On 08.06.21 13:26, Dongjiu Geng wrote: > On Tue, 8 Jun 2021 at 18:53, Jan Kiszka <[email protected]> wrote: >> >> On 08.06.21 11:20, Dongjiu Geng wrote: >>> Hi, >>> From the jailhouse design, the Private Per-CPU data should be not >>> accessed by other CPUs except the current CPU. But from the code[1] >>> and my test, it is not. For example, the CPU1 can access the CPU0's >>> private data, because hv_paging_structs already map it, and the >>> per-cpu page table will link the hv_paging_structs. is this a bug？ I >>> think the Private Per-CPU data should be not accessed by other CPUs >>> except the self CPU. >>> >>> [1]: err = paging_create(&hv_paging_structs, >>> paging_hvirt2phys(&hypervisor_header), >>> system_config->hypervisor_memory.size, >>> (unsigned long)&hypervisor_header, >>> PAGE_DEFAULT_FLAGS, >>> PAGING_NON_COHERENT | PAGING_HUGE); >>> >> >> You are referring to >> https://github.com/siemens/jailhouse/blob/6d9c51d0bd819689c00f6a3c38d3099f6eb9c657/hypervisor/paging.c#L678: >> Right, the initial mapping means everything is visible for all CPUs. But >> then comes >> https://github.com/siemens/jailhouse/blob/6d9c51d0bd819689c00f6a3c38d3099f6eb9c657/hypervisor/setup.c#L100 >> which setups up an alternative mapping (installed by arch_cpu_init()). >> That does not contain the private data structs of the other CPUs. You >> should be able to confirm that AFTER initialization, e.g. on a first VM >> exit after setup. > > Thanks very much for the answer, could you paste the code where it > removes the private data structs of the other CPUs? I do not find that > logic. After enabling jailhouse hypervisor and returning to the first > VM0. the CPU0 still can access other CPU's private data structs, as > shown in[1]. > > [1]: > @@ -261,8 +262,10 @@ int entry(unsigned int cpu_id, struct per_cpu *cpu_data) > return error; > } > > - if (master) > + if (master) { > + printk("10: > ----------------------------------------------------------%lx----\n", > per_cpu(10)->id_aa64mmfr0); > printk("Activating hypervisor\n"); > + } >
Ah, need to correct myself: https://github.com/siemens/jailhouse/blob/6d9c51d0bd819689c00f6a3c38d3099f6eb9c657/hypervisor/arch/arm64/setup.c#L103 So the disabling happens even later, in arch_cpu_activate_vmm. The alternative mapping we built up and installed in arch_cpu_init still contains full access as you noticed. Again, please try after the first vmexit. 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/a1b7be32-ecba-eeab-493f-3f80c9b00078%40siemens.com.
