On 2018-01-12 22:30, Gustavo Lima Chaves wrote: > * Jan Kiszka <[email protected]> [2018-01-12 13:15:23 +0100]: > >> On 2018-01-12 01:24, Gustavo Lima Chaves wrote: >>> Before this patch, that routine was being called needlessly a lot of >>> times on each new partition setup. We just need it once, though, and we >>> choose to do so at pci_remove_physical_device() (ivshmem has its own >>> call as well). >>> >>> Signed-off-by: Gustavo Lima Chaves <[email protected]> >>> --- >>> hypervisor/control.c | 1 - >>> hypervisor/include/jailhouse/pci.h | 1 - >>> hypervisor/pci.c | 11 ----------- >>> 3 files changed, 13 deletions(-) >>> >>> diff --git a/hypervisor/control.c b/hypervisor/control.c >>> index 52ef8733..c49532bc 100644 >>> --- a/hypervisor/control.c >>> +++ b/hypervisor/control.c >>> @@ -571,7 +571,6 @@ static int cell_start(struct per_cpu *cpu_data, >>> unsigned long id) >>> cell->comm_page.comm_region.cell_state = JAILHOUSE_CELL_RUNNING; >>> cell->comm_page.comm_region.msg_to_cell = JAILHOUSE_MSG_NONE; >>> >>> - pci_cell_reset(cell); >> >> This would mean we no longer reset devices on cell reset - that's not >> correct. > > I'm afraid I need more input on this, sorry. Isn't the expected flow > from cell management create, load, start, destroy? Does Jailhouse give > any guarantees if load/start are issued repeatedly with no destroys? > If we destroy, we have to hand devices over again from root to guests > and that's where FLR is covered.
There is a case of reload without destruction as well. You just stop the cell by setting it loadable, then reload the same or a newer version of the cell code & data and start it again. That won't even require the OK of other voting cells. Jan -- Siemens AG, Corporate Technology, CT RDA IOT SES-DE 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]. For more options, visit https://groups.google.com/d/optout.
