linux-kernel - Re: NFS crash, hashed pointers in backtrace

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] [day] [month] [year] [list]

Message-ID: <CAMuHMdVOa9Bxph=QBeNUJBZ=_pwGrr+aZr7kZ5WvhUNRQZMRow@mail.gmail.com>
Date:   Fri, 8 Dec 2017 14:19:07 +0100
From:   Geert Uytterhoeven <geert@...ux-m68k.org>
To:     Trond Myklebust <trondmy@...marydata.com>
Cc:     "anna.schumaker@...app.com" <anna.schumaker@...app.com>,
        "linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>,
        "linux-renesas-soc@...r.kernel.org" 
        <linux-renesas-soc@...r.kernel.org>,
        "linux-nfs@...r.kernel.org" <linux-nfs@...r.kernel.org>,
        "me@...in.cc" <me@...in.cc>
Subject: Re: NFS crash, hashed pointers in backtrace

On Wed, Dec 6, 2017 at 5:19 PM, Geert Uytterhoeven <geert@...ux-m68k.org> wrote:
> On Wed, Dec 6, 2017 at 5:10 PM, Trond Myklebust <trondmy@...marydata.com> wrote:
>> On Wed, 2017-12-06 at 15:31 +0100, Geert Uytterhoeven wrote:
>>> On Tue, Dec 5, 2017 at 5:02 PM, Geert Uytterhoeven <geert@...ux-m68k.
>>> org> wrote:
>>> Got another nfsroot crash:
>>>
>>> Unable to handle kernel NULL pointer dereference at virtual address
>>> 00000030
>>> pgd = 329e8f6e
>>> [00000030] *pgd=80000040004003, *pmd=00000000
>>> Internal error: Oops: 206 [#1] SMP ARM
>>> Modules linked in:
>>> CPU: 0 PID: 101 Comm: kworker/u4:1 Not tainted
>>> 4.15.0-rc2-koelsch-01166-g047d7d3248e08fc7-dirty #3762
>>> Hardware name: Generic R-Car Gen2 (Flattened Device Tree)
>>> Workqueue: writeback wb_workfn (flush-0:15)
>>> task: 8a5bf858 task.stack: e93c92bc
>>> PC is at nfs_page_async_flush+0x110/0x244
>
>>> static int nfs_page_async_flush(struct nfs_pageio_descriptor *pgio,
>>>                                 struct page *page)
>>> {
>>>         struct nfs_page *req;
>>>         int ret = 0;
>>>
>>>         ...
>>>
>>>         /* If there is a fatal error that covers this write, just
>>> exit */
>>>         if (nfs_error_is_fatal_on_server(req->wb_context->error))
>>>                 goto out_launder;
>>>
>>> c03bc644:       e595300c        ldr     r3, [r5, #12]
>>> c03bc648:       e5930030        ldr     r0, [r3, #48]   ; 0x30
>>> c03bc64c:       ebfffd1b        bl      c03bbac0
>>> <nfs_error_is_fatal_on_server>
>>>
>>> req->wb_context must be NULL.
>>
>> I'm confused. If your test involves only writing to a sysfs file, then
>> why is the NFS code involved at all?
>
> I don't think the second was related to sysfs.
>
>> Could this be a use-after-free?
>
> Possibly. I'm seeing other crashes, too. Looking into them...

Found it: https://lkml.org/lkml/2017/12/8/399
That one caused corruption (zeroing) of the 4th 32-bit word of a memory block,
which is consistent with the "ldr r3, [r5, #12]" loading NULL above.

So NFS is fine (as usual ;-), sorry for the fuzz...

Gr{oetje,eeting}s,

                        Geert

--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@...ux-m68k.org

In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
                                -- Linus Torvalds