Holger Adams | 21 Jun 2012 15:49
Favicon

OMAP4430 /Pandaboard using RT_PREEMPT - Experience?

Hello Everybody,

is somebody using the OMAP4430 with RT_PREEMPT and is willing to tell us 
about their experience regarding worst case latencies, stability issues, 
kernel/patch versions?

Thanks!
Holger

Emmanuel Deloget | 22 Jun 2012 11:26
Picon
Favicon
Gravatar

Re: OMAP4430 /Pandaboard using RT_PREEMPT - Experience?

I successfully booted a PB ES with a 3.2-something-rt kernel a while ago. I haven't tested much and I don't have pretty numbers to show out, but the board itself seemed to be stable. 


I guess I used the latest 3.2 and latest available rt patch at that time. I still have a 3.2.18-rt29 patch in my download directory, so I guess this was around that time. When building my image, I used the vanilla kernel + the rt patch + the wlan patch (http://elinux.org/images/8/8d/0001a-omap4-pandaboard-wlan-fix.patch). 

The image successfully booted on my Pandabaord ES and on a friend's Pandaboard. I guess that more recent kernels (with more recent RT patches) also boots, as there is a very limited activity on the RT patch itself - and recent kernels seems to boot quite easily on any Pandaboard. I will probably check 3.4 + rt in the coming weeks. 

Best regards, 

-- Emmanuel

On Thursday, June 21, 2012 3:49:56 PM UTC+2, Holger Adams wrote:
Hello Everybody,

is somebody using the OMAP4430 with RT_PREEMPT and is willing to tell us
about their experience regarding worst case latencies, stability issues,
kernel/patch versions?

Thanks!
Holger

Holger Adams | 28 Jun 2012 15:35
Favicon

Re: OMAP4430 /Pandaboard using RT_PREEMPT - Experience?

Hello Emmanuel,

Am Fri, 22 Jun 2012 02:26:53 -0700 schrieb Emmanuel Deloget:

> The image successfully booted on my Pandabaord ES and on a friend's
> Pandaboard. I guess that more recent kernels (with more recent RT
> patches) also boots, as there is a very limited activity on the RT patch
> itself - and recent kernels seems to boot quite easily on any
> Pandaboard. I will probably check 3.4 + rt in the coming weeks.

Thanks for you informations!

I tried following setup:
 - Kernel: 3.4.4
 - RT-Patch: 3.4.4-rt13 

# uname -a
Linux pdev 3.4.4-rt13 #1 SMP PREEMPT RT Thu Jun 28 11:34:25 CEST 2012 
armv7l armv7l armv7l GNU/Linux

# cyclictest -t1 -p 80 -i 500 -l 10000000000

policy: fifo: loadavg: 64.78 68.71 68.55 3/214 4153          
Min:     52 Act:   84 Avg:   85 Max:    5014

(around 7656995 cycles under heavy CPU and I/O-Load. Burnt my fingers 
touching the chip)

The worst case jitter increases about 5000uS if I start write/read 
operations onto the SD-Card. Is this phenomena known out there? I disabled 
all powersaving settings in the kernel, also the the high precision timer 
(not this 32KHz thing) is being used.

Best Regards,
Holger


Gmane