Re: [Qemu-devel] [PATCH 4/4] fixup! pc: implement pc-dimm device abstraction

["Michael S. Tsirkin" <mst@redhat.com>]

On Mon, Jun 16, 2014 at 06:33:43PM +0200, Igor Mammedov wrote:
> fix typecheck error in DEFINE_PROP_UINT64 when building with with clang,
> on OS X ram_addr_t is uintptr_t with size 8 bytes but compiler
> still refuses to accept type_check() macro as valid.
> 
> It's still fixing consequences of how ram_addr_t is defined
> but it's safer this way than to change ram_addr_t to uint64_t
> unconditionally last minute.
> 
> TODO:
> is to fix ram_addr_t on 32-bit host so that it could represent
> 64-bit guest addresses which is broken now if --disable-xen
> configure option was used.
> 
> Signed-off-by: Igor Mammedov <imammedo@redhat.com>
> ---
>  include/hw/mem/pc-dimm.h |    2 +-
>  1 files changed, 1 insertions(+), 1 deletions(-)
> 
> diff --git a/include/hw/mem/pc-dimm.h b/include/hw/mem/pc-dimm.h
> index 5f80d14..0f4a6ba 100644
> --- a/include/hw/mem/pc-dimm.h
> +++ b/include/hw/mem/pc-dimm.h
> @@ -52,7 +52,7 @@ typedef struct PCDIMMDevice {
>      DeviceState parent_obj;
>  
>      /* public */
> -    ram_addr_t addr;
> +    uint64_t addr;
>      uint32_t node;
>      int32_t slot;
>      HostMemoryBackend *hostmem;

This needs some thought.
HACKING explicitly says:
    Guest RAM memory offsets must use ram_addr_t, but only for RAM, it
    may not cover whole guest address space.

I think we need to validate addresses where they
are used/defined.

> -- 
> 1.7.1