Build for ARMv8 vm_minimal_sim build should not run QEMU

[axel-h]

Now that we have separate build and HW-test steps, it might make sense to avoid running the QEMU simulation as part of the build for vm_minimal_sim, and have a dedicated simulation step here. The issue I am currently facing is, that the simulation fails hafer a few seconds, then the the queue is stuck for 50 min to recognize the failure and terminate the simulation.

This is what I'm seeing from https://github.com/seL4/camkes-vm/actions/runs/4684601834/jobs/8300924624?pr=81

2023-04-13T01:43:13.7011661Z Requested labels: ubuntu-latest
...
2023-04-13T01:46:05.5385900Z ##[group]vm_minimal_sim
2023-04-13T01:46:05.5386227Z �[1m-----------[ start test vm_minimal_sim ]-----------�[0m
2023-04-13T01:46:05.6144625Z �[33m+++ ../init-build.sh -DAARCH64=TRUE -DPLATFORM=qemu-arm-virt -DCAMKES_VM_APP=vm_minimal�[0m
2023-04-13T01:46:27.7973409Z [407/407] Generating images/capdl-loader-image-arm-qemu-arm-virt
2023-04-13T01:46:27.7980266Z �[33m+++ tar czf ../vm_minimal_sim-images.tar.gz images/�[0m
2023-04-13T01:46:29.0389620Z �[33m+++ bash -c expect -c 'spawn ./simulate; set timeout 3000; expect "buildroot login:"'�[0m
2023-04-13T01:46:29.0516106Z spawn ./simulate
2023-04-13T01:46:29.1339565Z ./simulate: qemu-system-aarch64 -machine virt,virtualization=on,highmem=off,secure=off -cpu cortex-a53 -nographic  -m size=2048  -kernel images/capdl-loader-image-arm-qemu-arm-virt
2023-04-13T01:46:29.1343762Z 
2023-04-13T01:46:29.1346682Z ELF-loader started on CPU: ARM Ltd. Cortex-A53 r0p4
...
2023-04-13T01:46:30.6131863Z FAULT HANDLER: data fault from vm0.vm0_0_control (ID 0x2) on address 0x90, pc = 0x464014, fsr = 0x93800006
...
2023-04-13T01:46:30.6302854Z FAULT HANDLER:     +-- 0x0000000000400000 --
2023-04-13T01:46:30.6302985Z
...
...  now we are stuck for 50 min ...
...
2023-04-13T02:36:29.6319786Z �[1m-----------[ end test vm_minimal_sim ]-----------�[0m
2023-04-13T02:36:29.6320403Z ##[endgroup]
2023-04-13T02:36:29.6320682Z �[32mvm_minimal_sim succeeded�[0m
2023-04-13T02:36:29.6320829Z 
2023-04-13T02:36:29.6321087Z ##[group]vm_minimal_smp_tx1
...
2023-04-13T02:37:25.9190959Z Cleaning up orphan processes

[axel-h]

The quicker fix mit be changing set timeout 3000 to set timeout 300 as 5 mins should be sufficient also for the minimal VM scenario.

[axel-h]

Seems vm_minimal_sim is explicitly excluded in the hw builds at the moment, see https://github.com/seL4/camkes-vm/actions/runs/4715496818/jobs/8362557510?pr=87, this should remain disabled then, because QEMU simulation runs should be a separate step.

Successful tests: vm_minimal_tx1, vm_minimal_tx2, vm_minimal_smp_tx1, vm_minimal_smp_tx2, vm_virtio_net_arping_tx2, vm_virtio_net_ping_tx2

SKIPPED tests: vm_minimal_sim

[lsf37]

Makes sense, I'll see if there are any specific problems in making it a separate job.

[lsf37]

With seL4/ci-actions#332 it will be easier to manage workflows over multiple repos, so I'll defer this until we have gotten that working properly.