PREEMPT-RT

[Pierre Ficheux]

Hi,

I'm proud to announce I've integrated a first version of PREEMPT-RT
extension in Android-x86 (jb-4.2, kernel 3.8.0). It works fine, just
like on GNU/Linux. It would help to create RT apps (with RT
constraints) on top of Android.

rt.wiki.kernel.org

regards

--
Pierre

[安琪]

What is RT?

2013/5/25 Pierre Ficheux <pierre....@gmail.com>

--
You received this message because you are subscribed to the Google Groups "Android-x86" group.
To unsubscribe from this group and stop receiving emails from it, send an email to android-x86...@googlegroups.com.
To post to this group, send email to andro...@googlegroups.com.
Visit this group at http://groups.google.com/group/android-x86?hl=en.
For more options, visit https://groups.google.com/groups/opt_out.

[Pierre Ficheux]

RT = Real Time

Lots of industrial systems use realtime constraint (just like on
VxWorks, QNX, LynxOS). Linux kernel is not realtime but patches such
as PREEMPT-RT or Xenomai do the job much better. It's very helpful to
get it work on Android :)

regards

2013/5/25 安琪 <sayy...@gmail.com>:

--
Pierre

[bryanhundven]

On Saturday, May 25, 2013 12:56:24 AM UTC-7, Pierre Ficheux wrote:

RT = Real Time

Lots of industrial systems use realtime constraint (just like on
VxWorks, QNX, LynxOS). Linux kernel is not realtime but patches such
as PREEMPT-RT or Xenomai do the job much better. It's very helpful to
get it work on Android :)

Very interesting. I'm curious to know the performance, power savings, and/or how it would handle device drivers with weird timing code and lack of spin/wake-lock guards (i2c, spi, and other platform devices not normally used with android). Has this been done with any other android port? Did anything change that is part of the system and not in the kernel to make things work properly?

Not criticizing, just very curious. :D GJ!

-Bryan

[Pierre Ficheux]

It was just "tested" on Android-x86 for 3 reasons :

1- Android-x86 jb kernel (3.8.0) was close to the last PREEMPT-RT
patch version (3.8.4 and +). ARM based Android (Pandaboard) is based
on older kernel (3.0) quite different from Linux.

2- PREEMPT-RT works fine on x86, though it works not so bad on recent ARM

3- It was just a 1rst test :)

I just wanted to check whether Android will boot with such a kernel.
I used my favorite rt-tests tools (cyclictest, hackbench, ...) to
compare jitter with standard Android kernel. Looks good, just like on
GNU/Linux => small jitter
with PREEMPT-RT, sometimes "infinite" with standard kernel. It should
helpful to design device with Android based GUI and low latency at
kernel/system level.

regards


2013/5/29 bryanhundven <bryanh...@gmail.com>:

[Chih-Wei Huang]

Thank you for the effort.
Would you like to push the RT branch to
android-x86 repository so more people can test it?
Could you write docs for how to use or test it?


2013/5/29 Pierre Ficheux <pierre....@gmail.com>:

Chih-Wei
Android-x86 project
http://www.android-x86.org

[Pierre Ficheux]

I will do it that asap. I'm currently writing some docs.

regards

2013/5/29 Chih-Wei Huang <cwh...@android-x86.org>:

[Pierre Ficheux]

Hi,

My PREEMPT-RT patch is 710 kb, may be it's better to send a web link
on the list instead of file?

regards

2013/5/29 Chih-Wei Huang <cwh...@android-x86.org>:

[Bryan Hundven]

Pierre, All,

On 6/4/13 4:25 PM, "Pierre Ficheux" <pierre....@gmail.com> wrote:

>Hi,
>
>My PREEMPT-RT patch is 710 kb, may be it's better to send a web link
>on the list instead of file?

I personally think just sending a git patch to the group should be the
easiest, besides getting push access to the RT branch like Chih said and
just pushing the changes yourself.

I have some more comments below, inline.

>regards
>
>2013/5/29 Chih-Wei Huang <cwh...@android-x86.org>:
>> Thank you for the effort.

Indeed! I second that thanks!

>> Would you like to push the RT branch to
>> android-x86 repository so more people can test it?
>> Could you write docs for how to use or test it?

Docs would be great, and should have a minimal set of documentation with
the first patch, in my opinion.

>>
>>
>> 2013/5/29 Pierre Ficheux <pierre....@gmail.com>:
>>> It was just "tested" on Android-x86 for 3 reasons :
>>>
>>> 1- Android-x86 jb kernel (3.8.0) was close to the last PREEMPT-RT
>>> patch version (3.8.4 and +). ARM based Android (Pandaboard) is based
>>> on older kernel (3.0) quite different from Linux.
>>>
>>> 2- PREEMPT-RT works fine on x86, though it works not so bad on recent
>>>ARM
>>>
>>> 3- It was just a 1rst test :)

:)

>>>
>>> I just wanted to check whether Android will boot with such a kernel.
>>> I used my favorite rt-tests tools (cyclictest, hackbench, ...) to
>>> compare jitter with standard Android kernel. Looks good, just like on
>>> GNU/Linux => small jitter
>>> with PREEMPT-RT, sometimes "infinite" with standard kernel. It should
>>> helpful to design device with Android based GUI and low latency at
>>> kernel/system level.

I think this is great. I'm excited to see a patch.