Help testing arora,qt5 and sdl examples on RaspberryPI

[Reinier Millo Sánchez]

Hi Norman
That's right, this is the problem on Arora, QT5, and SDL examples

    [init] fb_drv: no route to service "Platform"
    [init] usb_drv: no route to service "Platform"

now I'm checking the platform driver and USB HID at the run scripts. I'm 
emulating the examples on QEMU to get the serial output. I have switched 
to Genode 14.08, but now in 14.08 the Demo example don't run on the real 
harware. I have tested successfully the Demo example from Genode 14.05, 
i'm goint to check the main difference between versions to try catch the 
problem.
Best regards

On 09/05/2014 02:58 PM, Norman Feske wrote:
> Hello Reinier,
>
>> I have switched to Genode's release 14.08, prepared again the ports and
>> compiled the examples in the same way, but when try to test it on RPI,
>> only the colored screen again.
>> Someone have tested successfully this examples on RPI hardware? I have
>> tested successfully the Demo example.
> the SDL and Qt5 examples are indeed not expected to work on the
> Raspberry Pi out of the box. Most run scripts still lack two things that
> are important for the RPi, namely the platform driver and the USB HID
> driver. Those components are included in the 'os/run/demo.run' script
> but are lacking in many others such as the 'libports/run/sdl.run'.
> However, you should still get serial output in any case. Normally, the
> serial output gives enough indications about the missing pieces. Given
> the demo.run script as example, complementing the other run scripts with
> the missing pieces should be straight forward.
>
> Has the switch to version 14.08 resolved the build problem?
>
> As another heads-up warning regarding the Rpi, the USB driver is still
> not perfect. There are two issues. First, the keyboard is not working
> reliably unless you set the number of used host channels to 8 instead of
> 2 using the 'dwc_param_host_channels_default' define in
> <genode-dir>/contrib/dde_linux-<hash>/src/lib/dde_linux/drivers/usb/host/dwc_otg/dwc_otg_core_if.h.
> Second, the USB driver draws an enormous amount of CPU time because it
> has to issue USB start-of-frame (SOF) transactions every microframe
> (that is every 250 microseconds). So when the USB driver is running,
> there is hardly any CPU time left for the other processes. On Linux,
> this problem has been somehow fixed by the so-called FIQ optimization. I
> am still pondering about the best way to relieve this issue when using
> the base-hw kernel.
>
> Best regards
> Norman
>

-- 
Lic. Reinier Millo Sánchez
Centro de Estudios de Informática
Universidad Central "Marta Abreu" de Las Villas

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.genode.org/pipermail/users/attachments/20140905/900c42a3/attachment.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: rmillo.vcf
Type: text/x-vcard
Size: 519 bytes
Desc: not available
URL: <http://lists.genode.org/pipermail/users/attachments/20140905/900c42a3/attachment.vcf>