2022-02-02 13:31:24 EST Todd Humiston: We have time for one question
2022-02-02 13:32:00 EST Renato Levy: mite
2022-02-02 13:32:03 EST Renato Levy: mute
2022-02-02 13:32:12 EST Danielle Stewart: Interessante Präsentation, vielen Dank!
2022-02-02 13:32:14 EST Robbie VanVossen: Do you have plans to add virtualization support to your RISC-V platform?
2022-02-02 13:33:29 EST Darren Cofer: Also, you can ask yourself questions.
2022-02-02 13:33:58 EST Danielle Stewart: ^ The ones that are sooo good no one else thought to ask.
2022-02-02 13:34:16 EST Darren Cofer: It's bad if you stump yourself, though.
2022-02-02 13:34:34 EST Ihor Kuz: that leads to research though
2022-02-02 13:36:09 EST Todd Humiston: @Robbie: Sebastian Eckl / Axel Heider•01:35 PM yes, but the virtualization specification was just ratified recently, so we will have to address this
2022-02-02 13:55:53 EST Darren Cofer: Nice talk!
2022-02-02 13:56:17 EST Olivier Engelkes: Great talk, thank you!
2022-02-02 13:56:22 EST Danielle Stewart: Agreed. Great talk, Todd!
2022-02-02 13:56:42 EST Valerio Senni: When you eventually get to functions that have no subcomponents how do you abstract the flows?
2022-02-02 13:56:54 EST Valerio Senni: do you assume all possible I/O flows?
2022-02-02 13:57:19 EST John Hatcliff: AADL provides model level notations for describing those flows.
2022-02-02 13:58:02 EST John Hatcliff: Then, you need to bring in code-level analysis linked to the model level flows that confirms that code level conforms to (refines) the model-level flow.
2022-02-02 13:58:04 EST Valerio Senni: That notation is then used in the HAMR generated implementation as an assertion/monitor?
2022-02-02 13:58:49 EST Valerio Senni: amazing thanks
2022-02-02 13:59:10 EST John Hatcliff: Valerio, yes, the idea (not implemented yet) is that information flow specs at the model level are compiled down to specs in the code.
2022-02-02 13:59:55 EST John Hatcliff: It may be interesting to note that KSU worked on code-level information flow analysis for SPARK Ada with AdaCore in a five year AirForce Project.
2022-02-02 14:00:25 EST John Hatcliff: The technology that we develop as part of that project was the basis of the primary implementation of information flow analysis in SparkAda
2022-02-02 14:00:46 EST Valerio Senni: Thanks John, it would be nice to have a followup
2022-02-02 14:01:27 EST John Hatcliff: Valerio, sure. That would be great.
2022-02-02 14:02:12 EST Todd Carpenter: Thanks, John!
2022-02-02 14:17:18 EST Hui Lu: Wow, a nice tool to build seL4 VMs! Is it publically available?
2022-02-02 14:20:27 EST Robbie VanVossen: it is a licensed tool available for purchase
2022-02-02 14:20:33 EST Robbie VanVossen: https://dornerworks.com/vm-composer/
2022-02-02 14:20:38 EST Ammar Salman: Do all VMs use the same Linux kernel? How customizable are the VMs themselves?
2022-02-02 14:22:13 EST Valerio Senni: How far you support general CAmkES apps? You mention the VirtServer template and it's not clear to me how general those apps can be.
2022-02-02 14:22:20 EST Valerio Senni: You mentioned zcu102 support, how much of the ultrascale+ devices you cover and how complex is the extension? More in general, how complex is to extend board support?
2022-02-02 14:24:37 EST Hui Lu: Are devices shared between VMs?
2022-02-02 14:24:55 EST Ammar Salman: ^ I presume not for encapsulation.
2022-02-02 14:26:17 EST Hui Lu: Got it. That makes sense.
2022-02-02 14:26:20 EST Olin Sibert: Do you plan on shared storage or network support across multiple VMs?
2022-02-02 14:26:25 EST Valerio Senni: I think VirtServers are meant for sharing devices, that should be the standard template
2022-02-02 14:26:43 EST Renato Levy: @Olin, see my talk tomorrow....
2022-02-02 14:27:15 EST Olin Sibert: @Renato wouldn't miss it
2022-02-02 14:27:46 EST Valerio Senni: I think you can still develop your own CAmkES server on the basis of the generated code
2022-02-02 14:28:28 EST Renato Levy: any plans for a simplified environment instead of a full Linux? like a lightweight DOCKER
2022-02-02 14:28:40 EST Ammar Salman: Does the system automatically detect the necessary resources and spawns VMs for them or are the VMs running at all times?
2022-02-02 14:28:46 EST Olivier Engelkes: Thank you for your great talk!
2022-02-02 14:30:02 EST Ammar Salman: Great talk all around. Thank you for sharing!
2022-02-02 14:30:06 EST Robbie VanVossen: Our early support for docker will be a prebuilt VM that supports containers with automation to install your container images and start up scripts
2022-02-02 14:30:21 EST Todd Carpenter: Great talk! That looks like a slick tool.
2022-02-02 14:30:40 EST Robbie VanVossen: But I am interested to see way we can provide more minimal container environments that have less reliance on Linux
2022-02-02 14:31:11 EST Ammar Salman: The infrastructure is there, so it is a natural progression.
2022-02-02 14:31:17 EST Renato Levy: @Robbie, I did meant actual dockers, what i meant was that to run a full OS like Linux for 1 app is a waste of resources. Something lightweight can do most of the work
2022-02-02 14:32:20 EST Renato Levy: sorry, I agree with you, i did not see your reply until after I typed 😂
2022-02-02 14:34:41 EST Robbie VanVossen: ah, I see. I think there are a few other options we will provide that helps with this. One is using an RTOS VM instead of Linux. We already support RTEMS as a guest and have FreeRTOS on our roadmap. Another option is custom CAmkES components like Valerio was asking about.
2022-02-02 14:35:12 EST Robbie VanVossen: Thank you for all of the great questions!
2022-02-02 14:35:37 EST Ammar Salman: Is there support to utilize custom camkes components instead of the VMs?
2022-02-02 14:37:41 EST Robbie VanVossen: Not yet, but it is a logic progression for the tool. We are looking at customer/community needs to help drive the next features we implement, so if you have any suggestions, feel free to send them my way
2022-02-02 14:38:38 EST Michael Doran: One thing that we can point out is that the tool generates an seL4 project for you that can be modified as needed