[GSoC] RTEMS Tester Improvements

[Joel Sherrill]

On Apr 1, 2017 8:12 AM, "Tanu Hari Dixit" <tokencolour at gmail.com> wrote:

Hello Chris, All,

I have added PyYAML as the tool I'll be using to parse the yml files
in my proposal. Thank you for clearing that. Can you please look at
the proposal and give comments on the methods I have proposed to solve
the various problems? (In particular the "Proposed Schedule" section).

I had questions regarding the simulator recipes that need to be
supported in rtems-tester and I would be grateful if they are
answered.

The listed bsps are those that are supported by sim-scripts and not
rtems-tester:

Blackfin/bf537Stamp support on skyeye
ARM/CSB337 support on skyeye
MIPS/CSB350 support on skyeye
M68K-Coldfire/CSB360
ARM/EDB7312 support on skyeye
Blackfin/ezkit533 support on skyeye
ARM/Gumstix support on skyeye
SPARC/LEON2 support on skyeye
ARM/RTL22xx support on skyeye
ARM/SMDK2410 support on skyeye
arm/lm3s6965 support on qemu
i386/pc386 support on qemu
ARM/GumStix Connex support on qemu
SPARC/LEON2 support on qemu
lm32/lm32_evr support on qemu
OpenRISC/or1k support on or1ksim
PowerPC/QemuPPC support on qemu
arm/realview_pbx_a9_qemu_smp support on qemu
m68k/uc5282 support on qemu


I see a bf537Stamp-skyeye.in in sim-scripts but I don't see it listed
in https://devel.rtems.org/wiki/Developer/Simulators/SkyEye. Also
couldn't find ARM/Gumstix and ARM/SMDK2410 listed there. Are these
configurations supported? Should I be adding support for these?


Skyeye is a sad project to me. They had a lot of functionality and ran a
number of RTEMS BSPs. Then they undertook an overly ambitious
rearchitecting and seem to have killed the project.

It might be worth checking the status of skyeye to see if it improved but I
wouldn't put it anywhere near your critical path or put a lot of time in
it. It may be that RTEMS needs to make a decision that we use old releases
or simply abandon it if the current source is broken and the project
inactive.




It is given on the same web-page that arm/csb337 will run hello.exe
and ticker.exe on Skyeye. Would it be worth adding support for this
bsp if they can run a few tests only?
It is written that RTEMS BSP Csb350 should work once support in Skyeye
comes further along. What does this indicate?


That we we're hopeful. :(


Are these the simulator recipes that the devs wanted to see in
rtems-tester (other than for running RTEMS on gem5  for sparc64/usiii
and arm/realview_pbx_a9_qemu BSPs) ? Should I be focusing on a few
relevant ones? If so, which ones are more important?


The uc5282 had working networking at one point. In general ai would say
assess and don't waste time doing more than that on any specific BSPs.
Focus on the ones that work to improve the overall capabilities. If we know
which BSPs need work, we can nibble at known BSP specific issues.

I hope this helps. It would be better to have high rtems-tester
functionality​ on a few BSPs than to get hung up on BSP specific issues. We
just need them recorded with tickets.


Thank you,
Tanu Hari Dixit.

On Mon, Mar 27, 2017 at 6:37 AM, Chris Johns <chrisj at rtems.org> wrote:
> On 27/03/2017 04:34, Tanu Hari Dixit wrote:
>>
>> Also, do you suggest against using PyYAML to parse .yml configuration
>> files (since that will add a dependency into rtems-tester)?
>
>
> PyYAML might be perfect for the task. We will need to find a suitable
> solution as part of the change to Yaml. The YAML support maybe added to
the
> rtemstoolkit.
>
> Chris
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.rtems.org/pipermail/devel/attachments/20170401/63d83f65/attachment-0002.html>