Follow @Openwall on Twitter for new release announcements and other news
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Date: Sat, 29 Oct 2016 01:27:38 -0400
From: David Windsor <dave@...gbits.org>
To: kernel-hardening@...ts.openwall.com
Cc: Kees Cook <keescook@...omium.org>, Hans Liljestrand <ishkamiel@...il.com>
Subject: Re: [RFC v2 PATCH 00/13] HARDENED_ATOMIC

On Fri, Oct 28, 2016 at 3:51 PM, Reshetova, Elena
<elena.reshetova@...el.com> wrote:
>>Done.  I added a line to the "HARDENED_ATOMIC Implementation" section of Documentation/security/hardened-atomic.txt describing the results of the benchmarks (no measurable performance difference).  You might want to add this line to the cover >letter as well.
>
> Thank you David! I will merge it tomorrow before sending a new rfc. I will also add the performance line to cover letter.
>
>> I know I promised to post the results of another set of benchmarks I performed, and I will do that soon.
>
> If you can send it before tomorrow, would be great! We can include them in v3 straight then.
> I would like to get v3 out tomorrow that people can look at it before the kernel summit.
>

The benchmark results don't have anything that will actually go into
the RFC.  The results were that there's no measurable performance
degradation, and this language has already been included in my update
to the documentation.  I'll post the benchmark results when I get a
chance, but they're really just for informational purposes at this
point.

> Best Regards,
> Elena.
>
> On Fri, Oct 28, 2016 at 9:02 AM, Reshetova, Elena <elena.reshetova@...el.com> wrote:
>>>> - add missing tests for atomic64 and local
>>>> - rebase on top of latest linux-next
>>>> - compile test and test run the whole thing in different
>>>> combinations
>>>> - send rfcv3 with also all atomic maintainers included for wider
>>>> blame/feedback
>>>>
>>>> Does it sound like a good plan for everyone?
>>>>
>>
>>> Actually, it doesn't look like I've updated Documentation/security/hardened-atomic.txt yet.  I need to fix the language explaining the x86 race condition to make it clear that we're discussing the SMP case.
>>> I also want to add a sentence somewhere (either in your cover letter or in the kernel documentation, or both), referencing the benchmark results and lack of demonstrable performance degradation.
>>
>> David, could you please push the changes you want to do to the documentation in separate commit to the top of hardened_atomic_on_next?
>>  I will cherry pick them to our new rebased branch hardened_atomic_next that we still working actively now.
>>
>> Best Regards,
>> Elena.

Powered by blists - more mailing lists

Confused about mailing lists and their use? Read about mailing lists on Wikipedia and check out these guidelines on proper formatting of your messages.