Hi Jan, so if I wanted to start a second Linux inmate cell and mount a rootFS installed on the second partition of my eMMC (/dev/mmcblk2p2) is it feasible or not? It's not still much clear...sorry. What do you mean *"In that case, you need one cell taking ownership of the controller, serving other cells as proxy."*?
Kind Regards. Stefano Il giorno gio 10 giu 2021 alle ore 10:53 Jan Kiszka <[email protected]> ha scritto: > On 10.06.21 09:47, laurent gauty wrote: > > Hi, > > > > I'm currently playing with a NXP dev board based on iMX8Mplus. I'm very > > interested by the approach you have developed in the jailhouse project > > in order to segregate different cells running different systems. > > But as i tried to install jailhouse and played with cells, i reached a > > point where i don't understand how jailhouse in intend to behave on > > persistent memory (i.e. eMMC / SD). > > > > The current configuration is to have several cells using the same SD > > card / eMMC, but relying on different partitions. And i think that most > > of the projects in embedded world will have to deal with only one memory > > device (SD, eMMC) for cost and space reasons. > > > > So here is my question: > > is there a way in jailhouse to isolate the persistent memory (for > > example partition) of each cells from each others ? > > If not existing today, do you plan for this kind of features in the > future ? > > > > The general issue with partitioning of storage is the partitioning of > the device access methods to it. With something like AHCI, NVMe or MMC, > you have to go through a central point, the controller, in order to talk > to different partitions of the same medium. In that case, you need one > cell taking ownership of the controller, serving other cells as proxy. > That would be the Jailhouse model of it: no device driver in the > hypervisor, rather using a cell as driver. To provide virtual access on > the physical storage, something like virtio-over-shmem + virtio-block > could be a pattern (though not yet production-grade). Or storage over > (virtual) network. > > An alternative to the proxy cell could be if you have a directly > memory-mappable storage that does not require coordinated access on a > central controller. Or if you had something like SR-IOV capable storage > controllers that could present themselves as multiple, thus > hardware-partitionable, devices to the host. None of both we've seen or > tried so far. > > HTH, > Jan > > -- > Siemens AG, T RDA IOT > Corporate 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/0d8757b9-c2f7-266f-8ded-709c632f955d%40siemens.com > . > -- 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/CAKpvFF10mGXWcmgj%2B%3DQy4ypi39cB-wQdtLVzn6OwppOtuLms7g%40mail.gmail.com.
