On 24.11.22 20:19, Yelena Konyukh wrote: > Hi All, > > Answering my own question :) : > > I have run the demo from https://github.com/siemens/jailhouse-images for > Raspberry Pie and simply collected the output of "cat /proc/iomem" > before running Jailhouse, then after enabling Jailhouse in the root cell > and in a Linux non-root cell. From that output it looks like Jialhouse > does 1:1 mapping between IPAs and PAs.
It does so in many cases, but the 2nd state MMU would permit a different mapping as well. However, we do not have broad SMMU (IOMMU) support on ARM SoCs so that we often have to keep a 1:1 mapping even for non-root cells to work around that. > > On a platform with Zynq UltraScale+ MPSoC, onto which Jailhouse has been > ported, I have experimented with memory accesses, trying to access from > the root cell the memory allocated to a non-root Linux cell and the > other way round - trying to access from the non-root Linux cell the > memory allocated to the root cell only. As expected, in both cases I > have got "Unhandled data write...FATAL: unhandled trap (exception class > 0x24)", while the cells can happily read from/write into physical > addresses belonging to their own memory regions specified in the > respective cells configuration. This appears to confirm the use of the > 2nd stage translations for restricting cells access to physical memory. Yes, primarily. For remapping physical addresses, see above. Jan -- Siemens AG, Technology Competence Center Embedded Linux -- You received this message because you are subscribed to the Google Groups "Jailhouse" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. To view this discussion on the web visit https://groups.google.com/d/msgid/jailhouse-dev/97c9d6b2-778d-5e20-fc2e-00537d9d5e07%40siemens.com.
