linux-kernel - Re: [PATCH 5/7] ramoops: Split ftrace buffer space into per-CPU zones

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: <CAGXu5jJhFfomZu6wbgokxAAnCwVwuP3GM_G+vZGWhywb2FFWVQ@mail.gmail.com>
Date:   Mon, 10 Oct 2016 17:00:50 -0700
From:   Kees Cook <keescook@...omium.org>
To:     Joel Fernandes <joelaf@...gle.com>
Cc:     LKML <linux-kernel@...r.kernel.org>,
        Steven Rostedt <rostedt@...dmis.org>,
        Anton Vorontsov <anton@...msg.org>,
        Colin Cross <ccross@...roid.com>,
        Tony Luck <tony.luck@...el.com>
Subject: Re: [PATCH 5/7] ramoops: Split ftrace buffer space into per-CPU zones

On Mon, Oct 10, 2016 at 4:59 PM, Kees Cook <keescook@...omium.org> wrote:
> On Sun, Oct 9, 2016 at 10:15 AM, Joel Fernandes <joelaf@...gle.com> wrote:
>> On Fri, Oct 7, 2016 at 10:28 PM, Joel Fernandes <joelaf@...gle.com> wrote:
>>> If FTRACE_PER_CPU flag is passed to ramoops pdata, split the space into
>>> multiple zones depending on the number of CPUs.
>>>
>>> This speeds up the performance of function tracing by about 280% in my tests as
>>> we avoid the locking. The trade off being lesser space available per CPU.  Let
>>> the ramoops user decide which option they want based on pdata flag.
>>>
>>> Signed-off-by: Joel Fernandes <joelaf@...gle.com>
>>> ---
>>>  fs/pstore/ram.c            | 70 +++++++++++++++++++++++++++++++++++-----------
>>>  include/linux/pstore_ram.h |  3 ++
>>>  2 files changed, 56 insertions(+), 17 deletions(-)
>>>
>>> diff --git a/fs/pstore/ram.c b/fs/pstore/ram.c
>> [..]
>>> @@ -391,14 +418,21 @@ static void ramoops_free_przs(struct ramoops_context *cxt)
>>>  {
>>>         int i;
>>>
>>> -       if (!cxt->przs)
>>> -               return;
>>> +       /* Free dump PRZs */
>>> +       if (cxt->przs) {
>>> +               for (i = 0; i < cxt->max_dump_cnt; i++)
>>> +                       persistent_ram_free(cxt->przs[i]);
>>>
>>> -       for (i = 0; i < cxt->max_dump_cnt; i++)
>>> -               persistent_ram_free(cxt->przs[i]);
>>> +               kfree(cxt->przs);
>>> +               cxt->max_dump_cnt = 0;
>>> +       }
>>>
>>> -       kfree(cxt->przs);
>>> -       cxt->max_dump_cnt = 0;
>>> +       /* Free ftrace PRZs */
>>> +       if (cxt->fprzs) {
>>> +               for (i = 0; i < nr_cpu_ids; i++)
>>> +                       persistent_ram_free(cxt->przs[i]);
>>
>> I am supposed to free fprzs[i] here, instead of przs[i]. Also need to
>> check flag and free correct number of zones. Will fix for v2, sorry
>> about that.
>
> I think the cpu id needs to be bounds-checked against the size of the
> allocation. In theory, CPU hot-plug could grow the number of CPUs
> after pstore is initialized.

Oh, also, this new flag needs to be explosed as a module parameter and
Device Tree flag as well, so it's available for more than just the
dummy driver. :)

-Kees

-- 
Kees Cook
Nexus Security