Re: [RESEND PATCH 01/32] memory: Add RAM_PROTECTED flag to skip IOMMU mappings

[Paolo Bonzini <pbonzini@redhat.com>]

On 07/05/21 07:24, Yang Zhong wrote:
>    Paolo, i will revert this patch, but in order to disable sgx epc memory region
>    as DMA target, we still need to skip this memory region while adding this memory
>    region.
> 
>    So, my plan is to revert this patch, and then add one new bool variable in the
>    struct MemoryRegion.
> 
>    The new patch as below and it will be as patch 2 in the next version
> 
>    diff --git a/backends/hostmem-epc.c b/backends/hostmem-epc.c
>    index 0c98812f0d..316e1ee688 100644
>    --- a/backends/hostmem-epc.c
>    +++ b/backends/hostmem-epc.c
>    @@ -67,6 +67,7 @@ void sgx_memory_backend_reset(HostMemoryBackend *backend, int fd,
>         }
> 
>         sgx_epc_backend_memory_alloc(backend, errp);
>    +    mr->sgx_epc_device = true;
>    }
> 
>    static void sgx_epc_backend_instance_init(Object *obj)
>    diff --git a/hw/vfio/common.c b/hw/vfio/common.c
>    index 5bc5d29358..7a53855436 100644
>    --- a/hw/vfio/common.c
>    +++ b/hw/vfio/common.c
>    @@ -538,7 +538,7 @@ static bool vfio_listener_skipped_section(MemoryRegionSection *section)
>     {
>         return (!memory_region_is_ram(section->mr) &&
>                 !memory_region_is_iommu(section->mr)) ||
>    +            section->mr->sgx_epc_device ||
>                 /*
>                  * Sizing an enabled 64-bit BAR can cause spurious mappings to
>                  * addresses in the upper part of the 64-bit address space.  These
>     diff --git a/include/exec/memory.h b/include/exec/memory.h
>     index 2816e52be3..9bba0a1590 100644
>     --- a/include/exec/memory.h
>     +++ b/include/exec/memory.h
>     @@ -463,6 +463,7 @@ struct MemoryRegion {
>          bool readonly; /* For RAM regions */
>          bool nonvolatile;
>          bool rom_device;
>     +    bool sgx_epc_device;
>          bool flush_coalesced_mmio;
>          uint8_t dirty_log_mask;
>          bool is_iommu;
>     
>    If this patch is not suitable to skip this memory region, any suggestion is
>    appreciated, thanks!

No, you can keep RAM_PROTECTED.  I was just remarking on the new bool 
argument, and suggesting passing the flags (RAM_SHARED and 
RAM_PROTECTED) directly  tot he memory_region_init* functions.

Paolo