Building leon3 prom image for testing on renode
Muhammad Sulthan Mazaya
msulthanmazaya at gmail.com
Sat Jul 29 02:12:28 UTC 2023
I have sent a new patch for this. For now, as I kept stucking on trying to
replace the prom with resc statements, I have decided to just add the prom
image binary to the rtems-tools and use it in the renode script.
- Mazaya
On Fri, Jul 21, 2023 at 4:13 AM Joel Sherrill <joel at rtems.org> wrote:
>
>
> On Wed, Jul 19, 2023 at 10:30 PM Alan Cudmore <alan.cudmore at gmail.com>
> wrote:
>
>> Hi Mazaya,
>>
>> On Tue, Jul 18, 2023 at 8:35 PM Muhammad Sulthan Mazaya <
>> msulthanmazaya at gmail.com> wrote:
>>
>>> Hi Alan,
>>>
>>> > Also, as you mentioned, the renode Zephyr example runs directly from
>>> RAM without a prom image. Is it that the RTEMS LEON3 BSP relies on
>>> initialization done by the boot prom (GRMON or MKPROM2 based boot code)?
>>>
>>> Do you or anyone on the devel mailing list know where/how can I check
>>> which kind of initialization rtems leon3 bsp rely on? Sorry, I am fairly
>>> new to this
>>>
>>> The prom image source code is pretty simple, but it would take a little
>> bit of time (for me at least) to decode since it is in SPARC assembly
>> language:
>> The code:
>>
>> https://github.com/TUT-ASI/leon3-grlib-gpl-mirror/blob/master/software/leon3/prom.S
>> And the prom.h header they use for the renode prom image:
>> https://github.com/antmicro/renode-rtems-leon3/blob/main/prom.h
>>
>> Essentially, it is setting up the processor including the stack pointer,
>> and minimal configuration (AMBA Plug and play?). The idea would be to
>> translate what the prom.S assembly program is doing into a set of .resc
>> statements to program configuration registers and set up anything that
>> RTEMS may need to run. It may be a subset of everything that the assembly
>> code does - It may come down to just one or two statements. I'm not being
>> specific with the steps, because I'm not sure myself.. I would have to take
>> some time to figure out exactly what it's doing.
>>
>
> I think we rely on grmon passing us the size of RAM which might be via the
> stack pointer we start on.
>
> There is also CLOCK_SPEED in start.S which is info from grmon which
> appears to be used in the clock driver.
>
> If we are relying on specific state of device setup, I don't know. I don't
> think we would by now given the use cases for this BSP.
>
>
>> Here is an example for the K210 where some of the peripheral registers
>> are programmed in the renode resc script:
>>
>> https://github.com/renode/renode/blob/master/scripts/single-node/kendryte_k210.resc#L10
>>
>> Replacing the prom with resc statements is one way to solve the problem -
>> by understanding the assembly and the memory mapped registers that are
>> being programmed. The other way to solve the problem is by figuring out the
>> way to supply the prom.bin file to users that need to run RTEMS/LEON3
>> binaries on renode. One is an embedded programming problem and the other is
>> an automation problem. I think both are valid solutions.
>>
>> Regards,
>> Alan
>>
>>
>>
>>> Thank you,
>>> Mazaya
>>>
>>> On Tue, Jul 18, 2023 at 11:27 PM Alan Cudmore <alan.cudmore at gmail.com>
>>> wrote:
>>>
>>>> Hi Mazaya,
>>>> I wonder if it is possible to initialize the processor to the state
>>>> where it can run the RTEMS image in RAM without using the prom image? The
>>>> processor initialization would have to go in the resc script before the
>>>> image is loaded and started.
>>>> Also, as you mentioned, the renode Zephyr example runs directly from
>>>> RAM without a prom image. Is it that the RTEMS LEON3 BSP relies on
>>>> initialization done by the boot prom (GRMON or MKPROM2 based boot code)?
>>>>
>>>> https://github.com/renode/renode/blob/master/scripts/single-node/leon3_zephyr.resc
>>>>
>>>> I guess it ends up being a trade off of tooling vs. trying to program
>>>> the necessary initialization in the resc script.
>>>>
>>>> Regards,
>>>> Alan
>>>>
>>>> On Tue, Jul 18, 2023 at 2:46 AM Muhammad Sulthan Mazaya <
>>>> msulthanmazaya at gmail.com> wrote:
>>>>
>>>>> Hi,
>>>>>
>>>>> I was trying to implement renode for testing leon3 using rtems-test. I
>>>>> noticed that from the example they provide
>>>>> https://github.com/antmicro/renode-rtems-leon3/blob/main/leon3_rtems.resc,
>>>>> it seems like the renode simulator needs the prom image of leon3 to do so.
>>>>> I don't think an additional build step before testing a particular bsp is
>>>>> something supported by the build config of rtems-tools + my mentors
>>>>> also said it shouldn't be on the rtems-tools. Is it okay to have RSB build
>>>>> that for us as part of the BSP?
>>>>>
>>>>> Thank you,
>>>>> Mazaya
>>>>> _______________________________________________
>>>>> devel mailing list
>>>>> devel at rtems.org
>>>>> http://lists.rtems.org/mailman/listinfo/devel
>>>>
>>>> _______________________________________________
>> devel mailing list
>> devel at rtems.org
>> http://lists.rtems.org/mailman/listinfo/devel
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.rtems.org/pipermail/devel/attachments/20230729/281572f8/attachment-0001.htm>
More information about the devel
mailing list