Received-SPF: pass (google.com: best guess record for domain of linux-kernel-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) client-ip=209.132.180.67;
MIME-Version: 1.0
References: <20191115191728.87338-1-jannh@google.com> <20191115191728.87338-2-jannh@google.com>
 <CALCETrVQ2NqPnED_E6Y6EsCOEJJcz8GkQhgcKHk7JVAyykq06A@mail.gmail.com> <CAG48ez2z8i1nosA1nGrVdXx1cXXwHBqe7CC5kMB2W=uxbsvkjg@mail.gmail.com>
In-Reply-To: <CAG48ez2z8i1nosA1nGrVdXx1cXXwHBqe7CC5kMB2W=uxbsvkjg@mail.gmail.com>
From:   Andy Lutomirski <luto@kernel.org>
Date:   Wed, 27 Nov 2019 21:23:44 -0800
Message-ID: <CALCETrXU-hetnH7CTz-Z2xPDAkawx6GdxGtYo0=Jqq1YnoXrWg@mail.gmail.com>
Subject: Re: [PATCH v2 2/3] x86/traps: Print non-canonical address on #GP
To:     Jann Horn <jannh@google.com>
Cc:     Andy Lutomirski <luto@kernel.org>,
        Thomas Gleixner <tglx@linutronix.de>,
        Ingo Molnar <mingo@redhat.com>, Borislav Petkov <bp@alien8.de>,
        "H. Peter Anvin" <hpa@zytor.com>, X86 ML <x86@kernel.org>,
        Andrey Ryabinin <aryabinin@virtuozzo.com>,
        Alexander Potapenko <glider@google.com>,
        Dmitry Vyukov <dvyukov@google.com>,
        kasan-dev <kasan-dev@googlegroups.com>,
        LKML <linux-kernel@vger.kernel.org>,
        Andrey Konovalov <andreyknvl@google.com>,
        Sean Christopherson <sean.j.christopherson@intel.com>
Content-Type: text/plain; charset="UTF-8"
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

On Wed, Nov 27, 2019 at 12:27 PM Jann Horn <jannh@google.com> wrote:
>
> On Sun, Nov 24, 2019 at 12:08 AM Andy Lutomirski <luto@kernel.org> wrote:
> > On Fri, Nov 15, 2019 at 11:17 AM Jann Horn <jannh@google.com> wrote:
> > > A frequent cause of #GP exceptions are memory accesses to non-canonical
> > > addresses. Unlike #PF, #GP doesn't come with a fault address in CR2, so
> > > the kernel doesn't currently print the fault address for #GP.
> > > Luckily, we already have the necessary infrastructure for decoding X86
> > > instructions and computing the memory address that is being accessed;
> > > hook it up to the #GP handler so that we can figure out whether the #GP
> > > looks like it was caused by a non-canonical address, and if so, print
> > > that address.
> [...]
> > > +static void print_kernel_gp_address(struct pt_regs *regs)
> > > +{
> > > +#ifdef CONFIG_X86_64
> > > +       u8 insn_bytes[MAX_INSN_SIZE];
> > > +       struct insn insn;
> > > +       unsigned long addr_ref;
> > > +
> > > +       if (probe_kernel_read(insn_bytes, (void *)regs->ip, MAX_INSN_SIZE))
> > > +               return;
> > > +
> > > +       kernel_insn_init(&insn, insn_bytes, MAX_INSN_SIZE);
> > > +       insn_get_modrm(&insn);
> > > +       insn_get_sib(&insn);
> > > +       addr_ref = (unsigned long)insn_get_addr_ref(&insn, regs);
> [...]
> > > +}
> >
> > Could you refactor this a little bit so that we end up with a helper
> > that does the computation?  Something like:
> >
> > int probe_insn_get_memory_ref(void **addr, size_t *len, void *insn_addr);
> >
> > returns 1 if there was a memory operand and fills in addr and len,
> > returns 0 if there was no memory operand, and returns a negative error
> > on error.
> >
> > I think we're going to want this for #AC handling, too :)
>
> Mmmh... the instruction decoder doesn't currently give us a reliable
> access size though. (I know, I'm using it here regardless, but it
> doesn't really matter here if the decoded size is too big from time to
> time... whereas I imagine that that'd matter quite a bit for #AC
> handling.) IIRC e.g. a MOVZX that loads 1 byte into a 4-byte register
> is decoded as having .opnd_bytes==4; and if you look through
> arch/x86/lib/insn.c, there isn't even anything that would ever set
> ->opnd_bytes to 1. You'd have to add some plumbing to get reliable
> access sizes. I don't want to add a helper for this before the
> underlying infrastructure actually works properly.

Fair enough.  Although, with #AC, we know a priori that the address is
unaligned, so we could at least print "Unaligned access at 0x%lx\n".
But we can certainly leave these details to someone else.

(For context, there are patches floating around to enable a formerly
secret CPU feature to generate #AC on a LOCK instruction that spans a
cache line.)

--Andy