Re: [Xenomai-core] [Adeos-main] New I-pipe patch for ARM needs testing.

[Jan Kiszka <jan.kiszka@domain.hid>]

[-- Attachment #1: Type: text/plain, Size: 1318 bytes --]

BOUIN Alexandre wrote:
> Same tests but for AT91SAM9263-EK board : 
>  - one under DD load, latency set to 45µs
>  - another one under Calibrator load, latency set to 30µs
>  - and the last under Test_switch load, latency set to 45µs
> Xeno-test traces are available and the end of this mail.
> 
> We used kernel 2.6.20.13, Xenomai snapshot from 08/06/2007, adeos-ipipe-2.6.20-arm-1.7-03.patch included plus 2.6.20-at91.patch.
> 
> A little correction on macb and ipipe patch was needed : 
>  - macb ethernet controler is not detected on startup
>  - we added some corrections to make 9263 usable under ipipe
>  - owing to changes in at91 patch (clk are now registered on startup) we made a little correction for future adeos-ipipe-2.6.21-arm.patch.
> 
> You will find all added patches attached to this mail.
>  <<adeos-ipipe-2.6.20-arm-AT91.patch>>  <<2.6.20-at91-macb-9263.patch>>  <<adeos-ipipe-2.6.20-arm-AT91.patch>> 
> Latencies are a little worst than on other boards. We presume that it's due to the large number of peripherals.

<pure theory without looking at details here>
Every additional IRQ source can cause additional delays of your
benchmarked event. All those IRQs may once in a while line up just in
front of your RT-IRQ, delaying it yet a bit more.
</end of theory>

Jan


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 250 bytes --]