Hello all, mostly Kent,
I read your blog post about rumpkernels on sel4: https://research.csiro.au/tsblog/using-rump-kernels-to-run-unmodified-netbsd...
and after experimenting with the rumprun apps (such as the ethernet one https://github.com/seL4/camkes/tree/master/apps/rumprun_ethernet) I think it is a great addition to sel4.
So I have few questions. I need to run two different rumpkernels, each with its own address space, connected only through dataports/events.
1- does the current rumpkernel backend support having this separation? According to your thesis http://www.disy.cse.unsw.edu.au/theses_public/16/kentm.pdf, all rumpkernel threads run in a single sel4 thread - is that correct?
2- Right now, I can combine multiple rumprun components in one application (lets say *runmprun_hello* and *rumprun_pthreads*). butI can have only one rumprun_platform_layer per application, correct? An example below:
*assembly {* * composition {* * component rumprun_platform_layer rrpl;* * component rumprun hello;* * RUMPRUN_META_CONNECTION(hello, rrpl)*
* component rumprun_pthreads pthreads;* * RUMPRUN_META_CONNECTION(pthreads, rrpl)* * }*
* configuration {* * hello.rump_config = {"cmdline": "hello World!" };* * RUMPRUN_COMPONENT_CONFIGURATION(hello, 0)*
* pthreads.rump_config = {"cmdline": "./pthreads_test 4" };* * RUMPRUN_COMPONENT_CONFIGURATION(pthreads, 1)* * }* *}*
If I instantiate another *rumprun_platform* component, it wont compile.
3 - how can I go around defining/adding my own rumpkernels? Lets say I want to build a rumpkernel with rust support https://github.com/rumpkernel/rumprun-packages/tree/master/rust (you have an sel4 example here https://github.com/SEL4PROJ/rumprun-sel4-demoapps/tree/master/userapps/rust) and integrate it with the rumprun_hello https://github.com/seL4/camkes/blob/master/apps/rumprun_hello/rumprun_hello.camkes#L19 app.
Regards Michal