenko@...ux.intel.com>,
        christophe leroy <christophe.leroy@....fr>,
        Linus Torvalds <torvalds@...ux-foundation.org>,
        Rasmus Villemoes <linux@...musvillemoes.dk>,
        "Tobin C . Harding" <me@...in.cc>, Michal Hocko <mhocko@...e.cz>,
        Sergey Senozhatsky <sergey.senozhatsky@...il.com>,
        "linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>,
        Michael Ellerman <mpe@...erman.id.au>,
        "linuxppc-dev@...ts.ozlabs.org" <linuxppc-dev@...ts.ozlabs.org>,
        Russell Currey <ruscur@...sell.cc>,
        Stephen Rothwell <sfr@...abs.org>,
        Heiko Carstens <heiko.carstens@...ibm.com>,
        "linux-arch@...r.kernel.org" <linux-arch@...r.kernel.org>,
        "linux-s390@...r.kernel.org" <linux-s390@...r.kernel.org>,
        Martin Schwidefsky <schwidefsky@...ibm.com>
Subject: Re: [PATCH] vsprintf: Do not break early boot with probing addresses

On Wed 2019-05-15 09:23:05, Geert Uytterhoeven wrote:
> Hi Steve,
> 
> On Tue, May 14, 2019 at 9:35 PM Steven Rostedt <rostedt@...dmis.org> wrote:
> > On Tue, 14 May 2019 21:13:06 +0200
> > Geert Uytterhoeven <geert@...ux-m68k.org> wrote:
> > > > > Do we care about the value? "(-E%u)"?
> > > >
> > > > That too could be confusing. What would (-E22) be considered by a user
> > > > doing an sprintf() on some string. I know that would confuse me, or I
> > > > would think that it was what the %pX displayed, and wonder why it
> > > > displayed it that way. Whereas "(fault)" is quite obvious for any %p
> > > > use case.
> > >
> > > I would immediately understand there's a missing IS_ERR() check in a
> > > function that can return  -EINVAL, without having to add a new printk()
> > > to find out what kind of bogus value has been received, and without
> > > having to reboot, and trying to reproduce...
> >
> > I have to ask. Has there actually been a case that you used a %pX and
> > it faulted, and you had to go back to find what the value of the
> > failure was?
> 
> If it faulted, the bad pointer value is obvious from the backtrace.
> If the code avoids the fault by verifying the pointer and returning
> "(efault)" instead, the bad pointer value is lost.
> 
> Or am I missing something?

Should buggy printk() crash the system?

Another problem is that vsprintf() is called in printk() under
lockbuf_lock. The messages are stored into printk_safe per CPU
buffers. It allows to see the nested messages. But there is still
a bigger risk of missing them than with a "normal" fault.

Finally, various variants of these checks were already used
in "random" printf formats. The only change is that we are
using them consistently everywhere[*] a pointer is accessed.

[*] Just the top level pointer is checked. Some pointer modifiers
are accessing ptr->ptr->val. The lower level pointers are not
checked to avoid too much complexity.

Best Regards,
Petr