Closed rajnesh-kanwal closed 1 year ago
probably need to update
tellus
too?
tellus
uses the register_shmem
function in sbi-rs which takes in shmem_ptr
and converts it to pfn, with register_shmem
updated tellus
doesn't need any changes.
I can add a test for ProbeFeature
if that's what you are suggesting.
Ah, I was looking at the CI failure (which is because Atish's PR isn't in yet).
I am thinking to add CoVE related features which can be probed as well.
SBI_NACL_FEAT_COVE_GPR -- This will describe the the scratch layout for GPRs
SBI_NACL_FEAT_COVE_CSR
This will align both specs more cleanly and clearly describe where CoVE supersedes NACL description. It will also allow other TSM/host hypervisor implementations to not use the NACL features if they wish to. It allows extensibility for any possible future use case as well.
Any thoughts ? This can be done a separate PR as spec change also need to drafted.
I am thinking to add CoVE related features which can be probed as well.
Doesn't probe just say if those features are present? i.e. if we want to know whether or not the shmem scratch area has the COVE layout, don't we need just one "feature" that indicates we have COVE?
I have pointed sbi-rs to https://github.com/rivosinc/sbi-rs/commit/9b83bc49608177d092848f41880064feef5886fd.
I am thinking to add CoVE related features which can be probed as well.
Doesn't probe just say if those features are present? i.e. if we want to know whether or not the shmem scratch area has the COVE layout, don't we need just one "feature" that indicates we have COVE?
One will work too. I was suggesting two that aligns better in terms of current spec alignment (scratch area/ CSR area). We can describe the scratch & CSR access exception under the same feature.
In the recent updates to Nacl extension [0], feature probing was introduced and allows guest hypervisor to probe which features are supported by the host hypervisor.
Also, the SetShmem FID argument has changed from pfn to full address of the shared memory page.
[0] https://lists.riscv.org/g/sig-hypervisors/message/256